A permanent effect of temporary immigration on economic growth

Abstract

Immigration can help to lessen the burden of ageing for the welfare states of most Western economies. To show this, we develop a decomposition framework for Gross Domestic Product (GDP) per capita which deals with the impact of both ageing and immigration on economic growth. Using a Vector Error Correction Model (VECM) for the Netherlands during 1973 to 2009, we demonstrate the empirical relevance of some crucial interactions between elements of that decomposition. The conclusion is that even temporary immigration may help to alleviate the ageing problem through a positive long-term contribution to employment, wages and GDP per capita, as long as the immigrants are able to participate in the labour force in tandem with the native population. Unfavourable short-term effects should be avoided through a gradual phasing in of immigration policies.

Keywords:

• ageing
• immigration
• growth
• unemployment
• skills

JEL Classification::

• E24
• F22
• O15
• O52

Acknowledgements

We thank Frank Cörvers for his contribution to the initial research on this article and his help with the data in the current version.

Notes

¹ The data sources are presented in Appendix 1.

² For microeconomic evidence, see Vandenberghe and Waltenberg (2010) and the literature reviewed therein. See also OECD (2008, Box 1.7) for a nuanced view.

³ This is elaborated in Muysken and Ziesemer (2011).

⁴ We consider immigration to be a process which is largely unrelated to emigration and we therefore do not include the latter.

⁵ Other criteria allow for one to four lags. However, having two lags in the error correction model should allow us to avoid a serial correlation bias. Using more lags reduces the degrees of freedom even further.

⁶ Identification of the r = 5 cointegrating vectors requires at least r − 1 = 4 restrictions setting coefficients in the long-term relations to zero (see Patterson, 2000, Chap. 14). As we have six variables in each long-term relation four zeros lead to bi-variate long-term relations.

⁷ We use a permanent shock in Muysken and Ziesemer (2011) with a net migration variable rather than gross immigration used in this article. The results are very similar indicating the robustness of our results.

⁸ Further details for the regression are presented in Muysken and Ziesemer (2011).