This article is a guest contribution by John P. Hussman, Ph.D., Hussman Funds.
Last week's stock market advance clearly reflected relief over an uptick in the ISM manufacturing Purchasing Managers Index (which advanced slightly to 56.3) and fewer employment losses than expected in the August report. Still, the market's reaction was selective from the standpoint of economic data, reflecting the tendency to seize on any piece of comforting news. Objectively, last week's data was mixed at best, as the broader-based ISM services PMI dropped from 54.3 to 51.5, paced by less growth in new orders, and a shift to contraction in the employment component. The U6 unemployment rate, which includes discouraged workers, moved up to 16.7%. Total hours worked were stagnant. New unemployment claims also remained high, at 472,000, though down slightly from the upwardly revised 478,000 claims the previous week (initially reported at 473,000).
The elevated focus of investors on small bits of news is palpable, and can be observed in the recent series of days where more than 90% of stocks have been either up or down. These strong responses to day-to-day news are somewhat understandable, because a further economic downturn is not well-priced into the market, and would most likely prompt a whole host of policy dilemmas. News that appears to reduce that risk is met with relief and short-covering, while news that validates economic concerns causes a spike in risk aversion.
Over the course of the market cycle, one of the primary areas of risk for stocks (and conversely, one of the best periods for Treasury bonds) is typically the "recognition window" where economic activity begins to deviate from the upward trend that is priced into the market, and investors begin to recognize that an economic downturn is, in fact, likely. In my view, the instant relief provoked by the manufacturing PMI and the employment report was an overreaction to data that is still very early in that window. The typical lead times between deterioration in reliable measures such as the ECRI weekly leading index (and several of our own measures) and deterioration in "coincident" economic activity tend to be on the order of 13-26 weeks. For most economic indicators, we are not there yet. This is why I emphasized two weeks ago that "the much earlier deterioration in economic measures is not encouraging, but it also opens up the possibility that we may see some misleading 'improvement' in the data in the next few weeks before we get into the more typical window of deterioration."
Given that comment, I was a bit dismayed last week by a flurry of inquiries we received just after the ISM data was released, asking whether the slight uptick in the manufacturing index was a "game changer" that removed any concern about further economic risks. The same happened in response to the August employment report. If investors who read these comments weekly - and knew that we could observe misleading "improvement" in the data - were still prone to interpret last week's data as some sort of "all clear" signal, it's no surprise that investors may have done the same on a much broader scale.
Unfortunately, this "all clear" impression is based on a lack of appreciation for lead-lag relationships, and the misconception that economic evidence and market movements come in on an exquisitely perfect timeline. The tone of economic data does not turn all at once. There are leading indicators, "coincident" indicators, and lagging indicators. Once leading indicators have clearly deteriorated, it takes a while for coincident indicators to follow. During the interim, small positive surprises in coincident indicators are unreliable. There is also some amount of variability in the timeline. From this perspective, my sense is that investors have abandoned the concern about further economic weakness prematurely.
To clarify the picture that I suspect we may see within the next few months, the chart below presents the historical relationship between the ECRI Weekly Leading Index (growth rate), the Philadelphia Fed Index (2 month smoothing), and the ISM Purchasing Managers Index (manufacturing). As I've noted before, movements in the ECRI leading index tend to precede movements in the PMI by about 13 weeks. Similarly, the Philly Fed index tends to lead the PMI, but only by a month, on average. The red line on the chart below presents the weighted average of the WLI and the Philly Fed Index, shifted ahead by those typical lead times, while the blue line depicts the Purchasing Managers Index.
Note that the latest reading on the PMI was fairly flat, but from the standpoint of the typical profile, was certainly not an outlier. Based on the ECRI and Philadelphia Fed data, however, the prospect of continuing strength in the PMI is not good.
Look closely at the graph above, however, and you'll note that it would be wrong to assume the PMI will follow the ECRI and Philly Fed data with exact precision. Observe, for example, that in 1974, the ECRI and Philly Fed index broke down sharply, but it took 7 months (beyond the normal window) until the Purchasing Managers Index abruptly plunged below 50. By then, the S&P 500 had cratered to one of the worst bear market lows of the post-war period. Likewise, if you look at the downturn in early 2008, the ECRI and Philly Fed index gave a timely warning, but the ISM data held near the 50 level for several months longer than usual before finally plunging. In the intervening period, the S&P 500 lost 20%.
Suffice it to say that it is premature to interpret last week's somewhat benign data as an "all clear" signal for the economy. Yes, this time may be different, and we may somehow skirt evidence that has historically been reliable, but we don't have a clear logical justification based in other data to support a rosy view. Even when statistical relationships are quite strong, the fact is that "coincident" evidence of economic weakness does not follow the leading indicators with flawless precision. We work with probability distributions - not forecasts - and distributions (picture a bell curve) have variation. There is no way to remove this uncertainty, and it is dangerous to assume that last week's data have done so. From my perspective, economic risks continue to be quite serious.
