Contenu du sommaire : Qu'a-t-on appris sur la relation salaire-emploi depuis Keynes?

The question of the wages-employment relationship stemmed from the statistical invalidation by Dunlop and Tarshis, immediately after the publication of the General Theory, of a conjecture formulated by Keynes on the alleged negative correlation between real and money wages. All three authors tried from the beginning to reconcile theoretical views and statistical observations by resorting, almost exclusively, to imperfect competition. Actually, these efforts culminated, at least in Dunlop's case, in a true research programme on macroeconomic analysis of imperfectly competitive economies, a programme which was basically implemented only in the eighties and the nineties. The present paper discusses this episode from the theoretical point of view and casts a look at its modern echoes, namely at the two main ways of reconciling the Keynesian conjecture with the stylized facts : increasing marginal returns due to internal economies and cyclical

New Classical Economics led to the collapse of the famous « neoclassical synthesis » and to an agreement around the idea that microeconomics foundations to macroeconomics were necessary. Since the central element of the neoclassical synthesis was the assumption of nominal wage rigidity, neokeynesian macroeconomics tried to find a correct description of microeconomics appropriate to keynesian results. This paper shows how this attempt failed : neo-keynesians tried to find microeconomics foundations to nominal wage rigidity, but found only explanations for real rigidity, and then exhibited an equilibrium unemployment with classical characteristics, very far from Keynes' approach.

The research program called « Economics of Conventions » aims at revisiting the century- long separation, within mainstream economics, between the issues of coordination, rationality and ethics. In the realistic world of bounded rationality, the rules are a fundamental means of coordination. An important subset of rules consists in conventions (of which a restrictive definition was provided by the american philosopher David Lewis), but every kind of rule is generally uncomplete, so that it is necessary to resort to collective representations associated with their expected correct application. Since the agreement upon these representations can be viewed as a conventional phenomenon, conventions in economics become quite comprehensive. From a microeconomic point of view, wages are the results of rules. Does that mean that they are conventional rules ? Many authors have exploited the idea that the amount of wages was determined by a convention. This idea is confronted by insuperable obstacles, suggesting the relation with conventions should be looked for at another level. Any firm, whatever its model of management, rests on a quasi social-contract, linking the dynamics of wages and labour productivity : the wage rules, empirically very diverse, are interpreted through this conventional pact. We define three pure models of business firms (merchant, industial, domestic), or three « dynamic conventional equilibria » . They correspond to three profiles of the employment/wage schedule, decreasing, flat, increasing. From a macroeconomic point of view, the diversity of the employment/wage schedule, at the level of firms, explains the fact that the trend of real wages makes very weak forecasts of the trend of employment -unlike the trend of effective demand. The predictive superiority of this last index comes from the betting system on which all the firms are built, together with the identity which binds the change of the macroeconomic variables. The concept of dynamic macro-equilibrium appropriate to the « Economics of Conventions » splits in two pieces : in the short run, the constancy of unemployment and interest rates, whatever their value, appeal to the neo-keynesian thesis of possible multiple equilibria ; in the long run, we define a « rule-equilibrium », when applying the received rules for modifying transaction confines the sequence of the « short-run dynamic macro-equilibria » within a « corridor », which does not erode the collective agreement upon « normal conjoncture », especially with respect to the goal of full employment.

Contemporary research is built upon a strong substituability between labour and capital in response to the signals of relative prices, specially of real wage. "Régulation theory" takes into account the rough complementarity of production factors, as evidenced by econometric studies, and proposes the notions of wage labour nexus in order to capture the density of institutions and coordinating mechanisms, other than prices, which govern labour adjustments. Thus as many wage/employment elasticities are observed as institutional architectures. It was positive within the Fordist growth regime, but it may be negative for small open economies under strong external competition. The paper provides a panorama of the main results obtained by long run historical studies, international comparisons, as well

The research program called « Economics of Conventions » aims at revisiting the century- long separation, within mainstream economics, between the issues of coordination, rationality and ethics. In the realistic world of bounded rationality, the rules are a fundamental means of coordination. An important subset of rules consists in conventions (of which a restrictive definition was provided by the american philosopher David Lewis), but every kind of rule is generally uncomplete, so that it is necessary to resort to collective representations associated with their expected correct application. Since the agreement upon these representations can be viewed as a conventional phenomenon, conventions in economics become quite comprehensive. From a microeconomic point of view, wages are the results of rules. Does that mean that they are conventional rules ? Many authors have exploited the idea that the amount of wages was determined by a convention. This idea is confronted by insuperable obstacles, suggesting the relation with conventions should be looked for at another level. Any firm, whatever its model of management, rests on a quasi social-contract, linking the dynamics of wages and labour productivity : the wage rules, empirically very diverse, are interpreted through this conventional pact. We define three pure models of business firms (merchant, industial, domestic), or three « dynamic conventional equilibria » . They correspond to three profiles of the employment/wage schedule, decreasing, flat, increasing. From a macroeconomic point of view, the diversity of the employment/wage schedule, at the level of firms, explains the fact that the trend of real wages makes very weak forecasts of the trend of employment -unlike the trend of effective demand. The predictive superiority of this last index comes from the betting system on which all the firms are built, together with the identity which binds the change of the macroeconomic variables. The concept of dynamic macro-equilibrium appropriate to the « Economics of Conventions » splits in two pieces : in the short run, the constancy of unemployment and interest rates, whatever their value, appeal to the neo-keynesian thesis of possible multiple equilibria ; in the long run, we define a « rule-equilibrium », when applying the received rules for modifying transaction confines the sequence of the « short-run dynamic macro-equilibria » within a « corridor », which does not erode the collective agreement upon « normal conjoncture », especially with respect to the goal of full employment.

Real Business Cycle models try to explain the labor market dynamics in a walrasian framework. The volatility of employment is explained by an intertemporal substitution of leisure in the cycle. However, these models suffer from several imperfections, mainly the too high level of the wage-employment correlation and the too weak volatility of employment relatively to wage. Extensions in a walrasian framework, exploring the labor supply behavior or including other sources of impulsions, do not really succeed in improving the ability of this kind of models to replicate the observed fluctuations. Only the introduction of real and nominal rigidities allows to give a better explanation of the labor market dynamics, even if the intertemporal choice framework is always in the lines of the RBC theory methodology.

This historical review is intended to evaluate the econometrics contribution to the analysis of the relationship wage/employment. The first part is mainly devoted to the macroeconomic modeling "as it was in the old days". The second part is concerned with the developments of the "new" econometrics dealing with the relationship wage/employment. Or more precisely, it is mainly devoted to the four following points : the improvements in dynamic specifications concerning labour demand, the modeling in terms on prices and wages stting (models socalled WS/PS), the studies using microeconomic data, and lastly the analyses which distinguish the different qualifications of the labour force. On the other hand, we do not mention the developments relative to the relationship between the labour supply and the wage levels. However, these developments constitute an important part of econometrics applied to labour economics. The conclusion sums up the historical overview in five points.

A large number of labour market policy programmes rely on the reduction of labour costs. Therefore the link between wage/labour costs on the one hand, and employment on the other hand, can be analyzed through the results stemming from active labour market policy evaluation. On the ground of a survey of existing studies, this contribution presents successively micro- and macroeconomic impact studies. It focuses on the methodological debates in this field, mat are necessary to the understanding and interpretation of the results. The main conclusion would be that reducing labour costs through various types of subsidies might be efficient in terms of job creations, but in a limited proportion, the extent of which is dependent on the content of the programme (pure job subsidy, or comprehensive programme including training ; targeting). At the macro level, one can state, on the basis of econometric studies, that active labour market policies considered as a whole have a wage