The Limits Of One-Indicator Recession Analysis

The search for a silver bullet in business-cycle analysis is a hardy perennial. But it’s a search that’s almost certainly destined for failure.

No single indicator is infallible in the quest to identify a new recession. Some pundits suggest otherwise, but history isn’t kind on this front. A recent example: US industrial production. Several observers of the macro scene were quick to point out last year that the extended run of negative year-over-year changes in output was a sure sign that another recession was fate. The reasoning was that red ink for production in annual terms for more than a few months was always linked to an NBER-defined downturn – based on the historical record going back to the 1920s.

By that logic, the economic outlook looked quite grim late last year. The annual decline in industrial production in Nov. 2016 marked the 20th straight month of decline. By some accounts, nearly a century of economic history suggested that the jig was up. Every case of slumping industrial activity for that length of time was linked with a recession.

But as we now know, industrial output has rebounded in recent months and the odds that a recession has started is virtually nil at the moment, based on data published through Apr. 2017. Tomorrow’s another day, but the available figures in hand– across a broad spectrum of indicators – betrays no sign of recession risk.

The failure of industrial production to signal a new recession may be surprising, but relying on one indicator to provide a robust measure of the business cycle is an accident waiting to happen. But let’s not limit this caveat to industrial activity. The same can be said of other indicators that are lauded as dependable benchmarks of the trend.

Indeed, some analysts are smitten with the ISM Manufacturing Index, convincing at least one investment strategist to proclaim that this metric is all you need to accurately gauge the state of the US economy. But here, too, recent history has raised doubts about the dependability of this data when used in isolation.

Others say that Mr. Market is the solution. Recall, for instance, that the year-over-year slide in the stock market in late-2015 and early 2016 was widely seen as a harbinger of a new recession. But the market’s decline turned out to be a false alarm.

Another seemingly reliable measure of business cycle risk – real M0 money supply – has stumbled recently too. So-called high-powered money has been contracting in year-over-year terms for over a year – a dive that’s usually associated with recession. But so far, the economy hasn’t cried uncle.

In fact, it’s always different when it comes to evaluating the real-time risk of recession. Every downturn is different in terms of catalysts. As such, it’s hopeless to think that one or even two indicators will suffice for recognizing early on that the economy is slipping over to the dark side. History is quite clear on this point. The mix of triggers that create downturns keeps changing, which means that the signals that provided early warnings in the past may not be useful in the future.

Don’t misunderstand. Looking at the indicators above — and many more — is essential. The mistake is thinking that we can cherry-pick the numbers.

What accounts for the popularity of trying to synthesize recession analysis down to one data set? Simplicity. Looking to one metric has obvious appeal since it relieves us of the hard work of analyzing a broad set of indicators.

It’s a pleasant thought, but the challenges of developing timely and reliable recession-risk metrics requires more than simply gazing at the stock market or checking in with the monthly updates on industrial production.

If there’s one exception to these caveats surely it’s the Treasury yield curve. Much has been written of the power of comparing the yield spread between, say, the 10-year Note and the 3-month T-bill. By some accounts, this measure has a perfect record of predicting new recessions. Perhaps, but should we assume that the yield curve will be as dependable in the future?

The extraordinary state of monetary policy at the moment raises doubt. Combine that with the fact that the Federal Reserve has learned or thing or two in recent decades about its widely documented role in playing a key role in creating conditions that are ripe for economic slumps. As such, it’s reasonable to wonder if the Fed will be successful in gently squeezing monetary policy without triggering a recession. If so, the yield curve’s stellar history as a business cycle indicator may stumble.

The lesson, of course, is to look beyond a handful of indicators to evaluate recession risk. The gold standard is the three-month moving average of the Chicago Fed National Activity Index, a multi-factor benchmark that dispensed relatively early real-time warnings for the last two recessions.

But nothing’s perfect. The Chicago Fed index is published with a lag on a monthly basis. The challenge, then, is to develop analytics to anticipate this benchmark’s signals in advance. One effort that’s provided encouraging results in recent years is The Capital Spectator’s monthly reports on the business cycle – here’s last month’s update. A more ambitious project is offered in The US Business Cycle Risk Report, which is published weekly (or more frequently during volatile economic conditions).

There’s still no guarantee, of course. The next recession could surprise even the most sophisticated models. But the probability of getting blindsided can be substantially reduced by looking beyond one or two indicators.

Some critics might ask: What’s the relevance from an investment perspective? By the time it’s clear that a recession has started is it too late to adjust portfolios? Not necessarily, assuming that you’re using a robust recession-monitoring model. As I noted recently, most of the S&P 500’s decline in the last two recessions followed real-time recession warnings from the three-month average of the Chicago Fed National Activity Index.

It could be different in the future, of course. But the vintage history of the Chicago Fed data (and related efforts) looks encouraging. That’s hardly an iron-clad guarantee, but it’s a dramatic improvement over what you can expect from analyzing recession risk with one or two indicators.