Economists are citing higher recession risks, with the latest reported April poll conducted by The Wall Street Journal putting the probability of the economy being in recession sometime in the next 12 months at 28%, up from 18% in January and just 13% a year ago. However, as of late May few forecasters had “pulled the trigger” and adopted a recession as their base case. Since historically economists have not been very good at forecasting recessions, might they be wrong now? And what might motivate them to initiate recession forecasts?
Economists are poor forecasters. Consider the track record of consensus forecasts compiled by the Federal Reserve Bank of Philadelphia in its Survey of Professional Forecasters (SPF) over the seven recessions before the sudden and unexpected 2020 pandemic recession. In only three of those seven previous downturns did the professional forecaster consensus correctly predict the outcome. Moreover, in all those seven recessions the quarter of deepest real GDP growth decline was always worse than predicted by the consensus. Even more worrisome, in none of those seven downturns did the consensus correctly anticipate the first negative real GDP growth quarter by more than two quarters in advance.
A variety of reasons explain professional forecasters’ struggles with prognosticating recessions. For one, many of the surveyed economists who work at investment banks operate in organizations with protocols requiring coordination across research teams, including investment strategists. The aim is to present credible, unified, and non-conflicting views to clients. But that may come at a cost, including less ability to ‘stick one’s neck out’ or even a reluctance to deviate much from consensus forecasts of competing firms.
In addition, some forecasters at securities firms exhibit asymmetric loss functions. Senior executives at these firms and their clients generally make more money in bull than bear markets. Their economists may perceive greater reputational and job risk in incorrectly forecasting a recession and an ensuing bear market than in not correctly forecasting a recession at all. There is an added element of forecaster angst at securities firms—judging whether their views are already discounted in forward-looking stock and bond prices. If that is the case, then changing to an ultimately mistaken recession forecast may be perceived as an especially costly blow to one’s professional reputation.
In a fast-changing environment with as many mixed signals as in the current situation, the above considerations can hinder a forecaster’s ability to make timely and successful recession forecasts.
Aside from institutional and psychological constraints, using historical experience to project a downturn is inherently difficult. To be sure, pre-recession environments have some similarities (e.g., excess investment or financial leverage, or inflation and more restrictive financial market conditions). But in current circumstances some worrisome factors might be temporary.
One example is COVID19-related disruption to global supply chains. Such dislocations have curbed supply and boosted prices. Whether they linger depends in part on the severity of further waves of infection, challenges economists are ill-equipped to address.
Similarly, surging energy prices could reverse if there is a de-escalation of the Russia/Ukraine war—another hard-to-predict event risk. And even if oil prices do not materially recede anytime soon, economists can disagree among themselves about how much damage to the US is inflicted by higher oil prices. This is because the US is much less dependent on oil imports than in the past. For an energy surplus country like the US, rising oil prices redistribute income rather than reduce it, with more difficult-to-evaluate impacts on total spending. Most probably, however, producers flush with cash will be less likely to spend enough to compensate for the fall in purchasing power among consumers, who have few options to save on driving, or on heating or cooling their homes.
In this challenging forecasting environment, what could motivate a more widespread adoption of recession forecasts?
Weekly initial claims for state unemployment insurance are important. Job losses are consistent with slowdowns and recessions. However, given tight labor markets and a ‘war for talent’, employers may be hesitant to lay off workers. Hence, alongside jobless claims, forecasters should focus on consumer spending and confidence data, as well as on labor demand signals such as The Conference Board®−Burning Glass® Help Wanted OnLine® (HWOL) Index.
Economic forecasting is about more than drawing inference from the past. Macro models must be updated for changes in the economy. Qualitative judgment is essential as a partner to quantitative analysis. And so too is courage—the willingness to step away from the crowd when circumstances warrant.
1. For details, please see Inside The Crystal Ball—How To Make And Use Forecasts published in 2015 by Maury Harris.
2. See “Rational Bias in Macroeconomic Forecasts” by Bennett, Geoum and Laster in a July 1996 Federal Reserve Bank of New York Research Paper.