Energy Policy Through a Gender Lens: The Impact of Wind Power Feed-In Tariff Policy on Female Employment

Abstract

In light of the United Nations Sustainable Development Goals (SDGs), particularly Goal 5 (Gender Equality) and Goal 13 (Climate Action), integrating gender dimensions into climate adaptation strategies can foster more inclusive and equitable development pathways. While China’s rapid expansion of wind power has been a central component of its climate action strategy, gender considerations in energy policies remain largely overlooked. This study utilizes data from 31 provinces from 2008 to 2021 to assess the impact of wind power Feed-in tariff (FIT) adjustment policy in 2014 on female employment in China. A difference-in-differences (DID) approach is employed using a continuous treatment variable. Due to data availability, we focus on urban unit employment, which reflects mainly formal employment. The results show that FIT subsidy reduction in 2014 substantially reduced female employment at the 1% level, while men were not impacted. The underlying mechanism is validated by observing a significant decline in overall wages and that only female employment in the service sector was notably affected. Wage reduction likely leads to a decrease in demand for service-oriented products, a sector where women dominate, thus amplifying the gendered impact. By providing empirical evidence and theoretical insights, this study highlights the gendered effects of energy policy as part of climate change mitigation efforts and underscores the need to align Goal 5 with Goal 13 through more inclusive and gender-sensitive energy policy design.

1. Introduction

As embodied in Goal 13 of the United Nations Sustainable Development Goals (SDGs), climate change adaptation plays a vital role in reducing exposure to climate risks, alleviating social vulnerabilities, and enhancing human well-being [1]. However, the extent to which climate action can be aligned with other SDGs remains uncertain, given the complex and sometimes conflicting interactions between goals [2]. Structural gender inequality rooted in social norms significantly enhances women’s vulnerability to climate change across sectors and regions [3], highlighting the critical intersection between SDG 13 (Climate Action) and SDG 5 (Gender Equality).

As one of the primary instruments for implementing climate action, renewable energy (RE) policies are often designed without attention to gender dynamics [4], implicitly assuming that the benefits and harms are distributed equally between women and men. Such gender blindness could solidify or even intensify existing inequalities while undermining efforts to coordinate different SDGs. An example lies in the employment impact of RE policies: Initially promoted for their environmental attributes, RE policies have received increasing attention since the late 20th century for their promising potential in job creation, both in political rhetoric [5,6] and academic discussions [7,8,9]. Nevertheless, this does not seem to yield more gender-equal employment patterns. The International Renewable Energy Agency (IRENA) [10] reported that, of the estimated 11.5 million renewable energy jobs worldwide in 2019, only 32% were held by women. Cai et al. [11] built up an extended input–output model to study the distributional employment impacts of RE development in China’s power sector under the policy scenario and found that from 2011 to 2020, 58.6% of new job opportunities were taken by men, compared to 41.4% by women, exacerbating gender imbalances in the labor market. Also, Costa and Veiga [12], who employed an econometric model using data from 278 Portuguese municipalities during 1997–2017, found that local RE investment contributed more to reducing male unemployment than female unemployment.

Previous studies have focused on the gendered distribution of employment opportunities arising from the expansion of renewable energy, confirming that such benefits are not equally shared between men and women. A subset of the literature has offered more in-depth insights into the prevailing patterns of women’s employment in the renewable energy sector [13,14], revealing the structural barriers that systematically limit their participation and progression within the industry [15,16]. Currently, there has been limited scrutiny of the policies themselves [17,18,19]. A top-down gender perspective of energy policy design and decision making is absent in many emerging economies including China, which constitutes a critical gap. Addressing this missing piece can contribute to more nuanced policy design and implementation in the energy sector, both enriching global research and providing context-specific insights from China’s unique experience.

Therefore, in this paper, we examine the gendered employment impacts of China’s most significant incentive RE policy, the Feed-in Tariff (FIT) policy, during its rapid development of wind power. Utilizing panel data from 31 provinces in China during the period of 2008 to 2021, a difference-in-differences (DID) approach is employed with a continuous treatment variable, measured by the cumulative wind power installed capacity in each province in the year preceding the subsidy reduction. This approach is grounded in the assumption that regions with greater wind resources are more likely to have higher cumulative wind power installations, and thus be more exposed to the effects of subsidy reductions. The main results show that the implementation of the FIT subsidy reduction policy in 2014 substantially reduces female employment at the 1% level, while men are not impacted.

The contributions of this study are as follows: (1) From a quantitative perspective, this study provides empirical evidence on the impact of the wind power Feed-in Tariff (FIT) subsidy reduction policy on female employment and explores the underlying mechanisms driving these effects. (2) Through a gender lens, it offers valuable insights into how energy policy may inadvertently produce gendered outcomes, thereby informing more inclusive policy design in the context of China—one of the leading renewable energy developers worldwide, yet still one that is facing persistent challenges in achieving gender equality in the labor market. (3) Finally, this study highlights the complex interactions between SDG 13 (Climate Action) and SDG 5 (Gender Equality). By revealing unintended gendered consequences in the labor market, this research contributes to a more nuanced understanding of how different SDGs may intersect—sometimes in tension rather than synergy—thus calling for more integrated and equity-oriented approaches to sustainable development.

2. Literature Review

2.1. Previous Studies on the Impact of FIT and Theoretical Framework

A feed-in tariff (FIT) is a fixed payment for a defined amount of produced energy, often incorporating various subsidies. It is widely recognized as a useful mechanism for incentivizing RE investments by providing a guaranteed minimum price for RE generators and offering a long-term power purchase agreement with the grid [20]. Unlike general government subsidies, FIT subsidies are often characterized by a gradual reduction over time [21].

Most scholars agree that the FIT policy plays a significant role in promoting investment in renewable energy, thus facilitating its rapid development [22,23,24,25]. The subsequent effects on the labor market could be twofold: In terms of the benefits, RE investment creates many ‘green jobs’ in RE sectors [26,27,28,29,30]. However, subsidies by FIT also increase energy prices for firms and private households, and crowds out the labor force in energy-intensive industries [31,32].

To clarify the underlying causal relationship, we present a simplified theoretical framework that illustrates how FIT subsidies may generate gendered labor market outcomes by separately examining the labor demand and labor supply sides (Figure 1).

In the first place, there are several key terms calling for clarification. One important distinction is between ‘gross’ and ‘net’ jobs. Gross effects include only positive job creation and typically overlook opportunity costs [33], such as the layoffs resulting from the closure of coal-fired power plants [34]. In contrast, net employment offers a more comprehensive metric, as it accounts for jobs that might be displaced elsewhere in the economy. The majority of existing studies suggest that RE policy interventions result in net employment gains. This is supported by several meta-analyses, for instance, Stavropoulos and Burger found that, among 30 studies on the net employment effects of RE expansions, 22 reported positive net job creation, four negative, and four a mixture of positive and negative [35]. Our analysis is grounded in net employment changes, assuming that job creation outweighs job losses, following previous evidence.

Another commonly used categorization, as proposed by Wei et al. [36], distinguishes between ‘direct’, ‘indirect’, and ‘induced’ employment effects. Direct employment is associated with wind power installations, such as the construction and operational phase of a wind farm. Indirect employment refers to jobs generated along the supply chain to accommodate RE expansions such as the manufacturing of wind turbine components, while induced employment refers to the changes in spending resulting from changes in labor income in RE sectors, which extends far beyond the scope of direct and indirect effects.

The theoretical framework starts with labor demand and then moves on to labor supply, although in reality, the two occur simultaneously and interact with each other. On the labor demand side, FIT subsidies stimulate RE investment, resulting in increased labor demand within the RE sectors, primarily through direct and indirect employment. The associated rise in labor income further drives demand for induced employment across the broader economy. Meanwhile, investment in traditional energy-intensive sectors declines, leading to a modest contraction in these sectors, which is outweighed by the job creation of RE investments according to the above assumption. Here, the gender heterogeneity of labor demand is mainly related to gender segregation across sectors. One of its manifestations is that women are concentrated in the service sector [37]. As increased labor income boosts household consumption [38], the resulting induced labor demand predominantly emerges in the service sector [39], thereby having a greater impact on women.

Overall wages rise due to booming labor demand, which feeds back into the labor supply. Neoclassical labor supply theory, where individuals allocate time between work and leisure to maximize utility, suggests that rising wages affect labor supply via two opposing forces: the substitution effect (work more as leisure costs more) and the income effect (work less as income rises) [40]. Most studies indicate that the labor supply elasticity in China remains significantly positive, and that the substitution effect exceeds the income effect [41]. Therefore, higher wage rates increase the labor supply. Compared with men, women exhibit higher labor supply elasticity, as supported by a substantial body of existing research [42,43]. This renders women more sensitive to wage fluctuations and more likely to exit the labor market when wages fall. The impact of electricity prices on labor supply primarily operates through a reduction in real wage levels, shifting the income constraint curve inward with a slight counterbalancing effect.

When the FIT subsidy decreases, the directions of the above steps are reversed. Subsidy reductions suppress RE investment, leading to a decline in both direct and indirect labor demand in RE sectors. This results in lower overall wages and dampens the induced labor demand, particularly in the sector where women are the majority. Additionally, the wage drop further decreases women’s labor supply, ultimately causing a significant decline in their employment levels.

This study aims to examine several hypotheses:

(1)

Baseline

H1a. 

Wind power subsidy reduction policy leads to a significant decline in female employment.

H1b. 

The decline in male employment is either insignificant or less pronounced than that of female employment.

(2)

Mechanism

H2a. 

Wind power subsidy reduction policy leads to a significant decline in the overall wages of workers.

H2b. 

Wind power subsidy reduction policy only leads to a significant decline in female employment in the service sector.

2.2. Wind Power FIT Policy in China

The Chinese government has been actively promoting renewable energy in recent years, particularly wind power technologies. By the end of 2021, China’s cumulative installed wind power capacity has reached 328.5 GW, accounting for 39.2% of the global share, maintaining its position as the world leader for several consecutive years [44]. Among all the policies issued by the Chinese government, the feed-in tariff (FIT) mechanism was considered the most impactful instrument to boost the wind power industry [45].

The development of FIT policy promoting wind power expansion has undergone several stages. Initially, the pricing mechanism for RE was government pricing and bidding [46]. In 2009, the official FIT system started when the National Development and Reform Commission (NDRC) issued a notice for onshore wind projects which divided the country into four zones based on wind energy resources and construction conditions (Figure 2) and set the corresponding FIT rates (

Table 1. FIT rates for onshore wind power projects in 4 zones from 2009 to 2021.

Issue Year	Zone 1	Zone 2	Zone 3	Zone 4
20 July 2009	0.51 1	0.54	0.58	0.61
31 December 2014	0.49	0.52	0.56	0.61
22 December 2015	0.47	0.5	0.54	0.6
26 December 2016	0.4	0.45	0.49	0.57
21 May 2019 (End in 2021)	0.34	0.39	0.43	0.52

¹ The units are CNY/kWh.

). Areas with abundant wind resources had lower FIT rates. Subsidies for wind power projects were covered by the Renewable Energy Development Fund, which was primarily financed through a surcharge on electricity consumption collected from end-users nationwide. The subsidy duration was 20 years from the moment that projects connected to the grid.

On 31 December 2014, wind power subsidies were first reduced, with the adjustments taking effect on 1 January 2015. The FIT rate remained unchanged in Zone 4 at 0.61 CNY/kWh, as fewer wind power plants were built in these areas. Onshore wind projects approved after 2015 would adhere to the updated FIT standard. Conversely, projects approved before 2015 but not yet implemented were allowed to retain the previous FIT rates, provided that they achieved grid connection and power generation before January 2016. Similar subsidy reductions occurred in 2015 and 2016. On 21 May 2019, the “Notice on Improving the Wind Power Feed-in Tariff” announced that the benchmark FITs would be replaced by guide prices. Starting from 1 January 2021, newly approved onshore wind power projects were fully market-based, with no further state subsidies provided.

3. Methodology

3.1. Research Model

The Difference-in-differences (DID) model is a common research method for policy impact evaluation. The advantage of the DID model is to avoid endogenous problems by using a treatment–control comparison over time. A continuous DID is an extension of the traditional DID approach that allows for treatment intensity to vary continuously rather than being binary. It has been employed in a number of influential economic studies, for instance, Nancy Qian [47] used regional variation in tea cultivation as a continuous treatment variable to examine how increases in female income affect the sex ratio. This paper applies this method to analyze the effect of FIT subsidy reduction policy in China on female employment. All provinces were affected by the policy at the same time, but with varying levels of treatment intensity.

We focus on the first FIT subsidy reduction policy in 2014. Since the units of analysis are provincial data and some provinces contain different wind resource zones, using the subsidy reduction amount as a treatment intensity variable is not applicable. Instead, we use the cumulative installed wind power capacity by province in 2014 as a proxy variable. This is because provinces that had a higher wind power capacity before 2014 are mainly located in Zone 1–3 (See Figure 3), causing them be more severely affected by the FIT subsidy reduction policy.

Based on the existing literature, we establish the following empirical model:

erate_i,t = α₁ + γwind_i × post_t + α₂wind_i + α₃post_t + α₄X_i,t + η_i + μ_t + ε_i,t

(1)

where erate_i,t is the female employment rate of province i in year t; wind_i is the cumulative wind power capacity in province i in 2014; post_t is a dummy variable, which equals 1 after 2014 and equals 0 before 2014. The interaction term (wind_i × post_t) is to test whether the policy effect exists. X_i,t is a series of economic and demographic control variables that may affect female employment. Finally, η_i is the provincial fixed effect. μ_t is the year fixed effect and ε_i,t is the error term.

It is necessary to control for time-varying factors that may result in systematic differences between the treatment and control groups for a DID model. The control variables include the per capita GDP growth rate (growth), which reflects the region’s economic development, and government expenditure (exp) and population density (dpop), which may influence both wind investment and employment [48]. Trade openness (ope) and foreign direct investment (fdi) are also included, as they may affect FIT allocation decisions and local labor market conditions, particularly through export-oriented industries [49], sex screening [50], and a more equitable corporate culture [51].

To better isolate the policy effect on female employment, we also control for factors that may affect women’s labor supply and demand, such as female education levels (edu); marital status (mar); and the dependency ratio (dep), which may either reduce labor force participation by increasing women’s caregiving burden [52], or increase it by exerting more economic pressure [53]. Finally, the urbanization rate (urb) is included, given that the scope of this study is limited to urban units. The specific measurements of these variables are provided in Section 3.3.

3.2. Data

This study utilizes a panel dataset covering 31 provincial-level administrative regions in China (excluding Hong Kong, Macao, and Taiwan) from 2008 to 2021. The data are mainly drawn from the China Labor Statistical Yearbook, the China Population and Employment Statistical Yearbook, the China Statistical Yearbook, and the China Electric Power Yearbook. The data on average wages are calculated using 2008 constant prices, using the Consumer Price Index in the China Statistical Yearbook. The data on wind power capacity and wind power generation in 2014 are log-transformed to approximate a normal distribution and reduce the observed skewness. All data analyses in this study are conducted using Stata 16.0.

3.3. Variables

The dependent variable is female employment. Due to data availability, the scope of female employment is restricted to urban areas. Following Wang [54], the variable is calculated as the number of female employees in urban units (this could be subgrouped into specific sectors and industries) to the total urban female labor force at year-end, in order to control for variations in regional female age structure on female employment. The estimated female labor force includes women aged 15 to 64 years. The specific calculation is shown in Equation (2):

erate = num/[pop × ratio × urb × (1 + 1/dep)]

(2)

where erate is the female employment rate; num is the numbers of female employees in urban units; pop refers to the number in the total population; ratio is the proportion of women in the total population; urb is the urbanization rate; dep represents the proportion of the non-working population (0–14 and over 65) to the working-age population (15–64).

We also calculate the male employment rate (mrate) for comparison using a similar process as in Equation (2). In this case, the number of female employees (num) is replaced with that of male employees, and ratio refers to the proportion of men in the total population.

Table 2. Variables, measurements, and date sources.

Variable	Symbol	Measurements	Source
Female employment	erate	See Equation (2)	CLSY, CSY, CPESY 1
Wind power capacity	wind	Cumulative wind power capacity in 2014	CEPY
GDP growth	growth	Real per capita GDP growth rate	CSY
Government expenditure	exp	Government general public budget expenditure/GDP	CSY
Trade openness	ope	Total imports and exports/GDP	CSY
Foreign direct investment	fdi	Actual foreign direct investment/GDP	CSY
Education	edu	Proportion of female employees with a college degree or above in urban units	CLSY
Marriage	mar	Proportion of married women to the female population over 15 years	CPESY
Dependency	dep	Proportion of children (0–14) and elderly (over 65) to the working-age population (15–64)	CPESY
Urbanization	urb	Proportion of urban population	CPESY
Population density	dpop	The population per unit area	National Bureau of Statistics, Provincial Statistical Yearbooks
The proportion of female employment	erate1	Proportion of women employees in urban units	CLSY
Wind power generation	wind1	Wind power generation in 2014	CEPY
Wage	wage	Average annual salary of employees in urban units	CLSY

¹ CLSY stands for the China Labor Statistical Yearbook; CSY stands for the China Statistical Yearbook; CPESY stands for the China Population and Employment Statistical Yearbook; CYPY stands for the China Electric Power Yearbook.

summarizes the measurements and sources of the main variables used in this study. The descriptive statistics are shown in

Table 3. Descriptive statistics.

Variables	Mean	Std. Dev.	Min.	Max.	N
erate	0.2232	0.0624	0.1262	0.4713	434
wind (104 kW)	311.87	439.13	1	2100	434
mrate	0.3522	0.0843	0.1571	0.7359	434
erate in 3 sectors
agricultural	0.0056	0.0120	0.0000	0.0859	434
industrial	0.0667	0.0343	0.0201	0.2022	434
sevice	0.1509	0.0605	0.0812	0.4143	434
erate within the service sector
transportation	0.0081	0.0034	0.0010	0.0222	434
IT	0.0048	0.0062	0.0012	0.0545	434
trade	0.0139	0.0106	0.0048	0.1087	434
catering	0.0056	0.0046	0.0009	0.0248	434
finance	0.0123	0.0059	0.0051	0.0513	434
real_estate	0.0052	0.0043	0.0000	0.0257	434
rental_service	0.0057	0.0081	0.0000	0.0521	434
research	0.0050	0.0052	0.0015	0.0368	434
environment	0.0042	0.0016	0.0019	0.0138	434
resident_service	0.0013	0.0016	0.0000	0.0090	434
education	0.0378	0.0111	0.0198	0.0815	434
sanitation	0.0210	0.0129	0.0115	0.1944	434
entertainment	0.0029	0.0022	0.0012	0.0148	434
social_organization	0.0246	0.0213	0.0072	0.1604	434
growth	0.0827	0.0330	−0.0360	0.1880	434
exp	0.2711	0.2116	0.0870	1.3792	434
ope	0.2618	0.3079	0.0076	1.7991	434
fdi	0.0207	0.0198	0.0001	0.1079	434
edu (female)	0.1819	0.1176	0.0030	0.6840	434
edu (male)	0.1722	0.1028	0.0020	0.8600	434
mar	0.7348	0.0361	0.5687	0.8020	434
dep	0.3761	0.0728	0.1927	0.5779	434
urb	0.5661	0.1393	0.2261	0.8960	434
dpop (103 ppl/km2)	307.91	448.00	2.19	2748.93	434
erate1	0.3706	0.0296	0.3114	0.4465	434
wind1 (108 kWh)	73.59	117.94	0	967	434
wage (104 CNY)	5.09	1.99	2.06	14.28	434

.

4. Results and Discussion

4.1. Baseline Results

Table 4. Baseline results (dependent variable: female employment vs male employment).

	Female		Male
	(1)	(2)	(3)	(4)
wind × post	−0.00818 *** (0.0024)	−0.00688 *** (0.0021)	−0.00719 ** (0.0035)	−0.00590 (0.0037)
Controls	NO	YES	NO	YES
Year FE	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
N	434	434	434	434
Adj. R2	0.2622	0.4219	0.3286	0.3815

Notes: **, and *** represent significance levels of 5%, and 1%, respectively. Robust standard errors clustered by province; SE shown in parentheses. The control variable edu in the regression equation of male employment refers to the proportion of male employees with a college degree or above in urban units. Other control variables remain the same as in female employees.

reports the OLS estimation results of the DID model, presenting specifications both with and without control variables. For female employment, the coefficient of the interaction term (wind × post) is significantly negative at 1%, indicating that the implementation of the FIT subsidy reduction policy in 2014 substantially reduces female employment. After incorporating a series of control variables that might affect the estimation, the significance of the interactive term remains robust, though the magnitude of the coefficient declines (See Column (2)). In contrast, male employment appears largely unaffected by the subsidy policy in 2014. While Column (3) initially suggests a statistically significant negative average treatment effect at the 5% level, this effect vanishes once time-varying confounders are accounted for, indicating potential selection bias.

Overall, the baseline results indicate that the subsidy reduction policy exerts a detrimental impact on female employment, whereas male employment remains stable. This supports the hypotheses H1a and H1b proposed in Section 2.2.

4.2. Robustness Tests

4.2.1. Parallel Trend Test

A key assumption of the DID approach is the “parallel trend hypothesis”, which requires that the treatment and control groups follow similar time trends prior to the implementation of the policy. In this study, instead of a binary indicator to define the treatment and control groups, we employ a continuous variable to capture treatment intensity. Following the method proposed by Li et al. [55], we construct 14 time-relative dummy variables (from post⁻⁶ to post⁷) spanning the period from 2008 and 2021, with 2015 defined as post¹. Each post variable represents a scenario where a specific year is considered as the policy intervention point; for instance, if 2012 is treated as the policy intervention year, the corresponding post variable takes the value of 0 prior to 2012 and 1 thereafter. These post variables then interact with the contemporaneous wind power capacity (wind × post*) and are included in Equation (1) for regression. By plotting the estimated coefficients of these interaction terms, we can assess whether the parallel trend assumption holds true.

Figure 4 illustrates the dynamic impact of the FIT adjustment policy on female employment. The coefficients show that prior to 2015, no statistically significant policy treatment effect is observed at the 5% level. It is only after 2015 that the subsidy reduction effect becomes significantly negative, particularly in 2019. These findings support the validity of the parallel trend hypothesis, thereby confirming the robustness of the baseline results.

4.2.2. Replacing the Measurements of Key Variables

We also conduct a robustness check by replacing the measurements of two key variables: the dependent variable and the continuous treatment proxy variable. In the baseline analysis, the dependent variable, female employment, is measured as the proportion of women in the labor market relative to the total estimated female labor force in each province. In the robustness test, this is substituted with the proportion of women in the total workforce (erate1), without accounting for the age structure of each province. The treatment intensity proxy is replaced with total wind power generation (wind1) instead of wind power capacity in 2014. Columns (2) and (3) of

Table 5. Robustness test results (female employment as the dependent variable).

	(1) Benchmark	(2) Erate1	(3) Wind1	(4) 2008–2019
wind × post	−0.00688 *** (0.0021)	−0.00395 ** (0.0015)	−0.00810 *** (0.0023)	−0.00535 ** (0.0023)
Controls	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
N	434	434	434	372
Adj. R2	0.4219	0.6636	0.4300	0.4472

Notes: **, and *** represent significance levels of 5%, and 1%, respectively. Robust standard errors clustered by province; SE shown in parentheses. The first column corresponds to Column (2) of Table 4.

present the main results. Compared to the baseline in Column (1), the coefficients of the interaction terms exhibit consistent significance and direction, proving that the baseline results are robust.

4.2.3. Shorten the Time Window

The baseline regression employs a sample spanning from 2008 to 2021. Given that the COVID-19 pandemic that emerged in 2020 may have adversely impacted employment [56] through widespread lockdowns and economic uncertainty, we exclude the years 2020 and 2021 from the sample as part of the robustness analysis. Column (4) of Table 5 reports the results using a shortened time window, and the estimated policy effect remains significantly negative.

4.3. Heterogeneity Results

4.3.1. Gender Norms

Considering that the underlying causes of occupational gender segregation are closely tied to gender norms, a topic extensively explored in the sociological literature, the heterogeneity analysis in this study first seeks to identify a suitable proxy for provincial-level gender norms. In regions where gender norms are more conservative, occupational segregation tends to be more pronounced, and women are more likely to withdraw from the labor market. Consequently, the policy impact on female employment might be more significant in these regions.

Some studies utilize data from the China General Social Survey (CGSS) to construct a proxy for gender norms at the provincial level by aggregating respondents’ scores on five statements reflecting gender-normative attitudes (e.g., “Men are inherently more competent than women” or “Women should be laid off first during economic downturns”) [57]. However, the CGSS survey covers only 21 provinces, primarily in the eastern region where the wind power capacity is relatively limited, which we consider inadequate for this study.

As an alternative, we adopt a widely recognized proxy: the Confucian cultural tradition [58], which embodies conservative gender ideologies, including the moral doctrines of the “Three Obediences and Four Virtues” that have profoundly influence China throughout history. We measure this variable by counting the number of Confucian schools, academies, and temples in each province. The data are drawn from historical sources such as the Comprehensive Annals of the Qing Dynasty and local gazetteers from the Ming and Qing Dynasties. To ensure comparability across provinces, we normalize this variable by dividing it by the provincial population density.

To examine potential heterogeneity in policy effects, we interact the Confucian cultural traditional proxy (gender_norm) with the existing interaction term in the baseline regression model (Equation (1)), forming a triple interaction term wind × post × gender_norm. Column (1) of

Table 6. Heterogeneous results across 3 sectors (female employment as the dependent variable).

	Gender Norms	Economic Area
	(1)	(2) East	(3) Central	(4) West
wind × post	−0.00460 ** (0.0021)	0.00330 (0.0061)	−0.0142 *** (0.0035)	−0.00555 * (0.0026)
wind × post × gender_norm	−0.00677 (0.0084)
Controls	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
N	434	154	112	168
Adj. R2	0.4666	0.5349	0.3797	0.4160

Notes: *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively. Robust standard errors clustered by province; SE shown in parentheses.

reports the results. The coefficient on the triple interaction is statistically insignificant, suggesting that the impact of the subsidy reduction policy does not differ across regions with varying levels of gender norm conservatism.

It should be acknowledged that the Confucian cultural tradition proxy may still not reflect the intricate nature of gender norms. As an inherently unobservable and multifaceted construct, gender norms encompass a broad spectrum of ideological, cultural, and structural dimensions, the interpretation and manifestation of which vary considerably across contexts. These norms may relate to whether women should be economically independent, or even more fundamental issues, such as the legitimacy of prevailing gendered divisions of labor. Accordingly, the findings here should be viewed as indicative rather than definitive and must not be overgeneralized beyond the scope of this study.

4.3.2. Economic Area

The tripartite division of China into eastern, central, and western regions is widely adopted in empirical research to capture structural disparities in economic development, industrial composition, and demographics. We follow the common classification, where the eastern region includes Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan; the central region includes Shanxi, Jilin, Heilongjiang, Henan, Hubei, Hunan, Anhui, and Jiangxi; and the western region includes Inner Mongolia, Chongqing, Sichuan, Guangxi, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, and Tibet. We group the sample provinces based on their regions and re-estimate Equation (1) to examine the coefficients of the original interaction term wind × post.

Columns (2) to (4) of Table 6 present the results, where significant effects of the policy intervention are observed only in the central and western regions, with the central region showing the most pronounced impact, significant at the 1% level.

4.4. Mechanism and Discussion

4.4.1. Reducing Overall Wages

Based on the mechanism analysis in Section 2.1, we primarily test two hypotheses in this part, one of which examines whether the subsidy reduction policy leads to a decrease in overall wage levels. We replace the dependent variable with the annual average wage in urban units (converted to constant prices in 2008) and perform the regression in Equation (1).

Table 7. Mechanism test (average wage as the dependent variable).

	(1)	(2)	(3)	(4)
wind × post	−0.222 *** (0.0673)	−0.202 ** (0.0812)	−0.181 *** (0.0616)	−0.181 *** (0.0616)
Controls	NO	YES	YES	YES
Year FE	YES	NO	YES	YES
Province FE	YES	YES	NO	YES
N	434	434	434	434
Adj. R2	0.9418	0.9039	0.9591	0.9591

Notes: **, and *** represent significance levels of 5%, and 1%, respectively. Robust standard errors clustered by province; SE shown in parentheses.

presents the results, where the overall wages decline significantly, regardless of whether different fixed effects are controlled for or time-varying variables are included, thus validating the H2a hypothesis and suggesting a potential wage transmission mechanism.

4.4.2. Reducing the Induced Labor Demand in the Service Sector

The second hypothesis explores whether employment losses are concentrated solely in the service sector, a sector predominantly led by women. We decompose the female employment rate (erate) calculated in Equation (2) into the agricultural, industrial, and service sectors, which are the three major economic sectors in China. This is measured as the number of employed women in each sector divided by the total estimated female labor force. Each sector is treated as a separate dependent variable, and regressions are performed according to Equation (1).

Table 8. Heterogeneity in 3 sectors (female employment as the dependent variable).

	(1) Total	(2) Agricultural	(3) Industrial	(4) Service
wind × post	−0.00688 *** (0.0021)	−0.00159 (0.0012)	−0.00124 (0.0010)	−0.00404 *** (0.0014)
Controls	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
N	434	434	434	434
Adj. R2	0.4219	0.2091	0.3962	0.6775

Notes: *** represent significance levels of 1%, respectively. Robust standard errors clustered by province; SE shown in parentheses.

presents the results, with Column (1) displaying the baseline results for comparison. Column (4) indicates that only female employment in the service sector is significantly negatively impacted by the policy at the 1% level, confirming the H2b hypothesis in Section 2.1.

We also present regression results for female employment across multiple industries within the service sector as the dependent variable, with a significant negative impact observed in six industries, including transportation, IT, real estate, rental services, and environmental and residential services (See

Table 9. Heterogeneity within the service sector (female employment as the dependent variable).

	(1) Transport	(2) IT	(3) Real_Estate	(4) Rental_Service	(5) Environment	(6) Resident_Service
wind × post	−0.000330 ** (0.0001)	−0.000453 ** (0.0002)	−0.000435 *** (0.0001)	−0.000544 ** (0.0003)	−0.000243 ** (0.0001)	−0.000184 ** (0.0001)
Controls	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES	YES
N	434	434	434	434	434	434
Adj. R2	0.2144	0.6352	0.7140	0.3054	0.1297	0.1641

Notes: **, and *** represent significance levels of 5%, and 1%, respectively. Robust standard errors clustered by province; SE shown in parentheses.

). These industries all exhibit characteristics related to the wind power value chain. For instance, the transportation of wind turbines and components, the need for intelligent dispatching systems in the IT industry due to the volatility of wind speeds, and the services of environmental impact assessments prior to wind farm construction. These results may implicitly suggest a decline in RE investments following subsidy reductions.

5. Conclusions and Implications

5.1. Conclusions

This study utilizes panel data from 31 provinces in China during the period of 2008 to 2021 to assess the impact of the wind power FIT adjustment policy in 2014 on female employment, in comparison to their male counterparts. A difference-in-differences (DID) approach is employed using a continuous treatment variable. The main results show that the implementation of the FIT subsidy reduction policy in 2014 substantially reduces female employment at the 1% level, while men are not impacted. Through several robustness checks including the parallel trend test, replacing the measurements of key variables and shortening the time window, we confirm that the negative impact of the policy is robust. Furthermore, the results of the heterogeneity analysis reveal no significant moderating effect of gender norms on this negative effect, which may be attributed to the inaccurate measurement of gender norms. Meanwhile, in the heterogeneity analysis by region, only provinces in the central and western regions show a significant decrease in female employment impacted by the FIT adjustment. Finally, we test the potential mechanism by observing a significant decline in the overall wage levels of all employees, and finding that only female employment in the service sector is notably affected. We infer that wage reduction likely leads to a decrease in demand for service-oriented products, a sector where women are predominantly employed, thus amplifying the gendered impact.

Since its adoption in 2015, SDG 13 has emphasized the need for urgent and effective action to tackle climate change and embed climate considerations into national development agendas. Although SDG 5 specifically targets gender equality, the two goals are closely intertwined in practice. Recognizing and responding to this intersection can generate broader development benefits. This study highlights a potential trade-off between SDG 13 (Climate Action) and SDG 5 (Gender Equality), demonstrating that structural gender inequality cannot be alleviated solely through low-carbon energy transitions that lack gender-responsive design. Without integrating gender considerations into climate policy, such efforts may inadvertently reinforce or even exacerbate existing disparities.

Sustainability issues are inherently value-laden from the very stage of problem definition, guided by a transformative perspective that envisions a shift toward a more rational and just state. At the heart of addressing sustainability challenges lies the question of how to manage complex socio-ecological systems amid multiple uncertainties and a plurality of values [59]. This calls for a thoughtful consideration of pragmatic sustainability—one that integrates environmental, economic, and social dimensions, and moves beyond abstract concepts to ensure concrete implementation supported by rigorous analysis [60]. Therefore, balancing different SDGs should be grounded in reflexivity [61] and a human-centered approach to sustainability which embraces the concept of pragmatic sustainability and ultimately advances the triple goals of economic performance, environmental protection, and social progress [62].

5.2. Limitations and Further Directions

Despite some robustness checks, the extent to which our conclusions can be generalized should be approached with caution. Firstly, the exclusion of rural and non-private sector employment due to data availability leads to a narrow representation of female employment, as it relies solely on formal urban employment and thus overlooks more than half of women workers. This could be addressed by integrating micro-level survey data with macroeconomic policy interventions, enabling a more nuanced analysis of the Chinese labor market as a whole. Secondly, this study only focuses on the quantitative aspect of female employment and does not capture changes in job quality, such as the occupational status, working conditions, and wage levels of women. Additionally, we are unable to directly distinguish among direct, indirect, and induced employment effects due to data availability. We infer induced effects based on the observed changes in service sector employment, where women are predominantly represented. This indirect approach inevitably constrains our ability to comprehensively capture the full spectrum of employment impacts resulting from RE policies. Future research could benefit from more detailed gender-disaggregated data that allow for a more thorough analysis of the gendered effects of RE policies across job numbers and types. Lastly, other potential mediating mechanisms related to firm behavior have not been considered in this study. For example, the reduction in FIT subsidies may lead to lower profits for firms, which could drive automation to replace female workers rather than male workers, as demonstrated in previous studies such as that carried out by Deng and Liu [63], in which they reported that automation in manufacturing has gendered employment impacts. Changes in recruitment gender preferences could also influence female employment. These aspects warrant further investigation in future studies incorporating micro-level firm data.

5.3. Policy Recommendations

Several practical policy recommendations could be drawn based on the findings of this study: (1) Adopting a gradual subsidy phase-out strategy with market-responsive mechanisms: To minimize labor market disruptions, policymakers should avoid abrupt and uniform subsidy withdraws. Additionally, mechanisms such as Feed-in Premium (FIP) could be adopted, as they provide financial stability while encouraging market participation. (2) Integrating gender-sensitive approaches into energy policy designs: Renewable energy policies should incorporate gender-sensitive frameworks, which requires both gender-disaggregated labor data and the active involvement of women. Furthermore, promoting gender diversity in decision-making processes can lead to a more inclusive, effective, and sustainable energy transition that benefits all segments of society. (3) Implementing policies to combat gender discrimination and support women’s employment: The Chinese government should adopt targeted measures to support women’s employment in the renewable energy sector and beyond. These policies should focus on eliminating workplace gender discrimination, implementing gender-equal employment regulations, and promoting vocational training and social support initiatives for women. Such measures would ensure that women have equal access to job opportunities in the emerging industries and that energy transition is equitable for all.