The high and persistent unemployment experienced by many industrialized countries in recent years has stimulated research into the workings of labour markets by macro and labour economists in both Europe and North America. A substantial body of theoretical and empirical work has both refined the various competing explanations of the sources of unemployment and provided a corresponding variety of policy recommendations for governments seeking to reduce it, but the results of these studies have reveal a remarkable lack of consensus. Labour economists disagree over whether levels of unemployment vary in response to the same factors in different countries, which policies will prove most effective in reducing unemployment even when its source is clearly identified, and the extent to which different policies may be appropriate to different countries on account of disparities in their political, social or economic institutions. These issues were addressed at a joint conference with the Consorcio de la Zona Franca de Vigo on `Unemployment Policy: How Should Governments Respond to Unemployment?', held in Vigo on 24/27 September. The conference was organized by Dennis Snower, Professor of Economics at Birkbeck College, London, and Co-Director of CEPR's Human Resources programme, and Guillermo de la Dehesa, Conseyo Delegado at the Banco Pastor and Vice-Chairman of CEPR. The Centre gratefully acknowledges the financial assistance and organization of this conference provided by the Consorcio de la Zona Franca de Vigo.

In `High Unemployment from a Political Economy Perspective', Gilles Saint-Paul focused on how political and economic factors can interact to determine different countries' levels of political commitment to overcoming unemployment; not only the US but also Japan, Portugal and Switzerland have successfully avoided the high and persistent unemployment that has afflicted many European countries over the last twenty years. First, labour market imperfections may lead to distributive conflicts between skilled and unskilled and between employed and unemployed workers; second, coordination failures may lead to further rigidities that spread until the economy is `sclerotic' which makes gradual reform of the labour market infeasible; and third, the unemployed are a poorly organized minority, so policies to reduce unemployment need the support of the employed, which increases with their own vulnerability to unemployment. Saint-Paul then discussed various policies that may provide politically realistic means of reforming labour markets: introducing a two-tier system of `rigid' and `flexible' contracts to reduce firing costs over the long run, replacing minimum wage legislation with improved transfers, reducing pay-roll taxation, and converting unemployment benefits into employment subsidies.

Samuel Bentolila (Centro de Estudios Monetarios y Financieros, Madrid) suggested that family links and altruism across groups will mitigate the distributive conflicts. Richard Jackman wanted to know if any empirical tests could confirm the political origin or determine the strength of the rigidities. Many rigidities do not favour the politically influential, so this is particularly important for the US, which has both a government structure that is susceptible to manipulation and a highly flexible labour market. Patrick Minford cited recent UK experience to argue that political entrepreneurship should undermine the inefficiencies arising from the conflicts Saint-Paul described.

In `The Unemployment and Welfare Effects of Labor Market Policy: A Comparison of the U.S. and U.K.', with Steve Millard, Dale Mortensen used a dynamic, general equilibrium model of job creation and job destruction to consider the effects of a variety of policy reforms. By applying the well-known Mortensen–Pissarides framework to UK and US data for 1983–92, he showed that adopting UK labour market policies would raise the US unemployment rate by more than 2 percentage points, while implementing US policies in the UK would reduce the unemployment rate by approximately 1.5 percentage points. He also assessed the effects on UK unemployment and economic welfare of a cumulative reform programme based on first reducing unemployed workers' benefit entitlement to a six-month period (as in the US), then eliminating the firing costs associated with employment protection legislation, and finally introducing either subsidies to encourage new hiring or measures to reduce the bargaining power of employed insiders. Their numerical simulations indicated that all these policies reduce the rate and duration of unemployment but only reducing the benefit period can reduce its incidence; further, they all raise both output and labour income.

Giuseppe Bertola (Università degli Studi di Torino and CEPR) welcomed the authors' explicitly dynamic equilibrium approach but noted that their model's specification restricts the interpretation of its results, in particular those concerning income distribution. Michael Burda (Humboldt-Universität zu Berlin and CEPR) suggested extending the model to incorporate hysteresis, investment and signalling.

In `The U.S. versus Europe Unemployment Puzzle: Is there a Tradeoff between Unemployment and Productivity Growth?', Robert Gordon noted that macro and labour economists in the US have concentrated on theoretical and empirical analysis of the decline in productivity growth and the rise of income inequality, while their European counterparts have focused mainly on identifying causes of and policies to reduce high and persistent unemployment. US productivity growth has fallen drastically while unemployment has remained low, and European unemployment has risen while productivity growth has remained high, so this divergence of interests is entirely reasonable. Gordon argued that it is nevertheless misplaced, since there is a trade-off between reducing unemployment and raising productivity. Several institutional factors commonly cited as causes of Europe's higher unemployment – minimum wage legislation, unemployment benefits, higher taxation and regulation of labour and product markets – may also have raised its relative productivity growth. Increased legal and illegal immigration and greater competition from low-wage workers through relatively unrestricted imports have further exacerbated the widening disparity between the incomes of US high- and low-wage workers.

Charles Bean (LSE and CEPR) questioned the causality of this trade-off. While high unemployment may raise productivity growth in Europe, the widening of the US income distribution may also reflect technical progress. Conversely, Europe's minimum wage laws should arrest the widening of the income distribution but at the cost of higher unemployment. Bean then suggested that examining not only the levels of but also the changes in unemployment and productivity growth might shed light on this trade-off.

In `The OECD Problem of Low Wages & Unemployment: The Role of Welfare Support', Patrick Minford integrated a Heckscher–Ohlin–Samuelson model of international trade into a natural rate model of unemployment to consider the effects of an increase in competition from less developed countries. This raises the developed countries' overall welfare, but it also seriously affects their income distributions by reducing the relative wages of unskilled workers. Minford discussed a variety of possible countervailing policies that have been proposed to overcome the resulting social problem. The `negative income tax' appears to perform well in keeping the social peace but requires rigorous means testing; a flat-rate `basic income guarantee' (which is not means-tested and cannot be withdrawn) raises the marginal tax rate for the average family to unacceptable levels; while subsidies to low-wage jobs are less effective in relieving poverty than direct transfers to low-income households. Minford finally suggested decentralizing the control of income support following the Victorian model of charity, to enable local agencies to allocate welfare assistance selectively in a manner that maintains incentives and targets assistance on those most in need.

Jonathan Haskel (Queen Mary and Westfield College, London, and CEPR) noted that this reform would abandon the principle of equal treatment for equal situations and also induce rent-seeking behaviour by the disadvantaged. Alan Manning noted that the group of industrialized countries has expanded historically without suffering the dire consequences predicted by Minford's analysis.

In `About E. Phelps' Theory of Structural Slumps and its Policy Implications', Edmond Malinvaud (Collège de France) welcomed the serious attempt to establish a long-needed theory of medium-term unemployment set out in Edmund Phelps's recent book, Structural Slumps: The Modern Equilibrium Theory of Unemployment, Interest and Assets. While Keynesian and classical models play useful roles in accounting for short- and long-term unemployment respectively, there has been no corresponding theory to explain the medium-term unemployment that is currently afflicting the developed countries. Malinvaud also praised Phelps's non-monetarist approach and its emphasis on the real interest rate. He warned, however, that there are gaps in the analysis and that this approach should not be taken as the theory of medium-term unemployment. Specifically, the book focuses on efficiency wages with market clearing to account for job rationing or involuntary unemployment, since it takes no account of insider–outsider effects or the roles of union wage bargaining or social norms in labour markets. The book's equilibrium assumption that product markets clear, which generates its anti-Keynesian propositions, is also too restrictive. These factors, coupled with concerns about the neoclassical treatment of investment and savings, leave doubts about the model's generality. Malinvaud maintained that the theory must be tested against all its implications, so that more evidence is needed to derive policy implications from this approach, in particular those concerning the divergence of efficiency wages from competitive levels.

Klaus F Zimmermann (Universität München and CEPR) also applauded the Phelps volume but added that recent survey evidence by Charles Bean has shown that there is no single cause for the rise in Europe's unemployment, which reflects the fact that this theory is not complete. The large slow-down in the outflow from unemployment and its effects on labour market performance must also be examined at the microeconomic level. Zimmermann also welcomed Phelps's focus on interest rates but stressed the need to distinguish between short- and long-run rates. Edmund Phelps (Columbia University) replied that his work was not simply neoclassical economics plus efficiency wages, since it adopts a more modern approach that also takes account of imperfect information. More importantly, it is justified empirically, since the level of unemployment it explains is close to that observed in practice.

In `The Employment Content of Output Growth: A Cross-country Analysis', Ana Revenga (World Bank) and Samuel Bentolila showed that output growth induced much greater rises in employment in North America/Oceania than in Europe during 1960–90. They regressed this `employment intensity' of growth on proxies for the structural characteristics of a panel of 11 OECD economies and found that it depends critically on the extent of firms' market power in product markets and the institutional characteristics of the wage-bargaining process. They concluded that European governments that hope to reduce unemployment by promoting growth should adopt policies that aim to increase product market competition, decentralize wage bargaining, and reduce the costs of employment adjustment.

Christopher Pissarides (LSE and CEPR) pointed out that the substantial employment growth observed in the US may also reflect the large increase in its labour supply; moreover, technical progress is not exogenous, and studies of this type should consider the trade-off between the growth of productivity and that of employment. Brendan Walsh (University College Dublin) noted that labour supply is also important on account of cross-country differences in workers' movement from agriculture to industry, and he suggested that studies adopting this framework should also take account of European governments' subsidies to hi-tech, capital-intensive sectors.

Presenting `Labour Market Regulation and Unemployment', Paul Gregg (LSE) and Alan Manning noted that conventional labour market models assume that a reduction in unemployment requires a fall in real wages that can only be achieved through deregulation, and many studies adopting this framework have received empirical support. This evidence is much weaker than is generally believed, however, since the models have focused exclusively on sources of inefficiency that give market power to workers, such as trade unions, welfare systems and legislation governing minimum wages and employment protection. They have ignored the effects of the monopsony power that firms hold on account of imperfect information flows, even without the existence of `company towns'. While economists are easily persuaded that regulation can be justified only on equity grounds, since it reduces labour market efficiency relative to the ideal of perfect competition, a closer examination of deregulated labour markets reveals that they bear little relationship to this paradigm case. Power on both sides of the market must be considered before any firm conclusions can be drawn about the merits of deregulation; indeed, some form of regulation may be needed to maximize labour markets' efficiency.

Juan José Calaza (Université de Paris IX) added that firms' monopoly power and their conduct in the product market will exacerbate these effects, while Daniel Cohen remarked that recent empirical evidence regarding minimum wages supports Gregg and Manning's analysis. Steve Nickell agreed with the authors' assessment of the weak empirical evidence indicating that regulation raises unemployment but also stressed that the recent rise in unemployment still remains to be explained.

In `Gross Job Reallocation and Labour Market Policy', written with Pietro Garibaldi and Jozeph Konings, Christopher Pissarides used data on job reallocation from ten OECD countries during 1982–9 to investigate both its relationship with unemployment and growth and the impact of policies to promote labour market flexibility. His results indicated that countries with less job reallocation had more long-term unemployment (although the relationship with overall unemployment was ambiguous) and higher labour (and GDP) productivity growth. Job reallocation was also lower in countries with stricter employment protection legislation and `more generous' unemployment insurance systems (although a strongly negative association with the duration of such benefits masked a positive association with their level). Finally, Pissarides reported that there was no evidence to suggest that expenditure on industrial subsidies or active labour market intervention measures had any effect on job reallocation.

Brian Bell remarked that the observed cross-country differences in job reallocation may simply reflect differences in the composition of labour markets. Ana Revenga suggested incorporating developing countries into the study to enhance both the limited data set and the relevance of the results. Jacques Drèze stressed the need to consider the types of jobs analysed before drawing any inferences from the data on their reallocation.

In `Wage Subsidy Programs: Alternative Designs', Edmund Phelps stressed that the tax structure is also important: a pay-roll tax will exert a disciplinary effect on unemployment, while a value-added tax will not. The poor conditions of low-wage workers are a major and increasing social problem throughout the OECD area; unemployment is both unfair and inefficient, and it also generates negative spillover or neighbourhood effects. Phelps suggested offering tax incentives to enterprises to hire low-wage workers, which will reduce unemployment and raise wages, so that transfers will be paid to the working rather than the non-working poor. He maintained that the other available remedies would either not yield gains for generations or be less cost effective now; the cost of implementing his proposal in the US would be about $107 billion per annum – roughly equal to the expected cost of the proposed health-care reform. The main difficulty lies in the design of incentives that can prevent individuals from manipulating the system while avoiding placing an unnecessary burden on those higher up the social ladder; this requires graduated schemes (as opposed to flat rates) as well as strict accounting and policing.

Jeff Frank (Royal Holloway College, London) noted that this shift of resources away from the benefit system implies that payments will be less well targeted, since not all low-wage workers will require such assistance. Moreover, as Phelps notes, such a shift will `crowd out' or displace good workers by reducing incentives to invest in training, education and capital if there are more bad jobs and fewer good jobs. Dale Mortensen (Northwestern University) noted that this effect could go either way, depending on the extent to which high- and low-skilled workers are substitutes or complements; his own work suggested that an alternative policy of introducing a hiring subsidy would prove more efficient.

In `The Simple Economics of Benefit Transfers', Dennis Snower maintained that unemployment benefits discourage employment and thus form part of the problem they are intended to solve. He proposed reallocating these funds to improve employment prospects by issuing vouchers that enable the unemployed to transfer their benefit entitlements to employers that are willing to train and employ them. Linking the value of each voucher to the level of unemployment benefit, the durations of the holder's unemployment and subsequent employment as well as the amount of training provided by the employer should make this subsidy both costless and non-inflationary. Snower argued that his proposal would target such benefits effectively on the long-term unemployed and could be structured to minimize the displacement of the currently employed, although this would depend critically on establishing a smooth relationship between the voucher's value and the duration of unemployment.

Jacques Drèze (Center for Operations Research and Econometrics, Université Catholique de Louvain, and CEPR) noted that similar schemes in other EU member countries and elsewhere have reduced labour markets' transparency, in particular by creating large displacement effects. David Coe (IMF) noted that this programme may increase potentially costly `churning' in the labour market and must therefore be considered as part of a wider reform package.

In `Preventing Long-term Unemployment Through a Job Guarantee: Benefits and Costs of Workfare', Richard Jackman and Richard Layard (LSE) proposed restructuring the benefits provided to those out of work for twelve months or more; such `long-term' unemployed are not `productive' since they do not engage in search activity to fill vacancies and exert no disciplinary effect on the currently employed. Jackman and Layard suggested reducing long-term unemployment with a combination of carrot and stick: first, guaranteeing such workers jobs after twelve months by reallocating their unemployment entitlements to subsidize their employment; and second, withdrawing all benefits from those that refuse such jobs. As a result, the long-term unemployed get jobs while the short-term unemployed search harder. Provided the long-term unemployed do not get jobs at the expense of the short-term unemployed, the only costs of this programme are those incurred in its administration and policing.

Guillermo de la Dehesa observed that this proposal does not fully spell out who will provide the `guaranteed' jobs; making the government the employer of last resort would put a strain on the public sector and incur particular dangers reflecting the government's incentives to make its own policy work. Both Patrick Minford (University of Liverpool and CEPR) and Zmira Hornstein (Department of Employment, London) noted that this plan may even raise both unemployment and fiscal costs if there is sufficient substitution between employed and unemployed unskilled workers. Daniel Cohen (Centre d'Etudes Prospectives Mathématiques Appliquées à la Planification, Paris, and CEPR) and Alan Manning (LSE and CEPR) noted that recent French and turn-of-the-century British experiences with similar proposals were either ineffective or counterproductive.

In `Would Cutting Payroll Taxes on the Unskilled have a Significant Impact on Unemployment?', Steve Nickell (Institute of Economics and Statistics, Oxford, and CEPR) and Brian Bell (Nuffield College, Oxford) first examined the pattern of post-war unemployment across the OECD area, which appeared to indicate that the relative decline in the demand for unskilled labour had been inevitable; they estimated that this accounted for approximately one-fifth of the total post-war rise in long-run unemployment in the UK. They then considered the impact on unemployment of providing a wage subsidy to the unskilled by cutting their pay-roll taxation and found that this will have a beneficial effect – although probably a small one – when there are barriers to the acquisition of training or low-skilled wages are inflexible. It may nevertheless be worth pursuing, since the unskilled unemployment rate is relatively high and rising and it is also associated with significant social ills.

Robert Gordon (Northwestern University and CEPR) noted that such analyses of shifts in the relative demand for skilled and unskilled labour should also consider the effects of aggregate demand and alternative tax policies. Dennis Snower agreed and noted that Nickell and Bell's results rely heavily on their assumption that workers bear the entire tax burden, so that the choice of pay-roll taxation versus value-added tax is irrelevant; the proposed wage subsidy amounts to a tax on training and must therefore be integrated with policies on training.

In `Employment Protection Legislation: Employment Implications of State-Mandated Redundancy Pay', Alison Booth (Birkbeck College, London, and CEPR) considered the well-known result that firing costs stabilize labour demand over the business cycle and investigated the effects of fixing such costs by legislation and thus excluding them from pay negotiations. She showed that the wage resulting from a simple model in which both wages and firing costs are negotiated is equal to the opportunity cost of labour, which maximizes the bargaining surplus. `Negotiated' firing costs therefore stabilize employment in bad times, while they also reduce hiring in good times to an extent that depends critically on the firm's discount rate. Firing costs that are mandated by legislation will therefore usually reduce welfare, which suggests that the state should leave their determination to individual or collective bargaining. Booth cautioned, however, that other factors not captured in her model – including moral hazard that prevents firms from insuring against bankruptcy, for example – may nevertheless justify legislation in this area.

Gilles Saint-Paul (Département et Laboratoire d'Economie Théorique et Appliquée, Paris, and CEPR) argued that labour market dynamics may undermine the effectiveness of redundancy pay in insuring workers against the costs of dismissal, while firing costs that force firms to keep matches going may also promote efficiency by reducing turnover externalities. Juan Dolado (Centro de Estudios Monetarios y Financieros, Madrid, and CEPR) noted that these results will be affected by the size of the wedge between employers' labour costs and the wages workers receive and the existence of trigger responses to exogenous shocks. The analysis should also endogenize the capital stock, since firing costs may reduce the level of investment.

Presenting `Technical Development, Competition from Low-wage Economies and Low-skilled Unemployment', written with Henri Sneessens, Jacques Drèze identified and discussed three structural causes of the deterioration of the prospects of less-skilled workers in the technologically advanced countries. First, output growth has stagnated while productivity has continued to rise, leading to `deindustrialization' as the proportion of high-wage, blue-collar jobs in the labour force has fallen. Second, increased competition from low-wage economies has already exacerbated US trade deficits, and developments in Eastern Europe now threaten West European workers' wage and employment prospects. Third, many studies derived from detailed micro-data bases for specific industries or sectors have provided evidence of `skill-biased technical change', since the wage premium for higher education appears to be associated with the spread of micro-computers and the rising share of advanced technologies in the capital stock. Less-skilled workers have therefore experienced rising unemployment in Europe and falling real and relative wages in the US; the industrialized economies must now devise policies to reduce labour costs far enough to promote their employment while providing them with adequate incomes and proper incentives. Supply-side measures to raise skill levels by improving education and training provide the best long-term solution, while measures to boost the demand for `proximity services' – which often have a high intensity of low-skill labour – may also play a role. Drèze then turned to measures based on wage adjustment and maintained that minimum wage legislation accompanied by unemployment benefits and employment subsidies to the lower paid should prove more effective than issuing `participation income' to all adults and providing no unemployment benefits while allowing wages to remain fully flexible.

Olivier Blanchard (MIT) pointed out that designing these policies would involve considerable difficulties while their implementation would entail substantial costs. He also suggested considering their feedback effects on immigration and protectionism. Richard Layard suggested considering the German model of compulsory education as a promising alternative.