Value at Risk

Chapter 8
Value at Risk (VaR) is an attempt to provide a single number
that summarizes the total risk in a portfolio. It was
pioneered by J.P. Morgan and has become widely used by
corporate treasurers and fund managers as well as by
financial institutions.

As we shall see in Chapter 11, it is the measure regulators

have chosen to use for many of the calculations they carry
out concerned with the setting of capital requirements for
market risk, credit risk, and operational risk.
 DEFINITION OF VaR
VaR is used to make a statement about the maximum
potential loss a portfolio might experience within a specific
time frame, with a certain degree of confidence.

The statement looks like this: "We are X percent certain

that we will not lose more than V dollars in time T."
 Key Elements of VaR:
The statement looks like this: "We are X percent certain that we will
not lose more than V dollars in time T."

• V (VaR): This is the maximum potential loss we are confident will not
be exceeded. It's the key number VaR provides.

• T (Time Horizon): This is the period over which we are measuring

the risk (e.g., one day, one week, one month).

• X (Confidence Level): This is the probability that the actual loss will
be less than or equal to V (e.g., 95% confidence level means we are
95% certain the loss will not exceed V).
 Historical Perspectives on VaR
J.P. Morgan played a key role in popularizing Value at Risk (VaR) as
a risk measurement tool. Chairman Dennis Weatherstone found
daily risk reports overly detailed and unhelpful for top management.
He requested a simpler report focusing on the bank's total exposure
over the next 24 hours across its trading portfolio. Initially deemed
impossible by his team, they eventually adapted the Markowitz
portfolio theory to create a VaR report, which came to be known as
the 4:15 report, delivered daily at 4:15 p.m. after trading closed.
 Historical Perspectives on VaR
Creating the VaR report required extensive work, including daily
data collection on global bank positions, managing time zone
differences, estimating correlations and volatilities, and developing
computer systems, completed around 1990. The primary benefit
was that senior management gained a clearer understanding of the
bank's risks, enabling better capital allocation. By 1993, other
banks had also developed similar risk aggregation methods,
solidifying VaR as a key risk measure in the industry.
 Historical Perspectives on VaR
In 1994, J.P. Morgan broke the norm by releasing a simplified risk
model called RiskMetrics on the internet, providing data on
variances and covariances for various market variables. This move
sparked discussions on different VaR models and led to software
firms creating their own versions, some using RiskMetrics data. As a
result, VaR was quickly adopted by financial and non-financial
institutions. The BIS Amendment, based on VaR, was announced in
1996 and implemented in 1998. Later, J.P. Morgan's RiskMetrics
group became a separate company, developing CreditMetrics for
credit risks in 1997 and CorporateMetries for non-financial
corporate risks in 1999.
 VaR vs. EXPECTED SHORTFALL
VaR is an attractive measure because it is easy to understand. In
essence, it asks the simple question: "How bad can things get?" This
is the question all senior managers want answered. They are very
comfortable with the idea of compressing all the Greek letters for all
the market variables underlying a portfolio into a single number.
VaR is also relatively easy to back-test. However, when VaR is used
in an attempt to limit the risks taken by a trader, it can lead to
undesirable results. Suppose that a bank tells a trader that the one-
day 99% VaR of the trader’s portfolio must be kept at less than $10
million. The trader can construct a portfolio where there is a 99.1%
chance that the daily loss is less than.
 Expected Shortfall
Expected shortfall is a risk measure that offers better incentives for
traders compared to Value at Risk (VaR). While VaR answers, "How
bad can things get?", expected shortfall asks, "If things do get bad,
what is the expected loss?" It is defined by two parameters: T (time
horizon) and X (confidence level), representing the expected loss
over time T, given that the loss exceeds the Xth percentile of the
loss distribution. For example, if X = 99 and T is ten days with a VaR
of $64 million, the expected shortfall reflects the average loss over
ten days when losses exceed $64 million. Expected shortfall
promotes diversification and has superior properties compared to
VaR; however, it is less straightforward and harder to back-test,
making it more complex to understand and validate.
 VaR AND CAPITAL
Value at Risk (VaR) is utilized by both financial regulators and
institutions to determine the necessary capital reserves. Regulators
assess the capital needed for market risk as a multiple of the VaR
calculated over a ten-day period at a 99% confidence level, while
capital for credit and operational risks is determined using a one-
year horizon at a 99.9% confidence level. For example, if a
portfolio's VaR at a 99.9% confidence level for one year is $50
million, this indicates that, under extreme conditions (theoretically
once every thousand years), the institution might lose more than
$50 million in a year. Consequently, maintaining $50 million in
capital would provide a 99.9% probability of avoiding depletion of
funds within that year.
Suppose we are trying to design a risk measure that will equal the
capital a financial institution is required to keep. Is VaR (with an
appropriate time horizon and an appropriate confidence level) the
best measure? Artzner et al. have examined this question. They
first proposed a number of properties that such a risk measure
should have.' These are:
1. Monotonicity: If a portfolio produces a worse result than
another portfolio for every state of the world, its risk measure
should be greater.
2. Translation invariance: If an amount of cash K is added to a
portfolio, its risk measure should go down by K.
3. Homogeneity: Changing the size of a portfolio by a factor 2,
while keeping the relative amounts of different items in the
portfolio the same, should result in the risk measure being
multiplied by 2.
4. Subadditivity: The risk measure for two portfolios after they
have been merged should be no greater than the sum of their risk
measures before they were merged.
 CONHERENCE RISK MEASURES

A coherence risk measure is a function that assigns a single

number to a portfolio of assets, representing the overall risk of that
portfolio. This number is typically interpreted as the minimum
capital required to cover potential losses under adverse market
conditions.
 Choice of Parameters For VAR
The parameters for Value-at-Risk (VaR) are the key inputs that
determine the level of risk associated with a portfolio. They define
the specific conditions under which the VaR is calculated,
influencing the resulting risk measure. Here's a breakdown of the
key parameters and their meaning:.
1. Confidence Level: - Meaning: The confidence level represents
the probability that the actual loss will not exceed the calculated
VaR. It essentially sets a threshold for how much risk the investor is
willing to accept.
- Example: A 95% confidence level means that there is a 5% chance
of experiencing a loss greater than the calculated VaR.
 Choice of Parameters For VAR
2. Time Horizon: - Meaning: The time horizon specifies the period
over which the VaR is calculated. It determines the timeframe for
which the potential loss is being assessed.
- Example: A daily VaR assesses the maximum potential loss over a
single day, while a monthly VaR considers potential losses over a
month.
Impact of Autocorrelation
Autocorrelation, or serial correlation, occurs when data points
in a time series are correlated with previous data points. This
violates the assumption of independence in regression analysis,
leading to several problems
Marginal VaR, Incremental VaR, and Component VaR
Marginal VaR measures the sensitivity of a portfolio's Value at Risk
(VaR) to changes in the size of a specific subportfolio. It is
calculated as the partial derivative of VaR with respect to the
investment in the subportfolio. The marginal VaR is closely related
to the subportfolio's
beta, a measure of its systematic risk.

Incremental VaR measures the change in VaR when a specific

subportfolio is added or removed from the overall portfolio. It is
calculated by comparing the VaR of the portfolio with and without
the subportfolio.
Marginal VaR, Incremental VaR, and Component VaR

Component VaR, a concept in risk management related to the

breakdown of a portfolio's overall Value at Risk (VaR) into
contributions from individual subportfolios. Component VaR is
defined as the partial derivative of the portfolio's VaR with respect
to the investment in a specific subportfolio. It essentially measures
the impact of each subportfolio on the overall risk of the portfolio.
The text highlights two key properties of Component VaR:

1. The ith component VaR for a large portfolio is approximately

equal to the incremental VaR for the ith subportfolio. This means
that for large portfolios, the change in VaR due to adding or
removing a subportfolio is roughly equal to the component VaR of
that subportfolio.
2. The sum of all the component VaRs equals the total VaR for the
portfolio. This demonstrates that the individual contributions of
each subportfolio sum up to the overall risk of the portfolio.
Buy
Euler's theorem, which states that the sum of the products of each
subportfolio's investment and its corresponding component VaR
equals the portfolio's VaR. This theorem provides a useful tool for
allocating risk across different subportfolios.

Back-testing involves assessing how well a VaR model would have

performed in the past by comparing its predictions to actual
historical data. It's a crucial reality check to ensure the model is
reliable and not underestimating or overestimating risk.
The concept of bunching, which is a separate issue
from the number of exceptions. Bunching refers to
the situation where the exceptions are not
independent, meaning they are often bunched
together in successive days. This can be a problem
because it can lead to an overestimation of the VaR.
Thank you!