When will this horrible recession be over? According to one surprising source, it s over right now.

The source is Robert J. Gordon, an acclaimed macroeconomist and professor at Northwestern University. It s surprising to learn he thinks the recession is over, because he is one of seven members of the elite Business Cycle Dating Committee of the National Bureau of Economic Analysis . These are the people who decide officially, for the record books, when recessions begin and end usually many months after the fact, when the decision is really obvious. I m unaware of any previous case in which a member of this Committee has ever stepped forward and declared the end of a recession in real time.

Gordon bases his gutsy call on an indicator that he says the Committee never even looks at: claims for unemployment benefits. He s talking about the so-called jobless claims number that is released every Thursday morning before the market opens. Based on detailed data from state agencies, it reports the number of workers who have asked for unemployment benefits in the previous week. As Gordon points out, there is no other major macroeconomic statistics that comes out so frequently, and so close to real time.

According to Gordon s research, in every recession since 1974, the peak in jobless claims came within weeks of the bottom of the recession. This is a remarkable research result, in my opinion. I was impressed a year ago when economist Edward Leamer of UCLA wrote a paper that accurately explained recession timing with just three variables the unemployment rate, total payroll jobs, and industrial production. But Gordon has done Leamer two better. Gordon has it down to a single variable: claims. And because claims data is available nearly immediately, investors can use Gordon s insight to make actual trading decisions.

Claims are typically reported as a four-week moving average, to smooth out some of the random noise from week to week. All Gordon has done, really, is to make the simple observation that the peak in the four-week moving average coincides perfectly with the ends of recessions. I charted the data to prove it to myself, and he s right. Here it is:

One thing jumps out of the chart that has nothing to do with Gordon s indicator the fact that in this recession, we still haven t exceeded the number of claims in the 1981-82 recession. As bad as it seems today, we ve lived through worse, and not all that long ago. It was a lot worse in 1981-82, too, because the size of the work force was smaller then than it is today. So the same number of claims represents a larger percentage. Adjusted for the size of the work force, today s claims are just a little more than half of what they were at the 1982 peak.

Now let s ask a tough question about Gordon s indicator. How do you really know when there has been a peak in claims? Just because the four-week moving average turns down for a couple weeks, how do we know it won t just turn up again and go to new highs?

Gordon himself takes on this criticism. Writing two weeks ago when the four-week moving average was already 3.1% off its early April peak, he noted that the pattern of the decline in magnitude and timing nearly perfectly matched all the previous instances, in which no subsequent higher peak developed. So far he s right. With the new data released Thursday, the four-week moving average of claims is down 4.3%, so the early April reading is looking more and more like a real peak.

That said, yesterday s claims data did show a rise of 32,000 claims after two weeks of declining claims, which caused the four-week moving average to uptick a bit over the prior week. In other words, while it s down 4.3% from the peak, a week ago it was down 5.2% from the peak.

That s no reason to throw out Gordon s big idea. No one expects numbers like this to move only in one direction week after week. But a smart investor always looks for what might go wrong. Apparently the rise in new claims this week was due to layoffs at Chrysler, and that brings up an unpleasant memory a memory of the one time when Gordon s indicator got it wrong.

Take a look at the chart again. Note the recession of 1969-1970, in which a seeming peak in claims was then followed by another peak not a higher peak, but enough to delay the end of the recession. According to Gordon, this second peak was caused by a bitter strike at General Motors, which lasted 67 grueling days. Gordon says, This was a big deal at the time when GM had a 50% share of the U.S. automobile market and 400,000 members of the United Auto Workers.

It wouldn t make much sense for auto workers to go on strike at this point, given the near-death state of the industry. And the industry is a much less important element of the economy than it used to be. Nevertheless, the travails of GM and Chrysler wouldn t be making headlines if they didn t still count for something. Could the auto industry foil Gordon s indicator today just like it did in 1970?

Never say never, but I actually doubt it. There is too much other evidence that the worst is over for this economy. I continue to

that the only truly profound problem facing the economy has been the banking crisis, and all the evidence is that it has now passed. The government s stress tests of the banking system have identified the weak ones, and there is a coherent plan to bolster them.

This is confirmed by evidence from the most risk-sensitive markets. The global markets for credit default swaps what amount to insurance policies on risk in financial instruments of all kinds, from commercial real estate to emerging markets have recovered profoundly from their panic lows of two months ago.

I still think that the recovery from this recession will be slow and painful. The economy is going to have to fight the headwinds of the enormous government debt that has been loaded on to deal with the recession, and with the higher taxes and inflation that will flow from that. But I agree with Gordon. We ve seen the worst. The bottom is in.

believe