Using data released Friday morning, Jeremy Piger updated his estimates of U.S. recession probabilities through August of 2009. The results now suggest that the probability of recession was below 50% for both July and August. He notes that:
According to the model, the trough is June of 2009, with the peak being December 2007 (which matches the NBER peak). This makes the recession 18 months in duration, which would be the longest post-war recession (by two months over the 1973-1975 recession and the 1981-1982 recession).
Note that if I were to estimate a model that used only employment data, rather than the four variables highlighted by the NBER, the probability of an "employment recession" would still be quite high.