Active labour market policy, international organizations and domestic politics

The paper deals with active labour market policies (ALMP) in 22 established democracies of the OECD country group, in the period from 1985 to 2002. It seeks to explain variations in policy effort. The focus is on international impacts on domestic policy developments. Since the mid-1990s, both the OECD and the EU have been strong proponents of activation measures. I argue that the impact of these international organizations is conditional on the fit of the proposed policies with path-dependent national policies and politics. The OECD may succeed in developing efficient policy proposals, but it fails at the level of implementation. EU recommendations covary with increased effort in ALMP, but it fails to develop efficient policies. Keywords: Active labour market policydomestic impacts of the EUEUlabour market policyopen method of co-ordination Notes 1. In a recent contribution, Franzese and Hays Citation(2006) developed a modified diffusion model of ALMP. It is based on altering payoffs and the concept of strategic interaction between neighbouring national governments. The basic argument is that unemployment is spatially concentrated in regions that span national boundaries. The ALMP of a government in country A lowers the unemployment in the region that spans countries A and B, and hence the government in country B has an incentive to free-ride. The ALMP expenditures of countries A and B should thus be negatively correlated, and this is what the authors find. However, the authors consider only spending for labour market training programmes. There is only very weak empirical evidence that this scheme lowers unemployment (Kluve Citation2006). The model therefore lacks its economic foundation. In addition the model estimations are based, amongst other things, on the assumption that France, Belgium, and the Netherlands are contiguous with the UK, i.e. that they have a common regional labour market with the possibility of frequent regional labour migration. This seems to be a very heroic assumption. 2. OECD is an international organization that decides on the approval of policy advice by a consensus of the representatives of its member states. Recommendations are based on peer review procedures (Pagani Citation2002). Therefore, in principle, national governments can easily block the publication of unwanted suggestions. They do this less often than expected, though. The underlying reason for this is that OECD recommendations are frequently disregarded both in the domestic political process and by the national mass media. Hence, national politicians have little to fear domestically if they do not follow these recommendations (Armingeon Citation2004). 3. These are Australia, Canada, Japan, New Zealand, Norway, Switzerland, and the USA; the established democracies which made up the core of the EU in 1995 (Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Luxembourg, the Netherlands, Sweden, and the UK) and the new Southern democracies of the EU (Greece, Portugal, and Spain). 4. For descriptive purposes both coefficients have been multiplied by 100. 5. Note that this indicator is also standardized by GDP. If real GDP increases, this means that a stagnant figure of ALMP per unemployed indicates real absolute growth of ALMP per unemployed (all other things being equal). 6. This group includes the EU nations of 1995 minus Greece, Portugal, and Spain. 7. One could also argue that the established democracies of the EU countries had expanded their ALMP efforts particularly in the early 1990s, and that the EU adopted these national policy changes in 1997. If this were true, the correlation between group membership and change rate of ALMP should not differ in the later periods from the earlier periods. However, calculating these coefficients for the dataset with six periods, the only significant and positive correlations are in the last two periods for both operationalizations of ALMP. Similarly, in the dataset with three periods, the correlation between the first two periods is smaller than in the last two periods, though these coefficients were all not significant. 8. The controls were the lagged differenced variable, the lag variable (level), GDP growth, trade, the variable for ALMP effort, interaction terms of ALMP effort with welfare regimes, and the political complexion of the government in cases of employment ratio and long-term unemployment. For models with the difference of the share of good programmes of the total ALMP, the controls were the lagged dependent variable (difference and level), GDP growth, trade, and dummies for welfare regimes.