Aggregate Hours and Local Projections with Long-Run Restrictions

Abstract

I extend the local projections method to identify structural shocks using long-run restrictions. I show that the proposed estimator is substantially more robust than structural VARs (SVARs) to both the choice of lag length and the order of integration of the endogenous variables using data from a standard real business cycle (RBC) model. Benchmark simulations show that the proposed estimator can yield substantial reductions in both the bias and mean squared error of estimated impulse response functions relative to SVARs, particularly at short forecast horizons. In all cases, the proposed estimator correctly estimates the direction of the contemporaneous response and the shape of the full impulse response function, and can eliminate virtually all of the bias for some specification choices. Using my proposed estimator and data obtained from the Bureau of Labor Statistics, I then estimate the response of aggregate labor hours to productivity shocks. In contrast to much of the evidence based on SVARs, I find that labor hours rise in response to positive productivity shocks and subsequently follow a hump shaped profile. This result is robust to a number of specification choices and provides new evidence in support of the standard RBC model.