PARTIES, CABINETS, AND GENDER IN PARLIAMENTARY SYSTEMS

The idea that parliamentary math will be mirrored in cabinet math has a venerable history. The logic behind the idea traces to William Gamson (Reference Gamson1961), whose abstract surmise has been tested empirically in a number of settings throughout the years. Political scientists who have built on Gamson's foundations start with the assumption that political parties are the relevant participants in the bargaining scenarios that surround coalition formation. Gamson was concerned with the payoffs that participants receive as a result of coalition negotiations. He proposed that in any coalition negotiation process, a “participant will expect others to demand from [the] coalition a share of the payoff proportional to the amount of resources which they contribute to [the] coalition” (Gamson Reference Gamson1961, 376). In the context of multi-party negotiations over cabinet composition, each party's payoff is the proportion of total cabinet seats it receives. Resources, in turn, are the proportion of legislative seats that each party contributes to the coalition's total sum of legislative seats. Warwick and Druckman (Reference Warwick and Druckman2001, 627) succinctly summarize Gamson's law: “parties in a governing coalition tend to receive portfolios in one-to-one proportion to the amount of legislative support they contribute to the coalition.”

Table 1, which comes directly from Browne and Franklin's (Reference Browne and Franklin1973, 457) pioneering empirical study of Gamson's law, helps us to visualize Gamson's logic. The table assumes a parliamentary form of government and asks us to imagine a legislature that contains 100 members from three parties—A (45 seats), B (35), and C (20). There are seven possible governments that can come out of this scenario. Three of the seven governments (A, B, C) will be single-party minority governments; three others (AB, AC, BC) will be minimum-winning coalitions; and ABC will be a surplus majority coalition. In real-world scenarios characterized by a different number of parliamentary parties and/or different distributions of legislative seats, there will be other possibilities, e.g., single-party majority governments and/or minority coalition governments.

Table 1. Gamson's Partisan Proportionality Proposition

For the sake of simplicity, Table 1 presents four of the seven possible cabinets in the example sketched above—A; AB; AC; and ABC. In Browne and Franklin's words (Reference Browne and Franklin1973, 457), Gamson's law predicts that “the percentage share of ministries received by a party participating in a governing coalition and the percentage share of that party's coalition seats will be proportional on a one-to-one basis.” Thus, if A is the only party in government, A will control about 100% of the seats in the cabinet. If A forms a coalition with B, then A will control about 56% [45/(45 + 35)] of cabinet seats, and B will control about 44% [35/(45 + 35)] of the seats. And so on.

Browne and Franklin (Reference Browne and Franklin1973) conducted the first empirical and cross-national test of Gamson's logic in the context of coalition cabinets. Drawing on data from thirteen European parliamentary democracies between 1945 and 1969, they discovered a strong correlation (r = 0.926) between the percentage of legislative seats controlled by coalition partners and the percentage of cabinet seats each partner receives from cabinet negotiations. They found that 85.5% of the variance in the percentage of seats a party receives in the cabinet was explained by the percentage of legislative seats the party contributed to the legislative coalition. While subsequent generations of scholars have tested Gamson's logic in different ways (e.g., by taking into account the visibility/prestige/power of particular ministerial portfolios), Browne and Franklin's key finding has stood up well across a variety of contexts (Browne and Frendreis Reference Browne and Frendreis1980; Bucur Reference Bucur2018; Carroll and Cox Reference Carroll and Cox2007; Cutler et al. Reference Cutler, De Marchi, Gallop, Hollenbach, Laver and Orlowski2016; Druckman and Roberts Reference Druckman and Roberts2005; Falcó-Gimeno and Indridason Reference Falcó-Gimeno and Indridason2013; Laver Reference Laver1998; Schofield and Laver Reference Schofield and Laver1985; Warwick and Druckman Reference Warwick and Druckman2001).

While Gamson's law goes a long way toward explaining party seat shares in coalition contexts, we begin by stressing that cabinet formation is about more than party seats shares—it is also about which individuals ultimately receive cabinet seats. In proposing to gender Gamson's law, we are not suggesting that analysts expand their conception of cabinet formation participants. Party leaders remain the principal negotiators. They seek to assure that the distribution of seats accurately reflects the resources (legislative seats) that they bring to the process. We are proposing, rather, that party leaders are additionally concerned with which individuals fill cabinet seats and that they pay attention to gendered parliamentary arithmetic when deciding which individuals should fill those seats.

In this sense, we can view cabinet formation as involving two steps. In the first step, which Gamson discusses, principals decide how many cabinet positions will exist and how (in the case of coalition governments) the total number of seats will be shared. In the second step, which we prioritize, principals decide which specific individuals will take cabinet seats. The two steps are distinctive but involve a common point of reference: in both, characteristics of parties’ legislative delegations guide principals’ decisions.

In focusing on the second step, we draw from the expanding literature on gender in cabinets, which stresses that, regardless of government type (e.g., single-party, minimum-winning coalition, or otherwise), cabinets are key sites of political representation (Annesley, Beckwith, and Franceschet Reference Annesley, Beckwith and Franceschet2019; Claveria Reference Claveria2014; Jacob, Adams, and Scherpereel Reference Jacob, Scherpereel and Adams2014; Krook and O'Brien Reference Krook and O'Brien2012). The literature stresses, more specifically, that when party leaders assemble cabinets, they pay attention to a range of issues beyond whether participating parties receive a fair share of cabinet seats. Annesley, Beckwith, and Franceschet (Reference Annesley, Beckwith and Franceschet2019, 96), for example, stress that the selectors who, after elections, are in charge of constructing cabinets, see those cabinets as collective teams. They approach the project of who will sit in the cabinet with “a checklist of characteristics that must be present in the team. These characteristics apply to the collective cabinet, not to individual cabinet ministers.” The authors suggest that the characteristics that appear on party leaders’ checklists vary across space and time but that such characteristics generally fall within one of three categories: experiential (e.g., the extent to which members of the team possess expertise in specific policy areas), affiliational (e.g., the extent to which members of the team have personal relationships with the head of government and/or other party leaders), and representational (e.g., the extent to which members of the team represent specific parties, genders, geographic divisions, races, religious groups, social classes, and/or other relevant categories).

The notion that gender is a relevant representational characteristic in the context of cabinet formation has been reinforced by developments that have transpired over the course of the last half century (Krook and True Reference Krook and True2012; Paxton and Hughes Reference Paxton and Hughes2015). The 1975 Mexico City World Plan of Action, for example, called on governments to develop systematic strategies for increasing women's participation in elective and appointed offices by 1985. Subsequent decennial UN conferences (Nairobi 1985, Beijing 1995) reinforced and refined the emergent gender-balanced decision-making norm. The Beijing Platform for Action, for example, encouraged all UN states to achieve 30% women in decision-making posts. The last thirty years have also witnessed widespread experimentation, across a diverse range of countries, with quotas designed to increase women's representation in legislative office (Krook Reference Krook2009) and on corporate boards (Hughes, Paxton, and Krook Reference Hughes, Paxton and Krook2017). More recently, a number of nation-states (e.g., Niger) and sub-national bodies (e.g., various Spanish autonomous communities) have established gender quotas at the cabinet level, and sizeable political parties/alliances (e.g., the Left alliance in Poland, 2019) have campaigned on the idea of establishing cabinet quotas. Heads of government in an increasing number of states—Albania (2017), Bolivia (2010), Bulgaria (2016), Canada (2015), Cabo Verde (2014), Chile (2006), Colombia (2018), Costa Rica (2018), Ethiopia (2018), Finland (2007), France (2012), Grenada (2018), Iceland (2009), Liechtenstein (2014), Mexico (2018), Nicaragua (2017), Norway (2013), Rwanda (2018), Seychelles (2018), Slovenia (2017), South Africa (2019), Spain (2004), and Sweden (1994)—have established parity or majority-women cabinets, and their efforts have received widespread international attention.Footnote ¹ It is now rare, anywhere on earth, for a newly established executive leadership team not to be viewed through a gendered lens: “family photos” posted when cabinets take office become powerful symbolic representations of government leaders’ understandings of and sensitivity to gender concerns.

Taken together, these developments have had uneven effects across contexts and actors. In general, though, they have increased the relevance of gender as a representational characteristic—as a factor that matters to the principals who construct cabinet teams. We expect that this context—where party leaders are accustomed to consulting parliamentary delegational arithmetic and where domestic and international audiences are stressing gender as a representational category—affects party leaders’ approaches to step two of the cabinet formation process.

We suggest that parliamentary systems of legislative-executive relations provide the most appropriate setting for determining whether and how intersections between gender and partisanship unfold. Parliamentary systems are defined by a fusion of legislative and executive functions. Cabinets in parliamentary governments can only begin operating once a sufficient number of members of parliament have recognized their legitimacy. In this setting, we expect the party leaders who participate in the cabinet formation process to pay particular attention to the gendered personnel composition of party delegations in the legislature. We expect them to obey a gendered Gamson's logic.

Before specifying what precisely we mean by a gendered Gamson's logic, it is important to note that we are slightly adjusting the focus that is common in the Gamson's law literature—not just by prioritizing step two but also by considering single-party cabinets alongside coalition cabinets. Our basic idea—that the principals who fill cabinet slots attend to the gendered composition of parliamentary delegations—can pertain under conditions of both single-party and coalition government. Regardless of the partisan composition of a cabinet, principals’ appointment decisions will have a gender component. For the reasons discussed above, we expect that all principals will be aware of the gender-dimension of their appointment decisions.

Table 2 transforms the (step one-focused) Gamson's logic presented in Table 1 into a (step-two focused) gendered Gamson's logic. The structural (parliamentary system; 100-seat legislature) and partisan (A has 45 seats; B has 35; C has 20) components of the example scenario remain the same, but Table 2 also considers the number of seats controlled by women and men among each party's parliamentary delegation. In gendering Gamson's law, we propose that the percentage share of cabinet seats received by women participating in a cabinet and the percentage share of legislative seats held by legislators from members of the governing party or parties will be proportional on a (roughly) one-to-one basis.

Table 2. Gamson's Gender Proportionality Proposition

Under the conditions foreseen in Table 2, if party A forms a single-party minority government, roughly 33.3% [15/(15 + 30)] of the cabinet seats will be occupied by women, and approximately 66.7% of the seats will be occupied by men. If parties A and B establish a coalition, women will control roughly 31% {(15 + 10)/[(15 + 30)+(10 + 25)]} of cabinet seats, and men will control approximately 69% of the seats. If A forms a coalition with C—which is this legislature's least gender-balanced party—we expect the resulting cabinet to be comprised of about 25% women {(15 + 1)/[(15 + 30)+(1 + 19)]} and 75% men.

These considerations inform the paper's guiding hypothesis:

Existing literature on gender and cabinets is ambiguous with respect to the subjects on which cabinet selectors focus when they look to the legislature. Do they focus on the gender composition of the legislature as a whole? Or do they focus on the gender composition of the delegation/s of the governing party or parties?

Many studies find a positive association between women's full-chamber representation and women's representation in the cabinet (Barnes and O'Brien Reference Barnes and O'Brien2018; Claveria Reference Claveria2014; Davis Reference Davis1997; Escobar-Lemmon and Taylor-Robinson Reference Escobar-Lemmon and Taylor-Robinson2005; Reynolds Reference Reynolds1999; Siaroff Reference Siaroff2000). In Davis's (Reference Davis1997, 64) words, “whether or not a party appoints women to government seems to depend on the number of women parliamentarians from outside the party as much as—if not more than—the number who are party members.” While the association between percent women legislators and percent women cabinet members is empirically robust, scholars have not provided a clear account of the mechanism behind the association.Footnote ² Why should we expect the individuals making cabinet appointments to base their personnel decisions on the composition of the entire legislative chamber? Annesley (Reference Annesley2015, 619) addresses this question directly in the context of systems that restrict the pool of potential ministers to politicians who hold a legislative mandate. “It makes little sense,” she writes, “to correlate women's ministerial presence with the aggregate number of women in parliament, as it is only women from the governing party or parties, and not the lower house as a whole, who are in the ministerial eligibility pool.” Imagine the case of a victorious coalition whose participating parties have all taken steps to promote parity in their legislative delegations. Imagine, further, that no opposition parties have taken such steps. Under such circumstances, why would principals appoint a cabinet that looks more like parliament as a whole and less like the parliamentary delegations of the participating parties?

The Gamson's law literature suggests that when party leaders consider the cabinet, they focus more narrowly on the characteristics of the parliamentarians that belong to governing parties. Studlar and Moncrief (Reference Studlar and Moncrief1997) provide some support for the “narrow focus” idea in a study of gendered cabinet appointments in the Canadian provinces, where single-party governments are the norm. They discover that the two phenomena—the total percentage of women in provincial legislatures, on one hand, and the percentage of women within governing parties’ parliamentary delegations, on the other—are associated but that the latter, narrower variable is a particularly strong predictor of women's cabinet shares. “The proportion of women in the governing party,” they find, “far overrides all of the other variables [including the percentage of women in the legislature] as an influence on the percentage of women in the cabinet” (Reference Studlar and Moncrief1997, 76).

These considerations suggest that cabinet selectors are particularly attuned to the representational characteristics of the members of parliament who support the government and that the effect of women's seat shares in the set of government-supporting MPs may depend on the type of government in place.

To summarize the causal logic that informs our guiding hypothesis: parliamentary systems fuse legislative and executive powers and prioritize the construction of legislative delegations that support governing cabinets. Party leaders face two logically separable decisions in the cabinet formation process. In the first step, they determine how many cabinet positions will exist and how many cabinet seats each party will control. In the second step, they determine which individuals will occupy the aforementioned seats. Party leaders consider parliamentary arithmetic during both steps. In step one, they consider the relative power of parties within the set of cabinet-supporting legislators. In step two, they consider gendered representational patterns among the set of cabinet-supporting legislators. Both the number of seats partitioned to governing parties and the relative importance of gender as a checklist item depend on parliamentary math. When cabinet-forming principals look to the legislature, they see partisan constellations and gendered partisan constellations. They appoint cabinets whose partisan balances and gender characteristics mirror the partisan balances and gender characteristics of their legislative allies.

DATA, METHOD, AND RESEARCH STRATEGY

The percentage of women in the world's legislatures has held steady or risen every year for the past twenty or so years. The general picture seems to imply that different parties—regardless of ideology—are all on an upward trajectory. At present, though, no organization publishes the kind of comparative party-disaggregated data that would allow us to determine whether this is the case.

Thus, we have gathered party-disaggregated data on MPs and cabinet members in eighteen of the nineteen contemporary OECD countries that Elgie (Reference Elgie2018) classifies as having parliamentary systems. The primary reason we restrict our remit to OECD countries relates to data availability. It is easier to collect the names, party affiliations, and genders of legislators and cabinet members in OECD states over time than it is to collect such data from less-developed countries.Footnote ³ As noted above, we focus on parliamentary systems because such systems epitomize the fusion of legislative and executive powers. Chief executives in semi-presidential and presidential systems may consider gender as a representational characteristic, but the legislative-executive connection is particularly tight in parliamentary systems. Thus, such systems are likely to provide the most direct test of the gendered Gamson's logic sketched above. Our focus on OECD states also allows us to highlight relationships in contexts where liberal democracy is relatively well-established;Footnote ⁴ the gendered Gamson's logic assumes that the distribution of legislative seats reflects a relatively free and fair process of electoral contestation. We restrict our temporal remit to the 1990–2019 period for two reasons. First, as discussed above, the global gender-balanced decision-making norm only crystallized in the mid-1990s. While gender was certainly a relevant category before the 1990s, most principals interpreted it very differently after the United Nations Decade for Women (1975–1985) and the 1995 World Conference on Women in Beijing. Before the UN Decade for Women, the most prevalent gender rule was “ministers should be men.” Second, three of the 18 OECD states in our sample (Estonia, Hungary, Latvia) embarked on democratic transitions at the beginning of the 1990s; since the logic upon which our hypothesis rests assumes a democratic context, it would be inappropriate to include observations from these countries from moments preceding their respective transitions.

We focus exclusively on initial post-election cabinets. We define an initial post-election cabinet as the first cabinet formed after a general election. In calling such cabinets initial, we want to emphasize that our analysis does not account for cabinet reshuffles or cabinets that result from parliamentary acceptance of a new prime minister within a legislative session. We do not want initial to imply an identity between a particular post-election cabinet and the first post-election government formed by a particular prime minister. To clarify this point, consider Theresa May's cabinets in the United Kingdom. May's first cabinet (May I) came to power in the course of a legislative session. May's first initial post-election cabinet (May II) was formed on June 11, 2017, three days after a UK general election. May I was Theresa May's first cabinet, but May II was May's first initial post-election cabinet. We include May II (and exclude May I) in our dataset.

We code the partisan affiliations and genders of all legislators and post-election cabinet ministers since 1990 in Australia, Belgium, Canada, Denmark, Estonia, Germany, Greece, Hungary, Israel, Italy, Latvia, Luxembourg, the Netherlands, New Zealand, Norway, Spain, Sweden, and the United Kingdom. The most recent post-election cabinet included in our dataset is Ratas II in Estonia, which had an investiture date of April 29, 2019. Thus, the dataset includes all post-election cabinets with investiture dates falling between January 1, 1990 and April 29, 2019, inclusive. These parameters generate a dataset that includes 146 post-election cabinets across our eighteen countries.

For cabinet ministers, our data come from the websites of countries’ government offices, press coverage of cabinet investiture, the European Journal of Political Research's Political Data Yearbook (politicaldatayearbook.com) and/or native-language Wikipedia pages. For legislators, our data come from the websites of the respective legislative chambers, personal correspondence with the staffs at legislative chambers, and/or native-language Wikipedia pages.

Our dependent variable is the percentage of cabinet members in a post-election cabinet who are women. To determine the total number of cabinet ministers, we begin with the counts of cabinet positions reported by the Comparative Political Data Set (Armingeon et al. Reference Armingeon, Wenger, Wiedemeier, Isler, Knöpfel, Weisstanner and Engler2019). We include deputy heads of government (e.g., deputy prime minister) and exclude deputy portfolioed ministers (e.g., deputy minister of finance) and ministers of state. When we calculate the total number of cabinet posts, we follow O'Brien et al. (Reference O'Brien, Mendez, Peterson and Shin2015), who exclude the prime minister.Footnote ⁵ We calculate our critical independent variable, Gendered Gamson's percentage, through a three-step process. First, we determine the total number of legislators in the party or parties that participate in the cabinet. Second, we determine the number of women in that set of legislators. Third, we divide the latter number by the former number and multiply the quotient by 100.

For our statistical models, we use linear regression with ordinary least squares and control for institutional variables previously found to be associated with women's cabinet representation. We include controls, specifically, for variables related to the legislature, the cabinet, and a country's international context.