Reforming Europe's labour markets: political issues
While there are many reform proposals to reduce European unemployment, little attention is devoted to the political feasibility of these reforms and their impact on different social groups. In this paper we discuss these issues. We first argue that the role played by labor market rigidities in Europe is best understood as political, in the sense that they help build consensus within the ''core'' of society, at the expense of a ''periphery'' of politically and economically excluded individuals. We then go on and examine why it is politically difficult to enforce reforms of the labour market. We conclude by suggesting various strategies that may be taken advantage of to implement labour market reform.
Do rigidities enhance economic efficiency ?
In a world of imperfections, a case can be made in favor of some rigidities. The distortions they create may be useful in curing other types of market failures such as externalities and imperfect competition. However, the rigidities that we have in European institutions are certainly not meant to offset such market failures, and in many cases they would be a pretty far fetched instrument to achieve this goal. For example, it is often argued that firing costs, by reducing turnover, increase economic efficiency because they induce workers to invest more in on-the-job training and firm-specific human capital. It is true that in a world of imperfect labor markets, imposing firing costs may indeed have this effect. But, if we believe that there is not enough training provided and that an important component of it is on the job training, directly subsidizing it is certainly a much more appropriate instrument than increasing firing costs. Furthermore, the argument equally applies to voluntary quits, and nobody ever advocated to impose a tax on workers who quit their jobs. It is thus quite clear that while firing costs may have desirable side-effects, they were never meant to or designed so as to increase economic efficiency. Firing costs prevents job destruction in declining sectors; the value of training in these sectors is presumably very small. In many European countries, firing restrictions, minimum wage provisions and generous unemployment benefits were imposed by the unions so as to increase their bargaining power, often at the expense of outsiders. These outcomes were achieved either at times when unions were quite strong (such as the sixties), or when insiders wanted to protect themselves from redundancies and underbidding after the initial boost in unemployment triggered by the first oil shock.
Do they fulfill redistributive roles ?
Another common argument about rigidities is that they fulfill desirable redistributive functions. The minimum wage, for example, increases the welfare of the poorest workers. This raises two questions. First, do they indeed perform such roles ? Second, are they the most efficient instrument in doing so ? We believe the answer to both questions is no.(1) First, to the extent that rigidities exclude people from employment, they actually lower, not increase, the welfare of the poorest. Minimum wage regulations redistribute income not only from high-wage earners and capital to low wage earners, but also from the unemployed to the employed. Firing costs reduce both job creation and job destruction, thus reducing the unemployed's probability of finding a job and considerably increasing the duration of unemployment. It is hard to understand why a society that cares about equity and redistribution would support such institutions. Second, even assuming that rigidities perform such redistributive role, they are probably not the most adequate instrument for that. Rigidities magnify distortions by making employment much more reactive to taxes. For example, in a world without a minimum wage, a payroll tax (such as social security contributions) would be partly offset by a fall in after-tax wages. This fall would limit the rise in the total cost of labor and therefore dampen the adverse consequences of the tax. A minimum wage would prevent post-tax wages from falling, thus increasing the distortionary impact of the tax on employment. The same mechanism is at work with any institution which puts a floor on the workers' alternative options, such as unemployment benefits. It is more efficient to redistribute income via standard taxes and transfers without preventing the economy from attaining full employment.
So why have them ?
So are rigidities just the outcome of blindness and selfishness ? The answer is not so simple. We believe rigidities in European labor markets do fulfill an important role, which is similar to a redistributive role, but is motivated by political considerations, rather than normative ones. Rigidities are in fact a way of alleviating conflict by homogenizing a politically important ''core'' at the expense of an ill-organized ''periphery''. The employed are part of the core, while the unemployed are part of the periphery. In a heterogenous society - say a society with wide income inequalities - sharp distributive conflicts tend to arise. These conflicts may have adverse effects on the economy by leading, at the political level, to highly distortionary tax rates, populist and demagogic practices. Also collective violence such as riots, strikes and lockouts, and corruption may arise. Such conflicts and the associated outcomes are illustrated by the traditional experience of Latin America and the more recent one of Russia. By artificially creating diverging interests among identical people, rigidities paradoxically reduce conflicts. In a society where many poor try to expropriate a few rich, making some of the poor richer and the other poorer creates a convergence of interests between the former and the rich. If the poor who is made richer is the politically decisive group, then there will be less redistributive conflict. This is exactly what an institution such as the minimum wage achieves: it increases the income of those unskilled workers lucky enough to get a job, while reducing the income of those who end up unemployed. Collective decision is then shaped by diverging interests within the ''core'' of employed workers. These interests are less divergent than in the absence of rigidities. Society is therefore more homogenous from a political point of view. The problem, of course, is the existence of a periphery which is excluded both from the market, and, because it is an ill-organized minority, from the political system. Indeed, lack of political influence in the periphery is a prerequisite for rigidities to reduce polarization. This lack of political influence is achieved because the unemployed are a poorly organized minority which is quite heterogenous.(2)
The rationale for the ''European model'' has therefore little to do with equity, but has a lot to do with political stability. Note that the ''core'' is allowed to be larger when society is more homogenous. When exogenous forces such as technical change or international competition tend to widen income differentials, there is some social pressure to increase rigidities in order to preserve the core's homogeneity. The problem is that this reduces the size of the core, and that the periphery cannot be allowed to grow above some bounds without threatening the viability of the model. It is this concern, of course, which has suggested that another model could be tried and that labor market flexibility should be increased.
The difficulty of implementing reform
There are several reasons why, despite the frequent calls for liberalizing European labor markets, reform may be politically difficult to implement. First, there may be more losers than winners. Second, a large minority of winners may be easily identified, with the rest of the gains being much more diffuse in the remaining majority.
The employed are politically more powerful than the unemployed
Unemployment is usually thought of as a burning political issue. Yet when we look at the long run, we see that in most European countries, unemployment has steadily risen without causing major social or political disruptions. The reason is that even if unemployment is a painful experience for a large number of people, these people are a poorly organized minority and therefore are unlikely to influence political outcomes. The majority of the people are actually employed, so that they will be in favor of policies to fight unemployment only if their exposure to unemployment is large enough and/or if these policies are designed so as to favor their interests. Evidence from election results suggests that a sharp increase in unemployment is much more likely to lead to an adverse outcome for the incumbent government than a high, but stable level of unemployment. This is consistent with the view that unemployment is an issue not when it is simply high, but when large job destruction makes the employed more exposed to it. Similarly, the more the employed are sheltered, the more they would favor policies that protect their jobs and increase their bargaining power even if those policies introduce distortions that lead to higher unemployment.
So any attempt to reduce unemployment must face the reality that policy is actually determined by the interests of the employed, not the unemployed. In particular, many rigidities that are thought - although this is open to debate - to be responsible for European unemployment, such as minimum wages, unemployment benefits, firing costs and union wage setting, actually benefit (some of )the employed and are therefore difficult to remove.(3)
This logic explains why policies that are generally advocated as reducing unemployment are not implemented in practice. For example, social security contributions are often blamed as increasing the cost of labor and creating unemployment. But if firing costs prevent the employed from being fired, an increase in social security contributions will have little impact on them because they will not bear the burden of the induced reduction in labor demand. Those who lose are the unemployed and new entrants into the labor market, as firms reduce new hires due to the expected increase in labor costs. So raising payroll taxes is a politically efficient way of generating revenue.
Similarly, the proposal of reducing payroll taxes for the least skilled workers may run into the problem that it is a redistribution of income from a majority to a minority. The same issue arises with the idea of shifting from welfare to workfare, for example by using the unemployment insurance budget to fund millions of relief jobs for the long-term unemployed. The majority of the employed, for whom unemployment benefits is an insurance device, not a redistribution device, would lose from it because they are not prepared to take these jobs and their contributions would pay for them instead of their own income maintenance during unemployment spells.
Uncertainty about outcomes
Another reason why labor market reform may be blocked is that the gains from the reform may be unevenly distributed. First, they can be concentrated in the hands of a minority - for example skilled workers; in that case, however, there is scope for building consensus by redistributing income away from the minority of winners to the majority of losers. We return to that possibility below. Second, it may be the case that while the majority is expected to gain, people do not know whether they will actually be part of that majority (This argument is due to Fernandez and Rodrik, 1992).
Let us consider the following simple example: assume there are three groups in society, A, B, and C, each equally sized. Consider a reform such that group A loses 5 from the reform, group B gains 4 and group C gains 10. The aggregate gain from the reform is 10+4-5=9, so that efficiency requires that the reform be implemented. If everybody was certain about the group he will belong to after the reform, then those who anticipate to end up in groups B and C would support it, so that it will pass with a majority of two thirds. Consider now the case where people do not know which group they will belong to after the reform. More specifically, assume that members of group C know for sure they will remain in group C after the reform, while members of group A and B expect to end up in the other group with a 50 % probability. Then the expected gain from reform for both groups is (0.5* 4+0.5* (-5))=-0.5$. We see that reform is now repelled because groups A and B now lose from it on average.
We see that uncertainty then leads to a bias in favor of the status-quo when there is a well identified core of winners and a lot of mobility within the rest of society, which makes the remaining gains very diffuse and eventually lower than the losses.
This argument may be particularly relevant in the case of European labor markets: a large minority - skilled workers and the unemployed - know for sure that they will gain from a reform: the unemployed will find a job and, to the extent that there is some complementarity in production between skilled and unskilled labor, the skilled will enjoy higher wages and standards of living due to the increase in the number of unskilled workers who will be employed. The unskilled, however, are much less certain to gain because their wages might fall and they might be forced to relocate to other sectors. If all the unskilled knew whether they would gain or not from the reform, those who gain could form a coalition with the skilled and the unemployed in favor of the reform. But uncertainty may make the gains too diffuse to the unskilled, so that they might end up opposing reform.
Looking at the pattern of mobility across income groups confirms that argument: there is more mobility at the bottom of income distribution than at the top (See Quah (1994)), and we expect the top to be winners and the bottom to be losers. The richest are thus practically sure they will gain from the reform, while the rest of society will have much lower and uncertain gains.(4)
Extreme coalitions and multiple issues
The last reason why labor market reform is difficult to implement is that it is likely to benefit a very heterogenous coalition of capitalists, skilled workers, small entrepreneurs and unemployed workers. It is an ''extreme coalition'' which includes both the richest and the poorest people. These people have a common interest in increasing flexibility but diverging interests on almost all other issues including redistributive taxation, social policies and security policy. Because one votes on a package of policies rather than a single issue at a time, it is unlikely that these people will support the same political platform.
Strategies to implement labor market reform
Let us discuss some strategies to increase flexibility in the labor market in a ''politically viable'' way.
Side transfers
Since flexibility probably increases the efficiency of the economic system, there must exist a system of lump-sum transfers which, associated with liberalization, will make everybody better-off. This is of course at best a theoretical possibility. First these transfers are difficult to compute; second there will be considerable uncertainty about the appropriate level of these transfers(this uncertainty about the distribution of the gains from reform may lead to reform being ex-ante repelled, as we have seen above); third, and most importantly, the transfers that will prevail in equilibrium are themselves determined as the outcome of the political system. It is impossible for the government to commit over the future pace of taxes and transfers following reform, if anything because it is not certain to remain in power. So the people who lose from flexibility (meaning the price of the factors of production they supply will fall) may not get sufficient compensation through the political system, if they are weak. Or, more perversely, those who gained may fear to be taxed at too high a rate and find that rigidities are a less risky way to redistribute income. Take for example the case of the minimum wage. In a world where there are two types of labor, skilled and unskilled, the minimum wage can raise the unskilled wage while creating unskilled unemployment. At the same time the skilled wage falls because, to the extent that skilled and unskilled labor are complements, the marginal product of skilled labor falls since there are less unskilled employed. So the minimum wage redistributes income from skilled to unskilled labor, at the cost of creating unemployment. In the absence of taxes and transfers, eliminating the minimum wage would be favored by the skilled and the unemployed, and opposed by the unskilled employed. If taxes and transfers could be set in advance as part of a policy package, it would be possible to make everybody better-off by eliminating the minimum wage. But the most interesting case is when taxes and transfers are politically determined, through (say) majority voting. In that case it is reasonable to assume that in those three groups, the median voter will be unskilled employed. The tax rate will thus depend positively on the gap between the skilled and the unskilled wage. As a result, lifting the minimum wage regulation will automatically be compensated by an increase in redistributive taxation. This is somewhat a desirable outcome, because it tends to compensate the unskilled employed for the drop in their income, and thus contributes to building support in favor of reform among them.
However, the equilibrium tax rate may end up being so high that the skilled are actually those who suffer from the reform. This example illustrates the above argument that labor market rigidities may promote political stability by homogenizing the ''core'' of society (here the employed) at the expense of a politically unimportant ''periphery'' (the unemployed).
We have seen that by bringing back the unemployed into employment, labor market flexibility may make society more redistributive as a reaction to the implied widened income differences. Depending on how unemployment is distributed among income groups, however, one may also have a drop in the tax rate when the economy shifts to flexibility. This will be the case if unemployment is relatively evenly distributed across income groups, and if unemployment benefits are low. In that case rigidities drive the median voter's income down relative to the mean, so that the rigid economy is less conservative than the flexible one. This case is however less relevant to Europe than the other one, because unemployment benefits are relatively generous and unemployment is concentrated among the poorest.
To summarize, three cases may be distinguished:
1.Taxation may be insufficient so that not enough income is redistributed from winners to losers, implying reform may be blocked.
2.Taxation redistributes enough income to generate a wide enough consensus over the reform.
3.Taxation is overredistributive, so that a coalition of winners will actually block the reform.
Timing
Political viability may be achieved through an appropriate timing of the reforms. As we have discussed above, an important determinant of the political support for the reforms is the employed's exposure to unemployment. This exposure has more to do with the flow of job destruction than the level of unemployment. This suggests that reforming the labor market should be easier at times of large job destruction, that is to say during massive structural change or at the beginning of a sharp recession.(5) The optimal timing, however, heavily depends on which reform is considered; reducing firing costs through a two-tier system which maintains the incumbents' job security is more viable at times of large job destruction, but this is not true of an across-the-board reduction in firing costs (see the next sub-section).
Thus, it is when the economy undergoes large structural change, with both high job creation and high job destruction, that labor market flexibility is most likely to get political support. This ''virtue of bad times'' seems to have been exploited in Spain, where labor market reform was introduced in 1984 when Spain was in the middle of a period of massive job reallocation due to the inadequate production structure inherited from the Franco era.
Two-tier systems
An alternative strategy for making reform politically viable is to liberalize at the margin, for example by liberalizing new labor contracts while leaving existing ones untouched. Two tier systems have been used in many European countries to reduce firing costs. While the terms of existing contracts were left unaffected, it has been possible for firms to hire new people under so-called ''determined duration contracts''. These contracts are typically associated with a lower firing cost.
This strategy of creating a two-tier system has both advantages and drawbacks.
The advantages are as follows. On the economic side, the two tier system is almost as efficient, as far as raising labor demand is concerned, as a complete liberalization of the labor market. This is because in all firms who actually employ workers with flexible contracts, labor demand is determined by the "marginal cost of labor", that is to say the cost of the next worker to be hired or fired. It can be shown that this "marginal" worker is typically a worker under a flexible contract, so that the marginal cost of labor is the one associated with flexible contracts. Therefore - with the exception of those firms which do not employ "flexible workers" - labor demand is the same as if the whole of the labor market had been liberalized. On the political side, the two-tier system is much more likely to be supported by the employed than a one-tier liberalization. This is because they will enjoy the same degree of protection while benefitting from the higher rate of job creation in case they end up being unemployed.
Let us now turn to the drawbacks of two-tier systems.(6) First, if not properly designed, they can have perverse effects on wage formation; Bentolila and Dolado (1994) have studied the impact of temporary contracts on wage formation in Spain and shown that by increasing the permanent worker's job security, temporary contracts tended to have adverse effects on wages. This would not have mattered so much if the temporary wage was split from the permanent one, but the system was designed in such a way that temporary wages were linked to permanent wages, so that all wages increased. Second, it may be quite difficult, under these marginal schemes, to avoid crowding out, meaning that a large fraction of the jobs that are subsidized would have been offered by employers anyway. Similarly, employers might use temporary contracts for jobs they would have offered under permanent contracts if temporary contracts had not been available. Third, the two-tier system may intensify distributive conflicts within the workforce, thus increasing other forms of inefficiencies. For example ''secondary workers'' will be tempted to partly expropriate ''primary'' workers through redistributive taxation. This might particularly be a problem if temporary workers had little perspective to get a permanent job. Fourth, the two-tier system gradually affects the balance of power in society. As the stock of workers under temporary contracts develops, these workers will be more powerful politically and can be used by the government as political support for further reforms of the labor market. This is fine, but the incumbents and their unions will ex-ante recognize that their political influence will gradually be undermined by the increasing number of temporary workers, and for that reason oppose the change. That may lead to the reform being blocked, or to a different reform being passed. For example, in many European countries, the government managed to liberalize the use of temporary contracts only to a limited extent, since unions agreed to it provided restrictions were put to the use of these contracts. Such restrictions were designed so as to prevent the stock of these contracts from growing beyond a point were the political influence of the insiders would be threatened.
The Spanish experience, in that respect, is revealing (but similar issues arose in France, too). The government managed to liberalize the use of determined duration contracts in 1984, when the economy had experienced an unprecedented period of massive job destructions. In order to prevent the use of these contracts to be too widespread, the unions managed to restrict their renewal to two times, after which they had to be converted to rigid contracts. Nevertheless as much as 95 % of new hires were under flexible contracts, and the proportion of the workforce under DDC's quickly rose to 30 % in the early nineties. Meanwhile, the unions exerted repeated pressure to eliminate these contracts. But at the beginning of the nineties the government was in a situation where it could use these workers and the unemployment as a coalition to back further increases in labor market flexibility - even if consensus had to be reached to avoid a social explosion, it could strike a deal with the unions to trade the removal of temporary contracts against increases in flexibility. This is actually what happened as a reform of the labor market was engineered in 1994, but the reform turned out to be very timid, presumably because of the electoral fragility of the Gonzalez government at that time. Although temporary contracts were maintained for a while, their phasing out was agreed. The government has therefore lost an opportunity which may not be regained in the future.
Redistribution of property rights
We have seen above that nothing guarantees that a proper set of transfers will be implemented after the reform, and that this may lead to the reform being blocked. One way to get rid of this dilemma is to change the ownership structure of the economy so as to actually build these transfers in property rights prior to the reform . A simple example is when reform redistributes income from labor to capital (this issue may also arise with trade reform, or transition in Eastern Europe). Then a possible commitment device to ensure that appropriate transfers will actually be implemented is to redistribute capital so as to make its ownership more even. This may lead to the gains from reform being better shared and may therefore create a consensus over it. Schemes such as profit sharing should therefore be seriously considered as a means to increase the political feasibility of labor market flexibility, and more generally to build a consensus over free markets and the economic institutions of capitalism. This may sound like a very radical proposal, but we should remember that economic theory predicts that free markets make everybody better-off only if they are associated with appropriate transfers. In the absence of such transfers, large minorities or even the majority might lose, and there will be political opposition to capitalism.
Such redistribution of property rights is much more difficult to implement when flexibility is associated with transfers within the workforce, for example when it reduces wages for the unskilled and increases them for the skilled. One way to do it is to increase the educational level of the unskilled, a policy which is itself associated with many design problems and shortcomings.
The present paper suggests that properly taking into account political constraints considerably reduces the scope for labor market reform in Western Europe. High unemployment may therefore be something we'll have to live with for a long time. It makes it more important to carefully identify the margin of manoeuver for reform and to clearly spell out the society choices that lie behind this challenge.
REFERENCES
Fernandez, R. and D.Rodrik, ''Resistance to reform: status quo biasin the presence of uncertainty'', American Economic Review
Quah, D. (1994) ''One business cycle and one trend from many (many) disaggregates'', European Economic Review
Saint-Paul, Gilles (1993), ''On the Political Economy of labor market flexibility'', NBER Macroeconomics Annual
------------------------ (1994), ''Searching for the virtues of the European Model'', IMF Staff papers
NOTES
(1) See Saint-Paul (1994), for a formal argument.
(2) It is interesting to try to understand why this interpretation of rigidities does not apply to the United States. The American society is a pretty heterogenous one, yet it has not engaged in the sort of institutions which prevail in Europe. Why is that so ? There are three possible interpretations. One is that the U.S. is too heterogeneous, so that the periphery would be too large. If this was the only one, however, we would observe large distributive conflicts in the U.S., which is not the case. The second interpretation is that consensus is in fact achieved through mobility, meaning that society is in fact more homogenous than it looks like from a cross-sectional distribution of income. The third one, is that exclusion of the periphery is achieved politically but not economically; that is, measures are taken which prevent those political platforms which would suit the periphery from being available; for example, the communist party is illegal. This argument is consistent with the robust empirical evidence that the poor participate much less in elections than the rich. Since the ultimate goal is precisely such political exclusion, there is no need to introduce distortions in the labor market since it is achieved directly.
(3) For unemployment insurance that may seem paradoxical but realise that almost everywhere one has to have worked to be eligible, that the long-term unemployed and the new entrants have no benefits, so that the benefits are best fit for those with little exposure; furthermore, they also allow the employed to raise equilibrium wages
(4) Note, however, that the support for labor market flexibility among lower income groups will be enhanced by exposure to unemployment and the induced reduction in the tax rate needed to finance unemployment compensation.}
(5) The latter strategy is probably more problematic than the former, since a recession only lasts for a couple of years, and voters anticipate that job destruction will be low again in a delay which is short compared to the duration of the change. Furthermore, for the reform to get support it must be the case that voters expect it to have a substantial positive impact on job creation, which is unlikely in recessions.
(6) The following argument is analytically developed in Saint-Paul (1993).