Economics is called the “dismal science” because of its miserable economic forecasting record, especially when it comes to predicting recessions.
Last fall, hardly any economists predicted we would have a recession this year. To be fair, their rosy economic outlook for this year was because the Federal Reserve cut rates three times last year and the economy and financial markets were doing very well.
After decades of teaching monetary economics and seeing flawed recession forecasts, I realized the need for alternative forecasting techniques. Rather than using traditional indicators such as inverted yield curves or complicated algorithms and econometric models, I studied over 100 years of business cycle data to identify patterns in recession start dates, not unlike technical stock market analysis, or “charting.”
Identifying a consistent pattern over time to form predictions is known as a “calendar anomaly,” which is defined as a result “inconsistent with the present economics paradigm.” Calendar anomalies in finance are most well known in the stock market — for example, the “January effect” when stocks do best or the “September effect” when they do worst.
Using National Bureau of Economic Research data on the start date of every U.S. recession going back to 1854, I found there was an unusually high likelihood, nearly double of what would be expected, of one starting in years ending in 9 or 0. This probability increased considerably if there was also a presidential election, where there is always some uncertainty.
I coined the term “Turn of the Decade,” or TOD, effect in November 1990 for this calendar anomaly. It not only correctly predicted the 1990-1991 recession, but also the “Y2K Recession” of 2000-2001, although there was some controversy among many economists as to its precise starting date of the last quarter of 2000 or the first quarter of 2001.
The TOD effect did not predict the 2007-2009 Great Recession, but last November it resulted in an 80 percent prediction of a 2020 recession — the highest and most specific recession prediction at that time.
To be clear, this was not a prediction of a pandemic, major demand or supply side shocks, or significant stock market correction. Rather, it was a prediction with a high (80 percent) degree of certainty that there would be recession in 2020, but there was uncertainty as to what would be the cause.
It could be traditional factors such as inverted yield curves, trade wars, a global recession, uncertainty over the November presidential election, or some unknown “X” factor … that just happened to be the coronavirus.
A track record of correctly predicting three of the last four recessions, especially the current one, suggests the value of the TOD effect. But that is history, and what is of interest now is where the economy is going.
The coronavirus recession has caused major demand and supply shocks with serious economic consequences. Despite the record fiscal and monetary policy response to mitigate the economic impact of this recession, there is considerable uncertainty as to its depth and length, since it was caused primarily by a pandemic for which there is no vaccine or timeline.
The best insight on this recession’s likely start and end date, depth and shape is found in looking at 166 years of past business-cycle data covering all 33 U.S. recessions. The most infamous example of a pandemic — actually, the worst ever — was the 1918-1920 Spanish flu. It infected about one-third of the world’s population, resulting in some 50 million worldwide deaths, including 675,000 in the U.S.
That pandemic was followed by the 1920 recession that was caused by mainly economic factors. It was a severe recession, running 18 months from January 1920 to July 1921, and it included the November 1920 presidential election.
Fast-forward 100 years to today. The March 2020 start date of the current recession ends our record 128-month economic expansion since the last Great Recession, beating the previous 120-month expansion of the 1990s. That recession lasted 18 months, the average length of all recessions since 1854, and it exceeded the average postwar recession of 11 months.
With all the uncertainty around the economic and human impact of the COVID-19 virus, including a vaccine, this recession likely will last nine months through the remainder of this year. This best-case scenario will end with a happy, healthy and busy holiday season, with a reversal of the current “consumer distancing” from the retail and service sectors, which account for about 70 percent of our economy.
The worst-case scenario, with a steep “coronavirus curve,” delayed vaccine, and continued consumer distancing from restaurants, theaters, airlines, cruises, hotels, concerts, etc. would result in a deeper, 15-month recession into next spring. As was the case with the 1920-1921 recession, this likely will be “V”-shaped with the huge plunge in wholesale/retail sales, industrial production, real income and especially employment (the four key variables used by the NBER to determine a recession) occurring in the next three months.
The huge fiscal and monetary stimuli will have a lagged impact on slowing this free fall by the summer, when things should improve — hopefully with a flattened coronavirus curve from the current mitigation measures. The traditional V-shaped recession is very steep, going down for these three months and then gradually improving for the rest of the year.
When the economy begins its expansion next year — after either a best-case, nine-month or worst-case, 15-month recession — it hopefully will be a long-lived one, at least until 2029 and 2030, our next turn-of-the decade.
Kenneth H. Thomas, Ph.D., is president of Miami-based Community Development Fund Advisors, LLC. He taught finance at the University of Pennsylvania’s Wharton School for over 40 years.