In an interview with the Financial Times (March 16, 2010) France’s finance minister Christine Lagarde suggested that Germany is hurting its European partners by “putting very high pressure on its labour costs”. Her insinuation has sparked a vivid policy debate.
Economic theory provides a possible basis for her argument. Work by Davis (1988) shows that reducing minimum wages in a capital abundant country would hurt workers in a labour abundant trading partner. Such an argument is based on a comparative advantage trade model. However, Helpman and Itskoki (2010), Egger, Egger and Markusen (2009) and Felbermayr et al. (2009) have recently proposed models where trade is due to product differentiation and increasing returns to scale. In these models better labour market institutions in one country may benefit its trading partners.
The (stylised) facts
- Domestic and foreign unemployment rates are positively correlated across countries
- Domestic unemployment rates and foreign labour market distortions are positively correlated across countries
To illustrate these facts, it is convenient to summarise foreign variables as trade weighted averages. That is, foreign variables are weighted by a proxy for bilateral trade volumes. Using data from the OECD, the left panel of Figure 1 below shows natural logs of yearly domestic unemployment rates for 20 OECD countries (including France, Germany, Japan, UK, and the US) and the period (1990-2003) on the y-axis. On the x-axis, it plots corresponding foreign rates. A naive linear robust bivariate regression (taking account of outliers) yields a slope coefficient of 0.094 with an associated t-value of 3.77, which provides stylised evidence for the first fact. The right panel repeats this exercise, but plots the foreign tax wedge (the sum of unemployment benefit replacement rate and labour taxes, a commonly used measure of total labour market distortions) on the x-axis. There is again a strong positive relation, with a slope estimate of 0.075 and a t-value of 3.08, providing evidence for the second fact number1.
Figure 1. Foreign unemployment rates and labour market distortions are positively correlated to domestic unemployment
Figure 2 comes back to stylised fact number 1. It shows the share of positive, negative, and statistically insignificant bilateral correlations between unemployment rates of 20 OECD countries in the period 1963-2008, where short-run business cycles are purged by taking 5-year averages. We find that unemployment rates are usually positively correlated over time within country pairs. Negative correlations are very rare2.
Figure 2. Bivariate correlations of unemployment rates, 5-year windows.
While those data are suggestive, they call for a more formal econometric approach. We come back to this in the last section of this column.
One can distinguish at least four different channels through which labour market reforms in one country spill over to its trading partners.
The first link is the effect of labour market institutions on the pattern of comparative advantage, highlighted in models such as Davis (1988) or Davidson et al.(1999). A decrease in the effective wage floor (e.g. unemployment benefits) at home (Germany) decreases unemployment there, driving down the effective relative capital-labour abundance. A relatively capital-rich home economy now produces more of the labour-intensive good itself and imports less. Consequently, labour demand in the foreign country falls, which increases unemployment. This mechanism may rationalise the arguments of Mme Lagarde. However, it relies on a fairly special view of trade (the Heckscher-Ohlin model where Germany is more capital abundant than the rest of Europe).
A second potential link operates through a competitiveness effect. Lower wage costs at home increase Germany’s international competitiveness and make it harder for foreign firms to sell their output. This lowers labour demand and increases foreign unemployment. This argument, however, ignores that Germany’s income has also increased as lower German wages are the result of an efficiency-enhancing reform.
The third channel is an income effect. If labour market reform increases Germany’s employment and wage income, its total spending goes up. Some of the additional spending falls on imports from foreign countries, which increases labour demand and drives down unemployment there. Effects of this type operate in the new economic geography literature but have hardly been explored in models of trade and unemployment. The effect relies crucially on the use of a model where trade is due to product differentiation and increasing returns. This model generally explains trade between industrialised countries much better than the comparative advantage framework.
A fourth link compounds the second and third effects discussed above. It relies on firm selection and heterogeneous firms as in Melitz (2003). The idea is that competitive pressure is actually beneficial since it weeds out inefficient firms. Hence, when Germany lowers its wages and prices, inefficient producers in foreign markets are forced to exit. This makes room for the more efficient firms to expand so that average productivity goes up. Moreover, because Germany demands more exports from foreign markets, foreign exporting firms — which also happen to be the most productive ones – expand. This again increases average productivity. In Felbermayr at al. (2008) it is shown that higher average productivity implies more vacancy creation and hence lower structural unemployment.
Recent theoretical contributions by Helpman and Itskhoki (2010), Egger et al. (2009), and our own work highlight the importance of the third and fourth channels. In Felbermayr et al. (2009) we show that the size of spillovers is stronger if the reforming country is large and central.
The econometric test
In Felbermayr et al. (2009) we go beyond the descriptive analysis discussed in the introduction. The positive correlation of unemployment rates across space may be driven by common business cycles, by changes in labour and product market regulations and by other sources of country-specific heterogeneity. Moreover, there is no indication about how unemployment spillovers actually occur.
To account for these issues, we run standard cross-country unemployment regressions. We include measures of the domestic and the foreign output gaps as well as a host of orthogonal macro shocks and year dummies to deal with business cycle effects. We control for domestic and foreign product market regulation (including trade openness) and purge unobserved country-specific heterogeneity by adding fixed effects. We find the following results:
- Foreign labour market distortions increase the domestic unemployment rate. On average, domestic distortions are about 10 times more important than foreign ones.
- More central (and, hence, more open) countries are more strongly affected by foreign institutional changes. If trade openness converges to zero, spillovers vanish.
- Smaller economies are more strongly affected. An increase in the foreign unemployment rate by one percentage point increases the German unemployment rate by about 0.04 points while the Austrian rate would go up by twice as much.
The policy conclusions
Contrary to what some policymakers believe, better labour market institutions in one country ultimately benefit that country’s trading partners. This insight has implications for the conduct of reform in Europe. Most importantly, there is a case for coordinated action. If countries do not factor in the beneficial spillovers of reforms, they may not pursue reforms that would be socially desirable. Conversely, governments may submit to political pressure and introduce unemployment-increasing policies as the cost of this action is shared with trading partners. The case for cooperation is stronger in Europe than elsewhere, because low trade costs magnify spillovers. Moreover, in the long-run, there is little theoretical and even less empirical basis for a conflict of interests between reformist and non-reformist countries.
Davidson, C, L Martin, and S Matusz (1999), “Trade and Search Generated Unemployment”, Journal of International Economics, 48(2)
Davis, D (1998), “Does European Unemployment Prop up American Wages? National Labour Markets and Global Trade”, American Economic Review, 88(3): 478-494.
Egger, H, P Egger, and J Markusen (2009), “International Welfare and Employment Linkages Arising from Minimum Wages," NBER Working Paper 15196.
Felbermayr, G, J Prat, and HJ Schmerer (2008), “Globalisation and Labour Market Outcomes: Wage Bargaining, Search Frictions, and Firm Heterogeneity”, Journal of Economic Theory, forthcoming.
Felbermayr, G, M Larch, and W Lechthaler (2009), “Unemployment in an Interdependent World”, CESifo Working Paper 2788.
Helpman, E, and O Itskhoki (2010), “Labour Market Rigidities, Trade and Unemployment”, Review of Economic Studies, forthcoming.
Melitz, M (2003), “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity”, Econometrica, 71(6),1695–1725.
1 Using other periods generates some data consistency problems but yields similar results. Using levels instead of logs does not undo the relationship either.
2 Egger et al. (2009) look at the comovement of the aggregate EU rate of unemployment and the US one, and find that they are strongly positively correlated over time. They interpret this finding as evidence for fact number 1.