Studies that deal with the evolution of female political representation have provided valuable information regarding the narrowing down of the gender gap (Cheng and Tavits 2011; Beckwith 2007; Sanbonmatsu 2002; Studlar and Matlan, 1996). In line with this, Table 3 shows that 86 out of 119 Spanish regional parliaments (72.3%) have experienced an increase in the percentage of women representatives.

There have also been ten cases (8.4%) in which the percentage of women did not change, and another ten in which the figure actually decreased (in five cases, or half of this group, the fall took place in the 2^nd term). Finally, 13 instances (10.9%) belong to the category ‘within balance’. This embraces those cases that stay within the 40-60% interval and experience fluctuations that are not considered because they do not affect our classification of regional chambers.8

However, the focus on overall trends has diverted the attention from other features affecting the patterns of female representation, such as stability over time. Research on the latter is important for two reasons: first, trends in women’s political recruitment might be relevant or not depending on how stable they are; and second, stability might be associated to different models of incorporation of women into the chambers, and the combination of both is important to understand the evolution of women’s representation.

In order to make up for this derelict topic, we have distinguished between two patterns: consistently increasing up to the attainment of a balanced gender distribution; and unstable or inconsistent, a pattern in which both falls and rises in the percentage of women MPs can be traced.

We consider a trend to be consistently increasing if: (a) as long as the 40% threshold has not been reached, the evolution in the percentage of women MPs is not subject to fluctuations; notice that this definition is not over-demanding insofar as it only requires weakly (as opposed to strictly) increasing trends (i.e. Catalonia qualifies as stable in spite of its halt in the 5^th term); (b) once the 40% threshold has been reached, the number of women MPs in the region does not fall below this mark, and thus stays within the balanced group category. When regional chambers are balanced (40% to 60%), there is little room for the percentage of women MPs to keep growing. Regions that do not fulfil these conditions are therefore unstable.

According to our criteria, nine regions fit the ‘consistently increasing trend’ group and eight the unstable one. Two questions spring to our minds: first, are these patterns of stability related to the models of incorporation detected in the previous section?

Table 4 suggests that there is a close connection between the patterns of time stability and the models of incorporation of women: the forerunners always follow a stable path, while the laggards tend to conform to the unstable pattern (five out of six do so). These results do not allow us to establish a causal relationship between patterns and models but they do point to several lines of research that have not yet been undertaken: (i) to what extent is the relationship between stability and leadership valid for other cases?; (ii) does the relationship detected imply that the patterns of evolution can condition the models of incorporation?, for instance, is stability necessary to be a forerunner in women’s political access to the chambers?; (iii) or does the causal relationship run the other way round?

The second question is: are the regions that have been grouped together in this two-fold scheme similar or dissimilar in socioeconomic and cultural terms? Let us not forget that the literature has identified contextual factors of various sorts (such as income and education, access of women to the labour market, and religion) as conducive to augment (or decrease) women’s political presence (Studlar and Matland 1996). Focusing on two of these factors, GNP and religion, we may tentatively conclude that they do not substantially affect the percentage of female representatives in Spain. First, income (measured as GNP) is meant to be related to women-friendly environments, but the mean GNP of the Spanish regional forerunners (22,025) is below that of the mixed group (23,608). Second, Catholicism has been found to impinge negatively upon women’s political representation (Davidson-Schmich 2006; Cook 1985; Rule 1984). Although all Spanish regions are nominally Catholic and religion cannot therefore work as a discriminant factor, it has not prevented women from acceding in substantial numbers to any particular legislature over time either.

Women are entering the legislatures at varying paces worldwide and the rhythms may be different at the national and sub-national levels. The relative paucity by which women have been elected to the Congress in the United States (Lawless and Pearson 2008: 67) can be compared to the mounting speed at which they were becoming MPs in the 1980s’ and 1990s’ in Canada (Studlar and Matland 1996). In the US, “the first substantial jump in women’s candidacies happened in 1972, in concert with the rise of the women’s movement. The biggest jump in the number of female candidates occurred in 1992’s ‘Year of the Woman’” (Lawless and Pearson 2008: 70). In Canada, “the 1984 election was the turning point: the number of female candidates for all three major parties nearly doubled, and the percentage of female MPs jumped from five to ten” (Studlar and Matland 1996: 53-4). If we move to the sub-national level, women’s representation in the Canadian provinces along the 1975-1994 period has followed a gradual trend towards an increase of female candidates and legislators since the end of the 1980s (Studlar and Matland 1996). The percentage of women in state chambers in the US has grown approximately from six percent at the beginning of the 1960s to 22% at the turn of the century (Sanbonmatsu 2002).

When growth in women’s representation is unexpected and sudden, it may respond to international forces: a United Nations funding programme in Bangladesh might have paved the way for an increase in the percentage of women from 7 to 30%. Similarly, UN training in Vietnam might have contributed to a growth of women MP’s from 18 to 26%. Acts of political voluntarism can equally explain surprising gains in women’s political representation, as it happened when the Peruvian parliament approved gender quotas under the pressure of president Fujimori (Paxton, Hughes and Green 2006). Another example of abrupt change is that of the British Labour Party under the leadership of Tony Blair: the party obtained 418 seats (out of 659) in the House of Commons in 1997 and 101 women became Labour MPs (24.2%). A total of 120 women entered the lower chamber, a record number in British politics (Cowley and Childs 2003). Most legislatures, however, have followed an incremental pattern.

As a general rule, we expect slow and incremental changes to be more likely to sustain over time than fast and unexpected ones (unless gender quotas, which can bring about enduring changes ‘out of the blue’, are considered). In order to tell the former apart from the latter, we have labelled a change as abrupt when it departs by more than two standard deviations from the average change (± 2 * 5.8% +5.0% in our data); conversely, changes will be considered gradual if they lie within (-6.6% to +16.6%). Gradual changes are then divided into moderate (larger than one standard deviation, and hence outside the -0.8% and +10.8% bounds) and slow ones (comprised within this smaller interval).

Applying the former classification, four abrupt increases (in the Canary Islands, Cantabria and Castile La Mancha in the 5^th term, and in the Basque Country in the 8^th), and two abrupt decreases (in the 8^th term in the Canary Islands and the 9^th in the Basque Country) have been identified. Thirteen moderate increases (mostly in the 4^th and 5^th terms, with four cases each) and an equal number of moderate decreases (concentrated in the 2^nd and 8^th terms, with five cases each) have also been detected: two out of these five cases affect forerunner regions (Valencia and Madrid) which had nonetheless surpassed the 40% threshold in the 7^th election and did not fall below this benchmark in the subsequent election. Finally, 74 changes (61.1%) are slow (all of them being increases but one) and the remaining 15 ones (12.4%) show no change at all.9 Laggard regions, previously characterized by time instability in the process whereby their women MPs have been acceding to the regional chambers, have experienced relevant decreases (and no gains in certain cycles) as well. Table 6 helps to appreciate how the largest increases are concentrated in the 5^th term, while the largest decreases have taken place at the beginning or at the end of the period under study.

All in all, 42.1% women (mean) are now sitting (8^th term) in the regional chambers and this has been attained in only 31 years. This result is even more remarkable if put in a comparative perspective: in 1894, the first woman was elected to a state legislature (Colorado) in the United States, and Canada had its first female representative in a provincial chamber (Alberta) in 1917, and yet the mean percentage of sub-national women MPs in these countries is at present 24.5 and 23.4, respectively. In Germany, some Länder held their first regional elections in 1946, 34 years before the first electoral contest at the regional level in Spain, and yet they are still lagging nearly 10 points behind in female representatives (33.0%; own elaboration based on CAWP, PoC, and the web pages of the 16 Länder).

Does the relatively fast feminization of the Spanish regional chambers have to do with legislative gender quotas? In the Spanish regional context, they do not appear to have caused a dramatic change. So far, four regions have introduced quotas in the electoral legislation: the Balearic Islands (Law 6/2002) and Castile-La Mancha (Law 11/2002) have pioneered the move, and the Basque Country (Law 4/2005) and Andalusia (Law 5/2005) have followed suit. The first two regions, which approved 50% of positions for each gender, could not apply their laws in the 2003 regional elections because the PP state government lodged an appeal on grounds of unconstitutionality. The Basque law reserves a minimum of 50% of the party lists for women and allows exclusively-feminine candidacies. In Andalusia, the parity law has adopted the shape of zipper lists (Verge, 2008). Once in power, in 2004, the PSOE withdrew the appeal and the affected pieces of legislation were restored. However, the application of this legislation in the 8^th electoral term has overlapped in time with the putting into practice of the state 2007 equality law. Only one region could use its own regional law before the state law was enacted: the Basque country in 2005. The Basque law went hand in hand with a jump in women’s representation from 33.3% (2001) to 52.0% (2005) but, in the last elections held (2009), the percentage decreased to 44.0%. In Andalusia, where the regional zipper-law and the state parity law were applied in the 8^th term, the percentage moved up from 38.5% (2004) to 45.9% in 2008. The percentage of women MPs remained unchanged between 2007 and 2011 in the Balearic Islands (45.8%), and almost so in Castile-La Mancha (48.9 and 49.0%). Out of the four abrupt increases that have been detected for all regions, only one is related to gender quotas: the 33.3% to 52.0% increase in the Basque chamber.

Let us now turn to the state 2007 Law on Equality between Women and Men of the Socialist government of Zapatero (2004-11), heralded as a milestone in the path towards gender equality. This law mandates that in small districts (around 38%), where less than five candidates are elected, the presence of women and men on party lists should approach a perfect balance, whereas in those whose magnitude equals or is above five candidates, 40%-60% should be the rule. The law was first applied in the 2007 regional elections for the 13 regions which share a common regime, in 2008 in Andalusia, in 2009 in both the Basque Country and Galicia, and one year later in Catalonia.

Table 7 shows that the forerunner regions have been barely affected by the 2007 law. Their growth rate in women MPs after the law is the lowest (3.3%) and their growth differential is actually negative. The law has more important consequences for the laggard (8.0% growth rate) and the mixed groups (7.6%), both of which have positive growth differentials.

We can conclude that the consequences of legislative quotas are contingent on earlier developments: they do not seem to have contributed much to speeding up previously-initiated processes of women’s incorporation into the legislatures but, when regions lag behind the rest, legislation does encourage them to push up the percentage of women MPs.

Some recent studies on the impact of electoral quotas have cast doubts on their efficacy for achieving total gender parity. For example, Martinez and Calvo (2010) found that the PSOE and the PP have applied the 2007 law differently in the elections to Congress and that partisan reasons are therefore relevant when accounting for the insufficient effects of the law. At the municipal level in Catalonia, Verge (2008) concurred that parties have interpreted the law in a discriminatory way for women. Both studies acknowledge that the 2007 law has reduced the gender gap but have simultaneously advocated the introduction of zipper lists and clear mandates regarding the position of the candidates on the party lists in order to attain full parity. Similarly, Ortega and Trujillo (2011) revealed that party quotas and the diffusion of gender recruitment practices across parties are more likely to have had an impact upon the increasing feminization of the national Congress and the Senate than compulsory legislation. Even though these studies focus on the party and our research has the regional chamber as the unit of analysis, we have reached a somehow different conclusion. It is true that the early introduction of party quotas by the left (the PSOE did so in 1988 and the IU a year later) might have reduced the party leaders’ reluctance to incorporate more women into their lists. Different authors have suggested that strategic reasons can explain this: when women cease to be perceived as liabilities and parties that do not field enough women run the risk of being outflanked by their competitors, traditionally-minded political leaders might start promoting women (Sainsbury 2010; Davidson-Schmich 2006). In our case too, the fact that Castile-La Mancha has exhibited a high percentage of women MPs (49%) in 2003, 2007 and 2011 seems to be connected with the zipper lists used by the Socialists rather than with the quota laws. However, the identification of three regional models of incorporation of women MPs has allowed us to obtain a more nuanced conclusion: parity laws have different effects depending on the region to which they are applied. By leading the process in this incorporation, forerunners are little affected by the law, but all the other chambers have seen their percentages in women MPs grow at the same time as the 2007 piece of legislation was enacted. Would this result have taken place if the law had not been in place? Probably not. Further, the law prevents backsliding in the future because it will not allow the share of women to decline below the critical 40% threshold.10

Gender-balanced parliaments are attained when the percentage of MPs for each gender lies between 40 and 60%. In the last electoral term covered (year 2011, for 13 out of the 17 regions), all Spanish regional parliaments are either close to, or have reached, the 40% balanced benchmark. Only seven chambers have failed to do so, but even the region with the lowest percentage of female representatives (the Canary Islands) falls short by less than five percent. Among those (10/17) which have balanced parliaments, five have touched the 45% benchmark and Castile-La Mancha is on the verge of mathematical equality (49%). These figures suggest that there is little room left for further systematic increases in women representation in the Spanish regional parliaments –as a matter of fact, the 2007 Law places a cap of 60%. If this is so, the growth rates in the percentage of women MPs should be expected to fall.

Figure 1 shows that this is indeed the case. While the growth in the relative presence of women was initially very timid (1.3% in the 2^nd term), it increased at a mounting pace, reaching a peak of almost two digits (8.7%) in the 5^th term (precisely the one in which three out of the four abrupt increases have been identified). Although the average proportion of women has kept on growing since then, it has done so at a decreasing rate (down to 6.3% in the 6^th term, 5.2% in the 7^th, and reaching a bare 2.4% in the 8^th term).

Figure 1 also provides some further clues as regards the effects of legislative quotas on the 17 regional chambers. The global growth rates in women MPs after the enactment of the 2007 law was 5.2% in 2007 and 2.4% in 2011. These rates are lower than the ones found before the law came into existence, which means that compulsory gender quotas have not had a full impact because they have been applied when most chambers were on the brink of reaching a gender-balanced status. The law has simply rubberstamped a process that had been initiated previously in certain chambers, although it has contributed to pushing up the percentage of women MPs in laggard and mixed regions. The relevant effect of the law upon these two groups is in line with the importance attributed by different studies to legislative quotas (Martínez and Calvo 2010; Verge 2008; Mateo 2005; Paxton 1997). However, the paradox of the 2007 parity law is that it could limit the potential growth in the number of female representatives beyond the 60% threshold in certain chambers.