Professional survey forecasts and expectations in DSGE models

This paper asks whether Survey of Professional Forecasters (SPF) data can be efficiently integrated into medium-scale DSGE models, and whether models with imperfectly rational expectations based on Adaptive Learning (AL) outperform the standard Rational Expectations (RE) hypothesis when survey forecasts are used as observables. The authors work with quarterly US data spanning 1981q2–2019q2, using the Philadelphia Fed Real-Time Data Set (first and second releases) alongside SPF nowcasts for inflation, consumption, investment, and output growth. The SPF nowcast is defined as a prediction formed in the middle of period t+1 for period t+1 given information for period t, making it a suitable proxy for the model-based expectation E_t y_{t+1}.

Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach

The paper estimates an extended New Neoclassical Synthesis model on US data covering 1966:1-2004:4 using Bayesian likelihood methods, with seven observable series – real GDP, hours worked, consumption, investment, real wages, prices and the short-term nominal interest rate – and seven orthogonal structural shocks: total factor productivity, risk premium, investment-specific technology, wage mark-up, price mark-up, exogenous spending and monetary policy. The model carries sticky Calvo price and wage setting with backward indexation, habit formation in consumption, investment adjustment costs, variable capital utilisation and fixed costs in production. Three questions organise the paper. First, does the model describe the data? Compared against unrestricted VARs and a Sims-Zha Bayesian VAR over the full sample (with 1956:1-1965:4 as a training sample to standardise priors), the tightly parameterised model beats the unrestricted VARs decisively and is matched by the best BVAR(4) on marginal likelihood; in a rolling out-of-sample RMSE exercise over 1990:1-2004:4 the two are comparable one quarter ahead but the structural model “does considerably better than both the VAR(1) and BVAR(4) model” over horizons up to three years, with the improvement “quite uniform across the seven macro variables.” Second, which frictions earn their place? Cutting the Calvo probability for prices or for wages to 0.10 each costs about 50 in log marginal likelihood; removing investment adjustment costs costs about 160; reducing habit formation is costly but much less so; shutting off variable capital utilisation “comes at no cost”; and restricting price indexation to 0.01 actually improves the marginal likelihood. Third, what drives the US business cycle? Within a year, output is dominated by the exogenous spending, risk premium and investment-specific shocks, which together account for more than 50 percent of forecast error variance; beyond two years the productivity and wage mark-up shocks account for more than half, with the wage mark-up dominant in the long run, while monetary policy shocks “contribute only a small fraction of the forecast variance of output at all horizons.” Inflation is driven by price mark-ups in the short run and wage mark-ups in the medium to long run, which the authors attribute to a very small estimated slope of the New Keynesian Phillips curve and to an aggressive estimated policy response. A positive productivity shock reduces hours worked immediately and significantly, turning positive only after two years, and does so even under flexible prices and wages, which the authors trace to habit persistence and capital adjustment costs. Finally, sub-sample estimates for the “Great Inflation” (1966:2-1979:2) and the “Great Moderation” (1984:1-2004:4) show most structural parameters stable, the standard deviations of productivity, monetary policy and price mark-up shocks lower in the second period, the policy response to the output gap level halved and no longer significant, and a counterfactual exercise attributing the fall in volatility mainly to milder shocks rather than to policy or structural change.

Rafael Wouters | Macro Paper Warehouse

Professional survey forecasts and expectations in DSGE models

Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach