Is the Fed’s news perception different from the private sector’s?

ABSTRACT

The recent literature on monetary policy has dedicated considerable attention to modelling agents’ processing of information about the future in real time. This paper contributes to this growing strand by investigating the implied differences in the so-called news shocks estimated from the standard New Keynesian dynamic stochastic general equilibrium (DSGE) model using the real-time data sets from the Survey of Professional Forecasters (SPF) and the Federal Reserve’s Greenbook (GB) forecasts. Alternative specifications with either the SPF or GB forecasts aim to delineate the differences in the private sector’s and the Fed’s expectations of future macroeconomic outcomes and identify the differences in their perception of news shocks. Our results indicate that while the demand news shocks have very similar distributions in the two datasets, the monetary and cost-push news shocks from the models estimated on the GB data tend to be larger than those from the SPF. These findings suggest that the Federal Reserve’s forecasting methods allow for more variation in future outcomes than the SPF’s. These findings mesh well with the extant literature on the superiority of the Fed’s forecasts relative to the private sector’s and provide a structural explanation for the source of this superiority.

KEYWORDS:

• New keynesian model
• monetary policy
• anticipated shocks
• real-time forecasts
• greenbook forecasts
• SPF forecasts

JEL CLASSIFICATION:

• E31
• E32
• E52

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The views expressed in this article are solely those of the authors and do not reflect the position of the Federal Reserve Bank of Dallas or the Federal Reserve System.

² Brissimis and Magginas (2017) briefly discuss the role of alternative forecasts in a New Keynesian DSGE model similar to the one employed in this paper. However, the only focus on the role of difference in forecast estimates in the monetary policy reaction function, as opposed to how these differences affect all of the model’s endogenous variables.

³ See Orphanides (2001) for the seminal evaluation of the role of real-time forecasts in monetary policy rules. Best and Kapinos (2016) evaluate alternative modes of specifying forward-looking monetary policy rules and find that this functional form provides a good fit with the ex post data.

⁴ For details on the specification of the Metropolis-Hastings algorithm refer to Chib and Greenberg (1995).

⁵ The SPF forecasts are currently provided by the Philadelphia Fed and were previously collected by the American Statistical Association (ASA) and the National Bureau of Economic Research (NBER). The GB and RTDSM data are also available from the Philadelphia Fed website.

⁶ In the collection of the Real Time Data Set for Macroeconomics, the output variable changes in 1992 from GNP tp GDP. Therefore, we are using for our estimation the GDP growth rate before 1992 and the GDP growth rate thereafter.

⁷ See Milani and Rajbhandari (2014) for the details of merging the RTDSM and SPF datasets and related timing assumptions.

⁸ Similar discussion of the differences between the two datasets in terms of forecast errors is available upon request.

⁹ Since the ex post values of the federal funds rate and the 3-month T-bill rate are virtually identical, we only use the latter in these graphs for visual clarity.

¹⁰ We also estimated a model where the central bank is responding to 4 quarters ahead Greenbook forecasts of inflation and output growth $(k=4)$