Central bank independence, wage bargaining, and labor market performance: new evidence.

• Author: Feldmann, Horst

Using data on 20 industrial countries over the period 1982 to 2003, this article finds central bank independence to favorably affect both unemployment and employment rates. The size of these effects appears to be substantial, particularly in the long term. In contrast to some of the previous literature, the article finds that the favorable effects of central bank independence do not depend on the degree of wage bargaining centralization or coordination. Furthermore, it finds that higher centralization as well as higher coordination of wage bargaining may also have favorable direct effects on labor market performance.

JEL Classification: E24, E58, J50, J64

1. Introduction

   As Rogoff (1985) argues, an independent and conservative central bank (i.e., one with an anti-inflation bias) can overcome the inflation bias resulting from the time-consistency problem analyzed by Kykland and Prescott (1977) and Barro and Gordon (1983a, b) and can thus achieve low inflation at no unemployment cost. Empirical research corroborates this hypothesis, finding a robust negative correlation between central bank independence and inflation and generally no correlation between central bank independence and real variables (for surveys, see Eijffinger and de Haan 1996; Berger, de Haan, and Eijffinger 2001; Cukierman 2008).

   Unfortunately, this literature does not take into account the institutional structure of labor markets, wage bargaining institutions in particular. Until recently, the effects of wage bargaining institutions on labor market performance have been analyzed in a separate literature, which, in turn, abstracts from the institutional structure of monetary policy, specifically, from central bank independence. The work of Calmfors and Driffill (1988) has been most influential in the analysis of wage bargaining institutions. They hypothesize that unemployment will be comparatively high, and employment comparatively low, if wages are negotiated at the industry rather than at the firm or national level. If wages are negotiated at the firm level, wage increases will be constrained by competition in the product market. If they are negotiated at the national level, trade unions will internalize the negative externalities of excessive wage increases. Therefore, either low or high levels of wage bargaining centralization will lead to wage restraint and thus to favorable labor market outcomes. By contrast, wage bargaining at the industry level suffers from both the absence of competitive pressures and a lack of internalization of negative externalities, leading to excessive wage hikes and thus comparatively high unemployment and low employment. As Aidt and Tzannatos (2002, pp. 104-6) point out in their survey of the relevant literature, the evidence for the Calmfors and Driffill hypothesis is weak.

   More recently, several studies have taken the institutional features of both monetary policy and wage bargaining into account. Because of stark differences in assumptions and modeling strategies, their hypotheses differ substantially from each other. Some of them are diametrically opposed to each other. The most important hypotheses are the following. Iversen (1998a, b) argues that the best unemployment performance occurs when either wage bargaining is centralized and the monetary regime is accommodating or when wage bargaining is decentralized and the monetary regime is nonaccommodating. In the former case, low unemployment will be achieved by an effective coordination of wage, monetary, and employment policies. In the second case, the credibility of the central bank's commitment to low inflation will lead to moderate wage hikes. Soskice and Iversen (1998, 2000) hypothesize that, if unions are sufficiently large, a less accommodating central bank moderates unions' wage demands by raising the fear of unemployment among their members. Consequently, unemployment is lower and employment higher.

   According to Hall and Franzese (1998; Franzese and Hall 2000; Franzese 2001, 2002), the more coordinated the wage bargaining process, the more responsive wage setters will be to signals from an independent central bank, and thus the less likely it is that the central bank has to resort to restrictive monetary policies that would increase unemployment. In other words, Hall and Franzese argue that any unemployment cost of central bank independence decreases as the coordination of wage bargaining increases. Furthermore, they argue that the beneficial effects of centralized (or, equivalently, coordinated) wage bargaining on unemployment pointed out by Calmfors and Driffill (1988) should increase with increasing central bank independence. A more independent central bank, so the argument goes, can reinforce the coordination of wage bargaining and threaten more credibly that it would respond to excessive wage hikes by raising interest rates.

   Cukierman and Lippi (1999) hypothesize that a Calmfors-Driffill (1988) hump-shaped relationship between unemployment and wage bargaining centralization is more likely to arise when central bank independence is small and when unions are inflation averse. Furthermore, they hypothesize that in countries with highly independent central banks, decentralization of wage bargaining is likely to reduce real wages and thus unemployment. Finally, they argue that, by alleviating inflationary fears of unions, a more independent central bank may induce higher real wage demands, leading to higher unemployment. Integrating the different mechanisms developed in the previous literature in a unified theoretical framework, Coricelli, Cukierman, and Dalmazzo (2006) find that, for realistic values of the relative aversion of trade unions to unemployment and inflation, higher central bank independence lowers unemployment. Specifically, if the relative aversion of unions to unemployment versus inflation is sufficiently larger than this relative aversion at the central bank, higher central bank independence induces unions to moderate their wage claims, reducing unemployment. In an extension of this model designed to capture strategic interaction between fiscal and monetary policies in the presence of unionized labor markets, Cukierman and Dalmazzo (2006) find that higher central bank independence and higher centralization of wage bargaining lead to lower unemployment.

   As explained in the next section, the results from previous empirical studies are also mixed. Furthermore, their methodologies suffer from various shortcomings. Using data from 20 industrial countries, this article goes beyond previous empirical research in several respects. First, whereas previous articles study the effect of either centralization or coordination of wage bargaining, this article studies the effects of both. Second, whereas some previous studies fail to estimate the effect of interaction between central bank independence and wage bargaining centralization or coordination, this article estimates both the direct effect of central bank independence and the interaction effects. Third, whereas almost all previous studies control for the impact of only very few other factors that have been found to affect labor market performance, this article uses a large number of controls covering all major factors. Fourth, in contrast to almost all previous studies, we conduct many robustness checks. Fifth, whereas most previous studies' sample periods end around 1990, this article uses data from the years 1982 to 2003, thus covering important changes in central bank independence that occurred in more recent years. Finally, whereas previous studies estimate the effect on the unemployment rate only, this article additionally estimates the effect on the employment rate.

   The next section briefly surveys the literature empirically analyzing the labor market effects of both central bank independence and wage bargaining institutions. Section 3 describes the data set and methodology. Section 4 presents and discusses the regression results. Section 5 concludes.

2. Previous Empirical Research

   The first article to analyze the labor market effects of both central bank independence and wage bargaining institutions is Bleaney (1996). He uses data on 17 industrial countries, averaging the data for two separate periods: 1973-1982 and 1983-1989. For neither period does he find a statistically significant effect of central bank independence on unemployment. The estimates for a variable measuring wage bargaining centralization, trade union power, and the degree of employer organization suggest that a higher degree of corporatism, as measured by this variable, raises unemployment. Bleaney does not estimate the interaction between central bank independence and corporatism, nor does he employ any control variables.

   Using annual data on 10 industrial countries for the period 1973 to 1993, Iversen (1998a) finds both central bank independence and wage bargaining centralization to exert a favorable direct impact on unemployment. He also finds an adverse effect of interaction between central bank independence and wage bargaining centralization, suggesting that central bank independence increases unemployment at high degrees of wage bargaining centralization and that wage bargaining centralization increases unemployment at high degrees of central bank independence. These results support his theoretical argument. (1) Iversen controls for the average unemployment rate across OECD countries, the yearly growth in each country's export market, lagged unemployment, and Finland's economic crisis at the beginning of the 1990s. Additionally, he uses country dummies for Austria and Japan as he is unsure about these countries' degree of central bank independence.

   Covering 18 industrial countries and the period 1955 to 1990, Hall and Franzese (1998) estimate regressions using country averages, decade averages, and annual data. (2) They find central bank independence to have an adverse direct effect on...