Explanation:
When backtesting a VaR model at the 99% confidence level, exceptions (breaches) are expected to be very rare (only 1% of the time). This makes it statistically difficult to determine whether a model is accurately capturing risk or if it is flawed, resulting in a low-power statistical test. By shifting to a 95% confidence level, the expected number of exceptions increases to 5%. Having more data points (exceptions) increases the power of the statistical test, reducing the probability of making a Type II error (failing to reject a faulty model). Consequently, decisions regarding the rejection or acceptance of the VaR models become much more reliable.
Ultimate access to all questions.
Q.45 A newly recruited intern has been tasked with backtesting a firm’s VaR model. Since inception, the firm has always calculated 1-day VaRs at the 99% level of confidence. In a marked departure from the recommendations set by the Basel Committee, the intern has proposed a shift to the 95% level of confidence. What will be the implication of making this switch?
A
Decisions on whether to accept or reject VaR models based on backtesting will be more reliable
B
The power of statistical tests will decrease
C
The probability of making both type I and type II errors will increase
D
The 95% VaR model is less likely to be rejected using backtesting than the 99% VaR model
No comments yet.