TABLE OF CONTENTS In this chapter, we discuss various stages of refinements of the plain-vanilla VAR discussed in Ch. 26. Increasingly realistic aspects will be included, with the final aim to obtain a risk measure that is more useful in active risk management. The first set of improvements give what is termed in this book "Improved Plain Vanilla VAR" ( IPV-VAR). We then define a series of further improvements to produce "Stressed VAR" and finally "Enhanced/Stressed VAR" ( ES-VAR) [1]. We close with some miscellaneous topics including subadditivity issues, and also an integrated form of VAR.
The following table summarizes the next stage, including refinements past the PV-VAR to obtain the IPV-VAR, or Improved Plain-Vanilla VAR. These refinements are often included in current implementations of VAR.
| Quantity Compared | Plain Vanilla VAR | Improved PV VAR |
|---|---|---|
| Convexity | Not Included | Included via Grid |
| Time Scale dt | Uniform (10 days) | Variable (liquidity) |
| Cutoffs for d tx ? | Not included | Included (Judgment) |
| Time Period: x ?Data | Recent (1 to 3 yrs) | Recent or Variable |
We describe these improvements in the IPV-VAR one at a time.
Convexity exists in all option products, and even to some extent in discount factors. Convexity effects can be included in a VAR calculation if a grid of exposures is available. A given variable x ? is changed by discrete amounts to values on a grid, { x (Grid) ?}, for example x (Grid) ?
