Business Cycles

Business Cycles: Facts and Theory

Mark Huggett1

1 Georgetown

March 8, 2017
Business Cycles

One of the most controversial questions in

macroeconomics is what explains business-cycle
fluctuations?

Economists largely agree on what are the key facts

describing business-cycle fluctuations. Thus, it is
fairly clear what business-cycle theorists are trying
to explain. However, a unifying explanation for
these facts is still hotly debated. Business-cycle
policy is also hotly debated for the same reason.
Business Cycles

Business-Cycle Facts
According to Burns and Mitchell (1946, p. 3), business cycles
are

“... a type of fluctuation found in the aggregate economic activity

of nations that organize their work mainly in business enterprises:
a cycle consists of expansions occuring at about the same time in
many economic activities, followed by similarly general recessions,
contractions, and revivals which merge into the expansion phase of
the next cycle; this sequence of changes is recurrent but not
periodic; in duration business cycles vary from more than one year
to ten or twelve years; they are not divisible into shorter cycles of
similar character with amplitudes approximating their own.”
Business Cycles

Business-Cycle Facts

Economists nowadays characterize business-cycle

facts quite differently from Burns and Mitchell. A
common approach is to divide a time series
(y1 , y2 , ..., yn ) into trend and cycle components.
Business-cycle facts are then properties of the cycle
component ytcycle :

yt = yttrend + ytcycle
Business Cycles

Business-Cycle Facts

Modern view: What are the key facts?

Business-cycle facts describe the statistical regularities of
the cycle component.

Key properties
1. amplitude of fluctuations
2. comovements
3. lead and lag patterns
Business Cycles

US Log GDP 1948‐2012: Data and Trend
‐7.6
1940 1950 1960 1970 1980 1990 2000 2010 2020
‐7.8

‐8

‐8.2
Log Units

‐8.4

‐8.6

‐8.8

‐9

‐9.2
Year

Log GDP Trend
Business Cycles

Business-Cycle Facts

How to define the smooth trend line?

We will follow the literature and use the Hodrick-Prescott
filter to define the smooth trend line. Intuitively, this is a
way to extract the long-run growth portion of a series. It
can also be viewed as a way to take away “low frequency”
components from the data.
Business Cycles

Business-Cycle Facts
How to define the smooth trend line?
Given data (y1 , ..., yT ), Hodrick and Prescott defined the
trend (y1trend , ..., yTtrend ) as the values for the trend that
minimize the objective below when λ = 1600:

T
X T −1
X
(yt − yttrend )2 + λ trend
[(yt+1 trend 2
− yttrend ) − [(yttrend − yt−1 )]
t=1 t=2

This is a mechanical procedure that passes a trend through

data.
Business Cycles

Business-Cycle Facts

The next two figures plot the cyclical components

of output, consumption and investment in the US
as well as the cyclical components of output, labor
hours and labor productivity.

We will then be especialy interested in the

magnitude of the cyclical fluctuations and how
correlated the cyclical fluctuations in various
aggregate series are with output.
Business Cycles

US Business Cycles: Output Components
0.15

0.1
Business Cycle Component

0.05

0
1940 1950 1960 1970 1980 1990 2000 2010 2020

‐0.05

‐0.1

‐0.15
Year

Consumption Investment Output

Business Cycles

US Business Cycles: Output, Labor and Productivity
0.06

0.04
Business Cycle Component

0.02

0
1940 1950 1960 1970 1980 1990 2000 2010 2020
‐0.02

‐0.04

‐0.06

‐0.08
Year

Output Labor Hours Labor Productivity

Business Cycles

Business-Cycle Facts

Next we consider a study by Kydland and Prescott

that documented the magnitude of cyclical
fluctuations in US data and their correlation with
the cyclical component of output. Their data runs
from 1954- 1989. They apply the same definition of
trend as the series plotted earlier.
Business Cycles
Business Cycles
Business Cycles
Business Cycles

Key US Facts:

1. Labor hours are about as variable as output and are

strongly procyclical. The standard deviation of output is
1.71, whereas labor is 1.47. Thus, a typical deviation from
trend for output is 1.71 percent.
2. Labor productivity is procyclical.
3. Consumption is less variable than output but investment
is much more variable than output.
4. Consumption and investment are procyclical.
Government spending is acyclical.
5. Measures of money are procyclical but prices are
countercyclical.
Business Cycles

Business-Cycle Facts
Other Data Issues:
1. Have business-cycle fuctuations changed in magnitude or
nature over time in US data?

There is a literature that debates whether business-cycle

fluctuations in the US were larger before the Great
Depression than after WW II. There is also a literature
that documents smaller business-cycle fluctuations in the
US after 1984 but before the Great Recession.

2. Is the systematic movement in quarterly GDP (the

seasonal cycle) smaller or larger than business-cycle
fluctuations?
Business Cycles

Business-Cycle Theory

A good place to begin discussing business-cycle theory is by

discussing the procyclical productivity puzzle.

Procyclical Productivity Puzzle:

The data tell us that at business-cycle frequencies labor
hours move alot (SD=1.5 percent) and are procyclical but
that capital input moves little and is acyclical. If we adopt
a theory that features a neoclassical aggregate production
function, then what are some different possibilities to
produce procyclical labor productivity (Y /L)?
Business Cycles

A Problem:
Any theory with an unchanging, CRS production function
Yt = F (Kt , Lt ) with diminishing marginal products will be
problematic. We will see that this holds regardless of what
are the fundamental shocks (e.g. animal spirits,
government spending shocks, ...) driving business cycles
other than technology shocks and regardless of whether or
not consumer behavior in the face of such shocks is rational
or irrational.

Qualifying Proviso: capital input varies little in percentage

terms.
Business Cycles
Business Cycles

Two Possible Solutions:

1. Yt = At F (Kt , Lt ) and At varies.

2. Yt = F (Kt , Lt ) w/ increasing returns to labor

The next slide plots the same two data points on two
graphs. The data points are consistent with procyclical
productivity. The possible “solutions” proposed involve the
aggregate production function passing through the data
points, holding capital input unchanged.
Business Cycles
Business Cycles

Business-Cycle Theory

A large class of business-cycle theories are based on

“impulses” or shocks on the one hand and
“propagation mechanisms” on the other hand. This
follows Knut Wicksell who reportedly said:

“If you hit a wooden rocking horse with a club, the

movement of the horse will be very different to that
of the club.”
Business Cycles

Business-Cycle Theory
Impulses: technology shocks, government
spending-tax-default shocks, animal spirits,
erratic monetary policy, uncertainty shocks, news
about the probability of various shocks, sunspots,
financial shocks.
Impulses are viewed as exogenous

Propagation Mechanisms: physical capital,

human capital, future consumption is a normal
good, balance sheets of a firm, inventories,
input-output linkages across industries
Business Cycles

Business-Cycle Theory

While there are many different business-cycle

theories to consider, we will consider just two. They
will be vastly different in theoretical structure and in
policy implications.

* Theory 1: Technology Shock Theory

* Theory 2: Keynesian Animal Spirits Theory

Business Cycles

Theory 1: Life-Cycle Model w/ Technology

Shocks

Main Ingredients:
- rational choice: Yes ... consumers optimize
- technology: Yt = At F (Kt , Lt ) = At Ktβ L1−β
t
- impulses: At - technology shocks
- shock propagation: capital accumulation +
future consumption is a normal good
Business Cycles

Theory 1: A Permanent Increase in Technology Assume:

1. The Life-Cycle Model is the way the economy works.

2. Economy is at steady state at t = 1
3. Assume A1 = 1, A2 = 2, A3 = 2, A4 = 2, ...

To figure out what happens ... graph the law of motion.

 Business Cycles

n2.png
 Business Cycles

Permanent  Technology Shock: Life‐Cycle Model

35

30

25
Axis Title

20

15

10

0
0 2 4 6 8 10 12
Axis Title

Output Capital Wage Investment Technology

png
Business Cycles

Theory 1: Basic Story

The growth rate of technology At (as measured by the
Solow residual) is variable in data. This theory implies high
A is associated with high output Y , high wage W and high
Y /L. This pattern is supported in data.

The life-cycle model abstracts from a labor-leisure choice.

Thus, it does not get labor hours to vary. Procyclical labor
hours can be achieved by adding such a choice to the
model. However, much of the cyclical variation in labor
hours in the data comes via changes in employment
-unemployment. Modeling unemployment is way beyond
what we can do in this course.
Business Cycles

Theory 1: Policy

If the Life-Cycle model with shocks is the theory

of business cycles, then attempts to smooth out
business cycle fluctuations will not (with positive
real interest rates) lead to Pareto improvements.
Recall the Proposition established in the chapter
on the life-cycle model.
Business Cycles

Theory 2: Old-Time Keynesian Story

Main Ingedients:

- rational choice: None ... Keynes abandons

microeconomics
- technology: no production function
- impulses: animal spirits of investors are
exogenous impulses
- shock propagation: unclear as the model is
static and does not model time periods
Business Cycles

Rational Choice: None

“The fundamental psychological law, upon which

we are entitled to depend with great confidence
both a priori and from our knowledge of human
nature and from the detailed facts of experience,
is that men are disposed, as a rule and on the
average, to increase their consumption as their
income increases, but not by as much as the
increase in their income.” - Keynes (1936,
Chapter 8, p. 96.).
Business Cycles

Source of Shocks: Animal Spirits

“Even apart from the instability due to speculation, there
is the instability due to the characteristic of human nature
that a large proportion of our positive activities depend on
spontaneous optimism rather than on a mathematical
expectation, whether moral or hedonistic or economic.
Most, probably, of our decisions to do something positive,
the full consequences of which will be drawn out over many
days to come, can only be taken as a result of animal
spirits - of a spontaneous urge to action rather than
inaction, and not as the outcome of a weighted average of
quantitative benefits multiplied by quantitative
probabilities.” - Keynes (1936, Chapter 12, p. 161)
Business Cycles

Stylized Keynesian Model

C + I + G = Y - NIPA identity
C = a + b(Y − T ) - consumption (a > 0 and
0 < b < 1)
I - determined by animal spirits
T = G - assumption of a balanced budget

Solve for Output Y :

a+I+(1−b)G
a + b(Y − G) + I + G = Y or Y = 1−b
Business Cycles

Theory 2: Basic Story

The economy is buffeted by animal spirits of
investors. Thus, investment moves exogenously.
Times of low investment are times of low output
as the model implies (under the assumptions
above): Y = a+I+(1−b)G
1−b .
Times of low output are taken to be indicative of
a poorly functioning economy. A formal welfare
analysis is not undertaken (as it would in other
areas of economics) using the Pareto criteria and
the utility functions of the consumers.
Business Cycles

Theory 2: Policy

Typical advice: increase government spending (or

reduce taxes) in a recession.

Balanced Budget Multiplier: ∆Y

∆G =
1−b
1−b =1
Y = a+I+(1−b)G
1−b ⇒ ∆Y = (1−b)∆G
1−b

∆Y 1
Unbalanced Budget Multiplier: ∆G = 1−b >1
Y = a−bT1−b
+I+G ∆G
⇒ ∆Y = 1−b
Business Cycles

Government Spending Multipliers:

Keynesian Model: An increase in government spending

(whether balanced budget or deficit financed) increases
current output as both multipliers are POSITIVE. If the
multiplier is positive but less than 1 then (G, Y ) go up but
C falls. Model does not have dynamic multipliers as the
model is static.
Business Cycles

Government Spending Multipliers:

Life-Cycle Model: A multiplier is defined as the change in
output (at some horizon) due to the increased spending divided
by the change in government spending. To determine
multipliers, the output path under two different government
spending plans must be calculated. Basic Conclusion: current
multiplier is ZERO but the multiplier is NEGATIVE in all
future periods as the taxes/borrowing financing extra spending
depresses the capital stock.
NOTE: If a labor-leisure decision is added to the life-cycle
model, then the output multiplier could be POSITIVE. This
could occur if the negative income effect (from the increased
taxes funding government spending) leads to an increase in
labor that is sufficiently strong to offset the fall in savings from
the increase in taxes.
Business Cycles

Multipliers: Empirical Work

It is not surprising that different theories have
completely different multipliers or that one
theory could produce either positive or negative
multipliers under different assumptions.
Romer and Romer (2010) argue, using US data,
that the contemporaneous tax multiplier is
roughly zero but is negative at longer horizons.
Thus, an exogenous increase in taxes leads to
lower future output. We will discuss their work in
chaper 7.
Business Cycles

Gains to Business-Cycle Smoothing

Are Gains to Business-cycle Smoothing Large?

Although the source(s) of business-cycle fluctuations are
controversial, could we figure out whether the maximum
potential gain to (somehow) eliminating these fluctuations
is large?

Robert Lucas’ back-of-the-envelope calculation was that the

maximum potential gain was worth about $8.50 per person
per year! We will review the logic behind his calculation.
Business Cycles

Gains to Business-Cycle Smoothing

Digression: Expected Utility Theory

Ω = {1, 2, ..., N } - N possible distinct states of nature.

(x1 , x2 , ..., xN ) - A gamble that pays off xi units of
output if state i = 1, ..., N occurs.
(p1 , p2 , ..., pN ) - pi is the probability that state i occurs.
E[x] = N
P
i=1 xi pi - the expected value of a gamble
u(x) - the utility obtained for a sure payoff of x units
E[u(x)] = N
P
i=1 u(xi )pi - the expected utility of a
gamble
Business Cycles

Example 1: [Coin Toss]

Ω = {H, T }
(xH , xT ) = (90, 110)
(pH , pT ) = (0.5, 0.5)
u(x) = log(x)

Expected utility:
u(xH )pH + u(xT )pT = log(90) × 0.5 + log(110) × 0.5
Business Cycles

Example 1: [Coin Toss]

If the utility function u displays diminishing marginal utility,

then getting the expected value (or mean) of a gamble as a
sure payout is strictly prefered to taking the gamble. Utility
functions with this property display risk aversion.
How valuable is geting the mean for sure?

log(100) = log(90(1 + λ)) × .5 + log(110(1 + λ)) × .5

log(1+λ) = log(100)−log(90)×0.5+log(110)×0.5 = 0.00218

λ = 0.005 implies getting the mean is worth 1/2 percent of

consumption
Business Cycles

A Simplified Version of Lucas’ Argument:

1. Assume the government engineers the smooth trend
consumption rather than a risky consumption process with
the same period-by-period mean.
2. Assume agents are risk averse.
3. Quantify the gain achieved by such business-cycle
smoothing.
4. Need to measure consumption variation. A one standard
deviation movement of the cyclical component of aggregate
consumption from trend is about a 1.25 percent movement
over 1954-89.
5. Need knowledge of u.
Business Cycles

Specific Assumptions:

E[u(c)] = u(clow )P (low) + u(chigh )P (high)

Probability: P (low) = P (high) = 1/2
1−γ
Utility: u(c) = c1−γ , where γ > 1
Consumption: clow = 98 and chigh = 102
Business Cycles

Gain (λ) to getting c = 100 for sure?

E[U ((1 + λ)c)] = U (100)

(102(1+λ))1−γ (98(1+λ))1−γ (100)1−γ
1−γ (1/2) + 1−γ (1/2) = 1−γ

1001−γ
⇒ (1 + λ)1−γ =
1021−γ (1/2) + 981−γ (1/2)

1001−γ
⇒λ=[ ]1/(1−γ) − 1
1021−γ (1/2) + 981−γ (1/2)
Business Cycles

Table: Gain λ to Eliminating Aggregate Fluctuations:

Coefficient of Risk Aversion (γ) Compensation (λ)
γ=2 λ = .00040
γ=4 λ = .00080
γ = 10 λ = .00199

Lucas: $8.50 is a tenth of a percent of consumption/person in US

in 1983.
Table above says that if risk aversion γ ≤ 4, then the gain is less
than a tenth of a percent of consumption.
Business Cycles

How Risk Averse Are You?

Pick the row best matching your risk aversion. Each row describes
the maximum percentage of total wealth that a consumer with
1−γ
utility function U (c) = c1−γ is willing to give up to avoid an
even-odds gamble of gaining or losing a fraction α of total wealth.

Percentage of Wealth Given Up to Avoid a Proportional Wealth

Gamble of Size α
Coefficient of Risk Aversion (γ) α = 10% α = 30%
γ=1 0.5% 4.6%
γ=4 2.0% 16.0%
γ = 10 4.4% 24.4%
γ = 40 8.4% 28.7%
Business Cycles

Potential Reasons Why Lucas’ number is so

small:
1. Aggregate consumption movements over the
business cycle are small after WWII.
2. Individual households experience much larger
consumption fluctuations than those in aggregate
data.
3. Eliminating or reducing aggregate fluctuations
may reduce (but not eliminate) individual
fluctuations. If so, should start calculations with
the magnitude of individual fluctuations not