Wage discrepancies in developed and developing countries can vary for a multitude of reasons - some of which are tolerable, while others are not. Wages serve as crucial economic resources for wage workers, and thus, wage inequalities are reflective of economic disparities. They also bear significance for individual and household income; consequently, any form of wage inequity can intensify household income inequalities. Wage inequalities contribute to the uneven distribution of income among a population. These disparities, if deemed discriminatory, can lead to dissatisfaction, reduced productivity, recurrent labor unrest (such as strikes or industrial actions), calls for wage negotiations, high labor turnover, and even brain drain in some instances. Both theoretical and empirical data indicate that some of the factors causing wage differentials across the world are discriminatory. For instance, it qualifies as labor market discrimination when wage discrepancies occur among workers with similar qualifications and job roles due to their gender, race, or whether they work in the private or public sector. Such wage differentials have been a focal point for public policy over the past decades, with labor market policies being especially concerned if wage gaps widen due to unequal economic development (Ayentimi et al. 2020) and certain forms of deliberate or systemic discrimination. Worldwide, it is widely acknowledged that pay differences are influenced by factors such as firm-level productivity and wage-setting power (OECD 2021), the level of workers' skills or human capital, geographical location (Pritchett and Hani 2020), gender, unionization, the level of the minimum wage (Teulings 2003; Kristal and Cohen 2017), race, and the private-public segmentation of the labor market.
In the global north, wage inequality trends in the United States (US) and most European countries have been mixed since the 1980s. While the US saw a surge in wage inequalities, Europe experienced a mix of reasonably stable, rising, and falling wage trends due to the efforts of egalitarian institutions in the EU that prevented wage adjustments (Krugman 1994). In part, the rise in wage inequality in the US and Europe can be attributed to the demand for skilled workers and returns to education – consistent with the human capital theory by Adam Smith (Orazem and Vodopivec 1995; Van Reenen 2011). On the other hand, institutional factors such as unionization and minimum wages are believed to be associated with wage inequality because, historically, unions have had a balancing effect on the distribution of wages (Chaykowski and Slotsve 2002; Freeman 1980; Freeman and Medoff 1984). The increase in wage inequalities is believed and indeed estimated to be a result of a decline in union membership or unionization (Fortin and Lemieux 1997).
From the perspective of the global south, the literature on wage inequalities tends to lean on the classical theory of demand-supply and the Heckscher-Ohlin model to explain the increase in wage inequalities in developing countries. Scholars have posited that the openness of developing countries to trade has led to a rise in skilled premia, subsequently resulting in wage inequalities (Behar 2016; Khalifa and Mengova 2010). The focus of literature on wage inequalities in the global south appears to be on the effects of international trade on wages and wage inequalities. Could it be that wage inequalities in the economies of global south countries are not a function of institutional factors like minimum wages and unionization or human capital factors such as experience, education, job training, and the like? If not, why are the contributions of unionization, minimum wage, gender, race, and human capital to wage inequalities in many developing countries (including Ghana) not widely researched, even though these analyses are needed to inform policy proposals (Blunch and Verner 2004)? South Africa has the majority of the literature on the effects of race and unionization on wage inequalities in Africa. Perhaps due to its long history of Apartheid, research on wage inequalities along the lines of race, gender, unionization, and sector is significantly represented. Studies by Kerr and Wittenberg (2021) have examined how unions in South Africa raise wages and affect wage inequalities. Similarly, studies by Schultz and Mwabu (1998) show that union workers receive relatively higher wages than non-union workers in South Africa.
In relation to gender wage gaps, Asia has the most significant share of gender wage literature in developing countries – whereas Africa has the least (Khalid 2017). Most African wage inequality studies reveal that historically, females have received lower wages than males (Appleton and Hoddinott 1999; Boahen and Opoku 2021; Bhorat and Goga 2013; Nordman and Wolff 2009; Ntuli and Kwenda 2020). Clearly, while wage inequality has been a concern for global development, not much empirical evidence has been gathered on the issue in Africa, compared to developed countries. An important question that arises is where Ghana fits into this literature. Are labor market outcomes such as wages, wage inequalities, and public sector employment a function of human capital (experience, education, job training, etc.), institutional factors (unionization and minimum wage), or discrimination, and to what extent? Do these outcomes vary in Ghana along gender lines and across sectors (private/public)? This paper aims to provide answers to some of these questions.
Ghana's escalating public sector wage bill has led some analysts to suggest that the potential overpayment of public sector workers relative to their counterparts in the private sector with the same level of human capital and job characteristics could be the explanation. Younger and Osei-Assibey (2017) propose that public sector workers should receive similar wages as their colleagues in the private sector with the same qualifications to reduce the public sector wage bill to sustainable levels. According to their findings, on average, private sector workers earn lower wages than their colleagues in the public sector with similar qualifications. Furthermore, nearly two decades ago, Vernor (1999) proposed that productivity and wages were mismatched in Ghana. Thus, it remains as relevant today as it was then to investigate whether the Ghanaian employee is paid based on his or her "productivity" as measured by the stock of their human capital, experience, education, training, etc., or job characteristics. Or is the compensation structure discriminatory? And what factors determine wages and public sector employment for paid employees? This is also important since some public sector employees in recent times have complained about their wages – arguing that their wages do not correspond with their effort/productivity (Acquah 2022).
Previous studies on wage inequalities in Ghana have focused on understanding the effects of trade openness and skilled-bias technological change on wages (Görg et al. 2001), the gender-wage gap in the manufacturing sector (Abegaz and Nene 2018), and on examining gender-wage gaps using different selection models – correcting for double selection bias (Boahen and Opoku 2021). However, no study has investigated the private-public wage gap in Ghana using the Blinder-Oaxaca decomposition method. Many of the few wage gap studies in Ghana have primarily considered the gender wage gap, with only a handful of them focusing on sector wage differentials. As a result, there is relatively little empirical evidence on private-public sector wage differentials – 'sector wage gap' studies in the country. In relatively recent times, strides have been made by Younger and Osei-Assibey (2017) in investigating whether public sector workers were paid more than private sector workers, controlling for human capital and other observable characteristics. Their study broadly considered both formal and informal public and private sector workers. However, the informal sector in many economies, including Ghana, unlike the formal, lacks the laws and regulations to demand responsibility and compliance from its actors (Kanbur 2009). Therefore, for the comparability and validity of results, the wage gap must be estimated only among formal private and formal public sector workers. More studies are also needed to strengthen the internal and external validity of such empirical findings, especially because private-public sector wage gap analysis is problematic in Africa (Khalid 2017; Duodu 2019).
This paper contributes to the literature by limiting the estimation of the sector wage gap to formal private and formal public sector workers using the Blinder-Oaxaca decomposition method. It also controls for education, sector, gender, and their interactions when estimating the wage levels of workers. I examine the factors associated with the wages of private/public sector workers and estimate the private-public wage gap in Ghana. I use data from the seventh round of the Ghana Living Standard Survey (GLSS7) – a nationally representative household dataset on living conditions in Ghana – to investigate the factors associated with wage inequalities among paid employees. The GLSS captures wage workers' information, making it the most appropriate dataset for my study. This is mainly because, since 2017, there has been no major public policy to address wage inequalities in Ghana; thus, the characteristics of wages, wage inequalities, and public employment remain largely unchanged in the Ghanaian labor market.
Data and descriptive statistics
The data used in this paper were derived from secondary data from the most recent nationally and regionally representative household survey – The seventh round of the Ghana Living Standard Survey (GLSS7). It is a nationally representative household dataset on the living and economic conditions in Ghana. The sampling design employed was a two-stage stratified sampling design. Based on the ten previous administrative regions, one hundred thousand enumeration areas (EAs) were assigned to each region using the probability proportionate to size (PPS) and served as the primary sampling units (PSUs). In the second stage, a total sample size of 15,000 households was realized after selecting 15 households per EA across the nation. The data include information from 14,009 households who were successfully interviewed. Out of 51,295 individuals interviewed, 31,305 were within the legal working age (15 to 60 years), since the minimum age for admission into employment in Ghana is 15 years, and the retirement age is 60 years. Given that the study focuses on measuring wages and considering the challenges related to gauging the wages of unpaid employees, the emphasis was on the wages reported by paid employees.
Therefore, to obtain the analytical sample, only respondents who were between the working ages of 15 to 60 were included. In the end, the analysis was based on 876 observations drawn from the pool of paid employees in the private and public sectors who had received wages or salaries within the Greater Accra Metropolitan Area (GAMA). The key variables used in the study include wages, sector, gender, education, flexible work, and job security. The wages variable is obtained from a question asking respondents to indicate how much payment they received for their job. The payments were in different time units, but they were all converted into monthly payments. In deriving the sector variable, this research used a question that asked for the primary sector respondents were working and derived a dummy variable; the dummy sector variable takes the value of 1 if public and 0 if private.
The public sector in this study is defined to include all workers engaged in the government sector, including parastatals. The private sector constitutes workers who were reported to work in private entities, NGOs (local and international), cooperatives, and international organizations. Also, in this study, a formal sector employee is considered to have a legally enforceable contract and is a contributor to the Social Security and National Insurance Trust. Gender is defined based on the sex of the individuals; as a dummy, it takes 1 for males and 0 for females. Respondents were asked to provide their highest level of education attained. From that question, a variable measuring respondents' educational attainment was generated. The education variable has three values – basic education (kindergarten, primary, JHS/JHS, middle school), secondary education (SSS/SHS, secondary, Vocational/Technical), and tertiary education (Teacher/Agric/Nursing training college, polytechnic, university, professional).
Empirical model
This study adopted the wage decomposition method proposed by Blinder (1973) and Oaxaca (1973), the wage equation is stated as
$${l}{n}{{W}}_{{i}}= {\alpha }+ {{\delta }{P}}_{{i} }{+ {\beta }{X}}_{{i}}+{{\epsilon }}_{{i}}$$
The \({W}_{i}\) denotes the monthly wage of the i-th worker, \(ln{W}_{i}\) is the natural logarithm of the monthly wage of worker i. \({P}_{i }\)is a dummy which indicates whether an i-th worker is a public sector worker, \({X}_{i}\) is a vector of the individual characteristics or the control variables that include age, gender, education, and regular permanent job. Also, \(\alpha , \delta , and \beta\) are the vector of coefficients. The last term is the disturbance term \({\epsilon }_{i}\). Thus, the terms in the model, on the one hand, explain the differences in the pay structure between the two sectors (discrimination) and, on the other hand, explain the differences in the endowments of employees irrespective of the employment sector.
This study adopts the two-step estimation proposed by Heckman (1979) to correct selection bias by including the correction terms into the wage equation.
$${l}{n}{{W}}_{{i}{p}}={ {\alpha }}_{{p}}{+ {\beta }{X}}_{{i}{p}}+{{\epsilon }}_{{i}{p}}$$
$${l}{n}{{W}}_{{i}{v}}={ {\alpha }}_{{v}}{+ {\beta }{X}}_{{i}{v}}+{{\epsilon }}_{{i}{v}}$$
The subscripts p represents the public sector and the v, private sector. Using the Ordinary Least Squares (OLS), the study assumes that the \({{\epsilon }}_{{i}{p}}\) and \({{\epsilon }}_{{i}{v}}\) have no conditional mean for the OLS to be an un-bias estimator. Hence, after correcting for the selection bias, the wage gap between employees in the private and public sectors becomes.
$$\overline{ln\left({W}_{p}\right)}-\overline{ln\left({W}_{v}\right)} ={( \stackrel{-}{X}}_{p}{ - \stackrel{-}{X}}_{v}){\widehat{\beta }}_{p} +({\widehat{\beta }}_{p} - {\widehat{\beta }}_{v}){ \stackrel{-}{X}}_{v}+({\rho }_{p}{\gamma }_{p} - {\rho }_{v}{\gamma }_{v})$$
Rho (\(\rho )\) is the coefficient of Mill’s ratio term gamma (\(\gamma\)). The term \({( \stackrel{-}{X}}_{p}{ - \stackrel{-}{X}}_{v}){\widehat{\beta }}_{p}\) is the portion of the wage gap that is explained by the endowment characteristics of the individuals \(({\widehat{\beta }}_{p} - {\widehat{\beta }}_{v}){ \stackrel{-}{X}}_{v}\) is the portion of wage difference that is unexplained and due to discrimination, whereas the term \(({\rho }_{p}{\gamma }_{p} - {\rho }_{v}{\gamma }_{v})\) reflects the gap as a result of selection bias.