Lecture 5: Swaps in practice

Mark-Jan Boes

February 17, 2023

 Part I

Introduction

Lecture 5
Mark-Jan Boes 1
Introduction

If we would live in 2007 right now, we would be pretty much done

with the formal treatment of swaps and the usage of swaps in
hedging interest rate risk.

The fixed income world was considered to be risk free meaning

that the text book world closely resembled reality.

For example, the term structure implied by swap rates was close to
the term structure implied by Bunds, Dutch government bonds or
French government bonds.

We could take a lot of interest rate curves as a proxy for the risk
free interest rate curve.

Lecture 5
Mark-Jan Boes 2
Introduction

Hence, pension funds could construct the nominal matching

portfolio as follows:
provide the pension fund cash flows to an asset manager
give the swap based discount curve to this manager
ask the manager to replicate the return on the liabilities
the manager enters into a bunch of interest rate swaps and
uses the cash to generate the floating payment of swaps
the portfolio replicates carry, duration, curve positioning and
convexity of the liabilities
So, the end result would be a simple portfolio that tracks the
development of the liabilities very closely.

But then 2008 happened and the world experienced the

consequences of taking credit risk.
Lecture 5
Mark-Jan Boes 3
Introduction

Counterparty risk enters the problem in different ways:

counterparty credit risk in the interest rate swap contract:
what happens if the counterparty in the swap contract goes
bankrupt? And what is the impact of counterparty credit risk
on valuation?
the floating rate of an interest rate swap usually is a rate
that contains a compensation for counterparty risk

In the remainder of this lecture we’ll dig further into the issue of
counterparty credit risk, with the focus on interest rate swaps.

Lecture 5
Mark-Jan Boes 4
 Part II

Swap valuation

Lecture 5
Mark-Jan Boes 5
Swap valuation

We can express the valuation formula of an IRS (receive fixed, pay

floating) as follows:
n
Vtswap =
X  
P(t, Ti ) × N × K − ft:Ti−1 :Ti .
i=1

Hence, take the interest rate curve from the market, calculate
discount factors, calculate forward rates and pricing follows easily.

In traditional pricing it is not necessary to calculate forward rates,

i.e. the formula above can be rewritten such that the forward rates
disappear (see previous lecture: decomposition of interest rate
swap in a fixed coupon bond and a floating rate bond).

Lecture 5
Mark-Jan Boes 6
Swap valuation

The valuation formula works because we have assumed the

absence of counterparty credit risk, i.e. both parties in the
contract can be sure that the other party is going to pay the
coupons.

In the valuation formula the absence of counterparty credit risk is

explicitly visible because the swap rate K at all Ti is discounted
with the price of the risk-free zero-coupon bond Ti .

Lecture 5
Mark-Jan Boes 7
Swap valuation

Implicitly we have made a lot of assumptions in the pricing

approach of the previous lecture, among which:
there is no risk that one of the parties in the deal doesn’t pay
all inputs are derived from the same curve: the basis for the
discounting curve and the forward curve is exactly the same
the floating rate is a risk free rate

These assumptions were close to reality before 2008. However,

today we know better than that...

Lecture 5
Mark-Jan Boes 8
Risks in swaps

Is counterparty credit risk relevant in practice?

Suppose that the swap portfolio of a pension fund has a DV01 of

EUR 20 million and that interest rates drop with 100bps.

Then the market value of the swaps would have been risen with
approximately EUR 2 billion (ignoring convexity).

Also imagine that all those swaps would have been traded with one
counterparty (Lehman Brothers?!)...

Would you sleep during the night?

Lecture 5
Mark-Jan Boes 9
Swap valuation

The biggest risk we have to deal with in practice when using

interest rate swaps is counterparty credit risk. This risks leads to
a number of questions:
suppose the floating rate in the swap is the risk free rate but
there is the possibility that the counterparty goes
bankrupt, how would that influence the value of the swap?
suppose you would be able to get rid of all counterparty credit
risk but the floating rate in the swap contains credit risk,
how would that impact valuation?

Lecture 5
Mark-Jan Boes 10
Swap valuation

The answer to the first question is already complex: we need to

assess the creditworthiness of the counterparty and determine
how this risk should be included in the swap rate and in the value
of the swap contract after inception.

The second situation is most relevant for institutional investors:

later in this lecture we will look into this question in more detail.

We’ll first introduce counterparty credit risk into the problem and
then look at what we do in practice to make the contract risk free
again.

Lecture 5
Mark-Jan Boes 11
 Part III

Swaps in the new reality

Lecture 5
Mark-Jan Boes 12
Swaps: counterparty credit risk

As said, marking the swap positions to the market leads to

substantial counterparty credit risk.

Two consequences for the balance sheet:

Direct consequences for P&L: If the counterparty in the deal
defaults, then the other party loses actual market value of the
swaps and the swap position.
The non-defaulting party loses exposure to interest rate
movements: (1) the market could move away from this party
and (2) this party wants to buy back this exposure under
unfavourable market circumstances

Lecture 5
Mark-Jan Boes 13
Swaps: counterparty credit risk

If no measures are taken to mitigate counterparty credit risk the

traditional valuation formula is simply wrong.

Why? Because in traditional valuation only risk free bonds and risk
free interest rates are used but we just said that the instrument is
not risk free.

Suppose I would enter into an interest rate swap with a bank, I

receive fixed, I am non-defaultable but the bank is not. Would I
allow the bank to pay me a risk free rate as fixed rate?

No! I would definitely require a higher rate (or would be willing to

pay less in floating). I would be charging the bank a kind of credit
spread.

Lecture 5
Mark-Jan Boes 14
Swaps: counterparty credit risk

A different, and more fundamental, approach is to determine the

default probability of the defaultable party and apply a formula
that could possibly look like:
n
Vtswap =
X  
P(t, Ti ) K × Qt (τ ≥ Ti ) − ft:Ti−1 :Ti
i=1

where Qt (τ ≥ t) is the risk-adjusted probability that the

counterparty defaults after time t.

In this specific case the fixed rate payer is not risk free. In the
example of the previous slide this is the bank.

I calculate K such that the value of the total structure is equal to

zero at initiation.
Lecture 5
Mark-Jan Boes 15
Swaps: counterparty credit risk

Hence, in the case of a non-zero default probability, i.e. in the

presence of counterparty credit risk, the fair swap rate K differs
from the fair swap rate K in the text book world.

Connection with the replication argument: the swap can be

decomposed in a short position in a non-defaultable floating rate
bond and a long position in a defaultable coupon bond.

As a result, a swap rate K is not necessarily a good proxy of the

risk-free rate.

Final remark: in the credit risk literature, the difference between

the swap value in a risk-free setting and a setting with
counterparty credit risk is called credit value adjustment.

Lecture 5
Mark-Jan Boes 16
Swaps: counterparty credit risk

In order to mitigate counterparty risk, collateral arrangements

can be put in place.

Collateral management: if the value of the swap is negative for

the counterparty, the counterparty must post the market value of
the swap as collateral to you. If the counterparty defaults you
become the legal owner of the collateral.

Most important details of the collateral agreements are (1) the

frequency of collateral posting and (2) the eligible collateral.

Detailed arrangements are written down in so-called ISDA / CSA

documentation.

Lecture 5
Mark-Jan Boes 17
Swaps: counterparty credit risk

A CSA defines the terms or rules under which collateral is posted

or transferred between swap counterparties to mitigate credit risk
arising from ’in the money’ derivatives positions.

Source: http://www.isda.org/

Lecture 5
Mark-Jan Boes 18
Swaps: counterparty credit risk

CSA contain agreements on:

eligible collateral (which currencies, bonds etc.)

interest obligation on posted collateral
thresholds

CSA agreements have consequences for swap valuation.

Lecture 5
Mark-Jan Boes 19
Swaps: counterparty credit risk

Intuitively, it works as follows.

Consider the situation in which you have a swap that is worth 20

million dollars.

Whoever your counterparty is, you would feel more comfortable if

the counterparty would post collateral to you: assets of which you
become legal owner in case the counterparty defaults.

Economic ownership stays with the party that posts collateral,

meaning that the receiver must return the coupons (in case of
bond collateral) or must pay interest (in case of cash collateral).

With this collateral, the asset has been made safer (i.e.
counterparty risk is reduced) and therefore has a higher value than
without collateral.
Lecture 5
Mark-Jan Boes 20
Swaps: counterparty credit risk

For all derivatives: the more often collateral and the better the
collateral is, the less a default of a counterparty will hurt you.

If for instance, collateral is exchanged on a daily basis and only

non-defaultable assets are allowed as collateral, there is virtually no
pain in case a counterparty defaults.

In credit language: loss given default is virtually zero.

The original pricing formula can be applied in this case (’full and
safe collateralization’).

Lecture 5
Mark-Jan Boes 21
Swaps: counterparty credit risk

Assume now that we are in a situation of full and safe

collateralization.

You might expect that the zero rates implied by fully collateralized
swap rates and risk free zero rates (e.g. zero rates implied by U.S.
Treasury bonds of German Bunds) would be very close.

In that case we could use the swap implied zero rates for all kind of
pricing exercises.

Lecture 5
Mark-Jan Boes 22
Swaps: counterparty credit risk

This is what we see in practice as of Jan 26, 2015:

Lecture 5
Mark-Jan Boes 23
 
Swaps: counterparty credit risk

How is this possible?

How can it be that three curves derived from instruments that are
virtually risk free, are so different?

Lecture 5
Mark-Jan Boes 24
Swaps: counterparty credit risk

It has to do with the following figure:

Yield (%)
6

5 6‐mths Euribor
6‐mths EONIA
4

‐1

Lecture 5
Mark-Jan Boes 25
Swaps: counterparty credit risk

Libor stands for London Interbank Offer Rate. It is intended to be

the average secured funding rate at which a contributor bank can
obtain unsecured funding in the London interbank market, for a
given period, in a given currency.

Each contributor is asked on a daily basis: at what rate could you

borrow funds, were you to do so, by asking for and then accepting
interbank offers in a reasonable market size just prior to 11am?

These indications are then collected and highest and lowest 25%
are removed.

Important: Libor is based on an opinion and not on actual

transactions.

Lecture 5
Mark-Jan Boes 26
Swaps: counterparty credit risk

Libor is best known for a scandal that broke in 2012: it emerged

that banks and traders had been illicitly manipulating Libor rates
for years.

As a consequence Libor rates, and reference rates in general, have

been reformed and/or will be reformed or will be discontinued:
Example: Euribor, which is the commonly used floating rate
in European interest rate swaps, has been reformed using a
hybrid approach which makes use of actual transactions as
much as possible
Example: Eonia (Euro OverNight Index Average) has been
replaced by Estr on January 1, 2022.

Lecture 5
Mark-Jan Boes 27
Swaps: counterparty credit risk

Floating rate in swaps usually is based on Libor (in Europe:

Euribor)

What does the picture show us?

6-months Euribor
Swap rate of a 6-months EUR swap in which the floating rate
is an overnight interest rate

For the overnight interest rate we use Eonia in Europe.

Lecture 5
Mark-Jan Boes 28
Swaps: counterparty credit risk

The picture clearly indicates that there are periods in which there
is a credit risk component in the Euribor rate, otherwise the
difference between couldn’t be so large at some points in time.

I.e., Euribor is not a good representation of the true risk free rate.

But that gives an answer to our earlier question: if in a risk free

interest rate swap one party needs to pay a rate (Euribor) that is
higher than the risk free rate why would she accept to receive the
risk free rate? She would definitely demand a rate that is higher
than the risk free rate.

Hence, zero rates implied by swap rates are definitely not a

good representation of the risk free zero curve.

Lecture 5
Mark-Jan Boes 29
Swaps: counterparty credit risk

What to do then? For pricing derivatives we need a risk free curve

for discounting.

We could try to use Overnight-Indexed-Swaps (OIS) although

there is a problem with liquidity on the long end of the curve.

What is an overnight indexed swap?

Very similar to a ’regular’ swap but the floating payments are

based on the overnight rate (Fed Funds in the US, Eonia in
Europe, Sonia in the UK)

We can bootstrap the discounting curve from OIS swap rates in a

similar way as from Euribor swap rates.

Lecture 5
Mark-Jan Boes 30
Swaps: counterparty credit risk

And swap curves as of January 31, 2014

Lecture 5
Mark-Jan Boes 31
Swaps: counterparty credit risk

A couple of important remarks on this:

The difference between an OIS swap rate and a Euribor swap
rate is not because of the difference in credit risk between
both swaps but because of a difference in credit risk between
the floating basis of the swaps.
The value of Euribor swaps with full collateralization, still the
most liquid type of swaps, depends on two curves: the Euribor
curve for future expected payments and the OIS-curve for
discounting.
Although there are a lot of confusing stories about this issue:
theoretically, you should always use the best proxy for the risk
free rate for discounting and make a proper adjustment in the
expected cash flows for credit risk.

In my opinion, this story has been made far too complicated.

Lecture 5
Mark-Jan Boes 32
Swaps: counterparty credit risk

Let’s go back to the original pricing formula:

n
Vtswap =
X  
P(t, Ti ) K − ft:Ti−1 :Ti
i=1

In today’s reality (under full and safe collateralization):

the discount factors P(t, i) are zero rates implied by OIS swap
rates
the forward rates are based on zero rates implied by Euribor
swap rates

We cannot apply the relation between zero rates and forward

rates anymore to simplify the valuation of the floating leg.

Lecture 5
Mark-Jan Boes 33
Swaps: counterparty credit risk

In the new reality it would be better to write the pricing formula as

follows (for Euribor-swaps and under full and safe collateralization):

n h i
Vtswap
X
= P(t, Ti )OIS K − ft:T
EURIBOR
:T
i−1 i
i=1

Very important: nowadays we need two different curves to

price an interest rate swap.

And...in contrast to the traditional pricing method we need to

calculate forward rates (from the Euribor-curve) over the full life
time of the swap.

Lecture 5
Mark-Jan Boes 34
Swaps: counterparty credit risk

The following steps should be taken to calculate the value of a

swap in the new reality:
Take OIS-swap rates from the market
Bootstrap the OIS zero curve; this gives the risk free curve for
discounting
Take Euribor-swap rates from the market
Bootstrap the Euribor zero curve
Determine the Euribor forward curve from the Euribor zero
curve
Calculate present value fixed leg using OIS discounting curve
Determine (risk neutral) cash flow projection floating leg using
Euribor forward curve
Discount (risk neutral) cash flows using OIS discounting curve.

Nowadays, this is market standard for swap pricing.

Lecture 5
Mark-Jan Boes 35
Swaps: counterparty credit risk

In the new reality an interest rate swap still has zero value at
initiation.

However, both legs of the swap have a value above par, i.e.
assuming a notional amount of 1, both the fixed leg and the
floating leg have a starting value that is greater than 1.

Lecture 5
Mark-Jan Boes 36
Swaps: counterparty credit risk

So, we started with the issue of counterparty credit risk:

we can hedge this risk by full and safe collateralization of
swap P&L
valuation becomes more complicated because the floating leg
is based on an interest rate that contains a credit risk premium

However:
collateral management introduces new risks, especially
liquidity risk and the risk of a big collateral pool
valuation seems transparent but still depends on collateral
arrangements

Due to time constraints I’ll only focus on liquidity risk.

Lecture 5
Mark-Jan Boes 37
Swaps: collateral liquidity risk

While mitigating counterparty risk, collateral agreements increase

liquidity risk due to the fact that you might have to post collateral.

In swap positions there is always some liquidity management

needed because of the regular coupon payments.

However, looking after liquidity on the balance sheet becomes

much more important in case of collateral arrangements.

Lecture 5
Mark-Jan Boes 38
Liquidity risk: an example

Consider the following pension fund:

total assets: 10bln
notional swaps (receive fixed): 4bln
duration swaps: 25 years
BPV swaps: 4bln x 25 / 10000 = 10,000,000
P&L swaps: 0
daily collateral management
eligible collateral: Euro cash
Assume that today interest rates increase with 30 bp. Then the
pension fund must post 10,000,000 x 30 = 300,000,000 Euro cash
as collateral.

This is 3% of total assets!

Lecture 5
Mark-Jan Boes 39
Liquidity risk: an example

Liquidity problems almost led to bankruptcy of Vestia, a large

Dutch housing corporation.

Vestia had quite a large position in swaps (pay fixed), because they
anticipated a rise in interest rates.

However, interest rates went down considerably.

Hence, the value of their swaps was negative so they had to post a
lot of collateral.

The asset side of the balance sheet didn’t have sufficient liquidity
to fulfill the collateral obligations.

Lecture 5
Mark-Jan Boes 40
Liquidity risk: an example

The Vestia got a lot of media attention. The media wrote about
’huge losses on derivatives positions’.

In essence, there is nothing wrong with a huge loss on derivatives

positions that are used as hedge. There should be an offsetting
gain elsewhere.

It is even not so wrong to build up a swap position to anticipate on

a rise of future interest rates in case you expect to have to raise
debt in the near future.

However, something went terribly wrong in decision making.

Vestia (or their financial advisors) should have done a proper

stress-test, in order to see if the derivatives position could
potentially lead to liquidity problems.
Lecture 5
Mark-Jan Boes 41
Liquidity risk: an example

So risk management failed....

But please notice that without any collateral arrangements there

would have been no problem at all with liquidity.

Hence, ironically mitigating counterparty default risk almost led

Vestia into bankruptcy.

Lecture 5
Mark-Jan Boes 42
Liquidity risk: an example

What could they have done instead if they really wanted to have
the market exposure in swaps?

entering into swaps without collateral arrangements

entering into payer swaptions (options on swaps): because a
premium is paid upfront there are no collateral obligations

I will discuss swaptions briefly in one of the other lectures.

Lecture 5
Mark-Jan Boes 43
Swaps: Liquidity risk

The obvious conclusion is that a proper assessment of liquidity risk

is of utmost importance in case there are any zero-cost derivatives
in the investment portfolio, i.e. not only counterparty credit risk is
important.

In practice we conduct a weekly liquidity stress-test: we combine a

shock on foreign exchange, interest rates and equities and
determine the possible liquidity and collateral needs.

Lecture 5
Mark-Jan Boes 44
Consequences

From the previous lecture you would think that building a

matching portfolio that follows the short term movements of the
liabilities is quite simple.

If you would use the swap curve (for liquidity reasons) as basis for
discounting future expected cash flows then you could build a
portfolio of swaps and cash that replicates the carry, (key rate)
duration and convexity of the liabilities.

In a world without credit risk the cash would generate the floating
rate in the swap contract and you wouldn’t need to be concerned
about a large market value in swap contracts.

Lecture 5
Mark-Jan Boes 45
Consequences

Credit risk changes everything:

which swap curve do you take as a basis to calculate the
present value of liabilities? If you’d use the swap curve with
floating rate 6-months Euribor then in the matching portfolio
credit risk needs to taken in order to replicate the carry
swaps lead to basis risk versus the liabilities because the value
of swaps depends on two curves and the liabilities only on one
curve
do you still like solutions with many swaps? Even in the
presence of full and safe collateralization such a solution gives
large exposures to uncontrollable risks as liquidity risk and
valuation risk
The third bullet explains why we usually see large positions in long
term government bonds in matching portfolios: this provides
diversification across different risks.
Lecture 5
Mark-Jan Boes 46
Consequences

Using many long term bonds in the matching portfolio leads to

tracking risk.

What happens for instance when:

the swap curve moves down
government curves remain unchanged

Well, then:
the value of the liabilities goes up, i.e. positive return on the
liabilities
if the matching portfolio contains many bonds, the movement
of this portfolio will be less positive than the liabilities
consequence: the portfolio generates an underperformance
versus the benchmark with an impact on the funding ratio
Lecture 5
Mark-Jan Boes 47
Consequences

Recall that the instantaneous return on a non-defaultable zero

coupon bond can be expressed as:
P(t + h) − P(t) 1
≈ yt h−(T −t)(yt+h −yt )+ (T −t)2 )(yt+h −yt )2
P(t) 2
A similar formula can be derived for a bond P̃ that has a spread s
versus P:
P(t ˜+ h) − P̃(t)
≈ (yt + st )h − (T − t)(yt+h − yt + (st+h − st ))+
P̃(t)
1
+ (T − t)2 (yt+h − yt + (st+h − st ))2
2
If we would then subtract the P from P̃ then we get (ignoring
second order term):
P(t + h) − P(t) P(t ˜+ h) − P̃(t)
− = s(t)h−(T −t)(s(t+h)−s(t))
P(t) P̃(t)
Lecture 5
Mark-Jan Boes 48
Consequences

Hence a spread widening (ds positive) has a negative impact on

relative return.

We see from the formula that the impact of spread movements is

related to the duration of the asset, i.e. long term bonds will give
more tracking risk versus a swap-based benchmark than short term
bonds.

Therefore, a replicating portfolio with many swaps and many short

term physical assets has the lowest spread duration and therefore
lowest tracking risk.

Lecture 5
Mark-Jan Boes 49
Summary

What are the main lessons from this lecture?

valuation of a simple interest rate swap is in practice much
more complicated than shown in text books
even in the case of full collateralization with safe assets in the
same currency as the swap the text book arguments do not
hold: in this case we already need two curves to determine the
swap’s fair value
the zero curve implied by OIS swap rates is a fair
representation of the risk free curve
theoretically speaking, this zero curve should always be used
for discounting under the risk neutral measure
while mitigating the one risk, other risks might become more
prominent
...and last but not least: never trust banks!
Lecture 5
Mark-Jan Boes 50
Literature

Recommended literature:
Slides
Paper Hull and White, LIBOR vs. OIS: The Derivatives
Discounting Dilemma, p.1-8 and Section VI (conclusions)
Article Risk Magazine: Goldman and the OIS gold rush
Hull (11th Edition): 7.11, 7.12, 9.1

Lecture 5
Mark-Jan Boes 51