Summarise the criticisms of VAR as a risk management tool.

VAR is one of the simple and widely used risk measures that attempt to summarise the total risk of the portfolio. Despite of its popularity within Financial Intuitions, Treasures and Fund Managers, there are frequent criticisms against its use which we will discuss in this part.

One of the criticisms is that VAR focuses on the risks around the middle area of the distribution and completely ignores the tail portion which is associated with large losses. (Glasserman, Heidelberger & Shahabuddin, 2002, P239). So, the probability of the portfolio losing side has not been evaluated enough. For example, the interpretation of VAR number $904,617 calculated previously for the bank’s portfolio is that there is a 99% probability that the maximum loss will not exceed $904,617. This may not be the case if the 1% loss is of significant amount and of unpredictable nature. Measuring rare events such as Bank Robberies and Natural disasters are almost impossible. The use of historical data in such context is not sufficient enough predict the future which can lead to excessive risk taking or not hedging property. It may turn out to be like an airbag in a car that works all other times but the time of an accident.

Another criticism is regarding the Subadditivety. The sum of VAR of two portfolios is actually larger than the sum of 2 VARs. The end result should be either equal or lesser than the sum because diversification actually reduces the risk. This violates diversification principle and it exists because VAR is not an actual loss amount but a quantile on the distribution of profit and loss. Accepting sum of VAR as a maximum loss amount can create a false sense of security for Senior Management and executives. A study by Danielsson & Jorgension 2005 found that VAR is normally subadditive for all fat tailed distributions except for distributions with super fat tails. Some examples of super fat tails are Options and currency exchange rates.

VAR number is powerful tool and it is also prone to misunderstanding because of its simplicity. For optimal risk management

VAR should be complemented by other risk measures because of the unexamined tails. Risk Managers need to ensure that there is a plan in place to limit losses. There are several measures such as Back Testing, Stress Testing, Stressed VAR, Standard Deviation, Mean Absolute Deviation, ARCH and GARCH. Back testing and Stress testing are 2 of the main measures that we will discuss in this part.

Back testing involves looking at the past data over a specific period of time and checking the frequency of losses exceeding the VAR amount. It can be computed using the following formula.

Stress testing should always complement VAR.

Historical Simulation This method employs historical returns data to assemble the cumulative distribution function, and does not place any assumptions on the shape of the distribution. A historical simulation simply sorts the returns by size. If the sample include 100 returns, the value at risk at a confidence of 95% is the fifth largest loss. Several criticisms are often made of this approach.
Two portfolios with same VAR may have very different expected shortfall.

■Historical simulation assumes that returns are independent and identically distributed. This not necessarily the case; real-world data often displays volatility clustering. ■Returns in the recent-past and far-past are given equal weighting. However, recent returns have greater bearing on future behavior than older returns. ■This method requires a large set of historical data for accuracy; this, however, is not always available. ■Because the method is entirely reliant on historical data, the result cannot be influenced by subjective information (as with Monte-Carlo simulation). This may be significant if a fund manager predicts large changes in the business environment.