Risk Management for Banking

Value at Risk (VaR)

Lecturer: Claudio Vallar
School of Economics and Finance
 A Single Number for Risk?

• A financial institution's portfolio depends on hundreds, or even thousands, of market

variables.

• But how can senior management and regulators have an idea of the total risk to which
the financial institution is exposed?

• Value at Risk (VaR) and Expected Shortfall (ES) are attempts to provide a single number
that summarizes the total risk in a portfolio.

• Now widely used by corporate treasurers, fund managers as well as by financial

institutions and more importantly, by regulators.

• VaR captures an important aspect of risk in a single number.

 Value at Risk (VaR)

When using the Value at Risk measure, we are making the following statement:
"We are X percent confident that we will not lose more than V in time T".

Time Horizon

Specified Confidence Level (X%)

VaR
Potential Loss

Normal Market Conditions

 Value at Risk (VaR)

Value at Risk (VaR) is a financial metric that

estimates the risk of an investment.

More specifically, VaR is a statistical

technique used to measure the amount of
potential loss that could happen in an
investment portfolio over a specified
period of time.
 The Gain and the Loss Distributions

VaR can be calculated from either the probability distribution of gains or the probability
distribution of losses during time T
 The Time Horizon

• Time horizon is the holding period, measured in days, which the portfolio is at risk of
adverse price movements.

• VaR is often reported over a one-day period.

• 10-day VaR has been generally used for capital calculation purposes.

𝑇 − 𝑑𝑎𝑦 𝑉𝑎𝑅 = 1 − 𝑑𝑎𝑦 𝑉𝑎𝑅 ∗ 𝑇

• This formula is true when the changes in the value of the portfolio on successive days
have independent identical normal distributions with mean zero. In other cases, it is an
approximation.
 VaR and Basel III

• Under the Basel II and Basel III frameworks, banks must hold sufficient capital to cover
various risks, including market risk using internal models like VaR.
• The 10-day VaR at a 99% confidence level is used to estimate the potential loss in the
trading book over a 10-day period.
• Banks must calculate VaR daily and hold capital sufficient to cover potential losses.

The Basel III accord raised the minimum capital requirements for banks
from 2% in Basel II to 4.5% of common equity, as a percentage of the
bank’s risk-weighted assets. There is also an additional 2.5% buffer capital
requirement that brings the total minimum requirement to 7%.
VaR – Example 1

● The gain from a portfolio over six months is

normally distributed with mean μ = $2 million and
standard deviation σ = $10 million.

● The 1% point of the distribution of gains is

𝑉 = 𝜇 + 𝑁 −1 𝑞 ∗ 𝜎
𝑁 −1 0.01 = −2.33
𝑉 = $2𝑚 − 2.33 ∗ $10𝑚 = −$21.3𝑚

● The VaR for the portfolio with a six-month time

horizon and a 99% confidence level is $21.3 million.
 VaR – Example 2
• All outcomes between a loss of $50 million and a gain of $50 million are equally likely (uniformly
distributed) for a one-year project.
• What is V such as that outcomes are higher than V with prob of 99%?
• The range of the distribution is from -$50 million to +$50 million.
• The “length” of the range is thus $100 million = +$50 million - (- $50 million).
• The average of the distribution is $0 million = (+$50 million + (- $50 million))/2.
• Each 1% of the range represents $1 million = 0.01 X $100 million.
• The maximum loss for a 1% confidence interval is thus the minimum value (-$50 million) plus 1%
($1 million) = -$49 million.
• The VaR for a one-year time horizon and a 99% confidence level is $49 million.
 How to Calculate VaR

There are three approaches that banks can use for building an internal VaR model:

Historical Simulation
Variance-Covariance Approach
Monte Carlo Simulations

Which method a bank chooses to use depends on the size and nature of their exposure to
the market risk.
 VaR – Historical Simulation
Historical Simulation uses historical returns on portfolio assets to create a distribution of
potential future portfolio profits and losses, from which the VaR can be determined.

How to calculate VaR:

1. Identifying the applicable risk factors for the portfolio.

2. Collecting time series data for the required number of days.

3. Generating P&L outputs for each data point of the time series.

4. Identifying the VaR corresponding to the required confidence level.

 VaR – Historical Simulation – Example

• Calculate the returns

• Construct the histogram of returns

• Find the VaR value using the Excel

formula =PERCENTILE.INC (array_returns,
X%) where X% is the bottom qth percentile
.
 Historical Simulation: Advantages and Limitations
Historical simulation has a number of key advantages and some limitations:

Advantages: Limitations:
• Nonparametric approach • Under- or overstated depending on
• No Normal Distribution the number of days in the chosen
• Widely applicable data set
• No reliance on valuation models • Slow to respond to volatility changes
• Easy to implement
• Data is generally available
• Easy to understand: “how bad
can things get?”.
 VaR – Variance-Covariance Methods

• The Variance-Covariance methodology that relies crucially on the assumption that

asset returns are normally distributed.

• As such, making the normality assumption means that VaR can be calculated once the
mean and the variance are known.

• The mean is generally assumed to be zero (or very close to zero) as VaR is usually
modelled over a short holding period such as one day.

𝑉𝑎𝑅1% = 𝜇𝑃 + 𝑁 −1 0.01 ∗ 𝜎𝑃

• In case of a single-asset portfolio, correlation does not need to be considered. However, in

practice, correlations need to be accounted for as portfolio contains multiple assets.
 VaR – Variance-Covariance Methods
• For a two-asset portfolio:

• For a Multi-Asset Portfolio:

• The variance of a portfolio can be calculated using different techniques:

o Variance-Covariance Matrix
o Single-index model
o Constant correlation model.
o Shrinkage methods
o Implied volatility
o Volatility models.
 VaR – Monte Carlo Simulation
Monte Carlo simulation involves the generation of a simulated distribution by running a
series of scenarios developed using a random number generator rather than historical data.
In other words, we generate a model for future stock returns and run numerous trials.

How to calculate VaR:

1. Identify the risk factors for the portfolio for which VaR is to be calculated

2. Specify the dynamics for the risk factors, such as correlations and volatilities.

3. Generate a range of scenarios for each risk factor using a random number generator
to generate values for the portfolio.

4. Identify the VaR value from the output series that corresponds to the required
confidence interval.
 Monte Carlo Simulation: Advantages

Advantages:

• It is possible to generate an infinite number of outputs.

• The data sets used can be chose for each risk factor independently of each other.

• It can cater for non-linear situations. This means that options and more complex
derivatives can be addressed.

• Different methods can be used to estimate parameters for different risk factors. For
example, volatilities can be estimated using different techniques such as implied
volatility and GARCH

• Sensitivity analysis and stress testing can be carried out, as values are hypothetical.
 Monte Carlo Simulation: Limitations

Limitations:

• Assumption Dependence: Monte Carlo simulations heavily rely on the assumptions

made about the probability distributions of asset returns, correlations, and volatilities.

• Limited Historical Data: Monte Carlo simulations typically utilize historical data to
model future scenarios.

• Complexity and Human Bias.

• Model Risk: The risk that the chosen model may not accurately represent the
complexities of the market or the underlying assets. Model risk can lead to significant
discrepancies between simulated outcomes and actual market behavior.
 Antecedents of the Crisis
• A key limitation of VaR is that it cannot provide any useful information about the size of
losses should extreme events occur, such as those seen during the 2007-08 financial
crisis.

• Regulators require banks to use the Expected Shortfall (ES) approach when making
capital calculations, rather than VaR.

• Nevertheless, VaR continues to play a key role in back testing, as well as a useful tool for
the management of market risk.
Risk Management for Banking
Expected Shortfall (ES)
Lecturer: Claudio Vallar
School of Economics and Finance
 VaR Breaks
• Where the VaR value of a portfolio is exceeded, this is known as VaR break.

• Meanwhile, averaging the value of losses for all VaR breaks in a given period is knowns
as conditional VaR (CVaR). This is also referred as Expected Shortfall (ES).

• It is important to note that VaR is not an indicator of how much would be lost if a VaR
break occurs, but how likely it is such a break will occur.

• This likelihood will be higher:

o The lower the confidence level (say, 95% rather than 99%)

o The longer the time horizon (say, ten days rather one day)
 Expected Shortfall (ES)
• Below we can see two different distributions of gains, with the same VaR.
• But clearly the left tail of the distributions are different.
• How can we measure the amount of risk contained in this part of the distribution that is
not considered by the VaR?

Expected Shortfall (ES) is the expected

loss during time T conditional on the loss
being greater than the VaR.
 Expected Shortfall - Example
• In order to calculate ES, it is necessary to calculate VaR first.
• ES is the expected loss during time T conditional on the loss being greater than the VaR.
• For example, suppose that X = 99%, T is 10 days, and the VaR is $64 million.
• The ES is the average amount lost over a 10-day period assuming that the loss is greater
than $64 million.
• Setting also an ES limit rather than only a VaR limit makes it less likely that traders will
be able to take the sort of position indicated by the figure below.
 Coherent Risk Measures
Properties a risk measure should have:
• Monotonicity: If one portfolio always produces a worse outcome than another, its risk
measure should be greater.
• Translation Invariance: If we add an amount of cash K to a portfolio, its risk measure
should go down by K.
• Homogeneity: Changing the size of a portfolio by λ should result in the risk measure
being multiplied by λ (if you double the size of your portfolio, you should double its
capital requirements).
• Subadditivity: The risk measures for two portfolios after they have been merged
should be no greater than the sum of their risk measures before they were merged
 Coherent Risk Measures – ES and VaR

• VaR satisfies the first three conditions but not the subadditivity one.

• ES satisfies all four conditions.

• The subadditivity condition states that diversification helps reduce risks.

• When we aggregate two portfolios, the total risk measure should either decrease or stay
the same.
ES – Example
ES – Example – each project separately
ES – Example – portfolio of 2 projects
Risk Management for Banking
More on VaR and ES
Lecturer: Claudio Vallar
School of Economics and Finance
 VaR and ES - Normal Distribution Assumption

When losses (gains) are normally distributed with mean μ and standard deviation σ,

𝑉𝑎𝑅 = 𝜇 + 𝑁 −1 𝑋 ∗ 𝜎

(𝑁−1 𝑋 )2ൗ
𝑒− 2
𝐸𝑆 = 𝜇 + 𝜎
2𝜋 1 − 𝑋

Where
• 𝑋 is the confidence level
• 𝑁 −1 𝑋 is the inverse standard normal distribution with μ =0 and σ =1
• (the book to simplify the notation used Y = 𝑁 −1 𝑋 )
 VaR and ES - Example
Suppose that the loss from a portfolio over a 10-day time horizon is normal with a mean of
zero and a standard deviation of $20 million.
● The 10-day 99% VaR is

● The 10-day 99% equivalent Expected Shortfall is

 VaR and ES - Changing the Time Horizon
• If losses in successive days are independent, normally distributed, and have a mean of zero:

𝑇 − 𝑑𝑎𝑦 𝑉𝑎𝑅 = 1 − 𝑑𝑎𝑦 𝑉𝑎𝑅 ∗ 𝑇

𝑇 − 𝑑𝑎𝑦 𝐸𝑆 = 1 − 𝑑𝑎𝑦 𝐸𝑆 ∗ 𝑇

• If positions are very liquid and actively traded, shorter term horizons for VaR and ES should
be privileged.
• If there is autocorrelation ρ between the losses (gains) on successive days, we replace 𝑇
with

𝑇 + 2 𝑇 − 1 𝜌 + 2 𝑇 − 2 𝜌2 + 2 𝑇 − 3 𝜌3 + ⋯ + 2𝜌𝑇−1
in the T-day VaR and ES formulas.
 VaR and ES - Changing the Time Horizon
• This will give us the ratio of T-Day Var to 1-Day Var
 Regulation – Time Horizon and Confidence Intervals
• Time horizon should depend on how quickly portfolio can be unwound.

• Regulators are planning to move toward a system where:

o ES is used and
o the time horizon depends on liquidity

• Confidence level depends on objectives. Regulators have used :

o 99% for market risk and
o 99.9% for credit/operational risk

• A bank wanting to maintain an AA credit rating might use confidence levels as high as
99.97% for internal calculations
Risk Management for Banking
Stress Testing
Lecturer: Claudio Vallar
School of Economics and Finance
 Stress Testing
A model is the core element of the processing stage of the stress testing process where
output values are calculated to one or more risk factors based on input values.

Inputs Processing Outputs

Values produced by stress tests must be checked to ensure they are in line with
expectations. Where any issues are identified, the cause should be determined, and a
decision made on what action(s) to take.
 Stress Testing Levels

In addition to bank-wide tests, banks Depending on the level, stress tests use one
conduct stress test at varying levels. or more risk factors.
These include: These include:

• Individual customers • Economic factors (such as changes in

• Portfolio GDP, unemployment rates,…)
• Products • Market factors (e.g., interest rates,
• Business Unit commodity prices, stock prices,…)
• Industry-specific • Nonmarket factors (e.g., political events,
natural disasters, …)
 Stressed VaR

• Stressed VaR (sVaR) is an enhancement of the traditional Value at Risk (VaR) measure.

• It is used to assess the potential loss a bank could experience in a hypothetical future
scenario similar to an extreme historical period of significant financial stress.

• This measure is part of the regulatory framework introduced by the Basel Committee on
Banking Supervision under Basel III to ensure banks hold sufficient capital to cover
potential losses during periods of financial turmoil.
 Calculation of Stressed VaR:
Calculation of Stressed VaR:

1. Selection of Stress Period: Banks identify a historical period of significant financial

stress. (Examples include the 2007-2008 financial crisis).

2. Data Collection: Historical market data from the identified stress period is collected.
This data includes price movements, volatilities, and correlations for the relevant risk
factors affecting the bank's portfolio.

3. Calculation Using Historical Data: Using the collected stress period data, banks apply
their VaR models to calculate the potential loss.

4. Confidence Level and Holding Period: Typically, sVaR is calculated at a 99%

confidence level over a 10-day holding period, aligning with regulatory requirements.
 Differences Between Stressed VaR and Standard VaR

Standard VaR: Stressed VaR:

• Uses recent historical data to calculate • Uses data from a historical period of

potential losses under normal market financial stress, capturing extreme

conditions. market conditions, ensuring that banks

are prepared for financial crises.

• Often used for daily risk management • Specifically required by regulators

and internal monitoring. under Basel III to ensure sufficient

capital reserves .

Capital Requirement = k * (VaR + sVaR)

Where k is a regulatory multiplier (typically set to 3).