WOMEN’S ECONOMIC EMPOWERMENT AND ECONOMIC GROWTH IN SUB -SAHARAN AFRICA

43  Download (0)

Full text

(1)

WOMEN’S ECONOMIC EMPOWERMENT AND ECONOMIC GROWTH IN SUB-SAHARAN

AFRICA

University of Amsterdam, Amsterdam School of Economics MSc Economics, International Economics

Name: Larissa de Lima Almeida Student number: 11764058 Date: 15.08.2021

Supervisor: Naomi Leefmans

Second reader: Dr. Dirk Veestraeten Email address: larissadla@live.nl Wordcount: 11303

(2)

1

Statement of Originality

This document is written by Student Larissa de Lima Almeida who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

UvA Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

2

Abstract

This paper investigates if empowering women economically results in more economic growth. With evidence from 49 Sub-Saharan African countries and the use of system-GMM estimations, I find that a higher share of girls participating in primary education can positively impact GDP per capita growth. The results also indicate that a higher share of women in SSA completing secondary education reduces population levels. This could indicate that, women who have secondary obtained an education make the decision to have less children or get the decision-making power to do so.

(4)

3

Table of content

1. Introduction ... 4

2. Literature Review ... 6

2.1 Where do the women of SSA stand now? ... 6

2.2 How can women's economic empowerment induce growth? ... 8

2.3 How can women's economic empowerment hamper growth? ... 13

3. Data and Methodology ... 15

3.1 Methodolody………..15

3.2 Data……….……19

4. Results ... 26

4.1 Results direct outcome variable: GDP growth………27

4.2 Results indirect outcome variables: Investment and Population…….29

5. Discussion ... 35

6. Conclusion ... 36

7. References ... 37

(5)

4

1. Introduction

The women in Sub-Saharan Africa (SSA) do not always enjoy the same economic liberties as the men do. Even though SSA has seen some improvements in the aspects of gender and income inequality, the region still suffers under relatively high levels of inequality. In 2019, Africa had an average score of 48.6% on the ‘Africa Gender Index’, on which 100% is full equality of women to men (African Development Bank, 2020). Also is it the highest scoring region on the Gender Inequality Index of the UN, implying the highest gender inequality in the dimensions covered by this index (UNDP, 2020). Moreover, SSA is still behind the rest of the world in terms of GDP per capita growth, which recently has seen some stagnation.

In the period of 2016-2019 there has been a small negative GDD per capita growth rate on average in SSA (World Bank, 2020). With 40% of its population living under the $1.90 poverty line in 2018, poverty is widespread (World Bank, 2021). Economic empowerment of women might be one of the factors which can help mitigate poverty and stimulate economic growth.

Women’s economic empowerment seems to have good effects on human development in general (Harari, 2017; Chattopadyay & Duflo, 2004; Mason

& King, 2001). Therefore, I intend to answer the research question of what is the effect of women’s economic empowerment on economic growth. We can define economic empowerment as “the capacity of women and men to participate in, contribute to and benefit from growth processes in ways that recognise the value of their contributions, respect their dignity and make it possible to negotiate a fairer distribution of the benefits of growth.

Economic empowerment increases women’s access to economic resources and opportunities including jobs, financial services, property and other productive assets, skills development and market information.’’ (OECD, 2011). Thus, women’s economic empowerment concerns stimulation of a more equal position of men and women in the market, whereas gender equality refers to equal rights, responsibilities and opportunities for men and women in every aspect.

There is consensus among economists that the effect gender inequality

(6)

5

has on economic growth depends on the structure of the economy. For example, Seguino (2000) finds that middle-income export-oriented economies experience positive effects of gender inequality in income on investments and exports. Firms in these countries can easily move to places where the wage is lower. However, several other studies find a contrasting result and emphasise that especially low-income countries can benefit from more gender equality, in aspects like income and education (Hakura et al., 2016; Bandara, 2015; Klasen, 2000). This result captures the importance of research on gender inequality and economic growth for SSA, since these countries mostly fall into the low-income category.

Another relevant result found by previous research is that we can see gender equality improve economic growth through direct effects: immediate increases in GDP per capita growth (Bandara, 2015). But it also may result in more long-term economic growth through indirect channels: like investment and population growth. By increasing for example decision- making power or ownership rights for women, we can expect to see more investment in the well-being and education of children (Donald, et al., 2021; Knowles, et al. 2002). Other studies focussed on the role women’s decision-making power can have on fertility rates. Evidence has been found for lowered fertility rates due to more gender equality in intrahousehold decision making, which in turn can induce more economic growth (Berik, et al., 2009).

Previous research on the nexus between economic growth and gender inequalities often is done theoretically, while there is a lack of empirical research done on this topic. Empirical research that has been done on the link between gender inequality and economic growth is to a large extent focussed on other aspects of gender inequality than covered in this paper.

The novelty of this study is that it considers several economic inequalities and also the implications these have for Sub-Saharan Africa. Except for Hakura et al. (2016) previous papers have not examined implications for SSA specifically.

The first of objective of this is paper to investigate what the implications are of economic gender inequalities for GDP per capita growth. Its second

(7)

6

objective is to shed more light on how economic gender inequalities can affect economic long-term growth through indirect factors. The question that will be researched in this paper is as follows: what is the effect of women’s economic empowerment on economic growth in 49 Sub-Saharan African countries, covering the time period 1972-2020.

In line with Klasen (2009) I will examine how economic growth is affected directly as well as indirectly by women’s economic empowerment. Several proxies for economic empowerment are regressed on the indirect outcome variables: investment and population, as well as on the direct outcome variable: GDP per capita growth. I will be using the following proxies for economic empowerment: inequality in employment, ownership rights, the level of decision-making power of the women and women’s education obtainment. The growth framework used is based on the neo-classical growth model (Barro, 1991; Barro & Mankiw & Sala-i-Martin, 1995; Barro

& Lee, 2001). Lastly, the use of system-GMM analysis will serve to correct for possible endogeneity.

In this paper I will start by exploring the current literature out there in section 2. On the topics of what the current stance is of woman in SSA regarding economic gender equality and the relation between women’s economic empowerment and economic growth. In section 3 the data and methodology will be elaborated. Section 4 contains the research results, which will be followed by the discussion in section 5. Lastly, section 6 provides a summary and conclusion of this paper.

2. Literature Review

2.1WHERE DO THE WOMEN OF SSA STAND NOW?

What does the current position of women look like in the Sub-Saharan African market compared to that of men? In this section I will start by painting a picture of the current situation in SSA regarding economic gender inequality. I will describe how the economic liberties of women compare to those of the men in SSA and what their current role is in the market.

(8)

7

In Africa, considering the economic and business dimensions, women have an average of 61.7% of the liberties that men do (AFDB, 2020). This translates to an economic inequality gap of 38.8%. There still is space for improvement, even though this is not only the case for SSA, we can see relatively low rates of economic empowerment for women in this region. In contrast, in the African Botswana women have 88.6% of the economic liberties that the opposite sex has (AFDB, 2020). Botswana is an African anomaly; it has experienced relatively steady GDP per capita growth.

Young girls are not being stimulated as much as young boys to invest in their education and their work life on average in African countries (African Development Bank, 2020). SSA has one of the lowest enrolment rates in secondary education for girls, which was at 34% in 2018. In comparison: in Europe the percentage is as high as 91% and in East Asia at 81%. As suggested by empirical research, equal stimulation of girls might improve long-run economic growth. It can prevent overinvestment in less-qualified men and too little investment in qualified women (World Bank, 2020).

Furthermore, women undertake a bigger portion of household work and caring activities. This leaves women with less room to invest in their working career and limits their ability to be able to stay in school and invest in their future (AFDB, 2020).

Moreover, women are the main food producers in Africa, they however lack the ability to invest in their business. Next to agriculture, are the women in SSA are mostly active in the informal sector. Women in Africa often lack ownership rights over assets, which results in less income security, when considering the agriculture sector (AFDB, 2020).

Furthermore, the weaker property rights for women result in less productivity due to underinvestment (Seguino & Were, 2014). Also, their ability to access credit through financial institutions is limited due to the lack of collateral. Stimulating women’s economic empowerment may increase their access to inputs and the investment in their businesses.

Women may experience more security with economic empowerment, which can go along with more investment as well. There have been efforts towards

(9)

8

improving women’s economic empowerment, by giving them access to credit with micro-financing. However, researchers found that micro- financing still limits their opportunities in the sense that it limits the size to which their businesses can grow (Dejene, 2007).

In conclusion, women experience constraints in economic opportunities in the areas of labour participation, unpaid work, economic prospects, decision making power and ownership rights.

2.2CAN WOMENS ECONOMIC EMPOWERMENT INDUCE GROWTH?

This section describes the several mechanisms through which economic empowerment can induce growth. Namely, decision making power, a bigger share of women participating in the labour force, unpaid labour, gender equality in education and income. In section 3 these mechanisms will be linked to how these might hamper growth.

2.2.1DECISION MAKING POWER

To start off, women tend to invest a greater portion of their income in the well-being of their children compared to men. Specifically, women’s ownership rights over assets have been found to result in more investment in the future of their children, especially more in girls (Duflo, 2003; Qian, 2008; Luke & Munshi, 2011). On the other hand, men spend a greater portion of their income on luxury products (Seguino & Were, 2014). Harari (2017) has shown, with evidence from Kenya, that providing women with inheritance rights has resulted in women having a bigger say in household decisions through increased bargaining power. She finds that introducing inheritance rights for women, resulted in roughly one extra year of education for daughters.

Furthermore, educated women possess more decision-making power.

Stimulating educating in girls can result in relatively higher long-term economic growth, through more influence on intra-household decisions.

More specifically, empirical results have suggested that female education goes together with increased investment in the household and has a significant positive effect on labour productivity, which in turn reduce labour costs per unit (Krueger, 1999; Barro, 2001; Sianesi & Reenen,

(10)

9

2003).

Donald et al. (2021) try to shed light on question of whether if matters if women are given decision making power (for example by her husband) in contrast to if she were to take power. They investigate this by using several answers in the Demographic and Health Surveys as a proxy for decision making in 23 countries in SSA. The main results found is that women taking power, rather than this being given to her by her husband, has a positive effect on the well-being of her children.

2.2.2LABOUR PRODUCTIVITY

One of the direct effects on economic growth we can see from women’s economic empowerment is the increase in labour productivity. We can see this happening through several channels; two of them are increases in female labour participation and a higher share of girls obtaining an education. By excluding women from the labour force, the labour potential of SSA is not fully utilized. In other words, the allocation of resources is not done in an efficient manner. Bandara (2012) has shown that labour productivity in SSA could increase by 0.3 to 0.5 percentage points annually from 1970 – 2010 in case of gender equality. This translates in lost output of 10% of the SSA GDP in that period (Bandara, 2015).

By using cross-country and panel growth regressions based on the work of Barro (1991), Klasen & Lamanna (2009) come to similar results.

Gender gaps in employment significantly reduce economic growth. Based on economic growth literature the following control variables are included:

population growth, openness, the investment rate, human capital and regional dummy variables to capture regional differences. In order to examine long-term growth, they treat the period 1960-2000 as a single observation per country. Additionally, in order to measure gaps in employment, data on labour participation is used. They conclude that changing the composition of the labour force to include more women can have positive effects on growth. One caveat is that measurement of labour force participation often differs per country and are prone to measurement errors. These results therefore need to be considered with caution.

(11)

10

Also Cuberes & Teignier (2016) state that using the full potential of the female labour force population is macro-economically more efficient. Talent is distributed inefficiently in the economy, due to gender inequality. The authors study gender gaps in labour force participation and its effects on aggregate productivity and income per capita. The outcome of the study portrays the negative effect of gender inequality on resource allocation, which in the case of OECD countries results in an income loss of 15 percent.

Forsythe, Korzeniewicz & Durrant (2000) suggest the relation of economic development and the status of women also works backwards. The status of women is measured with the UN’s gender-related development index. This index exists out of three components, namely: health and longevity, education and standard of living. It considered how women perform on these components. They find supporting evidence for the hypothesis that more economic development is beneficial to the status of women.

2.2.3UNPAID LABOUR

Another burden on GDP growth in terms of gender inequality is unpaid labour. Unpaid labour is defined as ‘’ Work that produces goods or services but is unremunerated. It includes domestic labour, subsistence production and the unpaid production of items for the market.’’ (OECD, 2013). Women are the biggest performers of unpaid labour. This creates the existence of so-called ‘time-poverty’. Spending a large majority of time on unpaid work leaves people with less time they could otherwise spend participating in the labour force. The result is that women in SSA tend to work more hours in total (paid plus unpaid) than men. In addition, women having greater responsibility in caring for the household also has adverse effects on the quality of work performed. They might have to bring their children to their working place, etc. (Kabeer, 2012).

Research found that time spend on unpaid work declines with rising income levels (Budlender, 2008). Greater gender equality in terms of unpaid work has been argued to stimulate growth. Similar to the case of women being given more ownership rights, freeing up women’s time from

(12)

11

unpaid labour, would have the effect of increased bargaining power of women within the household. As stated before, this can improve the opportunities for children, especially girls (Seguino, 2000).

2.2.4GENDER EQUALITY IN EDUCATION

There is consensus among researchers that gender equality in education results in greater economic growth. The impact of increasing women’s education seems to have different effects on GDP growth than the effect of educating men. As well as direct effects, like increased productivity, we can see more indirect effects, through improved child health, longer life- expectancy, lower infant mortality, reduced fertility and increased investment in children’s education (Knowles et al., 2002). In addition, not only the mothers themselves but also the daughters of empowered women have less children. Lower fertility rates are linked to higher child investment, as the household’s resources must be distributed over a lower number of people (Berik et al., 2009).

Klasen (2000) emphasises how SSA in particular can benefit from gains in educational equality. He also finds that inequality in education reduces economic growth through direct and indirect channels. Here direct effects are understood as the lowered quality of human capital and indirect effects may be the previously mentioned lowered child mortality and reduced fertility. By using instrumental variables regressions, he measures the indirect effect by using population growth as well as investment as outcome variables. To measure the direct effect, he used GDP per capita growth as an outcome variable, the focus lays on long-run growth for all regressions.

He concludes that economic growth in SSA could have been up to 0.9 percentage points faster than has been the case, would there have been balance in gender in education.

Another interesting result is the one found by Mitra, Bang, & Biswas (2015). They used a neo-classical growth regression model analysis, augmented with gender inequality, and looked at what the effects are of gender inequalities in economic opportunities and inequalities in economic outcomes. Developing nations suggestively benefit most from creating more equality in economic opportunities, like access to education, whereas

(13)

12

developed nations benefit most from equality in economic outcomes, like giving women a stronger voice in the political spheres.

2.2.5 EQUALITY IN INCOME

Figure 1: gender equality index (GII) and GDP growth

Source: UNSTATS (2020); World Development Indicators (2020)

Income inequality and gender inequality have been found to jointly hamper economic growth, this is especially the case for low-income countries, according to Hakura et al. (2019). As can be seen in figure 2, there is a negative correlation between GDP per capita growth and the Gender Inequality Index (this figure includes only SSA countries). This index covers the following dimensions: reproductive health, representation of women in parliament and education and labour market dimensions. There will be further expended on the details of this index in chapter 4. Hakura, et al.

(2016) find in that in SSA the missed growth annually is as high as 0.9 percentage points, in comparison with if the regions income and gender

(14)

13

equality were similar to that of the Southeast Asian Nations. Since the Sub- Saharan African market suffers from credit market imperfections and credit constraints, investing more in human capital (by stimulating gender equality) can help mitigate against shocks which cannot be fenced in by the market. Lastly, higher gender inequality would adversely affect gender gaps in educational attainment.

They came to this result by exploring what the effects of gender- and income inequality are on GDP growth. The authors hereby used the GINI index and the UN’s gender inequality index as proxies for income- and gender inequality measures, respectively. In order to correct for possible endogeneity, they used the system-GMM econometric method. Moreover, the model includes commonly used control variables in economic growth regressions: initial income, initial level of infrastructure, years of schooling, investment to GDP, inflation, institutional indicators and terms-of-trade changes.

In the theoretical framework by Blumberg (2005) increased income controlled by women would result in many of the already stated benefits women empowerment can have, since increased income control is one of the forms of economic empowerment. It, for example, strengthens the voice of women within the household, which in turn would result in, the already stated, benefits of lowered fertility, more investment in education and well- being for children, etc.

Moreover, in other theorical explorations of the nexus between economic growth and gender inequality, researchers highlight the importance of looking at long term effects, rather than only focussing on short term implications of gender equality. Gender inequality would hamper long term growth especially; it may even be so that more gender wage equality will lead to less short-term growth (Berik, et al., 2009).

2.3HOW CAN WOMENS ECONOMIC EMPOWERMENT HAMPER GROWTH?

Apart from the studies that find a positive effect of women’s economic empowerment, others find opposing results. This section discusses in what cases and how women’s economic empowerment might hamper growth.

(15)

14

According to the theoretical exploration of gender equality by Seguino &

Were (2014), more economic gender equality in the form of wage equality can hamper growth in the short-run. Specifically, for semi-industrialised economies where women are concentrated in labour intensive export industries and firms are mobile. This occurs for two reasons. Firstly, if female wages are increased, firms can easily be moved to another location where wages are lower. Secondly, product demand is often price elastic in labour intensive industries, due to the homogeneity of the product, there are several other substitutes. This implies a reduction in demand if female wages are increased. However, these effects can be (partly) offset if women’s marginal propensity to consume is higher and the propensity to import is lower.

In terms of gender income equality opposing results have been found by empirical studies as well. One of the reasons for this is the difficulty in finding a good measure for gender income inequality, the variables often suffer from measurement errors, particularly due the lack of complete data in older work. Forbes (2000) finds a positive relationship between gender income inequality and economic growth in the short- and medium run.

However, he states that this may not be the case for very poor countries, due to lack of data on these countries.

Furthermore, women who were excluded from education have lower labour productivity than men. We can see this happening in the informal sector in SSA, where relatively more women are employed and there is lower productivity. In the short-run, this possible lower productivity of women can result in less economic growth. On the other hand, investing in women now, can therefore result in more long-term growth, as there is space for potential productivity growth (Bandara, 2015).

Similar to other studies, Seguino (2000) incorporates indirect effects in her regressions. She does this by looking at the effects of gender income inequality on investments. The focus lays on middle-income export- oriented economies. She finds that gender inequality has a positive effect on investments as well as exports. These results capture the importance that the economic structure in the effects gender inequalities can have on

(16)

15

an economy. The argumentation behind this result is that lower wages for women can stimulate export through increased competitiveness and with

that export technology.

The main conclusions that can be drawn based on these studies are that more economic gender equality can result in more economic growth. This can happen through direct channels, like increased quantity of human capital, and indirect channels, like increased investment in children’s education. In contrast, older work gives us another view, namely a negative relationship between more equality and economic growth. This may be explained by the issue of measurement errors and the then not yet existent more sophisticated econometric methods. Also the structure of the economy matters. As suggested by the literature, middle-income countries with a higher rate of trade openness would have less benefit from more economic gender equality. Lastly, the time frame looked at matters to judge how beneficial economic gender equality is for economic growth.

Empowering women economically could result in less economic growth in the short-run.

3. Data and Methodology

Section 3.1 discusses the methodology used in this paper. It discusses what research questions is intended to be answered and how the research will be implemented. The regression equations will also be explained in detail.

3.1 METHODOLOGY

The research question which I intend to answer is: what are the effects of economic empowerment for women for economic growth in Sub-Saharan Africa? Based on past studies, I hypothesize that more economic empowerment for women (a stronger position and more power in the market), has positive effects on GDP per capita growth in SSA. In the current economic structure of SSA, there is space for higher levels of GDP growth as well as higher levels of gender equality. Improving women’s access to the market can result in better allocation of resources within the

(17)

16

economy and with that higher GDP per capita growth. Also, an increase in the quantity and quality of human capital helps to mitigate against shocks.

Due to lack of credit market access in SSA, this region especially can benefit from more economic growth through higher levels of human capital.

This paper considers how introducing equal ownership rights, increased decision-making power for women and stimulating their participation in the labour force influence economic growth. These can be achieved by stimulating girls to obtain an education, decreasing fertility rates, changing the legal system etc.

This hypothesise will be analysed using the economic growth model based on the work of Barro (1991) and Barro (2003). In line with Klasen &

Lamanna (2009) and Mitra, Bang & Biswas (2015) the neo-classical growth model will be augmented with several proxies for economic gender inequalities. As stated in the previous chapters the effects of women’s economic empowerment work through several channels. In order to get a complete view of how empowering women in SSA economically will influence economic growth, I will look at the impact of women’s economic empowerment on three outcome variables. Namely, GDP per capita growth, investment and population; these will be used as the dependent variable in the regressions. The regressions considering GDP growth per capita are the main estimations of this paper. These show the direct effect of women’s economic empowerment on economic growth. The following two regressions with investment and the population as its dependent variable show the indirect effects of women’s economic empowerment on economic growth.

The investment and population models are in accordance with the work of Seguino (2000) and Klasen & Lamanna (2009). For each economic gender inequality, a separate regression will be run, this will be the case for all outcome variable considered. There will be elaborated on each proxy for economic inequality and the details of the explanatory variables in section 3.2. Taking these points into consideration, the following models will be estimated:

(18)

17

Included are several explanatory variables with economic empowerment as its independent variable. The first explanatory variable included is lagged growth, lagged investment and lagged population to take into account autocorrelation of the depend variable. In table 1 below, explanation of the rest of the explanatory variables can be found. In the data section 3.2 there will be further expanded on the expectations of these variables.

Table 1: explanatory variables.

Abbreviation Variable explanation Source

Initial GDP The GDP value at the

starting year of the data.

World Bank

Invest Investment to GDP ratio. Penn World Tables (PWT)

GovExp Government expenditure to

GDP ratio.

PWT

Inflation Consumer price Index (CPI)

annual estimation.

World Bank

HC Human Capital: measured

by years of education.

PWT

Open Openness: the ratio of

exports plus imports to GDP.

PWT

Density The amount of people per

square km of land.

World Bank

GDP Real GDP using national

accounts growth rates.

World Bank Growthit = β0 + + β1EconEmpowit + β2Growthit-1 + β3InitialGDPit + β4Investit + β5GovExpit + β6Inflationit + β7HCit + β8Open + εit

Investit = β0 + β1EconEmpowit + β2Investit-1 + β3GDPit + β5Inflationit + β6Openit εit

POPit = β0 + + β1EconEmpowit + + β2POPit-1 + βGDPit + β4Density + β5Inflationit + β6HCit + β6Open + εit

\

(19)

18

For all three of the models it is necessary to correct for country-specific omitted variables to prevent endogeneity issues. In previous literature country-fixed effects are often included in the regressions to correct for these initial differences in technology, institutions, etc. (Islam, 1995). However, to fix the issue of endogeneity in a more efficient manner, the system-GMM model by Blundell & Bond (1998) will be implemented for all three of the models analysed (the economic growth-, investment- and population model). Which accordingly removes country-specific fixed effects.

Moreover, previous research has found that the relationship between women’s empowerment and economic growth goes both ways. For instance, from the evidence provided by Korzeniewicz & Durrant (2000) follows that economic growth also causes women’s empowerment. Consequently, reverse causality issues need to be accounted for. The use of the instrumental variable technique has been a popular one to fix this reverse causality issue in gender inequality and economic growth research. However, it suffered from criticism due to lack of strength of external instruments used. In order to address this possible endogeneity issue, the use of system-GMM introduced by Blundell &

Bond (1998) will be the most appropriate in this case.

I tested whether difference-GMM or system-GMM would fit the data better by using the method of Bond (2001). This is done by running fixed effects regressions as well as OLS regressions for each of the women’s economic empowerment proxies separately. If the coefficients of the first lag of growth are closer to the fixed effects estimates, the estimates are downwards biased. After running the tests, system-GMM turned out to be the most appropriate for this research. Also, because the panel data used is unbalanced, system-GMM is preferred over difference-GMM. Lastly, the issue of endogeneity is not fully solved with the use of difference-GMM Furthermore, one-step system-GMM will be used rather than two-step system-GMM. For part of the data there only are a small number of observations available. One-step system-GMM is the more reliable choice in the case of a small sample size. Especially for SSA obtaining a large data set can be a difficulty. Although two-step system-GMM has the benefit of more

(20)

19

precision over one-step system-GMM, one-step system-GMM comes with the benefit of correctly estimated values of the standard errors. Lastly, the internal instrument Growtht-2 will be used and is thought to be uncorrelated with the fixed effects.

I will be using yearly data for the sake of retaining a higher number of observations. The data used starts from 1972 till 2020 and includes 49 countries. Lastly, time dummies are included for years 1983 and 2009 in the economic growth regressions so that correlation of the standard errors is prevented (Roodman, 2009). In these years SSA saw a common downward trend in growth due to economic crises.

Due to concern of multicollinearity among the explanatory variables a VIF test is performed, the results show that there is little need for this concern.

All the VIF results are below or equal to 2. However, I will exclude government expenditures to GDP ratio from regressions to check for robustness, as this variable seems to be moderately1 correlated with the investment to GDP ratio.

Furthermore, there were some outliers detected in the data2. These have been replaced with the Winsor method. Also, after testing it can be concluded that heteroscedasticity is present, which will be corrected for with the robust option. Lastly, the AR(2) test suggests that there is no sign of autocorrelation.

1 There is a correlation of 0.686 between the investment to GDP ratio and the government expenditures to GDP ratio. The complete correlation matrix can be found in table 12 in the appendix.

2 Please find the figure of the outliers in the appendix in figure 4.

(21)

20

Figure 2: GDP per capita growth in SSA

3.2 DATA

This section includes a discussion of the data used for the research.

Specifically, it discusses how the data looks. Also, I will expand on the variables introduced in section 3.2.

Why these are used, how these are measures and where these are retrieved from.

OUTCOME VARIABLES

In this paper, the direct outcome variable is defined as GDP per capita growth. From figure 3 we can see that average growth in SSA has become more stable over time and remained positive since 1999 until 2016. In the 1980s Sub-Saharan Africa experienced an economic crisis, which can be seen by the negative growth in that period. In the year 1983 this region experienced the most negative growth, one of – 2.4%. A relatively smaller dip in economic growth can be seen in the year 2009, due to the global economic crisis. SSA’s GDP growth went from 6.4% to 3.1%. In relative terms, SSA was not hit as hard as the developed world by this crisis, since there was a lack of exposure to foreign capital markets (Fosu, 2013).

Moreover, from the summary statistics in table 1 it can be concluded that growth per country and over time can be very heterogeneous, with a maximum of 140.4% growth, due to oil field discoveries in Equatorial Guinea (IMF, 1999) and a minimum of -47.5% in 1994 Rwanda, due to the genocide. Although the focus of this paper lays on economic growth per capita and not on the actual level of GDP per capita, the level of GDP per capita varies more economic growth and therefore is also an interesting factor to look at and take in consideration. I will consider the use of this variable as a test of robustness.

The indirect outcome variables examined are investment and population, in accordance with the papers of Klasen & Lamanna (2009) and Mitra, Bang & Biswas (2015). The reason for including these specific

Source: World Development Indicators (2020)

(22)

21

variables as a measure for the indirect effects, is that women’s economic empowerment influences consumption decisions in addition to intra- household decisions, like in the decision to have children and expand the family. These in turn influence investment and population growth, as explained in the literature review ((Knowles et al., 2002; Berik et al., 2009).

The data on investment and population are both retrieved from the Penn World Tables dataset. Investment is measured as import plus production minus exports. Population is the population level in millions.

Table 2: Descriptive Statistics, outcome variables

Variable Obs. Mean Std. Dev. Min Max Growth per

capita 2191 1.491 6.769 -47.503 140.371

Investment

(:1000) 1992 1404.12 3548.278 .027 45411.01 Population

(:1000000) 1992 13.342 22.472 .055 200.964 Real GDP per

capita (:1000) 1992 3873.39 4539.892 266.845 24759.463

PROXIES FOR WOMENS ECONOMIC EMPOWERMENT

Female labour force participation

The first proxy for women’s economic empowerment which will be included is inequality in employment, which will be measured by the percentage of women participating in the labour force, in line with Klasen (2009). The data is retrieved from World Bank (2020). Previous literature (Bandara, 2012) has found this variable to positively and significantly affect GDP per capita growth in the SSA region. Currently the labour force in SSA is not used to its full potential. Empowering women to participate in the labour force can result in a more efficient way of allocating labour in the market and prevent investment in less qualified men and stimulate investment in more qualified women.

One caveat of comparing female labour force participation rates between countries is the difference in measurement as the concept of labour force can differ per country. Furthermore, we should keep in mind that a large portion of the women in African countries work in the informal sector. This

(23)

22

sector often is not considered when measuring the labour force, since this falls out of the scope of the employment surveys. Considering these points, the results regarding the labour force participation should be read with caution (ILOSTAT, 2020).

Ownership rights and Decision-making power of women

Based on the papers of Duflo, 2003; Qian, 2008; Luke & Munshi, 2011;

Donald, et. al, 2021, I hypothesize a higher level of decision-making power and equal ownership rights for women and men to positively affect economic growth. Based on the literature, I expect more equal ownership rights as well as more female decision-making power to improve the prospects and opportunities for the next generation and therefore affect growth indirectly, rather than directly. So by influencing population and investment. Furthermore, since ownership rights and higher levels of decision-making power, changes intra-household decision making regarding expansion of the family, I expect to see lower fertility levels. Thus, lower population growth.

The decision-making variable is retrieved from the Demographic Health Surveys (DHS). The variable measures “the percentage of women who say that they alone or jointly have the final say in none of the three main decisions (own health care, making large purchases, visits to family, relatives, friends)”. The ownership variable is retrieved from the World Bank Gender Stats Database. The variable looks at if men and women have equal ownership rights to immovable property. It takes the value of 1 it is a case of equal ownership rights and 0 if not.

As discussed in the literature review, women who obtained an education have more bargaining power within the household. Furthermore, education raises access to the market and therefore economic empowerment.

Therefore, the female secondary education completion rate will be included in the regressions as well as percentage of girls attending primary school3. Both as proxies for women’s decision-making power. Also the total secondary education completion rate and total primary school enrollment

3 Both of data of these variables are retrieved from the World Bank (2020) database.

(24)

23

will be included in the regressions as a way of comparison to the schooling rate of girls.

Gender inequality index

The gender inequality index measures gender inequality through three dimensions. The first is reproductive health; this dimension includes the indicators of maternal mortality and adolescent birth rate. The empowerment dimension compares the number of males and females with at least some secondary education and the number of women with seats on the parliament. Lastly, the labour market indicator looks at female and male labour participation. Higher values on the gender inequality index are equal to more inequalities between men and women. As can be seen in figure 4, in SSA there was a clear downward trend in gender inequality and thus more equality regarding these dimensions. However, after 2013 the trend started to slightly oscillate.

Figure 3: Gender inequality index (GII) and GDP growth

Source: UNSTATS (2020); World Development Indicators (2020)

(25)

24

Table 3 includes the women’s economic empowerment variables’

descriptive statistics. It can be seen that female decision-making power and the gender inequality index have a relatively small sample size.

However, for both variables there is enough variation in the levels it takes to be used for estimations. Also, education obtainment is used as a proxy for decision making power, which has a larger sample size. In terms of sample size, this variable can give a more accurate representation of how decision making affects economic growth.

Table 3: Descriptive Statistics: women’s economic empowerment Variable abbreviation Obs. Mean Std.

Dev. Min Max

Percentage of

women who say that they alone or jointly have the final say in none of the three main decisions (own health care, making large purchases, visits to family, relatives, friends)

decision 82 38.473 21.587 6.3 87.2

Labour force participation rate, female

lfp 1320 61.35 16.50 21.53 91.83

Completion rate secondary education pupils, female

secondaryfem 847 30.625 28.062 .224 116.045

Men and women have equal

ownership rights

ownership 2295 .623 .485 0 1

Gender Inequality

Index GII 157 .588 .078 .377 .75

The percentage of female primary education pupils

primaryfem 1206 91.894 37.61 9.918 261.196

CONTROL VARIABLES

Next to the above mentioned variables of interest also several other variables are included in the regressions which serve as control variables in the regressions. These are as follows.

(26)

25

Initial GDP per capita growth

The phenomenon ‘conditional convergence’ is an issue mentioned in several economic growth papers. Initial GDP per capita will be included in the regressions to account for this issue. Initial GDP is defined in this paper as the starting year of the data used for the regressions A negative sign of this variable is evidence for the convergence hypothesis. In this case richer countries grow slower than poorer countries (Barro, 1991). Note that real GDP is in millions of US dollars. The data on this variable is retrieved from the PWT dataset.

Trade openness (open)

The degree of trade openness is defined, in this paper, as the ratio of exports plus imports to GDP. According to the regressions of Barro (2003), I expect this variable to have a small positive effect on GDP growth and investment. The significance of this variable is depending on the size of the country, as bigger countries tend to rely more on trade. This variable is created with the data from the PWT dataset.

Real government consumption to GDP ratio (govratio)

Real government consumption has been found to have a negative relationship with growth and investment. The argumentation behind this is that higher government consumption reduces savings and growth through tax- or expenditure distortions (Barro, 1991). The variable is retrieved from the PWT dataset and measures government consumption in constant real prices of 2017. The value of real government consumption is equal to total final consumption expenditure minus expenditure on individual goods or services provided as social transfers to households.

Inflation

Due to the risk of high inflation rates in Sub-Saharan Africa, this variable especially I expect to have a large explanatory power. Also, inflation can be seen as a proxy for macroeconomic stability. Based on the research of Barro (2003), I hypotheses this variable to have a significant and negative effect on GDP growth, investment and population. Inflation is measured by the consumer price index (CPI). It reflects the annual change in prices of a

(27)

26

basket of goods and services. This variable is sourced from the World Bank database (2020).

Human capital (hc)

Empirical research has found that human capital as well as investment are highly explanatory for economic growth. It is predicted that higher human capital has a positive effect on economic growth for a given level of GDP per capita. The most traditional form of the measure of human capital is secondary education attainment. This paper will be using the human capital variable retrieved from the PWT data set, educational attainment is used to calculate the variable of human capital. Specifically, it uses the average years of schooling and a rate of return of schooling based on worldwide Mincer estimates to calculate the human capital index. This variable will not be included in the regressions with secondary education as the independent variable, due to collinearity issues.

Investment and investment to GDP ratio (inv and invratio)

The investment data is retrieved from the PWT dataset. It measures investment (import plus production minus exports) at constant national prices from 2017. Investment is predicted to have a positive effect on economic growth. Actual investment is used as the dependent variable in the regressions measuring the indirect effect of women’s economic empowerment. Whereas the investment to GDP ratio will be used as one of the explanatory variables in the economic growth regressions.

Population density (density)

Population density will solely be included in the regressions with population as dependent variable. Population density is expected to have a negative relation with population growth (Sibly, & Hone, 2002). This data is retrieved from the World Bank database.

4. Results

In section four present the results found by the above introduced regressions. In the main results section below the economic growth results are presented. This is the direct outcome variable. In section 4.2

(28)

27

results of the regressions including the indirect outcome variables (population and investment) are discussed. Lastly, in section 4.3 the robustness of the results are tested and presented.

4.1RESULTS DIRECT OUTCOME VARIABLE:GDP GROWTH

In table four the main system-GMM results can be found. These are the yearly estimates on GDP growth per capita. I will start by shortly discussing the control variables, which will be followed by a discussion of how the economic empowerment proxies affect growth, according to these estimations.

Table 4: System-GMM GDP growth per capita regressions on a yearly basis.

(1) (2) (3) (4) (5) (6) (7) growth growth growth growth growth growth growth

L.growth 0.494 0.204*** 0.212** 0.525** 0.0792 0.136* 0.185**

(1.71) (3.85) (2.93) (3.44) (1.42) (2.55) (3.41) InitialGDP -0.000131 -0.000200** -0.000198*** -0.000175** -0.0000495 -0.000234** -0.000169**

(-0.84) (-3.03) (-3.80) (-3.16) (-0.36) (-3.11) (-3.33) invratio 0.0102 0.0238 0.0104 -0.00165 0.0122 0.0207 0.0286 (0.62) (1.09) (0.81) (-0.27) (1.06) (1.14) (1.18) govratio -0.00495 -0.0405 -0.0146 0.00468 -0.0260 -0.0545 -0.0686 (-0.32) (-1.00) (-0.60) (0.36) (-0.91) (-1.08) (-1.08) inflation -0.0632 0.00100** 0.00137** -0.00458 0.00851 -0.0212 -0.0310 (-0.95) (3.18) (2.82) (-0.15) (0.33) (-1.00) (-1.50) hc -1.507 0.828 0.704

(-1.41) (1.89) (1.57)

open -0.830 0.940 1.606** 1.224 1.165 0.224 0.549 (-0.57) (1.01) (3.28) (1.50) (1.76) (0.15) (0.31) decision 0.0190

(1.19) ownership -0.314 (-0.31)

y1983 0.0489 3.616 7.690 2.874 (0.01) (0.28) (0.73) (0.34) y2009 -1.738 1.171 -4.631 -3.123 -2.321 (-0.70) (0.49) (-1.44) (-0.86) (-0.54) lfp -0.00323 (-0.20)

GII -1.721 (-0.62) secondaryfem 0.00558 (0.59) primaryfem 0.0300**

(3.22) totalprimary -5.52e-10 (-0.00) N 67 1122 811 127 632 879 1106 t statistics in parentheses

* p<0.05, ** p<0.01, *** p<0.001

(29)

28

To start off, lagged GDP per capita growth has a high explanatory value in regressions 2, 3, 3, 6 and 7. This variable significantly and positively affects economic growth, as hypothesized in section 3. Furthermore, Initial GDP has the expected negative and significant effect in regressions 2, 3, 4 and 6. This is a sign of conformation of the convergence theory. Simply put, it suggests that poorer countries grow faster than richer countries. In regression 3 trade openness is significant at a level of 5%. This variable positively affects economic growth. So a higher level of trade openness is expected to lead to more economic growth. Lastly, in regressions 2 and 3 inflation has a positive significant effect on GDP growth per capita. This positive relation might be explained due to the fact that system-GMM estimations are short-run estimations, like suggested by the previous literature (Fischer, 1983). The results indicate that there is a positive relation between GDP growth and inflation in SSA in the short-run.

Now considering the economic empowerment proxies: women’s decision-making power, equal ownership rights, the gender inequality index, secondary- and primary education. Merely the percentage of girls as primary school pupils has a positive and significant effect on economic growth. The effect is significant at a level of 5%. This result is in accordance with what expected beforehand. Based on the results of table 4, we expect to see a 0.0293 percentage point increase in GDP per capita growth after a 1 percentage point increase in the share of girls attending primary school, ceteris paribus. The total of primary school attendance (girls plus boys) is not significant. This shows that especially girls attending primary school can significantly increase economic growth. Remarkably, the percentage of girls attending secondary education is not significant, whereas the share of girls attending primary school is. This could be explained by the fact that the primary school attendance rate in SSA is below the world average and still has room to grow.

Although the primary school enrolment in SSA lays close to 100%, only 56%

of those students persist to the last grade in 2015, whereas the world average was 81% in that same year (World Bank). Also primary education has continuously been shown to have the highest rate of return socially and privately (Psacharopoulos & Patrinos, 2018).

(30)

29

The rest of the above stated women’s economic empowerment proxies do not have any significant effect on economic growth according to the results found in table 4.

4.2RESULTS INDIRECT OUTCOME VARIABLES: POPULATION AND INVESTMENT

In table 5 the results are presented of the regressions with population as its dependent variable. This is the first of the two indirect outcome variables.

Table 5: System-GMM population regressions on a yearly basis.

t statistics in parentheses

* p<0.05, ** p<0.01, *** p<0.001

First, lagged population has the highest explanatory value for the population level. It is significant at a level of 1 % in all 7 of the regressions.

Next, inflation has a significant and negative effect on population in regressions 3, 4 and 5. This means that higher inflation results in a lower population level. Research has linked economic instabilities, like high

(1) (2) (3) (4) (5) (6) (7) pop pop pop pop pop pop pop L.pop 1.027*** 1.027*** 1.027*** 1.029*** 1.029*** 1.029*** 1.029***

(30 16.65) (1690.03) (1910.29) (812.74) (1125.56) (1080.50) (1028.01) GDP -0.0000171 -0.00000469 -0.00000333 -0.00000533 -0.00000139 -0.00000160 -0.00000400 (-1.11) (-1.10) (-0.48) (-0.81) (-0.58) (-0.73) (-1.15) inflation 0.00120 -0.00000470 -0.0000190* -0.0012 -0.00122* -0.00106* -0.000829 (0.54) (-1.51) (-2.65) (-0.44) (-2.47) (-2.34) (-1.21) density -0.0000382 0.000104 0.0000774 0.000645 0.0000737 0.0000703 0.0000876 (-0.14) (0.78) (0.51) (1.96) (0.88) (0.90) (0.91) hc -0.0448 -0.0651 -0.0993

(-0.62) (-1.49) (-1.77)

open -0.0292 0.00703 0.0252 0.146 0.0182 0.0173 0.0185 (-0.41) (0.26) (0.61) (1.94) (0.81) (0.78) 0.76) decision -0.000424

(-0.54) ownership -0.0448 (-1.36) lfp 0.00123 (1.57) GII 1.296 (1.85) secondaryfem -0.00107*

(-2.10) secondarytotal -0.00110 (-1.97) primarytotal -0.00126 (-1.70) N 64 1021 720 109 614 661 629

(31)

30

inflation, to lowered fertility (Sobotka, Skirbekk & Philipov, 2011). This could be a possible explanation for this result.

When considering the economic empowerment proxies, female secondary education completion negatively affects population (regression 5).

The variable is significant at a level of 10%. Suggestively, more women completing secondary education results in a lower population level. This result corresponds with previous predictions and literature. Women obtaining an education is expected to correspond with more power for women in intrahousehold decision-making and educated women are thought to make the decision to have less children (Berik et al., 2009). In comparison, regression 6 is the regression considering total secondary education completion (males plus females). This variable has no significant effect on population. This is a sign of conformation that especially women completing secondary education corresponds with lower population levels.

The remainder of the economic proxies have no significant effect on the population levels.

(32)

31

Table 6 display the results with investments as its dependent variable. This is the second indirect outcome variable that is considered.

Table 6: System-GMM investment regressions on a yearly basis.

t statistics in parentheses

* p<0.05, ** p<0.01, *** p<0.0001

Again the lagged variable has the highest explanatory value. Lagged investment in this case. It positively and significantly affects investment in all regressions below. Moreover, in regression 4 the government consumption to GDP ratio has a significant effect on investment. It is significant at a level of 5%. A higher government consumption to GDP ratio would result in more investment. Lastly, inflation also positively effects investment. Considering that system-GMM estimations are short run estimation. This result seems to suggest that there is a positive relationship between inflation and investment in the short run.

The economic proxies themselves have no significant effect on investment.

(1) (2) (3) (4) (5) (6) inv inv inv inv inv inv L.inv 0.775*** 0.784*** 0.752** 0.994*** 0.967*** 0.995***

(7.69) (4.23) (3.49) (16.21) (5.66) (19.58) GDP 0.0867 0.00717 0.00938 -0.0172 1.278 0.00760 (1.01) (0.65) (0.64) (-1.21) (1.04) (0.90) govratio 3.653 5.264 6.696 2.928** 2.355 7.920 (0.92) (0.99) (0.99) (2.94) (0.36) (1.21) inflation 30.75 0.441*** 0.474*** -7.660 30.03 0.389 (0.89) (5.90) (4.34) (-1.26) (0.59) (0.18) open -1762.5 -59.53 -159.0 -210.4 1770.5 144.6 (-1.90) (-0.81) (-0.89) (-0.63) (0.49) (1.13) decision -4.915

(-0.44) ownership 322.3 (1.17) lfp 1.475 (0.26) GII -1008.3 (-0.80) secondaryfem -16.71 (-0.22) primaryfem 0.0783 (0.06) N 72 1333 948 123 625 859

(33)

32

4.3ROBUSTNESS

In table 7 the results of the system-GMM regressions with real GDP per capita as dependent variable can be found. These results are considered a test of robustness.

Table 7: System-GMM real GDP per capita regressions on a yearly basis.

t statistics in parentheses

* p<0.05, ** p<0.01, *** p<0.001

First, similar to the results found in table 6, inflation has a positive effect on the level of GDP per capita in regression 3. Now also human capital has a positive and significant effect on real GDP per capita in regressions 1, 2 and 3. In comparison to the GDP growth results in table 3 female labour force participation has a negative effect on the level of real GDP per capita. This result is opposite than what was expected. It may be explained by lower productivity of women. Especially if there were excluded from obtaining an education. For a 1 percent increase in the share of women participating in the labour force real GDP per capita decreases with 93.08, ceteris paribus.

(1) (2) (3) (4) (5) (6) (7) GDP GDP GDP GDP GDP GDP GDP invratio -4.353 3.309 -5.994 9.782 0.148 -0.279 1.481 (-0.85) (0.34) (-1.27) (0.93) (0.07) (-0.11) (0.52) govratio -0.643 -11.74 3.540 -30.50 -6.690 -6.626 -13.33*

(-0.17) (-1.42) (0.62) (-1.03) (-1.96) (-1.66) (-2.08) inflation -83.18 0.786 1.414** -50.68 -22.64 -32.37 -26.75 (-1.35) (1.89) (3.42) (-1.31) (-1.07) (-1.30) (-1.42) hc 3072.3* 5468.2*** 5682.6***

(2.07) (4.58) (4.79)

open 54.19 1604.5 1591.5 4683.7 1337.6 1912.6 4588.5 (0.03) (0.77) (0.90) (1.37) (0.98) (1.19) (1.93) decision 13.29

(0.69) ownership -1941.9 (-1.30) lfp -93.08**

(-2.77) GII -36136.3***

(-5.48) seondaryfem 109.8***

(5.08) secondarytotal 106.0***

(4.19) primaryfem 15.14 (1.03) N 67 1152 811 127 660 709 907

(34)

33

Also the gender inequality index has a significant effect in these regressions. It is found to be significant at a level of 1%. Since higher on the gender inequality index means more gender inequality, this result suggests that more gender inequality in the dimensions of reproductive health, empowerment in education and parliament and the labour market increases the level of real GDP per capita. If you move up 0.01 on the Gender Inequality Index the real GDP per capita is expected to decrease with 307.43, ceteris paribus.

Lastly, both female secondary education completion and total secondary education completion have a positive and significant effect, these are significant at a level of 1%. This shows that secondary education completion in its entirety has a possible positive effect on real GDP per capita, rather than the female share of secondary completion. Although the effect of female share completing secondary education is slightly bigger, this is not significant difference.

The share girls attending primary school has been consistently positive and significant in the OLS, fixed effects4 and the system-GMM results considering GDP growth per capita. Surprisingly, female primary school attendance does not affect real GDP per capita according to the table in 7.

4 The OLS and Fixed effects estimations can be found in the appendix in table 10 and 11.

Figure

Updating...

References

Related subjects :