Internal Stress Tests

I argued in my July 31st posting that there are rather obvious reasons why the stress tests that regulatory authorities produce for public dissemination might fall short of the severest stringency. Authorities consequently run the risk that, if their tests that are perceived to be undemanding, they might be discredited and even counterproductive, given that publication of their results is intended as a confidence-building exercise. But of course, they also run the risk that excessively demanding tests could damage confidence. Banking authorities must tread a fine line. The same certainly should not be true of stress tests that firms devise for strictly internal consumption. Here it is self-defeating to set the parameters of stress any lower than the limits of what the firm can withstand, since the purpose of performing the tests is precisely to determine what those limits are.

The Credit Crunch revealed that stress can be extraordinarily complex. Sources of risk that are normally treated as conceptually distinct – market stress, credit stress, liquidity stress, operational stress, etc. – all seemed to manifest themselves simultaneously and in bewildering combinations, making it virtually impossible to separate primary causes from subsidiary effects. The Credit Crunch provided ample evidence, if any evidence were needed, that simple models of stress do not reflect the complex realities of systemic crises.

As in other situations, it is sound advice to “follow the money.” While the disappearance of liquidity is not usually a cause of financial crises, it is their invariable accompaniment. It is also the main proximate cause of financial stress, even illiquidity is only a subsidiary effect of whatever event originally caused the crisis. Consider the chart below, which shows the development of one the of financial system’s many sources of leverage. From the second to the third quarter of 2008, $1.1 trillion in credit was withdrawn from the system. On I.M.F. estimates, this increase exceeds all but twelve nations’ G.D.P. and equals 1.9% of the global economy. Such a massive withdrawal of liquidity cannot have failed to affect every leveraged participant in the system, even though it followed directly upon an even more massive increase from the first quarter to the second. Clearly, it would be impossible to trace all the margin calls, forced liquidations and recourses to standby credit facilities that resulted from such wholesale de-leveraging.

At the level of the individual firm, if it is even moderately complex, it is not possible to create a comprehensive test that addresses all the inter-connected sources of stress that were revealed by the Credit Crunch. Testing for stress at a level of such near-universal crisis requires a number of scenarios that increases by the square of the number of position, creditor and debtor exposures the firm carries. This figure rapidly becomes enormous. However, the firm can reduce it to more manageable size by concentrating its attention on margin calls resulting from haircuts to the values of its positions, the callable portion of its other indebtedness and to the default risk among its debtors. This is still a substantial task, but then the stress revealed during the Credit Crunch was itself very substantial. The firm that can hypothetically stress itself to this extent without reaching the limits of its solvency is in extremely sound shape.