Does Local Political Representation Affect the Childcare Coverage Rate in Austrian Municipalities?

Abstract:Historically, the provision of childcare has been a forgotten area of Austrian family policy. During the last decade, much effort was made to catch up with other European countries, but notable differences persist between Austrian regions and municipalities. This article engages the following question: does local political representation affect the public childcare coverage rate in Austrian municipalities? Based on a unique longitudinal data set (2003–2018) containing yearly measures for 1,789 Austrian municipalities, several hierarchical regression models are calculated. The results reveal a positive effect of women and left-party share in the local councils as well as an interaction effect between the two. The findings suggest that women have an impact on the share of children in public childcare only in a right-wing-dominated political sphere.


Introduction
During the last two decades, expanding public childcare became one of the major goals for countries of the European Union.At a 2002 summit in Barcelona, the European Union declared a target of massively increasing the numbers of children in public childcare 1 by 2010 which nearly no Member State achieved.Today, however, more recent figures show that the EU average has nearly met the target. 2Nevertheless, major differences between the EU Member States persist, although close analysis reveals that the greatest differences are within rather than between states (for Germany, see Ma ¨tzke 2019; for Italy, see Brilli, Del Boca, and Pronzato 2016; for Austria, see figures 1 and 2 in this article).
Based on this fact, I analyze in this article the influence of local political representation on childcare expansion.The federal structure of Austria, as well as its well-developed welfare state, combined with the fact that Austrian municipalities have large leeway and responsibility in the area of childcare, makes it an interesting case for answering the research question of this contribution: does local political representation affect the childcare coverage rates in Austrian municipalities?
The existing literature suggests that municipalities with a left-wing mayor and/or a high share of left-wing council members will have higher childcare coverage rates (see, e.g., Mosimann and Giger 2008).Scientific results further find higher rates for municipalities with a female mayor and/or a higher share of female council members (see, e.g., Bratton and Ray 2002).For this article, I attempted to replicate these findings using hierarchical regression analysis.I calculated the models based on a unique data set containing independent variables, measuring the local political representation structure and socioeconomic factors in 1,789 Austrian municipalities between 2003 and 2018.
After the Introduction, a section about the case of Austria, including legal regulations and the context of childcare provision, will give an overview about the specifics of the political system, the status quo, and the development of public childcare in the country.I develop several hypotheses regarding the effect of gender and party balance of the municipal political representation, explain the dataset and method, and present and discuss the results.At the end, I summarize and unite the findings in the Conclusion, discuss several limitations, and highlight the most important results.

The Case of Austria
Austria is an especially interesting case for learning about the effects of local political representation on childcare coverage rates, first, because Austria is one of currently only twenty-five federally organized states in the world, 3 and second, because it is one of an even more limited number of well-developed welfare states, all of which have enacted and enforced different sets of family policies.Third, in contrast with other policy fields, family policy in Austria is primarily governed at the subnational level, namely, by the nine Austrian states and the 2,095 municipalities (Gemeinden), meaning that in Austria, nine different sets of legal regulations are in effect at the same time.These regulations contain provisions regarding group size, staff requirements, equipment, and general provisions for staff and parents.
However, for childcare at the local level, the individual municipalities are the main responsible entities that make decisions about facility operations (Baierl and Kaindl 2011).Two-thirds of all childcare facilities in Austria are fully public institutions managed and funded by the municipalities; the remaining one-third are run by private or religious institutions and cofinanced by municipalities; municipalities pay 80 percent of the total spending for childcare, and the remaining 20 percent is covered by parents' fees and state and federal subsidies (Mitterer, Hochholdinger, and Seisenbacher 2022).The data set I used for this contribution covers all fully funded and co-funded childcare institutions that offer regular all-year childcare by trained staff (for at least 30 weeks per year and at least 4 days and 15 hours per week).In contrast, fully private institutional childcare not co-financed by public money plays a minor role in Austria.Private childcare providers are responsible for only 2 percent of the total supply for children between 3 and 5 years of age 4 (Baierl and Kaindl 2021).
Overall, childcare in Austria is influenced by very different factors.As shown in figure 3, official EU policy outcomes and targets (2002 Barcelona summit, European Pact for Gender and Equality 2011-2020) might put the Austrian state under pressure, and childcare in neighboring countries could influence the national Austrian childcare strategy.The influences are passed on to subnational levels by the provision of financial support, and legal regulations ("15a-agreements") dictate childcare provided at the subnational level; at the municipality micro level, factors such as demographics, gender norms including employment rates, and political affiliation influence the demand and support for public childcare facilities.
In addition to the reasons already mentioned, the country's highly diverse childcare conditions make another important argument for focusing on Austria to analyze national family policy mechanisms.For just one example, preschool coverage across Austrian municipalities varies between 0.8 percent and 100 percent in the dataset for this contribution.Additionally, unlike in other countries, this outcome diversity is not influenced by any historical geographic divide such as East versus West in Germany (e.g.Ma ¨tzke 2019) or the north versus the south in Italy (e.g.Oliver and Matzke 2014).On the contrary, figure 1 reflects that the highly varying percentages of children in Austrian public childcare do not follow specific geographic patterns but rather are an incoherent patchwork of policies.The findings indicate that factors other than geographic location and historical influence are more important for the current situation of institutional childcare in Austria.

Partisanship and Childcare Coverage
The impact of political party on substantive political representation has long been a topic of discussion.At least since Hibbs (1977), many scholars have attempted to prove that party politics have impacts, whereas others have aimed to show that partisanship is not relevant for political outcomes and output (e.g.meta-analysis of empirical studies by Bandau and Ahrens (2020)).A very persistent claim is the positive correlation between a high share of left-wing parties in parliaments, cabinets, or councils and corresponding higher spending on welfare and social policy in general and childcare in particular (for the German La ¨nder, see, e.g., Andronescu and Carnes 2015;Busemeyer and Seitzl 2018; for Swiss municipalities, see, e.g., Mosimann and Giger 2008).
The mechanisms behind this claim are rooted in the historical idea that social democratic parties have been the driving force for welfare state expansion in Western Europe during the post-World War II period (Seeleib-Kaiser, van Dyk, and Roggenkamp 2009).Additionally, public childcare expansion in particular has been a tool for promoting emancipative values and reconciling work and family life, which have become important targets for left-wing parties (Morgan 2013).Today, this relationship persists, bolstered by the fact that parties still do have diverse preferences regarding childcare and female employment: Whereas left-liberal parties prefer (income) equality between men and women, rightconservative parties tolerate (income) inequality (Hieda 2013).
Following Hieda's (2013) two-dimensional classification, the three most important local parties in Austria level can be classified as either social-liberal (SPO ¨, Social Democratic Party of Austria) or right-conservative (O ¨VP, Austrian People's Party; FPO ¨, Freedom Party of Austria).Together with the empirical fact that "a left-liberal government raises its budget for childcare services while a right-conservative government does not" (Hieda 2013: 483), I proposed the following hypotheses regarding the relationship between local representation and the share of children in public childcare: H1a: The average share of children in public childcare is lower in municipalities with a higher share of right-wing party members on the municipal council.
H1b: The average share of children in public childcare is lower in municipalities with a right-wing mayor.
A theory contradicting those hypotheses that parties matter is the new politics of Pierson's (1996Pierson's ( , 2001) ) welfare state approach.Pierson argues that party politics is no longer relevant for social-political reforms and concludes that socioeconomic factors such as demographic change and economic development are more important determinants.This approach emphasizes the importance of including all kinds of socioeconomic variables in statistical models, which according to Pierson should explain most variance.

Gender and Childcare Coverage
In addition to party, gender in political representation is a central variable of interest in this article.The transition from strict gendered roles to more fluid, comprehensive ideas under which men and women are equally responsible for paid work and family care is one of the important preconditions for changes in political representation.Based on this ongoing social transition from single-male-breadwinner to dual-earner families (Lewis 2006(Lewis , 2008)), the impact of female political representation is still quite dynamic; increasing the shares of women in political office is an ongoing process.
Research about the impact of female representation in political institutions on social policy often finds substantive associations between female representation and public benefits, such as more generous welfare state spending (e.g.Bolzendahl 2011;Bolzendahl and Brooks 2007;Bonoli and Reber 2010).However, more recent research on the topic indicates that greater female political representation mainly encourages public service provision, not higher family cash benefits (e.g.Ennser-Jedenastik 2017).
The mechanisms responsible for these findings are rooted in the fact that male and female political representatives on average pursue different aims and goals.According to Funk and Philips (2019), these different aims and goals persist because women are still disproportionally affected by matters such as work-family balance and organizing care work.Although it is evident that "what counts as a women's issue is context dependent and varies across time and space[,] policy areas can be considered gendered if they disproportionately affect (i.e., benefit or harm) women as a group" (Funk and Philips 2019: 23).Similarly Wa ¨ngnerud (2009) argued that "women's interests are connected to how societies are currently constituted.If we look at contemporary societies, we see noteworthy differences between women and men in their everyday life situation" (Wa ¨ngnerud 2009: 53).
In Austria, women spend twice as many hours as men do on unpaid work such as childcare and domestic work (Ghassemi and Kronsteiner-Mann 2009).As a result, in 2020, almost every second woman (49.1 percent) but only 9 percent of men between 25 and 49 years of age worked part time (Statistik Austria 2020).Additionally, only 3.6 percent of men took parental leave in 2020, and that percentage had been decreasing steadily for the previous ten years (Mauerer 2022).In this sense, family policy and especially childcare can be considered gendered policy areas in Austria.
Under critical mass theory, women's representation in politics must exceed certain levels in order to have substantive effects.Furthermore, it is necessary to distinguish between female representation in governments or parliaments and women chancellors or mayors, critical actors who individually have great impact on municipal social policy decisions (e.g.Childs and Krook 2008).Empirical analysis shows that critical actors indeed have much impact at not only the national but also at the subnational levels.
Holman (2014), for example, found that female mayors have a large, positive influence on the likelihood a city that a city will fund social welfare programs in the United States, and that link between women's descriptive and substantive seems to hold at both local and municipal council levels.Bratton and Ray (2002), for example, found a strong positive correlation between the share of women in municipal councils and childcare coverage in Norwegian municipalities for the years 1975-1991.These considerations led to the following two hypotheses for Austria: H2a: Municipalities with a female mayor have higher childcare rates than municipalities with a male mayor.H2b: The higher the share of women on a given Austrian municipal council, the higher the share of children taking advantage of public childcare.
Many scholars question the homogeneity of political interests among women around the world and the resulting homogeneity of substantive representation (e.g.Davidson-Schmich 2011; Palaguta 2020).Others argue that female representation and activism played a key role only in specific contexts (e.g.Misra 2003).These considerations suggest that the effect of gender composition might be mediated by the partisan compositions of local councils.Because most researchers compare small-N countries and it is only possible to estimate interaction effects for large-N samples, results cannot be anticipated.Either the supposed positive effects of female share and left-wing share reinforce each other or the positive effect of female representation is especially strong for more right-wing local councils.Therefore, two hypotheses could be considered: H3a: The positive effect of a higher female share becomes stronger for a higher left-wing share in the local council.
H3b: The positive effect of a higher female share becomes stronger for a higher right-wing share in the local council.

Data and Method
There does not yet exist one dataset that contains full and structured information about regional and local policies in Austria, particularly about the party representation at the municipal level.The unique longitudinal data set I used for this contribution contains yearly data for municipal political representation (Walenta-Bergmann 2023) and childcare in Austria for the years 2003-2018.Furthermore, the dataset contains several socioeconomic indicators such as employment rate, municipal debts, and number of inhabitants.I compiled the data from different sources into a long-format dataset I could analyze using hierarchical regression models.
I obtained most of the fiscal and childcare outcome data from Statistics Austria, the independent, nonprofit institution responsible for managing Austria's official statistics.One of the very important sources that provide the foundation for the dependent variable used in this analysis are the records of the Kindertagesheimstatistik (KTHS), the national public childcare statistics.Additional data sources were the Public Employment Service Austria, the Austrian Chamber of Labor, and the statistics units of the nine Austrian states.Table 1 gives an overview of the descriptive statistics for the variables used in the hierarchical regression models.

Outcome Variables
The main dependent variables of this contribution were the share of children in public childcare and the share of children in full-time public childcare.As already mentioned above, I calculated this variable based on the KTHS data provided by Statistics Austria.These official statistics for public childcare in Austria have reflected the childcare situation since 1972.The last reform, which standardized the data and made some more variables accessible, was carried out in 2003.
One important limitation of using only the KTHS data is that it is impossible to distinguish between children living and being cared for in the same community and children living in one community or municipality and being cared for in another, for instance, because parents' home communities might lack childcare facilities; the result is that the dependent variable could have been distorted.Nevertheless, the average (full-time) share of children in public childcare can be considered the most appropriate measure of the differences in space and time (see Results section).To test the robustness of the results, I used preschool childcare spending as an alternative-dependent variable (see model ( 3)).

Explanatory Variables
To examine the influence of gender and political representation structure in Austrian municipalities on the provision of public childcare, I included multiple measures in the statistical models: the mayor's party, the mayor's gender, the partisan composition of the municipal council, and the share of women in the municipal council.I distinguished partisan composition as left (SPO ¨and Die Gru ¨nen) or right (O ¨VP and FPO ¨) majority, and I dummy coded the mayor's party in the model.Although Austria formally has more than these four parties, the other parties lacked significant municipal-level representation between 2003 and 2018.
Local political representation in Austria is centered around the democratically elected municipal council, consisting of 9-100 council members (depending on municipality size and state law).The municipal council is often called the "parliament" of the municipalities, although this designation is technically inaccurate because municipal councils do not have legislative power.Rather, the municipal boards-that is, the local governments-consist of deputies selected by the municipal councils who in fact cannot make The head of the municipal council and the municipal board is the mayor.He or she is directly elected by the people in six out of nine states and elected by the municipal council members in Vienna, Lower Austria, and Styria.Decisions regarding childcare facilities are made in the municipal council based on majority voting, which suggests that party share in the council is more important for childcare provision than the characteristics of individual mayors or municipal boards.However, I included mayors' characteristics in the model because mayors are responsible for implementing the municipal councils' resolutions as part of overseeing the entire administration of the municipality.

Control Variables
Additional variables can influence the provision of public childcare, and therefore, I included several demographic, (socio)economic, and political control variables in the models.First, I included four demographic variables as controls: the share of people with only compulsory education (nine years), the total population of the municipality, the share of people younger than six years, and the share of people older than sixty-four years.
Second, I included in the model variables measuring the (socio)economic situation in municipalities.Three were the total unemployment rate in the municipality as well as the separate rates for men and women, and the other two variables were debt per capita and spending per capita.These variables are crucial measures because a municipality's economic health determines whether it can provide adequate public childcare facilities.
Third, I included a political variable controlling for the existence of political competition within the municipal councils expecting a positive (e.g.Shen 2023) or negative (e.g.Boulding and Brown 2014) effect of political competition on service provision in the field of childcare.

Method
Regarding my dataset, the methodological combination of time and cross sections is generally a key issue in contemporaneous political research, and researchers have proposed various solutions to the theoretical and econometric problems that have arisen such as autocorrelation, heteroscedasticity, and panel heterogeneity.The methodological approach I used for this research contribution is multilevel modeling, also called hierarchical regression or mixed-effects modeling.
Multilevel modeling is a powerful methodological approach for analyzing large-N samples; it advances classical variance analysis and linear regression, enabling the researcher to analyze the correlations between different variables on more than one level, where additional levels could be for instance, time or geographic unit (state, municipality, etc.).From a technical perspective, multilevel modeling is adequate for clustered or longitudinal data; because my data fulfilled both of these properties, this approach was highly appropriate.However, it has to be mentioned that multilevel modeling poses multiple theoretical and econometric challenges, such as autocorrelation, heteroscedasticity, and panel heterogeneity, for which researchers have proposed a number of different solutions.
The most notable assumptions have been the linearity of relationships, the homogeneity of variance, and the normal distributions of residuals.Because a multilevel regression analysis fits a rectilinear line that explains most of the variance in the data, the data should follow a straight line; visual analysis can test for the assumption that the variance in the residuals is equal across groups (e.g.Garritzmann and Seng 2020;Stadelmann-Steffen and Bu ¨hlmann 2008;Tiemann 2009).In this study, the model-residual-predictor plot did not show any specific patterns, and therefore, the assumption of linearity was not violated; the normal distribution assumption was violated to some extent, particularly for the dependent variable, full-time public childcare.Nevertheless, this study's data did not violate the most important assumptions for multilevel models. 5Despite its methodological challenges, multilevel modeling was the best technical solution for analyzing this time-series municipality data set, and I anticipated that it would produce trustworthy and hopefully interesting results.
I tested two of the study hypotheses proposed earlier using a hierarchical regression model that included four levels of analysis: 6,051 different local government compositions of 1,789 Gemeinden, measured in sixteen different years of observations (2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018) in eight different states.The independent variables, the political representation structure of the municipalities and the control variables were coded with a one-year lag, and the total data set contained 24,875 observations.Because there were incomplete data regarding the gender of the mayors and the regional council incumbents, the final multilevel regression models included only 16,614 and 7,727 observations, respectively.Most of the missing data involved the regional councils of the states of Salzburg, Lower Austria, and Carinthia.

Results
Here, I first give a detailed overview of the existing subnational differences in childcare provision and political representation.Then, I present the results of the hierarchical regression modeling for testing the study hypotheses.

Subnational Variation in Childcare Outcomes and Political Representation
Dependent variable: Public (full-time) childcare ages 0-5 years.To obtain an overview of the public childcare situation in Austria's municipalities, I calculated the yearly share of children in public childcare between 2003 and 2018 for every local community, which was a significant advantage for my study's aim of investigating the development of public childcare over time.My method took into account not only the absolute changes in public childcare supply but also the changes in the number of children in the population, accurately reflecting, for instance, if the number of children in a municipality increased while the public childcare facilities remained the same.For time-series data, the share of children in public childcare provided a more realistic picture of the care situation than absolute figures.
Figure 1 shows notable differences in the percentages of children in public childcare across Austria's municipalities.For the whole period of observations (2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018), the share of children in public childcare was between 0.8 percent and 100 percent, and for the year 2018, the proportion ranged between 15.38 percent and slightly more than 100 percent.The proportion exceeded 100 percent for 458 of the 26,549 cases, I propose because some municipalities are so small, they do not have facilities, or because parents commute to work and place their children in facilities in different municipalities from where they live.In short, that is, in some municipalities, more children are receiving public childcare than live in the municipalities.Figure 1 graphically depicts the wide variations in the provision of public childcare facilities across the state of Austria.
Changing the share of children in public childcare to the share of children in full-time public childcare by excluding all children from the calculation enrolled for less than 8 hours a day gave a different, but equally interesting, pattern: The share of children in full-time public childcare is much greater in the east of the country as well as in and near bigger cities such as Vienna, Graz, and Linz.Overall, however, it is notable that full-time public childcare is much less commonly used than part-time care.
Independent variables: The regional representation structure and socioeconomic situation.The regional representation structure in Austria typically is very different from the representation structure found at the national and state levels, first because historically and still today, the Austrian People's Party (O ¨VP) (center-right) holds much more than 50 percent of all mayor's positions.As seen in figure 4, in 2018, 1,233 of the 2,096 Austrian communities (58.8 percent) were led by a member of the O ¨VP.In the same year, 462 mayors were affiliated with the Social Democratic Party of Austria (SPO ¨); only 37 were members of the Freedom Party of Austria (FPO ¨, populist radical right), and 365 were not officially associated with a political party.
It is important to note that the political independence of those 365 mayors cannot be claimed with certainty, primarily because there is little information regarding the partisanship of local politicians, particularly in Tyrol in western Austria because there is no legal requirement in that state for candidates for office to be affiliated with a specific party; therefore, it might be that the number of truly independent mayors is much lower than 365.For the final analysis, I excluded municipalities in the state of Tyrol because of these data limitations; as a result, the analysis refers to information on only 1,789 of the 2,095 Austrian municipalities.
The regional representation structure in Austria typically is very different from the representation structure found at the national and state levels not only regarding partisanship, but especially also regarding gender balance.Compared to the national level, the average share of women in power is extremely low at the municipal level.Currently, 72 of the 183 deputies in the national parliament, or 39.34 percent, are women, and figure 5 shows clearly that the number of female community leaders is far below 10 percent.Specifically, women led 5.1 percent of Austrian communities in 2010 and only 7.8 percent in 2018.
Right now, there are few systematic and publicly available data regarding the proportions of women in local councils around Austria, and what data are available are only available at the state level.This fact leads to geographic disparities in the availability, quality, and historical accessibility of the data.For the following analysis, I was able to collect data on women's regional council representation in six of Austria's nine states; no data were available for Carinthia, Lower Austria, or Salzburg.I could find data regarding the partisan composition of the regional parliaments for all nine states, but in the state of Tyrol, one can run for office without being affiliated with a specific party.Therefore, I excluded the Tyrolean data from the final models.
Despite the addressed problems of data availability and the timeconsuming process of data collection, it is of great value to look at the lowest level of Austria's federal representation, primarily because of the large variance in the dependent variable (between 0.8 percent and 100 percent).The second important reason is that, at the same time, the units of analysis (municipalities) are very similar in terms of cultural, historical, and institutional background; this allowed me to control for most of the external variance and ensured that the omitted variables bias would be low.A third important advantage of this study compared with other transnational studies on the impacts of political representation structures is the large number of units of analysis (large-N), which expanded the possibilities of statistical methods (e.g.interaction effects).

Analysis of Subnational Variation
Just as the presented descriptive statistics gave an overview about the variables used and their distribution, inferential statistics provided an answer to the research question through hypothesis testing.All multilevel models presented in table 2 include four levels of analysis: time is nested within cabinet terms, cabinet terms are nested within municipalities, and municipalities are nested within states.The intraclass correlation coefficients show that about 10 percent of the variance can be found at the time and state levels each; the remaining 80 percent is due to differences between municipalities and cabinet terms.The models explain between 15 percent and 20 percent of the total variance, most of it at the municipal and cabinet-term levels.
For the time dimension of my time-series cross-sectional data set, these results highlight that cabinet terms are a more decisive time unit for the effect of political representation on childcare provision than years.This is in line with the fact that the political representation is very stable across position terms, five or six years depending on the state.For the geographic dimension of my dataset, in contrast, these results highlight municipalities as more   with the corresponding childcare coverage.The effect ranged between À0.116 and À0.716, indicating that childcare coverage decreased by between 1.16 percent and 7.16 percent when the share of male members in the municipal council increased by 10 percent.
Regarding the two competing hypotheses predicting the interaction effect of right-wing party share and female share in the municipal councils, I could confirm H3b but not H3a.The interaction term was positive (see Supplementary table A5), which supported H3b (the positive effect of a higher female share becomes stronger for higher right-wing share in the local council) but rejected H3a (the positive effect of a higher female share becomes stronger for higher left-wing share in the local council).Indeed, as the marginal effects plot (figure 6) shows, the interaction between the right-wing party and female shares in the local council was highly significantly positive.It appears that the presence of women in municipal government in Austria only has an impact on the shares of children in public childcare when the municipal council is right-wing dominated.
This result is especially interesting because it suggests that the positive effect of female representatives on childcare provision is only relevant in more conservative political spheres.The effect diminishes, and indeed ultimately becomes insignificant, in a more left-wing, progressive political sphere.This result provides evidence for the claim that political interests, and hence substantive representation of women, is not homogeneous.The substantive interpretation of the interaction effect can only be speculated at this point.It might be that women situated in a more conservative, right-wing-dominated political sphere support the expansion of childcare facilities because they perceive childcare to be solely their responsibility.It might also be that their personal political activity strongly depends on the accessibility of childcare facilities because they have no support from their partners.In more progressive, left-wing-dominated councils, on the other hand both, men and women are equally responsible for childcare, resulting in overall higher shares of children in public childcare, but ultimately eliminating the effect of women.The result goes in line with the general transition from conservative strict gendered role models, to progressive, more fluid comprehensive areas of responsibility, with men and women being equally responsible for paid work and family care.It seems that childcare is not a women's issue per se, but it is perceived to be a women's issue only in specific contexts and environments.
Supplementary figure A2 shows the predicted shares of children in public childcare when there were 0 percent, 50 percent, and 100 percent women in the municipal council according to the share of a right-wing party.We see that the share of children in public childcare is highest (70 percent) for municipalities with 100 percent right-wing women members and the lowest (47 percent) for municipalities with 100 percent right-wing members but 0 percent women.That is, for left-wing-dominated municipalities, the share of women does not significantly increase or decrease the share of children in public childcare; the predicted share is about 60 percent for 0, 50 and 100 percent women on a council.

Conclusion
Starting from the notion that public childcare has become increasingly important for EU countries but that subnational differences between regions and municipalities seem to persist, this research answers the question of whether local political representation affects childcare coverage rates in Austrian municipalities.Based on a unique dataset of the local political representation structures and socioeconomic factors in 1,789 Austrian municipalities between 2003 and 2018, I calculated several hierarchical regression models.The results of these models showed that the gender and partisanship balance of the local councils, but not the characteristics of the associated mayor, correlated with the share of children in public childcare.More women on municipal councils were associated with more children in public childcare, and more right-wing party members were associated with fewer children in public childcare.These results indicate that the party representation structure and gender balance is an important predictor of family policy outcomes.
Hence, the first main contribution of this article is that I show the importance of the party representation structure and gender balance for the political outcomes in municipal family policy in Austria.The focus on the municipal level, and the resulting large-N sample, made it possible to take a closer look at the interaction effect of gender and partisanship in political representation.The second main contribution of this article is the resulting finding that female representation in municipal government only affects childcare provision and use in right-wing-dominated political spheres.
Despite these conclusive findings, this article does contain some limitations.The most important empirical limitation is the fact that comprehensive data were not available for the whole time span of interest (2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018).Additionally, as mentioned earlier, I excluded the Tyrolean municipalities from the model because that state's data lacked the party affiliations of the local office holders.Further, the data become more complete in later years.Both of these issues increase the possibility that my data could have been biased and need to be corrected for in further research on the topic.
A methodological limitation of this study is the question of whether one year is an appropriate time span for mapping the impacts of specific municipal council decisions; these decisions could make their impacts clear sooner than a year or much later, and we do not know whether it takes the same amount of time to implement certain decisions in all municipalities.We also do not know the state of quasi-random variations in the time between policy decisions and policy outcomes or if there are structural differences between, for example, larger versus smaller municipalities.
Another important study limitation is the theoretical and empirical difficulties of distinguishing between causes and effects.For instance, the female employment rate could be a consequence of rather than a cause of higher childcare use and not just a variable that positively affects childcare use.Nevertheless, this contribution to the literature is based on a very rich and comprehensive dataset, and unambiguously shows the importance of political party and gender distribution in municipal council makeups for childcare uptake in the federal states and municipalities of Austria.These compelling results should stimulate further research on the mechanisms between descriptive and substantive representation.

Figure 1 .
Figure 1.Percentages of children aged 0-5 years in any public childcare in Austria in 2018.

Figure 2 .
Figure 2. Percentages of children aged 0-5 years in full-time public childcare in 2018.

Figure 3 .
Figure 3. Levels of factors influencing public childcare outcomes in Austria.

Figure 4 .
Figure 4. Overview of the party affiliations of Austrian mayors in 2018.

Figure 5 .
Figure 5. Overview of male and female municipality leaders in 2018.

Figure 6 .
Figure 6.Marginal effects plot for the interaction between male share and right party share in local councils, including the histogram of the share of men in local councils.