Nowcasting Tail Risks to Economic Activity with Many Indicators

Abstract

This paper focuses on tail risk nowcasts of economic activity, measured by GDP growth, with a potentially wide array of monthly and weekly information. We consider different models (Bayesian mixed frequency regressions with stochastic volatility, classical and Bayesian quantile regressions, quantile MIDAS regressions) and also different methods for data reduction (either the combination of forecasts from smaller models or forecasts from models that incorporate data reduction). The results show that classical and MIDAS quantile regressions perform very well in-sample but not out-of-sample, where the Bayesian mixed frequency and quantile regressions are generally clearly superior. Such a ranking of methods appears to be driven by substantial variability over time in the recursively estimated parameters in classical quantile regressions, while the use of priors in the Bayesian approaches reduces sampling variability and its effects on forecast accuracy. From an economic point of view, we find that the weekly information flow is quite useful in improving tail nowcasts of economic activity, with initial claims for unemployment insurance, stock prices, a term spread, a credit spread, and the Chicago Fed’s index of financial conditions emerging as particularly relevant indicators. Additional weekly indicators of economic activity do not improve historical forecast accuracy but do not harm it much, either.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 29 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.