Jump to ContentJump to Main Navigation
Contagious RepresentationWomen's Political Representation in Democracies around the World$

Frank C. Thames and Margaret S. Williams

Print publication date: 2013

Print ISBN-13: 9780814784174

Published to NYU Press Scholarship Online: March 2016

DOI: 10.18574/nyu/9780814784174.001.0001

Show Summary Details
Page of

PRINTED FROM NYU Press SCHOLARSHIP ONLINE (www.nyu.universitypressscholarship.com). (c) Copyright University of NYU Press Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in NYSO for personal use (for details see www.nyu.universitypressscholarship.com/page/privacy-policy/privacy-policy-and-legal-notice). Subscriber: null; date: 22 January 2019

Contagion and the Adoption of National Quotas

Contagion and the Adoption of National Quotas

Chapter:
(p.100) 6 Contagion and the Adoption of National Quotas
Source:
Contagious Representation
Author(s):

Frank C. Thames

Margaret S. Williams

Publisher:
NYU Press
DOI:10.18574/nyu/9780814784174.003.0006

Abstract and Keywords

This chapter examines two types of national quotas: compulsory party quotas, which require parties to nominate a certain percentage of female candidates; and reserved-seat quotas, which designate a certain proportion of seats for female legislators. In a statistical analysis of the onset and the incidence of both types of national quotas, the study found that the strongest predicator of national quota onset and incidence is regional diffusion. This is highlighted in an analysis of Ireland, wherein the probability of quota adoption increased steadily as the number of countries in Europe with a quota increased. The evidence for the impact of contagion on national quota adoption is mixed; women's legislative representation has either no impact or a negative impact on the adoption of quotas. However, the study did find evidence that the adoption of voluntary party quotas prompted the adoption of compulsory party quotas, suggesting that, once parties adopt such quotas, resistance to a national-level commitment to gender equity dissipates.

Keywords:   national quotas, compulsory party quotas, reserved-seat quotas, female legislators, regional diffusion, quota adoption, contagion, voluntary party quotas, gender equity

In 2000, the warring factions in Burundi’s civil war signed a peace agreement in Arusha, Tanzania. The Arusha Accord ended a nearly decade-long civil war in which thousands of people died and many more thousands were displaced. Peace led to the writing of a new constitution, which was approved by referendum in February 2005. Article 164 of the new constitution erected a 30 percent reserved-seat quota for women in parliament (International Institute for Democracy and Electoral Assistance and Stockholm University 2006). The quota law substantially increased women’s representation in Burundi, increasing the representation of women in the legislature from 18.4 percent in 2004 to 30.5 percent in 2005 (Inter-Parliamentary Union 2011).

In 1999, Costa Rica adopted a compulsory party quota that required all parties to nominate women for 40 percent of the “electable positions” on party lists (International Institute for Democracy and Electoral Assistance and Stockholm University 2006). Electable positions were those positions in which a party had won a seat in the previous elections. Thus, female candidates were not simply tacked on to the bottom of the party list. The application of the quota appears to have worked. The percentage of women in the legislature in Costa Rica grew from 19.3 percent in 2001 to 35.1 percent in 2002 (Inter-Parliamentary Union 2011).

(p.101) The adoption of these national quotas in both Costa Rica and Burundi raises an important question: given quotas’ powerful positive effect on women’s legislative representation, why have other countries not adopted them?1 In this chapter, we attempt to answer this question by undertaking a statistical analysis of national quota onset and incidence. Our results clearly indicate that contagion significantly impacts the onset and incidence of national quotas. The effect of women in the legislature or the presence of a female executive has different effects on different types of quotas. Moreover, we show strong evidence that regional policy diffusion plays an important role in explaining quota adoption.

Besides contagion, our analysis exposes several other key factors that explain the presence of national quotas. Our results highlight important differences in the factors that lead to the adoption of different types of national quotas. The processes that lead to the implementation of compulsory party quotas and reserved-seat quotas are not identical. This finding is critical because it demonstrates the importance of avoiding the temptation to treat different types of quotas as identical institutions without noting important differences between them. In addition, our results clearly support the view that national quotas are a fast-track approach to gender equity (Dahlerup 2005; Norris 2006). Our analysis shows that national quotas are less likely to be adopted in wealthy, institutionalized democracies with strong records of women’s legislative representation than in countries without these traditions. Instead, they are adopted by countries as a way to overcome the gender gap quickly, without waiting for the traditional processes that have increased women’s representation in more developed democracies such as the Nordic countries.

We start by defining national quotas. Here we explain the differences between our two main types of national quotas ​— ​compulsory party quotas and reserved-seat quotas. We then discuss the effects of these quotas on women’s legislative representation by referencing both our previous analysis and the broader literature. At this point, we explain why we believe that contagion should influence the existence of national quotas. Our statistical analysis follows this discussion. Using a dataset of 159 countries, we model both the onset and the incidence of national quotas.

Defining National Quotas

Beyond voluntary party quotas, there are two other major types of quotas ​— ​ compulsory party quotas and reserved-seat quotas. Compulsory party quotas require parties to nominate a certain percentage of female candidates in (p.102) each election. They may also require other gender-equity solutions such as the placement of female candidates in certain positions on the list. For example, some quotas require that parties alternate the male and female candidates on their lists. Similarly, the size of the quota can vary quite extensively among different countries. Niger, for example, adopted a 10 percent quota in 2002, while France adopted a 50 percent quota in 2000.

Reserved-seat quotas go beyond regulating the process of candidate selection to reserving a certain portion of legislative seats for female legislators. While quotas increase the electoral opportunities for women, reserved-seat quotas affirm women’s participation in the legislative process (Meier 2000). Thus, reserved-seat quotas go much farther to guarantee the presence of women in the legislature. As with compulsory party quotas, the percentage of seats reserved for women can vary. In Bangladesh, 13 percent of seats are reserved for women, while Burundi reserves 40 percent of seats.

Compulsory party and reserved-seat quotas may differ in their commitment to women’s representation. They do, however, have one critical characteristic in common: they represent national attempts to deal with the inequality in women’s representation. Unlike voluntary party quotas that are adopted voluntarily by individual parties, both compulsory party and reserved-seat quotas are adopted through a national political process and impact all actors in the system. Often these quotas are simply laws adopted by the legislature, but in some cases they are enshrined in the national constitution at its formation or by amendment. Compulsory party and reserved-seat quotas can have a much broader impact on women’s representation than voluntary party quotas because of their broader scope. Because of this common national commitment, we refer to both of these quotas as national quotas.

The Effects of National Quotas

Do national quotas ​— ​either reserved-seat quotas or compulsory party quotas ​— ​impact women’s legislative representation? The existing research consistently finds that national quotas do in fact increase women’s legislative representation (Jones 2009; Tripp and Kang 2008). In addition, our empirical analysis in chapters 2 and 3 found that both compulsory seat quotas and reserved-seat quotas had a strong, positive impact of on women’s legislative representation, while reserved-seat quotas were also related to executive representation. The presence of national quotas in fact had a strong, substantive impact on the percentage of women in the legislature.

Other research has found, however, that the impact of compulsory seat (p.103) quotas can vary with other factors, in particular the electoral system. In Indonesia, the failure of the 30 percent compulsory party quota to elect more women is blamed on the semi-open list proportional-representation system that limited the success of female candidates (Siregar 2006). Compulsory party quotas are likely to be more effective in countries that utilize closed-list, high-district-magnitude electoral systems (Htun and Jones 2002; Norris 2004). Given that these systems generally increase women’s representation by lowering the costs of party nominations of women, it is not surprising that they may amplify the effects of compulsory party quotas. While compulsory party quotas are national quotas that may represent a broader commitment to gender equity than, say, voluntary party quotas, it is not always true that the quotas have a significant impact if, for example, they are weakly enforced. Several scholars note that such quotas are often not strongly enforced, allowing parties to avoid reaching the required quota for female candidates (Htun and Jones 2002; Norris 2004; Schwindt-Bayer 2009). Brazil, for instance, maintains a compulsory party quota; however, the law that created the quota contains relatively weak sanctions for parties that do not abide by it (Miguel 2008).

National quotas, by helping to overcome women’s underrepresentation, can certainly impact the public policy process, prodding it to provide more policies supportive of women’s interests. Yet, research does suggest that quotas at the national level can have negative impacts, as well. In their study of Argentina, Franceschet and Piscopo (2008, p. 393) argue that “quota laws complicate both aspects of substantive representation. Quotas generate mandates for female legislators to represent women’s interests, while also reinforcing negative stereotypes about women’s capacities as politicians.” Quotas, therefore, may increase women’s representation but may also undermine the reputation of female legislators who obtain their seats through quotas. That such women won their position through a process different from that followed by male legislators suggests to some observers that they obtained their seats more easily and, perhaps, in an undemocratic fashion.

Supporters of national quotas often cite the need to increase women’s representation in order to stimulate women’s political activity. Since national quotas can increase the level of women’s legislative representation, they may spur greater activity among women more broadly as political aspirants observe female legislators active on the national stage. Zetterberg (2009, p. 717), however, casts doubt on this phenomenon by showing, in a study of Latin American countries, that the presence of quota legislation “appears to not be positively associated with women’s political engagement.” Thus, the use of quotas may not spur women to more political activity.

(p.104) The Logic of National Quota Adoption

When do political actors push to adopt national quotas? Several factors may explain why actors move to enact either compulsory party quotas or reserved-seat quotas. Parties have often been seen as the primary obstacle to women’s legislative representation (Frechette, Maniquet, and Morelli 2008; Lovenduski and Norris 1993). In fact, Bonomi, Brosio, and Tommaso (2006) argue that quotas are unlikely to occur in systems where parties themselves are the main cause of women’s underrepresentation. Yet, in chapter 5, we showed that parties can overcome these obstacles to adopt voluntary party quotas. The logic behind parties’ decisions to support national quotas shows a striking similarity to the logic that underpins parties’ decisions to adopt voluntary party quotas.

Parties may opt to support or oppose national quotas for strategic reasons. Party actors may be uncertain about electoral outcomes, creating openings for those who support quotas to cast them as a tool to increase party vote (Baldez 2004). Thus, the need to compete may push parties to support national quotas as a way to woo voters. In fact, Krook (2005) notes that, in India, parties voiced support for national quotas publicly to curry favor with certain voters while at the same time doing little to implement them. Some parties may find it impossible to adopt a national quota if the quota does not fit well within preexisting patterns of party competition, as was the case in India (Randall 2006).

Resistance to quotas within the party can be overcome, however. In particular, the opposition may fold as the result of strong lobbying by female activists. There is evidence that national quotas have been adopted after successful organizing and lobbying by female activists in a number of countries (Bruhn 2003; Carrio 2005; Craske 1999; Powley 2005; Siregar 2006). In some cases, cross-party coalitions of female activists pushed for the adoption of quotas (Carrio 2005). Just as in the case of voluntary party quotas, the level of women’s political activity strongly impacts the adoption of national quotas.

The level of opposition to national quotas within a political party may also depend on the party’s ideology. Left-wing parties are often more supportive of national quotas, just as they are often more supportive of voluntary party quotas (Bruhn 2003). In France, while the feminist movement concentrated on other issues, women within the Socialist Party pushed the idea of quotas back in the 1970s (Krook 2005).

Opposition to national quotas may also depend upon the electoral system. For male incumbents in a closed-list proportional-representation system, (p.105) a compulsory quota represents a direct threat, given that parties will need to meet the quota by prioritizing female candidates at the expense of male candidates (Frechette, Maniquet, and Morelli 2008). Any type of candidate quota can threaten the reelection of male candidates; therefore, one would expect them to oppose such quotas. Yet, Frechette, Maniquet, and Morelli (2008) argue that male candidates’ fear of quotas is lower in single-member district-majority systems. Using the adoption of the 2000 Parity Law in France, they show that male incumbents benefited from the law because of an electoral bias that favors men. Male incumbents are, therefore, more likely to support a quota if they operate in electoral systems that do not allow women to obtain the full benefit of the quota because of electoral bias.2

In the previous chapter, we found that parties often adopt voluntary party quotas in response to the decisions by other parties to adopt them. Research on national quotas finds a similar process. Countries are more likely to adopt national quotas when they see other countries within their region doing so (Gray 2003; Powley 2005). Relationships between political activists in different countries often speed this policy diffusion process (Bruhn 2003). Powley (2005) points out that activists in the Rwandan Patriotic Front were exposed to gender-equity policies in Uganda and knew of the important role played by women in the African National Congress in South Africa. As other countries in the region adopt quotas, a political leader may fear being considered backward if he does not improve women’s representation (Tripp and Kang 2008). Thus, the adoption of a national quota can be impacted by regional policy diffusion.

The existence of regional patterns in national quota adoption points to another critical factor that may explain quota adoption ​— ​international norms. Existing research argues that the existence of international commitments to gender equality can influence the adoption of national quotas (Krook 2006a, 2006b, 2007a, 2009). The 1995 Platform of Action drafted by the UN Beijing Conference is the best example of this international commitment. The platform was greeted favorably by many political actors and governments (Tripp and Kang 2008). Thus, countries may adopt national quotas in order to live up to international expectations.

As one country in a region adopts such measures, others may follow, not because they agree with the normative commitment to gender equality but because they fear that not to adopt quotas while their neighbors do so would make them seem backward (Tripp and Kang 2008). In fact, feminist activists in the Nordic countries often pointed to the better records of other governments to push their own countries toward greater equality (Freidnevall, (p.106) Dahlerup, and Skjeie 2006). This competition to appear modern may compel countries to adopt quotas to avoid the appearance of falling behind regional neighbors.

Contagion’s Influence on National Quotas

The influence of a female executive or the presence of a substantial number of women in the legislature may be particularly important to the adoption of national quotas. Because these quotas, unlike voluntary quotas, are codified in law, they require a substantial number of political actors committed to increasing women’s representation. This commitment may be more likely where women have made some headway in gaining political office.

The adoption of national quotas may also be spurred by the adoption of voluntary party quotas. In Belgium, a compulsory party quota was adopted only after the adoption of voluntary party quotas by several parties (Meier 2004). As the number of parties with quotas increases, the resistance to such measures may dissipate. In fact, the success of these quotas can increase the pressure on other parties to adopt them. It may also impel activists to push for the next step ​— ​the adoption of a national quota. Research on the causes of national quotas points to causes that differ from those associated with voluntary party quotas. Activists who seek the adoption of voluntary party quotas often cite the low level of women’s representation as a justification for implementing such guidelines. Similarly, activists leverage this issue to spur the adoption of national quotas. Yet, the research on national quotas points out that their adoption often occurs, in part, as the result of broader crises and political concerns. Baldez (2006, pp. 104 – ​105) argues that, “in Latin American countries, support for gender quotas (as well as quotas for youth and sometimes for indigenous peoples) is closely linked to high levels of distrust in the political system.”

Reform and democracy in Latin America allowed this distrust to become institutionalized in quotas that improve women’s legislative representation (Baldez 2004). The deep changes in political institutions ushered in by democratic consolidation can create the necessary space for the endorsement of national-level quotas as a legitimate option to address systemic discrimination.

We can see a similar process at work in Rwanda. In the aftermath of the Rwandan genocide, the new constitution mandated that 30 percent of legislative seats were reserved for women. Feminists groups as well as nongovernmental organizations (NGOs) organized to push for the inclusion of a reserved-seat quota (Powley 2005). Yet, the support of the government and (p.107) of the Rwandan Patriotic Front (RPF) was essential for the inclusion of the quota provision, as well. Powley (2005) noted that the discrimination against the RPF and the Tutsi minority that backed the party made them more supportive of a reserved-seat quota as a legitimate method to address discrimination against women.

National quotas represent a more drastic, less gradual solution to gender inequality than voluntary party quotas. Baldez (2006, p. 103) states that “[t]heir appeal derives in part from the failure of more gradual efforts to change the masculine culture of politics.” These more gradual efforts are most often associated with the experience of the Nordic countries, where factors such as the use of favorable electoral systems and the early implementation of voluntary party quotas led to impressive increases in women’s representation. Dahlerup (2005) argues that many new democracies that have had histories of excluding women from politics are opting for a fast-track approach that involves the use of national quotas. This explains why we see reserved-seat quotas being adopted in places like Rwanda, Uganda, Niger, and Morocco but have no examples of reserved-seat quotas in Western Europe. In fact, advocates of national quotas may face fewer obstacles in new democracies. Dahlerup (2005, p. 148) notes that “history seems to prove that the implementation of a quota system is made easier in a new political system than in an older one, where most seats might be occupied, and consequently a conflict may arise between the interests of new groups and those of the incumbents.”

The Distribution of National Quotas

Using our dataset of democracies between 1945 and 2006, we can examine the patterns of national quota adoption. We present separate patterns for compulsory party quotas and reserved-seat quotas. In our data, we have information concerning twenty-one compulsory party quotas and seven reserved-seat quotas.

Figures 6.1 and 6.2 present the distribution of compulsory party quotas and reserved-seat quotas by year from our sample. Both figures clearly show that national quotas are a recent phenomenon. Our first compulsory party quota appears in 1991; our first reserved-seat quota appears in 1989. The figures also demonstrate that compulsory party quotas are more common than reserved-seat quotas.

When we examine the regional distribution of national quotas, we also see a distinctive pattern. Figure 6.3 presents the percentage of compulsory party quotas by region between 1989 and 2006. The first compulsory quotas (p.108)

Contagion and the Adoption of National Quotas

Fig. 6.1. Compulsory Party Quotas, 1991 – ​2006

Contagion and the Adoption of National Quotas

Fig. 6.2. Reserved-Seat Quotas, 1989 – ​2006

(p.109)
Contagion and the Adoption of National Quotas

Fig. 6.3. Compulsory Party Quotas by Region

were adopted in South and Central America; they spread to Europe in the early 1990s. By 2006, the majority of nations (58 percent) with compulsory quotas were located in either Europe or Latin America. Compulsory quotas spread to Africa and the Middle East in the mid-1990s; by 2006, 33 percent of compulsory quotas were located in these regions. Compulsory quotas did not appear in Asia until 1999. Conspicuous by their absence from the list of countries with compulsory party quotas are North American countries. The regional distribution of compulsory quotas gives credence to the fast-track argument, given that these quotas appeared in regions such as Africa and Latin America during periods in which democratic transitions took place. Yet, we also see evidence that compulsory party quotas exist in Europe in older, established democracies.

Figure 6.4 presents our sample’s regional distribution of reserved-seat quotas between 1989 and 2006. The regional distribution of reserved-seat quotas appears to reflect the fast-track explanation. All of the quotas in our sample are to be found outside Europe and North America. The first reserved-seat quotas were adopted in Asia; they later spread to South and Central America and Africa in the mid-1990s. By 2006, fully 43 percent of reserved-seat quotas were located in Asian countries. (p.110)

Contagion and the Adoption of National Quotas

Fig. 6.4. Reserved-Seat Quotas by Region

Modeling National Quota Adoption

What factors explain the adoption of national quotas? Do we have evidence that contagion matters? To help answer these questions, we use our dataset of democratic countries between 1945 and 2006. Using this dataset, we model the onset and the incidence of both compulsory seat quotas and reserved-seat quotas. Thus, we avoid the assumption that the factors that lead to the onset of one type of quota are the same as those that lead to the onset of the other.

The Onset of National Quotas

To model onset for both types of national quotas, we employ Cox proportional-hazards models. Thus, we model the time to the adoption of both quotas. We tested each of our models to ensure that they meet the proportional-hazards assumption. For independent variables that do not meet this assumption, we include an interaction term between the problematic variable and time (Box-Steffensmeier and Jones 2003). It is possible that the errors in our model will be correlated by country (i.e., across panels). To deal with the potential of heteroscedasticity, we calculate robust standard errors (p.111) clustered on individual countries. We employ the Breslow method to deal with ties.

We measure the effects of contagion with several variables. First, we include the percentage of women in the legislature (Inter-Parliamentary Union 2008a). We also include a variable indicating the presence of a female chief executive (Gleditsch and Chiozza 2009; Worldwide Guide to Women in Leadership 2009). If the presence of women in other democratic institutions influences the adoption of national quotas, then these variables should be positively correlated with the onset of national quotas.

Contagion may operate through the existence of quotas, as well. In our model for compulsory party quotas, we include the number of parties with voluntary party quotas. There is evidence that, in some countries, the existence of voluntary quotas spurs the adoption of compulsory party quotas.

For both the compulsory party quota model and the reserved-seat quota models, we need to control for regional policy diffusion. Existing research suggests that countries may adopt quotas in response to their adoption in other countries. Consequently, we include variables that measure the number of countries in the region with each type of quota. These variables should control for regional quota dissemination.

We control for socioeconomic factors by including both GDP per capita in constant U.S dollars and the rate of female labor force participation (OECD 2009; World Bank 2009). Countries with longer histories of universal suffrage may be more likely to adopt national quotas; therefore, we include years since universal suffrage as a control variable (Inter-Parliamentary Union 2008b). To control for electoral system effects, we include the natural log of district magnitude (Beck et al. 2001; Golder 2005; Johnson and Wallack 2008).

The research on national quotas argues that states may prefer to fast-track women’s representation. In addition, states that have recently transitioned to democracy may also be more likely to adopt them. To control for these processes, we include a variable that measures the number of years since a country’s most recent democratic transition. A transition occurs when a country is democratic at time t but was nondemocratic at t–1.3

We include several regional dummy variables, as well. Unfortunately, several of the regional variables we used in previous models proved to be collinear in our national quota onset models, so we were forced to drop some of these variables from the model. Thus, we can account for some but not all the regional variation that may influence adoption.

Table 6.1 presents the results of Cox proportional-hazards models of the onset of compulsory party quotas (Model 1) and reserved-seat quotas (Model (p.112)

Table 6.1. Onset of Compulsory Party Quotas and Reserved Seat Quotas

Variable

Model 1

Model 2

Compulsory Coeff.

Std. Errors

Reserved Seats Coeff.

Std. Errors

Log of GDP per Capita Constant U.S. $

1.122

0.342

0.630

0.301

Female Labor Force Participation

0.952**

0.018

1.117*

0.073

Years since Suffrage

0.992

0.013

0.996

0.020

Years Democratic

0.911**

0.035

1.010

0.018

Log of District Magnitude

1.087

0.152

0.540**

0.130

Percent Women in Legislature

0.941*

0.032

0.928

0.070

Female Executive in Year

0.617

0.712

16.357

29.532

N Party Quotas

1.433

0.344

N Regional Compulsory Party Quotas

1.973***

0.224

N Reserved Seat Quotas

6.128***

3.855

South and Central America

0.528

0.387

46.277*

102.955

Africa and Middle East

0.887

0.765

0.625

0.877

Asia

0.657

0.635

Years Democratic* Log of Time

1.002**

0.001

1.002**

0.001

Observations

2053

2148

Countries

109

109

X2

86.28***

52.94***

(*) p < 0.10,

(**) p < 0.05,

(***) p < 0.01

2). If we examine Model 1, we find several interesting results. There is no correlation between wealth and the onset of a compulsory party quota. Thus, the traditional relationship between wealth and women’s representation that we found in other areas is not present here. We also find no evidence that quota onset is related to the number of years since democratic suffrage.

The literature on national quotas suggests that they are more likely to be adopted by countries seeking to fast-track gender equity. This approach is often associated with new democracies. We find evidence to support this contention in our model of the onset of compulsory party quotas. An increase of one standard deviation in the years democratic variable decreases the hazard of compulsory party quota adoption by 97 percent. This finding, coupled with the absence of an independent effect on quota adoption of wealth and years since suffrage, supports the notion that the onset of a compulsory party quota is less likely to occur in more established democracies.

We find no evidence that the type of electoral system has an impact on the onset of compulsory party quotas. The results indicate that the hazard of quota adoption increases as district magnitude does; however, the variable is not statistically significant.

In terms of contagion, we find interesting and somewhat contradictory (p.113) results. On the one hand, increasing the number of women in the legislature and the presence of a female executive appear to delay the adoption of compulsory party quotas. Increasing the percentage of women in the legislature by one standard deviation decreases the hazard by just over 40 percent. The presence of a female executive decreases the hazard of compulsory quota adoption by just over 96 percent. Thus, we find little evidence that the presence of women in the executive or the legislature hastens the onset of compulsory party quota adoption. However, this result does lend support to the fast-track argument. The results here suggest that the absence of women’s representation spurs quota adoption. This makes sense, given the logic of the fast-track argument that views quota adoption as a response to weak levels of women’s representation.

We find no evidence that the presence of parties with voluntary quotas affects the onset of compulsory party quotas. The hazard ratio for the number of parties with quotas variable is greater than one, indicating that, as the number of parties with quotas increases, the hazard for compulsory party quotas also increases; however, the variable is not statistically significant. Consequently, we find that the number of parties with voluntary quotas has no independent effect on the onset of a compulsory quota.

Yet, there is strong evidence that regional policy diffusion exists. According to the results of Model 1, a one-country increase in the number of regional countries with a compulsory quota increases the hazard of adoption by 97.3 percent. Countries in regions where other countries have previously adopted a compulsory party quota will, according to our results, adopt such quotas themselves more quickly than countries in regions where few or no countries have adopted compulsory party quotas.

Model 2 in Table 6.1 presents our Cox regression results for reserved-seat quotas. The results demonstrate both similarities and differences between the onset of compulsory party and reserved-seat quotas. We find no evidence that wealth impacts the onset of a reserved-seat quota, similar to what we found in our analysis of voluntary party quotas.

Yet, we do find evidence that women’s political mobilization matters. In Model 2, a one standard deviation increase in female labor force participation increases the hazard of quota adoption by more than 340 percent. Thus, while higher rates of female participation reduce the hazard of compulsory party quota onset, they actually increase the hazard of reserved-seat quota adoption. This may reflect differences in the level of women’s political mobilization necessary to win adoption of the different types of quotas. Compulsory party quotas may be adopted in countries with weak women’s representation as a response to demands by elites or activists for greater equality. A (p.114) reserved-seat quota represents a stronger, more direct commitment to equality than does a compulsory party quota, since reserved-seat quotas require legislative seats not just the opportunity to compete for seats. Consequently, adoption of such quotas may require greater levels of women’s activism.

Neither the years since universal suffrage variable nor the years democratic variable is statistically significant. Thus, we find no evidence that a longer commitment to women’s suffrage predisposes a country to the early onset of a reserved-seat quota. In addition, countries with longer democratic histories are neither more nor less likely to adopt a reserved-seat quota.

We find that other socioeconomic and political factors such as wealth, years since suffrage, and democratic persistence have no impact on the adoption of quotas. The fact that time to quota adoption does not depend on these factors means that those countries with longer democratic histories, longer records of suffrage, or greater female political participation do not adopt quotas earlier than other countries. This supports the fast-track contention that countries with quotas are often those with weak levels of women’s representation.

Electoral systems do matter; however, they do so in an unexpected way. We find that the onset of a reserved-seat quota is sped up by the existence not of a more proportional electoral system but of a less proportional electoral system. An increase of one standard deviation in the natural log of district magnitude decreases the hazard of adopting a reserved-seat quota by 57 percent. This finding could result from the fact that reserved-seat quotas are a response to women’s underrepresentation, which is more likely to be found in less proportional than in more proportional systems.

We do not find evidence of contagion. Both the percentage women in the legislature and the female executive variables are statistically insignificant for the time to adoption of a reserved-seat quota. However, our statistical results provide further evidence of regional policy diffusion. A one-country increase in the number of regional neighbors with a reserved-seat quota increases the hazard of quota adoption by more than 80 percent. This finding supports the view that demonstration effects or competitive pressures push nations toward reserved-seat quotas when neighbors adopt similar quotas.

The Incidence of National Quotas

To model the incidence of national quotas, we create two variables, one indicating the existence of a compulsory party quota in the country in the year and the other indicating the existence of a reserved-seat quota in the country in the year. To model incidence, we use a multilevel mixed-effects logistic (p.115)

Table 6.2. The Incidence of National Quotas

Variables

Model 3

Model 4

Compulsory Party Quota Coeff.

Std. Errors

Reserved Seats Coeff.

Std. Errors

Log of GDP per Capita Constant U.S. $

0.044

0.117

–1.303***

0.328

Female Labor Force Participation

–0.035***

0.009

0.045**

0.021

Years since Suffrage

–0.002

0.006

0.037***

0.014

Years Democratic

0.004

0.004

–0.034*

0.018

Log of District Magnitude

0.286***

0.084

–0.567***

0.193

Percent Women in Legislaturet–1

–0.033**

0.015

–0.006

0.035

Female Executive in Yeart–1

–1.060**

0.538

1.179**

0.568

N Party Quotast–1

0.365***

0.095

N Regional Compulsory Party Quotas

0.419***

0.087

N Reserved Seat Quotast–1

0.857***

0.241

Constant

–19.615

1001.517

–15.506

1545.260

Region (Variance)

0.131

0.134

3.190

3.141

Observations

2168

2168

Countries

109

109

Wald X2

102.42***

47.91***

L.R. X2

5.19**

18.07***

(*) p < 0.10,

(**) p < 0.05,

(***) p < 0.01

regression model. This approach allows us to include both fixed and random effects. Since the countries are nested in regions, we use a model with a random intercept for different regions. The remaining variables are calculated using fixed effects.4

Model 3 presents the results of our mixed-effects model using the compulsory party quota dependent variable. Wealth is not correlated with compulsory quota incidence; wealthier countries are not more likely to adopt these quotas. We also find that the years since suffrage and years democratic variables are statistically insignificant. Consequently, neither an early adoption of universal suffrage nor a long history of democracy independently impacts the likelihood of a compulsory party quota.

Female labor force matters, however, in an unexpected fashion. The percentage female labor force participation variable is statistically significant but negatively correlated with compulsory party quota adoption. A 1 percent increase in female labor force participation decreases the odds of quota adoption by 3.4 percent. Thus, compulsory party quotas are more likely to be found in those countries with weaker levels of female labor force participation.

We do find evidence of electoral system effects. The natural log of district magnitude variable is statistically significant and positively correlated with (p.116) the probability of compulsory quota incidence. A one-unit increase in the log of district magnitude increases the odds of quota incidence by 33 percent.

We find that both women’s representation in the legislature and the female executive variables are statistically significant but negatively correlated with the incidence of compulsory party quotas. A 1 percent increase in women’s legislative representation decreases the odds of compulsory party quota adoption by 3.3 percent. Having a female chief executive decreases the odds of compulsory party quota incidence by 65.4 percent. Thus, we find that contagion in terms of the presence of women in the legislature and the executive branch discourages the adoption of compulsory list quotas.

We do, however, find strong effects based on the number of quota parties and the regional diffusion of compulsory party quotas. Increasing the number of quota parties by one party increases the odds of compulsory party quota incidence by 44 percent. Increasing by one the number of neighboring countries with a quota increases the probability of compulsory party quota incidence by 52 percent.

As expected, we find significant differences among regions based upon our random intercepts in Model 3. Table 6.3 presents values for the random intercepts.5 The values of the random intercepts demonstrate that compulsory party quotas are more likely to be adopted in South and Central America, in Africa, and in the Middle East. Countries in other regions are less likely to use such quotas.

Model 4 in Table 6.2 presents the results of our model of reserved-seat quota incidence. The results are interesting and demonstrate that the factors that account for compulsory party quotas are, on the whole, different from those that explain reserved-seat quotas. The GDP per capita variable is statistically significant and negatively correlated with the incidence of a reserved-seat quota. Thus, wealthier countries are less likely to adopt reserved-seat quotas. This finding is consistent with the fast-track literature. A one-unit increase in the GDP per capita variable decreases the odds of reserved-seat quota incidence by 72.8 percent.

Table 6.3. Values of Regional Random Effects, Model 3

Region

Random Effect

South and Central America

0.484

North America

–0.068

Europe

–0.386

Africa and Middle East

0.074

Asia

–0.070

(p.117) Countries that adopt reserved-seat quotas, on average, have both a longer history of suffrage and higher levels of women’s labor force participation. In our analysis, both the years since suffrage variable and the female labor force participation variable are statistically significant and positively correlated with the incidence of a reserved-seat quota. A 1 percent increase in female labor force participation increases the odds of quota adoption by 4.6 percent. A one-year increase in years since universal suffrage increases the odds of quota adoption by 3.7 percent.

We find further support for the fast-track argument as an explanation for quota adoption in the results of the years democratic variable. The variable is statistically significant and negatively correlated with reserved-seat quota incidence. A one-year increase in the number of years democratic reduces the odds of reserved-seat quota incidence by 3.4 percent. Thus, more established democracies are less likely to adopt reserved-seat quotas.

In Model 4, district magnitude is statistically significant and negatively correlated with the likelihood of a reserved-seat quota ​— ​the opposite of what we found for compulsory quotas. A one-unit increase in the log of district magnitude decreases the odds of a reserved-seat quota incidence by 43.2 percent. Again, this result is consistent with the idea that reserved-seat quotas are adopted in countries where there are obstacles to women’s representation and where groups are interested in finding a quick solution to the problem of women’s legislative representation. Thus, the incidence of a reserved-seat quota is not consistent with traditional electoral system arguments.

Do we have evidence of contagion in our analysis of the incidence of reserved-seat quotas? The results of Model 4 do in fact demonstrate just this. On the one hand, the level of women’s representation has no independent impact on quota incidence. The percentage of women in the legislature variable is negatively correlated; however, it is statistically insignificant. Thus, the existing level of women’s representation in the legislature did not impact quota incidence independent of other factors.

On the other hand, we see strong evidence in the presence of a female executive. The female executive dummy variable is statistically significant and positively correlated with the incidence of quotas. Having a female chief executive increases the incidence of reserved-seat quotas by 225 percent. Thus, the election of a female chief executive positively impacts the likelihood of a reserved-seat quota.

We also find strong evidence of regional policy diffusion. The number of countries in the region with a reserved-seat quota is both statistically significant and positively correlated with quota adoption. A one-country increase in this variable increases the odds of quota incidence by 135 percent. (p.118)

Table 6.4. Values of Regional Random Effects, Model 4

Region

Random Effect

South and Central America

2.639

North America

–0.001

Europe

–1.073

Africa and Middle East

–1.320

Asia

0.206

As we found in Model 3, Model 4 reveals significant differences among regions. Table 6.4 presents values for the random intercepts. We find that countries in Asia and in South and Central America are more likely to adopt reserved-seat quotas.

Discussion of Results

The empirical analysis presented here points to several interesting conclusions about the nature of national quota adoption and incidence. First, it is clear that the processes that drive compulsory party quota adoption are different from those that drive reserved-seat quota adoption. Wealth appears to play no role in the onset or incidence of a compulsory party quota; however, it is clear that poorer countries are more likely than wealthy ones to have reserved-seat quotas. We find differences in terms of the effect of electoral institutions. While countries with more proportional systems are more likely to have a compulsory party quota, they are less likely to feature reserved seat quotas. We also find differences in terms of the effect of two main elements of contagion ​— ​the percentage of women in the legislature and the presence of a female executive. While the percentage of women in the legislature negatively affects both the onset and the incidence of a compulsory party quota, it appears to have no independent impact on the incidence or onset of a reserved-seat quota. In addition, the presence of a female executive has a conflicting impact on the incidence of national quotas: while the presence of a female executive reduces the probability of the incidence of a compulsory party quota, it increases the probability of the incidence of a reserved-seat quota.

Second, the results do support the logic behind the fast-track argument. The fact that wealthy countries with long histories of democracy are not more likely to implement a national quota indicates that such quotas will not typically be found in more established democracies. In addition, the fact that women’s legislative representation is not positively correlated with the adoption of national quotas strongly supports the fast-track argument, (p.119) which contends that countries adopt such quotas to speed up the process of women’s inclusion. The fact that women’s labor force participation is negatively correlated with the incidence of a compulsory party quota further supports this point of view.

Yet, the results for reserved-seat quotas do suggest something a bit different. A higher level of women’s labor force participation increases the hazard of quota adoption, as well as the probability of quota incidence. Moreover, the incidence of reserved-seat quota adoption is more likely in countries with a female executive and a longer history of women’s universal suffrage. Does this suggest that the fast-track logic is incorrect? We argue that the logic is not so much wrong but that it may need more development. The adoption of a reserved-seat quota represents a powerful commitment to women’s legislative equality. Thus, it may represent a desire to increase women’s representation quickly in the face of significant obstacles; however, it may be possible only in countries with a commitment to women’s equality in other areas, as indicated by the longer term of suffrage or an elected female chief executive.

Finally, we have strong evidence that regional policy diffusion matters for both types of quotas. Countries located in regions with greater numbers of national quotas have a higher probability of adopting such quotas than other countries and have a higher probability of adopting a quota system. Thus, we see strong evidence that the adoption of national quotas is sped by regional, competitive pressures. We further examine the effects of these variables in the critical case of Ireland.

Gender Quotas in Ireland

In March 2009, the Irish Labour Party proposed a Gender Parity Bill to establish a compulsory party quota for elections to the Dáil. Parties would lose half of their public funding if they failed to grant women at least 20 percent of their electoral nominations. This threshold would increase to 33 percent in seven years, then 40 percent in another seven years. The bill contained a sunset provision that would remove the quota after twenty-one years. The bill, however, was not adopted. The issue of a gender quota was again raised by Fine Gael leader Enda Kenny one year later, in March 2010, but the idea was rejected by the Fine Gael parliamentary party. The focus on quotas was not new in Ireland. The National Women’s Council of Ireland had recommended a quota previously in 1992, and the Democracy Commission, an independent commission that examined the state of democracy in Ireland, advocated one in 2005. The National Women’s Council of Ireland suggested a 40 percent quota in 2002. The chair of the council, Grainne Healy, argued that, at (p.120)

Contagion and the Adoption of National Quotas

Fig. 6.5. Percentage of Women in the Legislature, Ireland versus Europe

the current pace of improvement, “we will have equal Dáil representation by 2036” (O’Doherty 2002, p. 5).

That political leaders and activists have suggested a compulsory party quota should not come as a surprise, given that the percentage of women in the Irish Dáil is significantly lower than their representation in the governing bodies of other European democracies. Figure 6.5 plots the percentage of women in the lower house of the Irish legislature and the average percentage of women in European legislatures between 1945 and 2006. The figure clearly demonstrates that Ireland remains significantly below the average European country in terms of women’s legislative representation.

What explains Ireland’s poor showing? Some scholars point to Ireland’s political culture. On the basis of a survey of female Irish legislators, Knight et al. (2004, p. 16) contend that, since there is significant agreement among respondents that child care is an important obstacle to women’s legislative careers, “traditional cultural attitudes” that prevent women’s equality remain active. For some researchers, traditional values that reinforce women’s conventional gender roles reflect, in part, strong Catholic traditions in Ireland (Galligan 2005; Galligan and Wilford 1999; Randall and Smyth 1987). Thus, the below-average level of women’s legislative representation in Ireland is in part a result of cultural traditions that create obstacles for women’s equality.

The notion that Irish cultural traditions undermine women’s legislative representation is not universally shared. McElroy and Marsh (2009, 2011) (p.121) analyze candidate-level voting results, candidate survey data, and voter surveys and find no sign of gender discrimination among voters. Their results suggest that voters are just as likely to vote for female candidates as they are to vote for male candidates. In addition, voters are not predisposed against supporting female legislative candidates. Thus, they conclude that, “[i]f there are too few women in Dáil Éireann, it is not down to the voters” (McElroy and Marsh 2011, p. 13).

The Irish electoral system may also explain the country’s poor record on women’s legislative representation. Ireland employs a single-transferable vote (STV) system in multimember constituencies. In this system, voters are asked to rank individual candidates. Often, voters must rank candidates from the same party. Thus, STV creates significant incentives for candidates to cultivate personal votes, and research shows that, while party is not unimportant, voters often cast ballots according to their views of individual candidates (McElroy and Marsh 2009). As discussed previously, systems with strong personal vote incentives have been found to undermine women’s representation in other contexts (Thames and Williams 2010). Yet, Ireland’s system does elect between three and five members per district; thus, it should be more proportional than, for example, a single-member district system. Current research seems split on the extent to which the STV system undermines women’s representation. Only three countries use STV at the national level ​— ​Ireland, Malta, and Australia. Two of these nations have a consistently poor record of women’s representation, while Australia is well above average (Hirczy 1995). Schwindt-Bayer, Malecki, and Crisp (2010) argue that STV has no generalizable effect on women’s representation given the differences between Ireland, Malta, and Australia. Their research finds that Irish female candidates face significant discrimination based, they argue, upon Ireland’s traditional political culture.

Other factors besides the electoral system may be at work. McElroy and Marsh (2009) point to supply-side factors. Potentially, differences in the level of ambitions of male and female candidates may explain why so few Irish women get elected to the Dáil. In addition, the fact that female candidates do not perform better than male candidates may give party leaders less incentive to nominate women (McElroy and Marsh 2009). Parties, especially larger ones that tend to feature lower turnover among candidates, may also simply not encourage female candidates (McElroy and Marsh 2011). Thus, parties and their leaders, not voters, may be the true obstacle to women’s representation.

Given what we know about compulsory party quota adoption, should we expect Ireland to adopt one? Table 6.5 compares the averages of Ireland and (p.122)

Table 6.5. Comparing Ireland to Europe

Variable

Ireland Avg.

European Avg.

Difference

P Value

Log of District Magnitude

1.4

2.2

0.794

0.000

Female Labor Force Participation

41.9

56.9

15.0

0.000

N Quota Parties

0.749

0.758

0.75

0.479

Percent Women in the Legislature

6.2

13.2

7.0

0.000

Note: P values from a difference of means t-test.

Europe for several of the variables from our dataset. If we simply average the percentage of women in the legislature across all years in our dataset for Ireland and Europe, we see a stark difference. Women’s representation in Europe was twice that in Ireland. Two factors we found in chapter 2 that were correlated with women’s representation ​— ​log of district magnitude and female labor force participation ​— ​are, on average, substantially lower among our Irish observations in than in our European observations. Ireland had only slightly fewer parties with gender quotas on average than did the other European countries.

Yet, Table 6.5 also suggests that it is not entirely out of the realm of what is to be expected that Ireland might adopt compulsory party quotas on the basis of the models we presented earlier. We did find that compulsory party quotas were more likely to be adopted in systems with greater district magnitudes. However, we also found that countries with lower percentages both of female labor force participation and of women in the legislature were more likely to adopt compulsory party quotas. Thus, we have some reason to believe that a compulsory party quota would be more likely in Ireland than in other European countries.

To assess further the potential for compulsory party quota adoption, we calculate predicted probabilities of compulsory quota incidence using the results from Model 3 in Table 6.2. First, we calculate the observed predicted probability for each year where we set all variables at the observed levels of Ireland in the year. Then, we calculate a simulated predicted probability where we set the log of the district magnitude variable at the European average in the year, with the rest of the variables set at the observed levels of Ireland in a year. Thus, we are simulating the effect of electoral reform on the probability of compulsory party quota incidence. Finally, for both probabilities, we plot the 95 percent confidence intervals around the predictions.

Figure 6.6 plots the observed and simulated predicted probabilities of compulsory quota incidence in Ireland from 1991 through 2006. The observed (p.123) predicted probability that reflects the estimates from our model using the Irish values suggests an increasing probability of compulsory party quota incidence. Given the fact that such plans have already been actively pursued in the Dáil, the increasing probability should not surprise us. We plot the simulated impact of increasing the log of district magnitude to European levels, as well. As we would expect, increasing the district magnitude does increase the predicted probability for every year. However, the impact is relatively modest and not statistically different from our observed prediction.

What about contagion and regional policy diffusion? The results from our model of compulsory party quota incidence indicate that quota adoption was positively correlated with both the number of quota parties and the number of countries in the region with compulsory quotas. We simulated the impact of increasing levels of both the number of party quotas and the number of regional quotas on compulsory party quota incidence in Ireland. To do this, we first estimated the predicted probability, as we did previously, using observed levels of the variables based on the Irish case. Then, we estimated a simulated probability by doubling the number of quota parties in Ireland for every year. The Labour Party and the Workers Party adopted party quotas in 1991; the Green Party adopted a party quota in 1992.

Contagion and the Adoption of National Quotas

Fig. 6.6. Simulated Effect of Increased District Magnitude

(p.124) Figure 6.7 graphs the observed predicted probability and the simulated predicted probability, which we calculated by doubling the number of quota parties. The simulated predicted probability is significantly greater than that of the observed probability for most of the years for which we calculated probabilities, in particular after 1992. Yet, the differences are not statistically different from the observed values at the 95 percent level. If we calculate 90 percent confidence intervals, the results are statistically significant.6 Thus, there is some evidence, based on our simulation, that contagion through party adoption of voluntary quotas could increase the probability of the incidence of a compulsory party quota in Ireland.

We can explore the impact of regional policy diffusion by simulating the impact of an increase in the number of countries with a compulsory party quota on the incidence of quota adoption in Ireland. We do this by increasing the number of European countries with a compulsory party quota by 20 percent. Figure 6.8 plots the observed predicted probabilities and the simulated ones created by increasing the number of countries in Europe with a compulsory party quota by 20 percent. Not only are the simulated probabilities substantially higher than the observed probabilities, but also the difference between them is statistically significant at the 95 percent level. Thus, if more countries in Europe adopted quotas, we would expect the probability of compulsory party quota incidence in Ireland to increase dramatically. The impact of policy diffusion in terms of the regional spread of national quotas would, according to our empirical model and simulations, increase the chances that Ireland would follow with a national quota.

Conclusion

National gender quotas, whether compulsory party quotas or reserved-seat quotas, are rare. Very few countries make the type of commitment to legislative gender equity that is embodied in a quota that mandates that women have a predetermined percentage of nominations in an election or seats in a legislature. In this chapter, we undertook a statistical analysis of the onset and the incidence of both types of national quotas. The results provide us with critical insights on why some countries utilize quotas and others do not. Our main insight is that compulsory party quotas and reserved-seat quotas are two different animals. The factors that explain one do not necessarily explain the other. They differ in terms of the impact of wealth, labor force participation, years since suffrage, district magnitude, the percentage of women in the legislature, and the impact of a female chief executive. Thus, when we (p.125)

Contagion and the Adoption of National Quotas

Fig. 6.7. Simulated Effect of Increased Number of Quota Parties

Contagion and the Adoption of National Quotas

Fig. 6.8. Simulated Effect of Increased Number of Quota Countries

(p.126) analyze national quotas, we need to keep in mind the important differences between them.

We do find evidence that national quotas represent an alternative, fast-track approach to gender equity. The fact that national quotas are not often found in wealthy, advanced democracies is an indication of this. In addition, the fact that women’s legislative representation has a null or negative impact on quotas further supports the argument that national quotas are more likely to be adopted in countries that want to address quickly gender imbalances.

Perhaps the strongest predictor of national quota onset and incidence is regional diffusion. We find strong evidence that the number of national quotas of either type in a region predicts both the incidence and onset of quotas in individual countries within that region. This finding is clearly highlighted in our analysis of Ireland, where we found that the probability of quota adoption there increased steadily as the number of countries in Europe with a quota increased.

What about contagion? The evidence for the impact of contagion on national quota adoption is mixed. Women’s legislative representation has either no impact or a negative impact on the adoption of quotas. Thus, we find that women’s legislative representation may actually hinder the adoption of quotas. We find that the presence of a female executive discourages compulsory party quota adoption but encourages reserved-seat quota adoption. Thus, there is no general impact of executive contagion on national quotas. We did find evidence that the adoption of voluntary party quotas spurred the adoption of compulsory party quotas, suggesting that, once parties adopt such quotas, resistance to a national-level commitment to gender equity dissipates.

Notes:

(1.) In chapter 2, we showed that such quotas significantly increase the number of women in the legislature.

(p.153) (2.) The idea of a male bias presented by Frechette, Maniquet, and Morelli (2008) contradicts much previous work that does not find such a bias (Darcy and Schramm 1977; Hunter and Denton 1986; Kelley and McAllister 1984).

(3.) We use data from Polity IV Project (2004) and Freedom House (2007) to determine not only democracies but also democratic transitions. Countries that move from below 6 on Polity2 or from “not free” to “partly free” or “free” on Freedom House scores between t–1 and t are coded as having undergone a transition.

(4.) For each model, we present the results of a likelihood ratio (L.R.) test. This test compares the results of our multilevel mixed-effects logistic regression models to a standard logistic regression model without random intercepts. A statistically significant result indicates that the inclusion of random intercepts produces different results. In both of our models, the tests are both highly statistically significant.

(5.) The values are the median Empirical Bayes predictions of the random effects created by our model. The values reflect the median posterior distribution of the random intercepts that include estimates from the model parameters (Rabe-Hesketh and Skrondal 2005).

(6.) We can also obtain statistically significant results by more than doubling the number of quota parties. For the period in our analysis here, doubling the number of quota parties increases it to six. Given that eight parties won seats in the 2002, for example, we believe that an increase to six is a significant but not completely unrealistic number. (p.154)