Did the recession end in May?

It’s looking more and more as if the recession in the U.S. ended in May. To support that claim I have looked at initial unemployment insurance (UI) weekly claims and the pattern that that time series has followed in previous recessions. The four-week moving average of not seasonally adjusted weekly claims has peaked less than eight weeks before the end* of the each recession since 1970 (see chart).

Click on the chart to enlarge.

The lead time was eight weeks in the 1973-75 recession, seven weeks in two recessions (1980 and 1981-82), six in 1970, five in 2001, and zero in 1991. The moving average declined essentially monotonically after each of those identified peaks.

During the current recession, the moving average saw a peak on the week that ended on April 4, 2009. That would put the end of the recession somewhere between April and May (zero to eight weeks later). And if the past is any indication, most likely in May.

The usual caveat applies: the pattern-recognition method is based on only six episodes of recession, and six is a very small number to make any claim on the significance of this finding. Moreover, each episode is spaced by several years from each other, so the hypothetical relationship between UI claims and the business cycle may have changed since last time.

I can’t help, nonetheless, to find some solace in that chart.

*Here I take the NBER dating of the business cycles. The NBER, however, provides the month in which a business cycle trough occurs, not the week. I take the liberty of dating the end of each recession on the last week of the month when the NBER identifies a business cycle trough.