Improving output gap estimation— a bottom-up approach

Abstract

We propose a multivariate Bayesian state space model to identify potential growth and the output gap consistent with the dynamics of the underlying production sectors of the economy and those of inflation and the labor market. Our approach allows us to decompose economic fluctuations and long-term trend growth of output and employment into its driving factors. Applying our model to the Swiss economy reveals substantial divergence among the considered production sectors – their contributions to gap and potential vary both in size and direction. Potential growth has been declining over the past two decades and the data points to labor market frictions and a well-identified Phillips curve. In a comprehensive real-time study, we review revision and forecasting properties of our estimate and compare it to established methods. Overall, we document several advantages of our sector gap model: a) It facilitates the interpretability of economic trends and cycles, allowing for more efficient policy actions, b) it has favorable revision properties compared to standard univariate filtering techniques and a baseline model without sectors, c) it is useful in forecasting output growth and inflation, and d) it produces economically meaningful potential growth rates.