Multivariate asymmetric loss functions of the European Central Bank

ABSTRACT

This study investigates asymmetric loss functions in the forecasts of real output growth and inflation published by the European Central Bank (ECB). In contrast to previous studies that examined each variable independently, this study evaluates the ECB forecasts in a multivariate framework, which accommodates the interaction of forecast errors of different forecast variables. The empirical results indicate that ECB is likely to produce under-predicted real output growth and over-predicted inflation forecasts for the current year. For the next-year forecast, ECB produces unbiased inflation forecast by applying multivariate framework, while the forecast of real output growth is over-predicted. Those forecasts are rational under multivariate asymmetric loss functions. Our results suggest that ECB attempts to maintain its credibility by providing unbiased inflation forecasts, and thus, puts first priority on price stabilization for medium term.

KEYWORDS:

• Monetary policy
• central bank communication
• asymmetric loss
• forecast evaluation

JEL CLASSIFICATION:

• E37
• E52
• E58

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ See Komunjer and Owyang (2012) for details.:

² Following Caunedo et al. (2020), this study uses p = 2. See Komunjer and Owyang (2012) for discussion on calibrating p.

³ See https://www.ecb.europa.eu/pub/projections/html/index.en.html for details. See also ECB (2016) for the procedures and organization to produce forecasts.

⁴ Univariate estimation yields results similar to those of the non-separable case. The estimations under different values of p broadly yield qualitatively consistent results: weaker asymmetry for $p=1.5$ and stronger asymmetry for $p=2.5$.

Additional information

Funding

This research was supported by a grant-in-aid from Zengin Foundation for Studies on Economics and Finance.