Jim Rickards, the former Long Term Capital Management general counsel who, as
principal negotiator of the rescue of LTCM knows a thing or two about Wall Street risk, will testify tomorrow at 10 am before the House Science Committee's Subcommittee on Investigations about the risks of financial modeling.
Here's a summary of the main points and an advance copy of his testimony:
- The financial crisis would not have happened but for the fact that bankers, investors, rating agencies and regulators all relied on quantitative models which told them risk was under control. These models go under the general name of "Value at Risk" or VaR.
- VaR is fatally flawed. It relies on three assumptions about (i) efficient markets, (ii) random price movements and (iii) risk distributed in a normal or "bell curve" shape. All three assumptions are wrong. Markets are not efficient, prices do not move randomly and risk is not distributed in a bell curve.
- There is a better model. It is built around a power curve rather than a bell curve. The power curve accounts for frequent large scale failures. It tells us that risk is a function of the size or scale of the system. But as we increase scale, risk does not grow in a straight line, it grows exponentially. This is why the system failed in such a spectacular fashion the past few years.
- If scale is the problem, it is also the solution. By de-scaling we can reduce risk to manageable levels. We do this in three ways: (a) reinstate something like Glass-Steagall to separate commercial from investment banking, (b) put all derivatives on exchanges where risk can be netted out and collateral maintained at safe levels and (c) reduce leverage by imposing stricter capital ratios for banks and brokers.
- There has been a lot of talk about this but little action. The identical lessons of the 1998 collapse of LTCM were never learned or quickly forgotten. Will that happen again?