Can economists help investors avoid recessions?
How often they accurately predict a recession, and how often they’re late to the party.
Recessions hurt stocks
Since the Kennedy administration, there have been eight recessions in the United States, as determined by the National Bureau of Economic Research. Five times, equities lost money after including the effect of inflation. Worse, none of the gains were large, but two of the declines were calamitous.
Figure 1: Stocks during recessions. Sources: National Bureau of Economic Research, Morningstar Direct
The average recession performance, shown in the final bar to the right, was an 8.8% real loss. (I wish I could make that bar a different color, since it summarizes the data, but to accomplish that feat, I must await a software upgrade.) It therefore behooves investors to avoid economic downturns.
But how? Stock prices might provide clues, by plummeting before the fact. However, most stock market indicators are specious. Over that same period, the era since President John F. Kennedy was elected, the S&P 500 has suffered 31 double-digit losses. Investors who treated market corrections as a sell signal would have been spun right round.
The survey says
Perhaps the solution to anticipating recessions, rather than awaiting their occurrence, lies instead with economists, who are trained for such tasks. Since 1968, the Federal Reserve Bank of Philadelphia has conducted a regular survey of professional forecasters. Among its questions is whether the nation’s real gross domestic product will decline during the next quarter. On 10 of those occasions, the forecasters assigned an average probability of more than 30%.
(Technically, this happened more than 10 times, but I ignored the instances when a high response followed a previously high response. After all, once recessions begin, economists tend to foresee future difficulties.
To cite an example, in August 1990, the economists assigned a probability of 17.9% for lower upcoming GDP for the next quarter. When they were asked the same question three months later, the figure was 41.5%. I therefore recorded November 1990 as the date of their recession prediction, bypassing their February, May, and August 1991 outlooks, which also surpassed the 30% mark.)
This is proper material for an analysis! First, the Philadelphia Federal Reserve graciously provides more than half a century of data. Thank you, Philadelphia. (Words that are not regularly spoken.) Second, unlike with stock market data, the economists issued few false alarms. Eight times, their recession forecasts corresponded with the actual events. Only in 1988 and 2022 did they cry wolf.
Problem number one: Behind the economy’s curve
Regrettably, “corresponded with” is not the same as “exactly matched.” The next chart shows the time elapsed between the forecast and the start of each recession. A positive bar indicates that the prediction preceded the event, while a negative bar means that the forecast came after the downturn had begun. (With this exercise, I ignored the erroneous signals of 1988 and 2022.)
Figure 2: National Bureau of Economic Research, Philadelphia Federal Reserve, author calculations (data as of August 11).
Hmmm. The forecasters have usually been late to the party. If the markets have been similarly dilatory, no problem, as investors will still have been able to exit their shares before prices declined. However, if equities tend to slump when recessions begin—or worse yet, before the economic slide commences—then the predictions will have arrived after their expiration dates.
Problem number two: Behind the stock market’s curve
We can test this proposition by measuring how stocks performed in the 12 months before the economists’ forecasts. Once again, all returns are presented in real terms.
Oh, dear. That is terrible. By the time that the economists rang their bells, stocks had suffered an average after-inflation return of negative 7.3% over the preceding year, which nearly matched their full recession losses. Moreover, the forecasters were lucky rather than astute during their best showing, in 2020. They took so long to sound the recession alarm that when they did, stocks had already rallied. They were rewarded for sleeping on the job!
In most cases, stock prices beat the forecasters to the bad news, meaning that investors who used such predictions to change their asset allocations were … a day late and a dollar short.
John Rekenthaler, Vice President of Research
Summary
It would be useful if 2020′s experience was typical, with stocks recovering when the recession forecasts were issued. Were that so, investors could treat economic forecasts as a contrarian signal. But the pattern is inconsistent. Sometimes, the economists have indeed foreshadowed a stock market rally. At other times, though, the recessions persisted long after the economists first recognized them, leading to further equity declines.
In short, predictions from economists about potential GDP declines would not have helped equity shareholders, had they integrated such information into their decisions. In most cases, stock prices beat the forecasters to the bad news, meaning that investors who used such predictions to change their asset allocations were, to use my father’s words, a day late and a dollar short.
Final note
To be sure, there are other ways to test the economists’ predictions. My choice of a 30% threshold was arbitrary. Unfortunately, adjusting the standards does not change the conclusion. Lowering the hurdle to, say, 20% leads to earlier recession predictions before the 1980 and 2007 downturns, which were beneficial, but doing so also creates several false alarms, each of which occurred during prolonged bull markets. Not helpful.
(Raising the forecast bar would have been worse yet. The economists were already late rather than early.)
Nor would changing the measurement approach improve results. Initially, I had thought about incorporating the magnitude of the forecasts, weighting a 70% recession probability more heavily than a 35% chance. But doing so would have done the economists no favors whatsoever. They were at their gloomiest in early 1991 and first-half 2009, both of which preceded both economic booms and powerful stock market rallies.
A final idea was to consider whether economists might be better at predicting recessions further into the future than one quarter. As evidenced by this article, however, they are not. Their accuracy drops sharply over time.