Disasters Everywhere: The Costs of Business Cycles Reconsidered

Abstract

Business cycles are costlier and stabilization policies could be more beneficial than widely thought. This paper introduces a new test to show that all business cycles are asymmetric and resemble mini-disasters. By this we mean that growth is pervasively fat-tailed and non-Gaussian. Using long-run historical data, we show empirically that this is true for advanced economies since 1870. Focusing on peacetime eras, we develop a tractable local projection framework to estimate consumption growth paths for normal and financial-crisis recessions. Introducing random coefficient local projections (RCLP) we get an easy and transparent mapping from the estimates to a calibrated simulation model with disasters of variable severity. Simulations show that substantial welfare costs arise not just from the large rare disasters, but also from the smaller but more frequent mini-disasters in every cycle. On average, and in post-WW2 data, even with low risk aversion, households would sacrifice about 15 percent of consumption to avoid such cyclical fluctuations.