The arrow of progress in economic analysis usually rolls forward with time. Yesteryear’s insights give way to new wisdom as dismal scientists discover data sets, develop analytical tools, and devise new models. But sometimes the process works in reverse. To wit, a new paper makes use of a nearly century-old statistical tool that was originally designed to measure similarities in human skulls.

What’s useful for cranium analysis appears expedient for tracking the economy’s ebb and flow, advise the authors of “A New Index of the Business Cycle,” written by researchers at MIT and State Street. The underlying concept is a bit dusty – the Mahalanobis distance, the metric of choice, first saw the light of day in 1927. But as the authors demonstrate for this Jazz Age relic, there’s macro-econometric gold here for nowcasting expansions and recessions.

Surprisingly, the Mahalanobis distance does a better job of identifying turning points in the economy in real-time vs. the Conference Board’s leading indicator – a benchmark explicitly designed (and revised over the years) to anticipate recessions and expansions.

“We argue that, unlike the Conference Board Index of Coincident Indicators, our index gives an independent assessment of the state of the economy because it is constructed from variables that are different than those used by the NBER to identify recessions”, write William Kinlaw (a senior managing director at State Street Associates) and two co-authors.

We argue that our index is more objective than the NBER’s identification of recessions because our index is strictly data-driven and therefore, free of bias and persuasion. We argue that our index gives a more objective assessment of the business cycle than the Conference Board Indexes because it is expressed in units of statistical likelihood. And we argue that our index is more informative than the Conference Board indexes because it explicitly captures information about the co-movement of economic variables.

The paper’s KKT Index (named for the first letter of the authors’ surnames) is also a model of efficiency, using just four data sets to analyze the US economy:

• Industrial production (1-year percentage change)
• Nonfarm payrolls (1-year percentage change)
• Stock market return (1-year performance)
• Slope of the yield curve (10-year Treasury rate less the Federal Funds Rate)

Limiting business cycle analytics to this quartet is debatable. For example, leaving out consumer spending – accounting for roughly two-thirds of the U.S. economy – sidesteps a critical macroeconomic factor.

Then again, KKT’s backtest certainly looks encouraging. How might the results fare with a broader, more representative set of macro factors? Unclear, although it’s a worthy topic for further study.

Meantime, KKT in its current form appears to have an edge vs. the Conference Board Index of Leading Indicators. “Exhibit 3 indicates that the Conference Board’s Index of Leading Indicators tends to coincide with recessions rather than anticipate them,” the paper observes. “The KKT Index rises leading up to every recession so that the combination of its trajectory and level provides a reliable indicator of the likelihood recession.”

(Click on image to enlarge)

Notably, KKT was predicting a 76% probability of a recession in the next six months (as of November 2019). By contrast, the leading indictor’s estimate of recession risk was much lower.

“An index level of 76% does not necessarily mean that the economy is currently in recession. Rather, we should interpret it as an indication of the potential for the economy to enter recession in the foreseeable future,” the authors explain. “Given historical guidance, the index should be close to 100% when a recession is imminent or underway.”

The question, of course, is whether KKT is, in fact, a genuine improvement over nowcasting recession risk relative to the Conference Board’s leading index? It appears to be, but that’s based on the rear-view mirror. The acid test is how the analytics fare in the out-of-sample tests. On that score, we’ll have results by May—the sixth month of the paper’s elevated recession estimate period.

If the economy isn’t in recession by May (or earlier), would that be a sign of failure for KKT? Not necessarily. A 76% recession risk is high, but it’s still below 100%. On the other hand, if the economy is still expanding through May, the news will give the Conference Board some new bragging rights for its Index of Leading Indicators.