Returns to Education and Earning Inequality Nexus: a Microeconometric Analysis for Pakistan

This study intends to shed light on the nexus between returns to education and earning inequality in Pakistan. For this purpose, the study utilizes two estimation methods namely ordinary least square and quantile regression method to demonstrate how returns to different level of education contribute towards earning inequality. The results show that education plays a significant role in determination of within group earning inequality at all level of education. Within group earning inequality is higher within the individuals having tertiary education as compared to the individuals having secondary and primary education. Inter-temporal analysis shows that the earning inequality does not remain constant within the education groups during 2005-07. Moreover, education also causes earning inequality between educational groups. The findings of the study reveal that education has a positive effect on within as well as between groups earning inequality. INTRODUCTION whether returns to education for individuals at upper tail


INTRODUCTION
whether returns to education for individuals at upper tail Distribution of earnings has long been an area of education for those who are at lower tail of earning interest among the economists. Initially, concentration distribution with same level of education. If there is a was focused on relationship between earnings inequality difference then it is concluded that earning inequality and economic development. The most important idea is present due to education. The presence of such on this relationship was established by [1]. Since the differences in returns to education has obvious work of Kuznets, there has been an interest in implication for the labor market. understanding the determinants of distribution of The main objective of this study is to examine earnings. The current literature on inequality underlines relationship between returns to education and earning education as a contributing factor towards earning inequality in Pakistan. The present study contributes to inequality [2][3][4]. Human capital model illustrates that level the existing literature on earning inequality by analyzing and distribution of education determines earning the returns to different levels of education as a source of distribution of a society [5]. earning inequality. To this end, earning distribution is The contemporary research on returns to education characterized by using two methods namely ordinary least has exposed that there exists a relationship between squares (OLS) and quantile regression methods. OLS returns to education and earnings inequality [6][7][8].
method assumes that the marginal effect of education on They have taken into account the heterogeneity in earnings is constant over the earning distribution. earnings due to education. To provide the evidence on Therefore, the effect of an extra year of education can be whether education is a contributing factor towards represented by a right word shift of the conditional earning inequality or not, they provide a distributional earning distribution. On other hand, quantile regression analysis of earnings of educated workers. In doing so, method measures effect of education on earnings at they utilize quantile regression to understand that of earning distribution are different from returns to World Appl. Sci. J., 24 (7): [885][886][887][888]2013 886 different parts of the distribution. As a result, it can describe changes not only in the location but also in the shape of the earning distribution. By combining the two methods, we attempt to measure the effect of education where x is the vector of explanatory variables and is the on earnings inequality within and between groups.
vector of parameters and u is random error term. That is, OLS returns measure the difference between Quant (y -x ) denotes the conditional quantile of y education groups, while differences in quantile returns given x. It is assumed that. In addition, it is also assumed show the earning differences between individuals having that above model is correctly specified and y and x are the same level of education but are located at different observed with no errors. The quantile regression, quantile. 0 < < 1, is defined as a solution to the problem: The layout of study is as follows: section 2 provides a brief review of literature on subject matter of the study. The empirical specification of earning function based on quantile regression is presented in section 3. Results and discussion are given in section 4. Finally in section 5, concluding remarks are provided.
By variation of , any quantile of the conditional To examine effect of different levels of education on [11] found the relation of schooling and wage earning, we estimate the extended Mincerian earning inequality for the period 1994-2001 in Spain. They found function. The empirical earning function is specified as that higher education was related with higher wage follows: inequality.
[12] estimated returns to education in urban Argentina for period 1992-2002. The results showed that men at higher quantiles had higher returns to education compared to those at the lower quantiles. For women where, is quantile being analyzed, LnW is the natural log returns were highest at the lowest quantiles. Moreover, of monthly earnings for the ith individual. Primary, the returns to education had increased during the period secondary and tertiary refer to dummy variables for under study.
primary, secondary and tertiary education. These where, x is completed years of schooling of the It reveals that effect of each level of education on individuals. Primary is equal to 1 if individual has education from 1 to 5 years of schooling and zero otherwise. Secondary is equal to 1 if individual has education from 6 to 12 years of schooling and zero otherwise while tertiary is equal to 1 if individual has education higher than 12 years of schooling (or higher than secondary education) and zero otherwise. No education is omitted category here. Z includes labor market experience, square of labor market experience, dummies for gender, marital status, region of residence, occupation and province of residence. The above specified earning model is estimated at nine deciles of conditional earnings distribution. The standard errors of estimates are obtained by bootstrapping with 100 repetitions. In addition to quantile estimations, we also perform OLS regression.
The study uses Pakistan Social and Living Standards Measurement (PSLM) Survey data for the period 2005-06 and 2007-08. Keeping in the view the standard definition of labor force, only individuals ranging from age 15 to 65 are kept in the samples.

RESULTS AND DISCUSSION
The specified earning function has been estimated at nine deciles for each of the two years. Only coefficients of primary, secondary and tertiary dummies in the earning function and their respective t-statistics are presented in Table 1. These results show that effect of each level of education on earnings is positive and statistically significant at each of the deciles analyzed for both of the years. Returns to each level of education are not equal at each decile. In other words, each level of education has a different effect on earnings across earning distribution. earning increases as we move from lower to upper quantiles of the earnings distribution. It implies that there is heterogeneity in returns to each level of education. However, this heterogeneity in returns for primary and secondary education is less as compared to tertiary education. The results show that effect of education at upper tail of earning distribution is higher than at the lower tail of earning distribution in the two years. Therefore, we can conclude that education is a factor which promotes within group earning inequality.
Changes in the returns to education for different levels of education over time can also be observed from the table. Furthermore, in both the years, returns to education are convex that is, effect of education on earnings tends to increase as the level of education increases. These returns are higher for tertiary education as compared to primary and secondary levels across all quantiles of earnings distribution. The pattern of higher returns as level of education becomes higher is also obvious by the OLS results. This confirms that education also causes earning inequality between different education groups.