• • • • • • • • • •

• • • • • • • • • •

• • • • • • • • • •

Fundamental Review of the

Trading Book
Learning Objectives
After completing this reading, you should be able to:

• Describe the changes to the Basel framework for calculat

• Explain the FRTB revisions to Basel regulations in the fol
ing market risk capital under the Fundamental Review of
lowing areas:
the Trading Book (FRTB) and the motivations for these
- Classification of positions in the trading book
changes.
compared to the banking book.
- Backtesting, profit and loss attribution, credit risk, and
• Compare the various liquidity horizons proposed by the
securitizations.
FRTB for different asset classes and explain how a bank
can calculate its expected shortfall using the various
horizons.

Excerpt is Chapter 27 of Risk Management and Financial Institutions, Sixth Edition, by John C. Hull.

205
In May 2012, the Basel Committee on Banking Supervision
shortfall is J.l + 2.338CJ'.4 (See Problem 18.2.) For non-normal
issued a consultative document proposing major revisions to
distributions, they are not equivalent. When the loss distribution
the way regulatory capital for market risk is calculated. This is
has a heavier tail than a normal distribution, the 97.5% ES can
referred to as the "Fundamental Review of the Trading Book"
be considerably greater than the 99% VaR.
(FRTB).1 The Basel Committee then followed its usual process
of requesting comments from banks, revising the proposals, Under FRTB, the 10-day time horizon used in Basel I and Basel
2
and carrying out Quantitative Impact Studies (QISs). The final 11.5 is changed to reflect the liquidity of the market variable
version of the rules was published by the Basel Committee in being considered. FRTB considers changes to market variables
3
January 2019. The internationally agreed implementation date that would take place (in stressed market conditions) over peri
is January 2023, but many jurisdictions indicated that they will ods of time reflecting their liquidity. The changes are referred

be a year or two later than that. to as shocks. The market variables are referred to as risk factors.
The periods of time considered are referred to as liquidity hori
FRTB's approach to determining capital for market risk is zons. Five different liquidity horizons are specified: 10 days, 20
much more complex than the approaches previously used by days, 40 days, 60 days, and 120 days. The allocation of risk fac
regula tors. The purpose of this chapter is to outline its main tors to these liquidity horizons is indicated in Table 18.1.
features.
FRTB specifies both a standardized approach and an inter-
nal models approach for calculating market risk capital. Even
18.1 BACKGROUND
when banks have been approved to use the internal models

The Basel I calculations of market risk capital were based on a approach, they are required by regulators to calculate required
value at risk (VaR) calculated for a 10-day horizon with a 99% capital under both approaches. This is consistent with the Basel
confidence level. The VaR was "current" in the sense that cal Committee's plans to use standardized approaches to provide
culations made on a particular day were based on the behavior a floor for capital requirements. As discussed in Section 26.4,
of market variables during an immediately preceding period of in December 2017, the Basel Committee announced a move to
time (typically, one to four years). Basel 11.5 required banks to a situation where total required capital is at least 72.5% of
calculate a "stressed VaR" measure in addition to the current that given by standardized approaches. It will achieve this by
measure. As explained in Sections 12.1 and 26.1, this is VaR 2028 with a five-year phase-in period. These changes are a
where calculations are based on the behavior of market vari culmina tion of a trend by the Basel Committee since the
ables during a 250-day period of stressed market conditions. 2008 crisis to place less reliance on internal models and to
To determine the stressed period, banks were required to go use standardized models to provide a floor for capital
back through time searching for a 250-day period where the requirements.
observed movements in market variables would lead to signifi
A difference between FRTB and previous market risk regulatory
cant financial stress for the current portfolio.
requirements is that most calculations are carried out at the
FRTB changes the measure used for determining market risk trading desk level. Furthermore, permission to use the internal
capital. Instead of VaR with a 99% confidence level, it uses models approach is granted on a desk-by-desk basis. Therefore
expected shortfall (ES) with a 97.5% confidence level. The mea it is possible that, at a particular point in time, a bank's foreign
sure is actually stressed ES with a 97.5% confidence. This means currency trading desk has permission to use the internal models
that, as in the case of stressed VaR, calculations are based on approach while the equity trading desk does not.
the way market variables have been observed to move during
In earlier chapters, we saw how the ways in which capital is
stressed market conditions.
calculated for the trading book and the banking book are quite
For normal distributions, VaR with a 99% confidence and ES with different. This potentially gives rise to regulatory arbitrage
a 97.5% confidence are almost exactly the same. Suppose losses where banks choose to allocate instruments to either the trading
have a normal distribution with a mean µ. and standard deviation book or the banking book so as to minimize capital. In Basel
CJ'. The 99% VaR isµ. + 2.326CJw' hile the 97.5% expected 11.5, the incremental risk charge made this less attractive. FRTB
counteracts regulatory arbitrage by defining more clearly than
1 previously the differences between the two books.
See Bank for International Settlements, "Consultative Document:
Fundamental Review of the Trading Book," May 2012.
2
QISs are calculations carried out by banks to estimate the impact of
proposed regulatory changes on capital requirements. 4
From equation (11.2), the ES for a normal distribution with meanµ,
3
See Bank for International Settlements, "Minimum Capital and standard deviation a is,, + aexp(-Y2/2)/[ (1 - X)]
Requirements for Market Risk," January 2019. where
Xis the confidence level and Y is the point on a normal distribution that
has a probability of 1 - X of being exceeded. This can also be written
2 Financial Risk Manager Exam Part II: Market Risk Measurement and Management
µ, + u-2f(VaR)/(1 - X) where f is the probability density function for the loss.
 iffl!ffl!ll:JI Allocation of Risk Factors to Liquidity The delta risk charge for a risk class is calculated using the
Horizons risk weights and weighted sensitivity approach described in
Section 13.6:
Risk Factor Horizon
(days)
Delta Risk Charge= LL > P;/3,oiW;W; (18.1)
; j
Interest rate (dependent on currency) 10-60
Interest rate volatility 60 where the summations are taken over all risk factors in the risk
Credit spread: sovereign, investment grade 20 class. The risk weights, W;, and the correlations between risk
factors, Pij• are determined by the Basel Committee.5 The sensi
Credit spread: sovereign, non-investment 40
tivities (or deltas), o;, are determined by the bank. In the case of
grade
risk factors such as equity prices, exchange rates, or commod ity
Credit spread: corporate, investment grade 40
prices, the deltas measure the sensitivity of the portfolio to
Credit spread: corporate, non-investment 60 percentage changes. For example, if a 1% increase in a com
grade
modity price would increase the value of a portfolio by $3,000,
Credit spread: other 120 the delta would be 3,000/0.01 = 300,000. In the case of risk
Credit spread volatility 120 factors such as interest rates and credit spreads, the deltas are
defined in terms of absolute changes. For example, if the effect
Equity price: large cap 10
of an interest rate increasing by one basis point (0.0001) is to
Equity price: small cap 20
reduce the value of a portfolio by $200, the delta with respect
Equity price: large cap volatility 20 to that interest rate would be -200/0.0001 = -2,000,000.
Equity price: small cap volatility 60 Consider how the risk weights, W;, might be set by regulators.
Equity: other 60 Suppose first that all risk factors are equity prices, exchange
Foreign exchange rate (dependent on 10-40 rates, or commodity prices, so the deltas are sensitivities
currency) to percentage changes. If W; were set equal to the daily
Foreign exchange volatility 40 volatility of risk factor i for all i, the risk charge in equation
(18.1) would equal the standard deviation of change in the
Energy price 20
value of the portfolio per day. If W; were set equal to the daily
Precious metal price 20 volatility of risk factor i in stressed market conditions (the
Other commodities price 60 stressed daily volatility) for all i, equation (18.1) wouId give
Energy price volatility 60 the standard deviation of the daily change of the portfolio
in stressed market conditions. In practice, the W; are set
Precious metal volatility 60
equal to multiples of the stressed daily volatility to reflect
Other commodities price volatility 120
the liquidity horizon and the confidence level that regulators
Commodity (other) 120 wish to consider. Suppose that the stressed daily volatility
of risk factor i is estimated as 2% and that the risk factor has
a 20-day liquidity horizon. The risk weight might be set as
0.02 x V20 X 2.338 = 0.209. (Note that the 2.338 multiplier
18.2 STANDARDIZED APPROACH reflects the amount by which a standard deviation has to be
multiplied to get ES with a 97.5% confidence when a normal
Under the standardized approach, the capital requirement is the distribution is assumed.)
sum of three components: a risk charge calculated using a risk
Now suppose that the risk factors are interest rates and credit
sensitivity approach, a default risk charge, and a residual risk
spreads so that deltas are sensitivities with respect to actual
add-on.
changes measured in basis points. The W; for risk factor i is set
Consider the first component. Seven risk classes (corresponding equal to a multiple of the stressed daily standard deviation for
to trading desks) are defined (general interest rate risk, foreign all i. If the multiple were 1, the formula would give the standard
exchange risk, commodity risk, equity risk, and three categories
of credit spread risk). Within each risk class, a delta risk charge,
5
vega risk charge, and curvature risk charge are calculated. Banks are required to test the effect of multiplying the correlations
specified by the Basel Committee by 1.25, 1.00, and 0.75 and then set
the capital charge equal to the greatest result obtained.

4 Financial Risk Manager Exam Part II: Market RiskChapter

Measurement and Management
18 Fundamental Review of the Trading Book I 207
deviation of the value of the portfolio in one day. In practice the
multiple is determined as just described to reflect the liquidity
18.2.2 Curvature Risk Charge
horizon and confidence level. The curvature risk charge is a capital charge for a bank's gamma
risk exposure under the standardized approach. Consider the
Vega risk is handled similarly to delta risk.6 A vega risk charge
exposure of a portfolio to the ith risk factor. Banks are required
is calculated for each risk class using equation (18.1). The
to test the effect of increasing and decreasing the risk factor by
risk factors (counted by the i and j) are now volatilities. The
its risk weight, W;. If the portfolio is linearly dependent on the
summation is taken over all volatilities in the risk class. The
risk factor, the impact of an increase of W; in the risk factor is
parameter 8; is actually a vega. It is the sensitivity of the value
7
of the portfolio to small changes in volatility i. The param W;B;. Similarly, the impact of a decrease of W; in the risk factor

eter Pii is the correlation between changes in volatility i and is -B;W;. To evaluate the impact of curvature net of the delta
effect, the standardized approach therefore calculates
volatility j, and W; is the risk weight for volatility i. The latter
is determined similarly to the delta risk weights to reflect the 1. W;B; minus the impact of a increase of W; in the risk factor, and
volatility of the volatility i, its liquidity horizon, and the confi
2. -W;B; minus the impact of a decrease in the risk factor of W;.
dence level.
The curvature risk charge for the risk factor is the greater of
There are assumed to be no diversification benefits between risk
these two. If the impact of curvature net of delta is negative, it
factors in different risk classes and between the vega risks and
is counted as zero. The calculation is illustrated in Figure 18.1.
delta risks within a risk class. The end product of the calculations
In Figure 18.1a, the portfolio value is currently given by point
we have described so far is therefore the sum of the delta risk
0. If there were no curvature, an increase of W; in the risk
charges across the seven risk classes plus the sum of the vega
factor would lead to the portfolio value at point C, whereas a
risk charges across the seven risk classes.
decrease of W; in the risk factor would lead to the portfolio
value at point
18.2.1 Term Structures A. Because of curvature, an increase of W; leads to the portfo
lio value at point D, and a decrease of W; leads to the portfo-
In the case of risk factors such as interest rates, volatilities,
lio value at point B. Since AB > CD, the risk charge is AB. In
and credit spreads, there is usually a term structure defined
Figure 18.1b, the risk charge is zero because curvature actually
by a number of points. For example, an interest rate term increases the value of the position (relative to what delta
structure is sometimes defined by 10 points. These are would suggest) for both increases and decreases in the risk
the zero-coupon interest rates for maturities of 3 months, factor. (Figure 18.1a could correspond to a short position in an
6 months, 1 year, 2 years, 3 years, 5 years, 10 years, 15 option;
years, 20 years, and 30 years. Each vertex of the term Figure 18.1b could correspond to a long position in an option.)
structure is a separate risk factor for the purposes of using
When there are several risk factors, each is handled similarly
equation (18.1). The delta of a portfolio with respect to a one
to Figure 18.1. When there is a term structure (e.g., for interest
basis point move in one of the vertices on the term structure is
rates, credit spreads, and volatilities), all points are shifted by
calculated by increasing the position of the vertex by one
the same amount for the purpose of calculating the effect of
basis point while making no change to the other vertices. The
curvature. The shift is the largest W; for the points on the term
Basel Committee defines risk weights for each vertex of the
structure. In the case of an interest rate term structure, the W;
term structure and correlations between the vertices of the
corresponding to the three-month vertex might be the largest
same term structure.
W;, so this would define an upward and downward parallel
A simplification is used when correlations between points on shift in the term structure. The delta effect is removed for each
different term structures are defined. The correlations between point on the term structure by using the i:l; for that point.
point A on term structure 1 and point B on term structure 2 are
The curvature risk charges for different risk factors are combined
assumed to be the same for all A and B.
to determine a total curvature risk charge. When diversification
benefits are allowed, aggregation formulas broadly similar to
those used for deltas are used with correlations specified by the
Basel Committee.
6
This works well because most of the value of a derivative is in many
cases approximately linearly dependent on volatility.
Banks
7 can choose whether it is percentage or actual changes in volatil
ity that are considered. 18.2.3 Default Risk Charge
Risks associated with counterparty credit spread changes are
handled separately from risks associated with counterparty
 (a)
Portfolio Portfolio value (b)

value

Risk factor i Risk factor i

lif.llWl11II Calculation of curvature risk charge for a risk factor

In Figure 18.1a, the curvature risk charge is AB; in Figure 18.1b, it is zero.

defaults in FRTB. In the standardized approach, credit spread

For example, vega and gamma risk do not have to be consid
risks are handled using the delta/vega/curvature approach
ered. This should make FRTB more attractive to jurisdictions
described earlier. Default risks, sometimes referred to as jump
such as the United States that have many small banks that tend
to-defau/t (JTD) risks, are handled by a separate default risk
to enter into only relatively simple transactions.
charge. This is calculated by multiplying each exposure by a
loss given default (LGD) and a default risk weight. Both the
LGD and the risk weight are specified by the Basel Committee.
For exam ple, the LGD for senior debt is specified as 75% and
18.3 INTERNAL MODELS APPROACH
the default risk weight for a counterparty rated A is 3%. Equity
The internal models approach requires banks to estimate
positions are subject to a default risk charge with an LGD =
stressed ES with a 97.5% confidence. FRTB does not prescribe a
100%. Rules for offsetting exposures are specified.
particular method for doing this. Typically the historical simula
tion approach is likely to be used. Risk factors are allocated to
18.2.4 Residual Risk Add-On liquidity horizons as indicated in Table 18.1. Define:

Category 1 Risk Factors: Risk factors with a time horizon of

The residual risk add-on considers risks that cannot be
10 days
handled by the delta/vega/curvature approach described
earlier. It includes exotic options when they cannot be Category 2 Risk Factors: Risk factors with a time horizon of
considered as linear combinations of plain vanilla options. The 20 days
add-on is calcu lated by multiplying the notional amount of the
Category 3 Risk Factors: Risk factors with a time horizon of
transaction by a risk weight that is specified by the Basel
40 days
Committee. In the case of exotic options the risk weight is 1%.
Category 4 Risk Factors: Risk factors with a time horizon of
60 days
18.2.5 A Simplified Approach Category 5 Risk Factors: Risk factors with a time horizon of
In this section, we have described the standardized approach 120 days
that the Basel Committee requires all large banks to use. In June As we shall see, all calculations are based on considering
2017 the Basel Committee published a consultative document 10-day changes in the risk factors. In Basel I and Basel
outlining a simplified standardized approach that it proposes for 11.5,
smaller banks.8 This has been included in the final January 2019 banks are allowed to deduce the impact of 10-day changes
document. The full approach is simplified in a number of ways. from the impact of one-day changes using a V10 multiplier.
In FRTB, banks are required to consider changes over periods
8
See Basel Committee on Banking Supervision, "Simplified Alternative
of 10 days that occurred during a stressed period in the past.
to the Standardized Approach to Market Risk Capital Requirements," Econometricians naturally prefer that non-overlapping periods
June 2017. be used when VaR or ES is being estimated using historical
simulation, because they want observations on the losses to be

Chapter 18 Fundamental Review of the Trading Book 209

independent. However, this is not feasible when 10-day changes Continuing in this way, we obtain equation (18.2). This is
are considered, because it would require a very long historical referred to as the cascade approach to calculating ES (and can
period. FRTB requires banks to base their estimates on overlap be used for VaR as well).
ping 10-day periods. The first simulation trial assumes that the
Calculations are carried out for each desk. If there are six
percentage changes in all risk factors over the next 10 days
desks, this means the internal models approach, as we have
will be the same as their changes between Day 0 and Day 10
of the stressed period; the second simulation trial assumes that described it so far, requires 5 X 6 = 30 ES calculations. As
the percentage changes in all risk factors over the next 10 days mentioned, the
use of overlapping time periods is less than ideal because changes
will be the same as their changes between Day 1 and Day 11
in successive historical simulation trials are not independent.
of the stressed period; and so on.
This does not bias the results, but it reduces the effective
Banks are first required to calculate ES when 10-day changes are sample size, making results more noisy than they would
made to all risk factors. (We will denote this by ES1.) They are otherwise be.
then required to calculate ES when 10-day changes are made
FRTB represents a movement away from basing calculations on
to all risk factors in categories 2 and above with risk factors in
one-day changes. Presumably the Basel Committee has decided
category 1 being kept constant. (We will denote this by ES2.)
that, in spite of the lack of independence of observations, a
They are then required to calculate ES when 10-day changes are
measure calculated from 10-day changes provides more relevant
made to all risk factors in categories 3, 4, and 5 with risk factors
information than a measure calculated from one-day changes.
in categories 1 and 2 being kept constant. (We will denote this
This could be the case if changes on successive days are not
by ES3.) They are then required to calculate ES when 10-day
independent, but changes in successive 10-day periods can rea
changes are made to all risk factors in categories 4 and 5 with
sonably be assumed to be independent.
risk factors in categories 1, 2, and 3 being kept constant. (We
will denote this by ES4.) Finally, they are required to calculate The calculation of a stressed measure (VaR or ES) requires
ES5, which is the effect of making 10-day changes only to cat banks to search for the period in the past when market variable
egory 5 risk factors. changes would be worst for their current port folio. (The search
must go back as far as 2007.) When Basel 11.5 was implemented,
The liquidity-adjusted ES is calculated as
a prob lem was encountered in that banks found that historical
data were not available for some of their current risk factors. It
(18.2) was therefore not possible to know how these risk factors would
have behaved during the 250-day periods in the past that were
where LHi is the liquidity horizon for category j. To understand candidates for the reference stressed period. FRTB handles this
equation (18.2), suppose first that all risk factors are in category by allowing the search for stressed periods to involve a subset of
1 or 2 so that only ES1 and ES2 are calculated. It is assumed that risk factors, pro vided that at least 75% of the current risk factors
the behavior of all risk factors during a 10-day period is indepen are used. The expected shortfalls that are calculated are scaled
dent of the behavior of category 2 risk factors during a further up by the ratio of ES for the most recent 12 months using all risk
10-day period. An extension of the square root rule then leads to factors to ES for the most recent 12 months using the subset of
the liquidity-adjusted ES being risk factors. (This potentially doubles the number of ES
calculations from 30 to 60.)

Banks are required to calculate ES for the whole portfolio as

Now suppose that there are also category 3 risk factors. The well for each of six trading desks. The ES for a trading desk is
expression VESf + ES would be correct if the category 3 risk referred to as a partial expected shortfall. It is determined by
factors had a 20-day instead of a 40-day liquidity horizon. We shocking the risk factors belonging to the trading desk while
assume that the behavior of the category 3 risk factors over keeping all other risk factors fixed. The sum of the partial
an additional 20 days is independent of the behavior of all expected shortfalls is always greater than the ES for the whole
the risk factors over the periods already considered. We also portfolio. What we will refer to as the weighted expected short
assume that the ES for the category 3 risk factors over 20 days fall (WES) is a weighted average of (a) the ES for the whole
is V2 times their ES over 10 days. This leads to a liquidity portfolio and (b) the sum of the partial expected shortfalls.
adjusted ES of: Specifically:

VESf + ES + 2ES WES= A X EST+ (1 - A) X LESPj

I
where EST is the expected shortfall calculated for the total port
folio and ESPi is the jth partial expected shortfall. The param
210 Financial Risk Manager Exam Part II: Market Risk Measurement and Management
eter A is set by the Basel Committee to be 0.5.

Chapter 18 Fundamental Review of the Trading Book 211

 Some risk factors are categorized as non-mode/able. Variance of U
Specifically, if there are less than 24 observations on a risk factor Variance of V
in a year or more than one month between successive observa
where U denotes the difference between the actual and model
tions, the risk factor is classified as non-modelable. Such risk fac
tors are handled by special rules involving stress tests. profit/loss in a day and V denotes the actual profit/loss in a day.9
Regulators expect the first measure to be between -10% and
The total capital requirement for day tis +10% and the second measure to be less than 20%. When there
are four or more situations in a 12-month period where the
max(WESr_1 + NMCr-1, me x WESavg + NMCav9l
ratios are outside these ranges, the desk must use the standard
where WESr_, is the WES for day t - 1, NMCr_, is the capital ized approach for determining capital.
charge calculated for non-modelable risk factors on day t -
1, WESavg is the average WES for the previous 60 days, and 18.3.3 Credit Risk
NMCavg is the average capital charge calculated for the non
modelable risk factors over the previous 60 days. The parameter As mentioned, FRTB distinguishes two types of credit risk expo
me is at minimum 1.5. sure to a company:

1. Credit spread risk is the risk that the company's credit

18.3.1 Back-Testing spread will change, causing the mark-to-market value of
the instrument to change.
FRTB does not back-test the stressed ES measures that are
2. Jump-to-default risk is the risk that there will be a default
used to calculate capital under the internal models approach for
by the company.
two reasons. First, it is more difficult to back-test ES than VaR.
Second, it is not possible to back-test a stressed measure at all. Under the internal models approach, the credit spread risk is
The stressed data upon which a stressed measure is based are handled in a similar way to market risks. Table 18.1 shows that
extreme data that statistically speaking are not expected to be the liquidity horizon for credit spread varies from 20 to 120 days
observed with the same frequency in the future as they were and the liquidity horizon for a credit spread volatility is 120
during the stressed period. days. The jump-to-default risk is handled in the same way as
default risks in the banking book. In the internal models
FRTB back-tests a bank's models by asking each trading desk to
approach, the capital charge is based on a VaR calculation with a
back-test a VaR measure calculated over a one-day horizon and
one-year time horizon and a 99.9% confidence level.
the most recent 12 months of data. Both 99% and 97.5% con
fidence levels are to be used. If there are more than 12 excep
tions for the 99% VaR or more than 30 exceptions for the 97.5% 18.3.4 Securitizations
VaR, the trading desk is required to calculate capital using the
The comprehensive risk measure (CRM) charge was introduced
standardized approach until neither of these two conditions con
in Basel 11.5 to cover the risks in products created by securitiza
tinues to exist.
tions such as asset-backed securities and collateralized debt
Banks may be asked by regulators to carry out other back-tests. obligations (see Section 26.1). The CRM rules allow a bank (with
Some of these could involve calculating the p-value of the profit regulatory approval) to use its own models. The Basel Com
or loss on each day. This is the probability of observing a profit mittee has concluded that this is unsatisfactory because there
that is less than the actual profit or a loss that is greater than is too much variation in the capital charges calculated by dif
the actual loss. If the model is working perfectly, the p-values ferent banks for the same portfolio. It has therefore decided
obtained should be uniformly distributed. that under FRTB the standardized approach must be used for
securitizations.

18.3.2 Profit and Loss Attribution

Another test used by the regulators is known as profit and
loss attribution. Banks are required to compare the actual
profit or loss in a day with that predicted by their models. Two
measures must be calculated. The measures are:
9
The "actual" profit/loss should be the profit and loss that would occur
Mean of U
if there had been no trading in a day. This is sometimes referred to as
Standard Deviation of V the hypothetical profit and Joss.

212 Financial Risk Manager Exam Part II: Market Risk Measurement and Management
18.4 TRADING BOOK VS. BANKING approved by their supervisors to use the internal models

BOOK approach, banks must also implement the standardized approach.

Regulatory capital under the standardized approach is based
on formulas involving the delta, vega, and gamma exposures of
The FRTB addresses whether instruments should be put in the
the trading book. Regulatory capital under the internal models
trading book or the banking book. Roughly speaking, the trading
approach is based on the calculation of stressed expected short
book consists of instruments that the bank intends to trade. The
fall. Calculations are carried out separately for each trading desk.
banking book consists of instruments that are expected to be held
to maturity. Instruments in the banking book are subject to credit
risk capital whereas those in the trading book are subject to Further Reading
market risk capital. The two sorts of capital are calculated in quite
Bank for International Settlements. "Minimum Capital Require
differ- ent ways. This has in the past given rise to regulatory
ments for Market Risk," January 2019.
arbitrage.
For example, banks have often chosen to hold credit-dependent
instruments in the trading book because they are then subject to Practice Questions and Problems
less regulatory capital than they would be if they had been placed
in the banking book. 18.1 Outline the differences between the way market risk
capital is calculated in (a) Basel I, (b) Basel 11.5, and (c) the
The FRTB attempts to make the distinction between the trading
FRTB.
book and the banking book clearer and less subjective. To be
in the trading book, it will no longer be sufficient for a bank to 18.2 Use footnote 4 to verify that when losses have a normal
have an "intent to trade." It must be able to trade and manage distribution with mean µ and standard deviation o· the
the underlying risks on a trading desk. The day-to-day changes in 97.5% expected shortfall is /.l + 2.3381T.
value should affect equity and pose risks to solvency. The FRTB 18.3 Explain why the use of overlapping time periods pro
provides rules for determining for different types of instruments posed by the FRTB does not give independent observa
whether they should be placed in the trading book or the banking tions on the changes in variables.
book. 18.4 What are the advantages of expected shortfall over value
An important point is that instruments are assigned to the at risk?
bank ing book or the trading book when they are initiated and 18.5 What is the difference between the trading book and
there are strict rules preventing them from being subsequently the banking book? Why are regulators concerned
moved between the two books. Transfers from one book to about specifying whether an instrument should be one
another can happen only in extraordinary circumstances. or the other in the FRTB?
(Examples given of extraordinary circumstances are the closing
18.6 How are credit trades handled under the FRTB?
of trading desks and a change in accounting standards with
regard to the recogni tion of fair value.) Any capital benefit as a
result of moving items between the books will be disallowed. Further Question
18.7 Suppose that an investor owns the $10 million portfolio
SUMMARY in Table 12.1 on July 8, 2020. Suppose that the 250
days ending April 30, 2020, constitute the stressed
FRTB is a major change to the way capital is calculated for mar period
ket risk. After over 20 years of using VaR with a 10-day time for the portfolio. Calculate the 97.5% expected shortfall
horizon and 99% confidence to determine market risk capital, using the over-lapping periods method in conjunction
regulators are switching to using ES with a 97.5% confidence with historical simulation and the cascade approach.
level and varying time horizons. The time horizons, which can Relevant data on the indices is on the author's website
be as high as 120 days, are designed to incorporate liquidity (see "Worksheets for Value at Risk Example"). For the
considerations into the capital calculations. The change that is purposes of this problem, assume that S&P 500 and FTSE
considered to a risk factor when capital is calculated reflects have a 10-day liquidity horizon, CAC 40 has a 40-day
movements in the risk factor over a period of time equal to the liquidity horizon, and Nikkei 225 has a 20-day liquidity
liquidity horizon in stressed market conditions. horizon. For each day during the stressed period,
The Basel Committee has specified a standardized approach consider the change in a variable over a 10-day period
and an internal models approach. Even when they have been ending on the day.

Chapter 18 Fundamental Review of the Trading Book 213