The trouble with VAR and other mathematical models of risk is that if it becomes the dominant paradigm, and everyone begins to use it, it creates distortions in the market, because institutions gravitate to asset classes that the model makes to appear artificially cheap. Then after a self-reinforcing cycle that boosts that now favored asset class to an unsupportable level, the cashflows underlying the asset can no longer support it, the market goes into reverse, and the VAR models encourage an undershoot. The same factors that lead to buying to an unfair level also cause selling to an unfair level.
Benchmarking and risk control through VAR only work when few market participants use them. When most people use them, it becomes like the portfolio insurance debacle of 1987. VAR becomes pro-cyclical at that point.
Every month and quarter, multiple reports on average hedge fund returns are released from several sources. However, it can be difficult to sift through the many returns to uncover the most consistent hedge funds. The good news is that Eric Uhlfelder recently released his "2022 Survey of the Top 50 Hedge Funds," which ranks the Read More
Sometimes I think the Society of Actuaries is really dumb. The recent financial crisis demonstrated the superior power of long-term actuarial stress-testing versus short-term quant models for analyzing risk. The actuarial profession has not taken advantage of this. Now, maybe some investment bank could adopt an actuarial approach to risk, and they will be much safer. But guess what? They won’t do it because it will limit risk taking more than other investment banks. Unless the short-term risk model is replaced industry-wide with a long-term risk model, in the short-run, the company with the short-term risk model will do better.
The reason why VAR does not effectively control risk is simple. VAR is a short-term measure in most of its implementations. It is a short-term measure of risk for short- and long-term assets. Just as long-term assets should be financed with long-term liabilities, so should risk analyses be long-term for long-term assets.
This mirrors financing as well, because bubbles tend to occur when long-term assets are financed by short-term liabilities. Risk gets ignored when long-term assets are evaluated by short-term price movements.
And, as noted above, these effects are exacerbated when a lot parties use them; a monocultural view of short-run risk will lead to booms and busts, much as portfolio insurance caused the crash in 1987. If a lot of people trade in such a way as to minimize losses at a given level, that sets up a “tipping point” where the market will fall harder than anyone expects, should the market get near that point.
The idea that one can use a short-term measure of risk to measure long-term assets assumes that markets are infinitely deep, and that there are no games being played. You have the capacity to dump/acquire the whole position at once with no frictional costs. Ugh. Today I set up a new client portfolio, and I was amazed at how much jumpiness there was, even on some mid-cap stocks. Liquidity is always limited for idiosyncratic investments.
The upshot here is simple: with long term assets like stocks, bonds, housing, the risk analysis must be long term in nature or you will not measure risk properly, and you will exacerbate booms and busts. It would be good to press for regulations on banks to make sure that all risk analyses are done to the greater length of the assets or the liabilities (and with any derivatives, on the underlying, not contract term).
By David Merkel, CFA of Aleph Blog