Testing for asymmetries in the preferences of the euro-area monetary policymaker

Abstract

This article tests for asymmetries in the preferences of the euro-area monetary policymaker with 1995:1–2005:2 data from the latest update of the European Central Bank's (ECB's) Area-wide database. Following the relevant literature, we distinguish between three types of asymmetry: precautionary demand for expansions, precautionary demand for price stability and interest rate smoothing asymmetry. Based on the joint generalized method of moments (GMM) estimation of the Euler equation of optimal policy and the aggregate supply-aggregate demand (AS-AD) structure of the macroeconomy, we find evidence of precautionary demand for price stability in the preferences revealed by the monetary policymaker. This type of asymmetry is consistent with the ECB's definition of price stability and with the priority of credibility-building by a recently created monetary authority.

¹CEMPRE–Centro de Estudos Macroeconómicos e Previsão - is a research centre supported by the Fundação para a Ciência e a Tecnologia, Portugal, which is financed by European Union and Portuguese funds.

Acknowledgment

We thank Fabio Canova's comments and suggestions to an earlier version of this article. The usual disclaimer applies.

Notes

¹CEMPRE–Centro de Estudos Macroeconómicos e Previsão - is a research centre supported by the Fundação para a Ciência e a Tecnologia, Portugal, which is financed by European Union and Portuguese funds.

² We thank Elvira Rosati, of the ECB, for providing this latest version of the AWMD.

³ In addition to reviewing the main European integration facts and the literature on the convergence of cycles in the EMU countries, Aguiar and Martins (2005) present econometric evidence of structural stability of an aggregate euro-area AS-AD system and of an aggregate policymaker's loss function coefficients since 1995.

⁴ One functional form often used in the recent literature, the linex function–see Nobay and Peel (2003) and Ruge-Murcia (2004)–behaves quite similarly to our threshold quadratic specification for realistic parameters, but results in a more complex Euler equation and could lead to empirical problems due to the limitations of the data sample.

⁵ This argument is, in fact, parallel to the reason that lies behind the standard use of a 3-month interest rate instead of a shorter money market rate, which would be closer to the actual policy instrument–the fact that i must, in this class of models, simultaneously play the role of policy instrument and connect closely to aggregate demand decisions.

⁶ Notice that the output gap is estimated with the whole available time-series of real output, 1970:1–2005:2, even though only gaps for 1995:1-2005:2 are used in the subsequent estimations in this article. All data and replication files are available from the authors upon request, including the Gauss codes for computing the output gap.

⁷ Our concept of quasi-real time output gap differs slightly from Orphanides and Van Norden's (2002) homonym concept, in that even though the kalman filter's one-step-ahead forecasts of trend output are computed from past data of output only, the filter uses the estimates for the UC model hyper-parameters that have been estimated using the whole sample.

⁸ The asymmetry tests results have proven to be robust to a cross-check (not reported here) consisting of testing for asymmetry in each loss function coefficient using as threshold variable all possible alternative target-variables in the loss function.

⁹ The disparity between our results and Surico's–which is beyond the scope of this article to scrutinize–could result from differences in the model as well as in data: Surico (2003) uses a different sample and periodicity (monthly data 1997:7–2002:10), a different computation of the output gap (two-sided Hodrick-Prescott-filter of industrial production) and a different structural macro model (purely forward-looking AS-AD model without any policy lags).

¹⁰Allowing also for asymmetry in the preferences regarding the output gap–results not reported for space conservation–does not change the statistical evidence in favour of a precautionary demand for price stability.

¹¹ Our framework would also allow for a joint test of the hypothesis of asymmetry in the policy-maker's preferences and of nonlinearity in the aggregate-supply equation. However, succeeding in such an econometric task with GMM estimation requires far more data than is currently available for meaningful analysis of aggregate euro-area policy-making, and thus we defer such exercise for future research.