Slides for the 5th Annual Macroprudential Conference: Macroprudential, June 2019 in Eltville [PDF]
We evaluate whether financial conditions can be used to predict risks of recessions and study their relation with real economic conditions. We focus on two financial condition indicators for the US: (i) the Chicago Fed’s NFCI that captures shifts in financial spreads and may provide short-term signals on the sudden activations of financial frictions; (ii) nonfinancial leverage indicators that could provide mediumterm information on the building-up of financial instabilities. We appraise their information content in two classes of models. First, we consider the quantile regression framework of Adrian et al. (2019) to assess near-horizon forecastability of recessions. Subsequently, in the spirit of the Basel Committee for Banking Supervision (2010), we propose a novel semi-structural time series model able to identify excess leverage in credit variables, as deviations of the credit growth from the output trend growth that are not due to the business cycle. Our results point to three stylised facts. First, the NFCI correlates with the left tail of the conditional distribution of GDP growth, but provides limited advanced information on growth vulnerability. Hence it is best thought of as providing coincident information on the evolution of recession episodes. Second, nonfinancial leverage exhibits fewer significant non-linearities in the conditional distribution but provides a leading signal at least for the 2008 recession. Finally, our stylised model provides a coherent narrative of excess credit growth in the US economy and identifies a timely indicator of “excess leverage” that can be used for macro-prudential purposes.