Quota Shocks: The Budgetary Implications of Electoral Gender Quotas

Transcription

1 Quota Shocks: The Budgetary Implications of Electoral Gender Quotas Amanda B. Clayton Harvard Kennedy School & Vanderbilt University Pär Zetterberg Uppsala University The rapid expansion of electoral gender quotas in the past two decades has been met with considerable scholarly and public attention. Despite this, there has been little empirical work examining the substantive effects of gender quotas across cases and over time. We make use of aggregate data on a wide sample of countries to examine how electoral gender quotas may affect changes in legislative spending priorities, a still largely unexplored dimension of women s substantive representation. We use time-series crosssectional data from a global sample of states during the peak period of quota implementation ( ) to examine how quotas affect subsequent changes in state spending priorities. Controlling for other possible determinants of state budgetary priorities, we find that substantial quota shocks associated with a high degree of change in women s parliamentary representation are associated with subsequent increases in state health expenditures, but no change in education, military, or aggregate spending. We provide preliminary evidence on mechanisms relating quota policies to women s substantive representation and the prioritization of health spending. Paper prepared for presentation at the 4 th European Conference on Politics and Gender (ECPG), Uppsala, June 11-13, 2015 Very Early Draft. Please Do Not Circulate or Cite 1

2 Introduction One of the most significant institutional developments in representative democracy in the last twenty years is the recent, rapid, and global diffusion of quotas for women in politics. Taken together, these policies have deeply transformed the composition of legislative bodies worldwide. Indeed, women s average share of parliamentary membership has nearly doubled between 1995 and 2015, moving from 11.3 percent to 22.1 percent in the last two decades (IPU, 2015); a trend made possible by the increasingly widespread use of gender quotas. Responding to this phenomenon, studies related to the causes and consequences of electoral gender quotas have become one of the fastest growing subfields in gender and politics research. Yet despite what has developed into a rich and substantial literature, still relatively little work has attempted to ascertain the substantive effects of quotas from a comparative perspective, in large part because of the difficulty of comparing the potential effects of quota policies across cases. Here we contribute to this literature in at least three ways: first, we present one of the first large-n state-level longitudinal analyses of the substantive effects of gender quotas. Second, by coding the quota variable in different ways, we single out the potential direct effects of quotas from their indirect effects. Third, we are able to make cross-national comparisons because we examine an outcome variable that to date has been largely overlooked in the quota literature: changes in budgetary priorities following quota implementation. Most studies that focus on potential substantive effects following quotas look for changes in policy outcomes, specifically whether quotas are followed by greater legislative attention to policy areas concerning women s rights or women s welfare (see e.g. Franceschet, Krook, & Piscopo, 2012; Franceschet & Piscopo, 2008; Goetz & Hassim, 2003). These types of outcomes are inherently difficult to study comparatively for at least two reasons. First, the ways in which members of parliament (MPs) affect the legislative process varies substantially across cases - and second, gender equality legislation may take varying lengths of time to pass (if passed at all), and their occurrence is not particularly frequent across cases. 2

3 Examining budgetary priorities as an outcome variable, then, has several advantages. First, there are consistent and comparable annual data available across states during the period in which the majority of quota implementations have occurred. Second, state spending is an essential government function that all states must accomplish each year, rather than a rare event. Third, controlling for total expenditures, budgetary allocations represent the distribution of somewhat finite resources, and therefore represent trade-offs that, at least to some degree, reflect politicians priorities in government investments. Examining choices in state expenditures represents a relatively new approach in the quota literature and, at the same time, examining the demographic composition of legislatures is a fairly unexplored dimension in the political economy literature on the cross-national determinants of state spending (see, however, Bolzendahl, 2009). To date, most country-level longitudinal studies on state spending priorities and on redistributive spending in particular have largely focused on states macro-economic incentives or political climates rather than budgetary priorities arising from MP preferences based on identity politics within legislatures. To examine the substantive effects of quotas, we measure both how quota implementation in general may affect state spending priorities, as well as how the depth of these policies - measured as the level of change quotas initiate in women s parliamentary representation are related to subsequent changes in state expenditures. Quota policies have had varying levels of effectiveness in their intended outcome to increase women s numeric representation in political decision-making (Tripp & Kang, 2008). Indeed, the descriptive changes in women s representation following quotas what we call here quota shocks can range from essentially non-existent (no increase in women s numeric representation) to dramatic (women s numbers increasing several fold). We argue that measuring subsequent changes in budgetary priorities following semi-exogenous quota shocks of different depths allows us to test if and how quotas affect the representative process in setting state budgets, one of the most basic government functions. This paper is organized as follows: Section 1 reviews the literature on gender 3

4 quotas, women s substantive representation, and states spending priorities. Section 2 presents our theoretical framework. Here we present several hypotheses that we derive from mechanisms relating quotas to state spending outcomes. We propose that quotas may send cues about legislative priorities to all members of parliament, that quotas allow more opportunities for female MPs to include their preferences in budgetary priorities in different ways, and/or that quotas provide increased attention to the budgetary preferences of female citizens. In Section 3 we detail our data and methodological approach, and in Section 4 we present our results. In sum, we find that high quota shocks are associated with subsequent increases in health spending both overall and when controlling for gross expenditures but not to significant changes in education or military spending or total budget size. Section 5 presents a discussion of our findings as they relate to our theoretical expectations, with a preliminary discussion of mechanisms. The final section concludes by outlining the future direction of this project. Literature review Women s Substantive Representation: Process vs. Outcome As quota policies have spread worldwide, scholarly attention to the consequences of these policies has grown in tandem (Franceschet et al., 2012; Krook, 2008; Schwindt- Bayer, 2010). Empirical research on the consequences of gender quotas on women s substantive representation often distinguish between the process of representation as distinct from the potential outcomes of this process (see e.g. Franceschet & Piscopo, 2008). Researchers concerned with the former have presented evidence that women s increased presence may bring in new less combative and more effective legislative styles into the political arena in both quota (Cowley & Childs, 2003) and non-quota (Anzia & Berry, 2011; Volden, Wiseman, & Witmer, 2013) settings. Other works has examined women s behavior in legislatures particularly around issues that disproportionately affect women. Some research has indicated that women, quota-elected and non-quota-elected alike, are more likely than men to voice concerns about women s welfare in plenary debates (Clayton, Josefsson, & Wang, 2014b; Xydias, 2007). Other work, however, has argued that the ability of quota-elected MPs to legislate effectively on women s issues 4

5 may be hindered by quota women s over allegiance to ruling parties, particularly in non-democratic or highly patriarchal settings (Clayton, Josefsson, & Wang, 2014a; Goetz, 2003; Longman, 2006), or by MPs stereotypes about the qualifications of women elected through quotas (Franceschet & Piscopo, 2008). When addressing the potential outcomes of electoral gender quotas, most political scientists have examined whether quotas have been followed by subsequent changes in gender-related public policy. For instance, in the first edited volume examining the substantive impacts of quotas, Franceschet et al (2012, pp., 12) refer to work on quotas and women s substantive representation as studies seeking to establish whether quota introduction increases the number of policies proposed, debated, and passed on behalf of women as a group. To date, research on whether quotas have been associated with advances in policy passages have been mixed. Some work has found that female legislators are more likely to sponsor or co-sponsor bills related to women s rights (Barnes, 2012; Franceschet & Piscopo, 2008; Kerevel & Atkeson, 2013) and that quotaelected women have been instrumental in the successful passage of these bills (Kerevel & Atkeson, 2013). Other work, however, reports mixed findings (Devlin & Elgie, 2008) or a general lack of success in advancements in gender-equalitarian legislation following quota implementation (Franceschet & Piscopo, 2008; Htun, Lacalle, & Micozzi, 2013). Measuring Outcomes and Defining Women s Interests Just like research on gender and political representation in general, work investigating if and how quota policies have affected outcomes related to women s substantive representation often build on the expectation that male and female citizens have different social and biological experiences, which leads to divergent political preferences (Sapiro, 1981; see also Young, 2000). Studies connecting gendered differences in political preferences to the legislative activities of female politicians often first specify women s interests through predefined issue areas that objectively and disproportionately affect women s welfare, such as maternal health or combating domestic violence (Clayton et al., 2014b). Other work measures women s interests in 5

6 areas that may also indirectly affect women given traditional gender roles, such as investments in child health (Miller, 2008). Rather than examining pre-specified interests, yet another group of scholars have attempted to first ascertain whether there are gendered differences in policy preferences through constituent surveys, and then measure the extent to which women s increased presence in legislative bodies affects these priorities (Chattopadhyay & Duflo, 2004). Importantly, this body of work reveals gender differences in legislative and spending priorities vary by context and are seldom uniform. For instance, Miller (2008), Swiss et al (2012) and Bhalotra and Clots-Figueras (2014) find evidence that female citizens prioritize increase public spending on health across a range of countries, whereas Gottlieb et al. (2015) and Schwindt-Bayer (2006) find that health is a gender-neutral policy area across African and Latin American cases respectively. Even in the same setting, there are disparate claims. For instance, while both examining evidence from India, Chattopadhyay and Duflo (2004) find that male citizens prioritize investments in education, whereas Clots-Figueras (2012) argue that education is a female-preferred policy area. Quota Effects on State Spending In this paper we focus on state spending priorities, a highly relevant but less researched outcome in the literature on quota effects. We argue this is an important dimension of women s substantive representation, as a state s ability to address gender inequalities relates not only to women s legal rights but also to legislative priorities when allocating finite financial resources to different policy areas. Although there is variation across states, spending priorities also reflect, at least to some degree, MPs preferences and influence. Despite the budgetary implications of electoral gender quotas, most work on female legislators influence in spending priorities have either excluded the role of quotas (Bolzendahl, 2009; Swiss et al., 2012) or included quotas as a methodological tool to address issues of endogeneity (e.g. Chattopadhyay & Duflo, 2004; Chen, 2010). Further, research on the cross-national determinants of state spending typically prioritize explanations related more to broader political contexts, such as the depth of democracy (Albertus & Menaldo, 2014; Huber, Mustillo, & Stephens, 2008) or macro-economic 6

7 variables, such as the effects of globalization (Wibbels, 2006), foreign direct investment (Avelino, Brown, & Hunter, 2005), and trade (Wibbels & Ahlquist, 2011), and less to matters of identity politics within legislatures. As a final relevant literature, we also draw on work that examines the microdeterminants of legislators ability to affect budgetary decisions as they work within different institutional constraints. A rich body of work examines how political institutions affect both the size and priorities of government spending (Besley & Case, 2003; Persson & Tabellini, 2005), related to, for instance, legislature size (Petterson-Lidbom, 2012), level of bureaucracy (Ting, 2012), bicameralism (Heller, 1997), and executive control in presidential versus parliamentary systems (Lienert, 2005). Although we limit our analysis to changes within rather than between states in this article, and thus hold political contexts relatively fixed, an important extension of this research is to understand how quotas affect budgetary outcomes when they are embedded within different types of political systems. Theory We focus our approach to understanding the budgetary implications of electoral gender quotas at the MP level. By having such an approach, we make the assumption that legislatures and legislators matter for how public spending is allocated to different policy areas (see e.g. Lienert, 2005 for an overview of the role of the legislature in budgetary processes). We argue that quota policies have the potential to change priorities and behavior within the legislature, which may result in a change in the allocation of public spending projects. We suggest at least two broad ways in which such changes may occur. First, quotas may affect outcomes directly by sending specific policy-generated cues to officeholders, thereby generating behavioral changes among elected representatives (c.f. Franceschet et al., 2012). Second, when properly enforced, quotas may change priorities and behavior indirectly by bringing more women into legislatures. If male and female MPs have different policy priorities, the female newcomers may act 7

8 differently than the men they replace, thus, generating a change in public spending priorities (Chattopadhyay & Duflo, 2004; Childs, 2004). Developing the argument, and starting with the former (direct) link between quotas and public spending, previous work has suggested the quota policies themselves may change legislators priorities and behavior regardless of the policy s strength. In particular, quotas may send cues to female MPs, increasing their mandate to represent women as an important constituent group. As a consequence of quota reforms, and thus by the politicization of gender in politics and society, female legislators are likely to perceive an obligation to act on the behalf of women (Franceschet & Piscopo, 2008). In this scenario, mandate effects may prompt changes in female MPs spending priorities regardless of whether the quota policy generates a large leap in the number of female legislators. H1a: Quota policies affect changes in budget allocations by giving female MPs a mandate to represent women s interests. In addition, quota policies may also change the priorities of male legislators toward areas that they believe better reflect women s preferences. All legislatures and parties that have made decisions about adopting gender quotas have been maledominated; thus, male politicians also have been exposed to, and participated in, quota debates. As a consequence, the introduction of quota policies may draw attention to gender equality issues more broadly and thus to issues that women find particularly important. Male representatives awareness about gender issues may therefore increase, which potentially may generate a reorientation of male MPs priorities (c.f. Miller, 2008). If the propositions about policy-generated cues and awareness raising of gender equality issues are true, we should find that the introduction of quotas causes a reorientation of spending priorities among both male and female MPs. After quotas are adopted, MPs should prioritize issue areas that they believe reflect women s preferences in budgetary allocations. 8

9 H1b: Quota policies affect changes in budget allocations by sending policy cues about women s legislative priorities to all MPs. The second (indirect) link between gender quotas and public spending suggests that an increase in women s substantive representation is caused through an increase in women s presence in political bodies. According to the argument, quotas need not only be adopted but also generate significant leaps in the number of women in parliament to bring about a change in public spending priorities. As more women enter legislative bodies through quotas, female newcomers may contribute to changing dynamics in politics and policymaking (see e.g. Chattopadhyay & Duflo, 2004; Chen, 2010). We suggest at least three ways in which an increased number of women may change public spending patterns. First, assuming male and female MPs have different legislative priorities, women s increased presence in political bodies will cause an aggregate preference shift in the legislature and, ceteris paribus, will affect collective legislative decision-making outcomes. The preference shift could, on the one hand, be a reflection of female representatives true preferences, due to the specific experiences that women bring into politics (c.f. Young, 2000). On the other hand, a preference shift may also be related to how female representatives perceive the interests of their female constituents, which they may believe correctly or not to have different legislative priorities than men (see also Franceschet & Piscopo, 2008). Thus, when attempting to be responsive to constituents needs, women in office may call for a slightly different agenda than their male colleagues. H2a: Quota policies affect changes in budget allocations through an increased number of female MPs, and an aggregate preference shift toward female-preferred policy areas. A difference in preferences, however, may not necessarily generate a shift in policy outcomes (see e.g. Tamerius, 1995). For instance, we cannot assume that gender concerns are always at the top of female legislators priority list and thus that issue preferences are translated into political action (see also Lee, 2007). Rather, quota-elected MPs may choose to act when they perceive that they are in a position to do so effectively. 9

10 A further way in which women s increased presence may change public spending is therefore to change the composition of formal preexisting institutions. As more women enter parliament through quotas, the supply of potential female committee members increases, and thus the likelihood that women are able to enter important committees such as finance or area-specific committees (e.g. education, labor, health, defense, etc.) that lobby the executive or the finance committee in order to get as big piece of the cake as possible. Through having different policy priorities than men and having access to influential positions in the legislature, women may be able to push for increased public spending in areas that they perceive are important for women. H2b: Quota policies affect changes in budget allocations due to gendered differences in legislative priorities and the increased number of female MPs in critical positions in the budgetary process. A third and final way in which an increased number of female legislators may change public spending is to mobilize collectively (including making use of informal structures) to have an impact on formal decision-making processes. Such mobilization includes the creation of new gendered institutions or a gendering of already established institutions. As regards the former, women can, for instance, create cross-partisan alliances (e.g. a women s caucuses) within the legislative body to lobby their male colleagues to give priority to certain issues and expenditures (e.g. Wang, 2013). With more women in office, the bargaining power of such formal or informal networks is likely to increase. An alternative strategy is to use the established structures of political parties. As more women achieve public office and thus attain leadership positions, women s possibilities to create or make use of powerful intra-party alliances is likely to increase. Such alliances may negotiate party positions, reorient party platforms, and potentially affect policy in ways that benefit women as a group (see also O'Brien, 2013). H2c: Quota policies affect changes in budget allocations by allowing women more opportunities to collectively mobilize on issues prioritized by women. If the propositions about women s descriptive representation are true, we should find that subsequent change in public spending are positively related to the magnitude of 10

11 the quota shock. Once quota policies generate significant leaps in the number of women elected to legislative bodies, more money should be spent within policy areas that are perceived as being particularly important for women; and less money should be spent on traditionally male policy areas. In the empirical analyses that follow, we test both for direct and indirect quota effects related to the implications stemming from these mechanisms. In addition, we contribute to the literature by examining the nature of any change in public spending as a result of the quota policy: that is, whether the change due to a change in budget size or to a reorientation of spending priorities (c.f. Besley & Case, 2003; Persson & Tabellini, 2005). Theoretically, it is crucial to understand whether quotas have the potential to redistribute funds from male-preferred to female-preferred policy areas, or whether the policy contributes to a recognition of female-preferred policy areas by increasing total spending (c.f. Fraser, 1998). Data and Methodological Approach Dependent Variables: State Spending by Sector We use time-series cross-sectional data on 152 countries (all states with a population over one million) to measure the budgetary effects of gender quotas across time. Our dependent variables consist of expenditure data by sector, measured as a percentage of GDP. The World Bank provides largely complete expenditure data on education, military, and health spending, as well as gross aggregate expenditures for the post-1995 period. Aside from relative data completeness, the post-1995 period is ideal for testing quota effectiveness, as the vast majority of quota implementations have occurred in the last twenty years. Indeed, in our sample only ten countries had fully implemented quotas before this period. This is due, in large part, to the momentum gained in global quota adoption after the UN s 4 th World Conference on Women held in Beijing in 1995, which for the first time at an international level recommended quotas as an effective means of increasing women s political representation. 11

12 Key Independent Variables: Operationalizing Quota Implementation We run two sets of models with different operationalization of quota implementation. The first is a simple dichotomous measure coded as zero for each year a country did not have a functioning quota and one for each year with an enforced gender quota in the single or lower house of parliament. Whereas up-to-date data on quota policies by country is publically available through the Quota Project, current data on years of quota adoption is not widely publically available. Building on information from the Quota Project, in addition to data compiled by Krook (2006) and Bush (2011) and other secondary sources, we construct our own coding of quota implementation by year. This measure includes all reserved seat and candidate list quotas that require over ten percent women on candidate lists or seats in parliament. We do not, for instance, include the eight percent reserved seat gender quota Egypt adopted in 1979 or the three percent quota Pakistan adopted in We also include voluntary party quotas, but only when they are regularly adhered to and when the total number of adopting parties holds at least thirty percent of the seats in the lower parliamentary house at the time of implementation. We do not, for instance, code Namibia as having a quota during the period because before 2015 only one small party (with less than 10 percent of seats) had an enforced candidate list quota at the national level. Importantly, however, we do not define quota implementation by the subsequent number of women elected to office. Even among voluntary political parties, subsequent changes in women s parliamentary representation in our data vary from a 0.7 percentage point increase in the number of female MPs elected (Romania s 2004 quota implementation) to a 13.7 percentage point increase (Ethiopia s 2005 quota implementation). Additionally, we code quota adoption as the year the quota was enforced; that is, the first year women who benefitted from the quota policy took office, rather than the year the quota was legally passed. Our dataset does not include countries such as Eritrea, which adopted a reserved seat quota in its 1997 constitution, but has not held elections 12

13 since this time. Similarly, we do not include, for instance, Uruguay, which adopted a national quota in 2009 that only took effect at the national level in the 2014 elections. To maintain consistency across cases, we code implementation as the first time a country (or major party) enforces a quota rather than separate measures for the cases in which quota laws have changed over time. Using these criteria, out of our sample of 152 countries, we code ten as having implemented quotas prior to 1995 (mostly among OECD-member states), and fifty-three implementations between 1995 and 2012 (mostly in the Global South). 1 In addition to a dichotomous measure of quota implementation, we also construct a new variable that measures policy strength. Here we measure the degree of change in women s representation that was initiated by the policy - what we call here, quota shocks. Quotas are intended to increase the numeric representation of women in politics and, as with most social policy, there are varying degrees to which this treatment is effective across states. The greatest quota shock in our data is Kyrgyzstan s reserved seat quota in 2007, which resulted in a move from zero women in parliament to 25.6 percent female representation in one election cycle. Afghanistan also experienced a significant quota shock, a 23.4 percent point change in women s parliamentary representation following the US-led invasion and new post-conflict constitution in On the other end of this scale, some countries have adopted quotas that impose requirements that are similar to current levels of female representation. For instance, when Panama enforced a candidate list quota in 1999, women s representation in parliament only increased from 9.7 to 9.9 percent. The mean quota shock in our data is a 9.7 percentage point increase. Women s mean parliamentary representation in the year preceding a quota shock in our data is 10.9 percent, meaning that, on average, when properly enforced, quotas have nearly doubled women s parliamentary representation in their first election cycle. 2 1 The ten pre-1995 implementers are: Argentina, Austria, Canada, Germany, Mozambique, Norway Netherlands, South Africa, Sweden, and Uganda. 2 By way of comparison, women s average parliamentary representation worldwide as of mid-2015 stood at 22 percent. 13

14 Although there is certainly variation in the level of treatment, in the aggregate, quotas have been quite successful in increasing women s descriptive representation worldwide. In 2012, the last year in our data, women held 26.9 percent of parliamentary seats in quota-implementing countries and 14.9 percent of seats in countries without a functioning quota policy. Figure 1 plots women s parliamentary representation centered on the year of quota implementation for each of the 152 countries included in our sample, and reveals the degree to which quota shocks have transformed the composition of parliamentary bodies worldwide in the past two decades. Figure 1 Here Controls: Additional Explanations for Shifts in State Spending In the regression models that follow, we also include a host of political, social, and economic covariates that the previous literature has indicated affect state spending choices. We include a dichotomous measure of democracy (a score of six or higher) and autocracy (a score of -6 or lower) from the Polity IV data. 3 We also include an index of cumulative democracy; measured as the total number of years a country has experienced democracy from 1945 to the observation year. In addition, we include the most recent Beck et al (2001) data on political institutions to record whether a state currently has a leftist ruling party, as well as a measure of the cumulative years of leftist party rule in the post-war period. We include a range of social and demographic variables from the World Bank Development Indicators, including the female / male labor force participation ratio. We include this variable on the assumption that women s participation in the labor force may be related to the political salience of women as a demographic group and, thus, the extent 3 Our results are also robust to inclusion of the standard polity score as well as different transformations of this indicator. 14

15 to which politicians factor women s preferences into political decision-making (see Iversen & Rosenbluth, 2010). The following models also include the percentage of the population that is urban, elderly (65 and over), and young (14 and under), as this may objectively affect the salience of funding particular types of social programs, such as health and education. We also include a host of economic variables from the World Bank Development Indicators, which various studies have indicated affect dimensions of state spending, including (all logged): GDP per capita, official development assistance (ODA) as a percentage of GDP, oil rents as a percentage of GDP, total trade (sum of exports and imports of goods and services) as a percentage of GDP, and foreign direct investment (FDI) as a percentage of GDP. We also model whether gender quotas are associated with general increases in budget size, as research on OECD cases has found evidence that women increasingly favor greater public spending and more redistributive policies in general (Edlund & Pande, 2002; Inglehart & Norris, 2003; Iversen & Rosenbluth, 2010; Lott & Lawrence, 1999). In the models that follow, all covariate values are lagged one year, except for gross spending which we theorize is more directly related to budgetary allocations for that year rather that the following year. Our data structure is time-series cross-section with an uneven panel. There are varying degrees of missingness across the included variables, with most missing values clustered in the measurement of our dependent variables (from 2 percent to 43 percent). In order to not loose unnecessary covariate information across model specifications, we create unique datasets for each dependent variable that list-wise deletes missing values of the dependent variable. Remaining levels of missingness are relatively low. All variables have fewer than five percent missing values, with the exception of oil revenue as a percent of GDP, which is missing from 16 percent of observations. We choose to use Amelia II for R to impute missing values rather than list-wise delete remaining observations. Estimation uncertainty due to imputation is included in the standard errors we report below. The results presented here are also robust to complete list-wise deletion. Results 15

16 We are interested in the extent to which quota policies are associated with changes in state budgetary priorities. Before turning to parametric estimates of these phenomena we first examine general descriptive patterns of state spending in both quota and non-quota cases across time. Given that states have adopted quotas at different times, it is difficult to visualize quota and non-quota descriptive patterns efficiently. In Figure 2 we present country-level time trends for health, education, and military spending separately for nonquota countries and quota countries however, the right hand panels do not reveal when during the eighteen-year period a country implemented the quota. Figure 2 reveals little variation in yearly trends between quota and non-quota countries related to education and military spending, but a slightly steeper upward trend for quota implementing countries as compared to non-implementing countries regarding health spending. More revealing among quota cases, Figures 3 5 show states spending patterns over time, centered on the year the quota was implemented. Whereas there do not appear to be great differences before and after quota implementation related to education spending, we see a slight downward trend related to military spending, although this seems to be driven by a handful of high military spending cases prior to quota implementation. We also note that military spending is particularly difficult to causally link to quota implementation, as gender quotas are often adopted in post-conflict situations. Figure 3 indicates that whereas health spending appears to be increasing in quota cases before implementation, the rate of change further increases in the post-quota period. We discuss the possible endogeniety of this relationship specifically that countries with already upward trending health spending are more likely to implement quotas when we present our results below. Figures 2 5 Here 16

17 In the models that follow, we use country fixed effects to measure how quota implementation and the depth of implementation are associated with changes in state spending within states over time. Fixed effects allow us to soak up country-level unobservable variables that may affect idiosyncrasies of state spending in ways that our models are not able to capture. The resulting coefficient estimates, then, tell us about within-country changes over time, rather than explaining variations in outcomes across cases. Our models also include a lagged dependent variable, allowing us to measure how quota implementation is associated with yearly changes in state spending priorities. We do this based on theory-driven interest in changes rather than levels of state spending, as well as for methodological reasons. Because state spending allocations are not static over time, de-trending our data allows us to observe whether the rate of change in our dependent variables alters from the pre to the post-quota implementation period. Table 1 shows these results for the dichotomous measure of quota implementation, and Table 2 show the results for quotas operationalized as the percentage point change in women s parliamentary representation following the initial quota shock. We drop the ten preexisting quota cases from the model data in order to obtain consistent estimates related to quota implementation during the post-1995 period. Coefficients in the first column under each dependent variable display model estimates without controlling for gross state expenditures, and those in the second column show results while controlling for this measure. Finally, Model 7 in both Tables 1 and 2 shows how quota implementation is associated with subsequent changes in total state spending. Tables 1 and 2 Here Table 1 reveals that a binary measure of quota implementation is not associated with significant changes in states budgetary allocations over time or with overall changes in aggregate spending. Table 2, however, indicates a positive association between the initial strength of the quota policy and subsequent increases in health spending, both with (Model 2) and without (Model 1) controlling for gross state 17

18 expenditures. These results suggest that strong quota policies are associated with a shift in legislative spending priorities, rather than an increase in the overall size of the budget. Indeed, Model 7 in both Tables 1 and Table 2 indicates that quota policies are not associated with an overall increase in gross state expenditures. Table 2 also reveals that the strength of quota implementation is not associated with significant changes in education or military spending. Related to effect size, the results from Model 1 on Table 2 indicate that moving up one standard deviation in quota effectiveness (a 6.8 percentage point increase in women s parliamentary representation) is associated with a yearly percentage point increase of health spending relative to GDP. Related to the mean quota policy in our data, this is equivalent to a percentage point increase in health spending relative to GDP over 8.3 years, the average quota policy s duration in our data. This is equivalent to a fifteen percent increase relative to this variable s mean value in our data (3.35 percent) over an 8.3-year duration. Robustness: The Possible Endogeniety of Quota Implementation One methodological concern related to our findings is the possibility that countries with already upward trending health spending might be more likely to adopt quotas. In this scenario, quotas are not causally related to changes in budgetary allocations, but rather are implemented as part of a host of reforms associated with a change in legislative priorities. Although by modeling changes rather than levels in the dependent variable we assuage some of these concerns, here we also run an event history analysis to test whether upward trending health spending is associated with quota implementation. Table 3 below shows the results of a cox proportional hazard model on the timing of quota implementation. Model 1 shows results including all cases of quota implementation in our data. Related to our finding that quota strength is a stronger predictor than a binary measure of quota implementation, Model 2 shows the results when only including cases with quota shocks higher than the mean value. Here we include average changes in health 18

19 spending in the three years before quota implementation, but our results are robust to one year and five year specifications as well. Table 3 Here Table 3 reveals that preceding changes in health spending are not associated with an increased likelihood of quota implementation, both across all included cases and among the high quota strength subset. We do note that in Model 1, countries with higher previous levels of health spending are more likely to adopt quotas, confirming our modeling choice to include country fixed effects and changes rather than levels of public expenditures. Discussion The results of our analysis show that gender quotas matter for public spending in one specific policy area: health. This finding supports previous work that has put improvements in health and increased health spending in relation to gender equality reforms such as women s enfranchisement (e.g. Miller, 2008) and to increased gender equality in parliaments (e.g. Chen, 2010; Swiss et al., 2012). On the other hand, we find that gender quotas are unrelated to government spending on education and to military expenditures. Importantly, not all quota implementations are followed by subsequent increases in health spending; rather these changes depend on the magnitude of the initial quota shock. In other words, by testing both for the possibility of direct (signaling) effects and for potential indirect effects (through an increased number of women in parliament), we are able to get a better understanding of the relationship. We can, thus, reject hypotheses H1a and H1b: It is not the cues that quotas send to MPs that matter, but rather the new composition of legislators that quotas generate (by replacing some male legislators with 19

20 female ones). In this regard, our findings suggest a clear link between gender quotas, women s descriptive representation, and women s substantive representation. By identifying an indirect relationship between gender quotas and public spending, the analysis also speaks and provides support to an influential but also criticized (see e.g. Childs & Krook, 2006) theory within gender and politics: critical mass theory (Kanter, 1977). It puts the number of women in parliament at the center of attention by claiming that a critical mass of female legislators is needed in order for women in parliament to be able to influence politics and policy. Although we are not able to identify a specific threshold level for when numbers matter, we are indeed able to show where (i.e. in which policy areas) the number of women in parliament plays a role. Conversely, by showing that quotas have no direct impact on aggregate-levels public spending, our analysis suggests that the presence of a few key (female) actors (or socalled critical actors) are not sufficient to influence budgetary processes (c.f. Bratton, 2005). Attempting to get a better understanding of how quotas through an increased number of women influence health expenditures, our models with total state spending as the outcome variable (Model 7 in Table 1 and Table 2) provide us with important pieces of information. By finding no impact of quotas on gross spending, we are able to conclude that quotas generate a reorientation of priorities and not an overall increase in public expenditures. In other words, women do not contribute to budgets by making them bigger but by reallocating financial resources from some policy areas to others. Theoretically, the reallocation of resources has important implications. It shows that quotas indeed contribute to shifting power from male MPs to female ones: They do not merely generate a recognition of women s interests and needs by adding funds to an existing budget; rather, they redistribute financial resources to areas that female MPs find particularly important (c.f. Fraser, 1998). In this respect, a larger number of women appear to disrupt established power balances in the legislature by increasing women s bargaining power. 20

21 Conclusion Using data on 152 countries across the globe, this paper has provided compelling evidence on the aggregate spending patterns of a wide range of states over time following quota implementation. More specifically, we have shown that gender quotas have budgetary effects on health spending. These findings point to several potential future extensions of this project, which is still in its early phases. First, as of yet, we do not have data that directly speaks to the causal mechanisms we identify, which are at the MP-level. Thus, we are not yet able to provide an answer as to whether hypotheses H2a-H2c are supported. However, we have obtained data on MP committee membership from a wide range of African cases, and we are hoping to obtain MP survey data from African and/ or Latin American cases. In future versions of this project, we hope to integrate these data into the paper to not only examine macro-patterns of quota effects (as we currently do), but also, at least tentatively, supply evidence relating directly to causal mechanisms. Another possibility might be to do a small-n analysis of a single country that has had a quota shock and has increased its health spending. Theoretically, we plan to integrate more closely literature on women s substantive representation with research on the role of the legislature in the budgetary processes. It is obvious that the power of the legislature over the budget varies across political systems (Lienert, 2005). We will take these differences into account in the next version of the manuscript. References Albertus, M., & Menaldo, V. (2014). Gaming democracy: Elite dominance during transition and the prospects for redistribution. British Journal of Political Science, 44(3), Anzia, S. F., & Berry, C. R. (2011). The Jackie (and Jill) Robinson Effect: Why Do Congresswomen Outperform Congressmen? American Journal of Political Science, 55(3),

22 Avelino, G., Brown, D. S., & Hunter, W. (2005). The effects of capital mobility, trade openness, and democracy on social spending in Latin America, American Journal of Political Science, 49(3), Barnes, T. D. (2012). Genderand Legislative Preferences: Evidence from the Argentine Provinces. Politics and Gender, 8(4), Beck, T., Clark, G., Groff, A., Keefer, P., & Walsh, P. (2001). New tools in comparative political economy: The database of political institutions. World Bank Economic Review, 15(1), Besley, T., & Case, A. (2003). Political institutions and policy choices: Evidence from the United States. Journal of Economic Literature, 41(1), Bhalotra, S., & Clots-Figueras, I. (2014). Health and the Political Agency of Women. American Economic Journal: Economic Policy, 6(2), Bolzendahl, C. (2009). Making the implicit explicit: Gender influences on social spending in twelve industrialized democracies, Social Politics, 16(1), Bratton, K. A. (2005). Critical Mass Theory Revisited: The Behavior and Success of Token Women in State Legislatures. Politics & Gender, 1(1), Bush, S. (2011). International Politics and the Spread of Quotas for Women in Legislatures. International Organization, 65(1), Chattopadhyay, R., & Duflo, E. (2004). Women as policy makers: Evidence from a randomized policy experiment in India. Econometrica, 72(5), Chen, L.-J. (2010). Do Gender Quotas Influence Women s Representation and Policies? The European Journal of Comparative Economics, 7(1), Childs, S. (2004). New Labour's Women MPs: Women Representing Women: Routledge. Childs, S., & Krook, M. L. (2006). Should Feminists Give Up on Critical Mass? A Contingent Yes. Politics & Gender, 2, Clayton, A., Josefsson, C., & Wang, V. (2014a). Present without presence? Gender, quotas and debate recognition in the Ugandan parliament. Representation, 50(3), Clayton, A., Josefsson, C., & Wang, V. (2014b). Quotas and Women's Substantive Representation: Evidence from a Content Analysis of Ugandan Plenary Debates. Paper presented at the American Political Science Association's annual conference, Clots-Figueras, I. (2012). Are Female Leaders Good for Education? Evidence from India. American Economic Journal: Applied Economics, 4(1), Cowley, P., & Childs, S. (2003). Too Spineless to Rebel? New Labour s Women MPs. British Journal of Political Science, 33, Devlin, C., & Elgie, R. (2008). The Effect of Increased Women's Representation in Parliament: The Case of Rwanda. Parliamentary Affairs, 61(2), Edlund, L., & Pande, R. (2002). Why have women become left-wing? The political gender gap and the decline in marriage. Quarterly Journal of Economics, 117(3), Franceschet, S., Krook, M. L., & Piscopo, J. M. (Eds.). (2012). The Impact of Gender Quotas New York: Oxford University Press. 22

23 Franceschet, S., & Piscopo, J. M. (2008). Gender Quotas and Women's Substantive Representation: Lessons from Argentina. Politics & Gender, 4(3), Fraser, N. (1998). From Redistribution to Recognition? Dilemmas of Justice in a 'Post-Socialist' Age. In A. Phillips (Ed.), Feminism and Politics. Oxford: Oxford University Press. Goetz, A. M. (2003). The Problem with Patronage: Constraints on Women's Political Effectiveness in Uganda. In A. M. Goetz & S. Hassim (Eds.), No Shortcuts to Power. African Women in Politics and Policy Making. London & New York: Zed Books. Goetz, A. M., & Hassim, S. (2003). No Shortcuts to Power: African Women in Politics and Policy Making. London: Zed Books. Gottlieb, J., Grossman, G., & Robinson, A. L. (2015). Do men and women have different policy preferences, and if so, why?, Afrobarometer WP153. Heller, W. B. (1997). Bicameralism and budget deficits: the effect of parliamentary structure on government spending. Legislative Studies Quarterly, 22(4), Htun, M., Lacalle, M., & Micozzi, J. P. (2013). Does Women's Presence Change Legislative Behavior? Evidence from Argentina, Journal of Politics in Latin Americe, 5(1), Huber, E., Mustillo, T., & Stephens, J. D. (2008). Politics and social spending in Latin America. Journal of Politics, 70(2), Inglehart, R., & Norris, P. (2003). Rising Tide. Gender Equality and Cultural Change Around the World. Cambridge: Cambridge University Press. IPU. (2015). Women in national parliaments. Accessed January 15, Iversen, T., & Rosenbluth, F. (2010). Women, work, and politics: The political economy of gender inequality: Yale University Press. Kanter, R. M. (1977). Men and women of the corporation. New York: Basic Books. Kerevel, Y. P., & Atkeson, L. R. (2013). Explaining the Marginalization of Women in Legislative Institutions. Journal of Politics, 75(4), Krook, M. L. (2006). Reforming Representation: The Diffusion of Candidate Gender Quotas Worldwide. Politics & Gender, 1(2), Krook, M. L. (2008). Quota Laws for Women in Politics: Implications for Feminist Practice. Social Politics, 15(3), Lee, F. L. F. (2007). Gender Differences in Campaign Materials and Politicians' Professional Orientations: The Case of a Hong Kong Election. Sex Roles, 57(11-12), Lienert, I. (2005). Who controls the budget: The legislature or the executive?, IMF Working Paper 05/115. Longman, T. (2006). Rwanda: Achieving Equality or Serving an Authoritarian State? In G. Bauer & H. E. Britton (Eds.), Women in African Parliaments. Boulder, CO: Lynne Rienner Publishers. Lott, J., & Lawrence, K. (1999). Did Women s Suffrage Change the Size and Scope of Government? Journal of Political Economy, 107, Miller, G. (2008). Women's suffrage, political responsiveness, and child survival in American history. Quarterly Journal of Economics, 123(3),