انعطاف ناپذیری دستمزد جزیی در مکزیک: شواهد از پرونده های امنیت اجتماعی

This paper analyzes the existence and extent of downward nominal wage rigidities in the Mexican labor market using data from the administrative records of the Mexican Social Security Institute (IMSS). This establishment-level, panel dataset allows us to track workers employed with the same firm, observe their wage profiles and calculate the nominal-wage changes they experience over time. Based on the estimated density functions of nominal wage changes, we are able to calculate some standard tests of nominal wage rigidity that have been proposed in the literature. Furthermore, we extend these tests to take into account the presence of minimum wage laws that may affect the distribution of nominal wage changes. The densities and tests calculated using these data are similar to those obtained using administrative data from other countries, and constitute a significant improvement over the measures of nominal wage rigidities obtained from household survey data. We document the importance of minimum wages in the Mexican labor market, as evidenced by the large fraction of minimum wage earners and the indexation of wage changes to the minimum wage increases. We find considerably more nominal wage rigidity than previous estimates obtained for Mexico using data from the National Urban Employment Survey (ENEU) suggest, but lower than that reported for developed countries by other studies that use comparable data.

There is a growing consensus about the importance of labor regulations and labor market policies in determining the productivity of labor itself as well as total factor productivity. While theory1 and evidence2 are increasingly pointing to the importance of a flexible labor market, the implementation of a labor reform to attain greater flexibility has been pending for the past two decades in Mexico and other Latin American countries. In fact, most of the labor regulations in this region date back to the first half of the 20th century and, while some countries have implemented partial reforms to increase the flexibility of their labor markets, these attempts have been halfhearted at best.3 The issue of labor market reform has become even more pressing in light of the increase in international competition from China and other South-East Asian countries, most of which have “at will” hiring and firing policies and thus very flexible labor markets. Furthermore, the United States and Canada—Mexico's main trading partners—, also have more flexible labor markets, with relatively few impediments to how labor is allocated within or between firms. This fact evidently places Mexico at a disadvantage relative to all these countries. One reason why this fact could be ignored in the past is that real wages in Mexico displayed a remarkable degree of downward flexibility, which provided much of the correction needed to regain competitiveness. This adjustment typically came through unexpected increases in inflation, which induced sharp decreases in real wages. As the macroeconomic environment in the region has become more stable and inflation gradually converges to the levels of developed countries, this clearing mechanism is no longer available. Thus, flexibility in the labor market has become a much more important issue. In this paper we focus on two particular types of labor regulation which may give rise to inefficiencies in the allocation of labor: downward nominal wage rigidity (DNWR) and minimum wages. Although these two topics have generally been treated separately in the literature, downward nominal wage rigidities cannot be analyzed independently of minimum wages in developing countries like Mexico, where the fraction of minimum wage earners is high relative to more developed economies and the practice of indexing wage changes to the minimum wage increases is pervasive. The detection of nominal wage rigidities usually begins by estimating the probability density functions of wage changes over consecutive periods (typically a year) using panel data on individual workers or jobs. The existence of DNWR is then first established through analyzing the shape of the density functions, as well as through estimating certain moments of the distributions, and how they relate to the rate of inflation. To the extent that wages exhibit DNWRs, the estimated densities should be asymmetric (right skewed), with few observations corresponding to negative wage changes and much of the density's mass piled up at zero, reflecting the fact that wage decreases are legally or institutionally prohibited so negative wage changes should not be observed. Furthermore, this pattern should be more pronounced at lower inflation rates to the extent that at high inflation levels real wage decreases might make nominal wage cuts unnecessary. The analysis is formalized through the estimation of sample statistics and measures of DNWR that try to compare the actual distribution of nominal wage changes with a hypothetical distribution of nominal wages under no rigidities (see below for a description of some of these tests).4 The existence and extent of DNWR has given rise to one of the longest standing debates in macroeconomics. Many macroeconomic models postulate the existence of such rigidities, and their results crucially hinge on this assumption since the effects of monetary policy largely depend on their existence and magnitude.5 Despite the importance of this topic, the empirical evidence on the existence and extent of DNWR is relatively recent. Moreover, the literature has thus far focused almost exclusively on developed countries.6 The first studies of DNWR were typically based on microeconomic data on wages from household surveys from the United States, such as the Panel Study of Income Dynamics (PSID) or the Current Population Survey (CPS), and focused on establishing the existence and extent of DNWR.7 The literature then shifted from using panel data from household surveys to using data from collective bargaining agreements by unions8 or panel data from firm- or establishment-level administrative records, such as the data used in the calculation of the Bureau of Labor Statistics' Employment Cost Index (ECI).9 The advantage of the second type of studies is that they are based on more accurate measures of wages, since they are obtained from administrative records and not from self-reported measures form household surveys. Even more accurate measures of wages and wage changes have been obtained using data for a small, non-random sample of firms that track both jobs and individuals.10 This paper is, to the best of our knowledge, the first to provide evidence on the existence and extent of DNWR for a developing country using establishment-level, panel data. In particular, it uses panel data on individual workers drawn from the administrative records of the Mexican Social Security Institute (IMSS) for every quarter over the period 1985:01–2001:04.11 This type of dataset provides more accurate measures of nominal wages and nominal-wage changes than either household survey data or manufacturing survey data and is more comparable to the firm- and establishment-level datasets recently used in Australia, Germany and the U.S. to measure DNWR. The only previous study on DNWR for the case of Mexico (Castellanos, 2003) estimates a series of standard measures and tests using microdata from the National Urban Employment Survey (ENEU) for the period 1994–2001.12 In particular, this study exploits the rotating panel structure of the ENEU data, which tracks workers over five consecutive quarters, in order to estimate the distributions of nominal wage changes. Her results suggest that nominal wages in Mexico exhibit few downward rigidities when compared to the United States, Canada or Australia.13 The main contribution of the present study is to bring together the measurement of DNWR and the rigidities introduced by the existence of minimum wages using a unified framework. In particular, this paper calculates several standard measures and tests that have been proposed in the literature in order to test for the existence and magnitude of DNWR. It then modifies the tests proposed by Kahn (1997) to account for the possible effect of a large fraction of minimum wage earners and widespread indexation of nominal wage changes to the minimum wage increases. To the best of our knowledge, this is the first time that tests of DNWR have been obtained for a developing county using this type of data, and the first time that the standard tests of DNWR have been modified to explicitly incorporate the effects of the minimum wages simultaneously. The issue of DNWR in Mexico is particularly interesting for at least two reasons. First, according to several measures of labor legislation flexibility (in hiring, in firing and in working conditions), Mexico is one of the countries with the most rigid labor laws in the world.14 In fact, the law specifically forbids nominal wage cuts. This is in contrast to many countries where DNWR may be the result of a social perception that nominal-wage decreases are “unfair”. Thus, while for most countries it remains an open question why nominal wages do not adjust downwards to help clear the labor market during recessions, it is less of a puzzle in Mexico since this provision is part of the legal framework. Second, the process of disinflation that the Mexican economy has undergone during the past two decades, and the resulting stabilization of nominal variables, may have increased the importance of DNWR. In particular, while DNWR may not have been relevant for firms during the period of high inflation, when real wages declined even in the absence of nominal wage cuts, the current environment of low and stable inflation makes this phenomenon potentially binding. Furthermore, the Mexican disinflation process provides the necessary variation in inflation rates needed to distinguish DNWR from other factors—such as menu costs and long-term contracting—that could affect the distribution of nominal wage changes. The rest of the paper is organized as follows. The next section provides an overview of the institutional and legal framework that regulates labor in Mexico. Section 3 describes the data used in the paper. Section 4 presents some simple statistics. Section 5 describes the econometric models we estimate. Section 6 presents the main findings of the paper. Finally, Section 7 presents the study's conclusions.

Using a unique panel dataset for all firms registered with the IMSS, we studied the distributions of changes in the log of nominal wages for employees who remain employed with the same firm from 1 year to the next. The analysis of the distribution of workers by wage level revealed the importance of the minimum wage in the Mexican labor market. In particular, it was shown that a large fraction of workers registered with the IMSS are reported to earn the minimum wage. This fact was corroborated by the finding that a large fraction of the changes in log wages were equal to the change in the log of the minimum wage. For the late 1980s, we also found that the wage changes of those earning substantially more than the minimum wage were indexed to the minimum wage. We also found that the fraction of employees whose nominal wages have decreased from 1 year to the next, while small, has been increasing over time. Thus, some private sector firms legally registered with the IMSS are increasingly being able to lower the nominal wages of their employees, despite the legal prohibition against doing so. In addition, we found evidence that it has been mostly employees at large firms that have been experiencing nominal wage decreases. This result seems surprising since one would suspect that large firms would be most affected by the legal restrictions against lowering nominal wages. To formally study nominal wage rigidities, we ran the test proposed in Kahn (1997). We also ran a modified version of this test, which considers the presence of a large fraction of workers whose nominal wage changes correspond exactly to the increase in the minimum wage. These tests have the advantage that they do not assume that the underlying distribution of nominal wage changes is symmetric. The results regarding nominal wage rigidities are consistent with those found in the literature. In particular, many workers experience no change in their nominal wage. Apart from a spike in the distribution of wage changes at zero, we estimate that reductions of the nominal wage are particularly uncommon. We therefore find evidence of downward nominal wage rigidity. Finally, when we break the sample into two parts, we find evidence that magnitude of downward nominal wage rigidities has fallen over time. More downward flexibility in the period after 1994 accords well with the fact that indexation was more prevalent during the years of the inflation stabilization plans than afterwards. It is also consistent with the decline in union density that has taken place over the same period. However, in the absence of any changes to the labor laws to promote market flexibility, this pattern remains an open issue for future research. The results of this paper regarding the importance of the minimum wage and the existence of downward nominal wage rigidity challenge the common perception that labor legislation in Mexico is actually innocuous in practice, since firms might easily be able to avoid complying with these regulations. In particular, for a large part of the group of legally registered, private-sector firms in Mexico, the rigidities embedded in the labor legislation seem to be important binding constraints. These rigidities will become more binding and relevant as the country and the region face increasing competition from other countries with more flexible labor markets, and as inflation continues to converge to the level of more developed economies. Thus, we believe that this topic deserves further consideration by both researchers and policymakers alike.