Are economic growth and the variability of the business cycle related? Evidence from five European countries

Abstract

We use a long series of annual data that span over 100 years to examine the relationship between output growth and its uncertainty in five European countries. Using the GARCH methodology to proxy uncertainty, we obtain two important results. First, more uncertainty about output leads to a higher rate of growth in three of the five countries. Second, output growth reduces its uncertainty in all countries except one. Our results are robust to alternative specifications and provide strong support to the recent emphasis by macroeconomists on the joint examination of economic growth and the variability of the business cycle.

Keywords:

• output growth
• output growth uncertainty
• GARCH

JEL Classification :

• C22
• E32

Acknowledgements

We are grateful to two anonymous referees for their helpful comments and suggestions. We also thank Keith Blackburn and Marika Karanassou for useful comments on an earlier draft. The usual disclaimer applies.

Notes

h _t is positive with probability one and is a measurable function of Σ _t−1, which in turn is the sigma-algebra generated by .

The sample period for France, Germany, Italy, Sweden and the UK starts in 1815, 1850, 1861, 1861 and 1860, respectively. In all countries, the sample period ends in 1999.

The autoregressive model of Germany includes two dummy variables: the first dummy captures Germany's separation in 1946 and the second the reunification of 1990.

An alternative channel regarding the effect of growth on growth uncertainty works via inflation uncertainty. Growth raises inflation and inflation uncertainty, which reduces growth uncertainty (the Taylor effect). We do not attempt to test for this channel and hence estimate a bivariate GARCH-M-L model, as our relatively short sample size is likely not to produce reliable results.

Inflation is measured by the annual growth in the Consumer Price Index (CPI).

That is, in equation (1) we add the term: where π _t−l denotes the inflation at time t−l. Similarly, in equation (2) we add the term: .

That is, in equation (2) we add the term: where D _t−1=1 if and 0 otherwise.

An alternative approach to account for structural breaks is the regime-switching model proposed by Hamilton (1989) and used by Bhar and Hamori (2003) to model the volatility process in the growth rate of Japanese real GDP.

The years of the Great Depression differ among countries and are taken from Romer (2007). They are 1930–1932 for France, the UK and Sweden, 1928–1932 for Germany, and 1929–1933 for Italy.