Skip to main content

Perils of Data du Jour Forecasting

Recessions are often recognized when they're about to end. It's no different with slowdowns.

Editor's note: This is a bonus story from Anirvan Banerji, whose commentary usually appears only on


. We're offering it today to

readers. This column was originally published on


at 7:51 a.m. EDT Monday. To read Anirvan's commentary every day, please click here for information about a free trial to


Friday's much stronger-than-expected employment report quelled talk of a new "soft patch" and concerns about a sharp slowdown -- or even a recession -- in the second half of this year.

Remarkably, such fears were stoked mainly by disappointing April releases on retail sales, industrial production, GDP and payroll jobs -- all coincident indicators that tell you nothing about future economic growth.

If you really believe there was a soft patch around mid-2004 and another one in early 2005, I invite you to look at GDP growth (four-quarter percent change, bottom line in chart).

Do you really see a soft patch anywhere? Or do you see what I see -- a fairly steady easing in growth over the past year following a downturn in the growth rate of ECRI's Weekly Leading Index (WLI, top line)?

But the markets have a data

du jour

approach to economic forecasting: Every day, they key off of so-called market-moving numbers that are implicitly placed on a bullish-bearish scale.

Never mind that the number may be a leading, coincident or lagging indicator, or none of the above. Or that it may relate to cycles in economic growth, inflation, or employment, but rarely all three. Such nuances are often lost in the rush to judge the data as bullish or bearish.

Most market participants just take the data

du jour

and extrapolate them indiscriminately into the future, sometimes using backward-looking data to generate unwarranted gloom.

The penalty incurred by such an approach is misjudgment of the direction of future economic growth. Eventually, markets do get it right -- at least well enough for Paul Samuelson to quip that the stock market has predicted nine of the last five recessions.

Remember the 1987 stock market crash? Many felt at the time that this was evidence of an imminent recession. The reality? GDP grew by almost 4% in 1988.

What 'Soft Patch'?
In reality, there's been nothing more worrisome
than an easing of growth

Click here for larger image.

Source: ECRI

A Predictable Easing in Growth

A year ago,

I wrote that "growth in the WLI has eased since last summer, and is now nearing a one-year low. Thus, U.S. economic growth is also set to moderate in the months ahead." As the chart shows, that's precisely what's happened over the past year.

Then, following the proclamation of last year's "temporary soft patch," I affirmed that

"a return to robust growth is nowhere in sight. This is no temporary 'soft patch' but a broad slowdown that will last at least until year-end. It's also part of a global industrial slowdown that's emerging in all the major economies."

That's pretty much what happened -- industrial production growth slowed in all major economies, as evinced by the growth rate of U.S. factory output slowing from 6.25% last August to 4% in March. But the data

du jour

approach led to talk of a new soft patch, chiefly because the markets had expected no such weakness. Remember, just over a month ago, most economists, oblivious to the global slowdown, were predicting 4%-plus first-quarter U.S. GDP growth.

In other words, the "soft patch" diagnosis emerged mostly from the disparity between reality and overly optimistic expectations. If you look objectively at the past year's data, what's evident is a sustained cyclical downturn in economic growth -- no more, and no less.

Typically, recessions are widely recognized just when they're about to end. As you can see, it's no different with slowdowns. Thus, we're probably much closer to the end of this slowdown than to its beginning.

But why were people so worried about second-half growth prospects? For one thing, most economic models, like most people, extrapolate from the recent past, implying that any weakness in the coincident indicators will intensify.

But some of the most persuasive evidence of a "soft patch" may have come from ostensibly sophisticated regression models that fall into the trap of forecasting by analogy.

Flawed Analogies

As I described the practice of

forecasting by analogy in late 2003, "analysts often assume that because the present resembles earlier periods in some way, it does so in other ways. They then predict that the current period will exhibit similar patterns. If the chosen analogy doesn't work, they hunt around for other analogies -- other periods this might better resemble. And if none of their analogies work, they proclaim a 'new paradigm.'"

Most regression models are driven by these kinds of historical similarities. Such models are nothing but fancy tools for forecasting by analogy that can be dangerously misleading without clear insights into the similarities and differences among historical episodes.

It's well known that a combination of oil price and interest rate shocks helped trigger the past five recessions. Thus, a regression model driven by those two variables would predict that the current upswing in interest rates, combined with the latest oil shock, will cause at least a sharp slowdown this year, if not an actual contraction. On the face of it, it's a plausible premise -- after all, the last time such a combination appeared, in the fall of 2000, I anticipated an

oil recession myself.

But there's a key difference. Back then -- and before every one of the four prior recessions -- the oil shock arrived when reliable leading indexes like the WLI showed the economy to be vulnerable to shocks. That's not the case today.

Our research into business cycles shows that it's the


of a cyclically vulnerable economy and external shocks that trigger recessions. Because that conjunction is not present today, oil prices $20 above the 2000 peak haven't produced anything close to a recession.

Of course, high oil prices are a drag on economic growth. But they aren't enough of a drag to produce a sharp slowdown or contraction without WLI growth turning decisively downward.

In fact, the ability of WLI growth to maintain a mild cyclical uptrend in recent months despite a plunge in investor optimism suggests that even if there's some near-term weakness, U.S. economic growth will hold up in the second half of 2005. Basically, our leading indexes reveal that the combination of accommodative monetary policy, accumulating corporate profits and strength in housing is likely to keep the economy afloat despite the shocks.

Divining the Future

In sum, the coincident indicators that so spooked the markets in recent weeks can't foretell the future direction of the economy. Ditto for simplistic regression models based on false analogies.

Both the data

du jour

approach and forecasting by analogy risk missing the forest for the trees. The best way to monitor economic cycles is to get a bird's eye view of the forest itself, on the basis of an array of specialized leading indexes, each based on a judicious combination of reliable leading indicators. Some of those leading indexes look even further ahead than the WLI, and help to clearly identify the upcoming undulations in the economic landscape.

Today, that tried and tested approach is telling us that we are approaching the end of the slowdown, and that growth will hold up in the second half. That's a far cry from the gloom and doom that's been roiling the markets.

Anirvan Banerji is the director of research for the Economic Cycle Research Institute, which was founded by Dr. Geoffrey H. Moore, creator of the original index of leading economic indicators (LEI) for the U.S. Department of Commerce. Banerji is on the economic advisory panel for New York City and is the co-author of

Beating the Business Cycle: How to Predict and Profit From Turning Points in the Economy

. At time of publication, neither Banerji nor his firm held positions in any securities mentioned in this column, although holdings can change at any time. Under no circumstances does the information in this column represent a recommendation to buy or sell stocks. While Banerji cannot provide investment advice or recommendations, he appreciates your feedback;

click here

to send him an email. has a revenue-sharing relationship with under which it receives a portion of the revenue from Amazon purchases by customers directed there from