The transmission of monetary policy and technology shocks in the euro area

Abstract

This article analyses the response of a set of euro area macroeconomic variables to monetary policy and technology shocks based on structural Vector Auto-regressions (VARs). The data set runs from 1970:1 until 2006:4 and includes a novel long-run series for hours worked per capita in the euro area. We find that real macroeconomic variables follow a hump-shaped response after monetary policy shocks and jump on impact after technology shocks. We also provide evidence that hours worked fall after a positive technology shock. These conclusions are robust to different sample periods and specifications of the variables.

Acknowledgements

The views expressed in this article are of the authors and do not necessarily those of Banco de Portugal. We thank the comments of two anonymous referees and participants at the Banco de Portugal's internal seminar.

Notes

¹ See also Erceg et al. (2005) for an assessment of the reliability of identifying technology shocks using long-run restrictions on a VAR.

² The data set includes all the 12 countries participating in the euro area in 2006.

³ The behaviour of the constructed series seems reasonable, namely when compared with the one of the ECB's estimate of euro area average hours worked published in the October 2004 Monthly Bulletin. The ECB used annual data from the European Labor Force Survey, which is available only at an annual frequency and for a relatively short time span.

⁴ Since the presence of deterministic trends affects the power of the unit root tests, and the presence of unit roots tampers inference on the significance of deterministic trends by altering the sampling distribution of the relevant test statistics, we also tried to apply the unit root tests after looking for the presence of deterministic trends. Although not reported here, inverting the testing sequence did not alter the outcome regarding the low-frequency properties of all our variables.

⁵ The reported tests include a constant, but no time trend.

⁶ This problem is well documented in Kwiatkowski et al. (1992) and Caner and Kilian (2001).

⁷ As suggested by Zivot and Andrews (1992), the break date is searched between the 15th and the 85th percentile of the sample.

⁸ The ADF test statistic for money velocity after removing the estimated quadratic trend is −3.75, thereby implying the rejection of the null of unit root at the 5% significance level.

⁹ These bands were computed by bootstrap simulation with 2.000 draws.

¹⁰ An initial negative response of inflation to an expansionary policy shock has been dubbed as the ‘price puzzle’. This effect has also been identified in several empirical studies for the US (Altig et al., 2005).

¹¹ This is so in spite of Galí (2004) using employment rather than hours as the measure of labour input.

¹² Notice that changing the definition of the labour input requires re-specifying productivity and nominal wage variables accordingly.

¹³ This latter measure is stationarised using a broken trend with a break date of 1974:1. As indicated by the Zivot–Andrews test, the resulting variable appears to be stationary.