Financial Variables and Macroeconomic Forecast Errors 
Following the Great Recession of 2007-09, much research focused on the measurement and predictive power for macroeconomic activity of financial variables meant to capture different aspects of the macro-financial landscape. Most of this work has centered on evaluating the importance of financial variables (primarily asset prices) in forecasting real economic activity.
The authors of this paper set out to assess the connection between finance and the macroeconomy by examining how a battery of financial variables fares in predicting macroeconomic forecast errors. The developed evidence sheds light on what types of financial linkages have been “missing” in many macro forecasting models to date. Such evidence could help inform policymakers’ and economists’ agendas for developing financial indicators and macroeconomic models that are better attuned to relevant financial developments.
Key Findings
- A large set of financial variables has only limited power to predict a latent factor common to the year-ahead forecast errors for real Gross Domestic Product (GDP) growth, the unemployment rate, and the Consumer Price Index (CPI) inflation for three sets of professional forecasters: the Federal Reserve’s Greenbook, the Survey of Professional Forecasters (SPF), and the Blue Chip Consensus Forecasts.
- Even when a financial variable appears robust across sample periods in explaining the latent factor, from an economic standpoint its contribution appears modest.
- These findings, however, speak more to limitations of some econometric procedures than to the financial variables’ modest predictive content for the latent factor.
- Several financial variables retain economic significance over certain subsamples; when non-linear effects are accounted for, these variables have a more important role in consistently predicting the latent factor over the business cycle.
Implications
Empirical methods better suited to capturing the intermittent nature of the relationship between financial variables and future economic activity can generate more positive findings. The threshold effects explored in the authors’ analysis suggest a more important role for several of the considered financial variables and reinforce previous results in the literature. Such threshold effects are especially relevant because they imply that financial variables become more important at explaining forecast errors when the economy is deteriorating. In other words, financial variables help predict economic activity at times when a forecast error may be particularly costly from a policy standpoint.
Abstract
A large set of financial variables has only limited power to predict a latent factor common to the year-ahead forecast errors for real Gross Domestic Product (GDP) growth, the unemployment rate, and Consumer Price Index (CPI) inflation for three sets of professional forecasters: the Federal Reserve’s Greenbook, the Survey of Professional Forecasters (SPF), and the Blue Chip Consensus Forecasts. Even when a financial variable appears to be fairly robust across sample periods in explaining the latent factor, from an economic standpoint its contribution appears modest. Still, several financial variables retain economic significance over certain subsamples; when non-linear effects are accounted for, these variables have an improved ability to consistently predict the latent factor over the business cycle.