The number that broke the bank

As blogee's know I have opined at length that the failure of the banks was driven by a rapacious demand for risk by their investors. The logic of this lies at the heart of the valuation process when dealing with near the money, high gamma businesses.

However, this does not explain how it was that the banks and the ratings agencies so badly mis-priced the sub-prime mortgage assets held by banks. The logic of the securitization process is that risk can be parcelled out via tranches and priced according to the risk appetite of different investors. The lowest priced, highest yielding tranches absorb the bulk of the risk of default flowing through the mortgage pipeline. The highest priced, lowest yielding tranches are assumed to be essentially risk free. It was these senior and super-senior tranches, with a first call on the pipeline, that were bought in large amounts by banks and other financial institutions. The way the pipeline was cut, i.e., the percentages of the mortgage asset pool going to the various tranches, determined the risk they were expected to carry.


The most senior tranches were designed to carry virtually no risk and it was these that found their way onto the balance sheets of the world's banks and it was the write off's of these securities which did the damage and led to the collapse in the system. So what was the problem - how was their risk so badly misjudged? An anonymous student asked me recently whether portfolio theory was examinable within the Advanced Financial Management syllabus. My answer is this: portfolio theory explains what went wrong with the banking system and, undoubtedly, if you don't know portfolio theory you don't know modern finance. Harry Markowitz's great work was not just in the quantification of risk but also in recognising that it is how individual risk interacts that determines the overall risk of the system. The way risks interact is through correlation and it is not understanding the correlation between defaults in the housing market that the ratings agencies got it all wrong.


The agencies had very little back information about the US property markets. Indeed, early studies of movements in real estate values gave little clue about the degree of correlation across the various states of the US. What turned out ot be the case is that the degree of correlation in property price movements was highly positive. Prices were falling in Florida and Washington State, and in California and in New York... Indeed, every where prices were in free fall.

Technically, not getting the default correlations right and poor lending practices led to a situation of over-valuation. But, there is another point which I talked about a long time ago on this blog. The collapse in the market was as much due to panic selling in the limited market for mortgage backed paper as it was to systematic over-valuation. Mark to market accounting means that when asset prices fall, irrespective of the reason, balance sheets must follow. That is OK when we have every confidence in the market fairly reflecting intrinsic value - but when markets go into crash mode it is worth considering whether a temporary halt to M2M would have been and would be in the future a wise thing to do,