Value-at-Risk (VaR)

How much can we lose?

Everything
correct, but useless answer.

How much can we lose realistically?

 What is the current Risk?
3

Bonds duration, convexity

Stocks volatility

Options delta, gamma, vega

Credit rating

Forex target zone

Total ?
4 Definition of VaR

 VaR is defined as the predicted worst-case loss at a specific confidence

level (e.g. 95%) over a certain period of time.
 VAR tells us the maximum loss a portfolio may suffer at a given
confidence interval for a specified time horizon.
 If we can be 99% sure that the portfolio will not suffer more than $ 10
million in a day, we say the 99% VAR is $ 10 million.
5 VaR
1

0.8

0.6

0.4
VaR1%
1%
0.2
Profit/Loss

-3 -2 -1 1 2 3
 Computing Value at Risk
6
 VaR is the product of :
 Value of portfolio
 Z factor ( depends on confidence level)
 Volatility
 Time ( VaR scales in proportion to square root of time)

 The usual practice is to calculate daily volatility by observing the daily

opening and closing prices of the portfolio over a period of time, say 6
months.
 Then we obtain the volatility for the actual period under consideration by
multiplying by the square root of the time.
 Thus the volatility for 5 trading days will be that for one day multiplied by √5.
 VaR at Goldman Sachs
7

Ref : Company Annual report

 Problem
8
The VAR on a portfolio using a one day time horizon is USD 100 million.
What is the VAR using a 10 day horizon ?
Solution
¨ Variance scales in proportion to time.
¨ So variance gets multiplied by 10
¨ And std deviation by √10
¨ VAR = 100 √10 = (100) (3.16) = 316
¨ (σN2 = σ12 + σ22 ….. = Nσ2)
Note; This approach is valid only when the daily variances are independent.
Note: Please refer to additional reading material for more solved
examples. They are important so go through them.
8
 Problem
9

If the daily VAR is $12,500, calculate the weekly, monthly, semi annual and annual
VAR. Assume 250 days and 50 weeks per year.
Solution
Weekly VAR ＝ (12,500) (√5)
＝ 27,951
Monthly VAR ＝ ( 12,500) (√20)
＝ 55,902
Semi annual VAR ＝ (12,500) (√125)
＝ 139,754
Annual VAR ＝ (12,500) (√250)
＝ 197,642

9
10 History of VaR

 80’s - major US banks - proprietary

 93 G-30 recommendations
 94 - RiskMetrics by J.P.Morgan
 98 - Basel
 SEC, FSA, ISDA, pension funds, dealers
 Widely used and misused!
 Risk Management Structure
11

Market data Current position

Risk Mapping

Valuation

Value-at-Risk

Reporting and Risk Management

 Risk Measuring Programs
Program Cost
CATS, CARMA $400K/yr
Algorithmics, Risk Watch >$1M
Infinity >$1M
J.P. Morgan, FourFifteen $25K/yr
FEA, Outlook $18K
Reuters, Sailfish ?
Theoretics, TARGA $75K
Bankers Trust, RAROC $50K/run
12
INSSINC, Orchestra $25-75K
 Unifying Approach
13

 One number
 Based on Statistics
 Portfolio Theory
 Verification
 Widely Accepted
 Easy Comparison
 How to measure VaR
14

 Historical Simulations

 Variance-Covariance

 Monte Carlo

 Analytical Methods

 Parametric versus non-parametric approaches

 Historical Simulations
15
 Historical Simulation uses the actual historical distribution of returns to
simulate the VAR of a portfolio.
 Real data plus ease of implementation, have made historical simulation a
very popular approach to estimating VAR.
 Historical simulation avoids the assumption that returns on the assets in
a portfolio are normally distributed.
 Instead, it uses actual historical returns on the portfolio assets to
construct a distribution of potential future portfolio losses.
 This approach requires minimal analytics.
 All we need is a sample of the historic returns on the portfolio whose
VAR we wish to calculate.
 Historical Simulations
16

How to Measure Returns

Steps for Calculations:
One year span
 Collect data
 Generate scenarios
 Calculate portfolio returns
 Arrange in order.

1% of worst cases
 % Returns Frequency Cumulative Frequency
- 16 1 1
- 14 1 2
17 - 10 1 3
-7 2 5
-5 1 6
-4 3 9
-3 1 10
What is VAR (90%) ? -1 2 12
0 3 15
1 1 16
2 2 18
4 1 19
6 1 20
7 1 21
8 1 22
9 1 23
11 1 24
12 1 26
14 2 27
18 1 28
21 1 29
23 1 30

Solution:
10% of the observations, i.e, (.10) (30) = 3 lie below -7
So VAR = -7
 Advantages and Disadvantages of
18
18
Historical simulation
 Advantages ¨Disadvantages
 Simple ¨Reliance on the past
 No normality ¨Length of estimation period
assumption
¨Weighting of data
 Non parametric
19
Qualitative & Quantitative Requirements for VaR
 Qualitative Requirements
 An independent risk management unit
 Board of directors involvement
 Internal model as an integral part
 Internal controller and risk model
 Backtesting
 Stress test
 Quantitative Requirements
 99% confidence interval
 10 business days horizon
 At least one year of historic data
 Data base revised at least every quarter
 All types of risk exposure
 Derivatives
 Simulation vs Variance Covariance
20 methods

 Simulation approaches are preferred by global banks due to:

 flexibility in dealing with the ever-increasing range of complex instruments in
financial markets
 the advent of more efficient computational techniques in recent years
 the falling costs in information technology
 However, the variance-covariance approach might be the most
appropriate method for many smaller firms, particularly when :
 they do not have significant options positions
 they prefer to outsource the data requirement component of their risk
calculations to a company such as RiskMetrics
 significant savings can often be made by using outsourced volatility and
correlation data, compared to internally storing the daily price histories required
for simulation techniques
 Variance-Covariance
21
 Means and covariances of market
factors VaR1%  V  2.33 V
 Mean and standard deviation of the
portfolio

 Delta or Delta-Gamma
approximation

 VaR1%= v – 2.33 v 1% 2.33

 Based on the normality assumption!


-2.33
 Problem
22
Consider a bond position of $ 10 million, a modified duration of 3.6
years and an annualized bond volatility of 2%.Calculate the 10 day
99% VAR assuming 252 business days in a year. At 99% confidence
interval the standard deviation critical value is 2.33.

Solution
 10 day volatility = .02X√(10/252)=.003984
 So 99% VAR = 2.33 X 3.6 X 10,000,000 X .003984
 = $ 334,186
23 What is Monte Carlo VAR?

 The Monte Carlo approach involves generating many price

scenarios (usually thousands) to value the assets in a portfolio over
a range of possible market conditions.
 The portfolio is then revalued using all of these price scenarios.
 Finally, the portfolio revaluations are ranked to select the required
level of confidence for the VAR calculation.
 Generate Scenarios
24
 The first step is to generate all the price and rate scenarios
necessary for valuing the assets in the relevant portfolio, as well
as the required correlations between these assets.
 There are a number of factors that need to be considered when
generating the expected prices/rates of the assets:
 Opportunity cost of capital
 Stochastic element
 Probability distribution
 Opportunity Cost of Capital
25

 A rational investor will seek a return at least equivalent to the

risk-free rate of interest.
 Therefore, asset prices generated by a Monte Carlo simulation
must incorporate the opportunity cost of capital.
26 Probability Distribution
 Monte Carlo simulations are based on random draws from a
variable with the required probability distribution, usually the
normal distribution.
 The normal distribution is useful when modeling market risk in
many cases.
 But it is the returns on asset prices that are normally distributed,
not the asset prices themselves.
 So we must be careful while specifying the distribution.
27 Calculate the Value of the Portfolio
 Once we have all the relevant market price/rate scenarios, the next
step is to calculate the portfolio value for each scenario.
 For an options portfolio, depending on the size of the portfolio, it
may be more efficient to use the delta approximation rather than a
full option pricing model (such as Black-Scholes) for ease of
calculation.
 Δ (Option) = Δ(ΔS)
 Thus the change in the value of an option is the product of the delta
of the option and the change in the price of the underlying.
 Other approximations
28
 There are also other approximations that use delta, gamma (Γ) and
theta (Θ) in valuing the portfolio.
 By using summary statistics, such as delta and gamma, the
computational difficulties associated with a full valuation can be
reduced.
 Approximations should be periodically tested against a full
revaluation for the purpose of validation.
 When deciding between full or partial valuation, there is a trade-off
between the computational time and cost versus the accuracy of the
result.
 The Black-Scholes valuation is the most precise, but tends to be
slower and more costly than the approximating methods.
 Reorder the Results

29
 After generating a large enough number of scenarios and calculating
the portfolio value for each scenario:
 the results are reordered by the magnitude of the change in the
value of the portfolio (Δportfolio) for each scenario
 the relevant VAR is then selected from the reordered list
according to the required confidence level

 If 10,000 iterations are run and the VAR at the 95% confidence level
is needed, then we would expect the actual loss to exceed the VAR in
5% of cases (500).
 So the 501st worst value on the reordered list is the required VAR.
 Similarly, if 1,000 iterations are run, then the VAR at the 95%
confidence level is the 51st highest loss on the reordered list.
30
Formula used typically in Monte Carlo for
stock price modelling
31 Advantages of Monte Carlo
 We can cope with the risks associated with non-linear positions.
 We can choose data sets individually for each variable.
 This method is flexible enough to allow for missing data periods to
be excluded from the VAR calculation.
 We can incorporate factors for which there is no actual historical
experience.
 We can estimate volatilities and correlations using different
statistical techniques.
32
Problems with Monte Carlo
 Cost of computing resources can be quite high.
 Speed can be slow.
 Random Numbers may not be all that random.
 Pseudo random numbers are only a substitute for true random
numbers and tend to show clustering effects.
 Monte Carlo often assumes normal distribution, But it can be
performed with alternative distributions.
 Results (value at risk estimate) depend critically on the models
used to value (often complex) financial instruments.
33
Comparison of different VAR modeling
techniques
 Risk Management
34

 Risk measurement
 Reporting to board
 Limits monitoring
 Diversification, reinsurance
 Vetting (Careful secret checks)
 Reporting to regulators
 Decision making based on risk
 Who manages risk?
35
Financial Institutions & Corporations Consulting
Consulting

 Citibank  AIG  Nike

 Oliver, Wyman and Co.
 Bank of England  General Re  Sony
 Willis Corroon
 CIBC  Swiss Re  Dell Computers
 Richard Scora
 J. P. Morgan  Aetna  Philip Morris
 Ernst and Young
 Bankers Trust  Zurich  Ford Motor
 Enterprise Advisors
 Kamakura
36 Examples of Risk Reports

http://www.mbrm.com/
slide
37 Home assignment

 Calculate your personal VaR

 Stress Testing
and
Back Testing VaR
 Back testing
39

 Backtesting is the process of comparing losses predicted by a value at risk

(VaR) model to those actually experienced over the testing period.

 It is done to ensure that VaR models are reasonably accurate.

 Risk managers systematically check the validity of the underlying valuation and
risk models by comparing actual to predicted levels of losses.

 The overall goal of backtesting is to ensure that actual losses do not exceed the
expected losses at a given level of confidence.

 Exceptions are number of actual observations over and above the expected
level. In the context of the VaR, the number of exceptions falling outside of the
VaR confidence level should not exceed one minus the confidence level. For
instance, exceptions should occur less than 1% of the time if the level of
confidence is 99%. Exceptions are also called exceedances.
 Examples of Backtesting
40

 Over 100 days, a good 95.0% VaR model will produce

approximately 5.0% * 100 days = 5 exceptions

 Over 1,000 days, a good 99.0% VaR model will produce

approximately 1.0% * 1,000 days = 10 exceptions

 Assuming there are 252 trading days in a year, a 95.0% daily VaR
should be exceeded about 13 days per year; 5% * 252 days = 12.6
days

 Assuming there are 252 trading days in a year, a 99.0% daily VaR
should be exceeded about 10 days over a four-year period; 1% * 4
* 252 days = 10.08 days
 Importance of Back testing
41

 Backtesting gives a “reality check” on whether VaR forecasts are properly

calibrated or accurate. Too many exceptions should prompt a recalibration of the
model and a thorough re-examining of assumptions, parameters, and the entire
modeling process.

 Backtesting provides the Basel Committee with a critical evaluation technique to

test the adequacy of internal VaR models.

 Backtesting helps bank regulators to verify risk models used by subject banks
and identify banks using models that may potentially underestimate risk, thereby
endangering the financial health of not just the bank but the industry in general
(This is particularly true for systematically important banks, SIBs).

It is not uncommon to find banks with excessive exceptions (more than four
exceptions in a sample size of 250) being penalized with higher capital
requirements.
 Difficulties in Back testing VaR Model
42

 There are several things that make backtesting a difficult task for risk managers.

 First, VaR models are based on static portfolios but in reality, actual portfolio
compositions are in a constant stage of change to reflect daily gains/losses,
expenses, and buy/sell decisions. For this reason, the risk manager should track
both the actual portfolio return and the hypothetical (static) return. In some
instances, it may also make sense to carry out backtesting using a “clean return”
instead of the actual return. Clean return is actual return minus all non-mark-to-
market items like fees, commissions, and net income.

 Second, the sample backtested may not be representative of the true underlying
risk. Since the backtesting period is just a limited sample, it would be a stretch of
reality to expect the predicted number of exceptions in every sample. At the end of
the day, backtesting remains a statistical process with accept/reject decisions.
 Basel Rules for Backtesting
43
Green zone - up to 4 exceptions
Yellow zone - 5-9 exceptions
Red zone - 10 exceptions or more
 Stress Testing
44  Stress testing serves to warn a firm’s management of potential adverse events arising from the firm’s risk
exposure and goes further to give estimates of the amount of capital needed to absorb losses that may result
from such events.

 Stress tests help to avoid any form of complacency that may creep in after an extended period of stability and
profitability. It serves to remind management that losses could still occur, and adequate plans have to be put in
place in readiness for every eventuality. This way, a firm is able to avoid issues like underpricing of products,
something that could prove financially fatal.

 Stress testing is a key risk management tool during periods of expansion when a firm introduces new products
into the market. For such products, there may be very limited loss data or none at all, and hypothetical stress
testing helps to come up with reliable loss estimates.

 Under pillar 1 of Basel II, stress testing is a requirement for all banks using the Internal Models Approach
(IMA) to model market risk and the internal ratings-based approach to model credit risk. These banks have to
employ stress testing to determine the level of capital they are required to have.

 Stress testing supplements other risk management tools, helping banks to mitigate risks through measures
such as hedging and insurance. By itself, stress testing cannot address all risk management weaknesses, nor
can it provide a one-stop solution.
 Stress Testing
45

 Stress testing involves analyzing the effects of exceptional events in the market on a
portfolio's value.
 These events may be exceptional, but they are also plausible.
 Their impact can be severe.
 Historical scenarios or hypothetical scenarios can be used.

Two Types of Stress Testing

Single-factor stress testing (sensitivity testing) involves applying a shift in a specific risk factor
to a portfolio in order to assess the sensitivity of the portfolio to changes in that risk factor.
Multiple-factor stress testing (scenario analysis) involves applying simultaneous moves in
multiple risk factors to a portfolio to reflect a risk scenario or event that looks plausible in the
near future.
 Regulatory Stress Testing
46  US, UK, and EU regulators require banks and insurance
companies to perform specified stress tests.

 In the United States, the Federal Reserve performs stress tests of

all the banks whose consolidated assets are over USD 50 billion.

 This type of stress test is termed as Comprehensive Capital

Analysis and Review (CCAR). Under CCAR, the banks are
required to consider four scenarios:

 Baseline Scenario
 Adverse Scenario
 Severely Adverse Scenario
 Regulatory Stress Testing
 The baseline scenario is based on the average projections from the surveys of the
47
economic predictors but does not represent the projection of the Federal Reserve.

 The adverse and the severely adverse scenarios describe hypothetical sets of events
which are structured to test the strength of banking organization and their resilience.
Each of the above scenarios consists of the 28 variables (such as the unemployment
rate, stock market prices, and interest rates) which captures domestic and international
economic activity accompanied by the Board explanation on the overall economic
conditions and variations in the scenarios from the past year.

 Banks are required to submit a capital plan, justification of the models used, and the
outcomes of their stress testing. If a bank fails to stress test due to insufficient capital,
the bank is required to raise more capital while restricting the dividend payment until the
capital has been raised.

 Banks with consolidated assets between USD 10 million and USD 50 million are under
the Dodd-Fank Act Stress Test (DFAST). The scenarios in the DFAST is similar to that in
the CCAR. However, in the DFAST, banks are not required to produce a capital plan.

 Therefore, through stress tests, regulators can evaluate the banks to determine their
ability to extreme economic conditions consistently. However, they recommend that
banks develop their scenarios.
 Stress Testing
48

Basel committee recognized the importance of stress testing in:

 Giving a forward-looking perspective on the evaluation of risk;

 Overcoming the demerits of modes and historical data;
 Facilitating the development of risk mitigation, or any other plans
to reduce risks in different stressed conditions;
 Assisting internal and external communications;
 Supporting the capital and liquidity planning procedures; and
 Notifying and setting of risk tolerance
 Key Elements on Stress Testing
49
 Stress testing is the ultimate responsibility of the management body of an
institution. The buck stops with senior management and the Board of directors.

 The management should have an in-depth understanding of the institution’s

risk profile and the potential impact of stress events.

 The stress testing program should be guided by clear, well-detailed policies

that outline the procedure to follow from start to finish, as well as describing the
role played by various employees.

 There should be consistent documentation of stress testing results and

methodologies

 There should be an independent review of not just the testing procedure but
also the results of the program.

 The ensuing learning outcomes seek to break down these key elements even
further.