Legislative Quota, Women Empowerment and Development. Evidence from Tanzania

Abstract

This paper analyzes whether the legislative women’s quota implemented in Tanzania has helped to reduce the existing gender gap in that country. We focus on a set of development indicators indicated by the literature and an analysis of female political activity. We exploit the variation in the number of female representatives across the 131 districts of Tanzania, employing a Difference and Differences approach including fixed effects and controlling for a number of socioeconomic variables. Our analysis indicates that the legislative women’s quota in Tanzania has led to significant reductions in the gender gap and improvements for women. The quota has effectively increased political participation in accordance with its goals, and the level of female representation continues to rise. We find evidence that the quota has reduced the gender gap in education for certain age groups, and we find indications of small improvements to female empowerment. In accordance with previous findings in other countries, we find that the increased female representation has led to substantial investments in water infrastructure that has greatly increased the number of people with access to clean water. While we do not find significant health impacts, this may be due to limitations in our dataset.

June 5 th, 2015

1 Introduction

The improvement of global gender equality and the empowerment of women worldwide is one of the eight UN millennium development goals. In the past two decades significant progress has been made in achieving this goal. According to the world gender gap report in 2014, the gender gaps in women’s educational attainment (94%) and in health and survival (96%) have almost been closed¹. In contrast, the gender inequality in economic participation and opportunity (60%), and in particular the gender gap in political empowerment (21%) remain far from being balanced Hausmann et al. (2014).

Although there has been great progress in some areas, gender inequality is still prominent in many societal aspects, particularly in the developing world. Women are often not granted the same rights and opportunities as men and are left with social and economic disadvantages, which have negative effects for an economy as a whole. Since human capital is one of the main drivers of an economy, the underuse of half of a country’s population can have far-reaching consequences for long-term economic growth and development.

Figure 1: Global Levels of Discrimination against Women

illustration not visible in this excerpt

Source: Social Institutions and Gender Index 2015

Figure 1 illustrates that discrimination against women is most prominent on the African continent. Africa lags behind most parts of the world in closing its gender gap on education and health, but is well ahead of many emerging regions on closing the gap in political empowerment. In order to redress gender inequality and its hampering consequences, a handful of African countries (e.g. Eritrea, Rwanda, Sudan, Tanzania and Uganda) have employed women quotas in legislature. Since many African countries have severe gender imbalances in legislature, but only a few are employing a quota system to address this disparity, it is crucial to evaluate the impact of such a policy in order to evaluate its usefulness for other developing countries.

In this paper we focus on the special seat system for women in Tanzania. Our main objective is to analyze the effects of this quota system on a set of development indicators and by these means to provide a sophisticated answer to the following policy question: Did the legislative women ’ s quota reduce the existing gender gap in Tanzania? In particular, we are interested in outcomes related to education, health, the quality of infrastructure and female empowerment.

Tanzania’s quota system was first introduced with relatively mild requirements in 1985, though the requirements were increased substantially for the 1995 elections to require female representation to account for at least 15% of traditional seats in parliament. This requirement was increased to 20% in 2000 and to 30% in 2005. Tanzania is a particularly interesting case to study for a number of reasons. Firstly, the prevailing patriarchal society, favoring segregate gender roles, makes it a good starting point to analyze the effect of the legislative women’s quota.

Furthermore, after the special seat system for women was first introduced, women managed to push for laws that address women’s concerns in several areas, such as maternity leave for mothers, a sexual offence bill, a law that promotes enrollment of women in tertiary education and a land law reform that addresses discriminatory practices against women (Meena (2003)). Furthermore, while data availability is typically a major issue for developing countries, data for Tanzania is readily available.

Besides the direct channel of more female-oriented policies, the quota might also induce an indirect change in women’s roles in society, i.e. a higher representation of women in politically influential positions might incentivize young women to pursue similar paths and at the same time lead to changes in cultural norms. However, the effectiveness of a legislative gender quota is debatable. Duflo (2012) concludes that a one-time impulsion of women’s rights is not sufficient in order to change entrenched political norms and values that discriminate against women, but instead further complementing measures are required.

In order to measure these effects we exploit the variation in the number of female MPs across the 131 districts of Tanzania employing a Difference -in- Differences (DiD) approach including fixed effects and controlling for a number of socioeconomic variables. For this purpose we are using various data sources. We are working with four extensive micro level datasets (Demographic and Health Surveys (DHS) with more than 178,000 observations), ranging from 2003 to 2012 and a self-generated database, which contains information about Tanzanian MPs for the past three legislative terms (2000, 2005, 2010). Using GPS data we match villages from the micro-level dataset with the districts of the country and the information on female representation by district.

We find significant evidence that the quota has reduced the gender gap in education for certain age groups, moreover we find indications of small improvements for female empowerment for some age groups. In accordance with previous findings in other countries, we find that an increase in female representation leads to substantial investments in water infrastructure that greatly increased the number of people with access to clean water. While we do not find significant health impacts, this may be due to limitations in our dataset.

The rest of this paper is structured as follows. Section 2 provides a literature review. Section 3 gives an overview of the quota in Tanzania, its implementation and its effect on female political participation. Section 4 looks at testable implications. Section 5 describes our empirical strategy and Section 6 describes our dataset in more detail. Section 7 provides our main analysis and section 8 provides our policy evaluation.

2 Literature Review

There is substantial research that analyzes the relationship between gender inequality and economic growth and development. The theoretical literature regarding gender inequality in education focuses on the insufficient exploitation of human capital. Klasen (2002) argues that a higher marginal return to education exists for girls, that if exploited could lead to substantial growth. Furthermore, higher education of women is expected to lead to both lower fertility and child mortality rates as well as a better educated following generation (Esteve-Volart (2004) ; Cavalcanti and Tavares (2007)).

The same argument is often applied when considering the effect of gender gaps in labor market participation on economic growth, i.e. that existing human capital is not being efficiently exploited (Klasen (2002)). Moreover, higher female employment has been shown to increase women’s bargaining power at home, which consequently might lead to higher investments in children’s health and education, fostering human capital formation of the following generation (Seguino and Floro (2003)). Finally, recent literature has argued that women tend to be less prone to corruption than men (Dollar et al. (2001), Swamy et al. (2000)). Hence, a higher female participation in the labor force and higher education for women may lead to less corrupt governance in business and policymaking.

Another line of research has demonstrated that increasing female political participation can reduce the gender gap in a variety of areas. Thomas (1991) as well as Besley and Case (2003) find evidence that increased political representation of women is correlated with different spending priorities, and Clots-Figueroa (2011) leverages close elections between men and women in India to show that women tend to invest more in education and make more pro-female policies. Aside from the direct effect of passing more female-oriented policies, there is increasing evidence that increased female representation can reduce the gender gap through its effect on social norms. Beaman et al. (2012) demonstrate that female leadership has an impact on adolescent girls’ career aspirations and educational attainments, which they attribute to a role-model effect. According to them, this role-model effect may influence girls’ notion of women’s status in society and thus may influence them to break with prevalent gender stereotypes. Therefore, being exposed to a female leader might increase girls’ ambitions and their propensity to enter male dominated areas. For rural India, the gender gap in aspirations closed by 25% for parents and by 32% for youths in villages that had exposure to a female leader for two election periods. Furthermore, in these villages the gender gap in educational attainment was eradicated and girls tended to spend less time on household work Beaman et al. (2012).

Recent literature has demonstrated that women quotas lead to increases in women participation in government. Yoon (2011) gives evidence that women quotas in Africa increase female legislative representation, and Jones (1998) finds similar evidence for Argentina. Dahlerup (2003) also documents the cases of Rwanda, South Africa and Costa Rica, where gender quotas have led to large increases of women representation in government.

Evidence on such quotas from a variety of settings indicates that required political represen- tation has an effect on policy choices and outcomes. Chattopadhyay and Duflo (2004) study a reservation policy for women in rural India. They find that gender-specific preferences of political leaders have significant effects on policy choices, implying that female political leaders better represent women’s preferences. In regions where women complained relatively more about specific types of infrastructure, women-led councils showed higher public spending for these types of infras- tructure. Beaman et al. (2010) use data from the Millennial Survey spanning eleven Indian states and show that on average, gender quotas result in increased investment in water infrastructure and education. Pande (2003), when looking instead at required political participation for various caste groups in India, finds increased transfers to those groups. On the other hand, Kotsadam and Mans investigate the effects of gender quotas in national elections in Latin America and find that while quotas substantially increased the number of women in parliament, they had no effect on political participation, public policy, or corruption.

Multiple studies demonstrate a change in cultural norms following the introduction of women quotas. Beaman et al. (2009) present evidence for changes in voter attitudes after being exposed to the quotas. According to their results, women were more likely to campaign and get elected conventionally in councils that were required to have a female leader in the previous two elections. Furthermore, reservation led to a decrease in gender discrimination by men. Beaman et al. (2010) further show that the likelihood that a woman speaks at a village meeting in India increases by 25% when local political leader positions are reserved for women. Furthermore, there is evidence that the effects of women’s quotas persist over time. Paola et al. (2015) show that gender quotas in Italy increase women’s representation in politics even after the quota was terminated. To our best knowledge, we are the first to quantitatively analyze the effects of the legislative women’s quota in Tanzania.

3 Quota in Tanzania

In order to better understand the effects of the quota system in Tanzania and to guide our micro-level analysis of outcomes, we first investigate the direct effects of the quota on female political participation in Tanzania. We use a three phase analysis consisting of 1) identifying the quota framework 2) evaluating the implementation of the quota and 3) analysing the political activity of female MPs.

3.1 Quota Framework

The quota in Tanzania was implemented to address large gender gaps in parliamentary represen- tation. High female participation in the struggle for independence and the nationalist movement attracted women to politics and helped motivate the need to address the gender gap in repre- sentation (Yoon (2008)). The quota is implemented through reserved seats called Special Seats.

The system was implemented in 1985, originally with 15 seats reserved for women. In 1995 the quota increased to require that 15% (37 seats in 1995) of the total number of traditional seats in parliament be added as special seats for women. In 2000 it was increased to 20%, and in 2005 it was increased again to 30% (Meena (2003); Yoon (2008); Yoon (2011)). The total size of parliament has been increasing over this timeline as well.

3.2 Implementation of the Quota

Table 1 shows the progression of the quota and female representation in parliament for the years 1985 to 2010. In each year the number of special seats women in parliament surpassed the level mandated by the quota. Furthermore, female representation continued to increase in the 2010 elections despite no increase in the quota. We are also interested in how the number of women elected to a constituency has changed over time. If the increased female representation resulting from the quota has caused more women to feel capable of leadership, or if the increased female representation has caused the public of Tanzania to have more faith in women as leaders, we might expect to see more women winning constituency seats. Indeed we find that the number of women elected to a constituency has also been increasing to keep track with the quota. These women made up between 17% and 19% of all women in parliament for each of the elections between 1985 and 2010.

Table 1: Women in Parliament

illustration not visible in this excerpt

Source: Yoon (2008), Keith (2011)

3.3 Political Activity of Female MPs

Thus far we have established that the quota has successfully led to a corresponding increase in female representation. In order to fully assess the impact of the quota, we next evaluate what these additional women have done once they have reached parliament. Ideally we would analyze the number and scale of policies put forth by female MPs compared to their male counterparts, as well as the pass rate of such policies. Furthermore we would identify any systematic differences in the type of policies put forth by men vs. women. Unfortunately we were unable to obtain data at this level of detail. Instead, we are limited to data on the gender makeup of parliamentary committees over the past four terms. Committees in Tanzania are made up of “[. . . ] several members of parliament with a specific goal and time-frame regarding a particular/distinct subject of concern” POLIS 2015

Under the assumption that MPs are active in the subject area of a committee they are on, a higher percentage female makeup of a committee would indicate that female MPs have a larger platform in that area. Tanzania Parliament’s website (2015) lists 31 committees with more than five members over the past four terms (1995, 2000, 2005, 2010). Committees with five or fewer members over this time period were dropped from this analysis so that only major committees are considered. These committees were grouped into ten overarching policy categories. Figure 2 shows the percentage female representation across the ten policy areas for the pooled data from 1995 to 2010. We do indeed see variation in committee makeup, with higher concentration in areas indicated by the literature like health and social welfare (Duflo (2012)), and lower concentration in more stereotypically male-dominated areas like foreign affairs, defense and security.

Figure 3 shows the evolution of the percentage of female representation, as well as the total number of female representatives in these committees over time. A few things are worth noting. First, while certain categories of committees tend to be made up by a higher percentage of women, there is no clear trend in this over time. In particular, no category appears to be becoming more female-centered over time. Second, the categories that have the highest female representation are mid-size categories. The Health category only consists of the HIV/AIDS Affairs committee, although some committees in the Social Welfare/Development category likely have health-related responsibilities. Third, there is a general increase in the number of women in the mid-to-large size categories.

4 Testable Implications

Based on our literature review and analysis of female political activity in Tanzania, we identify four channels through which the increase in female political participation is likely to affect outcomes in Tanzania, including 1) the direct effect of policy changes 2) the effect on social norms 3) the role-model effect and 4) the effect of incentives for re-election. The first three are stressed throughout the literature on both female representation and quotas, and the fourth is relevant as more female-oriented policies could encourage the support of more female voters in the future.

This last point is relevant as many special seats women attempt to win constituency seats later in their careers (Yoon (2008)). We further draw on the relevant literature and political analysis to identify relevant outcome areas where the increased female representation in Tanzania is likely to have an impact. These include 1) education 2) health 3) female empowerment and 4) water infrastructure. Previous studies have found positive impacts for each of these outcomes as the result of increased female representation. Furthermore, all four outcome areas are potentially affected by the two committee categories in Tanzania with the highest percentage of female representation, Health and Social Welfare/Development. We would ideally include additional outcomes indicated by the literature such as labor force participation, but we are limited by the data.

5 Empirical Strategy

This section describes the empirical strategy that is used to measure how an increase in political representation affects the gender gap in development outcomes. As the parliamentary quota is imposed simultaneously throughout Tanzania, there is no variation in the quota start date. Furthermore, while the quota extends to local councils, local government data for Tanzania is unavailable. Instead we use variation in female representation across districts to estimate

Figure 2: Female Representation in Committees

illustration not visible in this excerpt

Source: POLIS 2015

Figure 3: Evolution of Female Representation in Committees

illustration not visible in this excerpt

Source: POLIS 2015

the effects of the quota. In order to validate this approach it is important to understand how women become representatives and how these representatives are distributed across districts. If female representatives come overwhelmingly from one geographical area, we may not have the necessary variation for our analysis. Furthermore, if female representatives come predominantly from well-educated areas we may encounter problems of reverse causality.

[...]

¹ The figures reported here refer to the ratio of female to male outcomes.