Milevsky

Life Annuities:
An Optimal Product for
Retirement Income
Moshe A. Milevsky

Life Annuities: An Optimal Product for Retirement Income

available online at
www.cfapubs.org
ISBN 978-1-934667-56-9
90000

9 781934 667569
Electronic copy available at: http://ssrn.com/
abstract=2571379

RF Milevsky Cover 5-2013.indd 1-3

Electronic copy available at: https://ssrn.com/abstract=2571379 5/20/2013 8:18:09 AM
 Named Endowments The Research Foundation of
CFA Institute
The Research Foundation of CFA Institute acknowledges with sincere gratitude the Board of Trustees
generous contributions of the Named Endowment participants listed below. 2012–2013
Gifts of at least US$100,000 qualify donors for membership in the Named Endowment
category, which recognizes in perpetuity the commitment toward unbiased, practitioner- Chair Pranay Gupta Brian Singer, CFA
oriented, relevant research that these firms and individuals have expressed through Jeffery V. Bailey, CFA Lombard Odier Darier William Blair, Dynamic
their generous support of the Research Foundation of CFA Institute. Target Corporation Hentsch (Asia) Limited Allocation Strategies
Renee Kathleen-Doyle Blasky, Walter V. “Bud” Haslett, Jr., CFA Fred H. Speece, Jr., CFA*
Ameritech Meiji Mutual Life Insurance Company CFA CFA Institute Speece Thorson Capital
Vista Capital Ltd. Daniel S. Meader, CFA Group Inc.
Anonymous Miller Anderson & Sherrerd, LLP
Dwight Churchill, CFA Trinity Private Equity Group Charles J. Yang, CFA
Robert D. Arnott John B. Neff, CFA
Sunapee, NH Frank K. Reilly, CFA* T&D Asset Management
Theodore R. Aronson, CFA Nikko Securities Co., Ltd.
William Fung University of Notre Dame Wayne H. Wagner
Asahi Mutual Life Nippon Life Insurance Company of Japan London Business School Venice Beach, CA
Thomas M. Richards, CFA
Batterymarch Financial Management Nomura Securities Co., Ltd. James P. Garland, CFA Nuveen HydePark Group, Arnold S. Wood
Boston Company Payden & Rygel The Jeffrey Company LLC Martingale Asset
Management
Boston Partners Asset Management, L.P. Provident National Bank John T. “JT” Grier, CFA John D. Rogers, CFA
Virginia Retirement System CFA Institute
Gary P. Brinson, CFA Frank K. Reilly, CFA
Brinson Partners, Inc. Salomon Brothers
*Emeritus
Capital Group International, Inc. Sassoon Holdings Pte. Ltd.
Concord Capital Management Scudder Stevens & Clark
Dai-Ichi Life Company Security Analysts Association of Japan Officers and Directors
Daiwa Securities Shaw Data Securities, Inc. Executive Director Secretary
Mr. and Mrs. Jeffrey Diermeier Sit Investment Associates, Inc. Walter V. “Bud” Haslett, Jr., CFA Tina Sapsara
CFA Institute CFA Institute
Gifford Fong Associates Standish, Ayer & Wood, Inc.
Gary P. Brinson Director of Treasurer
Investment Counsel Association State Farm Insurance Companies Research Kim Maynard
of America, Inc. Sumitomo Life America, Inc. Laurence B. Siegel CFA Institute
Jacobs Levy Equity Management T. Rowe Price Associates, Inc. Research Foundation of
CFA Institute
John A. Gunn, CFA Templeton Investment Counsel Inc.
Jon L. Hagler Foundation Travelers Insurance Co.
Research Foundation Review Board
Long-Term Credit Bank of Japan, Ltd. USF&G Companies
Lynch, Jones & Ryan Yamaichi Securities Co., Ltd. William J. Bernstein Stephen A. Gorman, CFA Alan Marcus
Efficient Frontier Advisors Wellington Management Boston College
Company Paul O’Connell
Stephen J. Brown
Senior Research Fellows New York University Elizabeth R. Hilpman FDO Partners
Barlow Partners, Inc. Krishna Ramaswamy
Financial Services Analyst Association Sanjiv Das
Santa Clara University Paul D. Kaplan University of Pennsylvania
Morningstar, Inc. Andrew Rudd
Bernard Dumas
Robert E. Kiernan III Advisor Software, Inc.
For more on upcoming Research Foundation INSEAD
Advanced Portfolio Lee R. Thomas
publications and webcasts, please visit Stephen Figlewski Management Pacific Investment
www.cfainstitute.org/about/foundation/. New York University
Robert W. Kopprasch, CFA Management Company
Research Foundation monographs Gary L. Gastineau The Yield Book Inc. Robert Trevor
are online at www.cfapubs.org. ETF Consultants, LLC Macquarie University
Andrew W. Lo
William N. Goetzmann Massachusetts Institute of
Yale School of Management Technology

Electronic copy available at: http://ssrn.com/

abstract=2571379

RF Milevsky Cover 5-2013.indd 4-6

Electronic copy available at: https://ssrn.com/abstract=2571379 5/20/2013 8:18:09 AM
 Moshe A. Milevsky
Associate Professor of Finance,
Schulich School of Business, York University
Executive Director of the IFID Centre, Toronto

Life Annuities:
An Optimal Product for
Retirement Income

Electronic copy available at: http://ssrn.com/

abstract=2571379

Milevxky.indb 1
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:31 PM
 Statement of Purpose

The Research Foundation of CFA Institute is a

not-for-profit organization established to promote
the development and dissemination of relevant
research for investment practitioners worldwide.

Neither the Research Foundation, CFA Institute, nor the publication’s

editorial staff is responsible for facts and opinions presented in this pub-
lication. This publication reflects the views of the author(s) and does not
represent the official views of the Research Foundation or CFA Institute.

The Research Foundation of CFA Institute and the Research Foundation logo are
trademarks owned by The Research Foundation of CFA Institute. CFA®, Chartered
Financial Analyst®, AIMR-PPS®, and GIPS® are just a few of the trademarks owned by
CFA Institute. To view a list of CFA Institute trademarks and the Guide for the Use of
CFA Institute Marks, please visit our website at www.cfainstitute.org.
© 2013 The Research Foundation of CFA Institute
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior written permission of the copyright holder.
This publication is designed to provide accurate and authoritative information in regard
to the subject matter covered. It is sold with the understanding that the publisher is not
engaged in rendering legal, accounting, or other professional service. If legal advice or other
expert assistance is required, the services of a competent professional should be sought.
ISBN 978-1-934667-56-9
17 May 2013

Editorial Staff
Elizabeth Collins
Book Editor

Abby Farson Pratt Mike Dean

Assistant Editor Publishing Technology Specialist

Cindy Maisannes Christina Hampton

Manager, Publications Production Publishing Technology Specialist

Randy Carila
Publishing Technology Specialist

Milevxky.indb 2
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:31 PM
 Biography

Moshe A. Milevsky is an associate professor of finance at the Schulich

School of Business and a member of the graduate faculty in the Department
of Mathematics and Statistics at York University in Toronto, where he is also
the executive director of the not-for-profit Individual Finance and Insurance
Decisions (IFID) Centre. He has published 10 books and more than 60 peer-
reviewed articles in such journals as Insurance: Mathematics and Economics,
Financial Analysts Journal, Journal of Risk and Insurance, Journal of Banking and
Finance, Journal of Financial and Quantitative Analysis, Mathematical Finance,
Journal of Pension Economics and Finance, and Journal of Portfolio Management.
In 2006, together with Roger G. Ibbotson, Peng Chen, CFA, and Kevin X.
Zhu, Professor Milevsky was awarded a Graham and Dodd Scroll Award for
their Financial Analysts Journal article on life-cycle financial planning, “Human
Capital, Asset Allocation, and Life Insurance.” In 2003, he was granted two
national (Canada) magazine awards for his popular writing on personal finance.
Professor Milevsky is a 2002 fellow of the Fields Institute for Research in
Mathematical Sciences and has delivered more than 1,000 seminars and keynote
presentations around the world on retirement income planning. He holds an
MA from the Department of Mathematics and Statistics and a PhD in finance
from the Schulich School of Business, both at York University in Toronto.

Milevxky.indb 3
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:31 PM
Milevxky.indb 4
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:31 PM
 Contents
Foreword. . ................................................................................. vii
Preface...................................................................................... ix
Acknowledgments...................................................................... xi
1. Institutional Details................................................................. 1
Q1. What Is a Life Annuity, and What Flavors Do They
Come In?........................................................................ 1
Q2. How Long Have Life Annuities Been Available, and
Who Invented Them?....................................................... 4
Q3. How Are Life Annuities Related to Defined Benefit
Retirement Pensions?....................................................... 6
Q4. The Term Structure of Longevity-Contingent Claims:
What Do the Claims Yield?. . .............................................. 10
Q5. Historical Data: How Have Life Annuity Yields
Changed over Time?........................................................ 13
Q6. How Is Life Annuity Income Taxed, and Is It
Economically Neutral?...................................................... 16
Q7. Who Sells Life Annuities (in North America), and How
Are They Regulated?........................................................ 19
Q8. What Does the Insurance Company Do with the
Premiums?...................................................................... 22
Q9. Credit Risk: What Happens If the Company Goes
Bankrupt?. . ..................................................................... 24
Q10. Do the Credit Ratings of the Insurance Company
Affect Payouts?............................................................... 27
Q11. A First Look at Methuselah Risk: What If Annuitants
Lived for 969 Years?......................................................... 30
Q12. Are Life Annuities Popular, and What Is the Size of
the U.S. Market?.............................................................. 33
Q13. Is a Variable Annuity with a Guaranteed Lifetime
Withdrawal Benefit a Substitute for a Life Annuity?............ 35
2. Ten Formulas to Know............................................................. 39
Q14. What Is a Biological Mortality Rate, and How Is It
Measured?...................................................................... 39
Q15. How Are Mortality Rates Converted into Survival
Probabilities?.................................................................. 41
Q16. What Is the Benjamin Gompertz Law of Mortality?.......... 45
Q17. Valuation: What Is the Gompertz Annuity Pricing Model?. 49
Q18. What Are the Duration and Interest Rate Sensitivity
of a Life Annuity?............................................................ 52
Q19. What Is the Money’s Worth Ratio of a Life Annuity?........ 54

Milevxky.indb 5
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:31 PM
 Q20. Can You Afford to Wait? Introducing the Implied
Longevity Yield. . .............................................................. 57
Q21. What Is the Lifetime Ruin Probability from
Self-Annuitizing?............................................................. 59
Q22. How Does a Variable Immediate Annuity Work?.............. 62
Q23. What Is the Difference between a Tontine and a Life
Annuity?. . ....................................................................... 65
3. The Scholarly Literature........................................................... 69
The Life-Cycle Model and Life Annuities. . ............................... 70
Actuarial Pricing, Valuation, and Reserving. . ........................... 79
Optimal Product Allocation and Timing.. ................................ 83
Defining and Solving the Annuity Puzzle.. .............................. 94
The Money’s Worth Ratio around the World........................... 104
Other Institutional and Policy Literature. . ............................... 109
4. Conclusions and Final Thoughts................................................ 112
Final Takeaways of the Discussions . . ...................................... 112
Imagining the Life Annuity of 2020....................................... 116
Bibliography ........................................................................................... 119

This publication qualifies for 5 CE credits under the guidelines

of the CFA Institute Continuing Education Program.

Milevxky.indb 6
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Foreword

Of all the hurdles that individual investors face in saving for retirement, perhaps
the most challenging is the need to avoid running out of money before they
die. The technology that solves this problem—the life annuity—has existed for
centuries and, amazingly, predates ordinary stocks and bonds. A life annuity
is a contract in which an insurance company or other financial intermediary,
having received the investor’s capital, pays him or her a fixed income (which
may or may not be adjusted for inflation) for the rest of the investor’s life.
Could there be a more perfect marriage of need and ability, of demand and
supply? Yet, few people actually invest in life annuities; the only guaranteed
income they typically receive is the mandatory government-provided annuity
known in the United States as Social Security. Instead, investors widely believe
that life annuities are a “rip-off ” sold by unscrupulous insurance companies to
unsophisticated victims and refuse to consider them. Investors prefer to man-
age the savings decumulation (opposite of accumulation) process themselves.
To do so, many use such heuristics as the 4% withdrawal rule, which is guar-
anteed to fail in some small, but significant, percentage of scenarios.
To help investors understand life annuities as an invaluable tool for creat-
ing retirement income that cannot be outlived, Professor Moshe Milevsky of
York University in Toronto has produced a book that is delightfully entertain-
ing and richly informative. He seems to have forgotten the rule that techni-
cally detailed material must be a tough read.
Longtime readers of Research Foundation books will remember Milevsky
from a 2007 book he co-authored on the subject of how to achieve lifetime
financial security.1 In that work, the authors offered lifetime financial advice
that included life annuities as an important element in one’s tool kit. Here,
Milevsky focuses on this particular financial instrument. He recounts the long
history of life annuities (they existed in 1700 BCE and were popular, much
later, among Roman soldiers) and delves into the details of how insurers today
use modern statistical analysis to calculate the fair price of a life annuity.

Why Are Life Annuities Unpopular?

One of the obstacles to the use of annuities as a retirement tool may be the
perception that annuitization is an all-or-nothing decision: You either annui-
tize your wealth—exchange it for a guaranteed lifetime income—or you
manage decumulation yourself. Most people shy away from exchanging all
1
Roger G. Ibbotson, Moshe A. Milevsky, Peng Chen, and Kevin X. Zhu, Lifetime Financial
Advice: Human Capital, Asset Allocation, and Insurance (Charlottesville, VA: Research Foundation
of CFA Institute, 2007).

©2013 The Research Foundation of CFA Institute  vii

Milevxky.indb 7
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Life Annuities

their money (wealth or capital) for an income stream. The psychology goes
something like this: Almost everyone has been without money at some point.
Having escaped that predicament by accumulating a nest egg, they find part-
ing with it just too difficult, no matter what the promised reward.
Moreover, the exchange of assets for income is not made easier by the large
element of uncertainty surrounding whether one will live to receive the income.
Even if the creditworthiness of the insurer or annuity issuer is perfect, the
income stream continues only as long as the annuitant is alive. Thus, one might
exchange one’s life savings for an income stream that lasts only a year—or a
month. Of course, if the deal is priced fairly, this outcome is balanced by the
possibility of living to collect 40 or more years of payments when the annuity
issuer is only expecting to pay out for 20 years. If the investor does in fact live a
long time, his or her ability to collect annuity income is very important because
going back to work becomes a less practical strategy with each passing year and
is really quite unlikely at, say, age 105.

Making Retirement Assets Last a Lifetime

A more optimal solution, then, might be to buy a life annuity with some of
one’s wealth and hold the rest of it directly. In a recent Financial Analysts
Journal article, the authors propose a retirement structure in which the inves-
tor saves enough to live from age 65 to 85 by using a self-managed strategy
and also purchases a deferred life annuity in which the income does not begin
until age 85 to provide for those older years. 2
This structure is attractive partly because the deferred annuity is inexpen-
sive. It’s a bargain because the annuity is priced on the expectation that many
annuitants will not live to age 85 and many more will live to collect for only a
few years, so the issuer is on the hook for few payments. Annuitants who live
to a very old age will profit tremendously at the expense of everyone else in the
risk pool. At the same time, the annuitant keeps and manages most of his or
her capital during the critical early years of retirement, when the individual has
more opportunity for leisure-related spending and more options to earn money.
Milevsky presents the case for and against annuities structured with a
variety of terms, conditions, bells, and whistles. And he presents the material
with humor and clarity. The Research Foundation of CFA Institute is excep-
tionally pleased to present Milevsky’s book. I cannot imagine a reader who
will not be captivated by this tale.
Laurence B. Siegel
Gary P. Brinson Director of Research
Research Foundation of CFA Institute
Stephen C. Sexauer, Michael W. Peskin, and Daniel P. Cassidy, “Making Retirement Income
2

Last a Lifetime,” Financial Analysts Journal, vol. 68, no. 1 ( January/February 2012):74–87.

viii ©2013 The Research Foundation of CFA Institute

Milevxky.indb 8
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Preface

In a Wall Street Journal article published 18 April 2009, two veteran reporters,
Anne Tergesen and Leslie Scism, wrote:
For years, many retirees were content to act as their own pension managers,
a complex task that involves making a nest egg last a lifetime. Now, reel-
ing from the stock-market meltdown, many are calling it quits and buying
annuities to do the job for them. In recent months, sales of plain-vanilla
immediate annuities—essentially insurance contracts that convert a lump-
sum payment into lifelong payouts—have hit an all-time high. (p. C1)
They then went on to state:
While many investors have a general idea of what an annuity is, few under-
stand the strategies available for making these products a part of their hold-
ings. You have to figure out how much to buy, whether to put your money to
work immediately or gradually, and how to invest what remains. (p. C1)
With these concerns in mind—and motivated by the practical aspects of
the retirement challenge—I provide here an overview of the body of research
on life annuities, longevity insurance, and the role of these investments in the
“optimal” retirement portfolio. I start in Chapter 1 with a basic overview of
the main institutional aspects, discuss more advanced and somewhat technical
valuation issues in Chapter 2, and conclude in Chapter 3 with a comprehen-
sive and self-contained review of the scholarly financial and economic litera-
ture on life annuities. Each of these three main chapters should be of interest
to a distinct, but hopefully overlapping, group of readers.
The data, examples, and institutional features in this book are primarily
U.S. based, and “dollars” are U.S. dollars unless otherwise noted.3 Given my
dual citizenship and current academic base at York University in Toronto,
I contrast and compare elements with the Canadian market when possible
without distracting from the main narrative. As far as I am concerned, this
book is about life annuities in a North American context.
To make this potentially dry topic readable and accessible, the first two
chapters are structured in a question-and-answer form, with answers that are
each approximately 1,000 words in length. This format allows readers initially
to skip directly to questions and issues that interest them and perhaps catch
up on the rest at a later time. Because the fields of financial economics and

3
I am a co-founder and CEO of the Quantitative Wealth Management Analytics (QWeMA)
Group, which is a Toronto-based software consulting company serving the financial services
industry. Note that some of the tables in this book use data generated by QWeMA Group
employees. These instances are noted and acknowledged in the text.

©2013 The Research Foundation of CFA Institute  ix

Milevxky.indb 9
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Life Annuities

actuarial insurance, which are logically closely related, have tended to develop
separately, I focus on bringing finance practitioners and researchers up to
speed on the mechanics, dynamics, and economics of life annuities, without
requiring a degree in actuarial science or probability theory. So, my targeted
audience includes PhDs, CFA charterholders, MBAs, Certified Financial
Planners, Retirement Income Specialists, and other researchers, as well as
practitioners in private wealth management and executives at firms that create
and sell financial retirement products.
Given demographic trends, the decline in defined benefit pension cover-
age, and the widespread acknowledgment that current benefit projections for
government pension programs are in jeopardy, financial advisers must prepare
for their emerging role as personal pension plan managers. I believe—and
most scholars in the field for the past 50 years have argued—that life annuities
are a core component of the optimal retirement income portfolio. Hopefully,
after reading this book, you will agree.

x ©2013 The Research Foundation of CFA Institute

Milevxky.indb 10
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Acknowledgments

I would like to thank Walter (Bud) Haslett, CFA, executive director, and
Laurence B. Siegel, research director, of the Research Foundation of CFA
Institute for encouraging me to pursue this project and for providing ample
and constructive feedback as this book was created. I thank the Research
Foundation also for financial support. I would also like to thank my longtime
research co-authors—specifically, Narat Charupat; Peng Chen, CFA;
Huaxiong Huang; David Promislow; Chris Robinson; Thomas Salisbury; and
Virginia Young—with whom I have been thinking, researching, and writing
about retirement income and annuities for most of my academic life at York
University in Canada. Thus, without implicating any of them in my mistakes,
errors, and omissions, much of what follows can likely be traced to (long) con-
versations with members of a team I affectionately label my “Life Annuity
Group of Seven.”
In terms of the gritty details for this book, I thank Maxwell Serebryanny
and Dajena Collaku (the IFID Centre), Simon Dabrowski (the Quantitative
Wealth Management Analytics Group), Minjie Zhang (York University),
Lowell Aronoff (Cannex), and finally Edna Milevsky (Family Inc.), all of
whom provided comments on the manuscript and assisted with editing, sourc-
ing, research, and compilation.

©2013 The Research Foundation of CFA Institute  xi

Milevxky.indb 11
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
Milevxky.indb 12
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 1. Institutional Details

Q1. What Is a Life Annuity, and What Flavors Do They

Come In?
At the risk of sounding a bit eccentric, I will start by saying that a life annu-
ity can be described as a perpetual bond (i.e., one that never matures and pays
coupons forever) subject to a peculiar type of credit default risk, which I will
describe shortly. In other words, a life annuity is a type of high-yield corporate
bond. The yield on the life annuity bond is higher than the yield from a risk-
free government bond because of the extra default risk assumed by the holder,
which is a risk that grows over time. The life annuity bond is acquired with a
lump-sum premium, and the coupons or income payments are paid monthly
by the issuer (seller) to the holder (buyer). And like any interest-sensitive
financial product, its price, or value, is inversely related to interest rates. When
interest rates move higher, the annuity’s value falls, and vice versa; in periods
of (very) low interest rates, the value of a life annuity bond is (much) more
expensive. In fact, just as with conventional bonds, you can purchase real (i.e.,
inflation-adjusted) life annuity bonds as well as nominal ones. The real ones
are more expensive, but in exchange, the periodic coupon or income payments
are adjusted for price inflation.
Of course, highlighting similarities is useful only up to a point, and the
analogy of a high-yield corporate bond can take us only so far. Alas, in contrast
to a conventional corporate bond, in which the defaulting party is the issuer or
debtor, the party who causes payments to cease in the case of a life annuity is the
holder of the bond. Moreover—and this point is the most critical and important
distinction—the triggering credit event is death, not bankruptcy or insolvency. In
other words, when the holder of the bond (the creditor or buyer) dies, the issuer
(the debtor or seller) is no longer obligated to make coupon payments. Perhaps
only an actuary would think of this characteristic as a default risk analogous to
corporate bond default risk, but that is exactly what it is. In fact, many of the
models used by financial analysts in the realm of credit risk analysis were actu-
ally first developed by actuaries for use in valuing life annuities.
Now, although there are many variations on this basic theme, I will con-
tinue to use the language of bonds—default triggers, debtors, and creditors—to
help build the intuition for how a life annuity works. Nevertheless, from this
point on, I will dispense with the word “bond” and use the term “life annuity”
exclusively. I will also be careful to use the word “life” before “annuity,” to dis-
tinguish a true life annuity from instruments and vehicles that use annuity in
their (marketing) title but have nothing to do with bonds or income for the rest
of your life. So, please do not confuse “XYZ insurance company–sold annuity”

©2013 The Research Foundation of CFA Institute  1

Milevxky.indb 1
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Life Annuities

with “life annuity.” It is the difference between using the term “fund” and the
term “exchange-traded fund” (ETF). An ETF is obviously a fund, but calling
something a fund does not mean that you can buy or sell it on an exchange or
that its underlying holdings are stocks or bonds or anything else that you think
of when hearing the designation “ETF.”
Here is an example that should help explain how a life annuity works.
Suppose in early September 2012, a 65-year-old male annuitant was quoted a
figure of US$100,000 to purchase a life annuity paying $532 per month for the
rest of the annuitant’s life. He would receive $532 per month—which is $6,384
per year, and thus 6.4% of his $100,000 premium—as long as he remained alive.
The payments would be fixed in nominal terms and would not be adjusted for
inflation. And when he died, the payments would cease.
This price, which is usually quoted in terms of monthly payouts per
$100,000, is the actual average of the top five vendors in the U.S. market in early
September 2012.4 Notice that the 6.4% yield is higher than the 3% yield avail-
able from 30-year U.S. Treasury bonds at the time—or from any perpetual bond,
if such was available—(1) because of the mortality risk accepted by the annui-
tant and (2) because the 6.4% includes a return of capital over the annuitant’s
lifetime. That is, the annuitant takes the risk that he will die (early) and lose all
future payments. This particular—and stark—manifestation could be described
as a pure life annuity with no period certain (PC). Indeed, the annuitant (or his
beneficiary) has no certainty of receiving any payments.
Another way to think of it is as a gamble: If the 65-year-old male annuitant
lived for exactly 15.7 (100,000/6,384) more years, he would just barely get his
original investment back. Every year he lived beyond age 80.7 would be pure
“investment gravy”—of course, ignoring the time value of money or interest he
could have earned on a T-bond instead. And if the annuitant died before age 80.7,
he would “lose the bet,” having paid $100,000 and received less, perhaps much
less, in return. These breakeven winning and losing benchmarks are extremely
rough estimates but can be excused at this early stage of our exploration.
Mechanics aside, you do not need a degree in behavioral finance to hypoth-
esize that such a product—in which all is lost upon death—is not palatable to
most investors. This is the likely reason that most life annuities are purchased
with additional guarantees that provide assorted death benefits and/or that
stipulate that payments must continue (to someone) in the event of death.
Naturally, you do not need a degree in financial economics to appreciate why
those enhanced annuities will be more expensive; that is, the monthly payments
will be lower if all other things are equal. Exhibit 1 provides a high-level sum-
mary of the many bells and whistles available when purchasing or investing in a
life annuity.
4
The source of this material is the QWeMA Group, Toronto, Canada.

2 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 2
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Institutional Details

Exhibit 1.  Common Life Annuity Features

Terminology Explanation Popularity
Term certain Nothing to do with life or Only as a bond substitute.
longevity. Economically,
it is a portfolio of zero-
coupon bonds.

0 PC Payments cease at death, Rare form, viewed as too

even if it occurs soon after risky.
the original purchase date.
All is forfeited.

5-, 10-, 15-, or 20-year PC Original payments Common form purchased.

guaranteed to continue to
beneficiary up to a fixed
number of years.

Joint and survivor at x% Fraction of payment Common form purchased.

guaranteed to continue to
a survivor (spouse) while
that person is alive.

COLA at y% growth Cost of living adjustment Rarely purchased.

or CPI-linked, with (COLA) at a fixed
maximum cap percentage yearly or linked
to a consumer price index
(CPI). More expensive;
thus, less income.

Refund annuity In the event of early death, Increasingly popular

beneficiary gets refund of option.
original purchase premium
minus all payments
received.

Advanced life delayed Nonrefundable premium Small number of buyers

annuity (ALDA) or paid today, but income but growing rapidly.
deferred income annuity begins much later—
(DIA) assuming annuitant is still
alive.

Exhibit 1 is not necessarily an exhaustive list of all the permutations avail-

able, and you can actually purchase a combination of the listed features. Think of

©2013 The Research Foundation of CFA Institute  3

Milevxky.indb 3
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:32 PM
 Life Annuities

it as a cafeteria menu. Thus, for example, a couple—say a 65-year-old male and

a 62-year-old female—might spend $100,000 to purchase a life annuity with an
80% survivorship benefit and a 10-year PC. It might pay $400 per month while
both annuitants were alive. If the primary holder—the male, in this case—died
within 10 years of purchase, his spouse would continue to receive the $400 per
month. Then, once the 10 years from the purchase date had passed, if she were
still alive at age 72, she would continue to receive 80% of the $400, or $320, for
the remainder of her life. Note that if both died during the 10-year PC, the heirs
or beneficiaries would get $400 per month during the certainty period.
Notice the extra guarantees compared with the life-only version. The annu-
itants get only $400, compared with the previously described 65-year-old male
who would be entitled to $532 per month. This difference represents the trade-
off between risk and return and is at the heart of understanding life annuity
pricing. In exchange for taking on mortality risk—that is, the risk that you
might die “early” (and that your beneficiaries lose the original corpus5), you are
entitled to higher payments while you are alive. The more guarantees, refunds,
and options you add on to the life annuity, the more you water down the bene-
fits of longevity pooling: The income that you receive from a life annuity—above
and beyond your original principal and the interest it earns—is other people’s
money. It is a transfer, arranged by the insurance company through pooling of
risk, from people who die while you remain alive. So, if you are not willing to
forfeit your money to the pool when you die at some point by selecting long-
dated guarantees, then all you are left with is an expensive bond product.
In the next section, I will step back and discuss the history of these products.

Q2. How Long Have Life Annuities Been Available, and

Who Invented Them?
Nonspecialists are usually surprised to learn that life annuities existed and
were widely used long before bonds ever traded. In fact, national govern-
ments used annuity-like instruments to finance deficits well before they bor-
rowed by using the fixed-maturity bonds recognizable today. Life annuities
have a long and illustrious history going back thousands of years. Their exis-
tence predates common stocks, saving bonds, and, certainly, mutual funds and
ETFs. Yes, insurance companies—the only entities allowed to issue life annui-
ties nowadays—have been around only since the middle of the 18th century,
but churches, cities, and states issued life annuities (and tontines, which I will
touch on later) long before then.
In fact, if you had to pinpoint the first life annuity ever purchased (or
invented), it would probably be sometime around 1700 BCE, give or take a few
centuries. According to research cited by Kopf (1927), archaeologists in Egypt
5
Corpus is another way of saying original principal or investment.

4 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 4
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Institutional Details

uncovered evidence that a life annuity was purchased by a prince ruling the
region of Sint in the Middle Kingdom (1100–1700 BCE). The annuitant’s name
was Prince Hepdefal, but we know little else about the annuity itself, in what
units it was paid for, and whether it ended up being a good investment for him.
More recently—around the sixth century BCE—the Old Testament in
2 Kings, chapter 25, makes reference to the (life) annuity that was granted to
Jehoiakim, king of Judah, on his release from prison, by the king of Babylon. By
the second and third centuries CE, life annuities were quite popular in Rome,
where mutual aid societies of the Roman legions granted them to soldiers who
retired from military service at the age of 46. The life annuity’s ubiquity is con-
firmed by the Roman jurist Ulpianus, who created a pricing matrix for life annui-
ties based on the life expectancy of the annuitant. Although the prices themselves
are crude from today’s perspective, the document is popular with insurance his-
torians and is known as the Table of Ulpian. See Kopf (1927) for more details.
Over the next 1,000 years, primarily monasteries and churches sold and
dealt in life annuities. A well-known example (to insurance historians) is the
annuity sold in 1308 CE by the abbot of St. Denis, not far from Paris, to the
archbishop of Bremen. The archbishop paid 2,400 livres for the life annuity
and, in exchange, was granted 400 livres per year, which is a yield of 16.66%
for life and much more than you might expect today. Perhaps not surprisingly,
15 years after issuing the life annuity, the abbot, claiming that the amount was
usurious, contested the payment to the archbishop.
Life annuities were viewed by many as a legitimate way of receiving
interest without violating the laws and doctrines against usury. The rationale
was that the mortality risk taken by both parties to the transaction made the
instrument more of a gamble than a forbidden loan with interest. (Amusingly,
interest was banned but gambling was acceptable.)
One of the earliest regions in which cities themselves (rather than religious
organizations) issued life annuities was in the area of Flanders and Brabant
(modern-day Belgium). Historians there have located detailed life annuity
certificates (what we call “policies”) dating back to the years 1228–1229.
By the 16th century, the granting, or sale, of annuities was done primarily
by cities and governments to finance budget deficits. For example, in 1554, the
Dutch Republic borrowed 100,000 guilders by selling life annuities—probably
the first such sale by a government. The English did the same in 1693, during the
reign of King William and Queen Mary, to finance a war against France.
Interestingly, most of the life annuities issued during this period did not
offer age-based payouts. In other words, 30-, 60-, and presumably 90-year-
olds were all offered the same rate, which today seems preposterous. In fact,
scientists of that era predicted that this practice would lead to eventual prob-
lems. Edmond Halley, the famous British astronomer, wrote an influential

©2013 The Research Foundation of CFA Institute  5

Milevxky.indb 5
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Life Annuities

article in 1693 in which he formally priced a life annuity and showed how its
payout should depend on age. Another century—and many more scandals and
crises—would pass, however, before his ideas on properly pricing life annuities
gained currency (an excellent early example of policymakers and governments
ignoring the research of academics to the detriment of their citizens).
During the past 250 years or so, the sale of life annuities has been the
exclusive purview of life insurance companies. And as they slowly took over
the business, the insurance industry took better care to use scientific principles
when pricing and quoting annuities. They created the field of actuarial science,
hired actuaries, gathered mortality statistics, set reserves, and managed risk.
The first formal (incorporated) insurance company was the Equitable
Life Assurance Society of London. Initially, it sold life insurance—making
payments to widows and orphans—but it eventually graduated to selling life
annuities. Sadly, life annuities almost caused the company’s demise. In the year
2000, almost 240 years after receiving its royal charter, Equitable Life nearly
went bankrupt because of guaranteed annuity promises it could not afford to
keep. That story is told briefly in the later section on credit risk
Exhibit 2 provides a subjective bird’s eye overview of the history of life
annuities going back more than 3,500 years. Much more detailed and com-
prehensive reviews are available in Kopf (1927), Lewin (2003), and Poterba
(2005), which are the underlying sources for most of the material in Exhibit 2.
Toward the early part of the 20th century, life annuities became inter-
twined with retirement pensions. In 1918, the industrialist Andrew Carnegie
established in the United States the Teachers Insurance and Annuity
Association (the TIAA part of what is today TIAA-CREF) to grant life
annuities to retiring university professors and college teachers—many of
whom lived in poverty once they stopped teaching.
By the end of the 20th century, retirement pensions and life annuities
were often considered one and the same. The next section will discuss the eco-
nomic similarities and regulatory differences between the two.

Q3. How Are Life Annuities Related to Defined Benefit

Retirement Pensions?
Life annuities are often confused—justifiably—with defined benefit (DB)
retirement pensions. Both instruments entitle the holder (annuitant or pen-
sioner) to a guaranteed and predictable lifetime of income that cannot be out-
lived. The payments can be guaranteed for a fixed number of years, continue to
a spouse and/or to beneficiaries, and or be adjusted for price inflation. To the
novice (or theoretical economist), then, retirement pensions and life annuities

6 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 6
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Institutional Details

Exhibit 2.  Important Milestones in the History of Life Annuities

Year Event
1100–1700 BCE Egyptian Prince Hepdefal, based in Sint in the period of the
Middle Kingdom, acquires first recorded personal (life) annuity.

225 CE Roman law jurist Domitius Ulpianus creates first pricing matrix
for life annuities based on the life expectancy of the annuitant.

1554 The Dutch Republic (Holland) borrows 100,000 guilders by

selling life annuities, which is the first time they are issued by a
national government.

1671 Johan de Witt, the prime minister of Holland, derives a

mathematical relationship between the price of a life annuity
and term certain annuity, but unfortunately, he is subsequently
lynched by a Dutch mob in the aftermath of the war with
France.

1693 To go to war with France, the English government under King

William and Queen Mary tries to borrow a million pounds by
means of a tontine, a form of life annuity, that offers the same
guaranteed dividend at all issue ages. One nominee actually lives
to the age of 100. Also, the first asset pricing formula for a life
annuity that proves age should affect payouts is developed by
the English astronomer Edmond Halley. His pricing advice was
ignored for centuries.

1720 Wide-spread annuity fraud problems are uncovered in England

by relatives assuming the identity of dead annuitants and
claiming their income. New regulations are imposed.

1762 The Equitable Life Assurance Society of London is formed.

It is the first mutual insurance company to issue life insurance
and then life annuities. It closed to new business in 2001 after
massive losses on guaranteed annuities.

1918 Andrew Carnegie establishes Teachers Insurance and Annuity

Association of America (TIAA) to grant (deferred) life annuities
to retiring professors.

2012 More than $8 billion of individual life annuities are purchased

every year from more than 25 major insurance companies in the
United States and Canada.

©2013 The Research Foundation of CFA Institute  7

Milevxky.indb 7
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Life Annuities

should be viewed as the same thing. Yet, despite the similarities in terms of
what they do and how they work, there are some key differences between
them that are worth emphasizing.
First, an important aspect is that not all retirement pensions are actually taken
as life annuities. Many retirees who are entitled to a life annuity opt instead to
receive the payment in one large cash payout at the point of retirement. This choice
is known as “cashing out” of a pension plan or taking a “lump-sum” payment. This
option is not available with government pensions, such as Social Security (or the
Canadian Pension Plan, CPP), but a number of employer-based pension plans do
offer this choice. Taking the upfront cash when it is offered can be tempting.
Just as important is that not all retirement pension plans actually offer a
life annuity option at retirement. In many cases, a DB plan is simply not part of
the arrangement. In fact, defined contribution (DC) or money-purchase (MP)
plans usually offer their participants or members only a lump-sum option when
they leave or retire from service. If retirees want a life annuity, they must go
to the retail market and buy it. They are on their own. In fact, some observers
argue—and I am in this camp—that a retirement plan that offers its members
a lump sum of money at retirement with no provision to exchange the lump
sum for a guaranteed income should not be called a “pension plan” at all. It is a
retirement savings plan or, perhaps, a retirement investment plan. If there is no
life annuity at the end of the tunnel, then it is not truly a pension.
In my opinion, 401(k)s, 403(b)s, or IRAs (RRSPs or RPPs6 in Canada) are
not pension plans precisely because they do not provide guaranteed life annui-
ties to their members at retirement. DB pension plans that promise a periodic
stream of income at retirement (i.e., life annuities) are true pensions. (Some legal
scholars disagree with me on this question, but almost all economists agree.)
Interestingly, a number of large DB (proper) pension plans that had
promised their participants a lifetime annuity have recently offered their retir-
ees (i.e., those receiving income already) an option to cash out and receive
a lump sum. One of the more well-known cases involved General Motors
(GM), which in the summer of 2012 offered more than 40,000 of its salaried
retirees the option to stop their small monthly checks and receive, instead, one
much larger check. Chapter 3 and the literature review discuss how people
actually behave when given such choices.
In this case, moreover, for those retirees who opted to continue receiving their
monthly pension checks (i.e., their life annuities), the obligation to make those
payments was transferred to Prudential Finance, an insurance company. Either
way, the plan was for GM to wash its hands of the relationship with retirees.

6
Registered retirement savings plan and registered pension plan, respectively.

8 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 8
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Institutional Details

Although we do not know exactly how many ex-GM workers chose the
lump sum over the life annuity, this offer was unprecedented. Note that it was
given to people already in retirement—receiving their monthly income—as
opposed to those about to retire. Such offers are a growing trend and show the
close relationship between life annuities and retirement pensions.
Exhibit 3 displays some other aspects of the distinction between the two. For
example, in the United States, a corporate pension plan, such as GM’s plan, is
regulated (i.e., monitored and policed) by the federal government. The Employee
Retirement Income Security Act (ERISA) of 1974 is a federal law that sets
minimum standards for pension plans in private industry. Although the U.S.
Department of Labor (DOL) enforces ERISA, it does not require any employer
to establish a pension plan. It only requires that those who do establish pension
plans meet certain minimum standards. Think of the DOL as the watchdog.

Exhibit 3.  Life Annuities vs. Retirement Pensions: Key Differences and

Similarities
Key Feature Life Annuity Retirement Pension
Offered or sold by Insurance company. Employer or sponsor.

Purchase process Pay lump-sum premium or Based on work history, salary,

via installments. and employer generosity.

Basis of payouts Females must pay more for Based on years of service and
the same lifetime income salary. Companies cannot
because they live longer. discriminate on the basis of
gender or health.

How the money Insurance company invests Pension plan managers or

backing the income conservatively in “general sponsors must “manage
is invested account” and is subject to [assets] prudently.”
extensive regulation.

Regulation (in State regulators issue rules U.S. DOL and ERISA
United States) and guarantee funds. legislation.

Protection (in National Organization of Pension Benefit Guaranty

United States) Life and Health Insurance Corporation.
Guaranty Associations
(NOLHGA).

Value per dollar Lower: pays “retail.” Higher: bought “wholesale.”

©2013 The Research Foundation of CFA Institute  9

Milevxky.indb 9
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Life Annuities

In contrast, life annuities in the United States are sold by insurance companies
and are regulated by individual states, not by the federal government. Regulatory
policy is set by state legislatures, who then oversee state insurance departments,
which, in turn, enforce state insurance laws. Although some insurance regulators
are stricter than others—the New York office is notoriously vigilant—states tend
to coordinate these activities among themselves via the National Association of
Insurance Commissioners (NAIC), so it is not as though they all pull in com-
pletely different directions. (In Canada, the Office of the Superintendent of
Financial Institutions Canada is the primary regulator and supervisor of federally
regulated deposit-taking institutions, insurance companies, and federally regulated
private pension plans.) All of these differences might sound somewhat legal-
istic and esoteric, but the federal versus state perspective does create a consider-
able difference between retirement pensions and life annuities. For example, if an
insurance company goes bankrupt—a possibility worthy of its own section—the
individual states oversee an insurance association–funded guarantee fund to help
cover the losses. In contrast, if a pension plan runs into financial difficulty, the
federal government’s guarantee fund is the source of protection. (Neither of these
kinds of funds will cover all the retiree’s losses if they exceed certain limits.)
The regulation governing life annuities sold by insurance companies is far
more stringent than the regulation governing retirement pensions, although
they are similar economic instruments. Insurance companies must invest the
money backing the annuity income conservatively and under regulation, but
pension plan managers or sponsors are simply instructed to manage the assets
“prudently,” a vague order.
In summary, although an economist might consider a retirement pension
to be the same as a life annuity, subtle legal and regulatory differences char-
acterize the two. And if you ever have a choice between getting a lifetime of
income from an insurance company or from a pension plan, one of the things
you want to consider is who you want keeping an eye on your nest egg. If you
believe that your state is competent at regulating financial institutions, then
perhaps opt for the insurance company. If you prefer federal oversight, then
perhaps the pension plan is the way to go.
Of course, more important than these legal technicalities is how much
income you will actually receive, what the income level depends on, and how
to get more. That is the topic of the next section.

Q4. The Term Structure of Longevity-Contingent Claims:

What Do the Claims Yield?
Language is important, and when it comes to life annuities, the terminology
can be confusing. The “cost” of a life annuity can be described in two ways.
You can talk about the price of a life annuity or the payout from a life annuity.

10 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 10
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Institutional Details

The two are mirror images of each other and tend to be used interchange-
ably. Thus, for example, the price of obtaining a lifetime income of $1,000 per
month might be a premium of $270,000 at age 65. In that case, the focus is
on the cost of a given income stream. Or you can start with a given premium,
say $100,000, and then talk about the payout being $435 per month. In both
cases, the underlying economics of the transaction is the same. If you divide
$270,000 by $12,000 (the annual income stream, 12 × $1,000), you get the
same annuity factor of 14 in either case. Formally, the annuity factor is defined
as the cost of $1 of income per year for the rest of your life. So, the cost of
$1,000 or $10,000 or $100,000 per year is obtained by multiplying the annu-
ity factor by the desired amount of income.
In most financial transactions, we customarily talk about the price per
unit—for example, of an ounce of gold, a share of stock, or a carton of milk.
But, when it comes to life annuities, the discussion tends to be in terms of the
payout per $100,000 premium. This approach might seem odd at first; it is akin
to discussing how much milk you might be able to get each day in exchange for
a $100 one-time upfront payment. The convention to quote in terms of monthly
income per $100,000 of premium paid is probably a historical artifact. Rest
assured, if you want to buy an annuity, the insurance company will take any sum
of money—as long as it is not too small—and will probably send you a check at
the frequency you find most convenient. Rarely will you find volume discounts,
although if you buy as part of a group—as in an employer pension plan—you
will receive a better deal on the order of 10% or so. Also, you may have to pay
a fixed policy fee—which is embedded in the quote—regardless of the size of
the premium. So, there are some scale economies that reduce the fixed costs, but
they don’t really change the underlying mechanics of a life annuity.
Here is the key economic point: Whether you view the cost as price paid
or payout received, the ratio between your premium (what you paid) and your
annual income (what you get) will be the same and is called the “annuity yield.”
Table 1 displays what these payouts were in August 2012 as a func-
tion of the buyer’s age and gender and the guarantee period selected. This
table shows what I call the “term structure” of longevity-contingent claims;
some readers will recognize “term structure” from the literature and lingo
of the bond market. Note that these prices—or, better stated, payouts—can
change from week to week and often from day to day, just as bond prices
do. So, do not expect to get these exact rates if you plan to purchase a life
annuity any time soon. (I will later discuss where exactly these numbers
come from.) Interest rate changes have a big impact on payouts and can
happen anytime.

©2013 The Research Foundation of CFA Institute  11

Milevxky.indb 11
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Life Annuities

Table 1.   Monthly Life Annuity Payouts Available per $100,000 Premium

Age 0-Year PC 5-Year PC 10-Year PC 15-Year PC 20-Year PC
55 M = $431 M = $430 M = $427 M = $422 M = $414
   F = $416    F = $416    F = $414    F = $409    F = $403

60 M = $475 M = $473 M = $468 M = $459 M = $443

   F = $456    F = $455    F = $451    F = $443    F = $432

65 M = $532 M = $529 M = $519 M = $499 M = $470

   F = $507    F = $505    F = $497    F = $482    F = $461

70 M = $613 M = $606 M = $585 M = $544 M = $497

   F = $578    F = $574    F = $557    F = $527    F = $491

75 M = $729 M = $713 M = $664 M = $589 M = $516

   F = $686    F = $674    F = $635    F = $579    F = $513

80 M = $895 M = $865 M = $749 M = $623 M = $524

   F = $838    F = $809    F = $724    F = $617    F = $523
Note: M is male; F is female; PC is period certain.
Sources: QWeMA Group (August 2012). Based on data from Cannex Financial Exchanges. Average
is based on quotes from John Hancock Insurance, MetLife, New York Life, Nationwide, and Pacific
Life Insurance Company.

Table 1 works as follows: If, for example, you are a 65-year-old male and
want to guarantee that income payments will continue for at least 10 years—even
if you are not around to collect them—Table 1 shows that the (average) pay-
out you can obtain is $519 per month, which is $6,228 per year, or 6.23% of a
$100,000 premium. If you decide to buy (only) $75,000 worth of lifetime annui-
ties, your income will be the same 6.23% of $75,000, which is $4,672 per year, or
$389 per month. Similarly, if you want $1,000 of monthly income, you need to
pay $192,616 ($12,000/0.0623) in premium. The ratio is maintained.
Now, if you want to squeeze a bit more yield, or income, from your $100,000
life annuity premium, you can dispense with the 10-year PC and select the
0-year PC. This choice is slightly riskier because if you die within 10 years (i.e.,
before the age of 75), the insurance company will not have to continue mak-
ing any payment to you (obviously) or to your beneficiaries. In exchange for
that risk, the company (or the average company) will be willing to pay $532 as
opposed to $519 per month. The amount is an extra $156 per year as compen-
sation for the extra risk. Now, whether that amount is worth it for you person-
ally depends on your own circumstances and confidence in your health. But you

12 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 12
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Institutional Details

probably would not be surprised to learn that most 65-year-olds who purchase
a life annuity forfeit the extra $13 per month and select the 10-year PC. In fact,
the research suggests that annuitants are selecting longer guarantee periods and
other bells and whistles in exchange for reduced payment. Classical economists
do not quite understand why people do this because they are thereby giving up
the longevity pooling—which was the main point of buying the annuity. I will
provide more discussion of this issue in the literature review in Chapter 3.
All of these numbers might seem overwhelming at first, but there are a
number of important patterns in them you should note and understand. First,
females consistently get less income than males do. For example, at age 65
(with a 10-year PC), a female gets only $497 compared with $519 for a male.
The almost 5% difference is because females are expected to live longer than
males. In this case, you can see mortality risk at work. Males assume more risk
than females by purchasing a life annuity at age 65, and they get compensated
for this risk. Note also that the older you are when you spend $100,000 on a
life annuity, the more income you will receive. Again, the cause is mortality
risk. In fact, the one overwhelming takeaway from the matrix in Table 1 is the
way that mortality risk drives annuity quotes.
You might wonder how exactly an insurance company determines the
appropriate payout rate to apply at different ages and genders, and that will be
addressed in Chapter 2.
In conclusion, I remind the reader that the numbers in Table 1 are aver-
ages across a variety of insurance companies in late August 2012. Despite
being a competitive market, some companies quoted higher rates and some
quoted lower. The gap between companies can reach as high as 10% on any
given day, which reflects various companies’ appetite for the business. I will
delve more into this issue when I discuss credit risk.
In the next section, I will establish why and how these rates have changed
over time. Slightly more than a decade ago, all of the numbers in Table 1 were
50–75% higher, much to the chagrin of all retirees who are in the market to
buy life annuities these days.

Q5. Historical Data: How Have Life Annuity Yields

Changed over Time?
Sadly—at least for anyone in the market to buy—the payouts from life annui-
ties were at a historical low as of August 2012. In fact, payouts have been in a
long-term downtrend for decades, ever since interest rates peaked in 1981. Here
is a case in point. In late August 2012, a 65-year-old female could receive $500
in monthly income from a 10-year PC life annuity in exchange for a $100,000
premium. (Remember that the $500 number—like most payouts I quote—is the
average of the top five, or best, companies at the time.) Yet, a mere six years earlier,

©2013 The Research Foundation of CFA Institute  13

Milevxky.indb 13
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Life Annuities

in August 2006, a 65-year-old female could have obtained $630 in monthly

income for the exact same life annuity and premium. The difference is $130 per
month and $1,560 per year. So, today’s retiree, if he or she chooses to buy a life
annuity, enjoys 20% less income than was available 6 years earlier and almost 50%
less income than 15 years ago.7 One can only sympathize with those who are
annuitizing today versus a few years ago. Timing is everything.
I repeat just to be clear: For those who already purchased their annuities—5,
10, or 15 years ago—absolutely no change or reduction in their monthly income
has occurred. They are entitled to whatever was promised to them at the time
of purchase. It was guaranteed for life. The individual who purchases his or her
annuity today is the one stuck with less income compared with a few years ago.
This decline or reduction in payouts has two causes. One is obvious, and
the other is subtle. First and foremost, the downtrend is part of the story of
low interest rates in a struggling global economy. Life annuities are similar to
fixed-income coupon bonds, and just as the declining level of interest rates
has increased bond prices over the last few years, so too have they increased the
cost of life annuities. Remember, increasing the cost of a life annuity means
that it costs more to generate the same income; that is, the annuitant receives
less income for the same $100,000 premium. Ergo, payouts have dropped.
Note that back in the summer of 2006, well before the financial crises
erupted in 2007 and 2008, the yield on a 10-year U.S. government bond was
hovering around 5.20%. Thus, a $100,000 investment in this sort of bond would
have generated coupons of $5,200 per year. But in late August 2012, the same
risk-free government bond yield was closer to 1.5% and a $100,000 investment
in the bond would yield only $1,500 per year in coupon income. The drop of
almost 4 percentage points in bond yields is eventually transmitted to annu-
ity prices. (I discuss the pricing of life annuities and the role of interest rates
in much more detail in Chapter 2.) It is the big culprit, if you will, in the drop
in income from life annuities. And although life annuity payouts are driven by
many interest rates other than those on pure government bonds—for example, a
mixture of mortgage rates and corporate bond rates—all of these rates are much
lower than they were in 2006. Lower rates translate to lower payouts.
A more subtle reason for the decline in payouts over time has to do with lon-
gevity risk and population aging. One of the factors that determines how much
an insurance company pays out on life annuities is its estimate or projection of
how long annuitants might live. The longer it expects a cohort of 65-year-olds
to live—that is, the longer it must continue making payments—the less it can
afford to pay in exchange for the same $100,000 premium. Although detecting

7
High yields have not been restricted to the late 1970s and early 1980s, when government bond
interest rates were in double digits; in fact, in 1693, when the British government issued one of
the first annuities to the public, the government offered 14% payouts.

14 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 14
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:33 PM
 Institutional Details

or measuring population aging year over year is difficult, increases in life expec-
tancy do have an impact over time. People are living longer, and payouts will
be made for longer. In 2000, the typical 65-year-old would have been expected
to live for 20 years; in 2010, perhaps 21 years. Insurance companies have been
responding to this demographic change by (slowly) reducing the income they
are willing to pay. Thus, even if interest rates had remained relatively constant
for the last few years, the actual payouts on life annuities would have declined,
although by a much smaller but hard-to-define amount.
Figure 1 displays historical payouts, expressed as an annualized percentage
of the premium paid for males and females, for the eight years of 2004–2012.
The figure also displays the yield on a risk-free 10-year U.S. government bond
on the same dates. I selected the 10-year U.S. government bond rate as a proxy
for general interest rates, but I am not suggesting that it is the rate that deter-
mines how insurance companies price annuities.

Figure 1.  Life Annuity Payout Rate: Males vs. Females vs. 10-Year
Treasury Rate, 2004–2012
Rate/Yield (%)
9

8 Male

7 Female

4
10-Year T-Bond

0
04 05 06 07 08 09 10 11

Note: Yearly measurements taken on 14 September.

Source: Data collected on a weekly basis from a variety of vendors and processed with the help of
Cannex Financial Exchanges and the QWeMA Group.

©2013 The Research Foundation of CFA Institute  15

Milevxky.indb 15
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Life Annuities

A few things are worth noting in Figure 1, in addition to the obvious fact
that males receive a higher yield than females because of their shorter longevity.
First, as I mentioned earlier, the trend is noticeably downward over this period.
Second, life annuity payout rates are an average of 3.5 percentage points
more than 10-year U.S. T-bonds, although this difference tends to be variable
over time, especially during times of financial stress. So, the extra 3.5% is not
a bad rule of thumb.
Third, and as important as the other two points, interest rates tend to be
more volatile than life annuity payout rates. So, although the annuity rates do
change regularly, they do not “bounce around” as much as market interest rates.
Insurance companies are (probably) smoothing the ups and downs of market
interest rates and taking their time in adjusting payouts in response. The exact
mechanism by which this happens—and how exactly interest rates and longev-
ity expectations are merged to create annuity payouts—is discussed in a num-
ber of papers mentioned in Chapter 2 and in the literature review of Chapter 3.
In summary, life annuity payouts change from week to week and often
from day to day. Some companies offer better prices than others at differ-
ent points in time or to retirees at different ages. The trend, however, is clear.
Payouts today are much lower than they have been in the past, and unless
interest rates move back to higher levels, payouts will continue to be depressed
relative to historical averages.

Q6. How Is Life Annuity Income Taxed, and Is It

Economically Neutral?
Benjamin Franklin is quoted as having said that nothing in life is certain other
than death and taxes. Life annuities are a great example of both certainties. The
income will eventually end upon the death of the annuitant, and the same income
will indeed be taxed. Fortunately, however, not all of the income is fully taxable
because some of the money you are receiving was yours already. In other words,
some of the income will be excluded from taxes if you paid tax on the money pre-
viously and the income is fully yours. I will explain more about what this means
in a moment, but first let me give a brief overview of retirement and income taxes.
When you buy a life annuity to generate income in retirement, the source
of the funds—the origin of the $100,000 premium, for example—will affect
the tax treatment of the income. There are really only two possible sources.
First, the $100,000 you are using might come from a tax shelter, such as a
401(k), IRA, or other such account (in Canada, an RRSP or MP plan). When
you use these funds to buy the annuity, you have carried out a qualified life
annuity transaction. (In Canada, it would be called a “registered life annuity.”)

16 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 16
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Institutional Details

Remember that the defining characteristic here is that you have never really
paid any income tax on this qualified (registered) money yet. The money you con-
tributed to the account was deducted from your taxable income—perhaps a long
time ago—and you also never paid any income taxes on the gains as this money
grew over time. For this reason, when you reach retirement around the age of 70
or so, you are required to start withdrawing money from these accounts and pay-
ing the income taxes you never paid when you were working and contributing.
Here is the bottom line: If you use that tax-sheltered money to purchase the
life annuity, then all of the income from the annuity is taxable. Technically, you add
this life annuity income to your other taxable income—such as pension income,
employment income, or interest income—and pay whatever tax is due on the total
amount every year.8 Now, if you have little other income, or if you have large tax
credits and/or deductions that you can take advantage of, you may end up paying
very little tax. But 100% of the income is taxable. No exclusions or exemptions.
Alternatively, you can use a different pot of money to purchase a life annu-
ity. The $100,000 premium might come from “regular” funds that are not part
of a dedicated retirement tax shelter. This source buys you a nonqualified (non-
registered in Canada) annuity, and in this case, only a portion of the lifetime
annuity income is taxable. After all, most of the money you are getting back
was yours already and you have already paid tax on it. The actual amount that is
taxable versus the nontaxable amount is determined by the insurance company
that sells you the annuity on the basis of tax rules set forth by the U.S. Internal
Revenue Service (IRS) or, in Canada, the Canada Revenue Agency (CRA).
Here is an example to help you understand the so-called exclusion ratio so
that you can determine how much of the income will actually be taxable. The
basic economic principle at work is a comparison of how much you invested
in the life annuity (the $100,000 premium, for example) with how much you
expect to receive from the life annuity. In other words, the tax rules depend on
your life expectancy at the time of purchase and the number of payments you
anticipate receiving. Note that the emphasis is on what you expect to happen, not
what actually happens. The tax authorities wanted to keep things simple in this
case and fix the amount that is taxable as opposed to varying it from year to year.
Say that, at the age of 65, you invested $100,000 in a life annuity that
promises $550 per month, which is $6,600 per year, or a payout rate of 6.6%.
According to the relevant IRS actuarial mortality table—which is a topic for
another section—you can expect to live approximately 20 years from the age
of 65. If you make it to 85 (and no more), you will receive $132,000 in total
payments. Expectation is the key concept here. You are expecting $32,000
more than you paid. So, that is the amount of gain that is expected to be

8
Some special age-based tax credits or deductions might be available on pension income, but
they would be minor.

©2013 The Research Foundation of CFA Institute  17

Milevxky.indb 17
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Life Annuities

taxed. Of course, the IRS will not tax you on the $32,000 all at once, and
it cannot wait until you die (at age 85) to get the tax on the $32,000, so it
amortizes the amount over the remainder of your (expected) life. Each pay-
ment you receive is partially taxed in proportion to the ratio of 32 to 100. A
portion of total income, however, is not taxable; that is, it is “excluded” from
taxable income. For clarity, Exhibit 4 provides an example of how the exclu-
sion ratio is computed.

   Exhibit 4.  How Much of Nonqualified Annuity Income Is Taxable?

Life annuity premium (investment) $100,000
Age at time of purchase  65
Guaranteed monthly income for life $550
Life expectancy in months (IRS tables) 240
Total amount of payments expected 240 × $550 = $132,000
Exclusion ratio    $100,000/$132,000 = 75.8%
Monthly income that is taxablea   (100% – 75.8%) × $550 = $133.33

The entire $550 is taxable after you reach life expectancy, age 85.
a

One final wrinkle can make things a bit tricky in the United States
with the exclusion ratio. Although the majority of the income you receive
will not be taxable, this favorable treatment comes to an end once you
reach your life expectancy and have received $132,000 in total payments.
From that point onward, the entire income (which was $550 in the earlier
example) is fully taxable. The exclusion ratio goes to zero, so to speak. The
reason, or justification, is that every dollar you are now getting—after you
reach age 85—is definitely more than you put in, so 100% of it is taxed.
Your beneficiary can claim a tax credit, however, on your final (i.e., year
of death) tax return if you did not recover your entire premium while still
alive.
All these rules are a rather disconcerting and complicated way to tax
income, and a number of economists—including, for example, Brown,
Mitchell, Poterba, and Warshawsky (1999)—suggest some more-efficient
alternatives. A few states in the United States also impose a small (0.5–3.0%)
upfront premium tax when annuities are purchased with nonqualified funds, a
fact that is also worth noting in the discussion of taxes.
Surprisingly, however, and in contrast to the United States, Canada’s
tax authorities allow you to continue to exclude the same portion of the life
annuity income for as long as you live, even though you have received much
more than your original premium back in payments. This treatment provides

18 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 18
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Institutional Details

a generous tax break because the CRA uses outdated mortality tables that
assume a lower life expectancy—especially when compared with the U.S.
treatment, in which 100% of income becomes taxable at some point.9
Regardless of whether you use qualified or nonqualified funds and
whether you have lived long beyond your life expectancy, part of your life
annuity income will be considered taxable—regardless of where you live or
the jurisdiction in which you reside. So, you must consider your tax situa-
tion in general and other taxable income in particular before choosing a life
annuity for a retirement portfolio. And—all else being equal, if you have a
choice—get your (taxable) life annuity in Canada if you do not mind receiv-
ing your income in Loonies.10

Q7. Who Sells Life Annuities (in North America), and

How Are They Regulated?
Unlike opening a bank account, buying bonds from the U.S. Treasury, or par-
ticipating in a dividend reinvestment plan (wherein you buy stock directly
from a company), a life annuity is not bought directly from the product manu-
facturer. You cannot contact an insurance company via its toll-free number or
on its website and then buy a life annuity from the company as you would a
book, a pair of shoes, or an airline ticket. You must conduct the transaction
through a licensed insurance agent. The agent is your financial intermediary
for transacting with the insurance company. You give the agent your premium
or investment, which the agent then gives to the insurance company, which
then issues you the life annuity certificate or policy. In fact, buying a life annu-
ity is similar to buying life insurance, which also must be done through a
licensed agent. (There are exceptions to this rule; so-called term life insurance
can be purchased from companies directly.)
Becoming a licensed insurance agent is not a complicated process. It
usually involves a few months of study, an exam, and over time, some con-
tinuing education credits to retain the license. In the United States, the indi-
vidual states administer and manage the process, and the end result is a Life
and Health Insurance License. In Canada, agents must go through the Life
License Qualification Program, which is a course of study, and then take an
exam similar in scope and intent to the exam in the United States.
Note that a license to sell a life annuity (or any insurance for that matter)
is quite different from an educational designation or qualification, such as a
bachelor’s degree, CFA or CFP (Certified Financial Planner) designation,

9
At this point, I would like to offer a further disclaimer and suggest that you contact a tax spe-
cialist before making any irreversible decisions that might affect your taxes. The discussion in
the text here is intended as a first view.
10
“Loonie” is the nickname of the Canadian one-dollar coin.

©2013 The Research Foundation of CFA Institute  19

Milevxky.indb 19
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Life Annuities

or master’s degree. Yes, a person might have learned a bit of finance, insur-
ance, and economics in those programs, but to be able to actually sell a life
annuity and to receive a commission on that sale, the agent needs the insur-
ance license.
Just to be clear here, even if you call up a big life annuity company in the
United States directly—such as New York Life or MetLife (or even an invest-
ment firm, such as Vanguard or Fidelity, for that matter)—and ask to buy a
life annuity, it will refer you to an agent, possibly one of its own. You may
seem to be bypassing the agent, but in fact, one is always in the background.
Similarly, your financial adviser, broker, or wealth manager might be able to
get you a life annuity but only if that person has an insurance license in addi-
tion to a securities (investment) license.
All of this information is more than just an institutional technicality.
These licensing requirements have two implications in practice. The first is
that someone locally (in the state or province itself ) is regulating or over-
seeing sales practices. In fact, state insurance regulators review and approve
the actual prices at which life annuities are sold. This effort should provide
a measure of comfort in the purchase process. Among other requirements,
agents must ensure that the life annuity is suitable or appropriate for the cli-
ent before they recommend it.
Second, the existence of the insurance agent as an intermediary implies
that this person will be compensated—probably via commission—for selling
the life annuity. This commission can be anywhere from 0.5% to as much as
5% of the premium investment, depending on the company and product. Rest
assured that this fee will be embedded in the quoted payout and paid by the
insurance company to the insurance agent directly. So, you will not have to pay
it separately or to the agent, but it will not be revealed to you either. One could
argue that it should be disclosed and made transparent to the buyer, but that
debate is for another book and time.
Moving on to the insurance companies themselves (the ones who manu-
facture, manage, and guarantee the annuity payments): They are also regulated
by the insurance commissions. Every policy they issue, price they charge, or
innovation they ponder must be approved by the commission. The justification
for all of this red tape is that it provides a layer of scrutiny for the sole benefit
of the consumer. No surprise, then, that the insurance industry is viewed as
one of the most highly regulated businesses.
In return, during the financial crises of 2007–2008, state insurance
regulators—especially the commissioner of insurance in New York—were
extremely active in ensuring that policyholders were insulated and protected
from emerging problems. In one case, a well-known national insurance com-
pany was attempting certain financial actions that could have benefited the

20 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 20
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Institutional Details

owners and shareholders of the company but might have harmed the policy-
holders themselves. In this high-profile case, the commissioner rode to the
rescue and ensured that the widows and orphans and, for that matter, ordi-
nary retirees would continue receiving their annuities. So, this regulation and
oversight does occasionally have its upside.
Exhibit 5 lists the 12 largest insurance companies (according to their
annuity reserves) in the United States that actively sell annuities—and not
only life annuities. In addition to the various types of annuities they sell,
each of these companies issues hundreds of millions of dollars of life insur-
ance each year. Indeed, there are many types of annuities other than life
annuities—which is discussed in another section—and the industry has
come a long way from the old days of churches and parishes paying pensions.
At the top of the list, MetLife is holding more than $278 billion worth of
funds (known as reserves) that will eventually be used to pay annuities. The
second-largest company in terms of annuity reserves is holding more than
$179 billion in funds to pay these annuities.

Exhibit 5.  Largest Life Insurers in the United States Ranked by

Total Annuity Reserves Held as of 2010
Total Annuity Reserves
Rank Name (millions)
1 MetLife $278,089
2 TIAA-CREF 179,219
3 Prudential Financial 164,136
4 American International Group 157,048
5 Manulife Financial 146,525
6 ING North America 137,757
7 Hartford Life 129,629
8 Lincoln Financial 111,089
9 AXA Financial 99,725
10 AEGON SA 95,989
11 New York Life 81,721
12 Jackson National Life 79,172
Source: Based on ACLI tabulations of National Association of Insurance Commissioners
data.

In summary, buying a life annuity is somewhat more complicated—and

perhaps even more expensive—than going online and directly buying a bond
ETF or term deposit. In the next section, I will explore what exactly the
insurance company does with all the money it collects from annuitants, and
the reserves it holds, while the retirees and annuitants wait to receive their
monthly checks.

©2013 The Research Foundation of CFA Institute  21

Milevxky.indb 21
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Life Annuities

Q8. What Does the Insurance Company Do with the

Premiums?
Insurance companies collect life annuity premiums from annuitants and must
then invest the funds for many years, possibly decades, as they slowly pay
out the lifetime income. As noted previously, the annuity reserves that insur-
ance companies hold can amount to billions of dollars. It is natural to ask,
therefore, what exactly they do with all this money. The issue is not simply a
matter of curiosity; it concerns company solvency and regulation. After all, if
the companies lose the money, not only is the annuitant affected but, given
that those who are affected tend to be older and retired, so also is the public
at large.
Rest assured that insurance regulators monitor and ensure that the
funds are safely invested for the benefit of the annuitants, and the invest-
ments themselves are (also rather boring and) transparent. Table 2 provides
a snapshot of a typical insurance company investment (asset) portfolio. The
data are from the NAIC, which is an umbrella organization and advocacy
group consisting of all the various state insurance regulators in the United
States.

Table 2.  Percentage of Invested Assets by Size of Insurance

Company Investment Assets
Big Companies Small Companies
Asset Type (>$10 billion) (<$0.25 billion)
Corporate bonds 43.1% 36.1%
Structured securities 18.7 17.0
U.S. government bonds 18.2 35.1
Commercial mortgages 9.2 2.2
Corporate stocks 1.7 2.9
Other assets 9.1 6.7
Total 100.0% 100.0%
Source: Based on data from NAIC and Center for Insurance Policy and Research,
“Capital Markets Special Report” ( July 2011).

Table 2 reveals that most insurance company (admitted) assets are

invested in bonds. A small fraction of assets—overall, less than 3% of the
investment portfolio—is in common and preferred stocks. This tiny alloca-
tion to stocks—compared with what you or I might hold—is not by choice.
Indeed, the companies are forced by regulators and by common economic
sense to hold a conservative, safe portfolio.
Nevertheless, their investment portfolios are not, of course, immune from
problems. After all, corporate bonds, mortgage bonds, and even government
bonds do fluctuate in value, can default, and do have some economic risk

22 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 22
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Institutional Details

associated with them. But the key is that these investment portfolios have low
(stock market) betas. That is, the funds are not exposed to the vagaries of the
stock market and protect current and future policyholders.
Another interesting insight from Table 2 is the difference between the
holdings of large life insurance companies (those with investment assets
greater than $10 billion) and the holdings of small insurance companies.
Although some readers might expect small companies to have relatively riskier
investment holdings, the opposite seems to be true. They actually hold more
bonds, in general, and more U.S. government bonds, in particular. The impli-
cation is that smaller companies are probably earning less on their invest-
ment assets than the larger insurance companies. Are the bigger players in
the industry perhaps “reaching for yield” by investing in riskier corporate and
mortgage bonds with higher interest and coupon payments?
Needless to say, risk and reward are linked. Given their investment
portfolios, the smaller insurance companies probably cannot offer as high
a payout on life annuities because of their lower-yielding portfolios. Add
this to the fact that consumers are less likely to trust a smaller insurance
company to fulfill an income promise that might last decades, and it is not
surprising that the life annuity business is dominated by large companies.
(After all, it is not a year-by-year car insurance policy we are talking about;
it is the retiree’s livelihood.)
Note, however, that the numbers in Table 2 refer only to the insur-
ance companies’ own invested assets, also known as general account assets.
The general account is where they deposit and comingle all of the annuity
(and insurance) premiums they receive. The rates they offer on life (and
other) annuities will partially depend on the current yield of these general
account investments. If current bond yields are low, the companies clearly
cannot afford to pay as much on new premium deposits. But an insurance
company holds another type of account, the separate account, and in that
account, anything goes.
In fact, insurance companies may be holding hundreds of billions of dol-
lars in riskier assets, including domestic and international stocks, gold, com-
modities, real estate, and assorted mutual funds. But they are holding those
assets as a custodian, guardian, or trustee and in the name of individual clients.
They hold and manage them rather than own them, which is a subtle but big
difference. These assets are not included in Table 2. They are quite separate.
To understand the separate account versus the general account, imagine
having a bond certificate in your personal safety deposit box that is outright
owned by you and, at the same time, a bond, a stock, or even a collection
of gold coins that you are holding for a friend—but charging them a fee for
this service, one that is perhaps even based on the value of the investments

©2013 The Research Foundation of CFA Institute  23

Milevxky.indb 23
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Life Annuities

you are holding in trust. In both cases, the various investments are sitting in
your safety deposit box, but the asset owned directly by you is analogous to an
insurance company’s general account assets; the assets that you are simply a
guardian or trustee of would be separate account assets. Companies and their
regulators must ensure that both accounts are kept safe and secure, but more is
at stake in the general account.
In short, the investments an insurance company owns outright are con-
servative and tightly regulated. They cannot—and would not—take any
chances investing the funds in risky or dubious assets. In fact, what they
try to do is locate investments, such as bonds, that produce cash flows or
coupons that match the payouts they are obligated to make. This mission is
known as asset/liability matching and will be discussed more carefully in a
later chapter.
At a simplistic level, an insurance company uses the life annuity premiums
it receives from all the annuitants to purchase bonds. These bonds produce
coupon income, which the insurance company then uses to pay the annuitants.
Naturally, if you are not alive to get those annuity payments, the companies
give them to the survivors. The key is for the insurance company’s actuaries
to figure out how long it will be paying the annuitants, in aggregate, so it can
purchase bonds with the right maturities and durations.
First, however, consider what happens when insurance companies—and
the investments they hold—run into financial difficulties.

Q9. Credit Risk: What Happens If the Company Goes

Bankrupt?
Although insurance companies invest the premiums they receive conser-
vatively, they do occasionally run into financial difficulties and sometimes a
company goes bankrupt, is placed in receivership, or becomes insolvent. Most
often, insurance regulators intervene before things get too bad and assist in
transferring the policyholders’ assets to a healthy insurance company that
takes over the responsibilities. (I will explore this aspect in a moment.) If the
intervention is carried out early enough, all can turn out well.
Yet, some spectacular insurance company blowups have occurred in the
past few decades. For example, in August 1994, the fourth-largest insurance
company in Canada (and one of the top 30 in North America) was taken
over by regulators as it teetered on the edge of insolvency. The company was
called Confederation Life, and its name lives on in infamy for anyone in the
Canadian financial services industry. At the time, it had more than C$20 bil-
lion in assets, almost 5,000 employees (who lost their jobs), and more than
750,000 policyholders around the world. The main culprits in Confederation
Life’s failure were bad real estate investments and mortgage loans. Many

24 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 24
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Institutional Details

commentators have blamed incompetent insurance company executives and

regulators. McQueen (1997) provides an accessible and entertaining story of
the demise of Confederation Life.
In another notable and publicized failure, which took place in September
1983, an insurance holding company by the name of Baldwin-United filed
for Chapter 11 bankruptcy protection in the United States after insurance
commissioners in Arkansas and Indiana took over management of its insur-
ance subsidiaries. More than 165,000 policyholders had purchased high-yield
annuities from Baldwin-United, and the money was frozen for more than
three years while regulators and the courts picked up the pieces. The remain-
ing assets were transferred to another insurance company.
What happened to Baldwin-United—originally, a maker of pianos—was
that it extended itself far beyond its core competency (making pianos) by
selling annuities and it promised yields that far exceeded what it was earning
on the assets in its general account. This particular saga took years to resolve.
Another saga that has been ongoing for 20 years (yes, two decades) is
Executive Life Insurance Company of New York. This New York City–
based life insurer was placed in rehabilitation in April 1991 by the New
York Superintendent of Insurance, who filed a petition for the liquidation
of the company on 1 September 2011. The company’s problem? A book of
life annuities it took over from a DB pension plan.
Insurance company meltdowns are not limited to Canada or the United
States either. In fact, one of the oldest insurance companies in the world, the
Equitable Life Assurance Society, which was based in the United Kingdom
and founded in the 18th century, had a large business selling investment
plans with guaranteed annuity rates (GARs) attached. The GAR enabled
savers to convert their accumulated money into a life annuity at a rate that
was guaranteed in advance. So, for example, savers were guaranteed the abil-
ity to get 11% yields on their annuities regardless of actual interest rates
and mortality rates (the ratio of deaths within a subgroup of the population
expressed per 1,000 per year) at the time of annuitization. Because life annu-
ity rates plummeted in the late 1990s, Equitable Life faced serious financial
difficulties and then tried to renege on the GAR promises. The result was
that, in 2000, it had to (effectively) close down for business. To those famil-
iar with the lingo, the company had sold “naked put options” but had never
bothered to hedge or reinsure them.
The insurance industry has (hopefully) learned from these disasters.
Moreover, although these disasters were reported extensively in the media
at the time, I would like to point out that the situation in general is not as bad
or scary as it sounds from such headlines. Nowadays, annuitants and policy-
holders are protected by guarantee funds. Even if your insurance company goes

©2013 The Research Foundation of CFA Institute  25

Milevxky.indb 25
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Life Annuities

bankrupt and general creditors (unsecured bondholders) lose their money, the
chance is good that you and the other annuitants will get most of your prom-
ised income from other insurance companies.
State Guarantee Associations. The National Organization of Life
and Health Insurance Guaranty Associations (NOLHGA) is a voluntary
association made up of the life and health insurance guaranty associations
of all 50 states, the District of Columbia, and Puerto Rico. The organization
was founded in 1983, when the state guaranty associations determined that
a need existed for a mechanism to help coordinate their efforts to protect
policyholders when a life or health insurance company insolvency affected
people in many states. Basically, the mechanism is that when something goes
wrong, the stronger insurance companies in the industry pool their resources
and rescue the weaker company. It is similar to Federal Deposit Insurance
Corporation (FDIC) or Canada Deposit Insurance Corporation insurance
that protects bank depositors. Although there are some important differences
between “state guarantee funds” and the banking industry’s version, they per-
form the same function.
The individual state guarantee funds in the United States have a parallel
organization in Canada called “Assuris.” Exhibit 6 provides a brief overview of
what is covered and how to obtain more information about the limits.

Exhibit 6.  Guarantees and Protections Available to Life

Annuitants
United States: State Guarantee Funds Canada: Assuris
Between $80,000 and $500,000 Policy to be transferred to a solvent
premium coverage, depending on the company with the guarantee that
state in which the annuitant lives and annuitant will retain greater of $2,000
the insurance company is domiciled. per month or 85% of the promised
monthly income benefit.

Notes: These entities are funded by insurance companies collectively, not govern-
ments. More information can be found at www.nolhga.com and www.assuris.ca.

NOLHGA has been active recently in the following instances: when

Golden State Mutual Life Insurance was shut down by regulators in
California in September 2009; when Shenandoah Life entered receivership
in Virginia in February 2009; when Standard Life Insurance Company of
Indiana was taken over by Indiana regulators in December 2008; and when
London Pacific Life & Annuity Company was liquidated in July 2004. None
of these events, managed and coordinated with the help of NOLHGA, were
pleasant for policyholders, but policyholders were eventually compensated for
some of their losses.

26 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 26
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:34 PM
 Institutional Details

As far as the frequency of financial disasters is concerned, although hun-

dreds, if not thousands, of banks have failed (in the United States) over the
last decade, fewer than 10 annuity carriers have been taken over by the state
insurance regulators, and they have been small ones. So, the risk is certainly
present, but it is not as severe as in the banking industry.
In summary, insurance companies can run into financial distress, and
despite the fact that regulators and guarantee funds are available to smooth
things out in times of distress, the reluctance of (potential) policyholders to
invest or deposit more than the protected limits in any one company’s policy
is understandable. Indeed, the concept of diversification applies not only to
stocks and bonds but also to insurance policies, including life annuities. The
next section describes how default risk and the credit rating of an insurance
company can affect how much life annuities pay out.

Q10. Do the Credit Ratings of the Insurance Company

Affect Payouts?
Although policyholders are protected by the state-mandated guarantee funds
in the United States (Assuris in Canada), there is a strict limit on how much
is actually covered. So, if your life annuity premium is above the limit, you
are “at risk” if the company runs into financial distress. Moreover, even if
your premium is under the limit—which means you are 100% covered, in
principle—the possibility always exists of delays in regulatory resolution or
unwanted personal stress from the corporate distress. Therefore, as you might
expect in a well-functioning capital market, riskier companies—in terms of
their overall credit ratings—do actually pay out more on life annuities. The
effect is akin to corporate bonds paying higher yields than government bonds
or banks of different risks paying different rates on savings accounts, even
though they are all covered by the FDIC. Default risk (and stress) matters.
Here is an example. On 20 September 2012, a 65-year-old male was
quoted a life annuity of $527 per month (on a $100,000 premium with a
10-year PC) from the insurance company Genworth Life. The same indi-
vidual was only offered $507 per month, however, from New York Life. In
other words, Genworth was offering an extra $20 per month for the same
$100,000 premium, which is $240 per year, or 4% more income, than New
York Life. This difference is not trivial. So, why was one company paying
more than the other? Are not these life annuity payouts a commodity?
Given how easy it is to search for prices online, would not everyone go to
Genworth for the extra $240 per year? Part of the answer, as demonstrated

©2013 The Research Foundation of CFA Institute  27

Milevxky.indb 27
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Life Annuities

in Exhibit 7, is credit risk.11 Genworth is, in actuality or is at least perceived

to be, a riskier company—especially compared with New York Life—and,
therefore, must compensate the buyer with a better payout.
By riskiness, I do not mean volatility of the company’s stock price (in
the sense of, for example, having a higher beta); indeed, New York Life
stock is not publicly traded. I am referring to the overall creditworthiness
of the company as evaluated by independent credit rating agencies. These
agencies focus on such issues as the following: Will the company be able
to pay its debts on time? Does it have riskier investments backing its lia-
bilities? The bottom line is that New York Life is a safer company than
Genworth. So, it does not have to pay out as much.

Exhibit 7.  Monthly Income vs. Credit Rating, 20 September 2012

Income for Male A.M. Best
Rank Insurance Company at Age 75 Credit Rating
1 Genworth $676.30     A
2 American National  670.37     A
3 MetLife  662.98     A+
4 Nationwide  660.89     A+
5 Lincoln Financial  658.85     A+
6 Guardian  658.65 A++
7 New York Life  658.50 A++
8 Integrity Life Insurance (W&S)  655.64     A+
9 Principal Financial Group  653.54     A+
10 John Hancock  645.90     A+
11 Minnesota Life  645.19     A+
12 Pacific Life  644.45     A+
13 Symetra Life Insurance  640.66     A
14 Hartford Life  631.95     A
15 Lincoln Benefit Life Insurance  631.25     A+
16 Jackson National Life  614.91     A+
17 Protective Life Insurance  613.04     A+
18 American General Life  610.97     A
Average $646.34

Source: Based on data from Cannex Financial Exchange.

In the United States, there are three well-known credit rating agencies.
They are Moody’s Investors Service, Standard & Poor’s (S&P), and A.M.
Best. And although Moody’s and S&P have higher visibility and greater
Other reasons may be at work, of course, in the difference between what Genworth Life and New
11

York Life are offering. Genworth Life may have lower operating expenses or may be more com-
petitive; maybe it wants the business and is trying harder. But credit risk is probably the main factor.

28 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 28
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Institutional Details

name recognition, when it comes to insurance companies, A.M. Best is the

more comprehensive and widely cited. As Exhibit 7 shows, the A.M. Best
rating of Genworth is a single A whereas the rating for New York Life is
A++, two notches better.
Note in Exhibit 7 that, in general, the A++ companies do not offer the
best rates, although the risk-versus-payout relationship is not perfect: The
highest payout for a 75-year-old man was from Genworth’s and the lowest
payout was from American General Life, although both companies—with
a difference of more than $60 per month and $720 per year—had the same
A.M. Best rating.
Nevertheless, the A++ company is highly unlikely to ever be paying the
most, which is almost axiomatic if you think about it. In fact, in a formal
statistical regression of the highest monthly life annuity payout (the depen-
dent variable) on credit ratings (the independent variable), the relationship
is statistically significant and the slope is negative (with a p-value of 7%).
Higher-rated companies pay less. Figure 2 displays this relationship in a
graphical format.

Figure 2.  Relationship between Credit Rating and Annuity Payout

Monthly Payout at Age 75 ($)

680

675

670

665

660

655

650

645

640
A A+ A++
Credit Rating (A.M. Best)

Best Payout in Class Regression Line

©2013 The Research Foundation of CFA Institute  29

Milevxky.indb 29
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Life Annuities

Digging a bit deeper into the financial economics of the matter pro-
vides a good reason why higher-rated companies must pay out less on life
annuities than lower-rated companies do. Consider why a rating agency
would rate Genworth lower than New York Life. Recall that insurance
companies back, or hedge, their life annuities (and life insurance) with sim-
ilar investments in their general accounts. Some companies have general
account investments that are safer than others—mortgages and corporate
bonds—and that greater safety leads to a higher credit rating. Lower-risk
bonds tend to offer lower yields. Therefore, the companies with higher
credit ratings, and lower-yielding assets, cannot afford to pay as much on
their life annuities.
In the spirit of chicken-and-egg theorizing of which came first, you can
debate whether the higher-rated insurance companies decided to purchase
the lower-yield assets or whether it was the other way around—they get a
higher rating because of their lower-yielding (and safer) portfolios—but the
end result is the same.
When you purchase a life annuity, you have a choice of more than 20
insurance companies offering what is essentially a commodity product. The
only economic difference between a lifetime of cash from Company A and
from Company B is their chances of experiencing financial difficulties during
your life. This risk is reflected in the life annuity payout rate.
Remember, however, that if you keep your purchase under the state guar-
antee fund limits—$100,000 in most states—your income is protected even
if the company defaults on its (other) obligations. So, in some sense, credit
rating should not matter to you. You might as well go with the highest cash
flow. Of course, those planning to annuitize larger sums than a guarantee fund
limit might want to put the whole amount in the safer company to avoid the
difficulty of managing a large number of vendors simply to keep individual
purchases under the state limits.12

Q11. A First Look at Methuselah Risk: What If

Annuitants Lived for 969 Years?
The Old Testament makes reference to Methuselah, the grandfather of Noah,
who lived to the ripe old age of 969 years and was the oldest person mentioned
in what is known as the Hebrew Bible. The modern-day recordholder for lon-
gevity as of September 2012 was Jeanne Louise Calment, who was born on 21

I personally would probably forgo the extra $20 per month and stick to the A++ company
12

with my nest egg. I would hate to see my annuity provider in the financial headlines for the
wrong reasons, even if I am 100% covered. As any behavioral economist will remind you, peace
of mind is hard to quantify.

30 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 30
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Institutional Details

February 1875 and died on 4 August 1997. She was French—which may or
may not explain her extreme longevity—and lived a total of 122 years and 164
days. A far cry from 969, but impressive nevertheless.13
We do not know whether she ever purchased a life annuity from an insur-
ance company—although she did benefit financially from her longevity—but
the thought of having a Methuselah or even a Jeanne Calment among their
annuitants has struck fear into the hearts of insurance company executives
from time immemorial. What if people live longer than anticipated by the
actuaries? What happens if scientists find a cure for cancer or diabetes? Would
insurance companies be able to afford to pay annuitants for that much longer?
Would it place the company at risk?14
Some research a few years ago conducted by insurance analysts at
Moody’s, in which I participated, sheds some light on the matter.15 Table
3 displays the main results from their analysis and report. It provides some
indication of what might happen to company profitability (and credit risk)
if certain diseases were cured, mortality were reduced, and annuitants lived
longer than anticipated. Here is an example of how to interpret Table 3.

Table 3.  How Reductions in Mortality Affect Annuity Profitability

Unisex 55 Unisex 62 Unisex 70
Life Spread Life Spread Life Spread
Mortality Reduction Exp.a (bps) Exp.a (bps) Exp.a (bps)
Status quo    0% 82.9 +100 83.8 +100 85.6 +100
Stroke and
pneumonia –10 83.8 +85 84.7    +77 86.4 +60
Cancer and
diabetes –40 87.4 +39 88.1   +4 89.4 –67
Heart disease –80 97.7 –36 97.9  –111 98.6 –257
a
Expectancy.
Notes: The table displays the ex post spread calculated from a Moody’s model that would be earned
from an immediate annuity block of business assuming an ex ante desired spread of 100 bps.
Source: Based on data from Robinson and Fliegelman (2002).

13
Jeanne Calment and longevity risk will be forever linked by the fact that she engaged in the
peculiar French practice of selling her apartment with the proviso that the buyer could occupy
it upon the seller’s death. She sold it in 1965 at the age of 90 to a 47-year-old man who, despite
living 30 more years, did not outlive her.
14
Some insurance companies (such as TIAA-CREF in the United States) offer participating
annuities, in which the company is entitled to reduce payment to all retirees in the event of a
greater-than-expected increase in longevity. This tontine-like provision is rare; most companies
are on the hook for the payments.
15
Recall that Moody’s is one of the main credit rating agencies in the United States, so naturally,
this sort of question preoccupies them.

©2013 The Research Foundation of CFA Institute  31

Milevxky.indb 31
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Life Annuities

Consider an insurance company that sells a life annuity to a 55-year-old and

prices the annuity so that it makes a profit or spread of 100 bps (i.e., 1%;
a basis point is 0.01%). What this 100 bp spread means is that when the
insurance company is pricing the annuity, if the assets in the general account
earn 5%, for example, it expects to pay out 4% to the annuitant and keep the
spread. Basically, they keep 1 percentage point of the investment return on
assets as a profit margin.
Now, assume that scientists were able to completely eliminate all deaths
from strokes and pneumonia, which, according to biostatisticians, would
reduce mortality rates across the board by 10% at all ages. In this case, accord-
ing to the analysts’ model reported in Table 3, life expectancy at age 55 would
increase by about 1 year, from 82.9 to 83.8. Obviously, the insurance com-
pany would end up paying more in aggregate to its annuitants, but the realized
profit would be reduced only to 0.85%. In other words, the companies would
still earn a profit but not as much as they had previously. In fact, continuing
with the same logic, even if cancer and diabetes were completely eliminated,
the realized spread would drop to 0.39% but still be positive. What this means
is that, even if life expectancy at age 55 jumped from 82.9 to 87.4, the insur-
ance company could still earn an ex post profit.
This sort of theoretical exercise rests on many assumptions that are too
numerous to mention (I know because I helped create the model Moody’s
used), but there are some important takeaways from this simple thought
experiment. First, notice the columns labeled “Unisex 55,” “Unisex 62,”
and “Unisex 70” in Table 3. When the company sells a life annuity to a
55-year-old, it exposes itself to less risk than when it sells the annuity
to a 62-year-old or 70-year-old. But if mortality is reduced suddenly and
unexpectedly, this change will have a greater impact on companies with
older annuitants. Consider: If the company was expecting to make pay-
ments for 20 years, on average, and it had to make them for 22 years, the
relative increment of 10% is not great. But if the company was expecting
to make payments for eight years, on average, that extra two years of lon-
gevity translates into 25% more payments. Remember also that the com-
pany is paying the 70-year-old much more relative to the 55-year-old, so
this shock (extra years of payments) can definitely hurt profitability. The
main result shown in Table 3, however, is that the impact of the changes
would not be as dramatic as you might expect, at least for annuities sold to
young individuals.
Another important point worth noting relates to the fact that life insur-
ance and life annuities create opposing liabilities. When it comes to the
risk of selling annuities to Mr. Methuselah, if you also sold him a sizable
life insurance policy, your overall risk exposure would be neutralized. Yes,

32 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 32
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Institutional Details

the company would have to make life annuity payments for 969 years, but
then again, it would not have to pay out death benefits for 969 years. What
the company lost on the life annuity side, it might gain on the life insur-
ance side. Moreover, although the insurance company may not have sold life
insurance and life annuities to the same Methuselah, as long as the shock
to mortality affects an equal number of life insured and life annuitants, the
effect would be the same: One change offsets the other. This effect assumes,
of course, that the insurance company has an equal and equivalent dollar
exposure to life insurance and life annuities across the same age groups, but
the main idea holds true even if it does not. This effect, by the way, is called
a “natural hedge” and effectively implies that Methuselah risk is not as scary
as it sounds if the insurance company has a balanced portfolio of “long” and
“short” longevity risks.

Q12. Are Life Annuities Popular, and What Is the Size of

the U.S. Market?
Life annuities can be viewed as either extremely popular or highly unpopular
depending on your perspective and definition of “annuity.” For example, if you
consider DB pensions—government pensions, social security programs, and
the like—to be life annuities, then they are one of the most successful prod-
ucts (and programs) in financial history. According to the U.S. Social Security
Administration, almost 39 million retired Americans and their depen-
dents collected more than $45 billion dollars in benefits during one month
(December 2011) alone. The $1,234 average monthly benefit received adds up
to more than a half a trillion dollars in annual life annuity income.
The Canadian Pension Plan—which is equivalent to Social Security but
much less generous than the American version—pays $28 billion annually
to 5 million retired Canadians and their dependents, an average of $530 per
month.
In any given year, thousands of retirees who are part of a corporate or pub-
lic DB pension plan exit the labor force and actively select the annuity option
instead of the lump sum. Indeed, you cannot get better examples of the masses
enjoying their life annuities.
In contrast to the publicly mandated and employment pension programs,
however, the voluntary life annuity market in North America is small. In fact,
even in the universe of all annuities sold, life annuities are but a small fraction.
Table 4 provides an indication of the size of this market in relative and abso-
lute terms. It displays the volume of sales (i.e., premiums contributed by indi-
viduals) during a 12-month period and breaks down the numbers across three
broad annuity categories. The “fixed immediate annuities” category includes
our coveted life annuities.

©2013 The Research Foundation of CFA Institute  33

Milevxky.indb 33
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Life Annuities

Table 4.  Total Sales of U.S. Individual Annuities, Year Ending 30 June 2012
Sales Percent of
Product Type (billions) Sales
Variable annuity $150.7 68.2%
Fixed deferred annuity
  (including book, market, and indexed) 61.5 27.8
Fixed immediate annuity
  (including life annuities) 8.9 4.0
Total for 12 months $221.1 100.0%
Source: Based on data from the Insured Retirement Institute (September 2012). Data are from 55
insurance companies reporting.

At this point, I do not want to get lost in the details of the other annui-
ties listed in Table 4, which deserve their own sections; what follows is a brief
description of the broad categories.
A variable annuity (VA), which is the largest category of annuity when
ranked by sales in the United States, is essentially a mutual fund (or a col-
lection of funds) with some added insurance and financial guarantees. Some
industry participants will disagree with this simple description of a VA, but
you will have to trust me here, especially if this is your introduction to these
instruments. VAs are not really life annuities in any way that an insurance
or financial economist would accept. And although the insurance industry’s
own regulators and lawyers might consider them annuities, they are best
described as savings and accumulation products with, at most, an option to
convert the product into an income stream eventually. So, if and when a VA
is actually converted into income, it would be properly described as a life
annuity. Until that time, it is a savings and accumulation vehicle. (I will dis-
cuss more about VAs later.)
Moreover, few VAs end up being converted (annuitized) into lifetime
income streams. In most cases, they are cashed out, surrendered, or exchanged
for other annuities. Statistics from LIMRA (2010) indicate that only 1–3%
of variable annuities are ever annuitized.16 So, they are called “annuities,” but
they are very different from the life annuity with a 2,500-year history. Do not
confuse them.
As Table 4 shows, 68.2% of total “annuity” sales during the year were VAs.
It is the largest segment of the annuity market (and perhaps deserving of its
own book).
Fixed deferred annuities (FDAs) were 27.8% of sales. These products also
are primarily savings instruments in which individuals deposit premiums and
collect some form of interest gains, but unlike VAs, the funds in FDAs (1) are

LIMRA is the Life Insurance Management Research Association (located in Windsor,

16

Connecticut).

34 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 34
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Institutional Details

placed in the insurance company’s general account and (2) do not fluctuate on
the basis of the stock market. Think of this category as the insurance company
equivalent of a safe bank deposit but perhaps one in which the interest rate is
slightly better than that offered by the bank. The FDA category also contains
the option to convert into income. But as with the VA, few FDAs are annui-
tized, or converted into an income stream. They are purchased primarily as
accumulation or savings vehicles and then usually cashed out in full. FDAs are
often viewed as a type of tax-deferred certificate of deposit, which are sold by
banks and credit unions.
Finally, the category that is most relevant to this book is the fixed immedi-
ate annuity. As its name suggests, it is an annuity guaranteeing, or promising,
an actual income, usually starting immediately. During the 12-month period
ending 30 June 2012, a total of only $8.9 billion of these fixed immediate
annuities were sold in the United States. This is a mere 4% of the $221 bil-
lion total annuity sales and—at best—13% of fixed annuity (FDA plus fixed
immediate annuity) sales.
Moreover, what fraction of the $8.8 billion flowed into (true) life annuities,
in which payments are guaranteed for the remainder of an individual’s or cou-
ple’s life, is not easy to determine because the data are not available. Recall that
some annuities are purchased for a term certain that is not necessarily a lifetime.
Nevertheless, anecdotal evidence and discussions with industry experts indicate
that at least three-quarters of these sales are true lifetime contracts, perhaps with
PC or refunds attached. More granularity is difficult to obtain. (I hope that the
industry—or at least some of the larger companies—will release a more refined
breakdown of the broad category of fixed immediate annuities.)
In summary, annuities, broadly construed, are a multi-billion-dollar, perhaps
trillion-dollar, business. Social Security and pension programs are essentially life
annuities. The amount of money flowing into life annuities issued by insurance
companies, however, is small. It is small relative to the size of Social Security and
DB pensions, and it is small relative to the size of the overall insurance company
annuity market. It is only 4% of insurance company annuity sales.
So, from a statistical point of view, if you happen upon someone who
just purchased an “annuity policy” from an insurance company, there is a
96% chance it is not the type of annuity I have been discussing and advocat-
ing in this book.

Q13. Is a Variable Annuity with a Guaranteed Lifetime

Withdrawal Benefit a Substitute for a Life Annuity?
As I mentioned in the previous section, tax-deferred VAs, which are the bulk
of the annuity market in the United States, were initially promoted for the
favorable tax treatment and death guarantees they enjoyed. As these products

©2013 The Research Foundation of CFA Institute  35

Milevxky.indb 35
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:35 PM
 Life Annuities

have moved to include guarantees of minimum income streams that investors

could receive, however, those features have become critical selling points. In
this section, I will address the most popular type of income guarantee—the
guaranteed lifetime withdrawal benefit (GLWB). A VA product that contains
this guarantee often competes with a life annuity.
A GLWB rider on a VA allows the investor to lock in a minimum income
for life—similar to a life annuity or deferred income annuity—without sur-
rendering the capital irreversibly. Thus, these riders provide some of the retire-
ment longevity protection of a traditional annuity without surrendering upside
potential or liquidity. The best way to think of them is as a relatively more
expensive mutual fund with a complex path-dependent put option that allows
for a minimal withdrawal level. The guaranteed withdrawal level is less than
what a life annuity would offer, although the difference may not seem huge on
first examination. A GLWB might permit withdrawals of 5% for life, whereas
a life annuity issued to the same buyer at the same time might pay 7% or 8%
for life. Exhibit 8 provides an outline of the differences between a traditional
life annuity and a VA with a GLWB.

Exhibit 8.  Comparison of a Life Annuity with a VA + GLWB

Element Traditional Life Annuity VA + GLWB
Liquidity Little liquidity, especially in the event Possible surrender charges, but the
of no PC guarantee. account can always be liquidated.

Payout rate Function of age and interest rates. Lower than a life annuity by 1.5–2.0
percentage points.

Costs and fees Embedded commissions and fees are Various layers of fees within the
on the order of 1–2% of the premium. VA are difficult to disentangle, but
generally 1–3% of assets annually.

Other Primary focus of this research A shrinking number of insurance

monograph. companies are offering this product.

Here is a synopsis of the mechanics of the VA plus GLWB: The individ-

ual policyholder deposits, or rolls over from another VA, a sum of money into
an investment portfolio that is then allocated into a number of subaccounts
that contain stocks, bonds, and other generic investments. The portfolio then
grows (or shrinks) over time, depending on the performance of the underlying
investments. Any capital gains are tax deferred and eventually treated as ordi-
nary income. (In Canada, there is no tax deferral of gains.) Up to this point, it
might sound like a mutual fund.

36 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 36
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:36 PM
 Institutional Details

Then, at some future date—which is usually under the control of the

policyholder—the annuitant can start taking guaranteed withdrawals from
the account. Think of this income as a systematic withdrawal plan (SWiP)
at a nominal (i.e., not inflation-adjusted), nondecreasing level.17 The income
is guaranteed to never decline for the remaining life of the annuitant and, in
the case of a joint product, of the annuitant’s (younger) spouse. Thus, in con-
trast to a SWiP, if the annuity’s underlying investment portfolio (that is, the
account value) ever reaches zero, the guaranteed income will continue so long
as one member of the couple is still alive.
The guaranteed withdrawal rate is determined by the insurance company
issuing the GLWB at the time of sale. The guarantee amount is the product
of multiplying a guaranteed rate by the guaranteed base and is determined at
the point of first withdrawal. Moreover, if the investment portfolio happens
to grow even while undergoing these withdrawals, the guaranteed base might
reset to a higher level and hence generate even greater withdrawals. As far as
estate values are concerned, upon the second death, whatever is left over in the
account goes to the heirs, with the requisite tax implications depending on the
cost basis (and depending on whether the GLWB was inside a tax shelter).
GLWBs as thus described exist in a variety of alternative formats and are
often bundled with an array of other guarantees, ratchets, or step-ups linked
to death benefits and life insurance (all of which are beyond the scope of this
book). Regardless of the specifics, however, the basic GLWB ensures that
some withdrawals will continue for life regardless of whether the underlying
account has the funds to support them. In other words, fees and periodic with-
drawals are deducted from the VA account as long as there are funds available
there. But if those periodic withdrawals ever fully deplete this account, the
underwriter steps in and funds the remaining withdrawals for the lifetime of
the investor.
The periodic withdrawals provide downside protection, but some upside
potential remains for the underlying account to grow if markets perform
well. The investor preserves liquidity, because the underlying account value

A SWiP is a (dumb) mechanical liquidation rule that extracts a fixed amount of cash from
17

a retirement portfolio by selling assets to create a desired level of income, regardless of the
price level of markets. So, for example, if a retiree implements a SWiP for $50,000 per year
and, in one particular year, the dividends and interest from the portfolio are (only) $20,000,
then $30,000 worth of securities are sold to make up the income difference. Under a SWiP,
the systematic sale of $30,000 worth of securities ignores fundamental valuation levels and any
other market-timing rules. It is the mirror image of dollar-cost averaging, under which a fixed
amount of money is invested in securities on a regular basis independent of valuation levels.
Although many individuals view SWiPs as an alternative to life annuities, a SWiP can fully
deplete the portfolio whereas a life annuity cannot. The GLWB offers a SWiP with some insur-
ance protection—namely, that if the account value ever hits zero as a result of the depletions,
the insurance company will continue paying the annuity.

©2013 The Research Foundation of CFA Institute  37

Milevxky.indb 37
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:36 PM
 Life Annuities

may be withdrawn at any time (minus any surrender charges). Unlike a tra-
ditional life annuity, if the investor dies, his or her heirs inherit the remain-
ing account value.
The insurance companies manufacturing the relatively new generation of
VAs with a GLWB view the product as a private-sector replacement for DB
pensions in an increasingly DC world. Whether the GLWB is better than the
life annuity from the consumer’s perspective depends on a complex relation-
ship between the pricing of the guarantee, the retiree’s optimal consumption
strategy, and the existence of bequest motives.
In summary, although a smaller group of insurance companies are offering
them, the latest generation of VA contracts has been financially engineered to
provide an assortment of lifetime income guarantees that are meant to protect
the policyholder against what the industry has termed “sequence of returns
risk” and “longevity risk.” These terms refer to the chance that a retirement
portfolio from which cash is being withdrawn will suffer early losses and/or
the retiree will live longer than average. The common denominator of all these
insurance riders is that they contain an implicit put option on financial mar-
kets plus some form of longevity insurance, akin to a pure life annuity. Of
course, using the concept of put–call parity, they can also be viewed as call
options to annuitize at some variable strike price. The (anecdotal) “sales pitch”
for these products revolves around the idea that the guarantees permit inves-
tors to take on more investment risk than they would without the guarantees.18
Here is the bottom line: To the naked eye, the VAs with GLWB might
appear to have all the benefits of a life annuity—guaranteed income, risk
pooling—but without the costs associated with illiquidity and irreversibility.
However, although the GLWB product has merits, especially considering
the research evidence that it was initially underpriced, it is not a substitute
for pure life annuities because of its lower yields. For example, whereas a life
annuity might pay 6% to a 65-year-old, the GLWB rate under the same mar-
ket conditions would be in the vicinity of 4%.

A paper by Milevsky and Kyrychenko (2008) seems to indicate that, indeed, investors do take
18

on more risk when these riders are provided.

38 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 38
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:36 PM
 2. Ten Formulas to Know

In this chapter, I address the 10 main formulas that researchers use and that
practitioners should know in the life annuity literature.

Q14. What Is a Biological Mortality Rate, and How Is It

Measured?
When an insurance company issues or sells you a life annuity, the company
must try to predict when you will die, and payments will cease, so that it can
determine the appropriate monthly payments. Naturally, the longer an annui-
tant is forecast to live—and the lower the annuitant’s mortality rate—the
lower the payments must be. For example (with interest and the time value of
money ignored), if the annuitant is forecast to die in exactly 10 years, then the
$100,000 annuity premium must be returned to the annuitant in (only) 120
installments, which is $833 (100,000/120) per month (with interest ignored),
but if the annuitant is forecast to die in exactly 15 years, then the monthly
payment must be lower, $556 (100,000/180). So, pricing annuities is mostly
about predicting how long annuitants will live and payments will be made.
How does the insurance company make these predictions? What hap-
pens if it gets it wrong and you do not die exactly when you were supposed
to?
The answer to the first question is, of course, that because the insurance
companies are selling many life annuities to many different people, they do
not have to predict exactly how long you will live but, rather, how long an
individual member of a group will live on average. And forecasting the life
expectancy of a group is much easier than forecasting the exact length of life
for any one individual.
For example, suppose Client 5 lives beyond the group’s life expectancy,
but Client 17 does not make it to the group’s life expectancy. The more-than-
average number of payments made to Client 5 will be offset by the fewer-
than-average number of payments made to Client 17. (The client numbers are
completely arbitrary, and perhaps two shorter-lived annuitants—Clients 17
and 8—will offset one very long-lived annuitant—Client 5—but the idea is
the same.)
What makes this principle of offsetting risks work in an accurate manner
is the so-called law of large numbers. If the insurance company pools a large
enough number of annuitants with similar forecasted life expectancies, the
risk-offsetting process can take place with much more accuracy than would be
possible for a pool of just two or three people. Obviously, selling one annuity
to each of two 65-year-olds in the hope that they will cancel each other is a

©2013 The Research Foundation of CFA Institute  39

Milevxky.indb 39
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:36 PM
 Life Annuities

silly gamble. But if the company sells thousands of life annuities to a group
in their 60s, then the law of large numbers guarantees that, on average, even
though not in any individual case (except by coincidence), they really will live
to their life expectancy. So, all the insurance actuary has to worry about is the
behavior of the average in the pool.19 For this task, historical mortality pat-
terns of similar groups are helpful.
In fact, to be even more precise, insurance actuaries do not necessarily
focus on the life expectancy of the group but on year-by-year mortality rates.
In this manner, they can mix and match people of different ages in their
large annuity pool with the same risk-offsetting result applying on a year-
by-year basis.
Equation 1 displays the definition of a one-year mortality rate:
No. dying between age x and age ( x + 1)
qx = . (1)
No. alive at age x
Now, my objective is not to convert the reader into an insurance
actuary or to delve too deeply into the actuarial minutiae, but for clar-
ity, Table 5 displays historical mortality rates for a homogenous group of
annuitants who purchased life annuities from insurance companies in the
past few decades. Note that the table is based on a group’s (past) realized
experience, but the data are then used, with slight modifications, to forecast
current experience.20

Table 5.  Annuitant Mortality Table for Various Ages

Death between Ages Female Rate (qx ) Male Rate (qx )
55 and 56 0.00246 0.00453
60 and 61 0.00386 0.00643
65 and 66 0.00625 0.00994
70 and 71 0.01003 0.01698
75 and 76 0.01756 0.02830
80 and 81 0.03193 0.04604
85 and 86 0.05791 0.07328
90 and 91 0.10176 0.11221
Source: Based on data from the Society of Actuaries, an organization in the United
States that, among other responsibilities, tabulates and reports various types of mor-
tality tables; “Annuity 2000 Mortality Table” (www.SoA.org).

19
For a much more precise and proper mathematical definition of how the risk is reduced with a
large number of annuitants, please see the actuarial references mentioned in Chapter 3.
20
In simple terms, this table can be described as a static mortality table, one without any mor-
tality improvement projections, as opposed to a dynamic mortality table, in which a particular
cohort is modeled over time.

40 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 40
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:36 PM
 Ten Formulas to Know

Here is how to interpret the numbers in Table 5. Based on historical pat-

terns for a large group of individuals who purchased annuities, 1.003% of
70-year-old women died before reaching their 71st birthday. This fact can be
used to forecast that the probability a 70-year-old female will die before her
71st birthday is 0.01003 or 1.003%.
Here is another way of viewing Table 5. Given a large group of 70-year-old
females, the expectation is that 1% of them will die prior to their 71st birthday
and that the other 99% will survive to their 71st birthday. So, if the insurance
company has to pay each 71-year-old woman $1,000 at the end of the year, it
needs to charge only $99,000 to a group of 100 70-year-old women. This $990
is charged to any individual member of the group at the beginning of the year
(again, interest is ignored). The company does not know who the 1% will be,
but it is irrelevant. All the company needs is the average, so it can figure out
how much to charge the group as a whole.
If you look carefully at the mortality rates in Table 5, you will notice
a steadily increasing pattern, at the rate of between 9% and 11% per year
for both males and females. In other words, the mortality rate during age
(x + 1) is approximately 9–11% higher than the mortality rate during age
x. Were it not for this “law” of mortality—and if mortality rates were com-
pletely random from one age to the next—it would be extremely difficult
for insurance actuaries to forecast the life expectancy of a particular group
of annuitants.
In summary, the mortality rate is the probability that an individual from
a fairly homogenous group of insured lives will die during a given year. By
placing a large number of similar individuals together in a group, an insur-
ance company can accurately forecast how long it will be making payments to
the entire group, even if the behavior of individuals in the group is difficult to
predict. The forecast is based on a particular mortality table. The key is to pick
the right table.

Q15. How Are Mortality Rates Converted into Survival

Probabilities?
The mortality rate gives year-by-year estimates of the age-dependent prob-
ability of death for a given group, but for the purposes of pricing and valu-
ing life annuities, the mortality rate must be converted into a survival rate.
Technically, survival rates are a set of (declining) numbers that describe the
probability of living 1, 10, 20, and even 50 years into the future. And although
a 1% mortality rate for someone who is 70 years old obviously implies a
99% probability of surviving to the age of 71, extending those numbers to
more advanced ages is messy. Equation 2 describes how to convert mortality
rates into long-term survival rates under the assumption that the age-based

©2013 The Research Foundation of CFA Institute  41

Milevxky.indb 41
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:36 PM
 Life Annuities

mortality rates, qx, remain the same over time. The left-hand side is obtained
by multiplying together the quantities given by 1 less the mortality rate from
the current age until the age to which survival is being projected:
j =i −1
( )
p ( x, i ) = ∏ 1 − q x + j .
j =0
(2)

The expression p(x, i) represents the probability that an x-year-old, who

is alive, will survive to his (x + i) birthday. So, for example, p(75, 10) denotes
the probability a 75-year-old will survive 10 more years to his 85th birthday,
and p(80, 5) denotes the probability an 80-year-old will survive 5 more years
to her 85th birthday. Naturally, for the same set of underlying mortality rates,
the rate for survival from 80 to 85 is greater than the rate for survival from 75
to 85 because in the former, the person is already alive at 80. (The question is,
Who is more likely to reach his or her 85th birthday?)
The other expression in Equation 2 should be familiar from Equation 1;
qx+j denotes the probability that a (x + j)-year-old will die during the next
year, before his or her next birthday. For example, q65 is the probability that a
65-year-old will die before the 66th birthday.
Thus, whereas Equation 1 defined the mortality rate, Equation 2 shows the
probability of staying alive over a given period of time.
Here is a detailed example of how to use Equation 2: Assume that, based
on a given mortality table, the probability a 70-year-old will die before his
71st birthday is (q70 = 1.35065%); the probability a 71-year-old will die before
her 72nd birthday is (q71 = 1.50065%); and the probability a 72-year-old will
die before his 73rd birthday is (q70 = 1.66765%). (Never mind for now where
exactly these numbers come from or how they were estimated.) According to
Equation 2, the probability that a 70-year-old will survive for 3 more years
(i.e., not die before his 71st, 72nd, or 73rd birthday) is the product of the
quantities given by 1 minus these three individual rates. Using our notation
the answer is p(70, 3) = 93.777%.
Table 6 provides an entire vector of survival probabilities—for a given
mortality table—starting from age 70 all the way to the last age at which it is
assumed members of the group might still be alive. The mortality rates used in
one column of Table 6 are a 50/50 blend of “individual annuitant” mortality
rates for males and females observed in 2000; the table is thus often called a
“unisex table.” The survival probabilities would be slightly higher for females
and slightly lower for males.
Notice how the survival probability (in the last column) declines from a value
of 98.649%—which is the probability of surviving for 1 year—down to a value of
zero by the age of 115. Think of a continuous curve that begins at a value of 1.0
(probability of surviving for one small instant) and then gradually declines toward
zero (no one gets out of here alive). This decline is more rapid than exponential

42 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 42
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:37 PM
 Ten Formulas to Know

Table 6.  From Monthly Mortality Rates to Survival Probabilities as

of 2000
Mortality Rate
Survival from
50/50 Blend Age 70 to End
Age Female Male Unisex of Year
70 1.0034% 1.6979 1.351% 98.649%
71 1.1117 1.8891 1.500 97.169
72 1.2386 2.0967 1.668 95.549
73 1.3871 2.3209 1.854 93.777
74 1.5592 2.5644 2.062 91.844
75 1.7564 2.8304 2.293 89.737
76 1.9805 3.1220 2.551 87.448
77 2.2328 3.4425 2.838 84.967
78 2.5158 3.7948 3.155 82.286
79 2.8341 4.1812 3.508 79.399
80 3.1933 4.6037 3.899 76.304
81 3.5985 5.0643 4.331 72.999
82 4.0552 5.5651 4.810 69.488
83 4.5690 6.1080 5.339 65.778
84 5.1456 6.6948 5.920 61.884
85 5.7913 7.3275 6.559 57.825
86 6.5119 8.0076 7.260 53.627
87 7.3136 8.7370 8.025 49.323
88 8.1991 9.5169 8.858 44.954
89 9.1577 10.3455 9.752 40.570
90 10.1758 11.2208 10.698 36.230
91 11.2395 12.1402 11.690 31.995
92 12.3349 13.1017 12.718 27.925
93 13.4486 14.1030 13.776 24.079
94 14.5689 15.1422 14.856 20.502
95 15.6846 16.2179 15.951 17.231
96 16.7841 17.3279 17.056 14.292
97 17.8563 18.4706 18.163 11.696
98 18.9604 19.6946 19.328 9.436
99 20.1557 21.0484 20.602 7.492
100 21.5013 22.5806 22.041 5.841
101 23.0565 24.3398 23.698 4.456
102 24.8805 26.3745 25.628 3.314
103 27.0326 28.7334 27.883 2.390
104 29.5719 31.4649 30.518 1.661
105 32.5576 34.6177 33.588 1.103
106 36.0491 38.2403 37.145 0.693
107 40.1064 42.3813 41.243 0.407
(continued)

©2013 The Research Foundation of CFA Institute  43

Milevxky.indb 43
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:37 PM
 Life Annuities

Table 6.  From Monthly Mortality Rates to Survival Probabilities as

of 2000 (continued)
Mortality Rate
Survival from
50/50 Blend Age 70 to End
Age Female Male Unisex of Year
108  44.7860% 47.0893% 45.938% 0.220%
109  50.1498 52.4128 51.281 0.107
110  56.2563 58.4004 57.328 0.046
111  63.1645 65.1007 64.133 0.016
112  70.9338 72.5622 71.748 0.005
113  79.6233 80.8336 80.228 0.001
114  89.2923 89.9633 89.628 0.000
115 100.0000 100.0000 100.000 0.000
Source: Based on data from Society of Actuaries.

but not as fast as linear. The probability of a 70-year-old surviving to age 80 is

about 76%; the probability of that person surviving to 85 is approximately 58%;
finally, the probability of the person surviving to 100 is a bit less than 6%. Stated
differently, the mortality table indicates that 76% of a large group of 70-year-olds
(in the referenced group) will survive to age 80, that 58% of them will survive to
age 85, and that 6% of them will survive to age 100. Of course, these rates are all
projections for future mortality based on past trends, recent improvements, and so
on. As with trying to predict the odds of the stock market going up or down in
any given year or decade, all we have is historical data.
The underlying mortality table reflects the experience of the insurance
industry with annuitants—that is, retirees who actually purchase annuities.
These people tend to be healthier than population averages, experience lower
mortality rates, and, therefore, live longer. U.S. Social Security mortality tables
reflect a wider (and less healthy) group of individuals. Of course, regardless of
what mortality rates are used in the right-hand side of Equation 2, the cor-
responding survival probabilities, given those rates, are obtained by using that
equation.
The probability of surviving any given number of years depends critically
on the age on which the individual life is being conditioned. The probability of
surviving to age 90, for example, depends on the current age (as well as gender,
health, etc.). Someone who is 89 years old has a greater chance of making it
to age 90 than someone who is only 85 years old. This mathematical fact is
embedded in the logic of Equation 2. The probability of surviving to age 90
if you are 89 is (1 – q89), but the probability of surviving to age 90 if you are

44 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 44
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:37 PM
 Ten Formulas to Know

85 is (1 – q85)(1 – q86)(1 – q87)(1 – q88)(1 – q89), a smaller number. It is smaller

because in this case, we are multiplying the quantity (1 – q89), which is present
in both cases, by other numbers that are all smaller than 1.0.
To conclude, the “fundamental particle” used in pricing and valuing any
life annuity is the underlying mortality rate, presented in a mortality rate table,
which is sometimes referred to as the “mortality basis.” There are many dif-
ferent types of mortality tables, and like snowflakes, no two are exactly alike.
There are mortality tables for healthy and wealthy females and mortality tables
for unhealthy, unwealthy males. There are mortality tables for entire popu-
lations, and there are mortality tables for small groups of retired annuitants.
Whatever population the table represents, the rates in the table can be used to
create a unique set of survival probabilities based on Equation 2. Moreover, it
is easy to work backward and recover or extract mortality rates from survival
probabilities. In fact, as an exercise, you might try to recover the second-to-last
column in Table 6 from the survival probabilities.
Finally, anytime someone mentions or displays a survival probability
curve—whether or not it is within the context of life annuity pricing—you
should ask yourself, What was the underlying mortality table on which these
survival rates were based?

Q16. What Is the Benjamin Gompertz Law of Mortality?

The age-dependent mortality rates displayed in the first two columns of Table 6
might seem arbitrary at first, but they have a clear underlying pattern. The mor-
tality rates not only increase with age, but they also actually increase by almost
the same percentage amount every year. In other words, if the mortality rate was
q% at age y, then it was q(1 + z%) in year (y + 1), then q(1 + z%)2 in year (y +
2), q(1 + z%)3 in year (y + 3), and so on. Human mortality rates—regardless of
the particular mortality table you select—are for the most part an exponentially
increasing function of age. So, if you compute the logarithms of the annual mor-
tality rates, they can be approximated nicely by a straight line and determined by
(1) a slope parameter and (2) an intercept parameter. Think of it as a law of biol-
ogy. At the beginning of the 21st century, z is approximately 9%, so in any given
population, approximately 9% more individuals of a certain age will die this year
compared with last year. If 100,000 Americans who are 65 years old died in
2012; therefore, the estimate is that 109,000 Americans who are 66 years old
will die in 2013. Notice the age dependency and the link to the previous year.
Of course, I did not stumble on this law, nor is it a fluke of the data. This
biological observation was first made by the British demographer and actuary
Benjamin Gompertz (1779–1865), and it is today known as the “Gompertz
law of mortality.” His groundbreaking research on human mortality modeling

©2013 The Research Foundation of CFA Institute  45

Milevxky.indb 45
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:37 PM
 Life Annuities

was first published in 1825 in Philosophical Transactions of the Royal Society;

it is one of the foundational papers in the field of actuarial science and
demographics.
The reason this observation is more than actuarial trivia is that it gives
us a powerful analytical tool to compute survival probabilities to any age
as a function of only two basic parameters—the slope and intercept in the
previously mentioned straight line. This tool both simplifies the computa-
tional requirements and provides intuition for the annuity pricing formu-
las (yet to come).21
The exact derivation is beyond the scope of this book, but the concise
formula for the survival probability—from any age to any age—under the
Gompertz law of mortality can be written as follows:
 x−m 
(3)
( )
t /b  
ln  p ( x, t )  = 1 − e e b  ,

where t denotes the survival period, x denotes the current age of the indi-
vidual, and the parameters (m, b) denote, respectively, the modal age at death
and the dispersion coefficient of the age at death, both in years. These param-
eters are described in the following material, but a simple way to understand
them is to think of a baby who is born today with the most likely age at which
he or she will die being, for example, m = 80 and the (approximate) standard
deviation around that age being, for example, b = 10. The survival probability
itself—which is the main quantity of interest—is obtained by taking the expo-
nent of the right-hand side of Equation 3.
Here is a detailed example: Assume that you are currently 50 years old
and would like to estimate the probability you will live (at least) to the age of
90, which is 40 more years. According to the Gompertz law of mortality, this
probability depends on two parameters—m, the modal age at death (roughly
speaking, the age to which you can expect to live) and b, the dispersion coeffi-
cient of m. Thus, these two numbers can loosely be thought of as the mean and
standard deviation of the length of your lifetime, which is obviously a random
variable.
Keep in mind that the parameters (m, b) are characteristics of a population
of a group of people, so m = 80 means that a member of the population can
expect at the time of birth to live to (about) 80. The remaining life expectancy
for an individual at age x is a different concept, given by the conditional prob-
ability, which will be higher than (m – x).
21
In the early 21st century, with cheap and vast computational power available, actuaries tend to
use actual (discrete) mortality tables, rather than closed-form analytic laws, to price and value
life annuities. But these sorts of rules and laws were a godsend when calculations had to be done
by hand. More importantly, and as any financial economist will attest, being able to reduce the
price of a capital asset down to a few critical parameters is prized for its own sake.

46 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 46
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:37 PM
 Ten Formulas to Know

Remember that if T(x) represents the random number of years you will
live from age x onward—that is, the remaining lifetime random variable—then,
for example, the expectation E[T(65)] > E[T(45)] – 20. So, be careful to dis-
tinguish between expectations at age zero and conditional expectations at any
higher age.22
Technically, the modal age at death is the age at which you are most likely
to die. It is actually a few years higher than the median (the age at which
50% of people your age will have died and 50% will still be alive). The rea-
son is the skewness of the distribution. In simple terms, if the modal age at
death is m = 80 years and the dispersion value is b = 11 years, then according
to Equation 3, the survival probability to age 90 is 8.9%, which can also be
expressed as a 91.1% probability of dying prior to age 90. In contrast, with a
higher modal age at death, m = 92 (instead of m = 80) years in Equation 3, the
survival probability to age 90 increases to 44.4%. Note how the extra 12 years
of life (in the modal sense) add 35.5 percentage points to the survival prob-
ability. In fact, if you “believe” that your modal age at death is indeed m = 92
years, then, according to Equation 3, the probability of surviving to age 95
(from age 50) is 27.5% and the probability of surviving to age 100 and becom-
ing a centenarian is 12.9%. That probability is obviously optimistic, but thus
says the Gompertz law of mortality when m = 92 and b = 11. The problem is
that the inputs are almost certainly unrealistic.
Figure 3 provides a graphical indication of how the survival probabilities
under the Gompertz law of mortality are affected by the modal age at death,
m. In all four cases, the dispersion coefficient, b, is taken to be 11 years, but the
modal age at death ranges from m = 80 to m = 92. Notice how all four curves
start off at a value of 100% but decline toward zero. By the age of 110, all four
curves are close to zero. The difference between the individual curves is the
rate at which the probabilities decline toward zero. The curve with the lowest
m value declines at the fastest pace. From a qualitative perspective, the curves
look similar to the survival values displayed in the last column of Table 6. In
fact, I leave as an exercise for the reader to use Excel and locate the best fit-
ting parameters (m, b) that would minimize the distance between the survival
curve defined by Figure 3 and the numbers displayed in Table 6.
The question is, of course, which parameters to use in practice when trying
to forecast survival probabilities and/or trying to price life annuities. As you
can plainly see, changing the modal age at death by only a few years can have
a dramatic effect on the probabilities. The issue is analogous to using the log-
normal distribution to approximate long-term portfolio returns. The analytics

That is, m is unconditional, as opposed to a conditional moment of a random variable. Another

22

way to think of m and b is in purely geometric terms as the two degrees of freedom embedded
within slope and intercept of the logarithm of the mortality rate.

©2013 The Research Foundation of CFA Institute  47

Milevxky.indb 47
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:37 PM
 Life Annuities

Figure 3.  Analytic Law of Mortality: Gompertz Survival Probabilities for

Various Modal Values, m, at Death

Survival Probability (%)

100

90

80

70

60

50

40

30

20

10
0
50 54 58 62 66 70 74 78 82 86 90 94 98 102 106
Survival to Age

m = 92 m = 85
m = 90 m = 80

Note: The dispersion coefficient, b, is 11 years.

are well understood, but the parameters are debatable. Should you assume an
expected return of 4% from stocks or closer to 8%? The same issues apply
when it comes to mortality and longevity modeling under any parametric law
of mortality. And it is in this area that historical data and current mortality
rates are used to calibrate such a model. The good news is that the mortal-
ity models, such as the Gompertz law (or its extension, the Makeham law),
tend to “fit” the ages around retirement remarkably well. Although I do not
want to get caught up in the actuarial minutiae and demographic details, the
Gompertz law of mortality has withstood the test of time.
In summary, there are two ways of working with (and thinking about)
retirement survival probabilities, which are part of the DNA of annuity pric-
ing. The first approach is to start with a mortality table that is applicable to
a given population group and then compute (using Equation 2) the relevant
survival rates. This approach can get messy, is computationally cumbersome,
and is not intuitive, but it is actually the route preferred by insurance actuar-
ies who perform these calculations. The second—more elegant and certainly
easier—approach is to “select” the best fitting modal, m, and dispersion, b,

48 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 48
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:38 PM
 Ten Formulas to Know

values for the population or individual in question—based on a given mor-

tality table or population mortality—and then use Equation 3 to compute
the survival probabilities. The numbers are close because of the underlying
Gompertz law of mortality.
Fortunately, for all intents and purposes, the Gompertz approximation is
good when it comes to pricing life annuities at retirement, which is the subject
of the next section’s formula.

Q17. Valuation: What Is the Gompertz Annuity Pricing

Model?
What you pay for a life annuity—or the amount of income you can expect
for a given premium deposit—is determined in a competitive market based
on the interaction of numerous insurance companies. So, although the actual
price you pay is partially determined by the forces of supply and demand, a
strict mathematical relationship links mortality expectations and interest rates
to observed prices. This idea is akin to the concept of arbitrage in securities
markets, in which the prices you pay for securities with similar characteristics
should not vary much from one another.
The most basic pricing formula (or asset pricing equation) for a life
annuity—and probably the most important formula in this book—can be
expressed as follows:
g 1 ω− x p ( x, i )
a ( x, g , R ) = ∑ + ∑ . (4)
i
i =1 (1 + R ) i = g +1 (1 + R )i
The expression a(x, g, R) denotes the upfront “cost” of $1 per year for life,
starting at the age of x, guaranteed for g periods, given annual nominal inter-
est rate R. On the right-hand side, you see two terms: the guaranteed por-
tion and the life-contingent portion. In the life-contingent portion, which
is the rightmost term, the ratios of the survival probability, p(x, i), and the
interest rate factor, (1 + R)i, are added up until the end of the mortality table.
Technically, the sum terminates at the age of ω (omega), which denotes the
oldest possible age attainable.
Equation 4 differs from the standard present value formula by having the
survival-contingent probability instead of the standard $1 in the numerator
of the summation. So, you can think of Equation 4 as the present value factor
of $1 of income to be received for as long as you are alive. For example, if you
are 70 years old and the probability of surviving for 1 year is 97%, for 2 years
is 95%, and for 3 years is 92%, then the first three terms of the life-contingent
present value embedded in Equation 4 are [0.97/(1.05)1] + [0.95/(1.05)2] +
[0.92/(10.05)3], with the remaining terms declining in importance until the
final numerator is zero.

©2013 The Research Foundation of CFA Institute  49

Milevxky.indb 49
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:38 PM
 Life Annuities

To be precise, you can use any survival probability vector of numbers in

the numerator of Equation 4 and add up the terms to arrive at a(x, g, R). Yet,
a closed-form expression for the life annuity factor a(x, g, R) can be obtained
when mortality is assumed to obey the Gompertz law and payments occur
in continuous time. The process involves the incomplete gamma function,
Γ(A, B). The mathematics of this process go well beyond this book, but for a
pure life annuity with no PC, the model price is found with the Gompertz
annuity pricing model (GAPM):

a ( x, g , r ) =
(1 − e− rg )
+
bΓ {−rb, exp[( x − m + g ) / b]}
, (5)
r exp {[m − x]r − exp[( x − m) / b]}

where r denotes the continuously compounded interest rate, or r = ln(1 + R).

This equation is GAPM for a life annuity that is guaranteed for g years (see
Chapter 6 of Milevsky 2006 for a derivation). In contrast to many other asset
pricing models, the GAPM fits market prices quite well.
Table 7 displays the actual (market) payouts available from U.S. life
annuities in late August 2012—based on the average of the best five com-
pany quotes—and compares them with outputs from the GAPM. The
model parameters are (m = 89.81, b = 11.61) for males and (m = 92.06, b =
11.1) for females. The interest rate used in Equation 4 is R = 2.88% for both
males and females. For comparison purposes, this interest rate was approxi-
mately 0.60% lower than the 30-year fixed-rate mortgage at the time and
approximately 0.30% higher than the 30-year T-bond yield. The three model
parameters (m, b, R) required to fit market prices were selected by use of a
nonlinear procedure minimizing the squared distance between the 30 pay-
outs in the pricing matrix.
Note that the model and market prices (for monthly income) are within
$15 of each other. For example, if you plug the values of m = 89.81 and b =
11.61 into Equation 3 for survival rates and then plug those rates into annuity
pricing Equation 4 with a constant interest rate of R = 2.88%, the resulting
annuity factor at the age of x = 65 is (approximately) $15.461 per dollar of
lifetime income. So, if you spend $100,000 on such a life annuity, you will be
entitled to $6,468 (100,000/15.461) per year, which is $539 per month. This
model value is a mere $7 more (in terms of monthly income) than what was
offered by the (average) insurance company in August 2012.
To conclude, no formula in finance (even the Black–Scholes option pric-
ing formula) can provide a perfect fit to the observed price of a financial
asset traded in the market, but the “life annuity factor” described by Equation
4—and the GAPM of Equation 5—provides a reasonably good fit to quotes
offered by insurance companies. For those in need of a quick number,
Equation 4 is the key.

50 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 50
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:38 PM
 Ten Formulas to Know

Table 7.  Market Prices vs. GAPM: Life Annuity Payouts per $100,000 Premium,
August 2012
Age 0-Year PC 5-Year PC 10-Year PC 15-Year PC 20-Year PC
A. Males
55 Market = $431 Market = $430 Market = $427 Market = $422 Market = $414
Model = $431 Model = $430 Model = $426 Model = $419 Model = $409

60 Market = $475 Market = $473 Market = $468 Market = $459 Market = $443
Model = $478 Model = $475 Model = $468 Model = $456 Model = $438

65 Market = $532 Market = $529 Market = $519 Market = $499 Market = $470
Model = $539 Model = $534 Model = $520 Model = $498 Model = $469

70 Market = $613 Market = $606 Market = $585 Market = $544 Market = $497
Model = $620 Model = $611 Model = $585 Model = $545 Model = $498

75 Market = $729 Market = $713 Market = $664 Market = $589 Market = $516
Model = $730 Model = $712 Model = $661 Model = $592 Model = $522

80 Market = $895 Market = $865 Market = $749 Market = $623 Market = $524
Model = $882 Model = $842 Model = $744 Model = $633 Model = $537
B. Females
55 Market = $416 Market = $416 Market = $414 Market = $409 Market = $403
Model = $416 Model = $415 Model = $412 Model = $407 Model = $399

60 Market = $456 Market = $455 Market = $451 Market = $443 Market = $432
Model = $457 Model = $456 Model = $451 Model = $442 Model = $428

65 Market = $507 Market = $505 Market = $497 Market = $482 Market = $461
Model = $513 Model = $509 Model = $499 Model = $483 Model = $460

70 Market = $578 Market = $574 Market = $557 Market = $527 Market = $491
Model = $586 Model = $579 Model = $560 Model = $529 Model = $491

75 Market = $686 Market = $674 Market = $635 Market = $579 Market = $513
Model = $685 Model = $672 Model = $633 Model = $578 Model = $518

80 Market = $838 Market = $809 Market = $724 Market = $617 Market = $523
Model = $823 Model = $793 Model = $717 Model = $623 Model = $536
Notes: “Market” price is the average of actual market quotes for males or females. “Model” price is the out-
put of the GAPM.
Source: QWeMA Group analysis based on data from Cannex Financial.

©2013 The Research Foundation of CFA Institute  51

Milevxky.indb 51
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:38 PM
 Life Annuities

Q18. What Are the Duration and Interest Rate

Sensitivity of a Life Annuity?
A life annuity is a type of fixed-income product—one in which the cou-
pons are higher than those of government or corporate bonds, partly because
of the annuity’s illiquidity. This unique and personal longevity-linked bond
is subject to default (i.e., your death), at which point only a small fraction
of the original investment will be recovered by creditors (i.e., your heirs).
Continuing with the “defaultable bond” analogy, we can say that the life
annuity value is sensitive to mortality rates and interest rates. And as with
any traded bond, its price (or value) will fluctuate on the basis of changes in
interest rates. So, although a life annuity paying $1,000 a month cannot be
(easily) traded after it has been purchased—nor does it really have a market
value in the conventional trading sense—it does have an ongoing “theoreti-
cal” value, which declines as you age. That is, the closer you are to death, the
less the same $1,000 monthly income is worth and the less you should have
to pay for it. Over short periods of time, however, interest rate changes are
what drive life annuity price changes, an observation that brings us to the
topic of duration.
The duration of a life annuity is defined as the derivative of the annuity
factor with respect to changes in the valuation rate, scaled by the (negative)
annuity factor itself. Formally, it is expressed as follows:
−∂a ( x, g , R ) / ∂R −1 ω− x −( i +1)
D ( x, g , R ) = = ∑ i p ( x, i ) (1 + R ) . (6)
a ( x, g , R ) a ( x, g , R ) i =1
Equation 6 includes the usual variables, but I have modified the expres-
sion for the survival probability, p ( x, i ) , to embed guarantee period g. So,
the survival probability is p ( x, i g ) = 1  as long as i ≤ g and then jumps to
p ( x, i ) = p ( x, i )  when i > g. Also, time variable g can be measured in days,
weeks, months, or years. This shorthand notation allows me to avoid breaking
the summation into two distinct portions.
Here is what matters: The greater the life annuity’s duration, D(x, g, R),
the more sensitive it is to changes in interest rates. Moreover, as is similar
to the duration for any conventional bond, we can approximate the per-
centage change in the value of a life annuity by multiplying the (nega-
tive) duration by the relevant change in interest rates. This result can be
expressed as follows:
a ( x, g , R + ∆R ) − a ( x, g , R ) ∆a ( x, g , R )
= ≈ − D ( x, g , R ) ∆R. (7)
a ( x, g , R ) a ( x, g , R )

52 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 52
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:40 PM
 Ten Formulas to Know

I will provide numerical examples in a moment, but note that Equations 6

and 7 are versatile and can be used in many different ways. The survival prob-
abilities can be (1) taken from a mortality table, (2) assumed on the basis of a
given analytical law of mortality, or (3) implied from actual life annuity prices.
I will provide more on this subject later.
Table 8 provides values for the duration of a life annuity at various ages
and assuming different PC guarantees under the GAPM. Notice how dura-
tion uniformly declines with age but does not necessarily change in a predict-
able manner when the guarantee period is increased. Indeed, the derivative of
the duration with respect to the guaranteed period is indeterminate, but the
derivative with respect to age is negative.
Here is how to understand the numbers in Table 8. Take, for example, a
male retiring at the age of 65 and buying a life annuity with a 10-year PC.
The annuity factor under the GAPM, based on Equation 5, is $11.92 per dol-
lar of lifetime annual income. So, a $100,000 premium would translate into
$8,389 (100,000/11.92) in annual income, $700 in monthly income, or $161
in weekly income. These model value payout rates are slightly higher than the
rates in the summer of 2012.

Table 8.  Theoretical Duration of Life Annuity: Males vs. Females

(duration in years)
Age 0-Year PC 5-Year PC 10-Year PC 15-Year PC 20-Year PC
Age 55     M = 11.03     M = 11.00     M = 10.97     M = 11.00     M = 11.14
       F = 11.91        F = 11.90        F = 11.88        F = 11.89        F = 11.95

Age 60 M = 10.02 M = 9.98 M = 9.94     M = 10.02     M = 10.28

    F = 10.90     F = 10.88       F = 10.86        F = 10.88        F = 11.01

Age 65 M = 8.96 M = 8.90 M = 8.87 M = 9.03 M = 9.50

   F = 9.82    F = 9.78    F = 9.76    F = 9.83        F = 10.09

Age 70 M = 7.87 M = 7.78 M = 7.78 M = 8.12 M = 8.88

   F = 8.66    F = 8.61    F = 8.59    F = 8.77    F = 9.26

Age 75 M = 6.76 M = 6.64 M = 6.73 M = 7.37 M = 8.47

   F = 7.47    F = 7.39    F = 7.41    F = 7.81    F = 8.66

Age 80 M = 5.67 M = 5.52 M = 5.82 M = 6.85 M = 8.28

   F = 6.26    F = 6.15    F = 6.30    F = 7.07    F = 8.33
Note: In the GAPM, under a constant 5.46% interest rate, with mortality parameters
of (m = 88.15, b = 10.50) for males and (m = 92.63, b = 8.78) for females.
Source: Based on data from Charupat, Kamstra, and Milevsky (2012).

©2013 The Research Foundation of CFA Institute  53

Milevxky.indb 53
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:40 PM
 Life Annuities

Now, consider the duration numbers. If the underlying interest rate

were to suddenly increase by 0.50% (i.e., 50 bps), the same $700 monthly
income would cost less than $100,000. In fact, a $100,000 premium would
lead to more lifetime income because pricing interest rates have increased.
The question is, By how much would the cost of the $700 monthly income
decline? The answer lies in the duration of the annuity factor, which from
Table 8 is D(65, 10, 5.46%) = 8.87 years for a 65-year-old male with a
10-year PC. So, finally, according to Equation 7, the annuity factor would
decline by 4.435% ([8.87][0.005] = 0.04435). The same life annuity
(stream of income) would cost $95,565 ($100,000[1 – 0.04435]) instead
of $100,000.
Notice, again, how the duration at advanced ages is much lower than for
younger ages—for example, 11 years at age 55 versus 6 years at age 80—which
implies a much lower sensitivity to changes in interest rates as a person ages.
Practically speaking, retirees in their late 70s and early 80s who are interested
in acquiring a life annuity but are “waiting for interest rates to improve” might
be surprised to learn that interest rates increasing by a percentage or two will
not make much difference in their income. In a person’s 70s and early 80s,
mortality rates are what drive payouts.
In conclusion, the duration number provides a quick-and-easy indica-
tion of the sensitivity of a life annuity’s value to a change in interest rates. So,
if you want to forecast how much more income you might get—for a given
premium—­if rates increased by a few percentage points, duration is the closest
you will get to a crystal ball.
Of course, whether actual annuity prices will adjust to actual changes
in interest rates based on these duration values is an empirical question for
another book.

Q19. What Is the Money’s Worth Ratio of a Life

Annuity?
Up to this point, I have been careless with such terms as “market value,”
“model price,” and “theoretical cost” of a life annuity. As you might recall,
I displayed some market prices and some mathematically contrived values,
and I then discussed various combinations of the two. To add to the confu-
sion, I have also been vague about the exact mortality rates—or mortality
parameters—that should be used in annuity pricing models. This approach
has been deliberate, but in this section, I would like to take the opportunity
to clean up terminology and be more precise about the links between (1)
model values and (2) the market price of a life annuity. And although you

54 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 54
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:40 PM
 Ten Formulas to Know

might expect the two numbers to be close to each other—and the GAPM
provides values that are within a few dollars of market prices—in this sec-
tion, I will formally compare these two numbers.
The metric that connects the two is called the “money’s worth ratio”
(MWR). Formally, it is defined as follows:
Model value
MWR= . (8)
Market price
The numerator in Equation 8 is the result of a formal mathematical rela-
tionship mapping mortality, interest rates, and transaction costs into a single
output number. The denominator is the actual price paid by the purchaser of a
life annuity, or at least the number quoted for the annuity.
If we assume that model values are close to market prices, then the MWR in
Equation 8 should be in the vicinity of 1.0. A higher ratio—numerator greater
than denominator—obviously implies a better deal for consumers. And although
the words “money’s worth” might emit a strong aura of economic efficiency and
fairness, remember that the numerator involves a model with particular assump-
tions and the denominator is a snapshot of a price at a given point in time. Both
quantities are subject to biases, which is something I will discuss later.
Notwithstanding some of these concerns, the MWR has become the metric
used to measure, compare, and contrast the efficiency of annuity markets around
the world. To my knowledge, at least three dozen research articles have been
published in the past two decades—all cited in the bibliography—that have
examined the MWR in countries from Singapore to Chile. Much of this work
has been conducted under the auspices of the International Monetary Fund
(IMF). Countries and markets with high MWR (MWR > 1.0) are deemed to
provide good value for consumers, whereas countries where MWR < 1 are clas-
sified as providing less value.
Table 9 provides a limited sample of these studies from various
English-speaking countries. Notice how most numbers are less than 1.0, except
for Canada, which is a curious matter, but we can only speculate as to the reasons.

Table 9.  MWR in Various Annuity Markets

Annuitant Mortality Rate Population Mortality Rate
Country Male Female Male Female
United States 0.927 0.927 0.814 0.852
Canada 1.17 1.06 0.965 0.937
United Kingdom 0.96–0.98 0.94–0.93 0.90–0.86 0.90–0.85
Australia 0.986 0.970 0.914 0.910
Note: A number of studies have measured MWR in the United Kingdom; the num-
bers here are the range reported.
Source: Based on data from Nielson (2012).

©2013 The Research Foundation of CFA Institute  55

Milevxky.indb 55
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:40 PM
 Life Annuities

The MWR values are higher when annuitant mortality rates are used in
the numerator of Equation 8 than when population mortality rates are used.
Remember that annuitant mortality implies a much higher survival probabil-
ity than population mortality. This contrast gets to the heart of the issue: The
MWR is about comparing a model price with a market price. Naturally, if
the model price assumes that people are going to live (much) longer, then the
numerator will be greater; hence, the MWR, as calculated, will be higher. (I
will soon discuss why the MWR may not measure true “money’s worth.”)
Here is an example of how the MWR is computed: Start with a $100,000
annuity premium (investment) and assume that it would generate $500 per
month for life for a 65-year-old male with a 15-year PC. This number would
be an average quote from the relevant insurance companies. The MWR met-
ric is usually applied to a market as opposed to an individual company, so the
denominator of Equation 8 would be the $100,000 premium.
Moving on to the numerator: The $500 per month, or $6,000 per year, of
lifetime income is multiplied by the annuity factor pricing model—the GAPM
described previously with a given assumption for (1) mortality rates and (2)
interest rates. The mortality rates could be average population rates or healthier
annuitant rates. Similarly, the interest rate could be a fixed number or an entire
term structure of rates. In fact, these rates could be risk-free government rates or
risky corporate rates. Once all these assumptions (i.e., modeling decisions) have
been made, the resulting annuity factor is multiplied by the $6,000 income and,
finally, compared with the $100,000 premium. The ratio of these two numbers is
the MWR. Obviously, very different MWR values can be obtained depending
on the exact assumptions made for the annuity pricing model.
Although the MWR is widely used by researchers, people should keep in
mind a number of issues or concerns when interpreting results of such studies.
First, insurance companies with low credit ratings (which are likely to offer
more on life annuities) will report higher MWR values. Second, these num-
bers tend to be snapshots in time. Third, they are driven by model and pricing
assumptions involving both mortality and interest rate expectations over long
periods of time.
When compared with asset pricing models in financial economics, annu-
ity pricing models produce remarkably better fits with market prices. In fact,
using the GAPM, for example, you can fit market payouts to within a few
dollars, as displayed in Table 7. Nevertheless, this section is not meant to boast
about goodness of fit but, rather, to remind readers how critical mortality
assumptions are when pricing and valuing life annuities. Depending on your
mortality model, you might get MWR values that exceed unity by as much as
10% or fall short of unity by as much as 20%. What this means from a purely

56 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 56
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:40 PM
 Ten Formulas to Know

economic point of view should be clear. A life annuity is worth (much) less to
you personally if you are in average health (that is, health typical of the overall
population) than if you are in excellent (annuitant) health.
Finally, I recommend, in jest, that if you do have the choice, you should
buy your annuity in the country with the highest MWR.

Q20. Can You Afford to Wait? Introducing the Implied

Longevity Yield
Life annuities yield more income than non-mortality-linked fixed-income
instruments because the annuities’ benefits accrue only to the exclusive club of
survivors. In this section, I will present a new and intuitive method for under-
standing the magnitude of these benefits. The metric is called the “implied
longevity yield” (ILY) and is yet another way to think about mortality credits.23
To understand the context of the ILY metric, think in terms of the natural
option everyone has—and many use—to simply delay annuitization. As most
procrastinators would argue, why do something today if you can wait until
tomorrow, next week, or even next year? What is the urgency of now? The
same question might apply to buying a life annuity. It is costly. It is irrevers-
ible. What’s the rush? Bear with me for a moment and imagine the following
situation. You are 65 years old and have $100,000 available to finance your
retirement spending. So (perhaps after reading this book), you are thinking of
annuitizing. Should you wait?
Assume that if you were to annuitize today, you would receive $517 in
monthly income (with zero PC), which is $6,204 of yearly income, for life.
This outcome can also be expressed as a cash yield of 6.2% for life, which obvi-
ously includes the return of principal. Now, this embedded return of principal
makes it difficult to directly compare the 6.2% with, say, a 10-year government
bond yielding a coupon of (only) 2.4% on the same day as the annuity quote.
Alas, the 3.8% difference between the two numbers overstates the benefit
from the life annuity because of the blended mixture of interest and principal.
So, how do you (properly) compare the two? What is the actual spread over,
say, U.S. Treasury rates, which results from joining a longevity pool?
Following on the idea of procrastination, imagine that instead of buying
the $517 monthly annuity now, you decide to (1) wait for 10 years, (2) system-
atically withdraw the $517 from an investment portfolio in the meantime, and
then (3) try to purchase the same $517 life annuity at age 75. This approach
is called “self-annuitization,” and although forecasting how much the iden-
tical $517 monthly annuity might cost in 10 years’ time is difficult, a good
proxy is available—namely, today’s price. Suppose that on the same day as the
The ILY is now a registered trademark of Cannex Financial Exchanges. More information on
23

the algorithm behind the ILY can be found in Milevsky (2005b).

©2013 The Research Foundation of CFA Institute  57

Milevxky.indb 57
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:40 PM
 Life Annuities

quote for the $100,000 cash policy, a 75-year-old is quoted $704 in monthly
income for life—$8,452 in annual income and a life annuity yield of 8.45%.
The higher payout is caused by the shorter remaining life expectancy of this
person than you, higher mortality rate, and other factors. The bottom line is
that if you at 65 decide to wait 10 years and purchase the $517 annuity at
age 75 (and assuming prices remain exactly the same), you will need roughly
$73,400 ($6,204/0.0845) in 10 years. In the meantime, you have to be careful
not to spend more than $26,600 in principal.
So, to beat the annuity benefit of buying as a 65-year-old, you would have
to generate enough interest on the $100,000 that you would have $73,400 left
over at the end of 10 years with which to buy the life annuity while withdraw-
ing a $517 monthly income in the meantime.
Once the problem is posed in this manner, the answer boils down to a
basic problem in finance. It can be expressed as an internal rate of return
(IRR). Formally, the ILY is defined as
 a(x + t)
ILY = IRR t , a ( x ) , . (9)
 a ( x ) 
The symbol IRR(T, A, B), where T is the first argument, A is the second
argument, and B is the third argument, is shorthand notation for the IRR,
where the cash flow, A, is paid up front, then $1 is received annually for T
years, and then a final cash flow amounting to B dollars is received at the end
of T years.24
Table 10 displays this example, with the breakeven amount (or IRR) in
tabular form. It also provides the same analysis for a 65-year-old female. The
numbers are interpreted this way: A 65-year-old male would have to earn at
least 4.12% (every year) between the ages of 65 and 75 to be able to purchase
the exact same life annuity stream at age 75, assuming annuity prices remained
the same. This number is 2.43 percentage points greater than the risk-free U.S.
government bond listed on the same day on which the annuity quotes were
obtained. Therefore, joining the mortality pool between the age of 65 and 75
will yield an extra 2.43 percentage points above the safest asset you could have
purchased. If this does not seem like enough to you, then perhaps you should
wait to annuitize. The point is to convert the payout from the annuity into a
yield number that you can think of as a target to beat.
Note that for females, the ILY is lower than for males. The cause is the mor-
tality credits; that is, the number of people dying between ages 65 and 75 is lower
for females, so the subsidy is also lower. Similarly, if the life annuity you purchase
at age 65 contains guarantee periods, refunds, and joint-life options and if you
receive less than $517 per month (for males) or $485 (for females), the ILY value

24
For those readers who wish to express Equation 9 in Excel, it is RATE(T,1,–A,B,0,guess).

58 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 58
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:41 PM
 Ten Formulas to Know

Table 10.  The Implied Longevity Yield

Male Female
A. Monthly income on $100,000 for a
65-year-old who annuitizes $517.00 $485.00
B. Needed if person wants to purchase the
same income stream as a 75-year-old $73,400 $74,500
C. Annualized 10-year return required on
$100,000 to generate monthly income
in row A and have enough left over to
purchase row B 4.12% 3.77%
D. Yield on 10-year U.S. government
bond (November 2012) 1.69% 1.69%
E. Spread on bond yield 2.43% 2.08%
Notes: Payouts assume zero guarantee period. Average of 17 quoting companies as of 22 November
2012.

will be lower. In fact, many retirees purchase life annuities or annuitize a portion of
their nest eggs but add on various guarantees, which then greatly reduce and water
down the mortality credits. In many cases, the ILY values are little above govern-
ment bond yields. Once again, you are not really buying much of a life annuity if
the ILY is close to what you could get from the safest investment possible.
To conclude, even if you have no intention of annuitizing today (or in
the future), the ILY provides an alternative perspective on the threshold, or
benchmark investment return, required to beat the life annuity.

Q21. What Is the Lifetime Ruin Probability from

Self-Annuitizing?
In the previous section, I dealt with the return required to induce an investor
to delay annuitization. I described the breakeven rate known as the “implied
longevity yield” that you would have to earn, while waiting to annuitize, to
be able to purchase the same annuity stream at some fixed date in the future.
In this section, I continue this theme and address a related question: What if
you decide to forgo annuitization entirely? What is the probability you can
maintain a given withdrawal rate and standard of living while you are still
alive? Inspired by the insurance literature, this question has become known as
the “lifetime ruin probability” (LRP). It is the probability that your biological
lifetime will be longer than the financial lifetime of your portfolio.
Technically, the formula can be expressed as a present value:
 ω− x p ( x, t ) 
LRP ( w, x, µ, σ ) = Pr  ∑ t > w  , (10)
 ( )
 t =1 ∏ j =1 1 + R j 


©2013 The Research Foundation of CFA Institute  59

Milevxky.indb 59
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:41 PM
 Life Annuities

where the portfolio spending rate is 1/w (which is the inverse of initial
wealth w on the right-hand side), x is the initial age of the retiree, p(x, t)
is the survival probability curve, ω is the maximum age, and Rj is the real-
ized portfolio return in period j. The random variable Rj is a function of
the portfolio’s asset allocation and is summarized by the return–risk pair of
mean and standard deviation (μ, σ) in the LRP. For example, LRP(1/0.05,
65, 0.06, 0.20) denotes the lifetime ruin probability under an assumed
spending rate of 5% (i.e., $5 is spent annually for every $100 of initial prin-
cipal) for someone aged 65 investing in a portfolio that is expected to earn
6% with a standard deviation of 20%. The LRP for someone who is trying
to replicate the payout from a life annuity would, therefore, be denoted by
LRP[a(x, R) x, μ, σ].
The intuition behind Equation 10 is as follows: Assume for the moment
that Rj = R is constant. The left-hand term inside the square bracket is the
present value of a life annuity cash flow (that is, the annuity factor) because
the product Π tj =1 (1 + Ri ) in the denominator collapses to (1 + R)t. The entire
expression is the annuity factor. And if the annuity factor is greater than the
initial sum of money, w, available to finance spending, the individual is ruined.
Generally, when Rj is random, we can only talk about the probability that the
present value is greater than w, which is what Equation 10 is trying to capture.
So, Equation 10 is not an explicit formula. It describes an algorithm.
The LRP value displayed in Equation 10 can be computed in a number
of ways. A relatively easy methodology is to simulate a vector of Rj portfolio
returns and assume a particular mortality table, p(x, t), then count the number
of scenarios in which the mortality-weighted present value is greater than w.
This is the Monte Carlo approach to retirement income simulations. Another
(more accurate, in my opinion) approach is to analytically represent the LRP
as a solution to a partial differential equation and then use numerical schemes
to quickly and efficiently solve for the LRP. Although the exact methodology
is beyond the scope of this book, in the numerical examples that follow, I will
display results on the basis of this approach.25
Table 11 provides some examples, in which the reader can see the impact
of spending rates and asset allocation on the LRP. Remember that the retiree
is trying to replicate the cash flow from a life annuity until the account itself
goes broke and runs out of money.
For example, imagine a 65-year-old retiree with $1,000,000 in investable
assets who does not want to purchase a life annuity and, instead, would like
to withdraw $60,000 per year in inflation-adjusted terms from an investment
25
I used an Excel add-in created by the QWeMA Group, which computes a continuous-time
version of the LRP by solving the relevant partial differential equation. For more informa-
tion, visit www.qwema.ca, see the references in Milevsky (2012), or use the approximation in
Milevsky and Robinson (2005).

60 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 60
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:41 PM
 Ten Formulas to Know

Table 11.  LRP of Self-Annuitization at Age 65

100% Risky 80% Risky 60% Risky 40% Risky 20% Risky
Spending Stocks; 0% Stocks; 20% Stocks; 40% Stocks; 60% Stocks; 80%
Rate Safe Cash Safe Cash Safe Cash Safe Cash Safe Cash
2.0% 3.07% 1.41% 0.44% 0.06% 0.00%
2.5 5.71 3.27 1.49 0.43 0.05
3.0 9.14 6.13 3.64 1.78 0.72
3.5 13.23 9.98 7.15 4.94 3.99
4.0 17.81 14.67 12.03 10.38 11.60
4.5 22.70 19.98 18.05 17.86 22.56
5.0 27.75 25.69 24.83 26.61 34.22
5.5 32.83 31.56 31.94 35.65 44.66
6.0 37.82 37.40 39.00 44.23 53.22
6.5 42.66 43.05 45.71 51.86 60.00
7.0 47.26 48.40 51.90 58.38 65.35

portfolio. The $60,000 (inflating) withdrawals will be made from interest, div-
idends, and principal, if needed. Assume that the portfolio is invested 80%
in stocks—with an expected return of 6% and volatility of 20%—and the
remaining 20% is in cash yielding 1.5%. All returns are in inflation-adjusted
terms. In this case, the arithmetic mean return of the portfolio is μ = 5.10%
and the volatility is σ = 16%. The spending rate of 6% is equivalent to w =
1/0.06 = $16.666 per dollar of spending.
According to Table 11, the LRP—that is, the probability that the entire
portfolio will be exhausted before the random time of death—is 37.40%. This
probability is obviously high, and the spending rate of 6% is clearly unsustain-
able. Note that even if the asset allocation is increased to 100% stocks, which
has a higher expected return, the lifetime ruin probability is even higher.
It results in an LRP of 37.82%. Similarly, if the risk exposure is reduced to
60% stocks, the value of the LRP is 39%, which is even worse. In summary, a
$6-per-$100 spending rate from a portfolio at the age of 65 is unsustainable,
unless the retiree expects much higher returns from the stock market than
I have modeled here—an expected (arithmetic) return that exceeds 6% real
return. Such a case would be hard to make in today’s economic environment.
If the withdrawal rate is reduced to 3.5% while the asset allocation remains
80% stocks and 20% cash, Table 11 shows that the LRP drops to 9.98%, which
is a failure rate of approximately 1 in 10. This probability might not be accept-
able to everyone, but it is certainly more sustainable than a 6% spending rate.
Recall that all of these numbers and spending rates are in real (inflation-
adjusted) terms, which reduces sustainability in comparison with nominal
spending. If I were to assume a retiree spends a nominal (as opposed to real)

©2013 The Research Foundation of CFA Institute  61

Milevxky.indb 61
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:41 PM
 Life Annuities

$60,000 per year from a $1,000,000 portfolio, then I would have to modify the
(μ, σ, R) parameters and convert them into nominal terms. The LRP values
would fall.
A number of other qualitative insights are worth noting from Table 11.
First, the LRP is reduced with lower spending rates (1/w) and higher values
of w. The relationship is monotonic. The same is not the case with asset allo-
cation. Notice how at low spending rates, the LRP declines with increasing
exposure to safe assets. At higher levels of spending, the LRP is U-shaped
as a function of asset allocation. It is higher for much riskier and much safer
allocations and is minimized in between 100% stocks and 100% safe cash.
The intuition here is that further increasing exposure to stocks does not
necessarily improve the odds of success because of the higher shortfall risk
embedded in the portfolio.
In conclusion, the LRP is a summary risk metric that can help mea-
sure the sustainability of a retirement plan. The lower the LRP, the bet-
ter. I want to be careful not to advocate LRP minimization, however, as a
dynamic portfolio strategy. Rather, it should be viewed as yet another way
of quantifying the benefit of annuitization. If you purchase a life annu-
ity, the insurance company is on the hook regardless of how long you live
or how the stock market performs. So—in this section’s language—if the
insurance company’s credit is good, the LRP of the payout from a life
annuity is zero.

Q22. How Does a Variable Immediate Annuity Work?

After reading the previous section on the odds that a diversified portfolio
of stocks and bonds can beat a life annuity, you might naturally inquire
whether you can actually reap the benefits of both the equity risk premium
and mortality credits. Can you wrap a life annuity concept around a mix
of stocks and bonds? The answer is a resounding yes, and it is the topic of
this section.
Variable immediate annuities (VIAs), also known as “immediate vari-
able annuities” or “variable payout annuities,” are the “risky” counterpart to
the “safe” life annuity. These products allow annuitants to (1) receive a life-
time of income they cannot outlive but also (2) have the ability to earn vari-
able market-linked returns in the annuity structure. Technically, this feat is
engineered by an insurance company paying out annuity units, as opposed
to dollars and cents. Each year—or month, depending on the frequency of
payment—the actual annuity payment is adjusted up or down in accord with
how its underlying portfolio of stocks and bonds has performed. The number
of units is fixed. Their value fluctuates. Moreover, the annuitant selects a bal-
anced mixture of stock and bond funds (or subaccounts, in the parlance of

62 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 62
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:41 PM
 Ten Formulas to Know

insurance) and also picks a hurdle rate that must be achieved before payments
will increase. If markets do better than the hurdle, cash flow increases. If it
falls short, income shrinks.
The mechanics of a VIA seem tricky and obscure at first, so Equations
11a and 11b are provided to clear the fog. They display exactly how the
annuity payments are adjusted on the basis of the investment portfolio’s per-
formance. Equation 11a addresses the baseline initial payout, c0, as a func-
tion of the hurdle rate, and Equation 11b addresses how the payment is
adjusted over time:
W
c0 = ; (11a)
a ( x, R )

ci −1 (1 + Ri )
ci = . (11b)
1+ R
The symbol W denotes the initial premium (or wealth) used to pur-
chase the variable immediate annuity. The familiar symbol a(x, R) denotes
the standard annuity factor under an “assumed” rate of interest (that is, the
required hurdle rate). Finally, the actual return earned by the underlying
investment portfolio is denoted by Ri and will determine the actual annu-
ity payment. Notice that when Ri > R, the next period’s payment, ci, will
increase, when Ri < R, the next payment will decline, and when Ri = R, the
payment will remain unchanged.
Table 12 provides some numerical values for the baseline, Year 1, and Year
2 outcomes under various assumed investment returns (AIRs). For example,
suppose you are a 65-year-old (male or female) who has purchased a VIA
and has voluntarily selected an AIR of R = 4%. According to the GAPM and
Equation 11a, with parameters m = 87.25 and b = 9.5, the baseline annuity
payment is $7,820 per year. So, in theory, if the underlying investments in
the VIA earned a fixed constant Ri = 4% every year (forever), your life annu-
ity payment would remain at $7,820 per year. Now, assume that during the
first year—that is, between ages 65 and 66—the underlying portfolio of sup-
porting investments earns R1= 0%, which is 4 percentage points less than the
hurdle rate, or AIR, of R = 4%. In that case, the annuity payment drops by 4%
and the annuity payment falls during the second year to $7,519. Following
through to the second year, if the investment return from the underlying port-
folio is a loss of R2 = –25%, then the Year 2 payment drops by another 29% to
$5,423.
Note that for every year in which the investments earn less than the AIR
hurdle, the annuity payment will drop. So, if you want to avoid a decline in
income, you need to select the lowest AIR possible. Of course, doing so will
increase the annuity factor, a(x, R), which then will reduce the amount of the

©2013 The Research Foundation of CFA Institute  63

Milevxky.indb 63
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:42 PM
 Life Annuities

Table 12.  Payout from a VIA for Year 1 and Year 2

Payout
With AIR = With AIR = 4%; With AIR = 6%;
2%; Baseline Baseline Year 1 Baseline Year 1
Year 1 Return Year 2 Return Year 1 Payout = Payout = Payout =
(%) (%) $6,420.00 $7,820.00 $9,340.00
R1 = –25 NA $4,720.59 $5,639.42 $6,608.49
R1 = 0 NA 6,294.12 7,519.23 8,811.32
R1 = +25 NA 7,867.65 9,399.04 11,014.15

R1 = –25 R2 = –25 3,471.02 4,066.89 4,675.82

R1 = –25 R2 = 0 4,628.03 5,422.52 6,234.43
R1 = –25 R2 = +25 5,785.03 6,778.15 7,793.03

R1 = 0 R2 = –25 4,628.03 5,422.52 6,234.43

R1 = 0 R2 = 0 6,170.70 7,230.03 8,312.57
R1 = 0 R2 = +25 7,713.38 9,037.54 10,390.71

R1 = +25 R2 = –25 5,785.03 6,778.15 7,793.03

R1 = +25 R2 = 0 7,713.38 9,037.54 10,390.71
R1 = +25 R2 = +25 9,641.72 11,296.92 12,988.39
Notes: The data are for a 65-year-old unisex. The GAPM parameters are m = 87.25 and b = 9.5.
NA = not applicable.

initial baseline payment. So, if you choose a low AIR, your annuity income
is less likely to be sabotaged by a down market, but it will be starting from a
relatively low level of income.
If you select an AIR of 6% and during the first two years of retirement
the underlying investment portfolio earns 25% in the first year and 25% in the
second year, then Table 12 indicates that your annuity payment will grow to
$12,988 by the time you are 67. In contrast, if you select, or opt for, a 2% AIR
and the market drops by 25% in the first year and 25% in the second year, your
annuity check will be $3,471 by age 67. Note the range of outcomes.
Another helpful way to think about a VIA is to view it as a fixed life annu-
ity, but one that is paid in a foreign currency (euros or yen, for example). The
payments are converted back into dollars at the prevailing exchange rate and
paid out to the annuitant. So, if the currency has depreciated since the last
payment was received, income will decline from a dollar perspective. But if
the currency has strengthened since the last payment, the dollar value will
increase. From a dollar perspective, the annuity income may appear to be fluc-
tuating, but from the perspective of the foreign currency holder (and payer),
the cash flows are fixed.

64 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 64
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:42 PM
 Ten Formulas to Know

The same logic can be easily transferred to the VIA. The payments fluc-
tuate in relation to the value of the underlying investment units, but from
the perspective of the insurance company, the annuitant is entitled to a fixed
number of units. In fact, insurance companies manufacture (and hedge) their
obligations under a VIA in this manner.
Certainly, because of the market’s behavior and the AIR selected, annui-
tants can expect a wide range of payments from a VIA. This wide range is prob-
ably one of the reasons that VIAs are not popular in practice among retirees.
They are a tiny fraction of the market. Many people believe that retirement is a
time for stable and predictable income. They do not want to be exposed to the
ups and downs of the stock market. Therefore, some observers have said that a
product with the words “variable income” and “annuity” in it is an oxymoron.
Of course, the annuitant can control the volatility of his or her income stream
by selecting a more or less conservative allocation for the underlying investments
supporting the annuity. In theory, he or she can allocate 100% of the money to
safe money market funds, but if that is the annuitant’s preference, a fixed imme-
diate annuity—that is, a generic life annuity—might be more suitable.
In summary, the benefits of both longevity-risk pooling (that is, of mor-
tality credits) and the equity risk premium may be obtained by purchasing a
VIA instead of a fixed immediate annuity. Surprisingly, however, VIAs are even
less popular than fixed annuities, possibly for behavioral reasons and possibly
because of product complexity. Either way, one thing is certain: A life annu-
ity is not an alternative to—nor does it compete with—a diversified portfolio
of stocks and bonds. Rather, annuity payout characteristics can be overlaid on
stock and bond returns (a VIA) or obtained independently of them (a fixed
immediate annuity).

Q23. What Is the Difference between a Tontine and a

Life Annuity?
According to the Merriam-Webster dictionary, a tontine is “a joint financial
arrangement whereby the participants usually contribute equally to a prize
that is awarded entirely to the participant who survives all the others.” So,
tontines and life annuities might appear to be similar, especially to those unfa-
miliar with longevity-contingent claims. Indeed, they both contain longevity
insurance and protection against living “too long.” Technically, both a tontine
and a life annuity are a form of debt from the point of view of the issuer
because a large sum of money is advanced to a financial institution, such as an
insurance company (or even the king or queen in medieval times), entitling
the investor to annual payments over the life of a nominee. The subtle differ-
ence between the two is exactly how those annual payments are determined
and what happens when the nominee dies.

©2013 The Research Foundation of CFA Institute  65

Milevxky.indb 65
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:42 PM
 Life Annuities

First, some notation is needed. Recall that for a fixed life annuity, the
a(x, R) denotes the annuity factor at age x under a valuation rate of R. (I have
eliminated guarantee period g to keep things simple.) Recall that the annuity
factor is the lump sum of money an x-year-old investor must pay in exchange
for $1 of income during the rest of the nominee’s life. In most cases, of course,
the investor is the nominee:
ω− x p ( x, t )
a ( x, R ) = ∑ . (12a)
t =1 (1 + R )t
A critical point here is that upon death, the obligor (e.g., insurance com-
pany, government, or even the king and queen) is exempt from further pay-
ments. Many variations of Equation 12a are discussed in the literature, but the
basic idea is the same: (1) The death of the nominee ends the life annuity, and
(2) the investor does not care (aside from credit risk) how many other people
have purchased similar life annuities. Every life is an entity unto itself.
With a tontine, however, an investor’s peers and fellow annuitants play an
important role. In a generic tontine, the investor is guaranteed to receive, for
example, $1 of income for as long as the nominee is alive but the actual income
received depends on the number of other tontine nominees who die during the
year. Each death increases the surviving investors’ income. So, for example, if
100 investor/nominees purchase tontine units paying $1 per year for life, the
nominee pool consists of $100 per year. Then, 10 years later, if only 50 nomi-
nees are alive in the tontine pool, the surviving investors get to share the $100,
which is a payment of $2 per investor. Then, 20 years later, if only 10 nominees
remain in the tontine pool, the surviving investors get $10 of income each.
With a tontine, you and the other nominees are guaranteed the $1 of income
but the upside potential is enormous as long as you are still alive. (No wonder
tontine schemes have spurred the imagination of crime writers for centuries.)
To investigate the math for the tontine, let o(x, R) denote the tontine fac-
tor at age x with a valuation rate of R. Each tontine unit—paying at least $1
for life—will cost o(x, R) dollars up front. So, in parallel with the definition of
annuity factor, the tontine factor is
ω− x 1
o ( x, R ) = ∑
t =1 (1 + R )t (12b)
−( ω− x )
1 − (1 + R )
= .
R
Note the difference between Equations 12a and 12b: o(x, R) > a(x, R)
because there is no survival probability, p(x, t) < 1, in the numerator of the ton-
tine factor. In fact, o(x, R) is simply the present value factor of a term certain
annuity paid over (ω – x) years. Intuitively, there is no mortality in the formula

66 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 66
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:43 PM
 Ten Formulas to Know

because the obligor must continue to make payments to the pool regardless of
who is alive. As long as even one nominee is still alive, the payment to the ton-
tine pool continues. The tontine is effectively a term certain fixed annuity to
the obligor, but with a miniscule amount of risk that depends on the longest
lived annuitant—the term of the annuity.26
To monitor the size of the tontine pool, let N(x, t) denote the number
of original tontine nominees at age x who are still alive in year t. Naturally,
N(x, t) will decline over time, and eventually, once the entire group has died
off, N(x, ω – x) will equal zero. The actual tontine payment to the survivors
will be a multiple of the ratio between the original nominees and the number
of survivors: N(x, 0)/N(x, t).
Here is an example of the mechanics. Let’s assume a group of x number
of 50-year-olds and a valuation rate of R = 6% per year. As for the mortality
probability, p(x, t), let’s assume that the individual survival probability obeys a
Gompertz distribution with the modal value m = 86.549 and dispersion value
b = 10. This parameterization implies that p(50, 45) = 10%, and everyone is
dead by age ω = 120.
Let’s assume now that you (as both investor and nominee) are 50 years old
and have a choice between purchasing a life annuity paying $1 a year for life
and a tontine paying (at least) $1 a year for life. According to Equation 12a,
the life annuity will cost $13.303. So, an investment of $1,000 will result in a
constant payment of $75.17 (1,000/13.303) for life. This payment is a yield of
7.517%, which is 1.5 percentage points above the 6% valuation rate because of
the mortality credits. So far, nothing is new.
The tontine, according to Equation 12b, will cost $16.3845 per unit and
will entitle the nominee to (at least) $1 of income per year for life, with the
potential for more depending on the survival of other nominees. An invest-
ment of $1,000 will lead to 1,000/16.3845 = $61.03 (at least) per year for life.
Table 13 displays results under the assumption that 1,000 people each con-
tribute $1,000 dollars for a total of $1,000,000. The $61,030 per year benefit
to the pool of nominees—assumed to be made for a total of 70 years—is split
among all survivors. The last survivor then keeps the entire $61,030 each year
until he or she dies. For example, although there is only a 61% chance that a
50-year-old nominee/investor will survive to age 80, if he or she does actually
live for 30 years but only N(50, 30) = 610 other members of the tontine pool
survive, the nominee/investor will be entitled to a payout of $99.98. This is
$25 more than what the life annuity would provide. And if a mere N(50, 30)
= 488 members of the original pool of 1,000 nominees survive, the tontine
income becomes $124.98.

More information is available in my forthcoming manuscript tentatively entitled “Tontines:

26

How a Fascinating but Neglected Annuity Scheme Can Help Reduce the Cost of Retirement.”

©2013 The Research Foundation of CFA Institute  67

Milevxky.indb 67
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:43 PM
 Life Annuities

Table 13.  Tontine vs. Life Annuity Payouts over Time: $1,000
Invested at Age 50
Estimated Annual Annual
Survival Actual No. of Tontine Annuity
Age Probability Survivors Payout Payout
Year 5 (i.e., age 55) 98.34% 1,000 $61.03 $75.17
983 62.07 75.17
787 77.58 75.17

Year 10 (i.e., age 60) 95.65 1,000 61.03 75.17

956 63.81 75.17
765 79.76 75.17

Year 30 (i.e., age 80) 61.04 732 83.32 75.17

610 99.98 75.17
488 124.98 75.17

Year 45 (i.e., age 95) 10.00 120 508.42 75.17

100 610.10 75.17
80 762.63 75.17

In the tontine, you gain from others’ misfortune, which creates some inter-
esting moral (and mortal) hazard problems.
To conclude, although tontines were extremely popular in medieval
Europe—and actually proposed by Alexander Hamilton, the first U.S. secretary
of the Treasury, as a way to reduce U.S. debt—they are currently illegal in most
developed countries. This book is not the place to delve into the reasons for the
tontine bans and whether they are justified in today’s environment, but you can
certainly see the appeal of tontines over annuities for someone like Methuselah.
Perhaps they will be in vogue again someday.

68 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 68
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:43 PM
 3. The Scholarly Literature

The scholarly (academic and practitioner) literature on the topic of life annui-
ties is vast and growing. As of late 2012, I counted somewhere in the vicinity of
2,000 research articles (based on Google scholar citations) written during the
past 50 years that could be considered part of an extended field of annuity lit-
erature. Thus, listing, mentioning, or giving credit to all of them is impossible. I
have done my best to narrow down the list of relevant research to approximately
200 key articles. The filters I used for inclusion were articles on life annuities that
would be relevant to private wealth managers, institutional asset managers, and pen-
sion plan sponsors as well as scholars conducting research in this area.
To refine and organize this task, I separated the list of 200 or so research
articles into six streams or subfields in the life annuity literature. Although
some overlaps exist between the groups, I believe the articles can be classified
along the following general lines:
1. The life-cycle model of saving and consumption, which acknowledges that
the length of life is random rather than fixed and studies conditions in a
world in which life annuities are not necessarily available. This stream of
literature is concerned with the impact of (pure) longevity risk on rational
consumer behavior and with the way consumers behave once life annui-
ties are introduced into the opportunity set. The first formal discussions in
the economics literature are Fisher (1930) and Yaari (1965). I also include
in this section articles that tie annuities to the capital asset pricing model
(Sharpe 1964 and other authors).
2. The pricing, valuation, hedging, and reserving of life annuities. If you seek a
first paper in this subfield, it is the key article by Halley (1693). Articles in
this genre are actuarial and technical in nature. They will be discussed and
referenced only insofar as they relate to the (theoretical) cost of annuities
for individuals.
3. The optimal allocation and timing of annuitization. This subfield is closest to
the traditional investment asset allocation literature in that it attempts to
derive, in a normative fashion, the optimal amount of personal wealth that
should be allocated to a life annuity and the best time (and age) to annui-
tize. This research is usually embedded in a rational, utility-maximizing
life-cycle framework similar to the multiperiod asset allocation literature
pioneered by Samuelson (1969) and Merton (1971). The Yaari (1965)
article is key to this stream as well the first group mentioned, but the focus
is on modern “portfolio choice” models as opposed to the theoretical opti-
mality of life annuities.

©2013 The Research Foundation of CFA Institute  69

Milevxky.indb 69
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:43 PM
 Life Annuities

4. The formulation and solution of the annuity puzzle. This group, by far the
largest subfield in the annuity literature, addresses the (puzzling) phe-
nomenon that few people actually choose to annuitize. The puzzle was
first identified formally by Modigliani (1986), but can actually be traced
to Huebner (1927). Although this genre is labeled a “puzzle,” most of the
recent articles I will reference argue that it might (now) be less puzzling
than previously thought.
5. The money’s worth ratio of actual annuity prices in the United States and around
the world. This subfield, first introduced by Friedman and Warshawsky
(1990), is an attempt to compare actual prices with model prices and to
measure goodness of fit, efficiency, and other metrics of concern to econo-
mists. The money’s worth ratio was explained in Chapter 2; the litera-
ture review will provide an opportunity to showcase the large numbers of
researchers who use this metric.
6. Articles that do not fit neatly into the preceding categories. No key or first arti-
cle comes to mind for this category. It is a bit of a catchall for research
articles that do not belong in the other categories and is presented last for
that reason.
In each category discussion, the key research articles are given chronologi-
cally. Generally, relevant excerpts provide key insights from the articles in the
authors’ own words.

The Life-Cycle Model and Life Annuities

Most experts agree that the economics of annuities research begins with the
life-cycle work of Fisher (1930)—in particular, his following comments:
The shortness of life thus tends powerfully to increase the degree of impa-
tience, or rate of time preference, beyond what it would otherwise be. This
is especially evident when the income streams compared are long . . . But
whereas the shortness and uncertainty of life tend to increase impatience,
their effect is greatly mitigated by . . . solicitude for the welfare of one’s
heirs. Probably the most powerful cause tending to reduce the rate of
interest is the love of one’s children and the desire to provide for their
good. (p. 52)
The next step in the evolution of this literature was the classical paper duo
by Yaari (1964, 1965). While still a doctoral candidate at Stanford University,
under the supervision of Nobel laureate Kenneth Arrow. Yaari was the first
economist to introduce annuities into the life-cycle model, which is com-
monly linked to Modigliani or Friedman. Yaari’s 1965 research paper is the
most widely cited research article in the life annuity economics literature.

70 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 70
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:43 PM
 The Scholarly Literature

In the 1960s, academic economists had not really given any thought
to how length-of-lifetime uncertainty—and the randomness of the length
of retirement, in particular—affects financial planning, saving, and invest-
ment behavior. Fisher and some others had some vague notions that old
age might make people (cranky and) impatient, but they had nothing con-
crete or formal.
Around the same time, modern portfolio theory, introduced by
Markowitz, was starting to catch on with academics. (It would be decades
before the idea reached Wall Street.) But even Markowitz, and his contem-
porary Sharpe, did not until recently address how the randomness of life
might affect economic behavior and portfolio construction. Two other eco-
nomic giants of the time, Friedman (1957) and Modigliani (1986), theo-
rized that consumers like to smooth their standard of living over time in
consideration of their lifetime resources. Neither of them said anything
about mortality and longevity. In most of their models and papers, people
died at a fixed and known time.
Yaari, writing his PhD at Stanford University in the early 1960s, started
his famous 1965 paper with the following words:
One need hardly be reminded that a consumer who makes plans for the
future must, in one way or another, take account of the fact that he does not
know how long he will live. Yet, few discussions of consumer allocation over
time give this problem due consideration. Alfred Marshall and Irving Fisher
were both aware of the uncertainty of survival, but for one reason or another
they did not expound on how a consumer might be expected to react to this
uncertainty if he is to behave rationally. (p. 137)
Yaari went on to describe how consumers would slowly spend down
their wealth in proportion to their survival probabilities and attitudes to
longevity risk and gradually reduce their standard of living—rationally. But
then, if you gave these same consumers the ability to purchase any type of
annuity desired, they would not have to reduce their standard of living with
age. They would, in fact, be able to hedge or insure against their longevity
risk. He then went one step further and derived the optimal “portfolio mix”
between regular market-based instruments (e.g., mutual funds) and their
actuarial counterparts (life annuities) as a function of an individual’s prefer-
ence for bequest versus consumption in his or her own lifetime. In modern
terms, he introduced what I call “product allocation” only a few years after
Markowitz introduced “asset allocation.” The Yaari paper has been cited
thousands of times by economic scholars in the 45 years since it was pub-
lished.27 Yaari, Sharpe, and Markowitz are all still alive.28
27
Quite justifiably, some people refer to Yaari as “the Harry Markowitz” of the annuity world.
28
Most recently, Yaari was president of the Israeli Academy of Sciences and Humanities.

©2013 The Research Foundation of CFA Institute  71

Milevxky.indb 71
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:43 PM
 Life Annuities

Figure 4 illustrates how one would rationally spend down wealth in a life-
cycle model, especially during the retirement years. Each of the four curves
represents a different level of longevity-risk aversion. The y-axis represents the
annual consumption rate, and the x-axis represents the age of the retiree. For
example, according to the graph, an individual with a very low longevity-risk
aversion (i.e., longevity-risk tolerant) would spend at a rate of $14 per $100
dollars in his first year of retirement. Then, as time passes, he would reduce his
consumption rate until the nest egg was depleted at age 90 and would live off
his pension of $5 per year. In contrast, someone with very high longevity-risk
aversion, which is the lowest of the four curves displayed, would start off spend-
ing much less in retirement—$10 per year—and she would reduce her spending
over time only very slowly, so she would still have liquid wealth and assets at the
age of 100. Figure 4 thus shows the essence of longevity-risk aversion: The fear
that you might live a very long time leads to you spending less as a result. I refer

Figure 4.  Life-Cycle Consumption during Retirement as a Function of

Longevity-Risk Aversion

Annual Consumption Rate ($ thousands)

14

13

12

11

10

4
65 68 70 73 75 78 80 83 85 88 90 93 95 98 100
Survival Probability

Longevity Risk Aversion, 1 (very low) Longevity Risk Aversion, 4 (high)

Longevity Risk Aversion, 2 (low) Longevity Risk Aversion, 8 (very high)

Source: Based on data from Milevsky and Huang (2011), p. 51.

72 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 72
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

the interested reader to Milevsky and Huang (2011) for more information on
how longevity-risk aversion affects spending rates in the presence of pension
and life annuity income.
To sum up, Yaari (1965) would have the individual reducing consump-
tion in proportion to his or her survival probability and then eventually deplet-
ing wealth and living off pension and annuity income alone. The greater the
amount of pension and annuity income, the earlier the person’s wealth would be
depleted. Similarly, the greater the person’s longevity-risk tolerance, the earlier
wealth would be depleted.
One of the most frequently quoted results attributed to Yaari (1965) is
that life annuities are not only an important component of a consumer’s port-
folio but should also actually form the entirety of the portfolio in the absence
of a bequest or legacy motive. He pointed out that the mortality credits are
simply too valuable to ignore. Yet, as noted earlier, few people actively choose
to annuitize any portion of their nest egg, much less all of it.
In a seminar that Yaari gave at the IFID Centre at the Fields Institute
in 2010 to commemorate his work in the area, he mentioned that his 1965
paper was originally intended to help resolve inconsistencies in neoclassical
economics and the apparent low spend-down rate of assets around retirement.
In that sense, his paper was intended as “positive” (to explain observed behav-
ior) as opposed to “normative” (to provide financial advice). In other words,
he never intended to write a manifesto on how people should behave in the
face of lifetime uncertainty—namely, that they should hedge longevity risk by
purchasing annuities. At the same time, he acknowledged that this model can
easily be inverted and used to offer guidance on how people should allocate
their assets around retirement.
The next influential paper in this literature was written by Hakansson
(1969). Echoing the work by Yaari (1964, 1965), Hakansson argued, “Any given
individual may be able to make himself better off both by the purchase of insur-
ance on his own life and the sale of insurance on the lives of others” (p. 444).
A further advance in the literature on life-cycle planning and lifetime
uncertainty was a 1973 paper by Fischer. He stated:
An individual who receives labor income is more likely to purchase insurance
than an individual who lives off the proceeds of his wealth. If insurance is fair,
then—in the consumption decision—future income is discounted at the safe
rate and weighted by the probability of being alive to receive it in reducing it to
comparability with wealth . . . An individual who lives off the proceeds of his
wealth is unlikely ever to purchase life insurance. An individual who receives
labor income is likely to purchase life insurance early in his life. In all simula-
tions the individual tends to sell life insurance late in life: Institutional rea-
sons why companies do not engage in such transactions exist. The purchase of

©2013 The Research Foundation of CFA Institute  73

Milevxky.indb 73
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 Life Annuities

annuities generates consumption and bequest patterns similar to those of the

simulations. Multiperiod term insurance either presents the individual with
arbitrage possibilities, or makes no difference to his welfare. (p. 148)
Another important paper in this vein is that of Richard (1975), who
embedded the Merton (1971) model into the Yaari (1964) model and stated:
A continuous time model for optimal consumption, portfolio and life insur-
ance rules, for an investor with an arbitrary but known distribution of life-
time, is derived as a generalization of the model by Merton (1971). The
investor is found to have a “human capital” component of wealth, which is
independent of his preferences and risky market opportunities and represents
the certainty equivalent of his future net (wage) earnings. (p. 187)
In other words, the existence of annuities and insurance allows for the
valuation of the human capital stream.
Barro and Friedman (1977) contrasted consumer choice under uncer-
tain lifetimes with the behavior that would arise if each individual’s lifetime
were announced at birth. In a model that included life insurance and excluded
investments in human capital, they found that
the expected utility under uncertain lifetimes exceeds that under known
lifetimes when the latter expectation is based on preannouncement sur-
vival probabilities. This conclusion emerges, first, because the model without
human capital contains no planning benefits from knowledge of the horizon
and, second, because the prior announcement of lifetimes forces risk-averse
consumers to undertake an extra gamble that they could otherwise avoid by
using life insurance. (p. 843)
Once again, the role of life annuities emerges in a life-cycle model by cre-
ating certainty out of horizon uncertainty.
Mirer (1979) was one of the first to document that retirees were not
spending down wealth at the rate you might expect from a life-cycle model.
They simply weren’t spending enough, possibly (given the random length of
life) because they were worried about outliving their wealth.
Davies (1981) investigated whether
the continued accumulation, or mild dissaving, observed among the retired
can be explained by uncertain lifetime. In the absence of annuities, after
an initial period influenced by borrowing constraints, under constant rela-
tive risk aversion, uncertain lifetime depresses consumption by a proportion
increasing with age if the elasticity of inter-temporal substitution in con-
sumption is “small.” Illustrative computations, based on actual income and
survival data, show that plausible elasticities are sufficiently small to give this
effect. The reduction in consumption is large enough to explain much of the
lack of decumulation by the elderly. (p. 561)

74 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 74
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

In other words, Davies believes that people who do not have access to life
annuities—or refuse to use them—need more money to finance their retire-
ment than they would if they had access to life annuities.
Kotlikoff and Spivak (1981) took a slightly different approach to the topic
by examining risk sharing in families as an alternative to annuities. They wrote:
Consumption and bequest-sharing arrangements within marriage and larger
families can substitute to a large extent for complete and fair annuity mar-
kets. In the absence of such public markets, individuals have strong economic
incentives to establish relationships which provide risk-mitigating oppor-
tunities. Within marriages and families there is a degree of trust, informa-
tion, and love which aids in the enforcement of risk-sharing agreements.
Our calculations indicate that pooling the risk of death can be an important
economic incentive for family formation; the paper also suggests that the
current instability in family arrangements may, to some extent, reflect recent
growth in pension and social security public annuities. (p. 388)
In other words, if you have a large enough family to help share the burden
and the risk, you might not need life annuities.29 This line of thinking was pur-
sued also in Kotlikoff, Shoven, and Spivak (1986).
Bernheim (1984) argued that
actuarial valuation of annuity benefit streams is theoretically inconsistent with
the assumption of pure life-cycle motives. Instead, we show that the simple
discounted value of future benefits (ignoring the possibility of death) is often
a good approximation to the relevant concept of value. This observation moti-
vates a re-examination of existing empirical evidence concerning the effects of
Social Security on personal savings, retirement, and the distribution of wealth,
as well as the proper computation of age–wealth profiles. (p. 1)
In a series of papers, Eckstein, Eichenbaum, and Peled (1985a, 1985b)
examined
the implications of the absence of complete annuity markets on the distribu-
tion of wealth and the welfare of agents who make savings decisions under
uncertainty regarding the length of their life . . . The absence of annuity
markets has, in addition to its other effects, potentially important implica-
tions for the equilibrium distribution of wealth. In particular, the existence of
annuity markets ensures a degenerate distribution of wealth across individu-
als.30 On the other hand, the absence of such markets results in the inequality
of wealth across members of the same generation . . . This inequality is not a
transient phenomenon; the unique steady state distribution of wealth is non-
degenerate. (1985b, p. 789)

29
And perhaps if you have enough annuity income, you might not bother acquiring a family.
30
In this context, “degenerate” means nonsmooth distribution.

©2013 The Research Foundation of CFA Institute  75

Milevxky.indb 75
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 Life Annuities

In a related paper, Eckstein, Eichenbaum, and Peled (1985a) explored the

implications of social security programs and annuity markets through which
agents, who are characterized by different distributions of length of lifetime,
share death-related risks. They found:
When annuity markets operate, a nondiscriminatory social security program
affects only the intragenerational allocation of resources. In the absence of pri-
vate information regarding individual survival probabilities, such a program
will lead to a nonoptimal intragenerational allocation of resources. However,
the presence of adverse selection considerations gives rise to a Pareto improv-
ing role for a mandatory nondiscriminatory social security program.31 (p. 303)
If a life-cycle model that assumes rational and optimizing behavior is to
have practical application, an extremely important question is, Are individuals
capable of formulating coherent expectations about their longevity and mor-
tality probabilities? Hamermesh (1985) addresses this point. He examined the
awareness of demographic changes by individuals as they projected their life
expectancies and survival probabilities. He studied whether their projections
were based on determinants that coincided with the evidence of epidemio-
logical and demographic studies. Hamermesh concluded:
Most important, I find that people do extrapolate changing life tables when
they determine their subjective horizons, and they are aware of levels of and
improvements within current life tables . . . They base their subjective life
expectancies disproportionately on their relatives’ longevity. (p. 404)
Along the same lines, Hurd (1989) found:
The consumption path is sensitive to variations in mortality rates, meaning
that mortality risk aversion is moderate and certainly much smaller than
what is typically assumed in the literature. The marginal utility of bequests
is small; therefore, desired bequests, which are estimated from model simu-
lations, are small on average. Apparently most bequests are accidental, the
result of uncertainty about the date of death. The parameter estimates imply
that although consumption and wealth paths may rise at early ages, eventu-
ally they will fall as mortality rates become large. (p. 779)
In an interesting paper that seems to contradict the spirit of the Yaari
(1965) result, Pecchenino and Pollard (1997) examined the effects of intro-
ducing actuarially fair annuity markets into a model of endogenous growth
with overlapping generations. They found:
The complete annuitisation of agents’ wealth is not, in general, dynamically
optimal; the degree of annuitisation that is dynamically optimal depends non-
monotonically on the expected length of retirement and on the pay-as-you-go
“Adverse selection” refers to the fact that annuitants live longer than the rest of the population.
31

A “nondiscriminatory social security program” is one in which redistribution is minimized.

76 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 76
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

social security tax rate. The government has an incentive to restrict the avail-
ability of actuarially fair annuity contracts, and can often move the economy
from a pay-as-you-go to a fully funded social security system via voluntary
contributions to a government sponsored, actuarially fair pension today accom-
panied by reductions in social security taxes tomorrow. (p. 26)
Clearly, introducing multiple (overlapping) generations into the life-cycle
model can actually overturn some of the established results regarding the opti-
mality of annuities in a model based on a single representative agent.
As I have argued in numerous places in this book, life annuities are a form
of pension. And a number of researchers in the life-cycle literature have built
on that idea. Sundaresan and Zapatero (1997) provide a framework linking
the valuation and asset allocation policies of defined benefit plans with the
lifetime marginal productivity schedule of the worker and the pension plan
formula. They stated:
Our model provides an explicit valuation formula for a stylized defined ben-
efits plan. The optimal asset allocation policies consist of the replicating port-
folio of the pension liabilities and the growth optimum portfolio independent
of the pension liabilities. We show that the worker will retire when the ratio
of pension benefits to current wages reaches a critical value which depends on
the parameters of the pension plan and the discount rate.32 (p. 631)
As I have stressed, individuals who are unwilling to trade the bequest
motive and liquidity in exchange for mortality credits will not value the life
annuity. This assertion can be shown rigorously in a life-cycle model. For
example, Jousten (2001) stated that “consumption is non-increasing in the lin-
ear bequest parameter for the simplest certainty case” (p. 149). He found the
same was not true for lifespan uncertainty. Jousten also studied the issue of
annuity valuation and found that “for a sufficiently strong bequest motive, the
true value of an annuity is equal to the actuarial value” (p. 149). In other words,
a standard fixed-income bond would be preferable for those individuals with
strong enough bequest motives.
It is not only the bequest motive that can affect annuitization in a life-
cycle framework. Babbel and Merrill (2007b) modeled individual behavior
“under the possibility of default by the insurer issuing annuities” (p. 1). They
found that even a little default risk can have a huge impact on annuity pur-
chase decisions. Furthermore, state insolvency guarantee programs can have a
big impact on the level of rational life annuity purchases.

Sundaresan and Zapatero (1997) thus set forth a kind of liability-based capital asset pricing
32

model (CAPM), a path taken by Waring (2004a, 2004b). In this literature, a life annuity (or,
more generally, a fixed-income instrument with cash flows matched to the investor’s liability) is
considered the risk-free asset, whereas in the original CAPM, cash is the risk-free asset.

©2013 The Research Foundation of CFA Institute  77

Milevxky.indb 77
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 Life Annuities

Yet another deterrent to annuitization in a life-cycle model would be

government pensions that, implicitly, already provide annuities. This situa-
tion might also affect retirement behavior. For example, Jiménez-Martín and
Sánchez Martín (2007) explored the effects of the minimum pension program
in Spain on welfare and retirement. They used data from the Spanish social
security system to estimate the behavioral parameters of the model and then
simulated the changes induced by the minimum pension in aggregate retire-
ment patterns. They found that the impact is substantial: “There is a threefold
increase in retirement at 60, the age of first entitlement, with respect to the
economy without minimum pensions, and total early retirement (before or at
60) is almost 50% larger” (p. 923).
Continuing on the relationship between social security pensions and life
annuities, Sheshinski (2007) demonstrated rigorously that “a public social
security system may be socially superior to private annuity markets” (p. 251).
Of course, the public social security system would have to be “large” enough to
fully cover even the wealthiest of individuals.
Now, because U.S. Social Security and the Canadian Pension Plan are
mandatory program, you might wonder how its compulsory nature affects
the welfare of retirees. Motivated by this issue, Gong and Webb (2008)
investigated the distributional consequences of mandatory annuitization.
Using the University of Michigan Health and Retirement Study data and
accounting for longevity-risk pooling within marriage and preannuitized
wealth, they found substantial redistribution away from disadvantaged
groups in expected-utility terms. They used a life-cycle model to calculate
the value each household would place on annuitization, based on the hus-
band’s and wife’s subjective life tables, and the household’s degree of risk
aversion and proportion of preannuitized wealth. Their conclusion is that “a
significant minority would perceive themselves as suffering a loss from man-
datory annuitization” (p. 1055).
A growing recent literature attempts to calibrate the life-cycle model to actual
data—for groups both with and without access to annuities—to examine whether
mandatory markets affect consumption. Hansen and İmrohoroĝlu (2008) stated:
In our calibrated model, if complete annuity markets exist, consumption
would increase over the entire life-cycle. When the annuity market is shut
down, consumption displays a hump shape where consumption peaks well
before retirement. Social security, because it substitutes for the missing annu-
ity market, causes consumption to continue increasing until after retire-
ment, although consumption still displays a hump shape. In particular, the
consumption profile displayed by our model peaks significantly later than
what has been estimated from U.S. consumption data. We find that these

78 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 78
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

conclusions are robust to the introduction of a bequest motive calibrated to

account for the fraction of wealth held by the elderly, although consumption
now peaks at a somewhat lower age. (p. 582)
Lachance (2012) analyzed a similar life-cycle model, one in which the
optimal wealth depletion time was derived and calibrated.
A study along the same lines is Finkelstein, Poterba, and Rothschild
(2009), in which the authors calibrated and solved a model of the U.K. com-
pulsory annuity market and examined the impact of gender-based pricing
restrictions. They wrote: “Our findings indicate the importance of endogenous
contract responses and illustrate the feasibility of using theoretical insurance
market equilibrium models for quantitative policy analysis” (p. 38). In other
words, much can be learned from the responses of individuals to a wide choice
of similar insurance policies covering differing risks.
Finally, McCarthy and Mitchell (2010), like Finkelstein et al. (2009),
focused on the role of adverse selection (the fact that annuitants live lon-
ger than the rest of the population) and examined equilibrium pricing and
demand implications. According to McCarthy and Mitchell, in the absence
of insurance company underwriting, adverse selection improves the mortal-
ity of annuity purchasers but worsens that of purchasers of other life insur-
ance products relative to the general population. (By “improve,” they obviously
do not mean that buying the annuity makes them healthier but, rather, that
the pool of buyers tends to be healthier when there is no underwriting.) They
explored the differences between mortality tables for this group in the United
States, the United Kingdom, and Japan. They found:
Adverse selection reduces mortality for both life insurance and annuities, con-
trary to what theory would suggest. This indicates that insurance company
screening of potential poorer risks in classic life insurance (“underwriting”) is
effective, possibly even eliminating any asymmetric information held by poli-
cyholders. (p. 120)
The common thread among the papers surveyed and discussed thus far
is their use of a life-cycle model of saving and consumption to investigate the
implications of length-of-life uncertainty, both with and without annuity mar-
kets. I now consider the second strand in the annuity literature, which has to do
with pricing.

Actuarial Pricing, Valuation, and Reserving

Possibly the first published paper to formally demonstrate how to price a life
annuity is Halley (1693)—yes, the Edmond Halley of comet fame. His paper
incorporated mortality into the time value of money. He wrote:

©2013 The Research Foundation of CFA Institute  79

Milevxky.indb 79
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 Life Annuities

It is plain that the purchaser ought to pay for only such a part of the value of
the annuity as he has chances that he is living; and this ought to be computed
yearly, and the sum of all those yearly values being added together, will amount
to the value of the annuity for the life of the person proposed. (p. 602)
More than 320 years of actuarial literature and models have flowed from
this statement.
Most observers trace the invention of the variable immediate annuity
(VIA) to Duncan (1952). His objective was to supplement fixed annuities,
which are susceptible to inflation, with annuities that would increase over
time. His proposal was to
provide pensions which will, insofar as possible, increase when prices are
high. Of course, this carries with it the probability of decrease when prices
are low. The purpose of this paper is to describe this novel type of cor-
poration, which presents some unusual and interesting actuarial problems.
A general explanation of the Fund will first be presented, followed by a
summary of the statistical data compiled in developing it. Finally, there is
given a detailed presentation of the actuarial plan to be used, involving the
accumulation units and annuity units on which the Fund’s annuities are to
be based. (p. 318)
The Gompertz annuity pricing model (GAPM), which was described in
Chapter 2, can be traced to Mereu (1962). He wrote that
the formula for evaluating annuities on a Makeham mortality table, should
be satisfactory for making calculations recognizing calendar year of birth
as a factor, provided the improvement in mortality is anticipated in such a
manner that the generation mortality table for each year of birth follows
Makeham’s Law. (p. 286)
A financial view of life annuity pricing is offered by Broverman (1986),
who examined aspects of the distribution of the internal rate for standard life
insurance and annuity contracts. Another important actuarial paper in the pric-
ing literature was written by Beekman and Fuelling (1990). They developed
a model for certain annuities that can be used when interest rates and future
lifetimes are random. A related actuarial paper is Frees, Carriere, and Valdez
(1996), who investigated the impact of dependent mortality models—often
called the “broken heart syndrome”—when valuing annuities. They found
that annuity values are reduced by approximately 5 when dependent mortality
models are used rather than the standard models that assume independence.
In other words, if market prices fully adjusted for this effect, married couples
would be able to obtain 5% more income than markets (and models) take into
account. Frees et al. used a GAPM, and Carriere, in particular, was an early
advocate of this model. In a related paper, Carriere (1999) wrote:

80 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 80
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

Using a no-arbitrage argument, the classical actuarial valuation formulas for

life insurance and annuities are consistent with no-arbitrage pricing, assum-
ing that the time of death is stochastically independent of the market prices
on bonds. (p. 339)
Moving on to the valuation of guarantees in variable annuity (VA) poli-
cies, which offer the option to annuitize, Milevsky and Promislow (2001)
argued, “The insurance company has essentially granted the policyholder an
option on two underlying stochastic variables: future interest rates and future
mortality rates” (p. 299). They developed a model in which both mortality and
interest rate risk can be hedged and the option to annuitize can be priced by
locating a replicating portfolio involving insurance, annuities, and default-free
bonds. Similar techniques were used by Milevsky and Posner (2001) to value
guaranteed minimum death benefit (GMDB) options in VAs. They wrote that
a simple return-of-premium death benefit is worth between one and ten
basis points, depending on gender, purchase age, and asset volatility. In con-
trast, the median Mortality and Expense risk charge for return-of-premium
variable annuities is 115 bps. In other words, consumers were overpaying for
this guarantee. (p. 93)
Mudavanhu and Zhuo (2002) argue that the lapse option, which is the
option to simply walk away from the VA and surrender its cash value, signifi-
cantly increases the value of the GMDB option. Both the lapse and the death
benefit options are much more valuable for middle-aged and older investors
than for younger investors because of the insurance fees.
Most of the subsequent research concluded that the embedded guaran-
tees are underpriced. Examples are Boyle and Hardy (2003) and Ballotta and
Haberman (2003). The guaranteed lifetime withdrawal benefit (GLWB) and
its close cousin, the guaranteed minimum withdrawal benefit (GMWB) with
a fixed-maturity horizon, was formally analyzed by Milevsky and Salisbury
(2006). The GMWB promises to return the entire initial investment, albeit
spread over an extended period of time, regardless of subsequent market per-
formance. The main result in Milevsky and Salisbury is that “the No Arbitrage
hedging cost of a GMWB ranges from 73–160 bps of assets. In contrast, most
products in the market only charge 30–45 bps” (p. 21). The authors concluded
their article by arguing that pricing—in 2005 and 2006, when the article was
written—was not sustainable and that GMWB fees would eventually have
to increase or product design would have to change to avoid blatant arbitrage
opportunities. This prediction did, in fact, come true as most large insurance
companies scaled back or completely withdrew from this market. In some
cases, no equilibrium fee will cover this risk.
Similar “underpricing” conclusions were reached by Chen, Vetzal, and
Forsyth (2008). They wrote:

©2013 The Research Foundation of CFA Institute  81

Milevxky.indb 81
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 Life Annuities

Only if several unrealistic modeling assumptions are made is it possible to

obtain GMWB fees in the same range as is normally charged. In all other
cases, it would appear that typical fees are not enough to cover the cost of
hedging these guarantees. (p. 165)
Related research was conducted by Dai, Kwok, and Zong (2008). They
used singular stochastic control methods for pricing variable annuities with
the GMWB and also examined optimal withdrawal policies for holders of the
VAs with the GMWB.
Shah and Bertsimas (2008) examined annuities with GLWBs and claimed,
“GLWB has insufficient price discrimination and is susceptible to adverse
selection . . . Valuations can vary substantially depending on which class of
model is used” (p. 1). Shah and Bertsimas echoed the works of Milevsky and
Salisbury (2006) and of Chen et al. (2008) by concluding that the product can
be challenging to hedge and should create concerns for the insurance compa-
nies offering the guarantees.
Another strand in the valuation, pricing, and hedging literature addresses
the concept of natural hedges and possible securitization of longevity risk
in life annuities and similar insurance products. Lin and Cox (2005) wrote,
“Securitization in the annuity and life insurance markets has been relatively
rare, but we have argued that this may change” (p. 247). In related work about
hedging longevity risk in life annuities, Cox and Lin (2007) argued, “The
values of life insurance and annuity liabilities move in opposite directions in
response to a change in the underlying mortality. Natural hedging utilizes this
to stabilize aggregate liability cash flows” (p. 1). This statement is more than
theoretical. The authors claimed to find empirical evidence suggesting that
annuity writing insurers who have more balanced business in life and annuity
risks also tend to charge lower premiums than otherwise similar insurers.
In a review paper on the analysis of longevity risk using stochastic model-
ing techniques from finance, Cairns, Blake, and Dowd (2008) offered a broad
review and considered the wide range of extrapolative stochastic mortality
models that have been proposed in the past 15–20 years. A number of models
that they considered are framed in discrete time and emphasize the statistical
aspects of modeling and forecasting. The 2008 paper by this trio is one of the
most comprehensive articles on the topic. In follow-up work, Dowd, Blake, and
Cairns (2011) proposed computationally efficient algorithms for quantifying the
impact of interest rate risk and longevity risk on the distribution of annuity val-
ues in the distant future. They made the argument that annuity values are likely
to rise considerably but are also uncertain. By “annuity value,” they meant the
actual cost of receiving $1 of income per year for life. Remember that the annu-
ity value is the inverse of the annuity payout. If annuity values are likely to rise,
according to the authors, then the monthly payouts (yields) are likely to decline.

82 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 82
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

Macdonald and McIvor (2010) fall into the category of predicting or fore-
casting future prices for life annuities but from a medical and biological per-
spective. They considered a number of gene variants that have been found to
affect longevity. Using an annuity pricing model, they found that possibly sig-
nificant uncertainty about annuity premiums may be overlooked if the stan-
dard errors of parameters estimated in medical studies are ignored by medical
underwriters. They concluded with some policy implications:
Such considerations may play an important part when the acceptability of
using a risk factor in underwriting is conditional on proof of its relevance
and reliability. This is the current position in respect of genetic information
in many countries, most prominently in the United Kingdom. (p. 1)
Related research was conducted by Kwon and Jones (2006). They
showed that “extended risk classification enables insurers to provide more
equitable life insurance and annuity benefits for individuals in different risk
classes and to manage mortality/longevity risk more efficiently” (p. 271).
Like Macdonald and McIvor (2010), they argued that “mortality differen-
tials resulted in a noticeable impact on actuarial values for different risk
classes” (p. 287).
Finally, for those interested in a textbook-level introduction to actuarial
pricing models in general and the valuation of more complicated life annui-
ties in particular, the two leading texts in this area are Promislow (2011) and
Dickson, Hardy, and Waters (2009). Both texts provide a comprehensive
introduction to actuarial mathematics covering deterministic and stochastic
models of life contingencies and covering advanced topics, such as risk theory,
credibility theory, and multistate models.

Optimal Product Allocation and Timing

The questions of (1) when to annuitize and (2) how much to allocate to a life
annuity were first addressed in the classic paper by Yaari (1965), who found
that in the absence of bequest motives and other market imperfections, 100%
of wealth should be annuitized, immediately. This paper was the first to offer
a recommended allocation to an insurance product—that is, “product alloca-
tion” in contrast to investment “asset allocation.” Numerous papers on optimal
allocations followed during the next 50 years, and I will do my best to survey
them in chronological order.
Buser and Smith (1983) wrote that
insuring against the loss of a claim on future earnings as a result of the wage-
earner’s death may be modeled as a portfolio problem in which the return on
a life insurance contract is negatively correlated with the return on the claim.

©2013 The Research Foundation of CFA Institute  83

Milevxky.indb 83
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 Life Annuities

The model yields a result which expresses the optimal amount of life insur-
ance in terms of two components: the value of the claim to be protected and
the investment characteristics of the insurance contract. (p. 147)
Their result is obviously quite general and could be related to both life
insurance and life annuities.
Sinha (1986) is another early paper to examine the impact of survival
probabilities, loadings, interest rates, and bequest motives on the demand for
life annuities in an optimal portfolio framework.
Brugiavini (1993) developed a model in which consumers have the option
to purchase annuities before learning their survival probability. The model
then allows consumers to recontract the initial choice after the resolution of
this form of uncertainty. Brugiavini shows that consumers purchase insurance
against their own survival probability type at a young age and then “do not
undertake further transactions” (p. 31). In this model, it is better to buy the
annuity early in life before you—and the insurance company—learn about
your health classification. These sorts of models assume that you can borrow
and lend at the same riskless rate during the entire life cycle. Although this
assumption might be somewhat unrealistic in practice, the important insight
from Brugiavini’s model is the optimality of buying annuities before the insur-
ance company suspects you are anti-selecting.
In contrast to this “buy them early” result, Yagi and Nishigaki (1993) focused
on the consumption aspect of annuities. In particular, they derived the demand
function for the annuities in the case in which the capital market is imperfect
and life annuities must be locked in for life and (usually) pay out a fixed or con-
stant income throughout the retirement period. They then proved that “the indi-
vidual holds assets not only in the form of actuarial notes, but also in the form
of monetary wealth” (p. 385). In other words, less than 100% is allocated to life
annuities—in contrast to the Yaari (1965) result—because of the irreversibility
of annuities and the inability to roll over differing amounts of actuarial notes.
On the topic of self-annuitization and the ability to beat the return from
the life annuity, a number of related research articles can be found. Khorasanee
(1996) considered two ways for a retiree to obtain a pension from a retire-
ment fund. The first is through the purchase of a life annuity providing a level
monetary income, and the second is through the withdrawal of income from
a fund invested in equities. He used deterministic and stochastic models to
assess the risks and benefits associated with each approach. In each case, the
projected cash flows were compared with those from a whole life annuity pro-
viding an income linked to price inflation. He concluded that
although each of the two options considered involves significant risks, each
method may be attractive to certain groups of pensioners, particularly those
with additional savings held outside the retirement fund. (p. 229)

84 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 84
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:44 PM
 The Scholarly Literature

Milevsky (1998) examined the do-it-yourself option and arrived at the

following conclusion: “In the current environment, a sixty-five-year-old
female (male) has a ninety percent (eighty-five percent) chance of beating the
rate of return from a life annuity, until age eighty” (p. 401).
The product allocation dimension was addressed by Kapur and Orszag
(1999). They derived the optimal investment decisions of an individual who
retires with a given level of assets and decides to invest in annuities and equities
to provide income in retirement. They found that the optimal portfolio decision
depends on risk aversion but, optimally, all individuals switch to annuities as they
age. Also, continuing to focus on the idea of an optimal annuitization date, they
found that the risk-adjusted losses from early annuitization can be significant.
Campbell, Cocco, Gomes, and Maenhout (2001) built a partial equilib-
rium life-cycle model calibrated to population parameters. They observed a
welfare gain equivalent to 3.7% of consumption from the investment of half
of retirement wealth into equities accompanied by a reduction in the U.S.
Social Security payroll tax rate to maintain the same average replacement
rate of income in retirement. In Cocco, Gomes, and Maenhout (2005), they
solved a realistically calibrated life-cycle model of consumption and portfo-
lio choice with nontradable labor income and borrowing constraints. They
then showed that a crucial determinant of borrowing capacity and portfolio
allocation is the lower bound for the income distribution. Essentially, both
of these studies “prove” that life annuities offered by a social security system
enable consumers to take on more financial risk than otherwise early in life.
And although this finding might seem intuitive or just plain obvious, it is
comforting to see that these insights can be embedded in a dynamic life-
cycle model of savings and consumption.
When self-annuitization is considered, the equity risk premium may exceed
the mortality credits—if only fixed annuities are available—but shortfall is always
a risk. This possibility was stressed in Albrecht and Maurer (2002). They wrote:
In comparison to private annuity products a self-annuitization strategy using
mutual fund withdrawal plans contains, in particular for high entry ages, a
substantial risk of outliving the individual’s wealth, as long as the benchmark
(the annuity) is based on a competitive investment return. (p. 284)
In terms of the optimal allocation in a VIA, Charupat and Milevsky
(2002) showed that for constant relative risk aversion (CRRA) preferences
and geometric Brownian motion dynamics, the optimal asset allocation dur-
ing the annuity decumulation (payout) phase is identical to that for the accu-
mulation (savings) phase. This finding is the classical Merton (1971) solution.
Blake, Cairns, and Dowd (2003) considered the choices available to a
member of a defined contribution (DC) pension plan at the time of retire-
ment for conversion of his or her pension fund into a stream of retirement

©2013 The Research Foundation of CFA Institute  85

Milevxky.indb 85
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

income. Their results suggest that the best distribution plan does not usually
involve a bequest but rather pays regular mortality credits to the plan member
in return for the residual fund reverting to the insurance company on the plan
member’s death. In other words, the best distribution plan is one in which
more consumption during life is traded for zero consumption at death. Such a
product is, of course, a life annuity with no period certain (PC) or death bene-
fit. But these authors did find that utility and welfare gains depend on the plan
member’s attitude toward risk. For highly risk-averse retirees, the appropriate
plan is a conventional life annuity. If the retiree has a stronger appetite for risk,
however, the optimal plan involves a mix of bonds and equities, with the opti-
mal mix depending on the plan member’s degree of risk aversion. Importantly,
they found that “the optimal age to annuitize depends on the bequest utility
and the investment performance of the fund during retirement” (p. 29). Once
again, the researchers decide there is an optimal age that depends on both per-
sonal preferences and economic variables.
This line of research was pursued by a number of researchers. For example,
Dushi and Webb (2004) used numerical optimization techniques to conclude
that it is optimal for couples to delay annuitization until they are aged 73–82
and, in some cases, never to annuitize. For single men and women, annuitizing
at substantially younger ages, between 65 and 70, is usually optimal. Households
that annuitize will generally wish to annuitize only part of their wealth.
Similarly, Gerrard, Haberman, and Vigna (2004) investigated the income
drawdown option and, looking for optimal investment strategies to be adopted
after retirement, allowed for periodic fixed withdrawals from the fund. Their main
conclusion is that “for a pensioner with a not too high risk aversion, the income
drawdown option should be preferred to immediate annuitization, adopting
optimal investment strategies with a sufficiently good risky asset” (p. 341).
One of the first normative product allocation models—one that actually
offered advice—involving annuities was developed in Chen and Milevsky
(2003). They offered a sort of separation theorem:
The first step of a well-balanced retirement plan is to locate a suitable mix of
risky and risk-free assets independently of their mortality contingent status.
Then, once a comfortable balance has been struck between risk and return,
the annuitization decision should be viewed as a second-step “overlay” that
is placed on top of the existing asset mix. And, depending on the strength of
bequest motives and subjective health assessments, the optimal annuitized
fraction will follow. (p. 71)
Although the Chen and Milevsky (2003) “separation” result is certainly
valid under the hypothetical conditions specified in their assumptions, you
could make an argument that once a life annuity is purchased and longev-
ity risk is hedged, the investor can afford to take on more investment risk. In

86 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 86
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

other words, the asset allocation can be tilted toward equity instead of bonds.
In fact, a paper by Milevsky and Kyrychenko (2008) indicated that in the con-
text of variable annuities, the presence of a life annuity put option does induce
greater risk taking in practice. The individuals who have more longevity insur-
ance within their variable annuities actually take on more investment risk.
In a fundamental extension and generalization of the classic Yaari (1965)
paper, Davidoff, Brown, and Diamond (2005) examined demand for life annui-
ties with market incompleteness. They found that some annuitization remains
optimal over a wide range of preference parameters but complete annuitiza-
tion does not. They also argued that utility need not satisfy the Von Neumann–
Morgenstern axioms and need not be additively separable for the Yaari (1965)
result to hold. Furthermore, annuities need not be actuarially fair; they only must
offer positive net premiums (i.e., mortality credits) over conventional assets.
In terms of the optimal timing of annuitization, Kingston and Thorp
(2005) wrote:
The desire to keep consumption above a specified floor creates an incentive to
annuitize earlier than otherwise. HARA (hyperbolic absolute risk aversion)
agents must maintain an escrow fund in the risk-free asset to cover future
subsistence, effectively shrinking the potential for wealth creation through
risky asset investment compared with CRRA (constant relative risk aversion)
agents, and making actuarially fair annuities more attractive. Secondly, diver-
gence between a retiree’s subjective assessment of their survival prospects and
the annuity provider’s objective assessment of their prospects will still add to
any delay. (p. 239)
A similar life-cycle approach was taken by Cairns, Blake, and Dowd
(2006), whose objective was to find the optimal dynamic asset allocation strat-
egy for a DC pension plan that takes into account the stochastic features of
the plan member’s lifetime salary progression and the stochastic properties
of the assets held in the pensioner’s accumulating pension fund. Shi (2008)
showed that in a DC plan, the freedom to optimally choose the annuitiza-
tion time can lead to an increase of certainty-equivalent wealth of up to 1.8%.
Thus, according to Shi, “The embedded annuitization option in the retirement
option value is of significant economic value to individuals” (p. 29).
In a comprehensive life-cycle model similar to that of Cairns et al. (2006),
Chen, Ibbotson, Milevsky, and Zhu (2006) developed a unified human capi-
tal–based framework to help individual investors with life insurance and asset
allocation decisions. The model provides several key results, including the fact
that investors need to make asset allocation decisions and life insurance deci-
sions jointly. In an expanded book by the same authors (Ibbotson, Milevsky,
Chen, and Zhu 2007), the authors showed how to integrate the entire personal
balance sheet into individual investors’ asset allocations through a systematic

©2013 The Research Foundation of CFA Institute  87

Milevxky.indb 87
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

joint analysis of (1) how much life insurance a family needs to protect human
capital and (2) how to allocate the family’s financial capital. They proposed a
life-cycle model that addresses the transition from the accumulation phase to
the saving phase and the role of immediate payout annuities. In the same vein
and spirit, Kaplan (2006) used simulation-based techniques to derive optimal
allocations to annuities.
Optimal Timing and the Option to Wait. Milevsky and Young
(2007a) were one of the first research teams to tackle the optimal timing of
annuitization within the framework of optimal stopping and stochastic con-
trol. They motivated their paper by arguing: “Due to adverse selection, acquir-
ing a lifetime payout annuity is an irreversible transaction that creates an
incentive to delay” (p. 3138). They then differentiated between all-or-nothing
situations and gradual situations. For the institutional all-or-nothing arrange-
ment, in which annuitization must take place at one distinct point in time
(e.g., retirement), they derived the optimal age at which to annuitize—namely,
the age at which the option to delay has zero time value. Then, for the more
general open-market structure, in which individuals can annuitize any fraction
of their wealth at any time, they located a general optimal annuity-purchasing
policy.
Their main conclusion is that an individual will initially annuitize a lump
sum and then buy additional annuities slowly. The idea of slow annuitization,
or a dollar-cost-averaging strategy, is also advocated and demonstrated in vari-
ous simulation-based studies, such as Soares and Warshawsky (2004). Milevsky
and Young (2007b) used preference-free dominance arguments to develop a
framework for locating the optimal age (time) at which a retiree should pur-
chase an irreversible life annuity. In this framework, the selection of time is a
function of current annuity prices and mortality tables. Then, using the insti-
tutional characteristics of annuity markets in the United States, Milevsky and
Young showed that annuitization prior to age 65 or 70 is dominated by tempo-
rary self-annuitization even in the absence of any bequest motives.
Along the same lines, but in a paper written for a more mathematical
audience, Stabile (2006) examined the optimal annuitization time and the
optimal consumption/investment strategies for a retired individual subject
to a constant force of mortality in an all-or-nothing framework. The author
showed that if the individual evaluates the consumption flow and the annu-
ity payment stream in the same way, then, depending on the parameters of
the economy, the annuity is purchased at retirement or never. The book by
Sheshinski (2008) offers readers a theoretical analysis of the functioning of
private annuity markets in a life-cycle model; the demand function for annui-
ties is derived, and various macroeconomic implications are examined. In a

88 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 88
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

follow-up paper, Sheshinski (2010) examined the implications of annuity tim-

ing and investigated some embedded options and options that offer partial
refunds in certain states of nature.
In a closely related paper written for a practitioner audience, Goodman
and Heller (2006) offered some caveats about the advice of delaying annui-
tization. They wrote: “If one tries to self-annuitize and draw down the same
level of income as payable under a life annuity, there will be more than a 50%
chance that he or she will run of out of funds while still alive” (p. 9). In other
words, the risk is substantial. The authors concluded that only a life annuity,
whether fixed or variable, provides the highest level of living income available
to a retired individual. In terms of the timing of annuitization, they stated that,
if no significant change in interest rates is expected, a 5-year delay from age
65 to age 70 results in about a 5% loss in future income. Furthermore, delay-
ing the start of life annuity income from age 65 to age 75, a 10-year delay, can
easily result in a 15% loss in future income. In their opinion, given the sharper
increase in mortality rates after age 70, “it pays to begin life annuity income no
later than at age 70” (p. 9).
Stevens (2009) examined the problem of systematic longevity risk, which
is the probability that hazard rates across different individuals might not be
independent, in which case the law of large numbers might lose effectiveness
in diversifying mortality risk. For example, a cure for cancer might reduce
mortality rates for the entire population, which would wreak havoc on such
pricing models as the GAPM. He claimed that because of the uncertainty in
the future prices of annuities, for an individual aged 65 to purchase an annu-
ity currently, instead of postponing the annuity purchase, is utility increas-
ing. This conclusion differs from much of the literature cited earlier. Stevens
claimed that this conclusion results from systematic longevity risk. He found
that it is optimal to purchase—at an earlier age than found in other research-
ers’ results—an annuity with a short deferral period.
In other words, not everyone agrees that delaying annuitizing to age 75 or
so is low risk. In yet another anti-delay advocacy piece, Dellinger (2011) argued:
“To the extent one’s objective is to maximize retirement income with the poten-
tial to keep pace with inflation while minimizing the probability of outliving
that income, delaying income annuity purchase can be suboptimal” (p. 1).
So, most authors agree that annuitization at some advanced age is
optimal—fully or partially—but researchers disagree about the exact age,
given the loss of liquidity and irreversibility. Interestingly, according to
Zeng (2010), the utility loss because of the irreversibility of a future life
annuity purchase is small for the retiree.

©2013 The Research Foundation of CFA Institute  89

Milevxky.indb 89
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

Devolder and Hainaut (2006) found that optimal consumption may be

split into two periods. During the first one, the budget constraint is inactive:
“An individual old enough and without any bequest motive should dedicate
a part of his wealth to purchase a life annuity” (p. 47). In a companion paper
(Hainaut and Devolder 2006), the authors went further and argued, “An inter-
esting observation is that optimal asset allocation still includes a life annuity if
the retiree wishes to pass on a bequest to his relatives” (p. 631).
In the continuous-time finance literature, the paper by Pliska and Ye
(2007) generalized the Yaari (1965) model to include risky assets, which is
similar to the work by Richard (1975). They modeled the optimal insurance
purchase and consumption under an uncertain lifetime for a wage earner
in a simple economic environment, successfully obtaining explicit solutions
in the case of CRRA utility. In a similar framework, Gupta and Li (2007)
found that high insurance charges can make the net return from the annuity
less than the return from other available investment assets—for example, the
risk-free asset. A related paper by Goda and Ramsay (2007) examined the
optimal guarantee—such as a PC in a life annuity—and showed, on the one
hand, that if the retiree’s bequest constant—which is the strength of retiree’s
utility of bequest, or the minimum inheritance he or she would like to leave
to heirs—is less than or equal to the lower bequest threshold, then a straight
life annuity without the guarantee period is optimal. On the other hand, if
the retiree’s bequest constant is greater than or equal to the upper bequest
threshold, then the maximum guarantee period is the best. Once again, the
intuition here is that if you want to leave something for the kids, make sure
to select a long PC.
A slightly different perspective on the important role of life annuities in
the optimal portfolio is offered by Babbel and Merrill (2007a). They wrote:
By covering at least basic expenses with lifetime income annuities, retirees
are able to focus on discretionary funds as a source for enjoyment. Locking
in basic expenses also means that the retiree’s discretionary funds can remain
invested in equities for a longer period of time, bringing the benefits of
historically higher returns that can stretch the useful life of those funds
even further. The key in all of this is to begin by covering all of the basic
living expenses with lifetime income annuities. Then, to provide for addi-
tional desirable consumption levels, you will want to annuitize a portion of
the remainder of your assets while making provisions for extra emergency
expenses and, if desired, a bequest. (p. 11)
Babbel and Merrill (2007a) provided strong advocacy for the role of life
annuities, and Babbel (2008) stressed the importance of life annuities for
females, who risk even lower standards of living than males in the absence of
such annuities. In yet another endorsement, according to Freedman (2008), life

90 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 90
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

annuities can be used to extend the left-hand (low-risk) side of the efficient
frontier. Reichenstein (2003), in a practitioner-oriented article, discussed the
pros and cons of annuitizing a portion of the retirement portfolio and presented
the by-now-familiar trade-off between reducing longevity risk and reducing the
amount of wealth available to beneficiaries. This issue is the single most impor-
tant trade-off involved in the decision to annuitize.
Horneff, with various co-authors, has written a series of papers geared
toward an academic audience on the optimal allocation and timing of annui-
tization: Horneff, Maurer, Mitchell, and Dus (2008); Horneff, Maurer,
Mitchell, and Stamos (2009); Horneff, Maurer, and Stamos (2008); Horneff,
Maurer, Mitchell, and Stamos (2010); and Horneff, Maurer, and Rogalla
(2010). In all of these papers, the authors used a utility-based framework to
measure the welfare gains from allowing a robust and differing set of dynamic
allocation strategies, including VIAs. The results, as I see them, are that the
optimal strategy is to purchase annuities during your working life and con-
tinue to shift wealth into annuities well into retirement and until the age of 80
or 85. The authors stated: “The investor who moves her money out of liquid
saving into survival-contingent assets gradually from middle age to retirement
and beyond, will enhance her welfare by as much as 50%” (Horneff, Maurer,
Mitchell, and Stamos 2009, p. 1688). The more relevant point is that they
found VIAs to have an important role to play in the optimal portfolio.
Recently, Kartashov, Maurer, Mitchell, and Rogalla (2011) used the same
techniques to examine variable investment-linked deferred annuities, which
offer both an investment element (in terms of a mutual fund–style subaccount)
and an insurance element (in terms of pooling longevity risks across the retiree
group). An earlier article offering similar suggestions about the role of variable
immediate—as well as real annuities—is Brown, Mitchell, and Poterba (2001).
Some research articles refer to annuities as (personal) longevity bonds and
then derive the demand conditions. An example is Menoncin (2008), who
demonstrated that the wealth invested in the longevity bond should be taken
from the ordinary bond and the riskless asset proportionally to the duration of
the two bonds. In other words, the funds to purchase a life annuity should be
obtained from the fixed-income portion of the investor’s portfolio.
As an alternative to annuitization, some researchers have proposed using
alternative investments, such as real-return inflation-linked bonds. An example of
this research is Shankar (2009), who proposed using Treasury Inflation-Protected
Securities (TIPS) and longevity insurance—also known as deferred income annu-
ities (DIAs) in some articles and as advanced life delayed annuities (ALDAs)
in others—that would guarantee real annual withdrawal rates in excess of 5%

©2013 The Research Foundation of CFA Institute  91

Milevxky.indb 91
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

without any risk of financial ruin. His strategy involves investing the retirement
savings in a combination of inflation-protected securities and longevity insurance
that would generate a predetermined, inflation-protected lifetime income stream.
Sexauer, Peskin, and Cassidy (2012) proposed a decumulation benchmark
comprising a laddered portfolio of TIPS for the first 20 years (consuming 88%
of available capital) and a deferred life annuity purchased with the remain-
ing 12%. This portfolio could be used directly by the investor (akin to index-
ing) or as a benchmark for evaluating the performance of a more aggressive
strategy. One of the motivating reasons for the “TIPS over annuity” sugges-
tion in the first 20 years of retirement is the concern about inflation risk that
real-return bonds can address; other concerns include credit risk and the lim-
ited benefit of mortality credits at younger ages (ages 65–85). In a follow-up
article, Sexauer and Siegel (2013) used the TIPS-plus-ALDA portfolio as the
(almost) riskless asset, or base case, in an overall financial planning framework
that spelled out, in language accessible to plan sponsors and human resource
officers, how (given wide flexibility in the savings rate) to accumulate a desired
level of guaranteed income for life.
In the product allocation literature, Pang and Warshawsky (2009), in com-
paring wealth management strategies for individuals in retirement, focused
on trade-offs regarding wealth creation and income security. They examined
a variety of strategies in a systematic and comprehensive manner. In a follow-
up paper, Pang and Warshawsky (2010) derived optimal equity/bond/annuity
portfolios for retired households that face stochastic capital market returns,
differential exposures to mortality risk, uncertain uninsured health expenses,
and differential social security and defined benefit (DB) pension coverage. In
both Pang and Warshawsky papers (2009, 2010), annuities play an important
role in the preferred strategies. Similarly, Koijen, Nijman, and Werker (2011)
studied the life-cycle consumption and portfolio choice problem while taking
into account annuity risk at retirement, and they also concluded that ignoring
annuity risk before and at retirement can be “economically costly” (p. 799).
Park (2011) used simulation techniques to examine how immediate and
deferred (longevity) annuities can affect probable income; he also took into
account long-term care risk. His results indicate that for a male retiring at
age 65 and facing investment and longevity risk who desires a 90% chance
of adequate retirement income with an immediate annuity could optimally
achieve that target by fully annuitizing his initial retirement wealth regardless
of different equity allocations in his portfolio. If this retiree is assumed to be
facing investment, longevity, and long-term care risk, however, he would need
to annuitize 80–90% (not 100%) of his initial retirement wealth; some portion

92 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 92
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

of his initial retirement wealth would have to be reserved to finance unex-

pected long-term care costs. This finding highlights a reason (and reminder)
to keep some liquid assets.
Continuing with the theme of health care risk, which life annuities have
difficulty addressing, Yogo (2011) developed a life-cycle model in which a
household faces stochastic health depreciation and must choose consumption,
health expenditure, and the allocation of its wealth between bonds, stocks, and
housing. The author calibrated a model to U.S. population data and showed
that the welfare gain from relaxing borrowing constraints on home equity is
5% of wealth at age 65. Similarly, the welfare gain from private annuitization
is 16% of wealth at age 65. Similarly, Peijnenburg, Nijman, and Werker (2012)
found that the timing of health cost risk is important. If out-of-pocket medi-
cal expenses can already be sizable early in retirement, empirically observed
low annuitization levels are optimal. If health cost risk early in retirement is
low, individuals would do better to save out of their annuity income to build a
buffer for health cost shocks at later ages.
The technical problem of when to annuitize or begin withdrawals from a
life annuity is intricately tied to the optimal timing of social security benefits.
Sun and Webb (2011) used the similarity to argue “that the optimal claim age
is between 67 and 70” (p. 907). They then used similar life-cycle models to
calculate that the amount by which benefits payable at suboptimal ages must
be increased so that a household is indifferent between claiming at those ages
and the optimal combination of ages can be as high as 19.0%.
In a recent contribution to the optimal timing literature, Di Giacinto and
Vigna (2012) proved that compulsory immediate annuitization is suboptimal
and the cost varies, depending on the risk aversion of the member, in rela-
tive terms between 6% and 40% of initial wealth. This paper—like a number
of others cited—supports making programmed withdrawals available as an
option to DC plan participants.
In an interesting paper that touches on the cost of illiquidity, Wang and
Young (2012) examined a world in which life annuities are cashable (that is,
they can be sold back to the issuer at fair market value). In such a model, irre-
versibility is no longer a concern. They solved for the optimal investment strat-
egy, optimal annuity purchase, and surrender strategies in this world. The paper
is technical and based on some hard-to-justify assumptions, but the findings
are interesting and thought provoking. They found that in such a model, indi-
viduals do annuitize more—no surprise—but actually surrender (or cash in)
in various circumstances. Like many of the papers mentioned, these results are
intuitive—perhaps even obvious if we consider that people are willing to pay
for something they are likely to use. The authors’ contribution to the literature
is the ability to build a formal model that can actually price these options.

©2013 The Research Foundation of CFA Institute  93

Milevxky.indb 93
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

Finally, in a paper that raises some doubt about the appropriate model for
dynamic life-cycle models of portfolio choice, including annuities, Charupat,
Kamstra, and Milevsky (2012) wrote:
Prices adjust gradually and over a period of several weeks and often months,
in response to certain—and not necessarily riskless—interest rate changes. In
particular, we find that changes to the 30-year U.S. mortgage rate provide a
better fit and indication of where annuity payouts are headed, compared to
the 10-year swap rate, for example. In addition, we find that the sensitivity to
interest rate changes (that is, annuity duration) is asymmetric. Annuity prices
react more rapidly and with greater sensitivity to an increase in the relevant
interest rate compared to a decrease. (p. 1)
In conclusion, more than 50 articles discuss the optimal timing of annu-
itization, and although authors, papers, and models provide different con-
clusions, the main result seems to be that at some advanced age—perhaps
as early as 60 or as late as 80—most consumers should have some of their
wealth in life annuities.

Defining and Solving the Annuity Puzzle

Given the overwhelming evidence in favor of annuitization documented in the
previous section, the reader may be surprised that so few individuals actively
purchase life annuities or purchase as few as they do. As I have pointed out,
the size of the voluntary life market is tiny relative to the much larger VA and
mutual fund market. The question is, why? It has been labeled the “annuity
puzzle” in the scholarly literature.
In my opinion, the first person to label this question as a formal puzzle, or
at least to puzzle over it, was Huebner (1927). In his book on insurance eco-
nomics, he wrote:
The prospect, amounting almost to a terror, of living too long makes neces-
sary the keeping of the entire principal intact to the end, so that as a final
wind-up, the savings of a lifetime, which the owner does not dare to enjoy,
will pass as an inheritance to others. In view of these facts, it is surprising that
so few have undertaken to enjoy without fear the fruits of the limited com-
petency they have succeeded in accumulating. This can be done only through
annuities . . . Why exist on $600, assuming 3% interest on $20,000, and then
live in fear, when $1,600 may be obtained annually at age 65, through an
annuity for all of life and minus all the fear . . . (p. 189)
To most researchers in this field, the annuity puzzle is more closely
associated with Modigliani (1986). In his Nobel Prize acceptance speech,
he stated: “It is a well-known fact that annuity contracts, other than in the

94 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 94
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

form of group insurance through pension systems, are extremely rare. Why
this should be so is a subject of considerable current interest. It is still ill-
understood” (p. 307).
The subsequent 25 years of scholarly literature have (1) attempted to solve
the puzzle, (2) made the puzzle even worse, or (3) claimed that the puzzle
does not exist. So, the literature is contradictory, but nevertheless, it is vast, and
growing. What follows are the key articles in this area.
Williams (1986) blamed high interest rates:
High interest rates decrease the demand for life annuities even if the poten-
tial annuitant does not believe that he or she could earn higher interest rates
than the pension administrator or individual life insurer uses to calculate the
lifetime income. Longer life expectances may also reduce the demand for
annuities, but their impact is weaker than the effect of higher interest rates.
If, as many persons believe, interest rates are not likely to return to earlier
lower levels in the near future, one would expect an increase in the demand at
retirement for lump sums under pension plans, individual deferred annuities,
and supplementary life insurance contracts. (p. 169)
Friedman and Warshawsky (1990) blamed the high loads and costs
embedded in annuities:
An explanation for this phenomenon is based either on the actuarially unfair
cost of annuities, importantly including the cost element arising from adverse
selection, or on the interaction of the unfair annuity cost and an intentional
bequest motive. (p. 152)
Bernheim (1991) blamed government social security:
I find that social security annuity benefits significantly raise life insurance
holdings and depress private annuity holdings among elderly individuals.
These patterns indicate that the typical household would choose to maintain
a positive fraction of its resources in bequeathable forms, even if insurance
markets were perfect. (p. 899)
Brown and Poterba (2000) blamed marriage:
The utility gain from annuitization is smaller for couples than for single indi-
viduals. Because most potential annuity buyers are married, this finding may
help to explain the limited size of the market for single premium annuities in
the United States. (p. 527)
Post, Gründl, and Schmeiser (2006) confirmed and reinforced the idea
that family risk sharing and the high loads on annuities are jointly to blame
for low levels of annuitization. Others have suggested that, although blaming
bequests and the desire for legacy might seem natural—because the annuity

©2013 The Research Foundation of CFA Institute  95

Milevxky.indb 95
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

is irreversible—these reasons might not be legitimate reasons for avoiding life

annuities. Brown (2001) wrote: “There is no evidence that bequest motives are
an important factor in making marginal annuity decisions” (p. 29).
Vidal-Meliá and Lejárraga-García (2006) concluded that few couples
would be willing to purchase life annuities once they had taken into account the
combined effects of market imperfections, the possibility of preexisting annui-
ties, and the bequest motive. Mottola and Utkus (2007) echoed these ideas:
It seems clear that there is a strong desire for married couples to de-annuitize
with many actively trying, by selecting products, to overcome the federally
mandated default of a joint-and-survivor annuity. (p. 8)
Purcal and Piggott (2008) wrote: “Results suggest that the bequest
motive is the strongest single deterrent to annuity purchase, followed by
social security” (p. 513).
Lockwood (2012) argued:
People with plausible bequest motives are likely to be better off not annuitiz-
ing any wealth at available rates. The evidence suggests that bequest motives
play a central role in limiting the demand for annuities. (p. 226)
Pashchenko (2010) examined a variety of explanations for low levels of
annuitization—most of them alluded to here—but then concluded, “Among
the traditional explanations, pre-annuitized wealth has the largest quantitative
contribution to the annuity puzzle” (p. 1).
Dushi and Webb (2004) continued with the idea of preannuitized wealth.
Using data from the Assets and Health Dynamics among the Oldest Old and
Health and Retirement Study panels, they concluded that “much of the failure
of the average currently retired household to annuitize can be attributed to
the exceptionally high proportions of the wealth of these cohorts that is pre-
annuitized” (p. 109). In other words, the preannuitized have too many annui-
ties already in the form of pensions and social security.
Bütler, Peijnenburg, and Staubli (2011) wrote:
Most industrialized countries provide a subsistence level consumption floor
in old age, usually in the form of means-tested benefits or income supple-
ments. The availability of such means-tested payments creates an incentive to
cash out (occupational) pension wealth for low and middle income earners,
instead of taking the annuity. (p. 1)
In an interesting forecast of future prospects for this market, Dushi
and Webb (2004) concluded: “We expect younger cohorts to have smaller
proportions of pre-annuitized wealth and project increasing demand for
annuitization as successive cohorts age” (p. 109). McCarthy and Mitchell
(2004), coming to similar conclusions, wrote that “the demand for annuities

96 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 96
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

is likely to rise in the future, driven by increased longevity, diminished pub-

lic and corporate pensions, and the availability of new annuity linked prod-
ucts” (p. 43).
Sinclair and Smetters (2004) argued that uncertain health expenditures
are to blame for the annuity puzzle. They wrote:
Individuals face a risk of health shocks which simultaneously cause large
uninsured expenses and shorten the life expectancy. The value of a life annu-
ity then decreases at the same time as the need for cash increases, undermin-
ing its effectiveness in providing financial security. When the risk of such a
health shock is substantial, it is no longer optimal for risk-averse individuals
with uncertain lifespans to hold all of their wealth in life annuity form, even
if annuity contracts are reversible, and bequest motives, transaction costs, and
adverse selection are absent. (p. 1)
Ameriks, Caplin, Laufer, and Van Nieuwerburgh (2011) also focused
attention on the importance of long-term care and medical risk. They wrote
that “demand for annuities would be far higher if they included some accept-
able form of long-term care insurance” (p. 520).
Not all countries experience the same low levels of voluntary annuiti-
zation as found in the United States and Canada. In Chile, for example, a
large number of retirees purchase additional annuities. James, Martinez, and
Iglesias (2006) documented that almost two-thirds of all retirees have annui-
tized, which is a high proportion compared with other countries. The authors
argued that the high rate of annuitization in Chile is the result of guarantees
and regulations that constrain payout choices, insure retirees through the min-
imum pension guarantee, eliminate other DB components, and give a com-
petitive advantage to insurance companies selling annuities.
For Switzerland, which offers a robust dataset for studying annuitiza-
tion, Bütler and Teppa (2007) empirically examined the levels of annui-
tization in DB pension plans. They concluded that low accumulation of
retirement assets is strongly associated with the choice of the lump sum
because of the availability of means-tested social assistance. They further
claimed that the sponsor’s default option is highly influential in the deci-
sion to annuitize.
Also writing about the Swiss annuity system, Avanzi (2010) claimed that
higher annuitization is observed because
these [factors] include annuitisation as a default choice, a high level of regu-
lation that fosters trust from the insured, a legal guarantee of benefits, the
absence of market risk borne by the insured, a favorable tax structure, sub-
stantial savings with flexible withdrawal options, home ownership, generosity
of benefits, as well as additional elements. (p. 155)

©2013 The Research Foundation of CFA Institute  97

Milevxky.indb 97
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

Continuing with the case of Switzerland, Bütler, Staubli, and Zito (forth-
coming) noted that in 2004, several Swiss insurance companies reduced the
conversion rate from wealth to income, which means that annuity factors
increased and the cost of guaranteed lifetime income increased. The authors
found that, as a consequence of the policy change, the fraction of individuals
choosing an annuity decreased by 16.8 percentage points.
Ganegoda and Bateman (2008) suggested that the thin and fading market
for life annuities in Australia might be the result of a “lack of consumer aware-
ness of the risks of not annuitizing” (p. 1).
In the United Kingdom, Inkmann, Lopes, and Michaelides (2011) set
out to
provide an in-depth empirical analysis of the characteristics of households
that participate (or not) in the U.K. voluntary annuity market. We document
that annuity demand increases . . . [as] financial wealth, education, and life
expectancy [increase], while it decreases [with increases] in pension income
and a possible bequest motive for surviving spouses. (p. 315)
In Italy, Cappelletti, Guazzarotti, and Tommasino (2011) found “a strong
demand for annuity products, at least with respect to the one that we observe
today, at current market prices” (p. 18).
One thing is for certain: There are substantial differences in annuitiza-
tion rates across countries and even within countries by various demographic
and socioeconomic groups. In an interesting experiment involving the U.S.
military, Warner and Pleeter (2001) studied a U.S. government program that
offered to more than 65,000 enlisted members separating from the military
(separatees) the choice between an annuity and a lump-sum payment. Despite
having been offered breakeven discount rates exceeding 17%, most of the sep-
aratees selected the lump sum, not the annuity. According to the authors, this
(irrationality) saved U.S. taxpayers $1.7 billion in costs because the annuity
was a much better deal.
In recent attempts to salvage the rational model, some have conjectured
that default and credit risk, rather than behavioral or bequest factors, might
be to blame for the annuity puzzle. For example, Jang, Koo, and Lee (2010)
claimed that “fear of the default risk of annuity providers may have hampered
growth of annuity markets” (p. 2). But using an equally sophisticated model,
Lopes and Michaelides (2007) suggested otherwise. Their model calculations
suggest that “a rare event [default] is unlikely to be the main explanation of
the annuity market participation puzzle” (p. 84). Schulze and Post (2010)
introduced an actuarial element to the discussion of the annuitization puzzle.
In their models, “consideration of aggregate mortality risk may alleviate, but
also intensify, the annuity puzzle” (p. 423).

98 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 98
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

Yaari without Strong Assumptions.  In a widely cited, comprehen-

sive, and influential paper, Davidoff, Brown, and Diamond (2005) argued:
While incomplete annuity markets may render annuitization of a large frac-
tion of wealth suboptimal, our simulation results show that this is not the
case even in a habit-based model that intentionally leads to a severe mis-
match between desired consumption and the single payout trajectory pro-
vided by an incomplete annuity market. These results suggest that lack of
annuity demand may arise from behavioral considerations, and that some
mandatory annuitization may be welfare increasing. It also suggests the
importance of behavioral modeling of annuity demand to understand the
equilibrium offerings of annuity assets. (p. 1589)
In other words, the culprit is behavioral.
This approach is echoed by Hurd, Panis, Smith, and Zissimopoulos (2004).
After extensive simulations based on the life-cycle model, they concluded
that “we need a new understanding of the motives for the lack of annuitiza-
tion, and possibly, a new theoretical structure” (p. 130). In a further reinforce-
ment of the puzzle, Chalmers and Reuter (2012), after examining the choice
between life annuities and lump sums made by 32,000 retiring public employ-
ees in the United States, found little evidence that retiree “demand for life
annuities rises when life annuity prices fall” (p. 1). In other words, even when
life annuities go “on sale,” consumers do not seem interested in them. In one
of the final nails into the rational coffin, Chalmers and Reuter found “strong
evidence that demand responds to recent equity returns” (p. 1). Thus, when
stock markets have gone up in the past few months, the demand for annuities
declines, and vice versa. Previtero (2011) presented evidence from corporate
(IBM) pension plan behavior that is consistent with employees extrapolating
from recent stock market returns. He argued that this myopic extrapolation
can result in a significant reduction in retirement wealth if, for example, indi-
viduals annuitize too early because of a market drop.
The Behavioral Angle. So, many recent researchers have been tak-
ing the behavioral angle. For example, Hu and Scott (2007) claimed: “Mental
accounting and loss aversion can explain the unpopularity of annuities by
framing them as risky gambles where potential losses loom larger than poten-
tial gains” (p. 71). Brown, Kling, Mullainathan, and Wrobel (2008) wrote, in
an important paper about the impact of framing, that
framing matters for annuitization decisions: In a consumption frame, annui-
ties are viewed as valuable insurance, whereas in an investment frame, the
annuity is a risky asset because the payoff depends on an uncertain date
of death. Survey evidence is consistent with our hypothesis that framing
matters: The vast majority of individuals prefer an annuity over alternative
products when presented in a consumption frame, whereas the majority of

©2013 The Research Foundation of CFA Institute  99

Milevxky.indb 99
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

individuals prefer non-annuitized products when presented in an investment

frame. To the extent that the investment frame is the dominant frame for
consumers making financial planning decisions for retirement, this finding
may help to explain why so few individuals annuitize. (p. 308)
Continuing the exact same reasoning, Brown (2009) wrote that “part of
the answer to why consumers are so reluctant to annuitize will probably be
found through a more rigorous study of the various psychological biases that
individuals bring to the annuity decision” (p. 1).
Bütler and Staubli (2011) wrote that “annuity payout choices are signifi-
cantly influenced by default options and peer effects” (p. 195).
Drilling down into the behavioral explanations, Gazzale and Walker
(2009) offered two plausible behavioral biases:
Our first hypothesis is a risk-ordering bias: retirees effectively overweight the
early risk (an early death) relative to the later risk (a longer-than-anticipated
retirement). Our second hypothesis is an endowment effect stemming from
loss aversion. (p. 21)
They went on to find support for these hypotheses in a laboratory setting
capturing many of the salient aspects of the annuity decision.
Also in the realm of behavioral explanations, Panis (2004) conducted an
extensive survey and found that
those with greater annuitization were more satisfied in retirement, and they
maintained their satisfaction throughout retirement. By contrast, retirees
without lifelong annuities have become somewhat less satisfied over the
years. The guaranteed income benefits may reduce anxiety about the risks of
outliving one’s savings and ending up in poverty. (p. 13)
In a recent NBER paper, Beshears, Choi, Laibson, Madrian, and Zeldes
(2012), reporting on a large-scale study in which individuals were given various
hypothetical annuitization choices, found that “allowing individuals to annui-
tize a fraction of their wealth increases annuitization relative to a situation
where annuitization is an all-or-nothing decision” (p. 1). Also, they empha-
sized the importance of framing. Moreover, although Beshears et al. (2012)
and related studies, such as Brown et al. (2008), do not report on real choices
by live people with actual money, presumably something can be gleaned from
how people respond in surveys to different framing of the same economic
situation. Indeed, most of the foundational studies in behavioral finance in
the 1980s and 1990s were assembled from similar surveys, experiments, and
hypotheticals.
Tax Treatment. Although not addressing the annuity puzzle directly,
Brown et al. (1999) examined the tax treatment of nonqualified annuities in
the United States and made a number of observations and suggestions to help

100 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 100
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 The Scholarly Literature

improve the efficiency of the market.33 They examined an alternative method of

taxing annuities that would avoid changing the fraction of the annuity payment
included in taxable income as the annuitant ages but would still raise the same
expected present discounted value of revenues as the current income tax rule. They
found “that a shift to a constant inclusion ratio increases the utility of annuitants
and that this increase is greater for more risk averse individuals” (p. 563). For a
discussion of income taxes in the United States and how they affect the relative
appeal of annuities versus fully taxable assets, see Babbel and Reddy 2009.
Brown and Poterba (2006) examined specifically the market for tax-
deferred variable annuities—which is much larger than the market for life
annuities—and identified two factors that have contributed to its success
and growth. The first is the opportunity for tax deferral, and the second is the
insurance features of variable annuities. They also found that VA ownership
strongly increases as income, wealth, age, and education increase.
Brunner and Pech (2008) offered a different perspective on the taxation of
life annuities. Considering a nonlinear tax on annuity payoffs, they found that
it can be used to correct the so-called distortion of the rate of return caused by
asymmetrical information.
In some countries and markets, income taxes create an additional incentive
(as opposed to barrier) to actual annuitization. Charupat and Milevsky (2001)
documented an intriguing tax arbitrage opportunity involving the lighter tax
treatment of life annuities in Canada. This opportunity might explain the (rel-
atively) higher demand for life annuities in Canada, per capita, than in the
U.S. market. But, of course, overall Canadian life annuity demand is still low
relative to what might be expected from a life-cycle model.
Policy Prescriptions. Some analysts have moved from posing and
solving the annuity puzzle to offering policy suggestions or prescriptions
on how to increase annuitization rates. Teppa (2011) offered the following
suggestion: “The annuitization puzzle may be alleviated by helping individ-
uals in better assessing their perceived longevity risk, rather than forcing
their actions” (p. 1). Direr (2010) suggested that “a minimal degree of flex-
ibility could well promote wealth annuitization by reducing the mismatch
between the desired consumption path and the annuity income stream” (p.
51). Scott, Watson, and Hu (2011) suggested product innovations as one
way to increase participation in the market. They claimed that product inno-
vation to concentrate on late-life payouts could improve participation. They
wrote: “The most promising area for large increases in (mortality credits) is
not lowering annuity costs but rather offering annuity products focused on
late-life payouts” (p. 238). This approach (known as an ALDA in Milevsky
For a discussion of income taxes in the United States and how they affect the relative appeal of
33

annuities versus fully taxable assets, see Babbel and Reddy (2009).

©2013 The Research Foundation of CFA Institute  101

Milevxky.indb 101
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:45 PM
 Life Annuities

2005a and DIA in industry jargon) was extensively analyzed and shown to
provide value in a life-cycle model by Gong and Webb (2010). Webb (2011)
offered similar policy suggestions.
Murray and Klugman (1990) suggested other innovations to improve
annuitization:
For older persons in relatively poor health, life annuities may not provide
a sufficiently high expected return to justify their use . . . [but] a market
should develop to provide underwritten life annuities to those with impaired
health. (p. 50)
In other words, one solution to the annuity puzzle (low annuitization
rates) would be to create an active market for impaired annuities sold to less
healthy annuitants.
Murtaugh, Spillman, and Warshawsky (2001) claimed that combin-
ing immediate annuities with long-term disability insurance would reduce
the cost of both and make them available to more individuals by reducing
adverse selection in the income annuity portion and minimizing the need
for medical underwriting for disability coverage. This approach is, implicitly,
another suggestion for improving annuitization rates, but Davidoff (2009)
questioned whether it would increase annuitization rates. His simulations
indicate that life annuities and long-term care insurance might be substi-
tutes rather than complements.
Creighton, Jin, Piggott, and Valdez (2005) analyzed the reasons for the
“failure of longevity insurance markets” (p. 417) and examined possible inno-
vations in both markets and public policy that might lead to a more vibrant
market, including pooled annuities that resemble a type of tontine, described
earlier. Mackenzie (2006) addressed the questions of whether annuitization
or other restrictions on distributions should be mandatory and, if so, whether
the provision of annuities should be privatized. Goldsticker (2007) proposed
a mutual fund/tontine hybrid vehicle. It would be a pooled fund serving as a
low-cost vehicle to provide annuity-like cash flows. Rotemberg (2009) pro-
posed a new instrument to be called a “mutual inheritance fund,” which would
be another tontine-like innovation. Other innovative solutions that might help
reduce the cost of providing life annuities, and thus spur demand, include the
“pooled annuity fund” proposal made by Piggott, Valdez, and Detzel (2005)
and by Bravo, Real, and da Silva (2009). The actuarial literature continues to
produce interesting innovations in this area.
Agnew, Anderson, Gerlach, and Szykman (2008) focused on the impor-
tant role of financial intermediaries and advisers in promoting and effecting
annuitization. In experiments, they found that women are more likely than
men to annuitize when offered actuarially fair annuities, which might be a
further indication that framing is important.

102 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 102
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 The Scholarly Literature

The report by Lieber (2010) and the report of the Council of Economic
Advisers (2012) discussed efforts by the Obama administration to promote
annuitization.
In a widely circulated proposal, Gale, Iwry, John, and Walker (2008)
argued that retirees should be given an opportunity to “test drive” a lifetime
income product, which would help retirees overcome existing biases, reframe
their view of lifetime-income products, and improve their ability to evaluate
their retirement distribution options. They proposed that a substantial por-
tion of assets in 401(k)s and other similar plans be automatically directed
(defaulted) into a two-year trial income product when retirees take distribu-
tions from their plans, unless they affirmatively choose not to participate.
They wrote:
Retirees would receive twenty-four consecutive monthly payments from the
automatic trial income plan. At the end of the trial period, retirees may elect
an alternative distribution option or, if they do nothing, be defaulted into a
permanent income distribution plan. (p. 3)
This suggestion has similarities to other default options in DC plans, such
as life-cycle funds and automatic savings plans.
Brown and Nijman (2011) offered similar suggestions, albeit to a
Dutch audience. Specifically, they suggested that, instead of compulsory
annuitization of all retirement wealth, individuals be required to annuitize
a minimum amount in a reliably inflation-indexed annuity and that some
additional amounts of annuitization be structured as automatic with an
opt-out provision.
Many other suggestions continue to be provided by practitioners, policy-
makers, and scholars on how to innovate in this market.
Yet, not all scholars agree that policymakers should actively increase annu-
itization. Feigenbaum and Gahramanov (2012) used a sophisticated overlap-
ping generations model to argue counterintuitively,
If households were to begin following the advice of most economists to annu-
itize, there would be short-term gains as households enjoy higher returns on
their savings. But later generations would be hurt as they stop receiving acci-
dental bequests. In the long run, everyone would be worse off. So we would
argue that policymakers should not implement measures intended to encour-
age annuitization. (p. 91)
In a similar paper using the rational life-cycle model with a more general
aversion to uncertainty, Bommier and Le Grand (2012) suggested, “A possible
reason for the low level of wealth annuitization may, therefore, simply be that
individuals are too risk averse to purchase annuities” (p. 28).

©2013 The Research Foundation of CFA Institute  103

Milevxky.indb 103
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Life Annuities

For those interested in reading more in this “life annuities are not necessar-
ily all good” literature, I suggest Fehr and Habermann (2008), who argued that
although young cohorts experience significant welfare gains, future generations
are hurt by lower bequest amounts. This is a different perspective, to say the least.34
I conclude the literature review of the annuity puzzle with a quote from
recent article by Benartzi, Previtero, and Thaler (2011):
The notion that consumers are simply not interested in annuities is clearly
false. Social Security remains a wildly popular federal program, and those
workers who still have defined benefit pension plans typically choose to retain
the annuity rather than switch to a lump-sum distribution. Furthermore,
when participants in defined benefit pension plans with built-in annuitized
payout are offered the opportunity to switch to a defined contribution plan,
most stick with what they have. The tiny market share of individual annuities
should not be viewed as an indicator of underlying preferences but as a con-
sequence of institutional factors about the availability and framing of annuity
options. (p. 161)
In a private conversation I had with Yaari, he stated that perhaps one of
the reasons many people did not appreciate the value of life annuities was
that personal tastes and preferences can change over time and they know it.
The current design of annuities might not allow retirees to adapt to changes
in their own tastes. These changes might be in legacy preferences or even for
spending more now versus later. One thing is certain, for those writing in
2012, the annuity puzzle is not as perplexing as it was 45 years ago.

The Money’s Worth Ratio around the World

As I explained in Chapter 2, one of the most important formulas in the life
annuity literature is the money’s worth ratio (MWR), which compares the
theoretical (fair) price of a life annuity with the actual market price. The higher
the ratio, the better the value from the life annuity. The MWR metric has been
used in many studies and across various countries. In this section, I will review
and summarize the key articles in this literature.
To my knowledge, the first paper to use the MWR in the context of life
annuities is Warshawsky (1988). He collected market annuity prices in the
United States over a period of almost 70 years and concluded:
Load factors on life annuities issued to 65-year-old males and females over
the period 1919 through 1984 have ranged from 10 cents to 29 cents per
dollar of actuarial present value. From 8 cents to 16 cents of these loads

In a telling comment along the same lines, Davidoff mentioned to me in conversation that,
34

precisely for this reason, the Davidoff, Brown, and Diamond (2005) American Economic Review
paper, Diamond selected the title “Annuities and Individual Welfare,” as opposed to using the
words “consumer” or “society” welfare.

104 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 104
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 The Scholarly Literature

represents the cost of adverse selection, and approximately 7.5 cents repre-
sents transaction costs. The cost of adverse selection increased during the
middle period of study, 1941–1962, on annuities sold to males, while the cost
of adverse selection continually declined for females. (p. 518)
The main purpose in computing the MWR in Warshawsky and many other
papers is to quantify the impact of adverse selection on the return from life
annuities.
A similar study was conducted by Friedman and Warshawsky (1990),
who concluded:
Expected yields offered on individual life annuities in the United States dur-
ing 1968–1983 were lower on average by 4.21–6.13 percent per annum or
2.43–4.35 percent per annum after allowing for adverse selection, than yields
on alternative long-term fixed-income investments. (p. 152)
In the same vein, Mitchell, Poterba, Warshawsky, and Brown (1999) con-
ducted an extensive study of annuity prices in the United States and found
that the average annuity policy available to a 65-year-old man in 1995 deliv-
ered payouts valued at between 80 cents and 85 cents per dollar of annuity pre-
mium. In other words, the MWR was much less than unity. They also found
substantial heterogeneity among annuity providers in the payouts per dollar of
premium payment and found that various companies offered prices that were
quite different from the average. They concluded that in the late 1990s, “from
the standpoint of potential purchasers, an individual annuity contract appears
to be a more attractive product today than 10 years ago” (p. 1316). Poterba
(2001) suggested: “Requiring all persons to annuitize their retirement account
balances at a specified age is one way to reduce the degree of adverse selec-
tion in the annuity market substantially” (p. 268). This suggestion is consistent
with Walliser (2000), who found that adverse selection caused by the fact that
annuitization is optional in most countries and jurisdictions increases annuity
prices by 7–10%.
In the U.S. market, Brown (2002) examined the impact on the MWR
of different mortality rates for different socioeconomic groups. He found
that during the payout phase of the annuity, mortality differences are also
important: “The MWR is lower for men than for women and for blacks than
for whites, and increases [with] an individual’s education level” (p.437). An
important conclusion from this paper is that if life annuities were manda-
tory and everyone paid the same price, there would be a substantial (exceed-
ing 20%) transfer of wealth from the shorter-lived group to the longer-lived
group. In related research, Carlson and Lord (1986) argued that any prohi-
bition in the use of gender as an insurance classification parameter is inde-
fensible, as it would also transfer wealth from shorter-lived groups (males) to
longer-lived groups (females).

©2013 The Research Foundation of CFA Institute  105

Milevxky.indb 105
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Life Annuities

In a wide-ranging international comparison, James and Vittas (2001),

from, respectively, the IMF and World Bank, examined the market pricing of
annuities and came to similar conclusions as Mitchell et al. (1999) regarding
the high value of annuities. They wrote:
Preliminary findings suggest that the cost of annuities is lower than might
be expected. When using the risk-free discount rate, MWRs of nominal
annuities based on annuitant mortality tables exceed 97% and even when
using population mortality tables they exceed 90%—neither the industry
commissions nor the effects of adverse selection appear to be as large as
anticipated. (p. i)
James and Song (2001), who conducted a similar international compari-
son, were careful to note that the MWR will depend on the particular assump-
tions used to value the theoretical annuity. They found that “when discounting
at the risk-free rate, MWRs for annuitants are surprisingly high—greater than
95% in most countries and sometimes greater than 100%” (p. 1).
Recall that the theoretical price of a life annuity, which is the denomina-
tor of the MWR calculation, involves assumptions regarding both mortality
and interest rates. The two effects are often difficult to disentangle in mar-
ket prices. For example, Mitchell and McCarthy (2002), using data from the
United Kingdom and the United States, claimed that
the relatively lower mortality among older Americans who purchase annui-
ties is equivalent to using a discount rate that is 50–100 bps below the U.K.
rate for compulsory annuitants or 10–20 bps lower than the U.K. rate for
voluntary annuitants. (p. 38)
Interestingly, James and Vittas (2001) did find that real (inflation-adjusted)
annuities—available in Chile, Israel, and the United Kingdom and now avail-
able in the United States (but rarely purchased) from a limited number of
insurance companies—have MWR values that are 7–9% lower than those of
nominal annuities. (For an examination of the pension systems—including the
role of annuities—in Australia, Chile, Denmark, Sweden, and Switzerland, see
Rocha, Vittas, and Rudolph 2011.)
The U.K. market is by far the largest life annuity market (by volume of
sales) in the world. Gunawardena, Hicks, and O’Neill (2008) showed that
the pension annuities market tripled in size between 1992 and 2007. In 2007,
premiums in the pension annuities market were more than £11 billion and
more than 400,000 contracts were sold. Demand for pension annuities is set
to rise further in the coming years because of a rise in the number of maturing
DC pensions. Cannon and Tonks (2009) computed the MWR of annuities
in the United Kingdom and found that, on average, the money’s worth over
the sample period for 65-year-old males was 90% and for 65-year-old females

106 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 106
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 The Scholarly Literature

was similar but slightly higher, 91%. Taking into account load factors associ-
ated with annuity contracts and making a comparison with other financial and
insurance products, this finding implies that annuities are fairly priced. Some
evidence indicates, however, that money’s worth has fallen since 2002. Cannon
and Tonks (2008) is another excellent source of information about life annui-
ties in general and the U.K. life annuity market in particular. For more about
the U.K. market, see also Telford, Browne, Collinge, Fulcher, Johnson, Little,
Lu, Nurse, Smith, and Zhang (2011).
Continuing with life annuities in the United Kingdom, Finkelstein and
Poterba (2002, 2004) found substantial evidence of adverse selection and ex post
mortality that was different for different socioeconomic levels and found that
the “money’s worth ratio increases with the length of the guarantee period”
(p. 45). An interesting finding—which is consistent with James and Vittas
(2001)—is that the MWR for an annuity product with a rising nominal payout
stream or an inflation-indexed payout stream was lower than that for a level
nominal product. The cause might be low demand, which reduces competition.
Brown, Mitchell, and Poterba (2002) found the same and wrote that “the mon-
ey’s worth ratio of nominal annuities exceeds the money’s worth of inflation-
indexed annuities both in the United States and in other countries” (p. 24).
Again, this phenomenon does not appear to be the case in Chile, which makes
it difficult to generalize. According to James, Martinez, and Iglesias (2006), the
MWR for indexed annuities in Chile is 98%. They suggest two reasons. First, in
Chile, indexed financial instruments in which insurance companies can invest
to hedge their risk are more widely available than in other countries. Second
(and a more convincing reason, I believe), the forced indexation requirements
eliminate adverse selection between nominal and real annuities.
For Australia, Ganegoda and Bateman (2008), using the MWR, found
that annuities represent poor value for money. They found that the MWRs of
Australian annuities are lower than international estimates and the total load-
ings are higher.
For Singapore, Fong, Mitchell, and Koh (2011) reported that the coun-
try’s Central Provident Fund, a national DC pension scheme, has mandated
annuitization of workers’ retirement assets and, as a result, the government-
offered annuities are estimated to provide MWRs exceeding unity.
Oddly enough—and consistent with the idea that each country and mar-
ket is quite different and segmented—James and Sane (2003) found that in
India, “unrealistically generous payouts with high money’s worth ratios far
exceeding 100 percent were offered until 2002.” They reported that “one par-
ticular company reduced rates by much more than warranted, leading to a
decline in MWRs to 90 percent, which was an increase in the load from less
than nothing to more than 10 percent” (p. 258).

©2013 The Research Foundation of CFA Institute  107

Milevxky.indb 107
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Life Annuities

Indeed, the process by which a market develops for life annuities is inter-
esting. The volume edited by Fornero and Luciano (2004), which includes
some of the previously mentioned studies, examined the evolution of the
European market in particular. The consensus appears to be that with the
aging of the population and the transition from DB to DC pension plans,
this trend will continue. This point is made by a variety of authors in the
Fornero and Luciano volume and by Cannon and Tonks (2005) in the con-
text of the United Kingdom.
Note that some notable problems arise when using the MWR for comput-
ing relative value and comparing markets and countries. This issue is emphasized
(in the context of Singapore) by Fong, Lemaire, and Tse (2011), who wrote:
It is necessary to consider the entire weighted distribution of annuity bene-
fits, instead of focusing exclusively on its expected value, the numerator of the
MWR metric. For instance, if the weighted discounted benefits are spread
over a large range of values, the overall financial attractiveness of annuities
may be less than what the MWR indicates. (p. 3)
This caveat is consistent with the evidence provided by Charupat, Kamstra,
and Milevsky (2012) that market annuity prices take time (often months) to
fully respond to changes in interest rates, which implies that a slice-in-time
calculation of MWR might be comparing today’s annuity price with yester-
day’s interest rate. See also Carson, Doran, and Dumm (2011), who examined
market discipline in the individual annuity market by measuring annuity con-
tract yields during the accumulation phase.
Finally, Rothschild (2009) looked back more than 200 years and examined
one of the oldest known datasets for evidence of adverse selection. Using data
from an 1808 Act of British Parliament that effectively opened a market for life
annuities, he found (not surprisingly) that even back in 1808, healthier people
purchased annuities and the less healthy and unhealthy stayed away from this
market. For more information about annuities and tontines from the Middle
Ages until the 20th century, the interested reader is referred to Jennings and
Trout (1982), Ransom and Sutch (1987), and Jennings, Swanson, and Trout
(1988). Given the widespread evidence over many centuries—offered by all of
these authors—that healthier individuals are the ones who purchase annui-
ties, Philipson and Becker (1998) made an interesting argument that, perhaps,
when introducing mortality-contingent claims into a life-cycle model, longev-
ity should be treated exogenously. Edwards (2012) provided more discussion of
the economics of lifespan variation.
In a clever historical study, Salm (2011) used changes in pension laws for
U.S. Union army veterans as a natural experiment to estimate the causal effect
of pensions and life annuities on longevity. Examining the effects of the pen-
sion laws of 1907 and 1912, which granted old-age pensions to Union army

108 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 108
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 The Scholarly Literature

veterans, he found that veteran pensions reduced mortality for both acute
and nonacute causes of death. So, the endogeneity of income and longevity-
contingent claims is not as farfetched as you might initially suspect. All of
these findings echo the famous Jane Austen quote from Sense and Sensibility
(published in 1811): “If you observe, people always live forever when there is
an annuity to be paid them.”

Other Institutional and Policy Literature

The final category in the life annuity literature is rather eclectic and a catchall
for articles that do not fall neatly into any of the other five categories.
In terms of the history of annuity pricing, annuity use, and annuity pop-
ularity, I recommend the early article by Kopf (1927) and the more recent
article by Lewin (2003). It is amazing how active the market for life annui-
ties was in the 17th and 18th centuries and interesting how involved famous
mathematicians and astronomers were in the development of pricing formu-
las. See, for example, the book by Bellhouse (2011) on the work of Abraham
de Moivre and the work of Leonhard Euler—the first, a statistician, and the
second, a mathematician of the first caliber—in the development of annuity
pricing models. Poterba (2005) wrote:
During a period of roughly three centuries, the major nation-states in
Europe relied substantially on the sale of life annuity contracts to finance
wars and other public expenditures. The nature of annuity products evolved
during this time, from simple contracts that paid the same amount to all
buyers, regardless of their age or gender, to more finely graded products that
more closely resemble modern private insurance annuities. Leading math-
ematicians of this period contributed to important advances in the pricing
of annuity contracts. The history of annuities also offers evidence of the role
of sophisticated speculation by private sector investors . . . Syndicates arose
to invest in annuity contracts when it was possible to profitably speculate
against the governments that were selling them. (p. 207)
Moving to institutional and regulatory matters, and fast-forwarding to the
twentieth century, Mehr (1958) discussed the regulation of VIAs and mused
about the proper agency to oversee the sale and distribution of these products.
The dilemma is whether to view them as securities or insurance or both. The
VIA itself was created in the 1950s and is described at length in Biggs (1969)
in the context of TIAA-CREF.
In the early 1990s, a number of researchers began viewing DB retirement
pensions as a form of longevity insurance, which was a term not used by earlier
researchers, such as Yaari (1965). First among them was Bodie (1990), who
wrote that “defined benefit pensions offer the most complete type of retire-
ment income insurance . . . [and] defined contribution pensions make sense

©2013 The Research Foundation of CFA Institute  109

Milevxky.indb 109
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Life Annuities

as a supplement” (p. 30). Blake (1999) wrote that because DC plans do not
offer longevity insurance, government has a role to play in helping develop
and expand the life annuity market. Blake was one of the first to suggest that
One key contribution of the government would be to supply long-term
instruments such as indexed bonds and survivor bonds that would enable
annuity providers to hedge risks that are beyond the resources and abilities
of private sector organizations to hedge effectively and economically. (p. 367)
The market for longevity-linked bonds has spawned its own growing lit-
erature. MacMinn, Brockett, and Blake (2006) is an excellent starting point.
Munnell, Golub-Sass, Soto, and Vitagliano (2007) reviewed the major
pension freezes during the period of 2005–2007 and explore the impact on
employees at different stages in their careers. Four possible explanations are
offered as to why employers are shutting down their plans: (1) to reduce
workers’ total compensation in the face of intense global competition; (2) to
maintain existing compensation levels in the face of growing health benefit
costs; (3) to avoid the market risk, longevity risk, and regulatory risk that
make DB pensions unattractive to employers; and (4) to reflect the fact that
traditional qualified pensions have become irrelevant to upper manage-
ment, who now receive virtually all their retirement benefits through non-
qualified plans. Whatever the reason, one thing is certain: As documented by
Drinkwater and Sondergeld (2004), “People are becoming less protected over
time from mortality risk, as evidenced by the decline in traditional pension
plan coverage” (p. 1).
Whether DC plans—even with generous matches and investment
returns—can ever provide the same benefits as DB pensions is debatable, as
is whether the private annuity market can replace DB pensions. Feldstein and
Ranguelova (2001) claimed that DC plans can provide the same expected
benefits as DB plans. Their analysis indicates that
the risk that future retirees would receive less in a pure defined-contribution
system, than this benchmark level of benefits, would be relatively small at
savings rates that would be substantially less than the future paygo [that is,
pay-as-you-go] tax rate that would be required to fund that benchmark level
of benefits. (p. 1116)
In other words, life annuities could be used to replicate the benefits of a
DB pension. For me, writing from the perspective of the year 2012, I wonder
whether the authors contemplated the dismal performance of the stock market
during the last decade and the current abnormally low level of interest rates.
For those who are interested in light and accessible but comprehensive
material on the U.S. market for life annuities, I recommend the Vanguard arti-
cle by Zahm and Ameriks (2011) or the report by Sass, Munnell, and Eschtruth

110 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 110
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 The Scholarly Literature

(2011). For those interested in more general aspects of aging and financial plan-
ning, see Weierich, Kensinger, Munnell, Sass, Dickerson, Wright, and Barrett
(2011). As far as the specifics of annuities are concerned, the books by Pechter
(2008) for individuals and by Olsen and Kitces (2009) for financial advisers are
two excellent references. Finally, for the most recent scholarly research, includ-
ing a number of papers by the authors reviewed here, see Mitchell, Piggott, and
Takayama (2011) or the recently published book by Warshawsky (2012), in
which a number of his research articles on life annuities are collected.

©2013 The Research Foundation of CFA Institute  111

Milevxky.indb 111
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 4. Conclusions and Final Thoughts

To conclude my review of the vast and growing subject of life annuities, I

want to point out a number of implications for practicing financial analysts
that are worth emphasizing. I will then turn to imagining the future for life
annuity products.

Final Takeaways of the Discussions

1. A life annuity can be viewed—and properly thought of—as a fixed-
income bond that pays monthly coupons without a fixed maturity value
or date. To the buyer, it looks like a portfolio of zero-coupon bonds
structured to provide constant payments as long the annuitant is still
alive. The periodic payments may be level, increasing at a predetermined
rate, or inflation indexed. Most importantly, the yield spread above the
risk-free rate is generated by the mortality credits embedded in the risk
pooling. Therefore, to replicate this enhanced yield by using conven-
tional traded instruments (e.g., regular bonds) is virtually impossible.
Moreover, for people at older ages, the implied longevity yield is almost
impossible to beat.
2. In general, the word “annuity” is a catchall term that does not really
mean anything until it is qualified with a proper label. Financial
economists, securities lawyers, insurance executives, and members of
the media often talk across each other and miss each other’s points
because they are referring to different products. For example, there are
equity-indexed annuities, tax-deferred annuities, variable annuities
(with and without guaranteed living benefits), fixed annuities, deferred
annuities, and, of course, fixed and variable immediate annuities. They
all have the word “annuity” in their titles, but few offer the raison d’être
of annuitization—that is, mortality credits.
3. Therefore, financial analysts and wealth managers must ensure that they
understand which kind of annuity they are actually looking at before they
decide to dismiss it or include it as part of a client’s retirement portfolio.
Even the best low-cost variable and fixed immediate annuities (i.e., those
that offer pure mortality credits) can be watered down if (1) guarantees,
(2) period certain (PC), or (3) refund options are added on, which are
unnecessary but often added to make the annuity product palatable to the
loss-averse retiree.

112 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 112
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Conclusions and Final Thoughts

4. A useful way to think of the benefits of a life annuity is as follows:

Imagine that you and a retired neighbor both invest $500 in a money
market account, with the macabre proviso that the account can be
cashed-in only when one of you dies. The survivor gets the entire
$1,000 plus any interest accrued, while the family of the deceased
inherits nothing. (You will recognize that this arrangement is a ton-
tine.) Now, assuming you are the survivor, your terminal investment
return on the $500—whatever and whenever that might be—will far
exceed the investment return from conventional stocks or bonds dur-
ing that period, even though the actual money was invested in cash. Of
course, the key to the supercharged return from cash is that you have
to survive to claim the mortality credits and assets of your neighbor.
For the millions of Baby Boomers retiring on a meager pension and a
depleted nest egg, however, this longevity-contingent claim is likely to
be the best hedge for their longevity risk. It is asset/liability manage-
ment on the personal balance sheet.
5. Longevity-risk aversion is distinct from f inancial-risk aversion.
Longevity-risk aversion is about the fear of living longer than expected
and having to reduce your standard of living in retirement as a result.
Individuals who are longevity-risk averse will probably consume less
of their wealth early in retirement and allocate more of their nest egg
to annuity products to protect against this risk. This characteristic is
akin to savers who are financial-risk averse allocating more of their
wealth to safer assets, such as bonds. Conceivably, those individuals
who are financial-risk averse are also likely to be longevity-risk averse.
In other words, counseling a retiree to buy more stocks because the
person could live to be a centenarian might be internally inconsistent,
at best, and an oxymoron at worst. Those who fear living a long time
should own annuities. Period.
6. Most financial, public, and insurance economists would agree (some-
thing that is rare) that life annuities, longevity insurance, and guaran-
teed pensions have an important role to play in the optimal retirement
portfolio. Noted economists—such as Brown, Mitchell, Poterba, and
Warshawsky (2001)—whose works have been described in this book
are only a few of those who have written extensively on the impor-
tance and role of these products in financing retirement. The debate
in the literature tends to be around (1) the optimal age, (2) the opti-
mal amount, and (3) the optimal type of product. Notwithstanding, all
of these researchers agree that life annuities are a legitimate and core
product for the optimal retirement portfolio.

©2013 The Research Foundation of CFA Institute  113

Milevxky.indb 113
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Life Annuities

7. The fact that life annuities are priced in a competitive market to

account for healthier, longer-lived individuals implies that an adverse
selection cost is built into these insurance products. It is not a mark-up
or loading, per se, but a reflection of the clientele who are interested
in acquiring life annuities. Nevertheless, buying annuities as part of a
group—or perhaps making annuities mandatory for a portion of an
individual’s retirement account—would reduce the cost to everyone.
If you can buy any insurance product in wholesale bulk as opposed to
individual retail, you will save for two reasons. First, some fixed costs
will be reduced, and second, and more importantly, the adverse selec-
tion costs are reduced.
8. Naturally, some individuals do not need any additional life annuities
because they are already sufficiently annuitized or overannuitized. For
example, anyone with a DB pension plan from an employer already has
a substantial portion of wealth preannuitized. If we add to this annuity
social security benefits—which can add up to a $30,000 real, or inflation-
adjusted, annuity per individual—clearly many retirees do not need any
more life annuity income. Moreover, if they have strong bequest motives,
their optimal (additional) allocations to longevity-contingent claims
should be close to zero. For wealthy, high-net-worth individuals for whom
social security provides only a tiny fraction of their cash flow needs in
retirement but who are not so wealthy that they can afford the legacy and
bequest motives they assert, life annuities are an important class of prod-
ucts for them to consider.
9. Those who delay claiming U.S. Social Security (or Canadian Pension
Plan) income to the latest age possible are effectively buying a real
(inflation-adjusted) advanced-life delayed annuity, with a survivor ben-
efit for the spouse. The implied longevity yield from such a strategy far
exceeds the rate of return available from real or nominal bonds in today’s
environment of ultra-low interest rates, especially for people in better-
than-average health. For them, delaying annuitization is optimal.
10. There is nothing unique or special about fixed immediate life annuities.
The best way to think of life annuities is as a mortality credit overlay
to a conventional asset allocation. And although most life annuities are
currently bond backed (and are thus effectively a fixed-income prod-
uct with some additional mortality credits), the underlying assets could
be stocks, cash, or even alternative assets. Variable immediate annuities
(VIAs), which are even more rarely purchased than fixed immediate
annuities, are an example. Figure 5 provides a graphical illustration of
this overlay concept.

114 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 114
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:46 PM
 Conclusions and Final Thoughts

Figure 5.  The Overlay Concept: Any Asset Class Can Be

Annuitized

Bonds

Life Annuity and

Stocks Mortality Credit Overlay Cash

Alternatives

11. In the past decade or so, a number of substitutes for—and competitors

with—life annuities have emerged in the form of guaranteed living benefits
attached to variable annuities. The early versions of these products (2002–
2006) were grossly underpriced and extremely generous, according to most
researchers who carefully examined the pricing. Thus, despite the common
perception that annuities were exorbitant, the embedded (put) options were
not priced high enough, especially for those who knew how to optimize the
value of these guarantees. Numerous insurance companies came close to the
precipice as a result of offering these guarantees. Thus, in recent years, ratio-
nalization has occurred in pricing and features. Many companies, given the
extreme difficulty in hedging the embedded options, most of which have
maturities of 30 years or more, have withdrawn from this market altogether.
12. Behavioral evidence is growing that retirees (and seniors) who are receiv-
ing life annuity income are happier and more content with their financial
condition in retirement than those receiving equivalent levels of income
from other (fully liquid) sources, such as dividends, interest, and system-
atic withdrawal plans. Indeed, with growing concerns about dementia
and Alzheimer’s disease in an aging population, automatizing the retiree’s
income stream at the highest possible level—which is partly what a pension
life annuity is all about—will become exceedingly important and valuable.
13. Credit risk, illiquidity, and low interest rates are three concerns that
often are expressed by potential annuitants. Yet, all three concerns do
not quite add up to an excuse for complete nonannuitization. The credit

©2013 The Research Foundation of CFA Institute  115

Milevxky.indb 115
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

risk is mitigated by state guarantee funds. Annuitizing only a portion of

your portfolio, for example, can solve the concern regarding liquidity and
access to cash in the event of a medical emergency. In addition, fears about
interest rates (what if they move up suddenly tomorrow?) apply to any
fixed-income instrument, not only annuities. Conservatively investing in
medium-term and long-term bonds while waiting for annuity rates to
increase might seem like the punch line of a joke for financial economists,
but this strategy is currently being implemented by many retirees.
14. As North American Baby Boomers march toward the random ends of
their life cycles, this relatively small market is likely to grow at much
higher rates than it has in the past. The questions are, (1) What will life
annuities look like in the future (2) and how will the features be framed to
make life annuities more appealing than today?

Imagining the Life Annuity of 2020

Perhaps the next step in the evolution of retirement annuities will be a return
to the past. Maybe the annuity of the year 2020 will pay out no cash at all but,
instead, offer an actual retirement service. Allow me to explain.
One of the lesser-known facts about retirement annuities is that, despite
their illustrious 2,500-year history, only in the past few hundred years have their
benefits been paid in cash. For the first few thousand years, retirement annuities
provided something more reliable and useful than nominal cash or coin: They
paid out in units of service. For example, as I mentioned, one of the first known
retirement annuities was documented in the Bible in 2 Kings 25:27–38:
And it came to pass that the King of Babylon did lift up the head of the King
of Judah out of prison . . . And he spoke kindly to him, and he did eat bread
continually before him all the days of his life. And his allowance was a daily
rate for every day, all the days of his life.
Thus, according to most insurance historians, one of the first retirement
annuities ever was paid out in units of dinner.
During the Middle Ages, wealthy landowners and merchants purchased
corrodes from monasteries and abbeys, which provided them with food, cloth-
ing, and often shelter for the rest of their lives. They had wealth; what they
wanted was services.
The English poet and author Geoffrey Chaucer (1343–1400), at the young
age of 35, so enthralled King Edward III that the king granted him a unique
annuity—one gallon of wine daily for the rest of his life, to be served in the
port of London. (That would be 16 glasses of wine per day, which gives me
new respect for the fact that Canterbury Tales got written at all.)

116 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 116
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Conclusions and Final Thoughts

Only in the 16th century did it become commonplace for retirement

annuities to be paid and received exclusively in cash. Until then, you pur-
chased the retirement annuity with cash but the benefit itself was denomi-
nated in units of consumption—immune to inflation and the risk of a debased
currency. Perhaps the time has come to consider offering such annuities again.
Consider the following: A few years ago, the U.S. Postal Service
(USPS)—an institution that itself might not be around in 2020—started
offering “Forever Stamps.” Although the cost of mailing a first-class letter
in the United States is currently 46 cents and the price has been steadily
increasing by approximately 1 cent per year, a Forever Stamp comes with no
fixed monetary value. The stamp—effectively a financial derivative—entitles
the holder to one unit of service, in perpetuity. The holder can use the stamp
to mail one first-class (38-gram) letter anywhere in the United States forever.
Pay the 46 cents today and, regardless of what it will cost to mail a first-class
letter next year or 10 years hence, the cost of the service is locked in.
For the USPS, this offer is not as big a gamble as you might expect. First,
by not having to print new and costly stamps every time the price goes up, the
USPS saves in printing costs for the millions of people who need one- and
two-cent stamps to make up the difference. At the same time, the USPS gets
to keep clients’ money—money it needs—while clients keep the stamps in
their drawers for the next decade. All parties are winners, which is critical for
true financial innovation to flourish. Other countries—Canada, New Zealand,
and Singapore—offer similar stamps.
Now, think of a retirement annuity for which the benefit is modeled on a
“forever” service. It might cover your water, gas, or electric bill for life. Maybe it
would be denominated in units of days in a five-star nursing home or units of
Lipitor, Fosamax, and Plavix. Or perhaps it is a glass of wine a day (a Chaucer
mini-annuity).
The bottom line is that money is merely a medium of exchange. Even if
you manage to find an entity that will guarantee you an income of $1,000 or
$10,000 per month for the rest of your life—and you actually trust that the
entity will be around for the rest of your life—you still run the risk that those
dollars will not be enough to purchase the services you really want. In the lan-
guage of finance, you are not only incurring credit risk and inflation risk, you
are taking on basis risk—that is, a mismatch between the retirement assets you
own and the liabilities that you face.
In today’s interest rate environment, perhaps the hunt for yield should be
abandoned in favor of the hunt for institutions that can guarantee the services
an aging population will need to sustain itself. Would you buy a retirement

©2013 The Research Foundation of CFA Institute  117

Milevxky.indb 117
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

service annuity from a utility company? How about from a pharmaceutical

company? An oil and gas company? Do you trust them—or like them—any
less than your local insurance company?
Sure, these companies would have to contend with a host of regulatory
issues if they suddenly jumped into competition with century-old insurance
companies and pension funds. But I believe it is high time for someone to be
disruptively innovative in the retirement income space.
How about another glass of Merlot while we wait?

118 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 118
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography
Agnew, Julie R., Lisa R. Anderson, Jeffrey R. Gerlach, and Lisa R. Szykman. 2008.
“Who Chooses Annuities? An Experimental Investigation of the Role of Gender,
Framing, and Defaults.” American Economic Review, vol. 98, no. 2 (May):418–422.
Albrecht, Peter, and Raimond Maurer. 2002. “Self-Annuitization, Consumption
Shortfall in Retirement and Asset Allocation: The Annuity Benchmark.” Journal of
Pension Economics and Finance, vol. 1, no. 3 (November):269–288.
Ameriks, John, Andrew Caplin, Steven Laufer, and Stijn Van Nieuwerburgh. 2011.
“The Joy of Giving or Assisted Living? Using Strategic Surveys to Separate Public Care
Aversion from Bequest Motives.” Journal of Finance, vol. 66, no. 2 (April):519–561.
Avanzi, Benjamin. 2010. “What Is It That Makes the Swiss Annuitise? A Description
of the Swiss Retirement System.” Australian Actuarial Journal, vol. 16, no. 2:135–162.
Babbel, David F. 2008. “Lifetime Income for Women: A Financial Economist’s
Perspective.” Policy Brief, Personal Finance, Wharton Financial Institutions Center.
Babbel, David F., and Craig B. Merrill. 2007a. “Investing Your Lump Sum at
Retirement.” Policy Brief, Personal Finance, Wharton Financial Institutions Center.
———. 2007b. “Rational Decumulation.” Working Paper No. 06-14, Wharton
Financial Institutions Center.
Babbel, David F., and Ravi Reddy. 2009. “Measuring the Tax Benefit of a Tax-Deferred
Annuity.” Journal of Financial Planning, vol. 22, no. 10 (October):68–83.
Ballotta, Laura, and Steven Haberman. 2003. “Valuation of Guaranteed Annuity
Conversion Options.” Insurance: Mathematics and Economics, vol. 33, no. 1
(August):87–108.
Barro, Robert J., and James W. Friedman. 1977. “On Uncertain Lifetimes.” Journal of
Political Economy, vol. 85, no. 4 (August):843–849.
Beekman, John A., and Clinton P. Fuelling. 1990. “Interest and Mortality
Randomness in Some Annuities.” Insurance: Mathematics and Economics, vol. 9, no. 2–3
(September):185–196.
Bellhouse, D.R. 2011. Abraham De Moivre: Setting the Stage for Classical Probability
and Its Applications. Boca Raton, FL: CRC Press.
Benartzi, Shlomo, Alessandro Previtero, and Richard H. Thaler. 2011. “Annuitization
Puzzles.” Journal of Economic Perspectives, vol. 25, no. 4 (Fall):143–164.
Bernheim, B. Douglas. 1984. “Life Cycle Annuity Valuation.” NBER Working Paper
No. 1511 (December).

©2013 The Research Foundation of CFA Institute  119

Milevxky.indb 119
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

———. 1991. “How Strong Are Bequest Motives? Evidence Based on Estimates of
the Demand for Life Insurance and Annuities.” Journal of Political Economy, vol. 99,
no. 5 (October):899–927.
Beshears, John, James J. Choi, David Laibson, Brigitte C. Madrian, and Stephen P.
Zeldes. 2012. “What Makes Annuitization More Appealing?” NBER Working Paper
No. 18575 (November).
Biggs, John H. 1969. “Alternatives in Variable Annuity Benefit Designs.” Transactions
of Society of Actuaries, Part 1, vol. 21, no. 61:495–517.
Blake, David. 1999. “Annuity Markets: Problems and Solutions.” Geneva Papers on
Risk and Insurance, vol. 24, no. 3:358–375.
Blake, David, Andrew J.G. Cairns, and Kevin Dowd. 2003. “Pensionmetrics  2:
Stochastic Pension Plan Design during the Distribution Phase.” Insurance:
Mathematics and Economics, vol. 33, no. 1 (August):29–47.
Bodie, Zvi. 1990. “Pensions as Retirement Income Insurance.” Journal of Economic
Literature, vol. 28, no. 1 (March):28–49.
Bommier, Antoine, and François Le Grand. 2012. “Too Risk Averse to Purchase
Insurance? A Theoretical Glance at the Annuity Puzzle.” SSRN Working Paper Series
No. 1991085 (16 July).
Boyle, Phelim, and Mary Hardy. 2003. “Guaranteed Annuity Options.” ASTIN
Bulletin, vol. 33, no. 2 (November):125–152.
Bravo, Jorge M., Pedro C. Real, and Carlos P. da Silva. 2009. “Participating Life
Annuities Incorporating Longevity Risk Sharing Arrangements.” Working paper.
Broverman, Samuel. 1986. “The Rate of Return on Life Insurance and Annuities.”
Journal of Risk and Insurance, vol. 53, no. 3 (September):419–434.
Brown, Jeffrey R. 2001. “Private Pensions, Mortality Risk and the Decision to
Annuitize.” Journal of Public Economics, vol. 82, no. 1 (October):29–62.
———. 2002. “Differential Mortality and the Value of Individual Account Retirement
Annuities.” In The Distributional Aspects of Social Security and Social Security Reform.
Edited by Martin Feldstein and Jeffrey B. Liebman. Chicago: University of Chicago
Press.
———. 2009. “Understanding the Role of Annuities in Retirement Planning.” In
Overcoming the Saving Slump: How to Increase the Effectiveness of Financial Education
and Saving Programs. 1st ed. Edited by Annamaria Lusardi. Chicago: University of
Chicago Press.
Brown, Jeffrey R., and Theo Nijman. 2011. “Opportunities for Improving Pension
Wealth Decumulation in the Netherlands.” Netspar Discussion Paper No. 01/2011-
008 (9 March).

120 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 120
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography

Brown, Jeffrey R., and James M. Poterba. 2000. “Joint Life Annuities and Annuity
Demand by Married Couples.” Journal of Risk and Insurance, vol. 67, no. 4
(December):527–553.
———. 2006. “Household Ownership of Variable Annuities.” In Tax Policy and the
Economy. Edited by James M. Poterba. Cambridge, MA: MIT Press.
Brown, Jeffrey R., Olivia S. Mitchell, and James M. Poterba. 2001. “The Role of Real
Annuities and Indexed Bonds in an Individual Accounts Retirement Program.” In
Risk Aspects of Investment-Based Social Security Reform. Edited by John Y. Campbell
and Martin Feldstein. Chicago: University of Chicago Press.
———. 2002. “Mortality Risk, Inflation Risk, and Annuity Products.” In Innovations
in Retirement Financing. Edited by Olivia S. Mitchell, Zvi Bodie, P. Brett Hammond,
and Stephen Zeldes. Philadelphia: University of Pennsylvania Press.
Brown, Jeffrey R., Jeffrey R. Kling, Sendhil Mullainathan, and Marian V. Wrobel.
2008. “Why Don’t People Insure Late-Life Consumption? A Framing Explanation
of the Under-Annuitization Puzzle.” American Economic Review, vol. 98, no. 2
(May):304–309.
Brown, Jeffrey R., Olivia S. Mitchell, James M. Poterba, and Mark J. Warshawsky.
1999. “Taxing Retirement Income: Nonqualified Annuities and Distributions from
Qualified Accounts.” National Tax Journal, vol. 53, no. 3 (September):563–591.
———. 2001. The Role of Annuity Markets in Financing Retirement. Cambridge, MA:
MIT Press.
Brugiavini, Agar. 1993. “Uncertainty Resolution and the Timing of Annuity
Purchases.” Journal of Public Economics, vol. 50, no. 1 ( January):31–62.
Brunner, Johann K., and Susanne Pech. 2008. “Optimum Taxation of Life Annuities.”
Social Choice and Welfare, vol. 30, no. 2 (February):285–303.
Buser, Stephen A., and Michael L. Smith. 1983. “Life Insurance in a Portfolio
Context.” Insurance: Mathematics and Economics, vol. 2, no. 3 ( July):147–157.
Bütler, Monika, and Stefan Staubli. 2011. “Payouts in Switzerland: Explaining
Developments in Annuitization.” In Securing Lifelong Retirement Income: Global
Annuity Markets and Policy. Edited by Olivia S. Mitchell, John Piggott, and Noriyuki
Takayama. London: Oxford University Press.
Bütler, Monika, and Federica Teppa. 2007. “The Choice between an Annuity and a
Lump Sum: Results from Swiss Pension Funds.” Journal of Public Economics, vol. 91,
no. 10 (November):1944–1966.
Bütler, Monika, Kim Peijnenburg, and Stefan Staubli. 2011. “How Much Do
Means-Tested Benefits Reduce the Demand for Annuities?” CESifo Working Paper
Series No. 3493 (28 June).

©2013 The Research Foundation of CFA Institute  121

Milevxky.indb 121
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

Bütler, Monika, Stefan Staubli, and Maria G. Zito. Forthcoming. “How Much Does
Annuity Demand React to a Large Price Change?” Scandinavian Journal of Economics.
Cairns, Andrew J.G., David Blake, and Kevin Dowd. 2006. “Stochastic Lifestyling:
Optimal Dynamic Asset Allocation for Defined Contribution Pension Plans.” Journal
of Economic Dynamics & Control, vol. 30, no. 5 (May):843–877.
———. 2008. “Modelling and Management of Mortality Risk: A Review.”
Scandinavian Actuarial Journal, vol. 2008, nos. 2–3:79–113.
Campbell, John Y., João F. Cocco, Francisco J. Gomes, and Pascal J. Maenhout.
2001. “Investing Retirement Wealth: A Life-Cycle Model.” In Risk Aspects of
Investment-Based Social Security Reform. Edited by John Y. Campbell and Martin
Feldstein. Chicago: University of Chicago Press.
Cannon, Edmund, and Ian Tonks. 2005. “Survey of Annuity Pricing.” Research
Report No. 318, Department for Work and Pensions, Leeds, U.K.
———. 2008. Annuity Markets. New York City: Oxford University Press.
———. 2009. “Money’s Worth of Pension Annuities.” Research Report No. 563,
Department for Work and Pensions, Leeds, U.K.
Cappelletti, Giuseppe, Giovanni Guazzarotti, and Pietro Tommasino. 2011. “What
Determines Annuity Demand at Retirement?” Bank of Italy Economic Working
Paper No. 805 (April).
Carlson, Severin, and Blair Lord. 1986. “Unisex Retirement Benefits and the
Market for Annuity ‘Lemons.’” Journal of Risk and Insurance, vol. 53, no. 3
(September):409–418.
Carriere, Jacques F. 1999. “No-Arbitrage Pricing for Life Insurance and Annuities.”
Economics Letters, vol. 64, no. 3 (September):339–342.
Carson, James M., James S. Doran, and Randy E. Dumm. 2011. “Market Discipline
in the Individual Annuity Market.” Risk Management and Insurance Review, vol. 14,
no. 1 (Spring):27–47.
Chalmers, John, and Jonathan Reuter. 2012. “How Do Retirees Value Life Annuities?
Evidence from Public Employees.” Review of Financial Studies, vol. 25, no. 8
(August):2601–2634.
Charupat, Narat, and Moshe A. Milevsky. 2001. “Mortality Swaps and Tax Arbitrage
in the Canadian Insurance and Annuity Markets.” Journal of Risk and Insurance, vol.
68, no. 2 ( June):277–302.
———. 2002. “Optimal Asset Allocation in Life Annuities: A Note.” Insurance:
Mathematics and Economics, vol. 30, no. 2 (April):199–209.
Charupat, Narat, Mark J. Kamstra, and Moshe A. Milevsky. 2012. “The Annuity
Duration Puzzle.” SSRN Working Paper Series No. 2021579 (14 March).

122 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 122
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography

Chen, Peng, and Moshe A. Milevsky. 2003. “Merging Asset Allocation and Longevity
Insurance: An Optimal Perspective on Payout Annuities.” Journal of Financial
Planning, vol. 16, no. 6 ( June):52–62.
Chen, Peng, Roger G. Ibbotson, Moshe A. Milevsky, and Kevin X. Zhu. 2006.
“Human Capital, Asset Allocation, and Life Insurance.” Financial Analysts Journal,
vol. 62, no. 1 ( January/February):97–109.
Chen, Zhuliang, Ken Vetzal, and Peter A. Forsyth. 2008. “The Effect of Modelling
Parameters on the Value of GMWB Guarantees.” Insurance: Mathematics and
Economics, vol. 43, no. 1 (August):165–173.
Cocco, João F., Francisco J. Gomes, and Pascal J. Maenhout. 2005. “Consumption
and Portfolio Choice over the Life Cycle.” Review of Financial Studies, vol. 18, no. 2
(Summer):491–533.
Council of Economic Advisers. 2012. “Supporting Retirement for American Families.”
Executive Office of the President.
Cox, Samuel H., and Yijia Lin. 2007. “Natural Hedging of Life and Annuity Mortality
Risks.” North American Actuarial Journal, vol. 11, no. 3:1–15.
Creighton, Adam, Henry Hongbo Jin, John Piggott, and Emiliano A. Valdez.
2005. “Longevity Insurance: A Missing Market.” Singapore Economic Review,
vol. 50:417–435.
Dai, Min, Yue K. Kwok, and Jianping Zong. 2008. “Guaranteed Minimum
Withdrawal Benefit in Variable Annuities.” Mathematical Finance, vol. 18, no. 4
(October):595–611.
Davidoff, Thomas. 2009. “Housing, Health, and Annuities.” Journal of Risk and
Insurance, vol. 76, no. 1 (March):31–52.
Davidoff, Thomas, Jeffrey R. Brown, and Peter A. Diamond. 2005. “Annuities and
Individual Welfare.” American Economic Review, vol. 95, no. 5 (December):1573–1590.
Davies, James B. 1981. “Uncertain Lifetime, Consumption, and Dissaving in
Retirement.” Journal of Political Economy, vol. 89, no. 3 ( June):561–577.
Dellinger, Jeffrey K. 2011. “When to Commence Income Annuities.” Retirement
Income Solutions Enterprise.
Devolder, Pierre, and Donatien Hainaut. 2006. “The Annuity Puzzle Revisited: A
Deterministic Version with Lagrangian Methods.” Belgian Actuarial Bulletin, vol. 6,
no. 1:40–48.
Di Giacinto, Marina, and Elena Vigna. 2012. “On the Sub-Optimality Cost of
Immediate Annuitization in DC Pension Funds.” Central European Journal of
Operations Research, vol. 20, no. 3 (September):497–537.
Dickson, David C.M., Mary R. Hardy, and Howard R. Waters. 2009. Actuarial
Mathematics for Life Contingent Risks. Cambridge, U.K.: Cambridge University Press.

©2013 The Research Foundation of CFA Institute  123

Milevxky.indb 123
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

Direr, Alexis. 2010. “Flexible Life Annuities.” Journal of Public Economic Theory, vol. 12,
no. 1 (February):43–55.
Dowd, Kevin, David Blake, and Andrew J.G. Cairns. 2011. “A Computationally
Efficient Algorithm for Estimating the Distribution of Future Annuity Values
under Interest-Rate and Longevity Risks.” North American Actuarial Journal, vol. 15,
no. 2:237–247.
Drinkwater, Matthew, and Eric T. Sondergeld. 2004. “Perceptions of Mortality
Risk: Implications for Annuities.” In Pension Design and Structure: New Lessons from
Behavioral Finance. Edited by Olivia S. Mitchell and Stephen P. Utkus. New York
City: Oxford University Press.
Duncan, Robert M. 1952. “A Retirement System Granting Unit Annuities and
Investing in Equities.” Transactions of Society of Actuaries, vol. 4, no. 10:317–344.
Dushi, Irena, and Anthony Webb. 2004. “Household Annuitization Decisions:
Simulations and Empirical Analyses.” Journal of Pension Economics and Finance, vol. 3,
no. 2 ( July):109–143.
Eckstein, Zvi, Martin Eichenbaum, and Dan Peled. 1985a. “Uncertain Lifetimes and
the Welfare Enhancing Properties of Annuity Markets and Social Security.” Journal of
Public Economics, vol. 26, no. 3 (April):303–326.
———. 1985b. “The Distribution of Wealth and Welfare in the Presence of Incomplete
Annuity Markets.” Quarterly Journal of Economics, vol. 100, no. 3 (August):789–806.
Edwards, Ryan D. 2012. “The Cost of Uncertain Life Span.” Journal of Population
Economics (February).
Fehr, Hans, and Christian Habermann. 2008. “Welfare Effects of Life Annuities:
Some Clarifications.” Economics Letters, vol. 99, no. 1 (April):177–180.
Feigenbaum, James, and Emin Gahramanov. 2012. “Is It Really Good to Annuitize?”
Working paper, Economic Series No. 2012 (1), Deakin University, Faculty of Business
and Law, School of Accounting, Economics and Finance (6 March).
Feldstein, Martin, and Elena Ranguelova. 2001. “Individual Risk in an
Investment-Based Social Security System.” American Economic Review, vol. 91, no. 4
(September):1116–1125.
Finkelstein, Amy, and James Poterba. 2002. “Selection Effects in the United Kingdom
Individual Annuities Market.” Economic Journal, vol. 112, no. 476 ( January):28–50.
———. 2004. “Adverse Selection in Insurance Markets: Policyholder Evidence
from the U.K. Annuity Market.” Journal of Political Economy, vol. 112, no. 1
(February):183–208.
Finkelstein, Amy, James Poterba, and Casey Rothschild. 2009. “Redistribution
by Insurance Market Regulation: Analyzing a Ban on Gender-Based Retirement
Annuities.” Journal of Financial Economics, vol. 91, no. 1 ( January):38–58.

124 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 124
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography

Fischer, Stanley. 1973. “A Life Cycle Model of Life Insurance Purchases.” International
Economic Review, vol. 14, no. 1 (February):132–152.
Fisher, Irving. 1930. The Theory of Interest: As Determined by Impatience to Spend Income
and Opportunity to Invest It. New York City: Macmillan.
Fong, Joelle H.Y., Jean Lemaire, and Yiu K. Tse. 2011. “Improving Money’s Worth
Ratio Calculations: The Case of Singapore’s Pension Annuities.” SSRN Working
Paper Series No. 1928323 (7 September).
Fong, Joelle H.Y., Olivia S. Mitchell, and Benedict S.K. Koh. 2011. “Longevity Risk
Management in Singapore’s National Pension System.” Journal of Risk and Insurance,
vol. 78, no. 4 (December):961–982.
Fornero, Elsa, and Elisa Luciano, eds. 2004. Developing an Annuity Market in Europe.
Northampton, MA: Edward Elgar Publishing.
Freedman, Barry. 2008. “Efficient Post-Retirement Asset Allocation.” North American
Actuarial Journal, vol. 12, no. 3:228–241.
Frees, Edward W., Jacques Carriere, and Emiliano Valdez. 1996. “Annuity Valuation
with Dependent Mortality.” Journal of Risk and Insurance, vol. 63, no. 2 ( June):229–261.
Friedman, Benjamin M., and Mark J. Warshawsky. 1990. “The Cost of Annuities:
Implications for Saving Behaviour and Bequests.” Quarterly Journal of Economics, vol.
105, no. 1:135–154.
Friedman, Milton. 1957. A Theory of the Consumption Function. Princeton, NJ:
Princeton University Press.
Gale, William G., J.M. Iwry, David C. John, and Lina Walker. 2008. “Increasing
Annuitization in 401(k) Plans with Automatic Trial Income.” The Retirement
Security Project, Brookings Institution.
Ganegoda, Amandha, and Hazel Bateman. 2008. “Australia’s Disappearing Market for
Life Annuities.” Centre for Pensions and Superannuation Discussion Paper 2008-1.
Gazzale, Robert S., and Lina Walker. 2009. “Behavioral Biases in Annuity Choice: An
Experiment.” Williams College Economics Department Working Paper Series No.
2009-01.
Gerrard, Russell, Steven Haberman, and Elena Vigna. 2004. “Optimal Investment
Choices Post-Retirement in a Defined Contribution Pension Scheme.” Insurance:
Mathematics and Economics, vol. 35, no. 2 (October):321–342.
Goda, Gopi S., and Colin M. Ramsay. 2007. “Determining the Optimum Guarantee
Period for a One-Life Retirement Annuity.” North American Actuarial Journal, vol. 11,
no. 3:100–112.
Goldsticker, Ralph. 2007. “A Mutual Fund to Yield Annuity-Like Benefits.” Financial
Analysts Journal, vol. 63, no. 1 ( January/February):63–67.

©2013 The Research Foundation of CFA Institute  125

Milevxky.indb 125
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

Gong, Guan, and Anthony Webb. 2008. “Mortality Heterogeneity and the
Distributional Consequences of Mandatory Annuitization.” Journal of Risk and
Insurance, vol. 75, no. 4 (December):1055–1079.
———. 2010. “Evaluating the Advanced Life Deferred Annuity—An Annuity
People Might Actually Buy.” Insurance: Mathematics and Economics, vol. 46, no. 1
(February):210–221.
Goodman, Benjamin, and Michael Heller. 2006. “Annuities: Now, Later, Never?”
TIAA-CREF Institute Trends and Issues.
Gunawardena, Dmitri, Christopher Hicks, and David O’Neill. 2008. “Pension
Annuities: Pension Annuities and the Open Market Solution.” Research Paper No. 8,
Research Department, Association of British Insurers.
Gupta, Aparna, and Zhisheng Li. 2007. “Integrating Optimal Annuity Planning with
Consumption–Investment Selections in Retirement Planning.” Insurance: Mathematics
and Economics, vol. 41, no. 1 ( July):96–110.
Hainaut, Donatien, and Pierre Devolder. 2006. “Life Annuitization: Why and How
Much?” ASTIN Bulletin, vol. 36, no. 2:629–654.
Hakansson, Nils H. 1969. “Optimal Investment and Consumption Strategies under
Risk, an Uncertain Lifetime, and Insurance.” International Economic Review, vol. 10,
no. 3 (October):443–466.
Halley, Edmond. 1693. “An Estimate of the Degrees of the Mortality of Mankind,
Drawn from the Curious Tables of the Births and Funerals at the City of Breslaw.”
Philosophical Transactions, vol. 17:596–610.
Hamermesh, Daniel S. 1985. “Expectations, Life Expectancy, and Economic
Behavior.” Quarterly Journal of Economics, vol. 100, no. 2 (May):389–408.
Hansen, Gary D., and Selahattin İmrohoroĝlu. 2008. “Consumption over the
Life Cycle: The Role of Annuities.” Review of Economic Dynamics, vol. 11, no. 3
( July):566–583.
Horneff, Wolfram J., Raimond Maurer, and Ralph Rogalla. 2010. “Dynamic Portfolio
Choice with Deferred Annuities.” Journal of Banking & Finance, vol. 34, no. 11
(November):2652–2664.
Horneff, Wolfram J., Raimond H. Maurer, and Michael Z. Stamos. 2008. “Life-Cycle
Asset Allocation with Annuity Markets.” Journal of Economic Dynamics & Control, vol.
32, no. 11 (November):3590–3612.
Horneff, Wolfram J., Raimond H. Maurer, Olivia S. Mitchell, and Ivica Dus. 2008.
“Following the Rules: Integrating Asset Allocation and Annuitization in Retirement
Portfolios.” Insurance: Mathematics and Economics, vol. 42, no. 1 (February):396–408.

126 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 126
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography

Horneff, Wolfram J., Raimond H. Maurer, Olivia S. Mitchell, and Michael Z. Stamos.
2009. “Asset Allocation and Location over the Life Cycle with Investment-Linked
Survival-Contingent Payouts.” Journal of Banking & Finance, vol. 33, no. 9
(September):1688–1699.
———. 2010. “Variable Payout Annuities and Dynamic Portfolio Choice in
Retirement.” Journal of Pension Economics and Finance, vol. 9, no. 2 (April):163–183.
Hu, Wei-Yin, and Jason S. Scott. 2007. “Behavioral Obstacles in the Annuity Market.”
Financial Analysts Journal, vol. 63, no. 6 (November/December):71–82.
Huebner, Solomon S. 1927. The Economics of Life Insurance: Human Life Values, Their
Financial Organization, Management, and Liquidation. New York City: D. Appleton-
Century Company.
Hurd, Michael D. 1989. “Mortality Risk and Bequests.” Econometrica, vol. 57, no. 4
( July):779–813.
Hurd, Michael D., Constantijn W.A. Panis, James P. Smith, and Julie M.
Zissimopoulos. 2004. “Pension Annuitization and Social Security Claiming.” In
Developing an Annuity Market in Europe. Edited by Elsa Fornero and Elisa Luciano.
Northampton, MA: Edward Elgar Publishing.
Ibbotson, Roger G., Moshe A. Milevsky, Peng Chen, and Kevin X. Zhu. 2007. Lifetime
Financial Advice: Human Capital, Asset Allocation, and Insurance. Charlottesville, VA:
Research Foundation of CFA Institute.
Inkmann, Joachim, Paula Lopes, and Alexander Michaelides. 2011. “How Deep Is
the Annuity Market Participation Puzzle?” Review of Financial Studies, vol. 24, no. 1
( January):279–319.
James, Estelle, and Renuka Sane. 2003. “The Annuity Market in India: Do Consumers
Get Their Money’s Worth? What Are the Key Public Policy Issues?” In Rethinking
Pension Provision for India. Edited by Anand Bordia and Gautam Bhardwaj. Noida,
India: Tata McGraw-Hill.
James, Estelle, and Xue Song. 2001. “Annuities Markets around the World: Money’s
Worth and Risk Intermediation.” Center for Research on Pensions and Welfare
Policies Working Paper No. 16/01.
James, Estelle, and Dimitri Vittas. 2001. “Annuity Markets in Comparative
Perspective: Do Consumers Get Their Money’s Worth?” In OECD Private Pensions
Conference 2000. Paris: OECD.
James, Estelle, Guillermo Martinez, and Augusto Iglesias. 2006. “The Payout Stage
in Chile: Who Annuitizes and Why?” Journal of Pension Economics and Finance, vol. 5,
no. 2 ( July):121–154.
Jang, Bong-Gyu, Hyeng K. Koo, and Ho-Seok Lee. 2010. “Default Risk of Life
Annuity and the Annuity Puzzle.” Working paper.

©2013 The Research Foundation of CFA Institute  127

Milevxky.indb 127
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

Jennings, Robert M., and Andrew P. Trout. 1982. The Tontine: From the Reign of
Louis XIV to the French Revolutionary Era. S.S. Huebner Foundation for Insurance
Education, Wharton School, University of Pennsylvania.
Jennings, Robert M., Donald F. Swanson, and Andrew P. Trout. 1988. “Alexander
Hamilton’s Tontine Proposal.” William and Mary Quarterly, vol. 45, no. 1
( January):107–115.
Jiménez-Martín, Sergi, and Alfonso R. Sánchez Martín. 2007. “An Evaluation of
the Life Cycle Effects of Minimum Pensions on Retirement Behaviour.” Journal of
Applied Econometrics, vol. 22, no. 5 (August):923–950.
Jousten, Alain. 2001. “Life-Cycle Modeling of Bequests and Their Impact on Annuity
Valuation.” Journal of Public Economics, vol. 79, no. 1 ( January):149–177.
Kaplan, Paul D. 2006. “Asset Allocation with Annuities for Retirement Income
Management.” Journal of Wealth Management, vol. 8, no. 4 (Spring):27–40.
Kapur, Sandeep, and J.M. Orszag. 1999. “A Portfolio Approach to Investment and
Annuitization during Retirement.” Mimeo. London: Birbeck College, University of
London.
Kartashov, Vasily, Raimond Maurer, Olivia S. Mitchell, and Ralph Rogalla.
2011. “Lifecycle Portfolio Choices with Systematic Longevity Risk and Variable
Investments-Linked Deferred Annuities.” NBER Working Paper No. 17505
(October).
Khorasanee, M.Z. 1996. “Annuity Choices for Pensioners.” Journal of Actuarial
Practice, vol. 4, no. 2:229–255.
Kingston, Geoffrey, and Susan Thorp. 2005. “Annuitization and Asset Allocation
with HARA Utility.” Journal of Pension Economics and Finance, vol. 4, no. 3
(November):225–248.
Koijen, Ralph S.J., Theo E. Nijman, and Bas J.M. Werker. 2011. “Optimal Annuity
Risk Management.” Review of Finance, vol. 15, no. 4 (October):799–833.
Kopf, E.W. 1927. “The Early History of the Annuity.” Proceedings of the Casualty
Actuarial Society, vol. 13, no. 28:225–266.
Kotlikoff, Laurence J., and Avia Spivak. 1981. “The Family as an Incomplete Annuities
Market.” Journal of Political Economy, vol. 89, no. 2 (April):372–391.
Kotlikoff, Laurence J., John Shoven, and Avia Spivak. 1986. “The Effect of Annuity
Insurance on Savings and Inequality.” Journal of Labor Economics, vol. 4, no. 3
( July):S183–S207.
Kwon, Hyuk-Sung, and Bruce L. Jones. 2006. “The Impact of Determinants of
Mortality on Life Insurance and Annuities.” Insurance: Mathematics and Economics,
vol. 38, no. 2 (April):271–288.

128 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 128
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography

Lachance, Marie-Eve. 2012. “Optimal Onset and Exhaustion of Retirement Savings

in a Life-Cycle Model.” Journal of Pension Economics and Finance, vol. 11, no. 1
( January):21–52.
Lewin, C.G. 2003. Pensions and Insurance before 1800: A Social History. East Lothian,
U.K.: Tuckwell Press.
Lieber, Ron. 2010. “The Unloved Annuity Gets a Hug from Obama.” New York Times
(29 January): www.nytimes.com/2010/01/30/your-money/annuities/30money.html.
LIMRA. 2010. “U.S. Individual Annuity Yearbook.” Windsor, CT.
Lin, Yijia, and Samuel H. Cox. 2005. “Securitization of Mortality Risks in Life
Annuities.” Journal of Risk and Insurance, vol. 72, no. 2 ( June):227–252.
Lockwood, Lee M. 2012. “Bequest Motives and the Annuity Puzzle.” Review of
Economic Dynamics, vol. 15, no. 2 (April):226–243.
Lopes, Paula, and Alexander Michaelides. 2007. “Rare Events and Annuity Market
Participation.” Finance Research Letters, vol. 4, no. 2 ( June):82–91.
Macdonald, Angus, and Kenneth McIvor. 2010. “Pensions and Genetics: Can
Longevity Genes Be Reliable Risk Factors for Annuity Pricing?” Scandinavian
Actuarial Journal, vol. 2010, no. 1:1–14.
Mackenzie, George A. 2006. Annuity Markets and Pension Reform. New York City:
Cambridge University Press.
MacMinn, Richard, Patrick Brockett, and David Blake. 2006. “Longevity Risk and
Capital Markets.” Journal of Risk and Insurance, vol. 73, no. 4 (December):551–557.
McCarthy, David, and Olivia S. Mitchell. 2004. “Annuities for an Ageing World.” In
Developing an Annuity Market in Europe. Edited by Elsa Fornero and Elisa Luciano.
Northampton, MA: Edward Elgar Publishing.
———. 2010. “International Adverse Selection in Life Insurance and Annuities.”
In Ageing in Advanced Industrial States: Riding the Age Waves, vol. 3. 1st ed. Edited
by Shripad Tuljapurkar, Naohiro Ogawa, and Anne H. Gauthier. New York City:
Springer.
McQueen, Rod. 1997. Who Killed Confederation Life: The Inside Story. Toronto:
McClelland and Stewart.
Mehr, Robert I. 1958. “The Variable Annuity: Security or Insurance.” Journal of
Finance, vol. 13, no. 3 (September):386–411.
Menoncin, Francesco. 2008. “The Role of Longevity Bonds in Optimal Portfolios.”
Insurance: Mathematics and Economics, vol. 42, no. 1 (February):343–358.
Mereu, John A. 1962. “Annuity Values Directly from the Makeham Constants.”
Transaction of Society of Actuaries, vol. 14:269–286.

©2013 The Research Foundation of CFA Institute  129

Milevxky.indb 129
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Life Annuities

Merton, Robert C. 1971. “Optimum Consumption and Portfolio Rules in a

Continuous-Time Model.” Journal of Economic Theory, vol. 3, no. 4 (December):373–413.
Milevsky, Moshe A. 1998. “Optimal Asset Allocation towards the End of the Life
Cycle: To Annuitize or Not to Annuitize?” Journal of Risk and Insurance, vol. 65, no. 3
(September):401–426.
———. 2005a. “Real Longevity Insurance with a Deductible: Introduction to
Advanced-Life Delayed Annuities (ALDA).” North American Actuarial Journal, vol. 9,
no. 4:109–122.
———. 2005b. “The Implied Longevity Yield: A Note on Developing an Index for
Life Annuities.” Journal of Risk and Insurance, vol. 72, no. 2 ( June):302–320.
———. 2006. The Calculus of Retirement Income: Financial Models for Pension and
Insurance. New York City: Cambridge University Press.
———. 2011. “Spending Retirement on Planet Vulcan: The Impact of Longevity
Risk Aversion on Optimal Withdrawal Rates.” Financial Analysts Journal, vol. 67, no.
2 (March/April):45–58.
———. 2012. The Seven Most Important Equations for Your Retirement. Toronto: John
Wiley & Sons Canada.
———. Forthcoming 2013. Tontines: How a Fascinating but Neglected Annuity Scheme
Can Help Reduce the Cost of Retirement.
Milevsky, Moshe A., and Huaxiong Huang. 2011. “Spending Retirement on Planet
Vulcan: The Impact of Longevity Risk Aversion on Optimal Withdrawal Rates.”
Financial Analysts Journal, vol. 67, no. 2 (March/April):45–58.
Milevsky, Moshe A., and Vlad Kyrychenko. 2008. “Portfolio Choice with Puts:
Evidence from Variable Annuities.” Financial Analysts Journal, vol. 64, no. 2 (March/
April):1–30.
Milevsky, Moshe A., and Steven E. Posner. 2001. “The Titanic Option: Valuation
of Guaranteed Minimum Death Benefit in Variable Annuities and Mutual Funds.”
Journal of Risk and Insurance, vol. 68, no. 1 (March):93–128.
Milevsky, Moshe A., and S. David Promislow. 2001. “Mortality Derivatives and
the Option to Annuitise.” Insurance: Mathematics and Economics, vol. 29, no. 3
(December):299–318.
Milevsky, Moshe A., and Chris Robinson. 2005. “A Sustainable Spending
Rate without Simulation.” Financial Analysts Journal, vol. 61, no. 6 (November/
December):89–100.
Milevsky, Moshe A., and Thomas S. Salisbury. 2006. “Financial Valuation of
Guaranteed Minimum Withdrawal Benefits.” Insurance: Mathematics and Economics,
vol. 38, no. 1 (February):21–38.

130 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 130
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:47 PM
 Bibliography

Milevsky, Moshe A., and Virginia R. Young. 2007a. “Annuitization and

Asset Allocation.” Journal of Economic Dynamics and Control, vol. 31, no. 9
(September):3138–3177.
Milevsky, Moshe A., and Virginia R. Young. 2007b. “The Timing of Annuitization:
Investment Dominance and Mortality Risk.” Insurance: Mathematics and Economics,
vol. 40, no. 1 ( January):135–144.
Mirer, Thad W. 1979. “The Wealth–Age Relation among the Aged.” American
Economic Review, vol. 69, no. 3 ( June):435–443.
Mitchell, Olivia S., and David McCarthy. 2002. “Estimating International Adverse
Selection in Annuities.” North American Actuarial Journal, vol. 6, no. 4:38–54.
Mitchell, Olivia S., John Piggott, and Noriyuki Takayama. 2011. Securing Lifelong
Retirement Income: Global Annuity Markets and Policy. Oxford, U.K.: Oxford University
Press.
Mitchell, Olivia S., James M. Poterba, Mark J. Warshawsky, and Jeffrey R. Brown.
1999. “New Evidence on the Money’s Worth of Individual Annuities.” American
Economic Review, vol. 89, no. 5 (December):1299–1318.
Modigliani, Franco. 1986. “Life Cycle, Individual Thrift, and the Wealth of Nations.”
American Economic Review, vol. 76, no. 3 ( June):297–313.
Mottola, Gary R., and Stephen P. Utkus. 2007. “Lump Sum or Annuity? An
Analysis of Choice in DB Pension Payouts.” Vanguard Center for Retirement Research,
vol. 30:1–10.
Mudavanhu, Blessing, and Jun Zhuo. 2002. “Valuing Guaranteed Minimum Death
Benefits in Variable Annuities and the Option to Lapse.” Working paper, Haas School
of Business, University of California, Berkeley.
Munnell, Alicia H., Francesca Golub-Sass, Mauricio Soto, and Francis Vitagliano.
2007. “Why Are Healthy Employers Freezing Their Pensions?” Journal of Pension
Benefits, vol. 14, no. 4:3–14.
Murray, Michael L., and Stuart Klugman. 1990. “Impaired Health Life Annuities.”
Journal of the American Society of CLU and ChFC, vol. 44, no. 5:50–58.
Murtaugh, Christopher M., Brenda C. Spillman, and Mark J. Warshawsky. 2001. “In
Sickness and in Health: An Annuity Approach to Financing Long-Term Care and
Retirement Income.” Journal of Risk and Insurance, vol. 68, no. 2 ( June):225–253.
Nielson, Norma L. 2012. “Annuities and Your Nest Egg: Reforms to Promote Optimal
Annuitization of Retirement.” C.D. Howe Commentary No. 358 (16 August).
Olsen, John, and Michael E. Kitces. 2009. The Annuity Advisor. 2nd ed. Cincinnati:
The National Underwriter Company.

©2013 The Research Foundation of CFA Institute  131

Milevxky.indb 131
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:48 PM
 Life Annuities

Pang, Gaobo, and Mark J. Warshawsky. 2009. “Comparing Strategies for Retirement
Wealth Management: Mutual Funds and Annuities.” Journal of Financial Planning,
vol. 22, no. 8 (August):36–47.
Pang, Gaobo, and Mark J. Warshawsky. 2010. “Optimizing the Equity-Bond Annuity
Portfolio in Retirement: The Impact of Uncertain Health Expenses.” Insurance:
Mathematics and Economics, vol. 46, no. 1 (February):198–209.
Panis, Constantijn W.A. 2004. “Annuities and Retirement Well-Being.” In Pension
Design and Structure: New Lessons from Behavioral Finance. Edited by Olivia S.
Mitchell and Stephen P. Utkus. Oxford, U.K.: Oxford University Press.
Park, Youngkyun. 2011. “Retirement Income Adequacy with Immediate and
Longevity Annuities.” Employee Benefit Research Institute, Issue Brief No. 357.
Pashchenko, Svetlana. 2010. “Accounting for Non-Annuitization.” Working Paper
No. 2010-03, Federal Reserve Bank of Chicago.
Pecchenino, Rowena A., and Patricia S. Pollard. 1997. “The Effects of Annuities,
Bequests, and Aging in an Overlapping Generations Model of Endogenous Growth.”
Economic Journal, vol. 107, no. 440 ( January):26–46.
Pechter, Kerry. 2008. Annuities for Dummies. Indianapolis: Wiley Publishing.
Peijnenburg, Kim, Theo Nijman, and Bas J.M. Werker. 2012. “Health Cost Risk,
Incomplete Markets, or Bequest Motives—Revisiting the Annuity Puzzle.” Working
paper.
Philipson, Tomas J., and Gary S. Becker. 1998. “Old-Age Longevity and
Mortality-Contingent Claims.” Journal of Political Economy, vol. 106, no. 3
( June):551–573.
Piggott, John, Emiliano A. Valdez, and Bettina Detzel. 2005. “The Simple
Analytics of a Pooled Annuity Fund.” Journal of Risk and Insurance, vol. 72, no. 3
(September):497–520.
Pliska, Stanley R., and Jinchun Ye. 2007. “Optimal Life Insurance Purchase and
Consumption/Investment under Uncertain Lifetime.” Journal of Banking & Finance,
vol. 31, no. 5 (May):1307–1319.
Post, Thomas, Helmut Gründl, and Hato Schmeiser. 2006. “Portfolio Management
and Retirement: What Is the Best Arrangement for a Family?” Financial Markets and
Portfolio Management, vol. 20, no. 3 (September):265–285.
Poterba, James M. 2001. “Annuity Markets and Retirement Security.” Fiscal Studies,
vol. 22, no. 3 (September):249–270.
———. 2005. “Annuities in Early Modern Europe.” In The Origins of Value: The
Financial Innovations that Created Modern Capital Markets. Edited by William N.
Goetzmann and K. Geert Rouwenhorst. Oxford, U.K.: Oxford University Press.

132 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 132
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:48 PM
 Bibliography

Previtero, Alessandro. 2011. “Stock Market Returns and Annuitization: A Case of

Myopic Extrapolation.” SSRN Working Paper Series No. 1787123.
Promislow, S. David. 2011. Fundamentals of Actuarial Mathematics. 2nd ed. Chichester,
U.K.: John Wiley & Sons.
Purcal, Sachi, and John Piggott. 2008. “Explaining Low Annuity Demand: An
Optimal Portfolio Application to Japan.” Journal of Risk and Insurance, vol. 75, no. 2
( June):493–516.
Ransom, Roger L., and Richard Sutch. 1987. “Tontine Insurance and the Armstrong
Investigation: A Case of Stifled Innovation, 1868–1905.” Journal of Economic History,
vol. 47, no. 2 ( June):379–390.
Reichenstein, William. 2003. “Allocation during Retirement: Adding Annuities to the
Mix.” Journal of the American Association of Individual Investors (November):3–9.
Richard, Scott F. 1975. “Optimal Consumption, Portfolio and Life Insurance Rules
for an Uncertain Lived Individual in a Continuous Time Model.” Journal of Financial
Economics, vol. 2, no. 2 ( June):187–203.
Robinson, S., and A. Fliegelman. 2002. “The U.S. Payout Annuity Market.” Moody’s
Investors Service, Special Comment (August).
Rocha, Roberto, Dimitri Vittas, and Heinz P. Rudolph. 2011. “Annuities and
Other Retirement Products: Designing the Payout Phase.” International Bank for
Reconstruction and Development, World Bank, Washington, DC.
Rotemberg, Julio J. 2009. “Can a Continuously-Liquidating Tontine (or Mutual
Inheritance Fund) Succeed Where Immediate Annuities Have Floundered?” Working
Paper 09-121, Harvard Business School.
Rothschild, Casey G. 2009. “Adverse Selection in Annuity Markets: Evidence from
the British Life Annuity Act of 1808.” Journal of Public Economics, vol. 93, no. 5–6
( June):776–784.
Salm, Martin. 2011. “The Effect of Pensions on Longevity: Evidence from Union
Army Veterans.” Economic Journal, vol. 121, no. 552 (May):595–619.
Samuelson, Paul A. 1969. “Lifetime Portfolio Selection by Dynamic Stochastic
Programming.” Review of Economics and Statistics, vol. 51, no. 3 (August):239–246.
Sass, Steven, Alicia H. Munnell, and Andrew Eschtruth. 2011. “Managing Your
Money in Retirement.” Financial Security Project, Boston College.
Schulze, Roman N., and Thomas Post. 2010. “Individual Annuity Demand under
Aggregate Mortality Risk.” Journal of Risk and Insurance, vol. 77, no. 2 ( June):423–449.
Scott, Jason S., John G. Watson, and Wei-Yin Hu. 2011. “What Makes a Better
Annuity?” Journal of Risk and Insurance, vol. 78, no. 1:213–244.

©2013 The Research Foundation of CFA Institute  133

Milevxky.indb 133
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:48 PM
 Life Annuities

Sexauer, Stephen C., and Laurence B. Siegel. 2013. “A Pension Promise to Oneself.”
Working paper.
Sexauer, Stephen C., Michael W. Peskin, and Daniel Cassidy. 2012. “Making
Retirement Income Last a Lifetime.” Financial Analysts Journal, vol. 68, no. 1 ( January/
February):74–84.
Shah, Premal, and Dimitris Bertsimas. 2008. “An Analysis of the Guaranteed
Withdrawal Benefits for Life Option.” SSRN Working Paper Series No. 1312727.
Shankar, S.G. 2009. “A New Strategy to Guarantee Retirement Income Using TIPS
and Longevity Insurance.” Financial Services Review, vol. 18, no. 1 (Spring):53–68.
Sharpe, William F. 1964. “Capital Asset Prices: A Theory of Market Equilibrium
under Conditions of Risk.” Journal of Finance, vol. 19, no. 3 (September):425–442.
Sheshinski, Eytan. 2007. “Optimum and Risk-Class Pricing of Annuities.” Economic
Journal, vol. 117, no. 516 ( January):240–251.
———. 2008. The Economic Theory of Annuities. Princeton, NJ: Princeton University
Press.
———. 2010. “Refundable Annuities (Annuity Options).” Journal of Public Economic
Theory, vol. 12, no. 1 (February):7–21.
Shi, Zhen. 2008. “Annuitization and Retirement Timing Decisions.” SSRN Working
Paper Series No. 1516834.
Sinclair, Sven H., and Kent A. Smetters. 2004. “Health Shocks and the Demand for
Annuities.” Congressional Budget Office Technical Paper 2004–09.
Sinha, Tapen. 1986. “The Effects of Survival Probabilities, Transactions Cost and the
Attitude towards Risk on the Demand for Annuities.” Journal of Risk and Insurance,
vol. 53, no. 2 ( June):301–307.
Soares, Chris, and Mark J. Warshawsky. 2004. “Annuity Risk: Volatility and Inflation
Exposure in Payments from Immediate Life Annuities.” In Developing an Annuity
Market in Europe. Edited by Elsa Fornero and Elisa Luciano. Northampton, MA:
Edward Elgar Publishing.
Stabile, Gabriele. 2006. “Optimal Timing of the Annuity Purchase: Combined
Stochastic Control and Optimal Stopping Problem.” International Journal of Theoretical
and Applied Finance, vol. 9, no. 2 (March):151–170.
Stevens, Ralph. 2009. “Annuity Decisions with Systematic Longevity Risk.” Working
paper.
Sun, Wei, and Anthony Webb. 2011. “Valuing the Longevity Insurance Acquired
by Delayed Claiming of Social Security.” Journal of Risk and Insurance, vol. 78, no. 4
(December):907–930.

134 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 134
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:48 PM
 Bibliography

Sundaresan, Suresh, and Fernando Zapatero. 1997. “Valuation, Optimal Asset

Allocation and Retirement Incentives of Pension Plans.” Review of Financial Studies,
vol. 10, no. 3 ( July):631–660.
Telford, Peter G., Bridget A. Browne, Ed J. Collinge, Paul Fulcher, Ben E. Johnson,
Wendy Little, Joseph L.C. Lu, Jeremy M. Nurse, Derek W. Smith, and Feifei Zhang.
2011. “Developments in the Management of Annuity Business.” British Actuarial
Journal, vol. 16, no. 3 (September):471–551.
Teppa, Federica. 2011. “Can the Longevity Risk Alleviate the Annuitization Puzzle?
Empirical Evidence from Dutch Data.” Working Paper No. 302, De Nederlandsche
Bank.
Tergesen, Anne, and Leslie Scism. 2009. “Getting Smart about Annuities.” Wall Street
Journal (18 April): http://online.wsj.com/article/SB123972531986417405.html.
Vidal-Meliá, Carlos, and Ana Lejárraga-García. 2006. “Demand for Life Annuities
from Married Couples with a Bequest Motive.” Journal of Pension Economics and
Finance, vol. 5, no. 2 ( July):197–229.
Walliser, Jan. 2000. “Adverse Selection in the Annuities Market and the Impact
of Privatizing Social Security.” Scandinavian Journal of Economics, vol. 102, no. 3
(September):373–393.
Wang, Ting, and Virginia R. Young. 2012. “Optimal Commutable Annuities to
Minimize the Probability of Lifetime Ruin.” Insurance: Mathematics and Economics,
vol. 50, no. 1 ( January):200–216.
Waring, M. Barton. 2004a. “Liability-Relative Investing.” Journal of Portfolio
Management, vol. 30, no. 4 (Summer):8–20.
———. 2004b. “Liability-Relative Investing II.” Journal of Portfolio Management, vol.
31, no. 1 (Fall):40–53.
Warner, John T., and Saul Pleeter. 2001. “The Personal Discount Rate, Evidence
from Military Downsizing Programs.” American Economic Review, vol. 91, no. 1
(March):33–53.
Warshawsky, Mark J. 1988. “Private Annuity Markets in the United States: 1919–
1984.” Journal of Risk and Insurance, vol. 55, no. 3 (September):518–528.
———. 2012. Retirement Income: Risks and Strategies. Cambridge, MA: MIT Press.
Webb, Anthony. 2011. “The U.S. Longevity Insurance Market.” In Securing Lifelong
Retirement Income: Global Annuity Markets and Policy. Edited by Olivia S. Mitchell,
John Piggott, and Noriyuki Takayama. Oxford, U.K.: Oxford University Press.
Weierich, Mariann R., Elizabeth A. Kensinger, Alicia H. Munnell, Steven A. Sass,
Brad C. Dickerson, Christoper I. Wright, and Lisa F. Barrett. 2011. “Older and Wiser?
An Affective Science Perspective on Age-Related Challenges in Financial Decision
Making.” Social Cognitive and Affective Neuroscience, vol. 6, no. 2 (April):195–206.

©2013 The Research Foundation of CFA Institute  135

Milevxky.indb 135
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:48 PM
 Life Annuities

Williams, Arthur C. 1986. “Higher Interest Rates, Longer Lifetimes, and the Demand
for Life Annuities.” Journal of Risk and Insurance, vol. 53, no. 1 (March):164–171.
Yaari, Menahem E. 1964. “On the Consumer’s Lifetime Allocation Process.”
International Economic Review, vol. 5, no. 3 (September):304–317.
———. 1965. “Uncertain Lifetime, Life Insurance and the Theory of the Consumer.”
Review of Economic Studies, vol. 32, no. 2 (April):137–150.
Yagi, Tadashi, and Yasuyuki Nishigaki. 1993. “The Inefficiency of Private Constant
Annuities.” Journal of Risk and Insurance, vol. 60, no. 3 (September):385–412.
Yogo, Motohiro. 2011. “Portfolio Choice in Retirement: Health Risk and the Demand
for Annuities, Housing, and Risky Assets.” SSRN Working Paper Series No. 1085306.
Zahm, Nathan, and John Ameriks. 2011. “Estimating Internal Rates of Return on
Income Annuities.” Vanguard Research.
Zeng, Lulu. 2010. “Optimal Consumption and Portfolio Choice for Retirees.” SSRN
Working Paper Series No. 1327989.

136 ©2013 The Research Foundation of CFA Institute

Milevxky.indb 136
Electronic copy available at: https://ssrn.com/abstract=2571379 5/22/2013 12:39:48 PM