To read the whole report was not easy for me, being not a practitioner, and consequently I cannot seriously summarize - not about questionable trading practices.
So, I concentrated on valuation practices, especially VaR. And it turns out to me that VaR was not misunderstood but manipulated?!
I have touched VaR in The Blame Games of 2008, and remember an assessment of Paul Wilmott: VaR is often used to hide risk.
Yes, if one manipulates the complete result, questions like, how do VaRs backtest? - are the break days randomly distributed? what happened in critical periods? ... are worthless.
But if a new model that increases risk is introduced, such simple tests should give alarms. But only if somebody is authorized to perform them on the bank's VaR cubes independently - risk management and controlling should do ...
Two among the seven recommendations in the report are related to valuation and VaR
- Federal regulators should investigate trading activities that cause breaches of VaR
- Investigate (risk) models that substantially lower risk
Clearly regulators require derivatives performance data, hedge documentation, strengthen credit derivatives valuations (I still think impossible with the most complex instruments), ...
In a side sentence they say use independent pricing services.
Maybe, I am naive, but I always thought such large banks have strong model validation and powerful risk groups?
If the governments cannot trust banks, when doing their own VaR, will their regulators do VaR calculations for them?
IMO, this must fail, because it made the system even more complex. Remember, safety mechanisms in a complex tightly coupled systems do not always make them safer. The worst kind of safety system is the one that gives false comfort.
However, great regulators shall not only command "VaR!", but give insight in what and how before reviewing.
