Essays on how gender equality in pay could be achieved in Bangladesh - an analysis with special reference to sustainable development goals

The gender pay gap is very often thought of as a significant barrier to a country’s development. The United Nations’ sustainable development agenda outlines the obligation of each country in the world to work towards the closure of the gender pay gap by 2030. This has even more importance for Bangladesh as the World Bank in 2019 ranked it as one of the five fastest growing countries in the world, while Bangladesh itself sets a target to become a developed nation by 2041. However, the Global Gender Gap Report 2022 highlights that Bangladesh stands 141st out of 146 countries in economic participation and opportunity. One of the most important reasons for this non-participation is the substantial gap/discrimination that women face in employment/pay. Considering the sex ratio in Bangladesh is 100.3, it may be impossible to reach its development target if the majority of women do not participate in the labour market. Therefore, to determine a holistic policy plan, our thesis explores the current situation of the gender pay gap/discrimination in Bangladesh in the following three papers.  

The first paper examines the shape of the total gender pay gap and its endowment and discrimination effects in Bangladesh throughout the pay distribution for the period 2010–2017. The existing studies on Bangladesh use only the conventional measures, such as mean, median, quantiles, etc. that have considerable limitations. The generalized entropy (GE) measures advocated by Maasoumi and Wang (2019) overcome these limitations. Therefore, following the recommendation of Schirle (2015), we use both the conventional as well as the GE measures in our first and second papers. We use the three latest rounds (2010, 2013 and 2017) of Labour Force Survey (LFS) data from the Bangladesh Bureau of Statistics (BBS) in our studies. However, it is argued in the literature that a woman’s decision to participate in fulltime employment is not random and therefore causes selectivity bias that requires correction to have unbiased estimates of the gender pay gap. Thus, we use the recently proposed quantile-copula method in our first two papers to address the selectivity bias. This quantile-copula method is superior to other similar methods. The GE findings reveal that the gender pay gap diminishes monotonically over the years, which conflicts sharply with results from the widely used mean measure, and this variation justifies the necessity for inclusion of the GE measures. The study also reveals that the gap is more intense in the lower pay levels (i.e., at the 10th percentile). After the selectivity bias is addressed, the relative position of women seems to be that they are worse off. In addition, the quantile-copula method can recover the reservation wage of the non-fulltime employed (for instance, part-time and unemployed) people. If we include these reservation wages in estimating the gender pay gap, in other words, if we value the non-economic activities of women, we find that their relative position improves. To the best of our knowledge, our paper is the first among studies of developing nations to use the GE measures as well as to estimate the gender pay gap that takes into account non-economic activities.  

Our second paper investigates the disaggregated scenarios of the gender pay gap in Bangladesh. There is only one study on Bangladesh that analyses the gender pay gap at the disaggregated level (i.e., by region). However, keeping in mind the importance of disaggregated statistics, the UN Sustainable Development Goals (SDGs) obligate every member country to monitor the gender pay gap by age group and by occupation. From this point of view, the paper attempts to analyse the gender pay gap by human capital variables (such as age group, level of education, training status), and spatial variables (such as region, division) from the supply-side perspective, and by occupation from the demand-side perspective of the labour market. To the best of our knowledge, this type of thorough analysis by a number of important subgroups and using the GE measures has not been conducted on any country. Our results reveal that, before the selection correction, on average women are paid better only at the entry level (i.e., age group 15–29 years) of their career. Another staggering finding is that in general the most highly educated women of Bangladesh suffer from being paid the least relative to their respective male workers. Training, on the other hand, is found to be of considerable benefit. We find that women residing in rural areas are better paid compared to those in urban areas and, more specifically, out of six divisions in Bangladesh, it is predominantly in Dhaka where female disadvantage is clustered. The remuneration of women is better in five categories of high- and mid-level occupations including managers, while that of men is higher in the other four categories of mid- and low-level occupations, such as clerical support workers, plant and machine operators and assemblers, elementary occupations. It is in the elementary occupations where men have nearly outright dominance throughout the distribution and over the entire period. The probable reason for this is that employers prefer to engage men in labour-intensive blue-collar jobs with higher pay. On the other side, the GE findings are mostly inconsistent with those of the conventional measures in terms of magnitude as well as direction of the gender pay gap, irrespective of subgroups. The selection correction, in contrast, delivers homogeneous and consistent results where men take comprehensive control of the pay structure during 2010–2017 for each subgroup and across the entire distribution, regardless of the measuring approach.  

Using the recentered influence functions (RIF) regression method proposed by Firpo, Fortin and Lemieux (2009), our third paper conducts a detailed decomposition analysis throughout the pay distribution at the aggregate as well as at occupational and industry levels over 2010–2017. This method facilitates decomposition of the gender pay gap into endowment and discrimination effects, not only at mean but also at quantiles, and allows identification of the individual factors responsible for a particular gap/discrimination. This method has several other advantages compared to related existing methods, but the extant studies in Bangladesh do not exploit its full advantage. The findings reveal that at the occupational level, the lower tail of the gender pay gap is dominated by the discriminatory attitude, while the upper part is due to discrimination when the occupations are male dominated (i.e., in elementary occupations and professionals) and are due to an endowment difference if the occupations are female dominated (i.e., technicians and associate professionals, and managers) in terms of the mean pay gap. This finding is interesting because women are found to upgrade their relative pay by acquiring more endowments, whereas the achievement of men is for the most part simply a result of discrimination. At the industry level, the discrimination effect is found to be dominant in Manufacturing and the endowment effect is dominant in Transportation and storage, while the dominance of the discrimination effect in Financial and insurance activities in 2013 was surpassed by the dominant endowment effect in 2017. Finally, we find that the factors that explain the changes in the aggregate level discrimination effect do not necessarily explain the occupation/industry level effect in the same way, and the factors in the same occupation/industry impact differently at different quantiles of the distribution.  

Our study has some important policy implications. To ensure that employers follow the related articles of equal pay in the Bangladesh Labour Act (BLA) 2006, the government must monitor gender-based pay structures closely and should focus on Dhaka and Chattogram megacities as well as on low- and high-paying areas and blue-collar jobs like elementary occupations. Moreover, education levels below tertiary and training reduce the gender pay gap, whereas tertiary education enhances the gender pay gap, so post-secondary education that includes training in various areas such as technical, vocational, and information and communications technology (ICT), rather than tertiary education should perhaps be encouraged. The existence of the sticky floor effect in the pay distribution clearly indicates that a minimum wage legislation is a must to introduce in the labour market of Bangladesh. On the other hand, creating enough childcare centres across the country will increase the participation rate in fulltime employment by women and, thus, eventually result in a reduction of gender pay inequality. The constant positive gender pay gap in the lower tail of the pay distribution (i.e., the sticky floor effect) points towards enforcing the minimum wage law in Bangladesh. Finally, though we use the latest and most appropriate methods and the latest data sets, still our study is not free from limitations. We do not analyse the gender pay gap by industry in our second paper due to some constraints, though it is highly recommended in the literature. In addition, we do not consider the selectivity bias that might have occurred due to non-random participation of the women in occupations. We draw attention to these issues as recommendations for future research.