In the workshop on `Income Inequality Redistribution and Institutions' researchers reviewed recent theoretical and empirical work to provide an explanation for the shifts in income inequality, the distribution of wages and unemployment. The workshop was organized in Vigo by Thomas Piketty (CEPREMAP and CEPR) and Gilles Saint-Paul (DELTA, Paris). Financial assistance and organization was provided by the Consorcio de la Zona Franca de Vigo.
In `Demand for Skilled Labour, Technical Changes and International Trade: the French Case 1970-93' Dominique Goux (INSEE, Paris) and Eric Maurin (INSEE, Paris) test different hypotheses that might explain the evolution of the French wage structure between 1970 and 1993. Using labour force surveys, they show that over that period the French labour market experienced an increase in the average skill level as well as a massive rise in unskilled unemployment. This evolution was accompanied by a marked decline in the wage premium for skills (defined by either education or work experience). These changes can be read as a price adjustment in response to an increase in the supply for skills, without necessarily resorting to any demand side story. Changes in the composition of the domestic demand led to an increase in the relative demand for skilled labour, while international trade and technical change seem to have played a minor role in increase in the demand for skilled labour. A general discussion underlined the problems relating to the increase in the French minimum wage over the same period. Juan Dolado (Bank of Spain) suggested that it might have had an important role in the wage compression which happened in France, while Gilles Saint-Paul pointed to the complex problems associated with estimating a constrained demand and supply simultaneous equations model.
Elaborating on the results documented in Maurin's paper, Francois Bourguignon (DELTA, Paris) and Michael Martinez (?) focused on the consequences of these labour market changes for the inequality in French family income. In `Decomposition of Changes in the Distribution of Family Incomes in France', they look at a standardized measure of family primary income (exclusive of taxes and transfers) which corrects for differences in family size. A given family income is determined both by the socio-demographic characteristics of the household and by the labour market outcome of each of its members. Changes in the distribution of family income can thus result from a change in the distribution of family characteristics among all families and/or from a change in the way the labour market rewards family members having given characteristics. The date shows that during 1979 and 1984, only a slight increase in family income inequality is noted. This reflects two opposite tendencies: on the one hand a massive rise in income inequality due to a decrease in wage differentials among employed individuals compensated by a rise in unemployment which mostly hurt low income families, and on the other hand a change in the socio-demographic composition of households (i.e. changes in family size). An ongoing rise in female participation, from 1984-89, had little impact on overall inequality since it was quite evenly spread among all income groups.
In `Cyclical Fluctuations in Relative Wages in a Model of Comparative Advantage' Coen Teulings (University of Amsterdam) develops a model where the labour market assigns workers to firms. Workers are characterised by different skill-levels and firms can be ranked according to the complexity of the good they produce. Highly skilled workers have both an absolute and comparative advantage in more complex firms. Taking other firms' wage policy as given, each firm determines its wage to influence the skill level of the workers it will hire. In equilibrium this results in the most skilled workers being allocated to the most complex firms, since those firms have the highest incentive to hire very skilled workers. In case of an aggregate shock all firms should adjust their nominal wage offer proportionally. However if the shock is perceived as an idiosyncratic one, no firm will find it profitable to adjust its wage offer, at first. Each firm will keep the same (optimal) quality of workers but will hire less of them. This disequilibrium will results, in a second stage, in firms now hiring workers more skilled than what is optimal to them (this effect amplifies as we move down the skill/complexity ladder) and in unemployment accumulating at the bottom of the skill distribution. This process changes the supply for skill faced by each firm, making it profitable to adjust wages downward. The larger the discrepancy between optimal quality of workers and the quality of workers actually hired in disequilibrium, the stronger the incentive for wage adjustment. Since disequilibrium accumulates at the bottom of the skill distribution, this model explains why the wage premium of skilled workers should move counter cyclically. Thomas Piketty noted that it is hard to justify the specific form of the adjustment process adopted in the paper.
In two connected papers, Laurence Rioux (DELTA, Paris) and Daron Acemoglu (MIT and CEPR) examine the consequences on labour market outcome, of an increase in skill discrepancies. The labour market is analysed within a search-theoretic framework and both models share the feature of having two kinds of labour input: skilled and unskilled workers. Both inputs can be used in any job but skilled workers are more productive. The labour market outcome is shown to depend on two parameters: the productivity gap between skilled and unskilled and the proportion of unskilled workers in total labour force.
In `Heterogeneity, Matching and Endogenous Labour Market Segmentation', Laurence Rioux (DELTA, Paris) develops a two-sided search model where unemployed workers meet with other unemployed and decide whether to match together. If they do, the match is infinitely lived and workers share the total output. Potential matches can be homogeneous or heterogeneous. Due to high complementarity in the production function, being matched with a skilled worker always yields a higher payoff for each type of agent. When a skilled worker meets with an unskilled s/he can either accept to engage in a match and get a low payoff or reject the match and wait to meet a skilled. In the latter case, s/he incurs a search cost but might obtain a higher payoff when matched. The fact that it sometimes pays off to wait, results in segregation between skilled and unskilled.
An increase in the proportion of skilled workers in the economy (which reduces the search cost when deciding to segregate) and/or an increase in the skill gap (which increases the payoff advantage of being matched with a skilled) makes a segregated outcome more likely. The consequences are higher (duration of) unemployment for both types of workers; and an improvement (fall) in the utility of the high (low)-skilled workers.
In `Changes in Unemployment and Wage Inequality: An Alternative Theory and Some Evidence' Daron Acemoglu (MIT and CEPR) focuses on the determinants of labour demand and shows that a widening skill discrepancy might lead firms to create separate jobs for skilled and unskilled workers. Each firm is made of a single worker and some physical capital. In the model, firms choose a level of capital and then post a vacancy. Once a worker is hired the match is infinitely lived. Since skilled workers are more productive, the optimal allocation would give a higher level of capital to high-skilled workers. Nevertheless, since both skilled and unskilled workers are likely to apply to any vacancy and since the cost of a vacancy is proportional on the level of capital in the firm, it might be optimal, under uncertainty, to decide ex ante not to discriminate. This is the case if the productivity gap is not too large and/or if high-skilled workers are relatively scarce. In this no-discrimination equilibrium, the level of capital on all jobs is the same and is greater than the amount allocated to low-skilled in the Walrasian equilibrium. The uncertainty inherent to the matching process narrows the earning differentials. If the proportion of skilled workers in the economy and/or the skill gap increase, it might become optimal to create separate jobs for each type of worker. Hence, an increase in the relative supply of skilled workers might trigger an adjustment in labour demand that overshoots the competitive tendency to a fall in relative wages. This rise in segregation might empirically explain the increase in wage differentials that happened in the US since the mid 1970's. Furthermore, the paper gives evidence that this explanation is consistent with the fact that firms' recruitment strategy became more choosy during the 80's.
David Cutler (Harvard University) and Edward Glaeser (Harvard University) question the influence of segregation (by race or skill) on individual outcome in `Are Ghettos Good or Bad ?'. In theory, segregation can have a positive or a negative impact. Segregation is undesirable because of negative spillovers associated with ghettos (lack of exposure to education, informational separation). The positive effect rests on the idea that segregating people with different tastes might make it locally profitable to produce goods which more closely match their tastes. Cutler and Glaeser focus on the impact of Black versus Non Black segregation. They regress individual outcome on a measure of segregation in the city. As a measure of individual outcome, they take the probability to graduate from high-school, the probability to be unemployed or out of the labour force, the hourly wage rate and, for women, the probability to be a single mother. For each measure of outcome, the first set of regressions shows a negative correlation between the degree of segregation in a city and the outcome of black people. On the other hand, there doesn't seem to be any impact of segregation on the outcome of whites.
Raquel Fernandez (New York University and CEPR) and Richard Rogerson (University of Minnesota) examine the effect of the education finance reform that took place in California in the 70's. At that time, California moved from a system of mixed local and state financing to one of effectively pure state finance and saw a fall in educational spending. The authors develop a political economy model that accounts for this evolution. In the model, communities with different levels of income (and consequently different preferred levels of educational spending) vote for a level of state funding. Under the pure state finance system, this results in a unique level of spending that is financed state-wide through a proportional tax on income. This tax has redistributionnal consequences which lead communities above mean income to vote for a tax rate below their optimal level of spending. Under the mixed system, communities still vote for a level of state-funding but are able to augment the state level of funding through a local tax. This partly enables richer communities to avoid there distributional consequences of implementing a high level of education spending. Thus, the model predicts that shifting from the latter scheme to the former would lead to lower average educational spending and lower inequality in communities spending. Hence, the reform unambiguously leads to lower education spending for richer communities, but the conclusion is less clear cut for low income communities. The authors then calibrate their model to assess the impact of the reform on the bottom end of the distribution and show that it lead to significant gains in the level of spending for a very small fraction of the population and to important losses for a substantial fraction.
The paper by Daniele Checchi (State University,Milano), Andrea Ichino (IGIER, Milano) and Aldo Rustichini (CORE, Louvain-la-Neuve) `More Equal but Less Mobile? Intergenerational Inequality in Italy and the US' is a comparative analysis of intergenerational social mobility in Italy and the Unites States. Using different indicators of mobility, one can show that even if the probability of moving up or down the social hierarchy is the same in each country, the changes in earnings associated with mobility are much higher in the United States than in Italy. To explain those differences the authors offer a generation model of educational investment. The paper shows that state financing leads to too few resources being allocated to talented children. This prevents gifted individuals to move far away from the others in terms of human capital level, which results in a more concentrated wage distribution and reduces the incentive for investment in human capital. Daniel Cohen (Université de Paris 1, ENS and CEPR) questioned the empirical measures of mobility used in the paper. Thierry Verdier noted that a state finance education system could partially foster mobility by providing education to talented children whose parents wouldn't have invested in human capital due to resource constraint.