The European experience of both high levels of unemployment and long individual unemployment spells is not shared by other industrialized countries like the US or Japan. There are also clear differences in the design of such institutions as unemployment benefits, minimum wages, payroll taxes, firing costs and labour union practices in these countries. These facts have stimulated research on the role of institutions in explaining unemployment discrepancies, a central focus of CEPR's research programme on `Product Market Integration, Labour Market Imperfections, and European Competitiveness'. Supported by the Commission of the European Communities under its Human Capital and Mobility programme, this programme is led by Dennis Snower (Birkbeck College, London, and CEPR) and divides into five teams organized around specific topics. The two teams led by Juan Dolado (Centro de Estudios Monetarios y Financieros (CEMFI), Madrid, and CEPR) and Samuel Bentolila (CEMFI and CEPR) met at a joint CEPR workshop on `Imperfections in European Labour Markets' held at CEMFI on 24/25 February.

The Impact of Labour and Product Market Institutions
In the first paper, `Product Quality and Worker Quality', John Abowd (Cornell University), Francis Kramarz and Antoine Moreau (INSEE) studied the relationship between product quality and worker quality using a model which, under certain conditions, establishes a direct link between product price, product quality and workforce quality. To measure product quality, the authors use the evolution of the product's price relative to its broader product group, while worker quality measures assess the firm's average person effect and personal characteristics effect from individual wage rates. According to their model, these two measures should be positively related under perfect competition but might not be under imperfect competition. Using detailed firm level data from the French Producer Price Index, they find a weak, generally positive, relationship between both measures of quality. Marc Van Audenrode (University of Quebec, Montreal) questioned some of the empirical evidence relating to the measurement of product quality. He showed that similar types of petrol in neighbouring US states had consistently different prices.

In `Fiscal Policy Coordination with Demand Spillovers and Unionized Labour Markets', Huw Dixon (University of York and CEPR) and Michele Santoni (University of York) analysed fiscal policy coordination within a general equilibrium model of the EU that allows for an equilibrium rate of unemployment. They show that there is a positive demand externality and that government expenditures are strategic complements: the optimal level of government expenditure in one country increases with the level of expenditure in another. The combination of both features implies that uncoordinated EU fiscal policy involves too little expenditure. This result contrasts with existing analyses which stress negative externalities, such as exported inflation and interest rate effects. Emmanuel Petrakis (Universidad Carlos III de Madrid) pointed out that by assuming a monopolistic market structure in output markets, the authors rule out interesting strategic interactions that might feed back on the bargaining process between employers and unions. Furthermore, he questioned the assumption of complementarity between private and public consumption in household preferences, and wondered how robust the results are to the alternative assumption of substitutability.

In `Unemployment and Job Security', Pilar Diaz-Vázquez (Birkbeck College, London) and Dennis Snower explored the effect of job security legislation on the average level of unemployment in the long run. Their main conclusion is that job security legislation promotes employment when macroeconomic shocks are transient and employees have little market power in the wage determination process. But when shocks are prolonged and employees have substantial power in wage negotiations, such legislation reduces employment. This might be an explanation for Europe's favourable unemployment performance vis-à-vis the US in the 1950s and 1960s and its unfavourable performance since the mid-1970s. Omar Licandro (Universidad Carlos III de Madrid and FEDEA) emphasized that the main result of the paper depends crucially on the wage set-up: an increase in firing costs induces a rise in wages because the wage curve is negatively sloped. In standard models this can never occur since the labour supply is positively sloped, however.

In `Markets, Institutions and Wages', Joop Hartog and Coen Teulings (Universiteit van Amsterdam) compared the estimations of individual wage equations for the US (a very decentralized bargaining system) with those for the Netherlands (a more centralized/corporatist setting). They find that four types of wage differentials in the Netherlands are smaller than in the US: those associated with industry affiliation, time size, bargaining regime and tenure. This contradicts the notion that a decentralized system has smaller non-competitive wage differentials than a corporatist system. The authors also extend the comparison to eight OECD countries and find that corporatism tends to reduce inter-industry wage differentials and flatten wage tenure profiles. Juan Francisco Jimeno (Universidad de Alcalá de Henares and FEDEA) argued that labour market institutions are largely endogenous and the countries' sizes could be particularly relevant in this case. Thus, comparing the US with the Netherlands is like comparing a national labour market with a local one.

In `Quitting Externalities, Employment Cyclicality and Firing Costs', Alison Booth (University of Essex and CEPR) and Gylfi Zoega (Birkbeck College, London) investigated the possibility that firing costs may be a second-best response to market failure. They develop a model in which workers have firm- and industry specific skills, and in each period there is some quitting. This makes the private discount factor, used by firms in making decisions about hiring and training new workers and firing incumbent workers, higher than the social discount factor. As a result of the quitting externality, firms underinvest in training and employment becomes too cyclical. Firing costs can reduce the distortion, since they induce the firm to internalize the worker's full value to society. Finally, since the extent of the quitting externality is likely to differ across sectors, the authors argue that there is a case for determining firing costs at the sectoral level. Michele Santoni argued that one problem is that the authors do not consider a standard argument whereby severance payments may reduce workers' resistance to redundancies, making employment more rather than less cyclical. A second problem, related to the prescription of sectoral firing costs, is that firms and unions may engage in rent-seeking activities which cause misallocation of resources among sectors.

Mismatch and Persistence in Labour and Product Markets

In `Unemployment Duration, Unemployment Benefits, and Labour Market Programmes in Sweden', Kenneth Carling, Per-Anders Edin, Bertil Holmlund (Uppsala University) and Anders Harkman (AMS) analysed the impact of Swedish institutions on unemployment duration. On the one hand, finite duration of unemployment benefits, by lowering workers' reservation wages, should raise the probability of leaving unemployment as the expiration date approaches. On the other hand, labour market programmes offer the unemployed either training courses or public jobs, which allow them to regain eligibility for benefits, thereby mitigating that effect. The paper's results indicate that the exit rate from unemployment increases by 170% around the time benefits are due to lapse. This effect is quite significant, but since the comparable figure for the US is about 400%, the authors conclude that, to some extent, active labour market programmes diminish the incentives for leaving unemployment. Jonathan Leonard (University of California, Berkeley) indicated that the observed increase in job finding is relative to non-receivers of unemployment benefits, a group of workers with special characteristics and whose job-finding rate actually falls. This, he argued, is not explained by the authors' framework.

In `The Duration of Unemployment and the Persistence of Wages', Jonathan Leonard and Marc Van Audenrode examined unemployment duration in Belgium. Their estimates for a sample of workers displaced by large-scale layoffs suggest that the main reason for long unemployment duration in Belgium is not a declining probability of leaving unemployment but a low probability at any point in time. This results from low job vacancy rates. The authors also compare wages before and after displacement, separating economy-wide, firm- and individual-specific wage components. Their results show that the individual-specific component is the most persistent, especially if the worker is displaced from a dying firm: there is clearly a lower stigma from that attached to being dismissed by a going concern. Leonor Modesto (Universidade Catolica Portuguesa and CEPR) questioned the impact of vacancy rates on unemployment duration as a pure demand-side effect, since vacancies should also affect reservation wages. Modesto also stressed the need for an assessment of the relative importance of labour market institutions as opposed to personal characteristics.

In `The Anatomy of Unemployment Dynamics', Jaap Abbring and Gerard van den Berg (Universiteit van Amsterdam), and Jan van Ours (Erasmus Universiteit) decomposed aggregate unemployment dynamics in France into its incidence (the likelihood of becoming unemployed) and duration components. The business cycle was found to affect the outflow from unemployment positively, but not the inflow, which is mostly a seasonal variable. There is a negative effect of unemployment duration on the probability of leaving unemployment after the first year and a half of a spell, but not before that time. In addition, observed probabilities of exit from unemployment decrease more slowly in recessions than at the top of the cycle. This contradicts the presumption that employers prefer hiring the short-term rather than long-term unemployed. Bertil Holmlund challenged the finding of no cyclical effects on the inflow into unemployment: the raw data suggest cyclical behaviour, and the inflow is measured inaccurately as the number of people who have been unemployed for less than a quarter.

In `Unemployment Duration and the Relative Change in Individual Earnings: Evidence from Austrian Panel Data', Karl Pichelmann (Institut für Hohere Studien, Vienna) and Monika Riedel (Universität Klagenfurt) studied the wage changes experienced by Austrian workers after an unemployment spell. Their estimates indicate that the duration of the unemployment spell has short- and long-run negative effects on wages, that there is a higher short-run loss if the worker also changed industry, and that there is no effect of previous job tenure length. Per-Anders Edin noted that the wage change effects are estimated jointly with an equation explaining unemployment duration. This is quite poor, since the latter depends only on the worker's age.

In `Regional Unemployment Persistence (Spain, 1976–93)', Juan Francisco Jimeno and Samuel Bentolila analysed the determinants of regional unemployment persistence in Spain. Their analytical framework shows that real wage flexibility, migration, and procyclical labour participation all reduce regional unemployment persistence, and have enhanced effects if all simultaneously move in the right direction. Their empirical results suggest an extreme degree of persistence of relative regional unemployment rates in Spain arising from the interaction of low migration, low cyclicality of participation, and low real wage flexibility. Gerard van den Berg pointed out that the authors need to trace the causes of unemployment persistence back to labour market institutions, housing market imperfections and government responses to regional shocks.