- Published on
Investment - Value at Risk
Companies in the financial services industry believe that the assets and securities they possess will provide them with a positive return. But they also need to quantify the possible loss on an investment if their expectations for the asset or security turn out to be erroneous. This potential loss is commonly assessed using a metric known as value at risk.
Use and Advantages of Value at Risk
Value at risk (VaR) was developed in the late 1980s and is now a commonly used statistic. It relies on statistical notions, such as standard deviation.
VaR gives an estimate of the least loss of value that may be predicted for a certain duration with a given level of probability.
For example, an asset management firm may estimate that a portfolio has a VaR of USD1 million for one day with a probability of 5%. This estimate suggests that there is a 5% risk that the portfolio will fall in value by at least USD1 million in a single day, assuming no further trading. In other words, a loss of USD1 million or more for this portfolio is likely to occur, on average, once in 20 trading days (1/0.05).
VaR offers various advantages:
It is a common statistic that can be utilized across diverse assets, portfolios, business units, businesses, and markets.
It is relatively easy to compute and well understood by senior managers and directors.
It is a valuable tool for risk budgeting if there is a common procedure for allocating capital across business units according to risk.
It is widely utilized and mandated for usage by several regulators.
Weaknesses of VaR There are also limitations inherent in the VaR measure of risk. VaR gives an estimate of the least, but not the maximum, loss of value that can be predicted. Referring back to the preceding scenario, the asset management business can expect a loss of at least USD1 million 12 or 13 times a year (5% of the about 250 trading days a year). VaR does not represent the highest loss of value the portfolio manager may anticipate to sustain in one day, and it does not guarantee that a loss in excess of USD1 million will not occur more frequently than a dozen times a year.
In fact, VaR generally underestimates the frequency and amount of losses, mostly due to erroneous assumptions and models.
First, VaR mostly depends on previous data to anticipate future expected losses. But past returns may not be a strong indicator of future returns. In addition, history is not helpful in forecasting occurrences that have far-reaching repercussions, but are unforeseen or deemed impossible – that is, black swan events.
Second, VaR makes an assumption regarding the distribution of returns.
For example, it is typically believed that returns are regularly distributed and follow the bell-shaped distribution. The use of historical data and the assumption of a normal distribution may perform quite effectively in normal market conditions but not during periods of market upheaval.
The global financial crisis of 2008 is a case in point. Until 2007, most banks had a low daily VaR, which provided them a false sense of security. Once the crisis hit, the number of days when trading losses exceeded the daily VaR and the magnitude of those losses were much larger than projected. Some banks stated that the frequency of losses was 10 to 20 times higher than the VaR estimates, and some banks experienced losses that considerably depleted their equity capital.
To counter these limitations, firms, particularly banks, often adopt alternative risk management approaches in addition to VaR. These complementing techniques include scenario analysis and stress testing, which focus on the influence of more extreme events that would not be adequately captured or examined by VaR.
For example, an asset management firm may perform a scenario analysis by identifying different scenarios for the economy (strong growth, moderate growth, slow growth, no growth, mild recession, and severe recession) and then determining how each scenario would affect the value of a portfolio and the firm’s earnings and equity capital.
The firm may also engage in stress testing by evaluating the consequences of extreme market conditions, like as a liquidity crisis, to make sure that it would be robust and survive the crisis.
It is worth mentioning that the problems associated to VaR apply to all measurements that rely on models. The danger emerging from the usage of models is collectively known as model risk. This risk is related with improper underlying assumptions, the unavailability or inaccuracy of historical data, data problems, and misapplication of models.
Companies in the financial services industry believe that the assets and securities they possess will provide them with a positive return. But they also need to quantify the possible loss on an investment if their expectations for the asset or security turn out to be erroneous. This potential loss is commonly assessed using a metric known as value at risk.
Use and Advantages of Value at Risk
Value at risk (VaR) was developed in the late 1980s and is now a commonly used statistic. It relies on statistical notions, such as standard deviation.
VaR gives an estimate of the least loss of value that may be predicted for a certain duration with a given level of probability.
For example, an asset management firm may estimate that a portfolio has a VaR of USD1 million for one day with a probability of 5%. This estimate suggests that there is a 5% risk that the portfolio will fall in value by at least USD1 million in a single day, assuming no further trading. In other words, a loss of USD1 million or more for this portfolio is likely to occur, on average, once in 20 trading days (1/0.05).
VaR offers various advantages:
It is a common statistic that can be utilized across diverse assets, portfolios, business units, businesses, and markets.
It is relatively easy to compute and well understood by senior managers and directors.
It is a valuable tool for risk budgeting if there is a common procedure for allocating capital across business units according to risk.
It is widely utilized and mandated for usage by several regulators.
Weaknesses of VaR There are also limitations inherent in the VaR measure of risk. VaR gives an estimate of the least, but not the maximum, loss of value that can be predicted. Referring back to the preceding scenario, the asset management business can expect a loss of at least USD1 million 12 or 13 times a year (5% of the about 250 trading days a year). VaR does not represent the highest loss of value the portfolio manager may anticipate to sustain in one day, and it does not guarantee that a loss in excess of USD1 million will not occur more frequently than a dozen times a year.
In fact, VaR generally underestimates the frequency and amount of losses, mostly due to erroneous assumptions and models.
First, VaR mostly depends on previous data to anticipate future expected losses. But past returns may not be a strong indicator of future returns. In addition, history is not helpful in forecasting occurrences that have far-reaching repercussions, but are unforeseen or deemed impossible – that is, black swan events.
Second, VaR makes an assumption regarding the distribution of returns.
For example, it is typically believed that returns are regularly distributed and follow the bell-shaped distribution. The use of historical data and the assumption of a normal distribution may perform quite effectively in normal market conditions but not during periods of market upheaval.
The global financial crisis of 2008 is a case in point. Until 2007, most banks had a low daily VaR, which provided them a false sense of security. Once the crisis hit, the number of days when trading losses exceeded the daily VaR and the magnitude of those losses were much larger than projected. Some banks stated that the frequency of losses was 10 to 20 times higher than the VaR estimates, and some banks experienced losses that considerably depleted their equity capital.
To counter these limitations, firms, particularly banks, often adopt alternative risk management approaches in addition to VaR. These complementing techniques include scenario analysis and stress testing, which focus on the influence of more extreme events that would not be adequately captured or examined by VaR.
For example, an asset management firm may perform a scenario analysis by identifying different scenarios for the economy (strong growth, moderate growth, slow growth, no growth, mild recession, and severe recession) and then determining how each scenario would affect the value of a portfolio and the firm’s earnings and equity capital.
The firm may also engage in stress testing by evaluating the consequences of extreme market conditions, like as a liquidity crisis, to make sure that it would be robust and survive the crisis.
It is worth mentioning that the problems associated to VaR apply to all measurements that rely on models. The danger emerging from the usage of models is collectively known as model risk. This risk is related with improper underlying assumptions, the unavailability or inaccuracy of historical data, data problems, and misapplication of models.
0 Comments