Gender Differences in the Labour Market Status, Wages and Occupations in Pakistan

Pakistan’s economy is facing the lowest female participation rates compared to the developed and other South Asian economies. Accordingly, there is an acute need for in-depth analysis of the role of women in Pakistan’s labour market. The thesis consists of three essays covering three important issues in the labour market of Pakistan. The first essay provides a comprehensive profile of labour market states of working and not working females and compares it with the male counterparts. The working and not-working groups are further enumerated into four categories each. Once these states had been defined, the demand and supply factors have been identified to capture the effect on the individual’s decision about labour market participation. Since, the dependent variable is discrete having four possible outcomes which are categorical, a Multinomial logit model is applied first by taking working states of females as the dependent variable against the explanatory variables and then considering not working states as an outcome. The same procedure is repeated for males. Given that the MNL model relies on the assumption of independence of irrelevant alternatives (IIA) two tests of IIA Hausman-Mcfadden (HM) test and the Small-Hsiao (SH) test have been performed. The main findings show that age has a positive and significant impact on all the states of working males and females in labour market with the exception of men as unpaid family helpers. For females being married, having more than 2 children, ownership of the house, residence within a joint family or belonging to an urban area, reduces the likelihood to participate in the paid employment. Conversely, for married men, or those who own a house or live in an urban area there exists a higher probability to be involved in paid employment. Education has a significant higher probability for females to participate in all the working states of employment. The second essay explores the gender wage gap. A counterfactual decomposition approach of Oaxaca-Blinder (OB) is applied that divides the wage differential into explained and unexplained components. As the wage structure is mainly influenced by working individuals which might make a selective group leading to biased and inconsistent results. Therefore, the estimates from probit regression equations estimating the probability of paid employment are used to construct the Inverse Mills Ratio (IMR) to correct the selection bias in the monthly wage equations. Number of infants and children (aged 5 or below and 6 to 10) serve as main instruments to identify the selection equation. The empirical findings suggest the existence of the gender wage gap in Pakistan. Individual’s age, level of education, sectors, occupations and regions are the key determinants. The decomposition results shows that the explained component is 41% and unexplained is 59% without taking into account selectivity. However, with the presence of selection effects in the wage decomposition equations the results are upward biased explaining 39% endowment effect, 77% coefficient effect and -16% selection effect. The extent of occupation differences across gender and regions is explained in the third essay. For a comprehensive spatial analysis, the occupational gap between males and females within nine occupations has been estimated separately for an overall Pakistan, its four provinces and respective districts by using the non-linear decomposition method. The empirical findings indicate that in the low paid jobs (such as clerks, sales persons, skilled workers in agriculture and fishery, craft and trade workers, plant and machinery operators and unskilled or elementary occupations) a major part of the gender differential is attributed to differences in the coefficients indicating substantial differences in attitudes towards males and females. However, almost 50 percent of the differences in high earning jobs (such as professional and senior officials) are explained by different characteristics. Pooled data constructed from the PSLM (2004-09) surveys has been used consistently throughout the thesis due to the fact that the chapters are inter-related. Nevertheless, the sample size and number of observations differ in each of the chapters on the basis of their specific objectives