Comments
[The Cost of Job Security Regulation: Evidence from Latin American Labor Markets]
Alejandra Cox Edwards: This paper aims at measuring the impact of labor market regulations on various indicators of labor market performance. To accomplish this objective, the paper gathers and systematically organizes a substantial amount of information on labor policies in Latin America and the Caribbean. The paper focuses primarily on job security regulations, which is thus also the focus of my comments.
The main challenges that researchers face in the economic analysis of labor market regulations are a lack of sufficient variation in labor policies within countries for testing hypotheses based on time series and a lack of comparable measures of labor market interventions across countries for testing hypotheses based on cross sections.
In the specific case of job security regulation, the authors' strategy is to develop a measure of firing costs that summarizes the entire profile of tenure and severance pay, using a common set of dismissal probabilities. The indicator of firing costs (the job security index) is presented in table 1, and the numbers are interpreted as follows: if the cost of job security is three monthly wages in Argentina, for example, then the expected cost of labor in that country is whatever wage is paid plus an up-front investment of three monthly wages on average. The key elements included in this measure are advance notice requirements and compensation for dismissal with and without cause.
This measure is an important contribution to the literature. It may prove to be as important for informing policymakers on the effect of labor interventions as was the concept of effective protection for informing policymakers on the true impact of tariffs. However, if the measure is to be taken seriously, it needs to incorporate changes in at least three areas.
First, the authors must find a way to distinguish the case of Mexico from the case of Chile. These two countries both appear with a number close to 3 in table 1. Both countries have established minimum severance and advance notice regulations, which results in the similar estimates, but [End Page 145] the two systems differ in key ways. Mexico places no real limits on short-term contracts, meaning that employers who enter into long-term contracts choose do so. In Chile, on the other hand, any short-term contract that is renewed becomes de facto a permanent contract. When long-term contracts are truly an option, as in Mexico, they are more likely to internalize the cost of severance.
Second, the calculation rests on some measure of turnover, and Heckman and Pagés assume the turnover rate of the United States in all cases. Although they may have legitimate reasons for preferring a unique measure of turnover, I question the appropriateness of using the same measure for all countries. My own work indicates that turnover varies significantly by sector. In particular, it is much higher in services than in manufacturing or the public sector. My question is this: should the estimated effect of job security on the cost of labor incorporate the sectoral distribution of employment in each country? I think it should.
Finally, the authors themselves question the validity of their job security measure given that severance payments are transfers from employers to workers and as such are, in part, a form of compensation. They recognize that "the impact of job security could be completely 'undone' with a properly designed labor contract provided that there are no restrictions on transactions between workers and firms." Lazear, who pointed in this direction with his 1990 paper, stated that the impact of job security could only be undone if dismissal costs were paid in all separation cases and if payments took the form of lump-sum or deferred payments. Lazear shows that under those conditions, severance payments have a well-defined counterpart in current salaries, and the contract can fully internalize the severance. 1 However, as Heckman and Pagés indicate in this paper, most Latin American and Caribbean countries have established severance payments that (a) are a multiple of the last salary, or the salary at the time of dismissal; (b) depend on the existence of just cause; and (c) do not apply in the case of voluntary quits. Therefore, the capacity to internalize the cost of severance is generally low. Some countries have moved closer to the Lazear conditions or are considering reforms along those lines. In particular, some countries have established severance in all separations (for example, Bolivia and Brazil) or severance based on individual accounts. If the job security measure could distinguish cases in which severance is [End Page 146] partially or fully internalized from cases in which severance cannot be internalized, then the truly distortionary component of mandated severance could be isolated, and the Lazear effect could be tested empirically.
The section on empirical evidence summarizes the results from a series of country studies. Based on these studies, the authors argue that job security reduces turnover, reduces employment, and changes the composition of employment to the detriment of youth employment. The authors build a cross-section time-series data set to test these effects. This section needs further work both in defining the variables used and in justifying the empirical specification. The data include Latin America and the Caribbean, European countries, and the United States. The discussion highlights similar results on employment level and composition in all three regions.
The paper also presents a series of equations that estimates the effects on unemployment rates. Here, the empirical specification should probably incorporate the differences in unemployment insurance systems across these three regions. Latin American and Caribbean countries typically do not have unemployment insurance systems, European countries have unemployment insurance systems financed by a flat payroll tax rate, and the United States has an unemployment insurance system financed by an experience-rated payroll tax. The incentives generated by the unemployment insurance system are likely to encourage longer unemployment spells and higher unemployment rates in Europe than in the United States, other things being constant. The lack of unemployment insurance systems in Latin America and the Caribbean may imply a lower unemployment rate.
To close, I congratulate Carmen Pagés and Jim Heckman for persevering on the difficult path of producing empirical evidence on the impact of job security regulations. Their work represents a step forward on at least three fronts: developing a measure of the cost of job security regulation; showing that job security regulations reduce turnover; and demonstrating that the negative (undesired) effect of job security is the reduction of labor demand. The evidence that job security tends to reduce turnover and employment at the same time suggests a possible link between these two variables. Some of the literature on labor market dynamics already points in this direction, and the paper should incorporate that evidence.
Pablo Guidotti: This stimulating paper surveys existing sources of information and provides new evidence on the costs of a specific set of labor policies. Specifically, it tries to determine how job security provisions in [End Page 147] labor legislation affect employment. The paper finds that contrary to what has been argued recently in the literature, labor market regulation designed to improve job security brings about significant costs in terms of employment, although the effect of such provisions on the rate of unemployment is rather weak.
To put this paper into perspective, it is useful to review briefly the main challenges facing labor markets in the major Latin American economies. I focus on Argentina, which at present has the highest unemployment rate in the region. Furthermore, Argentina pursues an exchange rate regime (namely, convertibility) that requires labor market flexibility to improve the economy's ability to respond to real external shocks.
In the 1990s the debate on labor markets covered four main areas. The first area relates to the role of unions in the process of labor contract negotiation. The central question here is how efficiency and labor flexibility in the economy are related to the prevailing type of collective bargaining, in particular whether it is centralized at the sectoral or economy-wide level or whether it is decentralized at the firm level. Although it is generally agreed that decentralization favors efficiency, in practice sectors featuring centralized collective bargaining agreements have sometimes introduced significant flexibility in their labor contracts. In the Argentine context, employers' representatives have voiced a wide spectrum of opinions on the convenience of decentralized negotiations; their views on this issue appear to be strongly influenced by the relative attitude that central unions and individual firms' union representatives adopt during negotiations. 1
The second area in the debate refers to the political power of unions, in particular their ability to push through legislation that imposes mandatory contributions (taxes) on firms to finance the unions' bureaucracy or the monopolistic provision of health services (obras sociales). The third area, which is strongly linked to the second, encompasses tax policy and, more specifically, to the level of social security contributions levied on firms. [End Page 148]
Finally, the fourth area refers specifically to labor legislation, focusing on the provisions in labor contracts that affect a firm's ability to hire and dismiss workers or that raise the cost of doing so. Severance payments, advance notification, and seniority provisions are central elements in this part of the debate. The discussion of the effects and costs of labor provisions is not independent from the consideration of legal risks for firms in countries where the courts have produced significant jurisprudence favoring workers in the litigation process.
Within this debate, Heckman and Pagés focus essentially on the last set of issues, and this is where they produce their main contributions. The paper briefly covers the issues of collective bargaining and labor taxes, but the analysis of these two topics is too brief to obtain significant insights into the debate. Also, I disagree with the paper's classification of payroll contributions which finance a pay-as-you-go system (which are equivalent to taxes) together with personal contributions to private pension funds (which are more like savings).
Turning to the main part of the paper, the authors' first contribution is to provide a comprehensive survey of job security provisions in Latin American and Caribbean countries, focusing on their general legal framework. The paper provides valuable information on whether labor contracts are regulated by specific labor codes or whether they are subject to common law, and it examines the major job security provisions such as advance notification, dismissal compensation (whether with just or unjust cause), and the existence of a seniority premium.
The authors then construct a job security index based on these elements. The index maps regulatory provisions into a cost measure as a percentage of wages. In this way, the authors claim, one can move from a qualitative to a quantitative analysis of the impact of these labor regulations on employment.
I find this section of the paper and the job security index useful. However, inspection of table 1 leaves me with some doubts about the implied ordering of different countries in their sample. It is somewhat surprising to find most European countries (which are famous for their inflexible labor legislation) ranked as having less labor market regulation than Latin American and Caribbean countries. Within Latin America it is quite surprising that Chile has one of the highest values of the job security index. Given the importance of the rankings presented in table 1 for the empirical analysis, [End Page 149] it would be useful for the reader to have a better idea of the sensitivity of the index to various assumptions.
In particular, because the index is constructed exclusively using provisions existing in labor codes, it generates a bias in favor of common law countries where effective job security provisions emanate directly from private contracts and the legal jurisprudence applied by courts. In some countries with fairly rigid labor legislation, de-facto labor flexibility has been brought about through collective agreements. Thus an excessively "legalistic" approach may miss some of the actual characteristics of labor markets in the region.
A second element affecting the construction of the index is the discount rate. The authors use a single discount rate for all countries in the sample. However, the relevant discount rate for calculating the expected discounted cost of dismissing a worker is significantly higher in developing than in industrial countries. A possible notion of the relevant discount rate in developing countries could be obtained using existing measures of country risk. Again, it would be useful for the reader to have a sense of how the values in table 1 are sensitive to changes in the discount rate.
The main objective of the paper is to estimate the impact of labor regulations on employment. The paper addresses this issue in two ways. First, it relies on existing estimates of labor demand elasticities obtained from a sample of studies to simply infer the effect of labor regulations on employment. The authors find that the static labor demand elasticities estimated in the literature range between zero and negative one, with most of the cases clustered between -0.2 and -0.6. Even assuming that workers may bear part of the cost of labor regulations in the form of lower equilibrium wages, these estimates of labor demand suggest prima facie a potentially significant effect of job security provisions on employment. Of course, the size of the effect is measured directly by the labor demand elasticity, although the values of the elasticities may be directly influenced by how efficiently the labor market functions. In countries with relatively rigid labor markets, price elasticities of labor demand may tend to be large relative to the output elasticities of labor demand. Compare, for example, the relative size of price and output elasticities of labor demand in Argentina and Chile. Thus the measure of the effect of job security provisions on employment may be influenced by other elements affecting the functioning of labor markets. [End Page 150]
The authors also survey a number of studies that attempt to measure more directly the impact of job security provisions on employment. They report that a number of studies find that such labor market regulations have a significant effect on turnover rates. The impact on average employment, however, appears to be somewhat weaker.
In view of these results, the authors construct a data set spanning industrial and Latin American countries to estimate the effect of job security provisions, as measured by the job security index, on employment, its composition, and the rate of unemployment. Their results convincingly show a significant effect of labor market regulation on employment rates. Their results also show a strong impact of regulations on youth employment. Somewhat surprisingly, the effect on unemployment in Latin American countries appears to be much weaker. The authors suggest that this finding may reflect the fact that in the absence of unemployment insurance or equivalent support programs, workers tend to either find jobs quickly or drop out of the labor force. This explanation, however, does not appear to be consistent with the fact that some Latin American countries currently report unemployment rates comparable to those of some European economies.
This paper is thus valuable in two ways. First, it provides an important survey of current labor legislation in Latin America and the Caribbean. Second, the results are relevant to the current policy debate, as they offer evidence that some provisions in labor legislation intended to protect workers may indeed end up causing lower employment rates.
Note for Alejandra Cox Edwards
1. Lazear (1990).
Note for Pablo Guidotti
1. In Argentina, the government observed that during discussions on proposed modifications to existing legislation, employers' representatives from the financial and construction sectors favored centralized over decentralized negotiations while the opposite was true in the case of representatives from industry. Also, employers' representatives from large firms were rather indifferent to this issue, while representatives from small firms expressed a strong preference for decentralized labor negotiations.
