چگونه چارچوب مجموع عرضه-مجموع تقاضا نوسانات بیکاری را بخوبی توضیح میدهد؟ مقایسه فرانسه-ایالات متحده

This paper reviews the ability of the traditional aggregate demand–aggregate supply framework to explain the unemployment fluctuations of the last three decades. A structural VAR model for the growth rates of labor productivity, inflation and unemployment is estimated on American and French data. By using long-run identifying restrictions, unemployment fluctuations are associated with conventional aggregate demand and aggregate supply shocks and with a supplementary residual innovation. One key finding is that the residual shock is far more significant in France than in the United States. The traditional macroeconomic synthesis proves then to be well suited for the American labor market while it leaves unexplained a large part of the French unemployment drift. This result questions the conventional prior that the heterogeneity in unemployment experiences lies in the magnitude of aggregate shocks or in their propagation mechanisms and calls for alternative explanations.

Explaining the upward trend in the French unemployment rate while the American unemployment rate has remained partially trendless over the last three decades is an ongoing challenge for macroeconomic theory. In particular this development has cast doubt about the ability of the traditional aggregate demand–aggregate supply (AD–AS) framework to be a common benchmark for explaining unemployment fluctuations. According to this macroeconomic synthesis of conventional classical and Keynesian arguments, unemployment fluctuations are mainly driven by aggregate demand and aggregate supply shocks in combination with some inertia in the adjustment process. While demand perturbations may have real effects in the context of nominal rigidities such as Taylor (1979) fixed-length wage contracts, supply shocks are likely to affect the unemployment rate as long as real wages do not adjust to clear the labor market. Actually such an interpretation seems well suited for explaining the initial rise in American and French unemployment rates during the 1970s and the early 1980s. Both countries suffered a protracted period of rising unemployment (from 2 to 8%) while they were successively hit by large adverse supply and demand shocks such as a total factor productivity slowdown and a stringent tightening of the monetary policy. However, such aggregate perturbations are expected to have only short-run effects on unemployment. Once nominal wages and real wages adjust to the new inflation rate or to the lower underlying level of labor productivity, the effects of aggregate shocks should vanish. This is precisely what occurred in the United States where the unemployment rate started decreasing by mid-1985. So why has the French unemployment rate kept increasing since then? This puzzle has fostered two classes of theories. The first one keeps the AD–AS framework as a benchmark reference. But one holds that the effects of aggregate shocks are longer in France because of more sluggish adjustments in the labor market. This argument refers to the hysteresis theory of Blanchard and Summers (1986) who claim that aggregate shocks are more persistent in European countries than in the United States because of insiders–outsiders effects. Moreover, France may have been affected by supplementary aggregate shocks from which the United States have been preserved such as the sharp increase in interest rates in the late 1980s (Fitoussi and Phelps, 1988). An alternative class of explanations leaves the reference to aggregate perturbations and focuses on more specific labor market shocks. In particular, a recent macrodynamic framework based on job-search models (Pissarides, 1990) and wage setting–price setting models (WS–PS) gives micro-foundations to unemployment fluctuations. Job-search models relate unemployment to imperfections in the employment setting process due to transaction costs and potential mismatch. This theory seems well fitted to account for the contribution of unskilled workers to the persistence of a high structural unemployment in Europe. Rather, WS–PS models focus on microeconomic imperfections in the wage setting process. Wages are set as a mark-up over the reservation wage and depend on key parameters such as the replacement rate of unemployment benefits, the minimum wage or the union bargaining power. In that context changes in the institutional setting of employment and wages lead to a permanent drift in the unemployment rate. In that perspective, Nickell (1997) claims that the labor market characteristics of European countries have undergone major reforms during the last three decades while they remained quite stable in the United States. This article provides an econometric contribution to this debate by evaluating the quantitative share of unemployment fluctuations left unexplained by adverse aggregate shocks. The purpose is to weight the contribution of the long list of alternative explanations rather than giving them precise economic interpretation. To this end, I follow a structural VAR (SVAR, henceforth) approach based on the seminal work of Blanchard and Quah (1989) who used long-run restrictions to identify reduced form innovations as structural shocks. However, while the two authors only focused on the contribution of aggregate demand and aggregate supply shocks to unemployment fluctuations, I introduce a residual orthogonal explanation by enlarging the size of the VAR. I take the unemployment growth rate and two key variables which are the most likely to capture the main supply and demand perturbations of the period, namely the growth rate of labor productivity and the growth rate of inflation. In order to identify the three structural shocks, I use the following set of long-run restrictions. Aggregate supply shocks are considered as innovations which may affect all three variables in the long-run. Aggregate nominal demand shocks may have permanent effects on the inflation rate and on the unemployment rate but do not affect the level of labor productivity in the long-run. The residual orthogonal shock captures innovations which may only have long-lasting effects on unemployment. Furthermore, by allowing both technological and nominal demand shocks to have permanent effects on the unemployment rate, I can test the empirical relevance of the hysteresis hypothesis. Several studies have already extended the number of potential unemployment perturbations by enlarging the original size of the Blanchard and Quah (1989) system. However, they generally keep the aggregate demand–aggregate supply framework as their theoretical benchmark. For instance Blanchard, 1989 and Dolado and Jimeno, 1997 decomposed demand shocks into price push and wage push while Gamber and Joutz (1993) disentangle aggregate supply perturbations into labor demand and labor supply shocks. Furthermore, they choose a set of variables and restrictions generally grounded on traditional theories of the wage setting process such as Fisher-Taylor wage contracts or insiders–outsiders wage bargaining à la Blanchard and Summers (1986). The modelling strategy of this article generalizes the previous works in several ways. First the residual orthogonal shock not only captures specific labor demand and labor supply shocks but also institutional perturbations which have never been analyzed in the context of a SVAR model. Second the identifying strategy is consistent with a broader class of wage-setting process. It may encompass traditional wage contracts as well as more recent wage curves in the lines of Blanchflower and Oswald (1995) or Layard et al. (1991). In that regard, the modelling strategy of this article is much more pragmatic. The propagation mechanisms are let free to decide which theory of labor market fluctuations is best suited for each country rather than imposing them a common benchmark. Applying the previous SVAR approach to American and French data for the period 1970–1998, I find that the residual shock does not explain much of the American unemployment fluctuations while it accounts for the main part of the French ones. This result questions the traditional prior that the divergence in unemployment experiences lies in the magnitude of aggregate shocks or in their propagation mechanisms. The heterogeneity is rather linked to differences in labor market characteristics since the two unemployment rates do not react to the same kind of perturbations. In particular, aggregate nominal shocks have little effects on the French unemployment rate, suggesting that the wage setting process is closer to a wage curve in this country. In that context, technological shocks or nominal demand perturbations play a minor role in unemployment fluctuations. By contrast, the large contribution of nominal shocks to the American unemployment variations is consistent with numerous econometric studies stressing the existence of a Phillips curve in this country. More generally these findings question the ability of the AD–AS framework to account for French unemployment fluctuations. I then suggest alternative explanations of the French unemployment drift by analyzing the contribution of the residual shock to unemployment history. I find that the evolutions of mismatch imperfections and unemployment benefits display the most significant correlations with this residual component of French unemployment history. The rest of the paper is organized as follows. In Section 2, I provide a brief overview of the main salient features of American and French labor markets in order to motivate the modelling strategy. Section 3 is devoted to the presentation of the structural VAR approach and to the identifying restrictions of shocks. In Section 4, I estimate the SVAR model and I quantify the effects of each shock through impulse–response analysis and forecast error-variance decomposition. Section 5 suggests economic interpretations of the residual shock based on an historical decomposition of unemployment fluctuations. Finally, Section 6 concludes.

This paper has reconsidered the relative contribution of aggregate demand and aggregate supply shocks to unemployment fluctuations. To this end a structural VAR containing the growth rates of labor productivity, inflation and unemployment has been estimated on American and French data over the sample 1970:1–1998:4. By using long-run identifying restrictions, the unemployment rate has been allowed to be permanently affected by a supplementary residual shock which captures the unemployment fluctuations left unexplained by the standard aggregate demand–aggregate supply framework. The main results of the paper can be summarized as follows. I find that while the residual shock is not significant in the United States, it drives the main part of the French unemployment fluctuations at all horizons. This result suggests that the aggregate demand–aggregate supply framework is well suited for the United Sates but poorly explains the French labor market. Besides, it reveals that the trend in the French unemployment is not brought about by more persistent effects of supply and demand shocks in this country. This finding is opposed to the Blanchard and Summers (1986) seminal prior that European labor markets display more hysteresis features than the United States. More generally it challenges the conventional wisdom that the heterogeneity in unemployment experiences lies in the magnitude of the aggregate shocks or in differences in their propagation mechanisms. Second, the paper reveals the difference in the employment setting and the wage setting process across countries by stressing the very low impact of nominal demand shocks in France compared to the United States. This result provides further support to the existence of a Phillips curve in the United States while the wage-setting process seems better matched by a wage curve in France. Finally, this article shows that the main contribution of the residual shock to French unemployment history takes place at a time of buoyant labor market reforms during the eighties. These findings call for future research focusing on direct shocks to labor market characteristics.