Forecasting U.S. Recessions by Using the Probit Mode

Xinyu Wang & David Constantin Trebbin

Student thesis: Master thesis

Abstract

The topic of a possible forthcoming recession was brought up in 2019and gained even more attention in early 2020, when the COVID-19 coronavirus outbroke to the world. Due to the massive impact a recession can cause to the economy, it is of great interest for many parties to forecast a coming recession.
The aim of this thesis is to examine the predictive power of different variables and to use them to forecast the probability of a future U.S. recession in three different forecasting horizons. In total, 20 variables were studied including stock market indicators, macroeconomic variables, interest rates and spreads. These variables were used in a probability forecasting model called the probit model to generate forecasted recession probabilities, which were then evaluated by using different performance measures.
Our results showed that the macroeconomic variables perform best in short horizons, and interest rates and spreads perform best in long horizons when using them as a single predictor variable or in combination with another variable. Although the stock market indicators did not provide strong predictive power when used alone, they were able to improve the forecasting performance of other variables when combined.
Lastly, we selected the best performing forecasting models, which were the combination of the composite leading indicators and change in unemployment rate, the combination of the composite leading indicators and the term spread between the ten-year treasury rate and the federal funds rate, and the term spread between the ten-year and the three-month treasury rate, for a forecasting period of three months, six months and twelve months respectively. These three models were applied for predicting the U.S. recession probabilities from April 2020 to March 2021, and their forecasts all indicated that there is a high chance of a recession in the time between June to August 2020.

EducationsMSc in Finance and Investments, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2020
Number of pages140