A0190
Title: On the real time predictive content of financial conditions for growth
Authors: Michael McCracken - Federal Reserve Bank of St. Louis (United States) [presenting]
Aaron Amburgey - Federal Reserve Bank of St. Louis (United States)
Abstract: Analytical, Monte Carlo, and empirical evidence is provided on the real-time predictive content of financial conditions indices, notably the NFCI, for quantiles of the distribution of U.S. real GDP growth. We do so by investigating two specific issues in the vulnerable growth literature. First, we construct (unofficial) real-time vintages of the NFCI. This allows us to conduct the out-of-sample analysis without introducing look-ahead biases that are naturally introduced when using a single current vintage. We then investigate the usefulness of asymptotic and bootstrap-based critical values for tests of predictive ability for nested models in the context of linear quantile regression. We find that for quantiles near the median, the asymptotic critical values can provide accurately sized tests in reasonable sample sizes. As the quantiles shift into the tails, the asymptotic critical values perform quite poorly even in large samples. Both the fixed regressor wild bootstrap and the tapered block bootstrap provide accurately sized tests across all quantiles even in modest samples.
