Macro or Top-Down approach to VaR and Expected Tail Loss estimation

based on analyzing the variability of profits and losses (P&L) for each risk-taking unit directly (e.g. trader, business unit, enterprise). In contrast with traditional approaches to VaR estimation based on simulating possible market conditions and revaluing the portfolio under those possible future states of the world, this approach is based on a “macro ” or top-down analysis of realized P&Ls. We will argue that this approach should be part of the arsenal of methods used by risk managers, and we will show how it can complement and in many cases improve the ability to forecast risk. Traditional VaR methods or “bottom-up ” approaches Traditional simulation methods to calculate VaR and other risk measures use a bottom-up approach by taking each instrument in a portfolio, and recalculating its value under simulated market conditions for a set of risk factors or prices reflecting possible states of the world. This approach is conceptually sound, but in order to implement it, we need to make many assumptions about possible market behavior for each risk factor individually and jointly, and how the portfolio will react under those conditions. The first generation of VaR models also assumes that the portfolio remains static (therefore ignoring possible exit strategies such as stop-losses or delta-neutral hedging), and usually don’t take into account intra-day trades, transaction costs, liquidity and counter party risks. Firms using these models have found them useful as a risk measure, but due to the number of assumptions, their validity can sometimes be put to question