Calculated Risk reports that S&P is increasing its forecasts for losses on subprime mortgages again. As I’ve said before, in principle this means that their expectations about the economy are worse today than they were yesterday. They’re not just saying that defaults will go up; they’re saying that they will go up by more than they thought before today.
I previously discussed why I think this is weird, and there are a number of good comments to that earlier post. My theory that they are trying to spread out the deterioration of their forecasts over several months to save face got some support. q and others explained that rating agencies lag the economy because they base their forecasts on published economic data. That may be true, and it may be the best explanation for what is going on, but if so it seems like a condemnation of the rating agencies, since their job is to estimate the likelihood of default, and one of the inputs to their models should be the economic situation. (Or maybe their job is to estimate default likelihood assuming “normal” economic conditions, and it’s up to the investor to adjust accordingly.)
Note that these are their macroeconomic forecasts, not revisions to ratings of specific bonds, so the bond-rating schedule isn’t the driving factor here.
By James Kwak
The Baseline Scenario 
I wouldn’t call this forecast vacillation a “condemnation” of the rating agencies. It’s just confirmation that the accuracy of a prediction based on predictive modeling that is driven by prior economic events (vs. the cycles of the moon, say, or the length of women’s skirts) will vary inversely as the volatility of the economy (or the length of skirts), even when that volatility is unidirectional (fast growth or fast shrinkage or big undulations have the same effect). There are mathematical methods to deal with this, but they work much better in a price-setting situation (where seller sets price and buyer takes it or leaves it). Anyway, to the extent that one’s forecast is based on prior statistical data rather than one’s crystal ball, it will be as inaccurate as the data it uses varies.
Yeah, the rating agencies are fallible, and, yes, as the data is published, they and their algorithms learn. They’re like us; they’re just supposed to be better at this than we are. To improve forecasting accuracy and reduce investment risk, just stabilize the economy – simple, not easy.
I put this link up before from the “NOW” program on PBS television. It explains some of the problems at the credit ratings agencies. Many employees knew the ratings were bogus, but who’s going to stop the gravy train??? They’ve done nothing to correct the incentives which create a big conflict of interest. The problem will rear its ugly head again in a very systemic way. The question isn’t if, but when.
Here is the link. After you jump there click where it says Video: Credit and Credibility or just click on the crumpled up dollar.
http://www.pbs.org/now/shows/446/
Hmmm. Didn’t S&P give Lehman a high rating just before it bombed? Not to mention other ridiculous ratings it awarded that contributed to the spiral and collapse of the the financial market. Now why should we listen to them? Seriously.
from bloomberg- Standard & Poor’s again boosted its projections for losses from U.S. subprime mortgages backing securities, reflecting increasing delinquencies and defaults amid slumping home prices and growing unemployment.
Philly Fed coincident indicators came out the 21st of July. When was the last time Moody’s this? It shouldn’t be that hard to to see if they are revising at certain times and see what is possibly in their model.