Hussman: Watch the lagging indicators
John Hussman is not buying the recent risk-on sentiment that has developed since September began. For example, retail sales numbers are to be released tomorrow. Marc Chandler has said that a number above consensus estimates of 0.3% growth "will likely be seen as further confirmation that that exceptionally soft patch the economy had slipped into (May and June retail sales contracted) has eased."
In Hussman’s view, recent jobs numbers point to further downside ahead. He writes:
Except for a burst of census hiring that briefly pushed payroll growth above trend during the second quarter of this year, job growth has been perpetually below trend over the past two years. During the post-war period, the civilian labor force has historically grown at about 0.15% each month, which currently implies that normal "trend" job growth should be about 225,000 jobs per month.
While last month’s labor report was favorably received by Wall Street, that reception was based strictly on the fact that job losses were not as bad as anticipated, given concerns about a "double dip" in the economy. The problem with this celebration, however, is that analysts continue to overlook the typical lags between deterioration in leading indicators and deterioration in coincident measures, much less lagging ones. As I’ve noted frequently in recent commentaries, the typical lag between deterioration in say, the ECRI Weekly Leading Index and the ISM Purchasing Managers Index is about 13 weeks, and sometimes longer. The typical lag with respect to new claims for unemployment is about 23-26 weeks (which puts the likely window of deterioration at about the October – November time frame), and the typical lag with respect to the payroll unemployment report is, not surprisingly, about 4 weeks beyond that. The critical risk area here extends for several months, not a few weeks.
The labor reports of the past three months cannot possibly be considered to be favorable from a macroeconomic perspective. The reason for this is that these reports were each more than 500,000 jobs short of what should have been expected.
Hussman then goes on to model out how jobs data are transmitted to the broader economy via an Impulse Response model.
For employment, a 1% shock in job creation or destruction (versus trend growth) tends to be followed over the following year by an additional 1% movement in jobs in the same direction.
He says the normal impulse response to the large job destruction of 2008 and 2009 would be further job losses for a time. But we are well past the propagation point for those job losses. Rather, in a normal recovery, 460,000 to 500,000 jobs would be created. So, we are well short of this. Why? Clearly this has something to do with consumer deleveraging and/or industrial recalculation as labour shifts away from depressed sectors of the economy awaiting the pick up in new growth sectors.
So what is the most likely outcome of this situation? In my view, the next three months represent the most serious window for the U.S. economy and labor market. The typical 23-26 week lag between leading indicator deterioration and new unemployment claims deterioration suggests that we may observe upward pressure on new claims for unemployment beginning about mid-October. As I noted last week, however, these lags can be somewhat variable, and the leading indicators tend to have a better correlation with price fluctuations in the securities market. By the time the coincident economic evidence is clear, securities markets have often completed a large portion of their adjustment.
On the positive side, while the next few months may provoke some further job losses, I suspect that the U.S. economy is already running fairly "lean" from an employment standpoint. The better leading indicators of economic activity do suggest a further round of cuts, but there is not nearly as much room for this as employers had going into the recent credit crisis. Employment losses have already been so profound that the typical impulse response actually creates something of a tailwind for the labor market, which hopefully will hold additional job losses to a tolerable level. That said, we shouldn’t rely on anything close to the typical impulse response, because generally speaking, past "tailwinds" for job growth after a recession have been heavily reliant on the fresh expansion in debt-financed, large-ticket spending. This includes fixed investment, autos, durable goods, and residential investment.
Overall then, we are facing the likelihood a fresh near-term deterioration in U.S. economic activity, as part of a longer multi-year adjustment, which is typical post-credit crisis behavior. My impression is that Wall Street is eager to treat the present cycle as a "V-shaped" recovery. We see little evidence to support that view, and the best evidence we do observe is more consistent with a double-dip (if not a continuation of a single ongoing recession).
This is not a V-shaped recovery. And Hussman says we need to watch the lagging indicators to ascertain what sort of impulse response the fresh batch of employment weakness will have on the real economy. The ISM numbers should be of particular note not just here in the US but in Asia where there has been softness.
However, as I indicated in the links this morning, there are other more bullish scenarios based on an end to consumer deleveraging. As Marc Chandler has suggested, the retail sales number out tomorrow will be a good test of this view, which is driving the risk-on trade. For his part, Hussman believes the greatest downside risk is more in terms of stocks than the economy because profit margins are rich. Since margins mean-revert, his conclusion is that the risk/reward at present valuation is poor. Much more at the link below.
Impulse Response – John Hussman