Inflation divergence in the euro area: the Balassa-Samuelson effect

Abstract

This study estimates the Balassa-Samuelson effect for 7 EU countries. According to this effect, inflation differential between the tradable and the non-tradable sector can be attributed to unbalanced productivity growth between the two sectors. For the euro area this implies that countries with higher relative productivity growth in the tradable sector will suffer from higher inflation. We test the Balassa-Samuelson effect by using cointegration techniques. The empirical analysis supports the existence of a Balassa-Samuelson effect for 6 of the sample countries.

Acknowledgements

I would like to thank Nikitas Pittis and Dimitris Malliaropoulos for valuable comments.

Notes

¹ A number of authors complement the Balassa-Samuelson model by introducing demand side effects in the short-run. For a survey of the literature on the Balassa-Samuelson effect see Froot and Rogoff (1995).

² Earlier empirical studies of the Balassa-Samuelson effect can be distinguished into studies that test the relationship between productivity differentials and dual inflation (Asea and Mendoza, 1994; De Gregorio et al., 1994a, b) and studies that test whether changes of the real exchange rates could be explained by productivity differentials between tradables and non-tradables (Balassa, 1964, 1973; Officer, 1976; Hsieh, 1982; Marston, 1987; Edison and Klovland, 1987; Froot and Rogoff, 1991).

³ See, among others, Alberola and Tyrvainen (1998).

⁴ This is the largest available sample since the ISDB database has not been updated after 1998.

⁵ The tradable sector includes mining and quarrying and manufacturing and the non-tradable sector consists of electricity, gas and water, construction, wholesale and retail trade, restaurants and hotels, transport, storage and communications, finance, insurance, real estate and business services and community, social and personal services. Agriculture is excluded because EC subsidies and the Common Agricultural Policy are likely to distort the relationship between productivity, prices and wages. Similarly, the government sector is excluded since there are no data on capital, and, consequently, we cannot compute factor productivity.