Monetary integration vs. real disintegration: single currency and productivity divergence in the euro area ^#

^#Paper submitted at the INFER-supported international workshop Asymmetries in Europe: causes, consequences, remedies, Department of Economics, University Gabriele d’Annunzio, Pescara (Italy), March 31–April 1, 2017

Abstract

Productivity slowdown plays a prominent role in the build-up of the euro area crisis. This phenomenon affected member countries asymmetrically, causing divergence in their productivity trends. Recent research traces this divergence back to monetary integration. After reviewing the arguments that link real “disintegration” of the euro area to its monetary integration, we assess them empirically by modelling the evolution of labour productivity using a panel of sectorial data. The results indicate that monetary unification may actually have fostered divergence in productivity trends, and suggest some economic policy measures that could prevent further divergence.

Keywords:

• total factor productivity
• models of trade with scale economies
• macroeconomic issues of monetary unions
• labour productivity
• labour contracts

JEL Classifications:

• D24
• F12
• F45
• J24
• J41

Notes

^# Paper submitted at the INFER-supported international workshop Asymmetries in Europe: causes, consequences, remedies, Department of Economics, University Gabriele d’Annunzio, Pescara (Italy), March 31–April 1, 2017

1. Canzoneri et al. (2002) give a less benign interpretation of inflation differentials, seeing them as a structural phenomenon that might be a source of conflict in the monetary union.

2. The sectorial composition of output plays a role in another strand of literature that relates the slowdown in productivity to structural change, and in particular to the increasing weight of low-productivity tertiary activities (e.g. Delli Gatti et al. 2012).

3. We are grateful to an anonymous referee for pointing this out to us.

4. In his productivity equation, Sylos Labini (1983) defines this as the “Smith effect”, tracing it back to Chapter III, Book I of Adam Smith’s Wealth of nations.

5. Besides the “Ricardo effect”, a different rationale for the relation between wage level and productivity, working through worker effort, rather than labour misallocation, is provided by efficiency wage theories (Katz 1986). We are grateful to an anonymous referee for this observation.

6. Since our equation takes into account different channels of transmission, referring to different theoretical models and concepts of productivity, following Cette, Fernald, and Mojon (2016) we decided to estimate the model for average labour productivity and total factor productivity. The results do not differ qualitatively (those for total factor productivity are reported in the online Supplementary material).

7. Releases 2016 and 2012 (see O’Mahony and Timmer 2009; and Jäger 2016, respectively).

8. Since this indicator is not available before 2003, we reconstructed it using the National retail interest rates (NRIR) N5 series (medium and long-term loans to enterprises) previously published by the European Central Bank. Where this was missing, we used the Lending rate data obtained from the World Bank.

9. Following the advice of an anonymous referee, we used the ULC-based REER index provided by the IMF (2017). Previous results obtained using the CPI-based measure provided by BIS (2017) are reported in the online Supplementary material. The results are robust to this change.

10. More specifically, we used an average of the EPR_V1 and EPT_V1 indicators, measuring the strictness of employment protection against individual dismissals for regular contracts and temporary employment, respectively. The average was weighted with the shares of temporary and permanent employment extracted from the OECD Labour Force Statistics (LFS). We used version 1 of each indicator as this version is available for a longer sample.

11. An anonymous referee suggested we use a composite indicator of governance quality, constructed as the simple average of several indicators. Although this strategy has been followed in major studies (e.g. Rodrik 2008), estimates in Lambsdorff (2003) and Nifo and Vecchione (2014) show that different institutional quality indicators affect productivity with different coefficients. Taking their simple average would amount to imposing an equality constraint on their coefficients, which could result in potentially biased estimates. A more promising estimation strategy is to construct a synthetic institutional quality index, as in Nifo and Vecchione (2014). We leave this for future research.

12. We tried to account for innovation using variables such as Business enterprise R&D expenditure and personnel by industry. However, the data provided by the OECD did not allow us to reconstruct a panel with enough observations in each sector for the estimation to be performed.

13. The panel need not to be balanced (i.e. T can vary across individuals), and both p and q can vary across individuals. Moreover, q can vary across regressors. The model can include individual deterministic components (such as linear trends or dummies). We omit these further generalizations to avoid notational clutter. The only important restriction is that the number of parameters must be such as to allow separate estimation of the model for each individual.

14. In a limited number of cases in which the estimation procedure did not converge, this dynamic specification was imposed.

15. The total number of observations depends among other things on the measure of productivity considered (data on TFP is missing for Italy in 2014) and on the dynamic specification selected, and is reported for each estimated equation.

16. Their simple correlation coefficient is equal to −0.43 with a Student’s t of −3.93.

17. An anonymous referee pointed out that using a ULC-based REER may indirectly account for the impact of labour market reforms on competitiveness through wage moderation. This could explain why in Table 1 the coefficients of EPL are generally smaller than in Table 2 of the online Supplementary material.