Fernanda Rivas and Máximo Rossi (*)

WAGE DISCRIMINATION IN URUGUAY (1991-1997)

Fernanda Rivas and Máximo Rossi (*)

Working Paper 7/2000 Department of Economics, University of Uruguay

Abstract

The purpose of this paper is study the evolution, during the nineties, of the wage differential between men and women, and in particular the labor market discrimination. We try to analyze if in a frame of increasing openness to international trade, decreasing inflation, wage negotiation decentralization and increasing wage inequality, the labor market discrimination against women has experimented any change. We estimate the total wage differential and decompose it in three components: I) men advantage or overpayment due to the existence of discrimination (favoritism), ii) women disadvantage (pure discrimination) and iii) differences due to different human capital levels and labor insertion. It is observed that the wage gap between men and women diminished. Although the three factors go in the same way to improve women condition in the labor market, it is the differences due to different human capital levels and labor insertion the main factor that contributed to narrow the wage gap during the period of study (1991–1997). The decomposition of the wage gap shows that the men start the period with an advantage in terms of human capital and labor insertion that soon disappears and stars to be favorable for women, being the main factor that explains the diminishing in the wage gap. Not only the advantage of men but also the disadvantage of women had very moderate changes thus in a frame of wage gap diminishing, at the end of the period, they remained as the main explanation of the wage gap.

JEL: J71)

(*) Departamento de Economía, Facultad de Ciencias Sociales, Universidad de la República (Uruguay) ( and respectively). We acknowledge comments from seminar of the Social Economics Network for Latin America held at Panama financed by Sida/SAREC (Sweden).

I.- - Introduction [1]

The aim of this paper is to study how the wage difference between men and women and, in particular, wage discrimination has evolved during the 1990s.

There is discrimination when individuals with the same capacities as human capital are not equally compensated due to features other than their productivity such as sex or race.

Many theoretical explanations have been given for the possible causes of discrimination. Becker (1971) particularly emphasizes the fact that discriminatory behaviour against women can be developed in imperfect competition markets. Becker himself (1995) has developed another model in which discrimination derives from women’s weaker effort in the labour market as a result of the role they play at home. Lundberg and Startz (1983) explain discrimination as a result of imperfect information. Blau (1972) considers women’s lack of continuity in the labour market to be an important variable for the existence of lower wages.

Other authors believe that discrimination can also take place in competitive markets (see Francois, 1998).

In the present paper we intend to examine whether the changes that have lately taken place in Uruguay (an increase in women’s participation in the labour market as well as in their human capital, facts which have developed in a context of a more open and competitive economy), have affected wage discrimination- and if so, to what extent.

We shall not test each of the hypotheses discussed throughout this paper, but we shall employ the latest developments of the traditional Oaxaca decomposition correcting for selection bias, an experiment that had not been carried out in previous studies on the subject for the case of Uruguay (Bucheli and Rossi, 1987, Furtado and Raffo, 1998).

The organization of this paper shall be as follows: in Section II we present a general overview on the subject; in Section III we deal with the methodology; in Section IV we present the source of information employed; in Section V we explain the results obtained and, finally, in Section VI, we point out the main conclusions.

II. - A General Overview.

During the period under study (1991-97), the Uruguayan GDP increased except for the year 1995. The economic policy during such period was mainly characterized by: i) a gradual opening of trade and the creation of a free trade zone (MERCOSUR) and ii) the implementation of a stabilization plan which resulted in an important decrease in inflation.

During this period there occurred some changes in the degree of centralization of wage negotiation: since 1990 negotiation has become decentralized and the union affiliation level has decreased.

Other studies on the evolution of income distribution such as the ones carried out by Bucheli and Rossi (1994) and Vigorito (1998), show that in the last fifteen years, said distribution has not undergone significant variations. The studies made by Bucheli and Rossi (1994), Miles and Rossi (1999), Vigorito (1998) shed light on the fact that such stable equality levels occur at the same time as other changes, namely, changes in the different income sources and within those sources themselves. More detailed analyses deal with pension distribution and wage compensation distribution.

A recent study by Gradín and Rossi (1999) shows that a period of some stability in terms of wage distribution was followed by a clear tendency towards an increase in wage inequality in Uruguay, both in the capital city and in the rest of the urban country (RUC), particularly since 1990. Although the initial inequality levels were similar, the gradual increase in inequality is higher in Montevideo than in the RUC.

The present paper analyses whether wage discrimination against women has undergone any changes within this general framework.

a)  Participation Rates in the labour market

The increase in women’s participation rate is one of the most interesting changes in the labour market as from the late seventies: so much so that it has increased in more than 50% (let us bear in mind that the average annual rate of women’s participation was 27.4% in 1969, and 46.1% in 1997.)

During the period under study the participation rate of women between 25 and 55 years old rose by 5.1 percentage points, that is from 64.6% to 69.7%. This proves that the historical tendency of women’s participation, which began in the early sixties, has continued.

Figures No. 1

Source : own processing on the basis of the ECH of the INE.

When analysing labour market participation according to the educational levels, we can identify different working behaviours (see Figures 2). The highest participation rates correspond to the highest educational levels, both for men and women. There has been an important fall in the participation rate of men with low educational levels, from 94% to 89.5%, while the other educational levels underwent a slighter fall: the second educational level decreased by 2.1 percentage points and the two highest ones, by 1.1 and 0.1. In the case of women, the participation rate rose for all educational levels except for the lowest one (primary school unfinished), which fell 0.3 percentage points. Participation rates for the next two lowest educational levels increased by 4.4, 5.1 and 0.4, the highest rates occurring at the intermediate levels (from 6 to 12 years of formal education).

Figures Nº 2

Source : own processing on the basis of the ECH of the INE.

b)  Composition of labour force between 25 and 55 years old for each educational level

During the period under study the proportion of workingmen with unfinished primary education fell, while the participation rate among those who had between 6 and 9 years of formal education remained constant and the rate for the highest educational levels increased slightly.

In the case of women, the proportion of workingwomen for the two lowest educational levels fell and it increased for the two highest ones. It is important to point out that in 1997, 22.5% of workingwomen had 13 or more years of education, while that was the case only for a 13.5% of men. The set of workingwomen with a higher level of education increased much faster than the set of men for the same level.

Table No.1 Composition of the masculine active population according to educational level

Source : own processing on the basis of the ECH of the INE.

Table No.2 Composition of the female active population according to educational level

Source : own processing on the basis of the ECH of the INE.

c) Earnings of men and women

To compare the evolution of the wage ratio between men and women we have considered the set of full-time workers (over 35 hours a week).

Figure No.3

Source : own processing on the basis of the ECH of the INE.

From the analysis of Figure 3 we can draw two conclusions. Firstly, that during the period under study men’s hourly wage was higher than women’s wage. Secondly, that such difference showed a decreasing tendency during the whole period. In graph 4 we see that the most dramatic fall in women/men wage ratio occurred for the set of workers with third level education.

Figure Nº 4

Source : own processing on the basis of the ECH of the INE.

III.- Methodology

The basic decomposition employed for this study is the Oaxaca (1973) and Blinder (1973) type. The procedure starts with a Mincer-type wage equation for men and women (sub-indices m and f used for men and women respectively).

A basic problem is to decide which is the non-discriminatory wage structure. In most studies, men’s wage structure is taken as the non-discriminatory one. In the present paper we shall use Oaxaca and Ransom’s structure, 1994 (Cotton’s -1988- salary structure is just a particular case of the Oaxaca and Ransom’s one).

The matrix of weights specified by said authors is:

where X is the matrix of observations for the whole sample and Xm is the matrix of observations for the sample of men.

The salary structure in the absence of discrimination would be the following:

This information allows us to estimate the total wage differential and to decompose it in three parts: i) men’s advantage or overpayment deriving from discrimination (favouritism); ii) women’s disadvantage (sheer discrimination) and iii) differences attributable to different levels of human capital and of labour market participation. Therefore, the decomposition bears the following structure:

where:

and correspond to gross wage differential, and are the means of the Inverse Mills Ratio estimations, and are the estimations of their coefficients in wage equations, and are the estimated coefficients of wage equations for men and women, is the non-discriminatory payment structure and are the characteristics average value.

The procedure was carried out correcting for selection bias according to Heckman (1979)[2] . The potential existence of selection bias occurs because the information employed may not be a random sample of the population, since only the set of individuals who participate in the labour market are taken into account, thus their reservation wage being lower than the offered wage. As a result of this, the estimators of the parameters per ordinary least squares are often inconsistent.

Basically, this procedure introduces the Inverse Mills Ratio in the wage equation, which estimates the probability of labour market participation and which measures the effect of sample selection. For the purposes of correction an estimation in two stages is required: the second stage is –as mentioned before- to introduce a new variable in wage equation; the first one is to obtain a parameter l from the estimation of a probit model of labour market participation.

IV.- Data

This paper is based on the Continuous Household Survey conducted by the National Statistical Office (INE) in Uruguay. This survey gives information on urban population in two major regions: Montevideo, the capital city, where more than half the total population lives, and the Rest of the Urban Country.

This survey has been conducted monthly, with the same layout, since 1981 and bears individual data on monthly wage income, non-wage income, age, sex, educational levels, occupation, working hours and other relevant variables.

For this paper we have taken into account data of individuals between 25 and 55 years old, analysing such data for Montevideo (M), the Rest of the Urban Country (RUC) and for the whole country (WC).

The variable under study is hourly earning (see Annex I), which has been deflated by the Consumer Price Index for March 1997.

The household survey provides information on the individuals’ experience at their present jobs, but does not provide any information about the individuals’ experience as a whole. In this sense we follows the traditional method of estimating potential experience (age minus years of education minus six). Disregarding unemployment periods, men’s potential experience as a proxy of the experience seems relatively reasonable. Women’s potential experience, however, shall appear in many cases as higher than the real one due to their tendency to get in and out of the labour market as a result of the responsibilities assumed at home. For this reason there occurs some bias in the coefficient estimation of the potential experience variable that can be ascribed to a discriminatory treatment. Bearing that in mind, we shall be cautious when analysing the results and we shall highlight the tendency rather than the absolute value of wage discrimination.

V.- Results

First of all, we show the changes that have taken place at the beginning and end of the period under study (see Figure 3).

Secondly, we present the evolution of the wage ratio between men and women (see Figure 4), of men’s advantage in terms of wages, of women’s disadvantage and of the differences in human capital and in labour market insertion.

If we analyze the total variations between the two ends of the period, it shall be observed that the wage gap between men and women has narrowed in Montevideo and the RUC. Nonetheless, such narrowing is more significant in Montevideo.

Despite the fact that the three factors mentioned (men’s advantage, women’s disadvantage and differences in human capital and in labour market insertion) seem to move, in general terms, towards improving women’s situation within the labour market, the reason for the narrowing in the wage gap, is the difference in human capital and in labour market insertion.