Saturday, 2 May 2009

CREDIT RATINGS AGENCIES BET ON GLOOM

McCreevey, the EU DG for Competition is pushing through a law to force the major credit ratings agencies, Moody's, S&P, and Fitch to publish and validate their models and ratings systems. If they want t stay in business they must become more transparant and ensure that the disastrous kinds of absolute failures that Moody's had to confess to in its securitisation ratings models (announcement 16 June 2007) are never repeated. The ratings agencies are under threat from governments and are critically in the gunsights of the G20 agenda. They are also going to be arraigned in class action suits. Their commercial survival is threatened. This fundamental problem going forward is not holding them back from rattling the cages of the markets with threatened downgrades and dire warnings. The latest is corporate bond defaults. I find Moody's forecasts just absurd and irresponsible. The banks are in the firing line along with the corporations if corporate bond defaults rise to the percentages forecast. It is in the power of banks directly, and governments indirectly via fiscal reflation, and sometimes directly e.g. GM, plus shareholding investors and institutional investors especially have yhet to be tapped by non-bank corporations for significant help. Therefore, corporate defaults are something that can be managed, mitigated, assuaged, and we are supposedly in the US and UK within 2-4 quarters of return to positive GDP growth. Therefore, what is the basis of Moody's estimates. It seems they have simple trend line forecasts and are not taking account of risk mitigation factors.
Moody's is rightly being heavily blamed for bug-ridden ratings models applied to securitisation issues that were indifferent to rising defaults.
But insofar as Moody's, S&P and Fitch feel the heat of major blame for the crisis, essentially for applying 'through-the-cycle ratings generally and not advising enough on actual short term risks, and like many forecasters who missed the turning points, they are now erring on the other side by being excessively gloomy, and as ever assessing risks gross oblivious to collateral and risk mitigation actions available to all inolved. As a risk rater, predicting more gloom has an obvious gain of believing that it is harder to be blamed for being doom-laden than for being euphoric, for bearish rather than bullish opinions. The assumption is that everyone currently will be much more forgiving of an overly pessimistic forecast than an overly optimistic one, so best be biased towards 'Roubini' scenarios. Of all the lawsuits in the pipeline against the Ratings Agencies, none are lawsuits for over-estimating defaults. If they keep on like this I for one can forsee precisely such lawsuits. They are not impossible to define.
As per optimistic forecasts, there are few benefits to accuracy; when has a risk manager earned bonus for predicting shock events won't happen that subsequently don't? Coming up with improbabilities (Nassim Taleb's 'Black Swans' building on Plato's white swans and John Stuart Mill's use of black swans as examples of inductive reasoning, or my own alterntive South American Swan theory, of swans that are only black from the neck up!) is above almost everyone's 'pay-grade' even that of CEOs and their Boards. Governments never publicly predict recessions. Everyone fears that gloomy forecasts risk becoming self-fulfilling. Yet, these are precisely the risks that the ratings agencies are now running as a knee-jerk reacion to all the criticism they have received; they are biting back with vengence!
The default rate for speculative grade corporate loans was 4.1% in 2008, which is 71st out of the 89 years of data on this series, below 2001, 1990, and 1970 peaks. The prediction (above) for a 16% default rate in 2010 is a record spike in defaults, higher than any year, including 15.4% in 1933. If Armageddon turns up, you can't blame Moody's for not warning us.
Moody's forecasting model is dubious and simplistic, based on few factors and context-free factors at that, too few observations and embarrassingly incomplete macro-economic models. The data is highly serially correlated, implying the number of observations vastly overstates the range of outcomes, typical monte-carlo.
They operated a new model in 2007 called the Credit Transition Model. This uses recent credit grade transitions, an unemployment forecast, plus yield spreads (currently high, especially when many famous brands are offering 50% higher coupon than bank rate plus LIBOR). Credit risk grade 'buckets' are broad and it takes a lot to transition to lower grades, but this has been eased and there is a wholesale trends of downgrade transitions, and so projecting those forward easily generates an increase in default rates. But, major corporations are split by sector are not highly granular buckets. Therefore it is actually not straightforward to translate PD rates and stress LGDs to produce actual value of defaults. The unemployment rate is going up, but how that correlates with defaults is non-trivial to model and forecast. There are likely to be some finger in the air assumptions here, especially when not part of a complete macro-economic model. Do High Yield spreads correlate with future default rates? If so then we would simply place corporate borrowers into risk grade PD buckets depending on their bond yields and forget eveything else?
Moody's predicted rising corporate defaults in 1998 that did not arrive. In 1990 and 2000, they did not predict the increases in defaults, they merely changed credit ratings as default rates arrived. To the extent default rates are serially correlated, the high yield spreads did not add much information, nothing much to be relied upon. Better would be to go back to basics and Du Pont Ratios.
A bond trading well below par is in trouble, but the hypothesis that aggregate high yield spreads predicts future default rates as based on historical data, is very doubtful. Nonetheless, the new model shows an eerily accurate forecast over the past 15 years (according to the Moody's Credit Transition Model 2007 document there was a successful backtesting: Beware a time series model designed to make forecasts that inevitably has to incorporate past trends and yet be a necessary simplification, hence the back-test is merely the model's slightly simplified way of representing the recent past, not evidence of an ability to forecast the past as it might forecast the future. Moody's Annual Default report 1999 published in January 2000, there was a prediction of future defaults falling in speculative grades to "between 6.0% and 4.5%", but "trending downward over the year", also with "increasing recovery rates". They also noted the US moving away from the 'Asian Crisis' of 1998, and so were oblivious to the next bubble burst and stock market crisis leading into recession. Recessions happen suddenly and aggregating economic forecasts always result in compromise smoothing of the data, on top of which is the problem that GDP data gets revised in hindsight severely for up to 2 years after the data is first published. The fact is that users of Moody's and the other ratings agencies models expect a superior insight based on very detailed annual reports analysis etc. commiserate with what they imagine to be the rigorous analysis of corporate credit ratings based on some models of Du Pont ratios. Instead what we have is a simple-minded model that any analysts could construct in a spreadsheet from readily available information, and very little of that!
So, like those AAA-rated circa 2003-mid-2007 Mortgaged Backed Securities, Moody's is giving the market exactly what it agrees with and wants right now, subjectively, not objectively, more of the same lack of intellectual courage and lack of academic rigour. It is hard to vouch for whether S&P and Fitch deserve to be tarred by the same brush. But, so far that seems a reasonable working assumption?
As Eric Falkenstein says in his seeking Alpha article from which I took much by way of inspiration for the above, "part of being a professional credit executive is knowing the difference between what you can predict and what you can't. Moody's should simply stop forecasting aggregate default rates. They don't add any value here and merely highlight that for anything many people already have an opinion on, Moody's is not a special lens, but rather a mirror." That I broadly agree with from an investor viewpoint. From a concern about banks' balance sheets, aggregate defaults are certyainly worth analysing and modeling and fporecasting, but clearly they require a lot of sectoral disaggregation and to be part of a fully worked out macro-economic model, which Moody's is clearly not resourced for!