Industrial production fell 0.1% in January, the Federal Reserve reports, delivering a moderately worse-than-expected reading for the start of the year. Is that a sign of trouble for the business cycle in 2013? Possibly, but we'll need to see more dark news from this and other indicators before the broad picture issues a clear warning. For now, it's best to consider the latest data point for industrial production as a normal fluctuation within a slow-growth context.
In any case, there's no mistaking the recent downshift. For the last three months, industrial production has been slowing, dipping into negative territory in January. Ditto for the manufacturing component, as the chart below shows.
But the trend looks brighter on a year-over-year basis. Industrial production rose 2.1% last month vs. the year-earlier level. That's hardly ominous, if it holds. The bad news is that the annual rate of growth has been slipping recently, and a 2.1% increase is tied with last October's rise as the lowest in nearly three years.
If the annual rate continues to decline in the months ahead, we'll have a stronger signal for anticipating trouble in the broader economy. Meanwhile, it's still debatable if industrial production's January weakness is a smoking gun. Indeed, a broad set of numbers still point to growth. That's the message in the capital and commodity markets, as I noted earlier today. There's also a fair amount of forward momentum in key economic indicators as well.
Meanwhile, let's briefly note that the latest industrial production update conflicts with the upbeat news in the January ISM manufacturing and services reports. The same can be said vis-à-vis the market numbers of late (see link above). In addition, the signals from the labor market continue to lean positive. Private payrolls are still posting modest gains through January and initial claims for jobless benefits are still drifting lower, nearing a five-year low in yesterday's report.
None of this should be confused with a forecast, at least not formally speaking. What it does imply is that as of today, the statistical case for declaring that a recession has started is still thin. New data in the days and weeks ahead may tell us differently, of course, and we should be open to modifying our analysis when we have a convincing case via the numbers.
Today's industrial production report may be a sign of the tipping point, but it could just as easily be noise. If it is a harbinger of doom for the business cycle, the clues will start infecting other indicators, in which case we'll have a more persuasive case for worrying.
Why not simply assume the worst now and beat the crowd? Because you could be wrong, and history suggests you're likely to be wrong if you're inclined to call business cycle turning points before there's a strong case for doing so. Minds will differ on what constitutes a "strong case," although a fair reading of a wide spectrum of the key financial and economic indicators tends to drop robust clues in the aggregate -- emphasis on the aggregate.
Granted, the best we can hope for with "nowcasting" recession risk is receiving de facto confirmations that the cycle has turned in the recent past -- two to three months after the fact. That's a problem, but it's a lesser evil than regularly going out on a limb and projecting turning points before you have a convincing statistical case to back up your assumption. Why do I say that? Because history, including recent history, is littered with ill-timed warnings that a new recession has started.
Critics charge that a two- to three-month lag for calling start dates on recessions is too long. OK, but let's compare that with the alternative: accepting a series of incorrect predictions over the last several years that another recession has started. That's hardly a viable alternative, and on many accounts it's quite a bit worse than a careful methodology of nowcasting recession risk.
Although nowcasting suffers from a lag, it's less likely to whipsaw us with flawed warnings and a steady stream of hefty opportunity costs. Pick your poison. And, no, opting to call turning points in real time is hoping for the impossible. Some analysts get lucky every now and again, but don't confuse that with a reliable strategy for business cycle analysis through time.