Volume 45 Number 177,
April-June 2014
Jobs, Salaries and Inequality in Argentina:
An Analysis of Distributional Determinants
Fernando Groisman*
Date received: February 28, 2013. Date accepted: July 5, 2013

Following the 2001-2002 crisis, the economy of Argentina experienced a strong recovery, which extended throughout the time period from 2003 to 2011. Looking at levels of job creation and salary dynamics, the labor market also showed favorable behavior. One of the unique features of the labor market in these years was the marked increase in the relative share of employed individuals with high levels of education. However, this did not lead to an increase in salary dispersion; rather, on the contrary, there was an ostensible reduction in salary inequality.

Keywords: Argentina, employment, salaries, inequality, salaried workers in the private sector labor market.

Some of the most significant economic recovery following the 2001-2002 crisis was seen in the growth of private sector salaried workers. Essentially, between 2003 and 2011, this subset of employees grew by more than 50 percent. The share of workers with completed secondary studies who were not heads of household grew relatively among this group, while the proportion of young workers between 18 and 29 years of age fell, and the proportion of those aged 30 to 39 grew (see Table 1). In terms of the features of these jobs, the significant increase in positions registered with social security is notable; that is, higher-quality jobs.1

This change in the composition of employment in an environment of full growth can be interpreted as the result of company preferences. This would suggest that salaries should have followed a similar pattern, that is, should have shown relative variations, highest for the most in-demand groups of workers. However, salary evidence partially contradicts this assumption (see Empirical Results, p. 69). It was demonstrated, for example, that the salaries of those who had not finished secondary school saw the greatest relative increases (see Table 2).

The coexistence of opposing trends in employment and relative salaries is indicative of the fact that the distributional balance is a priori indeterminate. In effect, the balance in terms of equality could be compatible with both an increase in the degree of income concentration — for example, if the effect of education on employment were to prevail — and with the opposite case, with decreasing inequality, if educational returns were to fall to such a point that the previous effect was discounted. In Argentina, there has been a reduction in the inequality of salary distribution between the extremes of the period 2003-20112 (see Table 3). We can also observe that the decrease in inequality was considerable, nearly six points in the Gini coefficient.

The debate surrounding the deeper causes of distributional changes is still open. An important part of this research aimed to analyze the increase in inequality in Anglo-Saxon countries over the past two decades of the past century. From there arose the emphatic role of technological change and its impact on the demand for skilled employment (Freeman et al., 1995; Acemoglu, 2002). Changes to the technology pattern would have led to a decrease in the demand for routine skilled workers, as computers became widespread (Krueger, 1993), who were traditionally located in the middle of the salary distribution (Autor et al., 2006). In keeping with that, the salary gap widened in the upper extreme, between the ninth and fifth deciles, as the differential was reduced in the lower part, between the fifth and first deciles. It is said that this distributional deterioration could be evidence of polarization in jobs that essentially came out of a bias in the demand for more highly skilled workers. The central argument of these approaches highlights that modifications in the salary structure, due to increases in salary returns on education and/or training, would have driven even greater salary concentration. On the other hand, there have been cases with a decrease on the return for higher education, which could be linked to slow introduction of technological progress, thereby limiting the demand for high skills (Izquierdo et al., 2007; Naticchioni et al., 2008 and 2009). In a different paradigm, other research has indicated that the composition effect should not be ignored. By this interpretation, even when the increased education and experience of the workforce drives the increase in salary spread, other factors may have had certain influence, such as a decrease in union membership (Lemieux, 2006 and 2008, Mosher, 2009). As a result, distributional changes may be responding to changes in salary returns as well as variations in the composition of the universe of workers.


This study used the methodology developed by Firpo et al. (2009), principally because it allows the use of a regression technique to estimate the effects of changes on independent variables (X) in different sections, or quantiles, of the distribution of the dependent variable (Y). The procedure can also be extended to estimate factors that influence various distributional indicators, such as the Gini coefficient, among others. In addition, it allows the breakdown of changes in these indicators in the style of the Oaxaca-Blinder methodology (Oaxaca, 1973; Blinder, 1973) for salary differences. In other words, it makes it possible to identify the influence of the endowment effect and return effect for each covariate included in the model.

Basically, the model consists of an unconditional quantile regression (uqr), whose main feature is that it produces a recentered influence function (rif) for the dependent variable, and then regresses it over the independent variables (X) (Firpo et al., 2009). Formally, uqr regressions allow for the estimation of the marginal effect over a given unconditional quantile of the change in an observed variable X. The dependent variable Y is a function h of the observed features X and the unobserved ε. Moreover, is the quantile of the unconditional distribution of Y.

The unconditional marginal effect is based on a two-step procedure which makes it possible, first, to estimate an rif function for each individual where the density of income fy is estimated using a kernel indicator. This procedure computes the effect of an independent variable on the probability of obtaining income over (and in) a determined quantile. In the second step, the rif estimate is regressed over the explanatory variables using ordinary least squares. In this way, the probability that a worker obtains a salary over a determined quantile is assumed to be linear to the observed characteristics.


This procedure has a clear advantage with respect to standard regression techniques — in other words, Mincer salary functions — that are limited to estimating the effects of independent variables on conditional average salary. These models capture the expected effect on the salary that a worker will receive in light of the modification of a characteristic included in the vector of independent variables — for example, the condition of a registered job or the educational level. However, the coefficients obtained in standard regressions tend to differ among people, depending on where they are located in different sections of the salary distribution.3

In the same way, the uqr method is also different from conditional quantile regressions (cqr). These models are appropriate for evaluating the degree of spread of the variable of interest (Y) within the different sub-groups of the population that is assumed to have homogenous composition. In effect, this technique is valid for estimating the influence of a given independent variable — for example, whether a given position is registered — on sets of individuals where it is assumed that they all have the same observed characteristics, that is, controlling for the rest of the independent variables.4

uqr regressions consider the effective location of individuals in the income distribution, without controlling for observed characteristics. The coefficients obtained in this way can therefore be interpreted as the effect on the salary for each income quantile if there were a global increase in the independent variables (X), for example, in education, or employment registration, by 1 percent. In other words, this technique allows us to estimate how the income distribution would be affected given modifications in the independent variables.

To clarify the interpretation of these results, one might think of two groups of workers, for example, registered and unregistered, each with an extreme scenario. In the first, there would be no difference in the average salary of both groups of workers, that is, there would be no inter-group inequality, but salary variance would be higher among unregistered workers than among registered. In this context, an increase in registration — and therefore a decrease in the number of unregistered — would reduce total salary inequality, due to the decrease in global variance as a result of the transfer of workers. In the second scenario, the situation is reversed: an absence of variance because all workers receive the same salary. Consider, moreover, that they all belong to the group of unregistered salaried workers. In this case, the registered workers, who are non-existent at the beginning, would have a higher salary than the unregistered salaried workers and there would be no intra-group variance, and therefore no salary differences between them. By increasing registration, that is, when an unregistered worker becomes registered, there would be an increase in overall inequality, as there would have been an increase in overall variance. Distributional deterioration would continue until the point at which an additional increase in registered workers would lead to a reduction in total variance. This situation would occur when there are quantitatively fewer unregistered workers than registered workers.

The fact that the transformation implied by the uqr technique is similar to a standard regression, with the difference being that the dependent variable (Y) is replaced by the recentered influence function of the quintile of interest, allows us to advance, unlike other decomposition alternatives (see Juhn et al., 1993; DiNardo et al., 1996; Machado et al., 2005) towards its decomposition following the Oaxaca-Blinder methodology (Firpo et al., 2011).5

This article analyzed full time private sector salaried workers — those who work more than 34 hours a week — aged 18 to 59 years. The dependent variable was the (natural logarithm of) the hourly wage. The way in which this study group was defined favored the identification of distributional changes in economic activities with higher productivity and helped in categorizing the types of institutional factors that may have had an influence on these changes. The estimates were calculated using the micro-databases of the Permanent Household Survey (eph), conducted by Mexico's National Statistics and Census Institute in the major urban areas of the country.


The Evolution of Employment and Salaries in the Private Sector

The labor market in Argentina is characterized by a high proportion of workers dependent on the occupational structure. By the end of 2011, practically eight out of every ten employees aged 18 to 59 was a salaried worker, 77 percent of whom worked in private companies.6 Between 2003 and 2011, there was accelerated growth in private sector salaried workers. At the same time, this growth was accompanied by a marked change in their makeup. It is important to note, in this sense, the increase of over nine percentage points in registered employment, which went from 62.4 percent in 2003 to 71.7 percent in 2011 (see Table 1). A direct method to weigh this transformation arises from considering that the net increase in registered jobs was equivalent to about 90 percent of the variation, also net, of the total private sector salaried workers in this period.

Other modifications to the composition of this group were notable as well. For example, the number of salaried workers in jobs that demand operational skills rose by 6.7 percentage points, mainly at the expense of employment in positions that did not require minimum skills and, to a lesser extent, with a relative reduction of workers in technical jobs.7 At the top of the pyramid, salaried workers with professional skills maintained their relative share at around 6 percent. Also, there was a significant reduction in employment among small enterprises with up to five employees, by 9 percentage points, in favor of medium-size and large companies, practically in equal measure. By sector, it would be useful to note a marked increase for construction, by 3 percentage points, which, however, did not lead to a relative reduction in industrial jobs, which were maintained on the order of 25 percent, in both 2003 and 2011.

In terms of socio-demographic features, there was an increase among those with completed secondary studies by 4.5 percentage points, with a reduction in the relative participation of salaried workers that had not completed this level. People with completed university studies saw their relative weight grow by less than 1 percentage point (see Table 1). The expansion of education among private sector salaried workers was far superior than that of the overall population. Effectively, the number of workers with a maximum educational level of completed secondary school increased by 60 percent between 2003 and 2011, while this figure was only 28 percent among the general adult population. In terms of age, which could be considered a close indicator of labor experience, there was growth of 2.7 percentage points in the age group of 30 to 39 years, with relative reduction among the youngest, between 18 and 29 years. There were no changes among those aged 40 to 59 years. In terms of household position, there was a considerable drop in heads of household, by 5.2 percentage points. Finally, it is important to note that there were not relevant changes in the gender makeup of this segment of workers.

Between 2003 and 2011, the average hourly wage of workers dependent on the private sector went up by five times in nominal terms, while purchasing power grew by about 35 percent.8 It has also been confirmed that, despite marked changes documented in the makeup of this group, the salary outlook for these categories was stable overall. Effectively, the salary gap did not change by age, household position and educational level. In terms of education, it should be noted that in reality, salary spread fell due to the fact that those with low levels of education saw salary increases with respect to the average, whereas those who had finished medium and higher levels of education saw decreases, especially in the latter group (see Table 2). In other areas, there was also a trend towards salary compression by sector of activity, geographic region and size of company. In the last case, it is interesting to emphasize that the salary of workers in the largest companies fell while the average remuneration of salaried workers in companies with up to five employees increased.

It should also be noted that the salary gap between registered and unregistered workers only worsened slightly (see Table 2). We can observe that the salaries of registered workers fell with respect to the average salary, but the salary of unregistered salaried workers fell even more.

In summary, the labor outlook in the private sector of the economy at the end of these eight years was characterized by marked discrepancy among the changes in the composition of salaried workers, in accordance with certain socio-demographic features and characteristics of the jobs, and the evolution of the salary structure. The composition revealed a strong trend towards greater relative participation of workers with higher education, experience and skills, who tended to find registered jobs in medium-size and large companies. On the other hand, the salary structure revealed strong overall stability and, in some variables (education and skills), a trend to compress the differentials.

On Factors that Determine the Hourly Wage

Identifying variables that have some influence on the distribution of income is generally done through a multivariable regression analysis estimated by least ordinary squares. These income equations, called Mincer equations in the literature (Mincer; 1974), provide a reasonable specification of the conditional average of salaries. With this procedure, based on human capital theory, it is possible to identify the independent effects exercised by certain personal attributes on the variability of salaries. It is also possible to expand the model with features relevant to certain jobs and companies. The results of this type of analysis in Argentina clearly show that salaries grew with education. People who had completed the secondary/university level of education received higher salaries than those who had completed primary and secondary school only. This effect was even stronger for heads of household, males and people over age 30, with respect to non-heads of household, women and young people aged 18 to 29, respectively.9 It is also interesting to highlight that the variables associated with features of jobs also had significant influence. In particular, this work verified the persistence of a high salary return associated with the quality of the job, measured by whether or not the job was registered, both in 2003 and 2011. This return, and it must be emphasized that it is related to an attribute of the job itself, rather than the worker, is indicative of segmentation in the labor market.10 In other words, when the rest of the covariates included in the model are identical — controlling for educational level, gender, size of company and other independent variables — salaried workers in unregistered jobs suffered a severe salary penalty. The coefficients associated with skills: professional, technical and operational, in this order, and the size of companies (large and medium-size), also saw positive results as compared to people in unskilled positions or smaller companies, respectively.

The evidence described here is useful as an initial approach to identifying the variables that have an impact on salary distributions. However, other methods should be used to specifically address how to evaluate factors that determine salary inequality (see Methodology, p. 66).

Factors that Determine Salary Inequality: Unconditional Quantile Regressions

Unconditional quantile regressions (uqr) were computed for quantiles 1, 5 and 9. In the sections that follow, the interpretation of the estimates was limited to the variables: registered employment and educational level. This decision was made because these factors showed the greatest explanatory power for salary spread.

Registration of Jobs

The progressive effect of registration on salary distribution can clearly be seen in the coefficients obtained through uqr regressions. In effect, the parameter associated with registration was consistently lower the higher up the unconditional income distribution. It can even be noted that this parameter was located at the limit of statistical significance in decile 9, where its value was 0.076 and 0.041 in 2003 and 2011, respectively. In the lower decile, this figure was 0.67 and 0.89 for the two years, respectively (see Table 3). In other terms, the differential tended to be null among those with higher salaries, while it was positive at the lower extremes. This is an expression of its equalizing effect. Workers in low-income positions will receive greater benefits from an equally distributed increase in registration among workers.

For comparison, this work also estimated models with conditional quantile regressions (cqr). These models are useful to evaluate intra-group salary spread, in this case for registered and unregistered workers, controlling for the influence of the rest of the model covariates. The coefficients obtained for registration, in both 2003 and 2011, were positive for all deciles under consideration (1, 5 and 9) and decreased as the income conditional distribution increased (see Table 5). In other words, the reward for registration among salaried workers with higher incomes was lower than that for lower income workers, controlling for educational level, age and other covariates.




The coefficients estimated in each of the unconditional quantiles, considered on their own, can be interpreted as an approximation of the rates of salary return. In addition, comparing the value reached by these coefficients for the diverse quantiles under consideration, a reasonable approximation of the composition effect emerges; that is, the effect that results from changes in the relative participation of workers depending on level of education. Focusing on analysis of this last aspect, the results obtained in 2003 for average level of education allow us to conclude that an increase in these workers, equally distributed throughout the entire salary distribution, would have a neutral effect on inequality. Note that the coefficients were not statistically different in all of the deciles considered. This behavior was maintained in 2011, but was somewhat weaker (see table 6). These results are in keeping with the already high percentage of salaried workers with completed secondary education. On the other hand, the distribution effects of changing the proportion of workers with completed higher education revealed a rather different scenario. Effectively, the relative increase in this group had a clear concentrational effect. Observe that the coefficients in decile 1 were 0.14 and 0.12 in 2003 and 2011, respectively, while in decile 9, these figures were 0.50 and 0.46 for the two extremes of the time period considered.

The Decomposition of the Distributional Change

The distributional effects of the various variables analyzed, considered individually, may be reinforced or neutralized when looking at their aggregate effect on inequality. It is therefore useful to comprehensively analyze the influence of each of these factors on salary inequality, and how this may have evolved from 2003 to 2011 (see Tables 6 and 7). This analysis is possible through a Oaxaca-Blinder decomposition of the temporary changes in distributional indicators (Firpo et al., 2009). This article used this procedure for the variation of the Gini coefficient and the Logarithmic Variance (see Methodology, p. 66).


The decomposition of the distributional change verifies that the endowment or composition effect had a favorable, but overall minimal, influence on reducing inequality. Salary returns had the greatest effect (nearly 85-90 percent of the reduction in inequality is due to this effect). In other words, if there had not been such changes in salary returns, that is, in the way in which certain features of workers and jobs were remunerated, the variation in salary inequality between 2003 and 2011 would have been very slight, barely over one point in the Gini coefficient.

Analyzing each of the factors that plays a role in this reduction, the influence of the value associated with the constant is important. Effectively, this salary coefficient was the biggest contributor to reducing inequality; note that this value was such that it surpassed the overall difference between distributional indicators considered. The salary return on the constant captures the portion of remunerations not associated with determinants included in the model.11 In other words, interpreting this result indicates that salaries in Argentina tended to increase with greater relative intensity for salaried workers in the lower part of the income distribution. This scenario is coherent with the regulatory and institutional environment during this time period that drove the recomposition of remunerations from the base of the salary pyramid, with relative autonomy both in terms of personal characteristics and jobs and/or companies.

The salary return on the quality of job, measured by whether or not the job was registered with social security, was the second most powerful factor in determining greater salary equality. This is compatible with a significant increase in registration, as already shown, which had a big effect on workers with lower levels of education, for example, who saw their salaries go up considerably due to this shift. To a lesser extent, salary returns associated with skills, size of the company and age, in that order, also played a favorable role in reducing salary inequality. Salary differences linked to the diverse regions of the country, activity sectors and education, behaved against reducing the salary spread. This is unsurprising when we take into account that institutional regulations tend to ensure minimum wages, which can be surpassed by companies. In this framework, it is feasible that the companies in regions/sectors of activity with comparative advantages or greater productivity are willing to pay higher salaries to more educated workers, especially in times of growth, as this period was.


This analysis of how employment and salaries evolved from 2003 to 2011 revealed opposing trends. On the one hand, the educational profile of private salaried workers shifted in favor of those with more education. The relative increase of the proportion of the workforce with medium and high education was rather greater than the increase observed in the same time period for the overall adult population. The bias in job demand for this segment of workers, however, did not translate into an increase in their salaries with respect to wages received by workers with low education. On the contrary, the period is notable in that it saw a decrease in the salary differential. The discrepancy between job dynamics, measured by certain features/attributes of the workers, and salary evolution, as described, was not exclusively limited to education. Similar patterns were found in other variables, such as skills. The number of operational and technical jobs increased, while salaries did not reflect this greater dynamism. And, while the number of jobs fell in small companies, average salaries in small companies did not fall behind the general pay structure.

Changes in the makeup of employment and the salary structure are correlated to the distributional balance. The sign will depend on which factor prevails, jobs or salaries, and its characteristics. Specifically, on the degree of salary spread (inter and intra-group) and the relative weights of the relevant sub-groups of workers. In Argentina, there was a strong decrease in salary inequality on the order of 15 percent with respect to the Gini coefficient observed in 2003. It was also confirmed that the decrease in salary concentration was due to an intense recomposition of incomes in the lower end of the distribution.

This behavior is compatible with the effects produced by certain institutional variables, such as the minimum wage and/or salary agreements among employees and employers that define basic wages for a sector of activity. These regulations, which usually have a direct effect on registered jobs, alongside other governmental regulations, like fixed amount salary increases, may counteract, neutralize or exacerbate the repercussions of changing job demand on salaries. The salary equations, estimated with unconditional quantile regressions, help evaluate the impact of socio-demographic variables and other features of jobs in different parts of the salary distribution and provide information on synthetic distributional indicators. In this way, we can draw conclusions regarding the effects of marginal changes on salary concentration.

This analysis provides evidence as to the influence of improving the quality of jobs. In effect, salary returns associated with lower quality jobs, measured by registration in the social security system, were a potent determinant of salary equality. The estimated models demonstrate that in both 2003 and 2011, workers on the lower end of the unconditional salary distribution would have benefited to a greater extent by increasing equally distribution registration among the set of workers. In this interpretation, it is important to consider that the increase in registration was of considerable magnitude — equivalent to 90 percent of the overall net variation of the set of private sector salaried workers — and, in this way, reached a relevant proportion of workers receiving low salaries. The increase in protected employment continued to reduce salary inequality among private sector workers, because those in a registered job were able to go up in the distributional order, reducing the overall gap. It is of note that the salary reward for registration, evidence of labor segmentation, did not fall between the extremes of the time period and that the average salary of these workers was far superior, by around 80 percent, than those of unregistered salaried workers.

Another piece of evidence is that despite the changing educational profile of the salaried workforce, there was a marked decrease in salary returns on education. Effectively, in practically all estimated salary models, these coefficients fell between the two extremes of the period. Consequently, it could be deduced that there would have been a relevant change in salary determinants, shifting from those focused on personal attributes to those centered on features of the job itself. This shift agrees with the role of institutional devices that tend to raise the lowest salaries regardless of the personal features of the workers. Examples in this sense would be the minimum wage, the spread of collective negotiating and growing union action. The results of the decomposition of the change in selected distributional indicators confirm this statement.


The author would like to thank the anonymous evaluators for their comments.


Acemoglu, Daron (2002), “Technical change, inequality and the labor market”, Journal of Economic Literature, vol. 40, no. 1, American Economic Association, March, pp. 7-72.

Autor, David H.; Katz Lawrence F., and Melissa S. Kearney (2006), “The polarization of the U.S. Labor Market”, American Economic Review vol. 96 no. 2, American Economic Association, May, pp. 189-194.

Blinder, Alan (1973), “Wage discrimination: Reduced form and structural estimates”, Journal of Human Resources, vol. 8, no. 4, The University of Wisconsin Press, Fall, pp. 436-455.

Di Nardo, John; Nicole M. Fortin, and Thomas Lemieux (1996), “Labor market institutions and the distribution of wages, 1973-1992: A semiparametric approach”, Econometrica, vol. 64, no. 5, The Econometric Society, September, pp. 1001-1044.

Firpo, Sergio; Nicole M. Fortin, and Thomas Lemieux (2009), “Unconditional quantile regressions”, Econometrica, vol. 77, no. 3, The Econometric Society, May, pp. 953-973.

Fortin, Nicole M.; Thomas Lemieux, and Sergio Firpo (2011), “Decomposition methods in economics”, in O. Ashenfelter and D. Card (eds.), Handbook of Labor Economics, Amsterdam, North-Holland, Elsevier, vol. 4A, p.p. 1-102.

Fournier, Jean-Marc, and Isabell Koske (2012), “Less income inequality and more growth — are they compatible?, seventh part: The drivers of labour earnings inequality — an analysis based on conditional and Unconditional Quantile Regressions”, oecd Economics Department Working Papers, no. 930, oecd Publishing.

Freeman, Richard B., and Lawrence F. Katz (1995), Differences and changes in wage structures, Chicago, University of Chicago Press.

Izquierdo, Mario, and Aitor Lacuesta (2007), “Wage inequality in Spain: recent developments”, European Central Bank, Working Paper Series 781.

Juhn, Chinhui; Kevin Murphy, and Brooks Pierce (1993), “Wage inequality and the rise in returns to skill”, The Journal of Political Economy, vol. 101, no. 3, Chicago, The University of Chicago Press, June, p.p. 410-442.

Krueger, Alan B. (1993), “How computers have changed the wage structure: Evidence from micro data”, Quarterly Journal of Economics, vol. 108, no. 1, Oxford University Press, February, pp. 33-60.

Lemieux, Thomas (2006), “Increasing residual wage inequality: Composition effects, noisy data, or rising Demand for skill?”, American Economic Review, vol. 96, no 3, American Economic Association, June, pp. 461-498.

________ (2008), “The changing nature of wage inequality”, Journal of Population Economics, vol. 21(1), Springer, January, pp. 21-48.

Machado, José A. F., and José Mata (2005), “Counterfactual decomposition of changes in wage distributions Using Quantile Regression”, Journal of Applied Econometrics, vol. 20, no. 4, Wiley, May-June, pp. 445-465.

Mincer, Jacob (1974), Schooling, experience and earnings, Columbia University Press, New York.

Mosher, James S. (2007), “U.S. wage inequality, technological change, and decline in union power”, Politics Society, vol. 35, no.2, pp. 225-263.

Naticchioni, Paolo; Andrea Ricci, and Emiliano Rustichelli (2008), “Wage inequality, employment structure and skill-biased change in Italy”, Labour, vol. 22, pp. 27-51.

________ (2009), “Far away from a skill-biased change: falling educational wage premia in Italy”, Applied Economics, vol. 42 (26), pp. 3383-3400.

Oaxaca, Ronald (1973), “Male-Female wage differentials in urban labor markets”, International Economic Review, vol. 14, no. 3, Wiley, October, pp. 693-709.

* Researcher at the National Council for Scientific and Technical Research (Conicet) at the University of Buenos Aires, Argentina. fgroisman@conicet.gov.ar, groismanf@hotmail.com

1 These jobs make up the segment of highest quality occupations. Because they were declared by employers, they enjoy the effective protection of labor standards, and workers in these jobs also receive higher salaries than those in unregistered precarious positions.

2 The choice of these two years provides a reasonably large interval to analyze distributional changes. It is important to note here that comparing the variables analyzed between the extremes of the period properly reflects the trends that took place throughout this time period.

3 Effectively, the repercussion of certain variables on salaries, for example, the maximum level of education completed, will vary among different individuals. It has been demonstrated that education is more valuable — that is, the coefficient associated to education is the highest — for workers that require education to reach high-income jobs than for workers in low-salary positions. The standard least squares regression techniques ignore this heterogeneity and provide an estimate of the average effect of education.

4 There are therefore still differences in the coefficients estimated for the diverse conditional income quantiles. These differences will express the features that are not being controlled, either non-observable and/or not surveyed. As such, this method provides an intra-group spread measure where it is not possible to extrapolate the determinants of overall inequality.

5 Changes in distributional indicators can be analyzed as the result of modifications that would have taken place to the makeup of the universe being analyzed between two points in time and/or to the salaries associated to the features of salaried workers and the jobs. Moreover, the method has recently been used to break down the distributional differences between countries (Fournier et al., 2012).

6 Estimate made by the author based on eph, Quarter iv, 2011.

7 The classification of jobs by required skills in the categories of: unskilled, operational, technical and professional, is carried out by the National Institute for Statistics and Censuses. This institute seeks to determine the knowledge and skill requirements needed to perform these jobs. In general terms, people who work in operationally skilled jobs mainly require manual skills, while those in technical or professional jobs use more theoretical knowledge. Likewise, the distinction between these last two areas is dependent on the complexity of the tasks performed in each job.

8 Estimates made by the author based on eph and provincial consumer price indices.

9 Data available for interested readers.

10 The alternative explanation to the segmentation hypothesis is centered around the thesis of individual choice. When applied to Argentina, it would imply concluding that a group of full-time private sector salaried workers, around 30-40 percent, would have voluntarily rejected the benefits of a protected job and a higher salary, around 50 percent. The strong job creation shown in the time period and the magnitude of the salary gap mean that this interpretation is not quite convincing. In addition, it could be argued that the influence of unobserved characteristics, like intelligence or skills, which would be expressed in productivity differentials, is an underlying reason for the salary discrepancies between groups. However, this argument would impose the condition of a pretty precise equivalence between productive/non-productive jobs and registered/unregistered positions.

11 In standard salary equations — Mincer models — this coefficient is generally interpreted as the reference salary for a group of individuals against which the comparison is made. In the Oaxaca-Blinder decomposition — like the one conducted for the salary difference between registered and unregistered workers — the interpretation of the salary return on the constant indicates the component of pure segmentation, or discrimination, if the analysis were comparing men and women. In the decomposition in this work, this figure reflects the distributional effect of changes in salaries that cannot be attributed to any of the independent variables.

Licencia de Creative Commons  Problemas del Desarrollo. Revista Latinoamericana de Economía by Instituto de Investigaciones Económicas, Universidad Nacional Autónoma de México
is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License

Published in Mexico, 2012-2019 © D.R. Universidad Nacional Autónoma de México (UNAM).
PROBLEMAS DEL DESARROLLO. REVISTA LATINOAMERICANA DE ECONOMÍA, Volume 50, Number 196 January-March 2019 is a quarterly publication by the Universidad Nacional Autónoma de México, Ciudad Universitaria, Coyoacán, CP 04510, México, D.F. by Instituto de Investigaciones Económicas, Circuito Mario de la Cueva, Ciudad Universitaria, Coyoacán,
CP 04510, México, D.F. Tel (52 55) 56 23 01 05 and (52 55) 56 24 23 39, fax (52 55) 56 23 00 97, www.probdes.iiec.unam.mx, revprode@unam.mx. Journal Editor: Moritz Cruz. Reservation of rights to exclusive use of the title: 04-2012-070613560300-203, ISSN: pending. Person responsible for the latest update of this issue: Minerva García, Circuito Maestro Mario de la Cueva s/n, Ciudad Universitaria, Coyoacán, CP 04510, México D.F., latest update: February 15th, 2019.
The opinions expressed by authors do not necessarily reflect those of the editor of the publication.
Permission to reproduce all or part of the published texts is granted provided the source is cited in full including the web address.
Credits | Contact

The online journal Problemas del Desarrollo. Revista Latinoamericana de Economía corresponds to the printed edition of the same title with ISSN 0301-7036