Financial'Market'Analysis'(FMAx)

Module'7

Introduction to Risk Management

This training material is the property of the International Monetary Fund (IMF) and is intended for use in IMF Institute for Capacity Development (ICD) courses.
Any reuse requires the permission of the ICD.
The'Relevance'to'You

You might be…

! An investor (reserve management, sovereign wealth fund, bank, insurance

company).

! A policymaker interested in financial intermediaries’ exposures and portfolio

losses (central bank, banking sector, pension funds, insurance companies).
The'Relevance'to'You

Historical background…

! JPMorgan (90s):

! 4:15 Report

! RiskMetrics: Technical document

Introduction to Risk Management: A Roadmap

What&is&VaR

Historical&Simulation

How&to&measure&VaR Delta8Normal&Approach

Monte&Carlo
Value'at'Risk'(VaR)

Characteristics of…

! Tries to estimate the level of possible losses over a given time period with a
certain probability.

! VaR summarizes the expected maximum loss over a time horizon within a
given confidence interval.
! For example, the 95% VaR loss is the amount of loss that will be exceeded only
5% of the time.
Value at Risk (VaR): How much can I lose? – 1

Source:(Bloomberg
Value at Risk (VaR): How much can I lose? – 2

Source:(Bloomberg
What we cover next…

What&is&VaR

Historical&Simulation

How&to&measure&VaR Delta8Normal&Approach

Monte&Carlo
Historical'Simulation

Characteristics of…

! Utilizes historical data.

! Generates histograms.
! Examines historical daily returns.
! Provides statistics (mean, minimum, maximum, standard deviation)

! Calculates VaR.
Historical Simulation VaR: How much can I lose?

Source:(Bloomberg
Value'at'Risk'(VaR)

The Advantages of VaR:

! It is intuitive.

! It states potential loss in terms of amounts (US$) rather than percentage

returns.

Choice of time horizon:

! Probability of loss is greater with longer time horizons.

! Typical choices: day, week, month

Different'Time'Horizons

Up to now, we have assumed that the horizon is 1 day, but what about longer
horizons?

We can extend the horizon (“holding period”) to N days in a simple way (assuming
each day is an independent observation).

! =!D N
2 2
! =!D N
Pros/Cons'of'Historical'Approach

! No dependence on any distributional assumptions; No need to estimate

volatilities or correlations.

! Accommodates fat tails.

! Accommodates assets whose payoff are non-linear.

! Assumes that the past is a good and reliable representation of the future.
DeltaGNormal'Method
An'Analytical'Framework'– 1
Assume that asset returns are normally distributed.
Their behavior can be fully described in terms of mean and standard deviation.

0.45

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0.00
-3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00
DeltaGNormal'Method
An'Analytical'Framework'– 2
Why this method vs historical simulation?

! Advantage… It is very easy to compute.

! Has limitations with comparison to historical simulation.

The assumption about normality could be inaccurate.
Specially with asymmetric returns
VaR with'Normally'Distributed'Returns

How far below the mean return is the threshold depends on how many standard
deviations the threshold is below the mean return.

Threshold(L* μ51.0σ μ51.65σ μ52σ μ52.33σ μ53σ

Pr(loss>L*) 0.16 0.05 0.023 0.01 0.001

There is nothing sacred about those multiples of sigma--we choose them because
they map easily onto the normal probabilities.
VaR with Normally Distributed Returns

Negative Returns 0 Positive'Returns

An'Example'with'One'Asset

Suppose we want to compute the 95% VaR.

The critical threshold is 1.65 standard deviations below the mean.

0.0001-1.65 • 0.00295=-0.00477 = 0.477%

VaR = 0.00477 • 200m=0.95m

Portfolio'VaR

When we have more than one asset in our portfolio, we can exploit the gains from
diversification.

! There are gains from diversification whenever the VaR for the portfolio
does not exceed the sum of the stand-alone VaRs.

! The VaR for the portfolio equals the sum of the stand-alone VaRs if and only
if the securities are perfectly correlated.
An'Example'of'Portfolio'VaR

Consider two securities:

! 30-year zero-coupon U.S. Treasury bond
! 5-year zero-coupon U.S. Treasury bond

For example…
Invest US$100 million in the 30-year bond
! Daily return volatility (std dev) !1 = 1.409%

Invest US$200 million in the 5-year bond

! Daily return volatility (std dev) !2 = 0.295%
Stand'Alone'VaRs versus'Portfolio'and'Diversification

Take the following…

95% confidence level

30 year zero VaR:

1.65 * 0.01409 * 100m = $2,325,000

5 year zero VaR:

1.65 * 0.00295 * 200m = $974,000

Sum of individual VaRs = US$ 3.299m

VaR of'the'Portfolio

! = w ! + w ! + 2w1w2 "1,2! 1! 2
2
p
2
1
2
1
2
2
2
2

Portfolio Variance:
(100*0.01409)2 + (200*0.00295)2 + 2(100*0.01409)(200*0.00295) * 0.88 = $3.797m

Portfolio Standard Deviation: σp = $1.948m

Portfolio VaR = 1.65 * 1.948m = $3.214m

Monte'Carlo'Simulation'Approach

Compute returns for the assets in the portfolio using computer generated random
numbers.

Random numbers are generated by assuming a distribution for the asset returns.

The joint distribution for the random number generation is chosen to match the
expected values, variances, and co-variances of the asset returns in the portfolio.

Similar to the Delta Normal approach, if portfolios have assets whose payoff are
linear.
Drawing the Monte Carlo Approach

R>0

0%
Horizon

R<0
Advantages'of'the'Monte'Carlo'Approach

Include the following:

! Allow for non-linear payoffs in the portfolio (options).

! Accommodates any distribution of risk factors.

! Allows the calculation of VaR confidence intervals.

Summary: Introduction to Risk Management

Risk, market risk and VaR

What&is&VaR

Historical&Simulation
! Excel exercises

How&to&measure&VaR Delta8Normal&Approach
! Calculating returns
! Histogram Monte&Carlo
! Standard deviation
Refinements'of'VaR:
1.'Backtesting
Is interested in answering…
Did the investment perform as VaR had predicted?

Compares the calculated VaRs with the actual related returns.

! With a 95% VaR bound, expect 5% of losses greater than the bound.
Approximately 12 days out of 250 trading days.

! If actual number of exceptions is “significantly” higher than 5%, model may be

inaccurate.
Backtesting – 1

Source: IMF Global Financial Stability Report, September 2007

Backtesting – 2

Source: IMF GFSR September 2007

Refinements'of'VaR:
2.'Basel'Committee'and'VaR G Guidelines
1. VaR computed daily, holding period is 10 days.

2. The confidence interval is 99 percent

! Banks are required to hold capital in proportion to the losses that can be expected to occur more
often than once every 100 periods.

3. At least 1 year of data to calculate parameters.

4. Parameter estimates updated at least once every month.

5. Capital provision is the greater of…

! Previous day’s VaR.
! Average of the daily VaR for the preceding 60 business days multiplied by a factor based on
backtesting results.
Stressed'ValueGatGRisk'Measure'Introduced'After'the'Crisis'

! “Losses in most banks’ trading books during the financial crisis have been
significantly higher than the minimum capital requirements under the former
Pillar 1 market risk rules.”

! “The Committee requires banks to calculate a stressed value-at-risk taking

into account a one-year observation period relating to significant
losses, which must be calculated in addition to the value-at-risk based on
the most recent one-year observation period.”

BIS, Revisions to the Basel II Market Risk Framework, 2011

Stressed'ValueGatGRisk'Measure

! Replicate a VaR calculation generated on bank’s current portfolio if relevant

market factors were experiencing a period of stress.

! Based on 10-day, 99th percentile VaR measure, with model inputs calibrated to
historical data from a continuous 12-month period of significant financial stress

! “The stressed VaR should be calculated at least weekly”.

BIS, Revisions to the Basel II Market Risk Framework, 2011.

 Stressed'ValueGatGRisk'Measure

Replicates a VaR calculation that would be generated if the relevant market factors
were experiencing a period of stress.

The capital requirement c:

VaR t −1 = VaR yesterday

VaR avg = average VaR over a 60 day period
sVaR t −1 = stressed VaR last available day m c = multiplication factor of 3 (minmum)
sVaR avg = average VaR over a 60 day stessed period ms = multiplication factor of 3 (minmum)
Refinements'of'VaR:
3.'Expected'Shortfall'(ES)
VaR tells us that a loss larger than “$X” will occur one day in 20 (with 95%
confidence), but does not tell us how big the losses that exceeds “$X” will be.
! (VaR does not indicate what happens in extreme market events.)

Assume the loss exceeds VaR…

Expected Shortfall –
1. Knowing the cut-off of what will happen C percent of the time (VaR),
2. Provides the average size of the loss when it exceeds the cut-off value.

How much could be lost if losses exceed VaR?

Expected Shortfall - E(-X | -X≤-VaR)

Source:(GFSR,(September(2007
BIS'Committee'G
Move'from'ValueGatGRisk'(VaR)'to'Expected'Shortfall'(ES)
Weaknesses identified with using VaR for determining regulatory capital
requirements: inability to capture “tail risk”.

Proposed in May 2012 to replace VaR with ES.

! ‘ES measures the riskiness of a position by considering both the size and the
likelihood of losses above a certain confidence level.’

Agreed to use a 97.5% ES for the internal models-based approach.

Has also used that approach to calibrate capital requirements under the revised
market risk standardized approach.
BIS'Committee'G
Move'from'ValueGatGRisk'(VaR)'to'Expected'Shortfall'(ES)
Recognizes that basing regulatory capital on both current VaR and stressed VaR
calculations may be unnecessarily duplicative.
! Proposed framework will simplify the capital framework by moving to a single ES
calculation calibrated to a period of significant financial stress.

Expected shortfall for the bank’s portfolio:

! Uses set of risk factors that explain at least 75% of the variation of the full ES model.

! Is calibrated to the most severe 12-month period of stress available over the
observation horizon.