Modeling Professional Recession Forecasts

Abstract

recently. Many observers have noted warning signs of a recession, such as an inverted yield curve and volatile oil prices, while other indicators give hope for a soft landing. Although statistical models and professional forecasts imply a much-higher-than-normal probability of recession, there is still a lot of uncertainty about the outcome. Professional forecasters synthesize many types of information to predict future economic quantities, such as GDP or inflation, and they often assess the uncertainty associated with their predictions. In the case of recession prediction, for example, forecasters estimate the likelihood of a recession within the next 12 months. As of August 2023, a typical estimate of the probability of a US recession within 12 months is about 60%, and it has been above 50% since October 2022. While economic forecasters don’t have crystal balls and can only assign probabilities to various outcomes, their forecasts contain enough information for major corporations and organizations to purchase them. This raises questions: What variables do forecasters look at? How do these variables translate into forecasts? One could investigate forecast methods, but forecasters are naturally reluctant to publicly detail their methods. In addition, professional forecasters often supplement many sorts of statistical methods with their own judgment to account for special circumstances, such as unusual weather or changes to tax laws or financial regulations. These facts make it difficult to define forecasting techniques. An alternative to directly studying forecast methods would be to replicate forecasts using public information. In other words, instead of predicting output growth, inflation, probability of recession, or some other variable, one might try to replicate the past forecasts of the variable using public data. This essay investigates what variables professional forecasters appear to use to predict the probability of recession in the next 12 months. We obtain our target variable—the estimated probability of a US recession within 12 months— from Consensus Economics, which surveys economic forecasters and reports the results of their surveys as historical data series. The series for the estimated recession probability Modeling Professional Recession Forecasts