有哪些方法可以對投資組合進行壓力測試?


18

Wikipedia列出其中三個:

  • 極端事件:假設歷史事件再次發生,則假設投資組合的回報。當前頭寸和風險敞口與歷史要素收益結合在一起。
  • 風險因素衝擊:以用戶指定的數量衝擊所選風險模型中的任何因素。因子暴露保持不變,而協方差矩陣則用於基於因子與衝擊因子的相關性來調整因子收益。
  • 外部因素衝擊:代替任何風險因素,衝擊任何指數,宏觀經濟序列(例如,石油價格)或自定義序列(例如,匯率)。使用回歸分析,可以估算出衝擊帶來的新因素回報。

但是我想知道是否有更全面的列表。

8

Well all that you have cited seems quite all you can do with scenario maybe I can add another one which is portfolio dependent. Instead of looking to arbitrary scenarios you first decompose the factor to which you portfolio is the most sensitive to, and then look for scenarios that are specifically impacting this combination of risk factors.

Anyway,scenario losses can be quite an effective tool in risk management because it shows in a simple way where is your risk standing but it shouldn't be viewed as a standalone indicator.

One flaw is that it usually tell you something about only a few scenarios,but you can't be really sure that other scenarios won't impact you more. Moreover the more you have scenarios the less you know how to order them with respect to each other or with respect to some kind of likelihood.

Otherwise in the field of capital requirement area, I start to ear about Stressed-VaR which means among other things that you raise in a considerable way correlations between risk factors (as it usually happens during crisis) as well as volatility of risk factors.

In the end Scenarios or Stressed-VaR, look a little bit like some "cuisine" with no real theoretical grounds to support the whole building and that's the problem.

Regards