Determinants of Gender Budgeting Practices: Evidence from Municipal Governments in South Korea

Abstract

In recent decades, many countries have introduced gender budgeting initiatives both at the national and subnational levels. Gender budgeting aims to enhance the gender responsiveness of government spending by integrating gender-disaggregated analyses and gender equality outcome targets into the budgetary process. Although gender budgeting has been practiced for decades, empirical studies on the topic (including on the potential determinants of gender budgeting practices) have been scarce. This paper exploits a unique data set from South Korea. In that country, a gender budgeting system was uniformly introduced to local governments in 2013, but the extent of its use was left to the discretion of municipalities. Drawing from the gender budgeting statements of 66 municipal governments during 2013–2018, I find significant variations in the amount and types of gender budgeting across municipalities. I further find that the amount and types of gender budgeting are significantly influenced by political factors, institutional support, pro-equality climates of civil society, and the socioeconomic and fiscal conditions of the municipalities. Overall, the results suggest that, both inside- and outside-government conditions that are conducive to budgetary reform facilitate the implementation of gender budgeting.

Keywords:

• Gender budgeting
• gender mainstreaming
• local government
• performance management
• South Korea

Disclosure statement

The authors report there are no competing interests to declare.

Notes

1 South Korea provides a powerful setting to empirically examine the antecedents of gender budgeting. Since the mandatory introduction of GB in 2013, the resulting activities by local governments have provided abundant data about the amounts and types of gender budgets. Korea stands out as the one country that requires governments (at multiple levels) to examine and disclose the gender impacts of the budget (Kolovich, 2018).

2 Each line division serves a separate administrative function, such as social welfare, culture, safety, health, etc.

3 Local governments in the United States that have adopted the UN Convention on the Elimination of All Forms of Discrimination against Women (CEDAW) include San Francisco, Cincinnati, Honolulu, Los Angeles, Santa Clara County, and Miami-Dade County (Moser & Korac, 2021).

4 Several studies in other disciplines (e.g., economics, gender studies) discuss the GB determinants but are mostly conceptual (Adeyeye & Akinbami, 2010; Budlender, 2000; Chaney, 2016; Costa et al., 2013; Dey & Dutta, 2014; Kolovich, 2018; O’Hagan, 2015; O’Loughlin & Newton, 2014; Rao & Kelleher, 2005) or survey-based (Otero-Hermida & Lorenzo, 2020), without comprehensive data analyses. Those studies suggest that political factors, institutional supports, a pro-equality climate, and economic conditions are determinants of GB practices (Nolte et al., 2021).

5 In particular, the government report Tracking Equity found that socioeconomic background has an important impact on transition from school to full-time work or study. In 2011, 53.6% (63.1%) of young women (men) living in the most disadvantaged areas were fully engaged in work or study after school. In contrast, 82.1% (83.2%) of the least disadvantaged women (men) were fully engaged in work or study.

6 Mayors and legislators of Korean local governments are elected by residents. For ease of exposition, “mayor” in this paper refers to the head of local governments at any level (e.g., governors of provinces, mayor of cities, heads of districts, etc.).

7 In 2016, performance budgeting was introduced to local governments, including a requirement that governments prepare and disclose a performance budget statement and performance evaluation report.

8 The Conservatism party ruled the central government from 2008 to 2017. Before 2008 and after 2017, the Progressive party took power.

9 The legislators do not take a leading role in the gender-budgeting process but rather assess the adequacy of the gender-equality outcome targets and the gender-disaggregated resource allocation plans during the planning phase. In addition, the legislators eventually pass the budget, including gender budgeting statements.

10 After the fiscal year ends, the Gender Settlement Book is prepared. The book discloses the performance evaluation of outcome targets, the gender-disaggregated beneficiary assessment, and the policy impacts, as well as the causes of underperformance/outperformance and the future directions for the program. The resulting data makes the Korean local governments as the powerful setting for empirical analyses on gender budgeting.

11 When a project is earmarked for GB, administrators assess the differential impacts on women and men, determine gender equality outcome performance targets, and plan the budget allocation. During the execution phase, the divisions in charge work to achieve pre-set gender equality outcome targets. In the audit and evaluation phase, the administrators evaluate the target-achievement rate and the gender-disaggregated policy outcome of each spending project that was earmarked for GB, and analyse the causes of failure or success.

12 Jung (2011) states that the Korean gender budgeting initiative aims (1) that central and local governments pursue the equality of women and men in their budgeting systems, (2) that local governments enhance their accountability and fiscal democracy in a broader sense, and (3) that governments achieve efficient allocation of resources by assessing and quantifying the gender-related impacts of public policy.

13 Among the 226 municipalities in total, I select 74 that belong to metropolitan cities, because these municipalities are more homogeneous in their economic development and fiscal structure than municipalities in provincial areas.

14 Pooling multiple observations from one municipality in a sample may lead the estimation results to be driven by a subsample of municipalities. However, municipalities are equally distributed in the sample in terms of their number of observations. In addition, the model controls for various time-variant variables (e.g., fiscal sustainability, socio-economic conditions, cultural factors, and political factors) that may affect the gender budgeting, mitigating the concern that municipal-specific characteristics of certain municipalities drive the results.

15 Throughout the empirical analyses, none of explanatory variables have VIF values that exceed 10. The average VIF value is 2.70. In addition, heteroscedasticity may not be a problem in this setting, because the dependent variables are transformed to have less skewness (e.g., log transformed or deflated).

16 The percentage of budgets earmarked for GB (among total budgets) has an intuitive appeal in that it shows the budget shares applied with gender perspectives. However, I argue that population is a better deflator, because some of the explanatory variables could simultaneously influence both GB amount and total budgeted expenditures (e.g., fiscal sustainability likely increases both GB amount and total budget; mayors with stronger support from voters can put forth GB projects while increasing total spending at the same time). These confounding effects would cloud the inferences on the relation between several explanatory variables and GB amount.

17 Since Korean governments in general have a two-party system (Conservatism vs. Progressive), it is reasonable to measure the mayors’ political orientation by their party affiliation.

18 As mentioned in the Section 3, the six types of articles are (1) presence of a separate department for gender-equality programs, (2) presence of a citizen council to facilitate public-private cooperation for equality policy, (3) official designation as a women-friendly city, (4) raising a separate fund for gender-equality policy, (5) existence of a formal feedback process for gender-equality policy, and (6) mandatory assessment of the gender equality status in the municipality.

19 The gender equality index is provided by a national research foundation (i.e., Gender Statistics Information System) is the weighted average of gender equality scores in eight dimensions. In particular, the equality score in the “economic participation” dimension measures (1) women’s economic activity participation rate, (2) the gender wage gap, and (3) the proportion of full-time positions among female and male workers. The score in the “women’s empowerment” dimension measures (1) women’s proportion in the local legislature, (2) female directors’ ratio in the local government, (3) female directors’ ratio in companies in the region, and (4) women’s proportion in the citizen council of local government. The score in “education opportunities” measures (1) years of education among women and men and (2) the college entrance rate among women and men. The score in the “welfare” dimension measures (1) the ratio of basic living recipients among women and men and (2) the proportion of national pensioners among women and men. The score in the “health” dimension measures (1) quality of life among women and men, (2) medical check-up rates among women and men, and (3) feelings of stress among women and men. The score in the “security” dimension measures (1) perceptions about public safety among women and men and (2) the proportion of victims of violent crimes among women and men. The score in the “gender-equality in housework” dimension measures (1) time spent on housework, (2) the gender ratio of “third child or more” (i.e., preference for a son above a daughter), (3) satisfaction with family relations among women and men, and (4) proportion of maternity leaves among women and men. The score in the “cultural life and leisure” dimension measures (1) leisure time among women and men, (2) satisfaction with leisure, and (3) internet access among women and men. Each indicator of a particular dimension takes on the value of 100% when women and men exhibit the exact same figures, 0% for the largest disparity. The score of a particular dimension is the average of individual indicators’ values.

20 One thousand Korean won is equivalent to 0.842 U.S. dollars as of September 2, 2020. The standard deviation of GIA/Population (67,620) also exceeds the mean value (62,494). The standard deviation of GT/Population (75,981) is 1.8 times greater than the mean value (42,879). Likewise, the standard deviation and mean of GB/Budget are 0.08 and 0.08, respectively. The standard deviation (mean) of GIA/Budget is 0.04 (0.04). The standard deviation (mean) of GT/Budget is 0.06 (0.03).

21 On average, GIA budgets account for 56% and GT budgets account for 44% of the total gender budgets in the sample.

22 On average, GB/Budget changes from year to year by −0.5% (p = 0.06). Considering that the average GB/Budget is 8%, the change’s magnitude is far from being substantial. Likewise, GIA/Budget on average changes by −0.4% (p = 0.08). GT/Budget on average does not change from year to year (p = 0.82). Combined, each municipality rarely experiences changes of GB amount over time.

Additional information

Notes on contributors

Sun-Moon Jung

Sun-Moon Jung is the assistant professor of accounting at Dongguk University-Seoul, Republic of Korea. She received her Ph.D. and Masters in Accounting from Seoul National University, and a Bachelor of Business Administration and a Bachelor of Fine Arts from Korea University. Her research interest lies in management control systems including performance measurement, incentive contracts, budgeting, and executive compensation.