What causes global recessions? A quantile regression approach

The aim of this thesis is to examine the determinants of global recessions. In particular, we want to understand which variables are actually informative to explain what will happen to extreme negative realizations of world industrial production growth, by employing a quantile regression approach, as proposed by Adrian, Boyarchenko and Giannone (2019). As potential predictive variables, we consider indicators suggested by the financial literature and economic theory, such as financial uncertainty, credit market frictions, U.S. monetary policy stance, oil and commodity prices. Consistently with an extensive literature, our main finding is that the most important determinant is financial uncertainty, especially with reference to the Great Recession. Moreover, at the one-year-ahead horizon, our proxy of credit market frictions turns out to be quite informative. The Real Commodity Price Factor measure displays a strong first moment effect regarding the Great Recession, at the one-quarter-ahead horizon. Oil prices seem to be informative with reference to the 2009 global recession at the one-year-ahead horizon, consistently with the 2007-2008 oil price spike documented by Hamilton (2009) as an important factor contributing to the early stages of the Great Recession. Interestingly, the proxy variables for oil shocks considered do not prove to be informative for predicting global economic downturns. Regarding the U.S. monetary policy and Term Spread, quantile regression does not seem to add information with respect to a linear model.