Working Paper: NBER ID: w16725
Authors: Jon Faust; Simon Gilchrist; Jonathan H. Wright; Egon Zakrajsek
Abstract: Employing a large number of real and financial indicators, we use Bayesian Model Averaging (BMA) to forecast real-time measures of economic activity. Importantly, the predictor set includes option-adjusted credit spread indexes based on bond portfolios sorted by maturity and credit risk as measured by the issuer's "distance-to-default." The portfolios are constructed directly from the secondary market prices of outstanding senior unsecured bonds issued by a large number of U.S. corporations. Our results indicate that relative to an autoregressive benchmark, BMA yields consistent improvements in the prediction of the growth rates of real GDP, business fixed investment, industrial production, and employment, as well as of the changes in the unemployment rate, at horizons from the current quarter (i.e., "nowcasting") out to four quarters hence. The gains in forecast accuracy are statistically significant and economically important and owe exclusively to the inclusion of our portfolio credit spreads in the set of predictors--BMA consistently assigns a high posterior weight to models that include these financial indicators.
Keywords: Credit Spreads; Economic Activity; Bayesian Model Averaging; Forecasting
JEL Codes: C11; C53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| option-adjusted credit spreads (G12) | growth rates of real GDP (O49) |
| option-adjusted credit spreads (G12) | business fixed investment (G31) |
| option-adjusted credit spreads (G12) | industrial production (L69) |
| option-adjusted credit spreads (G12) | employment changes (J63) |
| inclusion of credit spreads (G19) | predictive accuracy of economic activity measures (E37) |
| BMA forecasts (E37) | reductions in RMSFE (C51) |
| credit spreads (G12) | early indicators of recessions (E32) |