The IMF, toxic assets and the taxpayer

Dreaded toxic assets: the general perception is that a lot more is still lurking out there. According to the IMF and its latest “Global Financial Stability Report”, the bill for the world’s financial sector has now swollen to the enormous figure of USD 4,100bn; banks have to shoulder some USD 2,800bn. Although financial firms around the world have already recognized more than USD 1,300bn of losses (USD 950bn by banks), the “reality-gap” has in fact risen. So, with its new estimate the IMF did nothing less than remove the light at the end of the tunnel for banks. The figures sit oddly with the recent stabilization of the financial sector. However, the markets, jaded by an overflow of bad news in the past, seemed to decide simply not to listen and continued with the rally of bank stocks.

Could it be that the loss calculations are themselves pro-cyclical? Certainly it is important to look at the figures in more detail. The IMF analyzed four classes of assets: residential and commercial mortgages, consumer and corporate debt; for all classes, it looked at loans still sitting on banks’ balance sheets as well as at securities, including structured loans.

As in the past the IMF used market data to estimate losses for structured products, i.e. derivative indices like the ABX index. Of course, pricing on ABX is convenient. It is the only transparent and publicly available index for securitized subprime mortgages. However, ABX prices reflect the technical imbalance of supply and demand more than the fundamental value of the underlying assets. Trading is thin. Using ABX prices can produce loss estimates wide of the mark.

But it is not only that these indices may not be very reliable. More disturbing is the fact that the IMF, in calculating the losses for securities, seems to follow slavishly market sentiment, a blemish also embodied in our accounting regulations for capital market instruments. That sits ill with its new mandate, given by the G20, to inform policymakers around the world of market exuberances countering the pro-cyclical herd behavior of market participants. Early warning can only fulfill these expectations if based on independent analysis. How can an approach that derives its forecasts from actual market prices lead to new information that the market does not already know? Applying IMF’s methodology at the height of the last boom would have resulted in “forecasts” that the banking sector is massively overcapitalized.

In this regard, the IMF’s approach to estimate losses for loans, from prime residential debt to plain vanilla corporate loans, is more robust. Here, the IMF uses its own models to predict charge-off rates for different loan types. Given its gloomy economic outlook and the size of the global banking industry sitting on more than USD 40,000bn of loans, it is no surprise that the IMF now expects that loan losses will top USD 2,000bn.

On the contrary, it is pretty obvious that rising credit losses are likely if the real economy were to deteriorate further. But these are normal second-round effects. By sharply increasing loan loss provisions, banks are preparing themselves for worse to come. Credit losses triggered by an economic slowdown are by no means signs of a financial crisis or evidence of wrongdoings or structural failures. It seems as if the IMF has attempted to calculate all credit costs deriving form the current world recession. However, slapping such “normal” costs of banking onto the bill for the crisis only inflates loss estimates and is pro-cyclical.

 The full Working Paper can be downloaded below.

PDF ( 144 kb)