Effect of Women’s Paid Work Status on Their Decision-Making Autonomy: A Systematic Review and Meta-Analysis

Autonomy and Its Dimensions

The idea of women’s autonomy began with the concept of women’s status, initially conceptualized by Mason and Smith (2000) and then explored further by other researchers. Jejeebhoy (2000) defined autonomy as women’s control in various domestic spheres; Khan and Ram (2009) defined autonomy as the capacity to achieve well-being through the role of decision making; Haque, Islam, Tareque, and Mostofa (2011) defined women’s autonomy as their ability to make choices or take decisions in the household relative to their husband; and Tesema et. al. (2021) described autonomy from the perspective of healthcare decision making.

This shows that the concept of autonomy is multidimensional (Bayissa, Fitsum, Smits, & Ruben, 2018; Jejeebhoy & Sathar, 2001; Kishor, 1993; Malhotra & Mather, 1997; Rammohan & Johar, 2009) and therefore difficult to quantify. Traditionally, different societies have accorded different meanings to this concept. For example, in India, women’s autonomy is usually defined in the domestic sphere (Bloom, Wypij, & Gupta, 2001), while control of fertility and contraceptive use (Dyson & Moore, 1983; Jejeebhoy, 1995), freedom from domestic violence (Koenig, Ahmed, Hossain, & Mozumder, 2003), and bargaining power between couples in household expenditure (Beegle, Frankenberg, & Thomas, 2001) are also taken as dimensions of women’s autonomy. In addition, the ability to decide for oneself or the ability to influence other household members’ decisions are also considered important indicators of women’s autonomy. In other research, the dimensions of women’s autonomy have been categorized under three broad headings: economic autonomy, decision-making autonomy, and physical autonomy or mobility (Abedin & Arunachalam, 2021; Bloom et al., 2001; Jejeebhoy, 2002).

The fact that autonomy is frequently used as a term in the psychological literature is pertinent, as is its interchangeability with the term “bargaining power” in the economic literature. According to Donald, Koolwal, Annan, Falb, and Goldstein (2017) in a World Bank Policy Research working paper, autonomy should be distinguished from independence, as people may depend on others to support their autonomy. On the other hand, bargaining power is essentially related to one’s ability to influence decisions and act according to one’s preference in the psychosocial sphere of life and in relation to tangible material matters of utility to the decision-maker. Hence, decision-making autonomy can be said to be the most important part of women’s autonomy overall.

Paid Work Status of Women in the Developing World

In the developing world, the relationship between economic growth and women’s workforce participation is not linear, but rather ‘U’ shaped (Assaad, Hendy, Lassassi, & Yassin, 2020). Some countries have experienced a decline in the rate of women’s workforce participation despite increasing income, education, and overall social development. Verick (2014) concludes that higher education (beyond secondary level) is critical in employment outcomes, along with social norms and job opportunities. The International Labor Organization (ILO) associates women’s participation in the labor market with development, education, fertility rates, and access to child healthcare. Notably, these are among the most important determinants of women’s decision-making autonomy. The ILO has observed a puzzling trend in the labor force participation pattern of women across South Asian countries. To this end, they commissioned a Decent Work Team for South Asia in 2012 to conduct further research on women’s workforce participation and its complex relationship with economic and empowerment issues.

The Complex Relationship between Labor Force Participation and Women’s Autonomy

Theoretically, paid employment should improve women’s financial and decision-making autonomy (Buller et al., 2016; Cornish et al., 2021) and empower them through self-esteem and identity to enjoy utility from their preferences (Riley, 1998). This is supported by classical economists, who mention that in any socio-economic norm, an earning woman is placed in a potentially better position to participate in household decision-making (Dixon-Mueller, 1993; Kishor, 2000; Sen, 1987). However, many literature reviews have shown this relationship to be inconclusive. For example, the study by Dharmalingam and Morgan (1996) commented that women’s work in two southern Indian villages did not give them autonomy directly, despite giving them bargaining power. In a number of South Asian studies, researchers observed that even the highest-earning female members do not have economic autonomy or economic decision autonomy over the wages they earn (Philips, 2003; Samarasinghe, 1993). Similarly, Ristiana and Handayani (2018) stated that the use of labor force participation as a proxy for women’s autonomy may not be beyond criticism as in many cases, women’s paid employment did not improve their autonomy. This argument is especially likely to be true in patriarchal societies like those in South Asian and African countries, (Beath, Christia, & Enikolopov, 2013).

Aim of the Present Study

In the context of this discussion, it is important to examine the relationship between women’s work and their decision-making autonomy. With this purpose, we have thoroughly searched the existing literature to explore the determinants of women’s decision-making autonomy where women’s paid work status has been included in the scope of the study. As studies exclusively discussing these two variables are scarce, we have also extracted data from studies that have incorporated women’s paid work status among many other determinants of autonomy.

The primary aim of this review is to determine 1) the relationship between women’s decision-making autonomy and their paid work status, 2) the relationship of other major determinants of autonomy with women’s paid work status and their importance, 3) the relative importance of the effect of paid work status on women’s autonomy, in reference to other determinants, and 4) the effect size of paid work status using meta-analysis and compare it with the effect size of other major determinant(s) on women’s decision-making autonomy.

As discussed earlier, the ILO has observed the “South Asian enigma” regarding women’s workforce participation and empowerment in South Asian countries and recommended further research in this area. Hence, while conducting our meta-analysis, we need to explore the contrast between South Asian and non-South Asian countries in terms of the effect of women’s paid work status on their autonomy. As per the ILO (International Labor Organization, 2012) and the research by Manjula (2021), such a contrast between South Asian and non-South Asian countries exists because of the fact that, in spite of showing promising economic development, women’s status in both labor force participation and intra-family autonomy are declining in South Asian countries. Hence, the secondary objective of this study is to conduct a subgroup analysis of South Asian countries to identify how paid work affects women’s decision-making autonomy differently in these countries.

Sources of Information

Major Internet databases, including PubMed, JSTOR, Web of Science, IBSS, ASSIA, the Social Science Research Network (SSRN), and EconStor, were extensively searched. We also searched for relevant articles using Google Scholar, and the reference lists of identified articles were thoroughly checked for further relevant articles (Higgins et. al., 2019). The Comprehensive Pearl Growing technique was used in the initial search phase (Shojaati & Osgood, 2021) and at the end, and expert consultation was availed of to ensure that the search process was complete and concise. Using Mendeley reference management software (Simarmata, Dewi, Sila, Sele, & Shidik, 2021), all relevant studies were downloaded and duplicate references eliminated. Table 1 shows the different search techniques used in the study and the number of selected studies identified using each search technique.

Table 1

Source of Studies Included in the Review

Screening and Selection

Searches were carried out using the following keywords and their synonyms: “working women,” “women autonomy,” “women in labor force,” “developing countries,” “women decision,” “women employment,” and “paid work.” The process followed for the selection of the studies is presented in the PRISMA flowchart (Rethlefsen et. al., 2021), as shown in Figure 1. The total time taken to conduct this systematic literature review was approximately 14 months. The major activities involved in developing this study began with the preparation of the study protocol, which was started in April 2020 and took approximately two months. Thereafter, searching for all relevant online and offline sources, screening of the studies, and eligibility testing followed by inclusion assessment took approximately six months. After that, data sieving, data entry, and follow-up of missing information involved another two months, and analysis of data and preparation of the review report took another three months. As the study is by nature a systematic review, endeavoring to keep the review up to date by incorporating the latest relevant studies was a major challenge and caused overlapping in many of the aforementioned steps (Higgins & Green, 2011).

Figure 1.

Flow of Search Strategy (based on PRISMA statement template).

Initially, 326 documents were identified using various search techniques, of which 16 were removed as duplicates, and a further 243 studies as commentaries, opinions, editorial articles, discussions, review articles, or conference proceedings without full papers. The full text of the remaining 67 articles was assessed, and a further 45 were discarded, either for not having women’s decision autonomy as the outcome variable or for not incorporating women’s labor force participation as a variable in the study. The final set of 22 articles remained and were included in the qualitative synthesis. Among them, three articles had three sub-analyses of three different samples taken from different geographical locations (countries), giving a total sample size of 28 (n=28).

Since this systematic review measured the effect size of women’s decision autonomy, a thorough quality assessment was performed for the studies included in this review. For this purpose, the Critical Appraisal Skills Programme (CASP) checklist was used (Nematy, Namer, & Razum, 2022)—the CASP qualitative studies checklist (10 points) for qualitative studies, and the CASP cohort study checklist (12 points) for cohort studies, as both were found to be a good fit for inclusion in this systematic review. The checklists were appraised for three broad issues: (a) Validity of the results of the study (Are the results of the study valid?), (b) Appraisal of the results (What are the results?), and (c) Generalizability (will the results help locally?).

The inclusion criteria used to select studies for this review were as follows:

• 1. Original articles
• 2. Articles based on developing or low- and middle-income countries only
• 3. Women’s autonomy as the outcome variable, showing relationship with women’s employment or paid work
• 4. Articles published in or after 2000
• 5. Studies published in English.

Meta-analysis Method

Study Characteristics

A total of 22 articles were included in this review, of which three were conducted in three countries with different samples (see Table 2). For ease of analysis and explanation, each of those study was numbered separately, giving a total of 28 samples. These studies were conducted in 13 countries (see Table 4), of which three belong to the low-income group, eight to the lower-middle income group, and two to the upper-middle income group. Fifteen of the 28 are from South Asian countries. Only three of these studies used both quantitative and qualitative methods, and the remaining 25 used quantitative data only. In eight studies, primary data were collected through questionnaires, interviews, or a combination of both, and 20 studies used secondary data from national demographic and health surveys.

Table 2

Characteristics of the Included Studies(n=28)

Dimensions of Autonomy

Most studies have discussed women’s autonomy under three broad headings: decision autonomy, economic autonomy, and physical autonomy/mobility. All three dimensions were discussed in eight studies with ten samples. Apart from these three dimensions, one study introduced personal autonomy in three samples and family autonomy in two samples (Heaton, Huntsman, & Flake, 2005). One study measured women’s autonomy in various dimensions of the purchase of household items (Anderson & Eswaran, 2009), which is analogous to economic autonomy. Jejeebhoy and Green (2001) added freedom from threat as a component of autonomy, whereas Vaz, Pratley, and Alkire (2016) discussed autonomy as part of psychological well-being (see Table 2).

Components of Decision Autonomy

As previously mentioned, most studies have included decision autonomy as a component of women’s overall autonomy. Some studies have distinguished between “economic autonomy” and “autonomy of economic decision making” (Jejeebhoy & Green, 2001). The autonomy of economic decision making involves the purchase of daily household goods and making decisions about large purchases such as jewelry or land (Hakim, Salway, & Mumtaz, 2003; Jejeebhoy, 2002; Kabeer, Mahmud, & Tasneem, 2018; Lamidi, 2016; Regassa & Regassa, 2016). On the other hand, economic autonomy is described as having access to and control over available economic resources, such as owning and controlling the mobility of household valuables, planning for savings, and having a bank account (Bloom et al., 2001; Jejeebhoy, 2000; Samari & Pebley, 2018; Sathar & Kazi, 2000). The possession and disbursement of dowry is clearly contradictory to women’s empowerment, but studies show that it entails economic autonomy for married women (Jejeebhoy, 2000). Hence, the autonomy of economic decision making is observed to be a more empowering term than economic autonomy. Taking decisions about one’s own and one’s children’s health has been noted as another important aspect of decision-making autonomy and is discussed in detail in several studies. Making decisions about a child’s health not only raises a woman’s importance within the family, but also emphasizes her intellectual ability (Alemayehu & Meskele, 2017; Bloom et al., 2001). Seven studies have cited health decision-making autonomy as a major component of women’s autonomy (Alemayehu & Meskele, 2017; Hakim et al., 2003; Jose, 2008; Kamiya, 2011; Lamidi, 2016; Osamor & Grady, 2018; Senarath & Gunawardena, 2009) and some studies have developed indices of decision autonomy from different components to use as a continuous variable (Bloom et al., 2001; Sultana, 2011) (see Table 2).

Addressing Workforce Participation

Only one study has dealt exclusively with women’s autonomy and work as its primary objective (Ristiana & Handayani, 2018), highlighting the paucity of discussion of women’s work in the context of their autonomy. However, while exploring the factors of women’s decision autonomy, 19 of the 28 studies found a statistically significant positive relationship between women’s workforce participation and their autonomy in decision-making and another 4 showed an insignificant relationship (Bolivia sample of Heaton et al., 2005; Kamiya, 2011; Regassa & Regassa, 2016; Sathar & Kazi, 2000). Three studies mentioned that paid work is a significant contributor to autonomy, as opposed to unpaid work (Hakim et al., 2003; Kabeer et al., 2018; Ristiana & Handayani, 2018) and only one study stated that women’s employment has a significant negative impact on women’s decision-making autonomy (Samari & Pebley, 2018). Table 3 shows the detailed characteristics of the relationship between decision autonomy and the paid work status of women.

Table 3

Characteristics of the Included Studies(n=28)

Other Determinants of Autonomy

The search involved the paid work status of women for each result. These studies mentioned several other factors and their degree of association with women’s autonomy. The two major factors associated with this are women’s education in terms of secondary or higher education, and age. A total of 17 (60.71%) studies included education as a demographic covariate of women’s autonomy, of which 12 (42.9%) found it to be significant, one insignificant, and four found it to be inconsistent, that is, not significant in all geographic locations.

Women’s age was found to be another important factor; that is, higher age was associated with better autonomy. This was included in 16 studies, and it was observed as significant in 10, insignificant in four, and inconsistent in two studies. Family wealth, husband’s education, residence, household size, religion, parity, husband’s age, natal kinship, and women-headed families were other major factors associated with women’s autonomy (see Table 3).

Husband’s Education and Occupational Status

Since this study discusses women’s work as a primary determinant of autonomy, to which women’s level of education has been found to be the second-most important contributor, the husband’s education and occupation status have also been included in studies of married female respondents. Although eight studies have considered husbands’ education as a factor in women’s decision autonomy, only two studies show that it is a significant positive contributor to women’s autonomy (Alemayehu & Meskele, 2017; Kamiya, 2011), suggesting that educated husbands are more open to concepts of gender equality and give equal importance to their partners’ involvement in independent decision making. However, three studies showed an insignificant effect of husbands’ education on women’s decision autonomy (Heaton et al, 2005; Regassa & Regassa, 2016; Sathar & Kazi, 2000), two showed inconsistent effects (Anderson & Eswaran, 2009; Samari & Pebley, 2018), and one even showed a negative effect (Ristiana & Handayani, 2018). Due to such wide variation in the perceived effect of husband’s education on women’s decision autonomy, meta-analysis with husband’s education was not performed to measure effect size (see Table 3).

Only one study has taken husband’s occupation as a determinant of women’s decision autonomy (Osamor & Grady, 2018); however, the result shows a negative effect, suggesting employed husbands are more restrictive in relation to their partners’ decision autonomy (see Table 3).

Outcomes of the Meta-analysis

Out of the 12 studies selected for meta-analysis, 11 were found appropriate for calculating log odds ratios and standard errors for meta-analysis of decision autonomy (autonomy vs. no autonomy) as the dependent variable and women’s paid work status (paid employment vs. no paid employment) as the independent variable. The overall effect is highly significant (p < 0.001) with a net odds ratio of 1.26 (95% C.I.= 1.17-1.36). However, the heterogeneity estimates of the effect sizes of all eleven studies were high (I2= 90%, p < 0.001). Hence, a subgroup analysis was conducted based on the geographical location (South Asian countries vs. non-South Asian countries). Since the peculiarity of women’s paid work status in South Asian countries in contrast to non-South Asian countries has been discussed, the choice of this subgroup was made accordingly.

Table 4

Summary of observations(n=28)

In the subgroup analysis, five samples were from South Asian countries, of which three were from the same author (Senarath & Gunawardena, 2009) with different regression models for samples from Bangladesh, India, and Nepal. There were six non-South Asian studies, four from African countries, and one each from Indonesia and Tajikistan, both Asian countries. The South Asian subgroup again showed high effect heterogeneity (I2=96%, p < 0.001), which explains the high variability of the relationship between decision autonomy and women’s paid work status in South Asian countries. This might be due to regional differences in societal norms or differences in the research methodology. However, the odds ratio of decision autonomy in working women was statistically significant (OR=1.30, 95% CI= 1.16- 1.45, p < 0.001) and similar to the net odds ratio of all 11 studies. In contrast, the subgroup of non-South Asian studies had significantly higher homogeneity (I2=49%, p=0.08) and the total effect size for this subgroup was also statistically significant (OR=1.21, 95% CI= 1.09-1.35, p=0.0006). The differences in the subgroups were not statistically significant (p=0.38) despite the difference in heterogeneity between the two subgroups. This proves that although the South Asian subgroup has variability in effect sizes, the overall effect is not much different from that of a subgroup in which the differences in effect size are less. Figure 2 shows the forest plot and funnel plot of the effect sizes overall and for the subgroups.

Figure 2.

Meta-analysis of Decision Autonomy and Women’s Paid Work Status

Among the other variables in our synthesis, we see that secondary education, respondent age, and family wealth are important predictors of women’s decision autonomy. However, there are a high number of insignificant relationships between family wealth and women’s decision autonomy. Therefore, these variables were excluded from this meta-analysis. On the other hand, most studies categorized the “age” variable into different ranges. Hence, the odds ratios were not comparable.

It is observed that many studies have found level of education (secondary or higher) to have a consistent relationship with decision-making autonomy. Ten of the 12 studies included in the meta-analysis modelled the relationship between secondary education and decision autonomy. In the meta-analysis, we found that the effect sizes were relatively homogeneous (I2=47%, p=0.05) and the odds ratio of the total effect was 1.24 (95% CI= 1.19-1.30, p < 0.001). By coincidence, the odds ratios in the meta-analysis for educational level and women’s paid work status are very similar. Figure 3 depicts the results of the meta-analysis of women’s decision autonomy in relation to secondary or higher levels of education.

Figure 3.

Meta-analysis of decision autonomy and women’s education

Synthesis

The measurement of women’s autonomy has principally been based on their freedom to access and enjoy ownership of money and assets, independence of decisions in various aspects of the household, including purchase decisions, and freedom of movement. Such autonomy has been portrayed as a part of women’s empowerment. While autonomy is undoubtedly difficult to measure, a variety of indicators can be used to do so. However, no gender gap-measuring index has included women’s autonomy; rather, women’s political inclusion has been taken as a surrogate marker for empowerment.

Decision-making autonomy is the most important dimension of women’s autonomy, as it is the most discussed component and has many subdimensions. Economic autonomy may be influenced by the economic status of a woman’s natal kin, dowry, and increased control of household assets with increasing age. In addition, women’s paid work status and economic autonomy are said to be indistinguishable in many studies (Susilastuti, 2003; Varghese, 2011). Hence, a regression study of economic autonomy with women’s paid work status might be biased. On the other hand, the reach of decision autonomy covers every facet of a woman’s life, from the purchase of household items to the utilization of healthcare. The freedom to make decisions has also been related to freedom of movement. Hence, in this study, decision-making autonomy was measured as women’s autonomy in relation to other variables.

It is worth mentioning that among low- and low-middle-income group countries, South Asian countries have especially focused on studies of women’s autonomy. Societal norms contradict the economic growth of this region, and gender disparity in labor force participation is highly influenced by social norms (Asadullah, Savoia, & Sen, 2020; Jayachandran, 2020). Apart from South Asia, most studies are from Africa, especially Ethiopia and Nigeria, while Latin American countries have both higher Human Development and Global Gender Gap indices.

Studies that discuss decision autonomy and paid work status also emphasize secondary education level, age, and family wealth. The latter two have been observed to be nonsignificant in many cases. The World Economic Forum commented that India’s women cannot share and enjoy household wealth at all socioeconomic levels (Tandon, 2018). Thus, women’s economic participation is an independent contributor, despite the contribution of family wealth to women’s autonomy. This may raise questions regarding the modernization theory of gender development (Gwynne, 2009), as increased wealth does not ensure the empowerment and well-being of women. Rather, the outcomes of our synthesis reflect ideas from the dependency theory of gender inequality (Scott, 2021) and reflect a process whereby participants engage in an intra-household cooperative game to attain an equilibrium of autonomy through economic contributions by both genders.

Meta-analysis

The current study is the first to perform a meta-analysis of women’s decision autonomy and paid work status. Previous meta-analyses were conducted on women’s employment and fertility (Matysiak & Vignoli, 2008) and on the effect of gender proportionality on job performance (Mackey, Roth, Van Iddekinge, & McFarland, 2019). Hence, the present study may pioneer statistical approaches to the matter of women’s employment and their autonomy in decision-making. Initially, we observed considerable heterogeneity in the effect sizes of paid work status on decision autonomy, especially in South Asian countries. However, such heterogeneity does not obviate the effectiveness of meta-analysis, where the total effect has been seen to be highly significant even for the subgroups, and the odds ratios are also very similar, showing that women’s employment has a consistent effect on autonomy. Lewin (2013) concluded that heterogeneity in social science is obvious because social systems are composed of heterogeneous elements with categorical dynamism and ambiguity owing to their complex nature. On the other hand, our findings confirm that women’s employment has a positive effect on their decision autonomy. This implies that women’s work imparts autonomy in decision-making. In our study, the causal relationship of paid work to women’s autonomy is shown through an odds ratio of “overall effect.” Although such causal inference has a real effect, there are limitations to such an inference when accomplished through meta-analysis (Weed, 2000).

Besides women’s paid work status, education is an emergent and important component of decision autonomy, as per this study. Our results show that the effect of education is homogeneous among studies and makes a significant positive contribution to women’s decision-making autonomy. The overall effect shows that the odds ratio of the effect of education on decision autonomy is similar to that of paid work status. However, for South Asian countries, the odds ratio for education is marginally lower than that for paid work status, which highlights the necessity for further statistical study in this context. Existing research has different opinions on the relationship between women’s education and their work. Yousefy and Baratali (2011) observed that higher education has been positively related to increasing salaries among already working women. Although a higher level of education increases the employability of women, the actual labor force participation of women has not been seen to increase. Rather, in many Asian countries, higher levels of education among women results in a reduction or stagnation of female labor force participation levels; this is termed the MENA (Middle East & North Africa) paradox (Assaad et al., 2020). However, the importance of education in decision-making autonomy cannot be denied.

In a theoretical context, our work supports the idea that paid labor leads to sufficient intra-household bargaining (Doss, 2013), which translates into autonomy and endows women with resource allocation power. In contrast to relational autonomy theory (Mackenzie & Stoljar, 2000), we propose the possibility of a cooperative game inside the family when women have a paid job. Although further research is needed to determine whether such interaction increases the marginal utility of all agents, we propose that sustainable autonomy can exist in the family as a result of the equilibrium in resource allocation power.

Practical Implications of the Findings

Research shows that women’s economic participation leads to better autonomy (Buller et al., 2016; Cornish et al., 2021; Ortiz-Rodríguez et al., 2017). However, during crisis periods or tough times like famine, war, and poverty, women’s overall conditions deteriorate and patriarchal social norms get re-imposed (McLaren, Wong, Nguyen & Mahamadachchi, 2020). This is reflected in the recent COVID-19 pandemic, which has had a gendered impact on society. A report by Care International shows that during the COVID-19 pandemic, women were more likely to be thrown out of the labor market than men, and the resultant reduction in their financial contribution to the household decreased their decision-making autonomy (Buschmann & Fuhrman, 2020). Another report by United Nations Women showed that the loss of income for women forced them to assume a larger care work burden and increased the risk of intimate partner violence. They were “trapped” in the abusive situations and the loss of employment further undermined such women's autonomy, making it much more difficult or even impossible for them to flee from the violent situation at home (UN Women, 2020).

Limitations

This systematic review has certain limitations. First, there have been few studies on women’s workforce participation and decision autonomy, and this is especially the case for South Asian countries. Second, most of the studies in this area are relatively contemporary; thus, the historical perspective of women’s autonomy is not well understood. Third, qualitative data was difficult to find because most of the studies were conducted quantitatively using secondary data. Had qualitative data analysis been available, the dynamicity and scope of women’s autonomy could have been better understood. Fourth, as stated in the inclusion criteria, only studies published in English were included in this systematic review, which may have resulted in a range of potential sources being excluded.