The Digital Economy and Gender Disparities in Rural Non-Agricultural Employment: Challenges or Opportunities for Sustainable Development?

Abstract

The digital economy is emerging as a transformative force for advancing inclusive sustainable development in rural China, particularly in addressing gender disparities in non-agricultural employment. Using the Digital Inclusive Finance Index and China Family Panel Studies data, this paper analyzes the digital economy’s impact on the gender gap in rural non-agricultural employment. It finds that the digital economy boosts women’s employment and wage, narrowing the gap. This effect is more obvious among groups with medium-high human capital, younger people, married people, and those with kids. The digital economy narrows the gap through the following three mechanisms: reshaping skill demands, reducing info-search cost, and promoting domestic labor socialization. It is an engine for growth and a force for gender equality in rural employment.

1. Introduction

Globally, the booming development of the digital economy is profoundly reshaping the labor market, presenting both challenges and opportunities for advancing Sustainable Development Goals (SDGs)—especially SDG 5 (gender equality) and SDG 8 (decent work and economic growth). Especially in rural areas, it has provided unprecedented opportunities for narrowing the gender gap in non-agricultural employment. In China, there are significant differences between rural male and female labor forces in non-agricultural employment [1]. According to the Monitoring Survey Report on Migrant Workers released by the National Bureau of Statistics in 2023, among the total 297 million migrant workers in the country, men accounted for as high as 62.7%, while women only accounted for 37.3%. Moreover, the gender wage gap has shown a continuous widening trend [2].

Although capital is gender-neutral, numerous studies have shown that there is a gender gap in the labor market. A study analyzing data from the United States between 1980 and 2010 found that, while the gender wage gap (Female-to-Male Earnings Ratios) narrowed from 62% to 79%, it still persists [3]. Studies have shown that the gender wage gap has been increasing rapidly in recent years, and the gender wage gap in rural areas of China is significantly higher than in urban areas [4]. The apparent contradiction between capital’s gender-neutrality and persistent industry-wide gender discrimination stems from structural and behavioral biases embedded in economic systems. While capital itself is profit-driven, its allocation and management are shaped by societal norms, institutional practices, and cognitive heuristics that perpetuate gender disparities. Implicit biases held by employers, colleagues, and institutional norms lead to discrimination against women in hiring, promotion, and pay. Gary Becker’s The Economics of Discrimination posits that discrimination can arise from employers’ preferences, clients’ bias, or statistical discrimination, where gender serves as a proxy for perceived productivity [5,6]. According to United Nations Women, women perform 76.2% of unpaid housework and childcare, leading to fragmented working hours that make it difficult to hold full-time or demanding jobs. A study highlights that high-paying occupations—such as lawyers, investment bankers, and corporate executives—typically demand ‘long hours’, while women are more likely to opt for flexible but lower-paying roles due to caregiving responsibilities [7,8]. Organizational culture and industry norms maintain gender inequality through “gendered institutions” (e.g., masculine leadership models, “maternal punishment” in promotions). For example, pregnant women are often viewed as “high-risk” employees, leading to career interruptions [9].

Previous studies have shown that there is serious gender discrimination in most industries, which is even more obvious among low-income groups, highlighting the gender divide in the labor market. Studies have proven that the impact of the COVID-19 pandemic has further exacerbated the gender gap, and this could be due to women having to take on family caregiving duties [10,11]. The probability of female workers in the UK and the US losing their jobs during the pandemic was significantly higher than that of men [12]. This was mainly because, after schools and nurseries were closed, the care burden on mothers increased, and the working hours of female workers were reduced [13]. These studies provide valuable references for researching related issues in China. Studies have shown that agricultural labor and household chores, elderly care, the childbearing process, child care, the pressure of children’s further education, and so on have a restraining effect on the non-agricultural employment of female labor forces, leading to work interruptions and shortened working hours for women and widening of the gender income gap [3,14,15,16,17,18].

Guided by the socialist principle of “the people-centered philosophy of development,” the digital economy in China has the unique potential to redress historical inequalities by reshaping labor market dynamics. For rural women, who are often trapped in unpaid care work and low-skill agricultural roles, digital platforms offer pathways to bypass traditional gendered employment segregation. The rise of the digital economy in recent years is expected to become a transformative force to break down this gender barrier, providing new opportunities and pathways for promoting women’s labor participation, improving the quality of employment, and narrowing the gender wage gap [3,19,20,21,22]. Moreover, the digital economy has promoted the transformation of the employment structure from agriculture to non-agriculture, increased residents’ incomes, and is conducive to the participation of low-skilled labor in the non-agricultural employment market [23,24]. The increase in income has also changed individuals’ reservation wages, affecting the comparative advantages of both husband and wife and the division of labor within the family, thereby increasing women’s labor participation [25]. Nevertheless, some studies also suggest that digitalization has a crowding-out effect on jobs in some traditional industries, reducing the number of employed people, shortening the working hours of the employed labor force, and having a greater impact on women and low- and medium-skilled workers [26,27]. A study also shows that formal digital credit, contrasting to expectations, has led to an increase in the gender gap in financial inclusion [28].

It can be seen from relevant studies that, although digital technology has shown potential in narrowing the gender gap, the existence of the digital divide and the urban-rural gap means that the influence of the digital economy in rural areas still needs to be explored in depth. This study contributes to the literature on digital economy, gender equality, and rural employment by addressing the following three key research frontiers that distinguish it from prior studies: (1) Unlike most existing research on urban or aggregate gender employment gaps, this study analyzes rural China. Here, traditional gender norms and structural barriers challenge women’s economic participation. By linking digital economy impacts to SDG 5 and SDG 8, it bridges micro evidence with global sustainability goals, departing from earlier studies that overlook rural-specific digitalization pathways for inclusive growth, especially for marginalized groups like married women with caregiving duties. (2) Building on the digital economy’s skill-biased findings, this study uses the U.S. Department of Labor’s O*NET database to introduce a new occupational classification framework. By categorizing jobs into digital, physical, and social skill-based occupations, it reveals the following nuanced mechanisms: digitalization reduces gender gaps in all three domains, especially amplifying women’s advantages in social/cognitive tasks and mitigating their disadvantages in physical jobs. This analysis of skill demands enriches our understanding of how digital transformation reshapes gendered occupational segregation, an underdeveloped area in the prior literature.

In view of this, based on the index of digital financial inclusion constructed by the Institute of Fintech of Peking University and the data from the China Family Panel Studies (CFPS), this paper empirically examines how the digital economy can significantly narrow the gender gap in non-agricultural employment in rural areas in terms of both employment opportunities and wage incomes. It further reveals the mechanism of the digital economy from perspectives such as changing the demand for labor skills, reducing the cost of information search, and alleviating the burden of household chores. The contributions of this paper are as follows: First, through empirical research, it provides strong evidence for the potential of the digital economy in promoting gender equality in rural areas. Second, it uses the skill scores of various occupations in the O*NET database of the U.S. Department of Labor/Employment and Training Administration (USDOL/ETA) to subdivide the occupations that individuals are engaged in, to highlight the heterogeneous impacts of the digital economy at different occupational levels. Third, it analyzes the main mechanisms through which the digital economy narrows the gender gap, providing an important reference for policymakers to promote the fair development of rural women’s non-agricultural employment.

The remaining content of this paper is arranged as follows: the second part puts forward the research design framework; the third part examines and judges whether the digital economy affects the gender gap in non-agricultural employment in rural areas; the fourth part discusses the mechanisms by which the digital economy narrows the gender gap in non-agricultural employment and income; finally, the conclusions and policy recommendations are presented.

2. Materials and Methods

2.1. Model Setup

This paper attempts to verify whether the digital economy can narrow the gender gap in non-agricultural employment from the perspectives of employment and income. Based on this, the following model is constructed:

$${\mathrm{Y}}_{\mathrm{i},\mathrm{j},\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{DE}}_{\mathrm{j},\mathrm{t}-1}+{\mathsf{\beta }}_2{\mathrm{Female}}_{\mathrm{i}}+{\mathsf{\beta }}_3{\mathrm{Female}}_{\mathrm{i}}\times {\mathrm{DE}}_{\mathrm{j},\mathrm{t}-1}+{{\mathsf{\beta }}^{\prime }}_4{\mathrm{X}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\phi }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{j},\mathrm{t}}$$

(1)

This model contains two dependent variables (${\mathrm{Y}}_{\mathrm{i},\mathrm{j},\mathrm{t}}$): One is ${\mathrm{Nonagri}}_{\mathrm{i},\mathrm{j},\mathrm{t}}$, which indicates whether individual i in region j is engaged in non-agricultural work in year t. The other is $\ln ({\mathrm{wage}}_{\mathrm{i},\mathrm{j},\mathrm{t}})$, which represents the annual wage income of individuals after logarithmic processing. The independent variable ${\mathrm{DE}}_{\mathrm{j},\mathrm{t}-1}$ represents the degree of digital economy development in the location of individuals. In this study, Peking University’s Digital Inclusive Finance Index is used as a proxy for the development level of the digital economy. To reduce the impact of reverse causality, the digital economy variable is processed with a one-period lag. ${\mathrm{X}}_{\mathrm{ijt}}$ represents the control variables at the individual, family, and regional levels. At the individual level, these include the age of workers (Age), Party membership (Party), health status (Health), educational attainment (Education), marital status (Marriage), and so on. Considering that an individual’s access to the internet largely determines the impact of the digital economy on them, following the practice of Zhang et al. [29], this paper also controls for the individual’s Internet usage situation (Internet). In addition, existing studies have pointed out that doing housework and taking care of children can affect the employment and income of rural labor forces, especially female labor forces [3,21]. Therefore, this paper further controls for two variables, the length of time an individual spends on doing housework per week (Housework) and whether they take care of children (Care_Child). At the family level, drawing on the practices of Liu et al. and Zhang et al. [1,23], this paper controls for family size (Familysize), the support burden of raising children and elderly (Raise_Ratio), per capita net income (Fincome), the proportion of internet users (Internet_Ratio), and the highest educational attainment of other family members (Other_Edu). At the regional level, it controls for per capita GDP (GDP), the proportion of non-agricultural GDP (Nonagri_GDP), the proportion of fiscal expenditure in GDP (Fiscal_Exp), and the proportion of the balance of loans from financial institutions in GDP (Loan). Finally, ${\mathsf{\phi }}_{\mathrm{t}}$ is the year fixed effect, and ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{j},\mathrm{t}}$ represents the random disturbance term.

In Equation (1), ${\mathsf{\beta }}_1$ represents the overall impact of the digital economy on the non-agricultural employment/income of rural labor forces. If ${\mathsf{\beta }}_1$ is significantly positive, then it indicates that the digital economy can increase the probability of rural residents engaging in non-agricultural employment or raise their wage income, and vice versa. ${\mathsf{\beta }}_2$ represents the existing gender gap in non-agricultural employment/income. If ${\mathsf{\beta }}_2$ is significantly positive, then it indicates that the probability of men engaging in non-agricultural employment or their wage income is higher than that of women, meaning a gender gap exists. The coefficient of the interaction term ${\mathsf{\beta }}_3$ represents the gender-heterogeneous impact of the digital economy on the non-agricultural employment/income of rural labor forces, which can be used to measure whether the digital economy narrows or widens the original gender gap. A significantly positive coefficient ${\mathsf{\beta }}_3$ suggests that the digital economy exerts a stronger enhancing effect on women than on men, thereby demonstrating its role in narrowing gender gaps. If the digital economy can generally promote the non-agricultural employment of rural individuals as a whole, then the sign of ${\mathsf{\beta }}_1$ should be significantly positive. The core issue that this paper focuses on is the gender gap in non-agricultural employment in rural areas. Much of the literature has already confirmed that women are in a disadvantaged position compared to men in the labor market. Therefore, the sign of ${\mathsf{\beta }}_2$ should be significantly negative. Meanwhile, if the digital economy can narrow the gender gap, then the marginal effect of promoting the non-agricultural employment of women should be greater than that of men, and the sign of ${\mathsf{\beta }}_3$ is expected to be positive.

Considering that the model may omit variables that simultaneously affect individuals’ non-agricultural employment and the development level of the digital economy, this paper intends to further introduce instrumental variables to address the possible endogeneity problem. Referring to the practice of Zhang et al. [29], this paper selects the spherical distance from the location of an individual to Hangzhou (Distance) as an instrumental variable. The reason for choosing the distance to Hangzhou is that the development of China’s digital economy is largely reflected in the popularization of digital finance, and the development of digital finance represented by Alipay originated in Hangzhou. The geographical distance can be regarded as a natural experiment on the popularity of digital finance. In addition, we take the interaction term between the gender variable and this distance, Female × Distance, as another instrumental variable to control the possible endogeneity between gender and the digital economy.

2.2. Data Source

This paper uses the Peking University Digital Inclusive Finance Index as an indicator to measure the development of the regional digital economy. This index is released by the Peking University Digital Inclusive Finance Research Center and comprehensively reflects the development level of digital inclusive finance in China. Since digital finance is an important part of the digital economy and China’s digital finance development is at the forefront in the world, this paper uses this index as a proxy variable for the digital economy, which is also consistent with the practice in the relevant literature of using the Digital Inclusive Finance Index as a representative variable for the development of the digital economy [26,28]. Because the research involves rural labor’s non-agricultural employment and income, this paper uses the index at the county level, as it has a more direct impact on micro individuals. Subsequently, the index at the municipal level is used for robustness tests. The data at the family and individual levels are sourced from the China Family Panel Studies (CFPS), conducted by the Institute of Social Science Survey of Peking University, with the sample period covering from 2016 to 2020. The data at the regional level are obtained from the “China Statistical Yearbook” and the “China County and District Economic Statistical Yearbook”, including per capita GDP, the proportion of non-agricultural GDP, the proportion of fiscal expenditure in GDP, etc. For some missing data, they are manually filled in through the official websites of local governments. For the classification of the types of occupations in which individuals are engaged, the O*NET database of the U.S. Department of Labor is used. This database adopts the methods of questionnaire surveys and expert scoring to rate various skills required by all occupations (such as mathematical skills, programming skills, etc.). It is possible to judge how different occupations demand a certain skill based on the scores and, thus, to classify the skill preferences of occupations.

2.3. Sample Filter and Descriptive Statistics

As the primary concern of this paper lies in the influence of the digital economy on the rural labor force, only the samples of individuals whose residence was in rural areas and whose household registration was agricultural at the time of the survey are retained. Given that the dependent variables in this study are non-agricultural employment and wage income, the samples are required to be confined to those who are employed. In accordance with the legal retirement age, the samples are further narrowed down to the working-age labor force, ranging from 16 to 60 years old. Samples of the unemployed and those not participating in economic activities are excluded, ultimately yielding a total of 19,308 samples. To mitigate the impact of extreme values, winsorization is carried out at the 1% level for both the upper and lower tails of all continuous variables.

From the results of descriptive statistics as shown in

Table 1. Description of variables.

Type of Variables	Variable	Variable Definition	Observations	Mean	Standard Deviation	Min	Max
Dependent variable	Nonagri	Non-agricultural employed = 1, Agricultural employed = 0	19,308	0.394	0.489	0	1
Independent variable	DE	Digital Inclusive Finance Index	19,308	90.127	21.510	40.730	126.385
Individual characteristics	Female	Female = 1, Male = 0	19,308	0.463	0.499	0	1
	Age	Age of the interviewee in the year of interview	19,308	42.595	11.104	16	60
	Party	CPC member = 1, Otherwise = 0	19,308	0.073	0.261	0	1
	Health	Very healthy = 1, Relatively healthy = 2, General = 3, Relatively unhealthy = 4, Very unhealthy = 5	19,308	2.902	1.226	1	5
	Education	Illiterate/semi -illiterate = 0, Elementary school = 1, Middle school = 2, high school = 3, Bachelor’s degree and above = 4	19,308	1.504	1.156	0	4
	Marriage	Married = 1, Otherwise = 0	19,308	0.863	0.344	0	1
	Internet	Use internet = 1, Otherwise = 0	19,308	0.503	0.500	0	1
	Housework	The duration of housework (hours per week)	19,308	15.804	14.037	0	70
	Care_Child	Take care of children = 1, Otherwise = 0	19,308	0.139	0.346	0	1
Family characteristics	Familysize	Number of family members	19,308	4.714	2.025	1	21
	Raise_Ratio	Ratio of family members under 6 or over 60	19,308	0.196	0.188	0	1
	Ohter_Edu	Highest education level of other family members	19,308	2.239	1.118	0	4
	Internet_Ratio	The ratio of family members use internet	19,308	0.359	0.275	0	1
	Fincome	Ln (per capita net income of the family)	19,308	9.057	1.155	4.719	11.247
	Land	Ln (value of land the family owned)	19,308	8.366	3.978	0	12.989
County characteristics	GDP	Per capita GDP of the county	19,308	10.334	0.583	9.006	12.145
	Nonagri_GDP	The share of agricultural GDP	19,308	0.213	0.107	0.006	0.459
	Fiscal_Exp	The proportion of fiscal expenditure to GDP	19,308	0.335	0.282	0.049	1.727
	Loan	The proportion of the loan balance of financial institutions in GDP	19,308	1.249	0.619	0.411	4.029
Mechanism variables	STEM_Ocp	Digital occupation = 1, Otherwise = 0	19,308	0.072	0.259	0	1
	Physical_Ocp	Physical occupation = 1, Otherwise = 0	19,308	0.684	0.465	0	1
	Social_Ocp	Social occupation = 1, Otherwise = 0	19,308	0.154	0.361	0	1

, it can be observed that the samples under investigation are predominantly engaged in agricultural employment, with non-agricultural employment accounting for around 39.4%. The samples involved in digital-skills-demanding occupations (digital occupations), physical-skills-demanding occupations (physical occupations), and social-skills-demanding occupations (social occupations) constitute 7.2%, 68.4%, and 15.4% respectively (the specific definitions are elaborated on in detail below). The overall educational attainment of the samples is relatively low, with nearly 50% of the samples possessing a primary school diploma or below. Moreover, the vast majority of the samples are married. Half of the samples have had exposure to the internet. The average weekly duration of housework is 15.8 h, and 13.9% of the samples are involved in child care activities.

Panel A in

Table 2. T-test by grouping.

Panel A: T-Test by Grouping According to Gender
	Male			Female		Group Difference (Male–Female)
	Observations	Mean	Standard Deviation	Observations	Mean	Standard Deviation	Mean Difference	pValue
Nonagri	10,478	0.465	0.499	9039	0.316	0.465	0.149 ***	0.000
Age	10,478	42.369	11.264	9039	42.904	10.919	−0.534 ***	0.001
Party	10,478	0.104	0.305	9039	0.038	0.191	0.066 ***	0.000
Health	10,478	2.757	1.189	9039	3.068	1.248	−0.310 ***	0.000
Education	10,478	1.696	1.091	9039	1.289	1.190	0.408 ***	0.000
Marriage	10,478	0.829	0.376	9039	0.902	0.298	−0.073 ***	0.000
Internet	10,478	0.548	0.498	9039	0.456	0.498	0.092 ***	0.000
Housework	10,478	12.160	13.585	9039	19.957	13.309	−7.797 ***	0.000
Care_Child	10,478	0.038	0.190	9039	0.255	0.436	−0.217 ***	0.000
STEM_Ocp	10,370	0.077	0.267	8938	0.066	0.248	0.011 ***	0.002
Physical_Ocp	10,370	0.660	0.474	8938	0.713	0.452	−0.053 ***	0.000
Social_Ocp	10,370	0.135	0.342	8938	0.177	0.381	−0.041 ***	0.000
Panel B: T-test by grouping according to region
	Developing Digital Economy Region			Developed Digital Economy Region		Group Difference (Developing–Developed)
	Observations	Mean	Standard Deviation	Observations	Mean	Standard Deviation	Mean Difference	pValue
Nonagri	10,270	0.320	0.467	9038	0.478	0.500	−0.157 ***	0.000
STEM_Ocp	10,270	0.061	0.239	9038	0.085	0.279	−0.024 ***	0.000
Physical_Ocp	10,270	0.750	0.433	9038	0.610	0.488	0.140 ***	0.000
Social_Ocp	10,270	0.122	0.327	9038	0.191	0.393	−0.069 ***	0.000
Internet	10,270	0.479	0.500	9038	0.530	0.499	−0.051 ***	0.000
Housework	10,270	16.611	14.158	9038	14.888	13.843	1.723 ***	0.000
Internet_Inf	10,270	0.242	0.429	9038	0.268	0.443	−0.026 ***	0.000
GDP	10,270	10.057	0.480	9038	10.648	0.530	−0.590 ***	0.000
Nonagri_GDP	10,270	0.267	0.093	9038	0.152	0.087	0.116 ***	0.000
Fiscal_Exp	10,270	0.427	0.289	9038	0.230	0.233	0.198 ***	0.000
Loan	10,270	1.313	0.525	9038	1.177	0.704	0.136 ***	0.000
Panel C: Cross-analysis of non-agricultural employment
Variable: Nonagri	Developing Digital Economics Region			Developed Digital Economics Region		Group Difference (Developing–Developed)
Male	0.386			0.555		−0.163 ***
Female	0.240			0.404		−0.156 ***
Group difference(male–female)	0.145 ***			0.152 ***		0.007
Panel D: Cross-analysis of wage income
Variable: Ln (Wage)	Developing Digital Economics Region			Developed Digital Economics Region		Group Difference (Developing–Developed)
Male	10.179			10.302		−0.124 ***
Female	9.726			9.837		−0.111 ***
Group difference(male–female)	0.453 ***			0.465 ***		0.003

Note: *** represents significance at the 1 percent level.

conducts a t-test by grouping according to gender. The results show that, in the male samples, non-agricultural employment, education level, health status, and access to internet are significantly higher than those in the female samples. On the contrary, however, female samples bear more responsibilities for housework and taking care of children than male samples, and more women are in a married state. From the perspective of the types of occupations, men are more engaged in digital occupations, while women are more involved in occupations related to physical labor and social interaction, but the difference between them is not significant. Panel B conducts a t-test by grouping according to regions. Regions with a digital economy development level higher than the average are classified as developed digital economy regions; otherwise, they are classified as developing digital economy regions. The results indicate that samples in developed regions are more engaged in non-agricultural work, digital work, and social work, use the internet more frequently and obtain information through it, have relatively shorter durations of housework, and their economic development levels are also higher than those of developing regions. Panels C and D present a cross-analysis of non-agricultural employment and wage income. The results show that the gender differences in digitally developed regions are greater than those in less developed regions, but the gaps are very close, and further analysis in the subsequent text is needed.

3. Results

3.1. Benchmark Regression: The Impact of the Digital Economy on the Gender Gap in Non-Agricultural Employment in Rural Areas

Table 3. Digital economy and gender gap: benchmark regression.

	(1)	(2)
Model	Probit	Margin Effect
Dependent Variable	Nonagri	Nonagri
DE × Female	0.002 **	0.001 **
	[0.001]	[0.000]
DE	0.003 **	0.001 **
	[0.002]	[0.000]
Female	−0.349 ***	−0.085 ***
	[0.101]	[0.025]
Age	−0.036 ***	−0.009 ***
	[0.001]	[0.000]
Party	0.053	0.013
	[0.043]	[0.010]
Health	−0.039 ***	−0.009 ***
	[0.010]	[0.002]
Education	0.275 ***	0.067 ***
	[0.012]	[0.003]
Marriage	−0.074 **	−0.018 **
	[0.036]	[0.009]
Internet	0.291 ***	0.071 ***
	[0.032]	[0.008]
Familysize	0.014 **	0.003 **
	[0.007]	[0.002]
Raise_Ratio	−0.049	−0.012
	[0.068]	[0.016]
Other_Edu	0.033 ***	0.008 ***
	[0.011]	[0.003]
Internet_Ratio	0.059	0.014
	[0.058]	[0.014]
Housework	−0.019 ***	−0.005 ***
	[0.001]	[0.000]
Care_Child	−0.192 ***	−0.047 ***
	[0.035]	[0.009]
Fincome	0.240 ***	0.058 ***
	[0.012]	[0.003]
Land	−0.100 ***	−0.024 ***
	[0.003]	[0.001]
GDP	−0.071 **	−0.017 **
	[0.036]	[0.009]
Nonagri_GDP	−1.530 ***	−0.372 ***
	[0.145]	[0.035]
Fiscal_Exp	−0.243 ***	−0.059 ***
	[0.060]	[0.015]
Loan	−0.036 *	−0.009 *
	[0.019]	[0.005]
Constant	0.691 *
	[0.390]
Year fixed effect	Yes	Yes
Observation	19,308	19,308
Pseudo R2	0.355	-

Notes: The figures in parentheses are robust standard errors (the same below). ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

presents the results of the benchmark regression. Since the dependent variable, non-agricultural employment, is a binary variable, the Probit regression is adopted for the benchmark model. Judging from the results of the benchmark model in column (1), the coefficient of “Female” is significantly negative, indicating that men are more likely to obtain non-agricultural jobs than women. The coefficient of the interaction term “DE × Female” is significantly positive, which means that the promoting effect of the digital economy on women’s engagement in non-agricultural employment is significantly higher than that on men, demonstrating that the digital economy has the effect of narrowing the gender gap. Column (2) shows the marginal effects of the Probit regression. It can be seen from it that, with other variables controlled, the probability of men finding non-agricultural jobs is 8.5% higher than that of women, indicating that there is a relatively large gender gap in the non-agricultural employment market and that women are in a relatively disadvantaged position. The coefficient of the interaction term shows that the digital economy can narrow this gap. The promoting effect of the digital economy on women’s non-agricultural employment is 0.1% higher than that on men, meaning that when the digital economy grows by one standard deviation (21.51), the improvement effect on the probability of women’s non-agricultural employment will be 2.15% higher than that on men. Considering that the difference in the probability of non-agricultural employment between men and women is 8.5%, it indicates that, in terms of non-agricultural employment, the digital economy can narrow the employment gap between genders by about one quarter. It can be seen that the effect of the digital economy on narrowing the gender gap is economically significant.

3.2. Endogeneity Test

Considering the possible endogeneity issues, the results of the instrumental variable estimation are presented in

Table 4. Results of IV.

	(1)	(2)	(3)	(4)	(5)
Model	IV-Probit	2SLS			2SLS
Dependent Variable	Nonagri	DE	DE × Female	Nonagri	Ln (Wage)
DE × Female	0.017 ***			0.002 *	0.020 *
	[0.005]			[0.001]	[0.011]
DE	0.032 ***			0.009 ***	0.142 **
	[0.006]			[0.002]	[0.055]
Female	−1.610 ***	0.018	150.164 ***	−0.239 **	−2.177 **
	[0.457]	[1.849]	[1.935]	[0.115]	[1.095]
Hangzhou		−4.728 ***	1.893 ***
		[0.191]	[0.190]
Female × Hangzhou		−0.021	−8.633 ***
		[0.264]	[0.279]
Observation	19,308	19,308	19,308	19,308	6579
Pseudo R2	-	0.890	0.946	0.375	0.942

Notes: The figures in parentheses are robust standard errors. ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

. It can be seen that, whether the IV-Probit method or the 2SLS method is adopted, the interaction term between gender and the digital economy remains significantly positive, which once again verifies the role of the digital economy in narrowing the gender gap. Column (5) further analyzes whether the digital economy can narrow the gender wage gap. The results show that there is a relatively large wage income gap between women and men. Meanwhile, the development of the digital economy can not only narrow the employment gap but can also reduce the wage gap, thus promoting gender equality in the labor market.

3.3. Robust Test

Table 5. Results of robust test.

Panel A (Dependent Variable: Nonagri)
	(1)	(2)	(3)	(4)	(5)	(6)		(7)
DE × Female	0.016 ***	0.021 ***	0.024 ***	0.016 ***	0.007 **	0.016 **
	[0.006]	[0.005]	[0.006]	[0.005]	[0.003]	[0.006]
DE	0.037 ***	0.018 ***	0.027 ***	0.030 ***	0.021 ***	0.041
	[0.006]	[0.006]	[0.007]	[0.006]	[0.004]	[0.093]
Female	−1.664 ***	−2.012 ***	−2.351 ***	−1.502 ***	−1.564 ***	−1.576 ***		−0.075
	[0.523]	[0.487]	[0.550]	[0.455]	[0.600]	[0.599]		[0.059]
Internet_Time								0.002
								[0.002]
Female × Internet_Time								0.006 *
								[0.004]
Observations	13,762	17,749	18,393	14,139	19,308	19,308		6921
Panel B (Dependent variable: ln (wage))
	(1)		(2)		(3)		(4)
DE × Female	0.018 **		0.020 *		0.006 **
	[0.008]		[0.011]		[0.003]
DE	0.078 ***		0.142 **		0.017 ***
	[0.028]		[0.055]		[0.005]
Female	−2.236 ***		−2.289 **		−1.564 ***		−0.428 ***
	[0.773]		[1.089]		[0.600]		[0.052]
Internet_Time							−0.001
							[0.002]
Female × Internet_Time							0.001
							[0.003]
Observations	3963		6579		6579		3480

Notes: The figures in parentheses are robust standard errors. ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

conducts robustness tests. Panel A and Panel B respectively present the robustness test results when the dependent variables are non-agricultural employment and wage income. Firstly, the education level and work situation of the interviewee’s spouse have an impact on an individual’s employment. Therefore, in column (1), the sample is limited to married individuals, and variables regarding whether the spouse is employed and the spouse’s education level are controlled. The results show that the coefficient of the interaction term remains significantly positive. To rule out the influence of policies that affect both entrepreneurship and the development of the digital economy, in column (2), the samples of non-agricultural self-employment are excluded, and the results remain robust. Since the economic development level of a region is highly correlated with its digitalization level, and the digital construction in highly economically developed regions is also faster, there may be some city-level factors that affect both the digitalization process and an individual’s non-agricultural employment behavior. Therefore, in column (3), the samples of individuals whose place of residence is in first-tier and new first-tier cities are excluded, and the results show that they remain robust. First-tier cities include Beijing, Shanghai, Guangzhou, and Shenzhen. New first-tier cities include Chengdu, Chongqing, Hangzhou, Wuhan, Xi'an, Tianjin, Suzhou, Nanjing, Zhengzhou, Changsha, Dongguan, Shenyang, Qingdao, Hefei, and Foshan.

To exclude the impact of the COVID-19 pandemic, in column (4), the observation data of 2020 are excluded. It can be seen that, when the dependent variable is non-agricultural employment, the coefficient of the interaction term remains positive at the 1% significance level. In column (5), the Digital Inclusive Finance Index at the city level is used, and the corresponding macro indicators are also replaced at the city level. The significance and sign of the coefficient remain unchanged. In column (6), provincial fixed effects are further controlled, and the coefficient of the interaction term is significantly positive at the 5% significance level. When the digital economy is taken as the core explanatory variable, what is measured is the intention-to-treat (ITT) effect of the digital economy’s development on non-agricultural employment. To analyze the impact of the digital economy on individuals more accurately, in column (7), the weekly online time of individuals is selected as the proxy variable for the digital economy because, theoretically, the development of the digital economy is closely related to the usage of internet, and the more an individual is exposed to the Internet, the greater the impact of the digital economy on them will be. It can be seen that the results also remain robust

3.4. Heterogeneity Analysis

This paper delves into the potential heterogeneity in the digital economy’s role in narrowing the gender employment gap, as presented in

Table 6. Results of heterogeneity analysis.

	Low Human Capital	Medium-High Human Capital	Age<40	Age≥40	Married	Unmarried	No Child	At Least One Child
Panel A (Dependent Variable:Nonagri)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
DE × Female	0.016 *	0.020 ***	0.021 **	0.018 ***	0.017 ***	0.031	0.014 **	0.020 ***
	[0.009]	[0.007]	[0.009]	[0.006]	[0.005]	[0.019]	[0.007]	[0.008]
DE	0.027 ***	0.036 ***	0.044 ***	0.023 ***	0.033 ***	0.022	0.024 ***	0.039 ***
	[0.010]	[0.007]	[0.010]	[0.006]	[0.006]	[0.016]	[0.007]	[0.009]
Female	−1.712 **	−1.867 ***	−1.768 **	−1.958 ***	−1.728 ***	−2.587	−1.494 **	−1.991 ***
	[0.806]	[0.628]	[0.769]	[0.556]	[0.470]	[1.753]	[0.597]	[0.700]
Observations	9619	9689	7301	12,007	16,663	2645	11,390	7918
Panel B (dependent variable:ln (wage))
DE × Female	0.019 **	0.021	0.031	0.020 ***	0.019 **	0.063	0.030	0.020 *
	[0.009]	[0.019]	[0.060]	[0.008]	[0.009]	[0.173]	[0.019]	[0.011]
DE	0.026	0.208 *	0.476	0.041 **	0.091 ***	0.647	0.171 *	0.142 **
	[0.023]	[0.107]	[0.633]	[0.016]	[0.034]	[1.319]	[0.087]	[0.055]
Female	−2.266 ***	−2.201	−2.463	−2.463 ***	−2.316 ***	−4.333	−3.207 *	−2.289 **
	[0.849]	[1.773]	[4.970]	[0.755]	[0.863]	[13.153]	[1.841]	[1.089]
Observations	1796	3997	3886	2693	5045	1534	3415	6579

Notes: The figures in parentheses are robust standard errors. ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

. Previous studies have shown that rural human capital has a moderating effect on the role of digital finance [30]. Initially, in terms of the heterogeneity of the labor force’s educational level, the paper classifies those with primary school diplomas, illiteracy, and semi-illiteracy as low-human-capital labor force, while those with junior high school diplomas and above are categorized as medium-to-high human capital labor force. This classification results in a close proportion between the two groups. Regarding employment opportunities, the results in columns (1) and (2) indicate that, for both low-human-capital and medium-high-human-capital labor forces, the digital economy significantly narrows the gender gap in non-agricultural employment. This finding aligns with the research by Tian and Zhang [24]. The digital economy, on one hand, through digital industrialization, creates employment opportunities for low-skilled labor, such as online anchors and online customer service. On the other hand, via industrial digitalization, it increases the demand for low-skilled labor like food delivery riders and couriers. However, when it comes to wage income, the digital economy has a less significant impact on the gender gap within the medium-high-human-capital group. Nevertheless, it still has a positive effect on the overall income level, and it significantly narrows the gender gap among the low-human-capital labor force.

Secondly, from the perspective of the heterogeneity of labor age, the results in columns (3–4) show that the digital economy has a significant impact on the gender gap among labor groups below 40 years old and those above 40 years old. Although older women have relatively weak competitiveness in the labor market, the digital economy can still narrow the gender gap among labor forces of different age groups, once again demonstrating the inclusiveness of the digital economy. The results from the income perspective indicate that, within the labor group above 40 years old, the digital economy can narrow the gender wage gap, but it is not significant in the group below 40 years old.

Thirdly, from the perspective of the heterogeneity of marital status, this paper divides the samples into two groups, married and non-married. The non-married group includes those who are unmarried, divorced, and widowed. Columns (5–6) indicate that the digital economy has a more significant impact on the gender gap in married samples compared to non-married samples. The reasons are as described in the mechanism analysis section. The digital economy reshapes time allocation between genders, reducing gender differences in time utilization within couples. As a result, the labor force of rural married women is more released and flows into the non-agricultural employment market. Similarly, when the explained variable is wage income, the impact of the digital economy in the married group is more significant.

Finally, from the perspective of the heterogeneity of the number of children in the family, as shown in columns (7–8), compared to individuals without children, the digital economy has a stronger effect on narrowing the gap among individuals with children. This is because Chinese women often bear more responsibility for taking care of children, which has a negative impact on their employment. However, the digital economy changes the traditional family division of labor pattern, reducing the burden of women in taking care of children, thus alleviating the negative effect of children on women’s employment. Similarly, this conclusion also holds in the income dimension.

4. Discussion

Digitalization has not only improved work efficiency but has also changed the mode of work and production. Digital production has, on the one hand, reduced heavy physical labor in many jobs and, on the other hand, increased the demand for mental workers. Skill-biased technological progress will reduce the importance of physical skills and emphasize the demand for cognitive skills, thus leading to an increase in the prices of cognitive and social skills and a decrease in the price of physical skills [31,32,33]. Therefore, the digital economy can improve the status of women in the non-agricultural employment market and narrow the gender wage gap by changing the demand for labor skills. Currently, with the rapid progress of the industrial digital transformation, “digital housekeeping” has emerged. Users can easily book housekeeping services through Internet platforms, which alleviates the daily housework and can release female labor from household labor, changing the traditional gender roles in families. If wives have more advantages in work than husbands, then it will further change the task allocation in the non-labor market, thus increasing the non-agricultural labor participation rate of women. Therefore, the digital economy can narrow the gender gap in non-agricultural employment by reducing the time spent on individual housework [25]. In addition, the digital economy has reduced the cost of information search, promoted online job search behavior, improved job matching efficiency, increased women’s job opportunities, and reduced the duration of unemployment for the labor force, and its promoting effect on women’s employment is more significant [34,35]. Therefore, the digital economy can improve the gender difference in employment by reducing the search cost.

In view of this, this paper holds that there are the following three transmission mechanisms for the impact of the digital economy on the non-agricultural employment and income of rural labor: changing the demand for labor skills, reducing the time spent on housework, and reducing the cost of information search. On the one hand, in order to verify the mechanism of the digital economy’s impact on the demand for labor skills, this paper further analyzes the specific types of non-agricultural employment promoted by the digital economy, and thus constructs Equations (2) and (3). Among them, in Equation (2), ${\mathrm{Occupation}}_{\mathrm{i},\mathrm{j},\mathrm{t}}$ represents the occupation type of workers, including digital occupations, physical-preference occupations, and social-preference occupations. If engaged in the corresponding occupation, it takes the value of 1, otherwise 0. The specific definitions are elaborated upon below. The rest of the variables are the same as those in Equation (1), where ${\mathsf{\gamma }}_3$ measures the gender heterogeneity impact of the digital economy on rural labor engaged in different occupations; Equation (3) measures whether engaging in different occupations can alleviate the wage income gap between men and women, and ${\mathsf{\eta }}_3$ represents the impact of engaging in this occupation on the gender income gap.

$${\mathrm{Occupation}}_{\mathrm{i},\mathrm{j},\mathrm{t}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1{\mathrm{DE}}_{\mathrm{j},\mathrm{t}-1}+{\mathsf{\gamma }}_2{\mathrm{Female}}_{\mathrm{i}}+{\mathsf{\gamma }}_3{\mathrm{Female}}_{\mathrm{i}}\times {\mathrm{DE}}_{\mathrm{j},\mathrm{t}-1}+{{\mathsf{\gamma }}^{\prime }}_4{\mathrm{X}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\phi }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{j},\mathrm{t}}$$

(2)

$$\ln \left({\mathrm{wage}}_{\mathrm{i},\mathrm{j},\mathrm{t}}\right)={\mathsf{\eta }}_0+{\mathsf{\eta }}_1{\mathrm{Occupation}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\eta }}_2{\mathrm{Female}}_{\mathrm{i}}+{\mathsf{\eta }}_3{\mathrm{Female}}_{\mathrm{i}}\times {\mathrm{Occupation}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{{\mathsf{\eta }}^{\prime }}_4{\mathrm{X}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\phi }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{j},\mathrm{t}}$$

(3)

On the other hand, in order to verify the mechanism of the digital economy’s impact on the time spent on housework and the cost of information search, this paper constructs Equations (4) and (5). Among them, ${\mathrm{Z}}_{\mathrm{i},\mathrm{j},\mathrm{t}}$ includes two mechanism variables, namely the duration of housework (Housework) and the information acquisition ability (Internet_Inf). Housework refers to the duration of the respondent doing housework per week. Internet_Inf measures the ability of workers to obtain information through the Internet. The China Family Panel Studies (CFPS) adult questionnaire asked respondents about the importance of the Internet for their information acquisition, with scores ranging from 1 to 5 from low to high. If the answer is 0–4 points, then the value of Internet_Inf is 0, and if the answer is 5 points, then its value is 1. ${\mathsf{\zeta }}_1$ in Equation (4) represents the impact of the mechanism variables on the non-agricultural employment variables. ${\mathsf{\zeta }}_1$ in Equation (5) represents the impact of the digital economy on workers’ information acquisition ability or the duration of doing housework. ${\mathsf{\zeta }}_2$ represents the average gap between men and women in non-agricultural employment/income, and the interaction term coefficient ${\mathsf{\zeta }}_3$ represents whether the mechanism variables narrow or widen the gender gap.

$${\mathrm{Y}}_{\mathrm{i},\mathrm{j},\mathrm{t}}={\mathsf{\zeta }}_0+{\mathsf{\zeta }}_1{\mathrm{Z}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\zeta }}_2{\mathrm{Female}}_{\mathrm{i}}+{\mathsf{\zeta }}_3{\mathrm{Female}}_{\mathrm{i}}\times {\mathrm{Z}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{{\mathsf{\zeta }}^{\prime }}_4{\mathrm{X}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\phi }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{j},\mathrm{t}}$$

(4)

$${\mathrm{Z}}_{\mathrm{i},\mathrm{j},\mathrm{t}}={\mathsf{\delta }}_0+{\mathsf{\delta }}_1{\mathrm{DE}}_{\mathrm{j},\mathrm{t}-1}+{\mathsf{\delta }}_2{\mathrm{Female}}_{\mathrm{i}}+{{\mathsf{\delta }}^{\prime }}_3{\mathrm{X}}_{\mathrm{i},\mathrm{j},\mathrm{t}}+{\mathsf{\phi }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{j},\mathrm{t}}$$

(5)

The existing literature generally analyzes from the perspective of industries when conducting arguments, and the classification is rather broad, making it difficult to capture the demand for various skills by different occupations within the same industry. Referring to the method of Li [36], this paper uses principal component analysis to reduce the dimensions of the scores of four types of digital-related skills in each occupation in O*NET, namely mathematics, science, programming, and technical design, to form corresponding labor skill variables. Occupations with scores in the top one-third indicate that these occupations have relatively high requirements for the digital-related skills of practitioners. In this paper, these occupations are defined as digital occupations (STEM_Ocp). Similarly, this paper defines social preference occupations (Social_Ocp) using the scores of six abilities, such as occupational coordination, command, and negotiation, and defines physical preference occupations (Physical_Ocp) using the scores of nine abilities such as endurance, trunk strength, and static strength. For details of the skill dimensions used in the evaluation, please refer to

Table 7. Classification of occupation types.

Variable	Require Skills	Examples
STEM_Ocp	Mathematics, science, programing and technology design	Computer and applied engineering technicians, mathematics researchers, electronic engineering technicians, etc.
Physical_Ocp	Stamina, dynamic flexibility, extent flexibility, gross body coordination, gross body equilibrium, dynamic strength, explosive strength, static strength, trunk strength	Engineering construction workers, decoration workers, steel bar processing workers, etc.
Social_Ocp	Coordination, instructing, negotiation, persuasion, service orientation, social perceptiveness	Sales and marketing personnel, lawyers, heads of social organizations and working institutions, etc.

. The China Family Panel Studies (CFPS) adopts the Chinese Standard Classification of Occupations (CSCO) occupational codes for respondents’ occupations, while the O*NET database adopts the 2019 version of the Standard Occupational Classification Code (SOC) released by the U.S. Department of Labor. The codes of the two are matched according to the conversion rules provided by the official, and a small number of occupations that cannot be matched are manually matched. Eventually, the specific occupation types of each worker can be obtained.

Panel A of

Table 8. Mechanism analysis: the types of non-agricultural employment promoted by the digital economy.

Panel A:The Types of Non-Agricultural Employment Promoted by the Digital Economy
Model:Probit	(1)	(2)	(3)
Dependent Variable	STEM_Ocp	Physical_Ocp	Social_Ocp
DE × Female	0.003 *	0.005 ***	0.002 *
	[0.002]	[0.001]	[0.001]
DE	0.006 **	−0.008 ***	0.005 *
	[0.003]	[0.003]	[0.003]
Female	−0.174	−0.608 ***	0.283 **
	[0.152]	[0.109]	[0.118]
Observations	19,308	19,308	19,308
Pseudo R2	0.229	0.314	0.248
Panel B: The impact of different types of non-agricultural employment on the gender income gap
Model:OLS	(1)	(2)	(3)
Dependent Variable	Ln (wage)	Ln (wage)	Ln (wage)
STEM_Ocp	0.033
	[0.037]
Female × STEM_Ocp	0.129 **
	[0.061]
Physical_Ocp		−0.113 ***
		[0.025]
Female × Physical_Ocp		−0.005
		[0.054]
Social_Ocp			0.048
			[0.033]
Female × Social_Ocp			0.014
			[0.049]
Female	−0.468 ***	−0.464 ***	−0.461 ***
	[0.027]	[0.029]	[0.030]
Observations	6579	6579	6579
Pseudo R2	0.163	0.164	0.162

Notes: The figures in parentheses are robust standard errors. ***, **, and * represent significance at the 1, 5, and 10 percent levels, respectively.

presents the regression results of Equation (2), testing whether the digital economy affects non-agricultural employment by changing the preference for labor skill requirements. Column (1) indicates that the digital economy does increase the probability of rural labor engaging in digital-related occupations, and this promoting effect is stronger for female labor. As mentioned earlier, on the one hand, the digital economy has weakened the demand for physical labor in many positions, and on the other hand, the demand for cognitive skills has increased. Although studies have shown that males and females exhibit significant differences in aspects such as muscles, bones, and cardiopulmonary function after puberty [37], some studies also have proven that, at the same level of human capital, women are more comprehensive and meticulous in information understanding and processing and are better operators of information and communication technology equipment; thus, women are more likely to find digital-related jobs [38].

From the results in Column (2) of Table 8, it can be seen that, in jobs requiring physical skills, the probability of male employment is significantly higher than that of women, but the digital economy has narrowed the gender gap in physical jobs. This means that, as the requirements for digital and other skills in traditional jobs increase, women’s weakness in physical strength is weakened; thus, the gap between women and men in jobs requiring physical skills also narrows with the development of digitalization. Column (3) shows the impact of the digital economy on individuals engaging in social preference occupations. It can be seen that the digital economy is conducive to individuals finding social preference occupations, and its promoting effect on women is stronger than that on men. This may be because industrial digitalization will promote the servitization transformation of various industries, and women have a comparative advantage in social skills [39]. Digitalization has increased women’s market wages relative to men by stimulating women’s social skill advantages and weakening men’s physical skill advantages; thus, digitalization has a stronger promoting effect on women’s employment [36].

Panel B shows the results of Equation (3), reflecting the impact of rural labor engaging in different types of non-agricultural employment on the gender income gap. The results in columns (1–3) indicate that engaging in digital occupations and social preference occupations is conducive to increasing the wage levels of rural labor, while the effect of physical preference occupations is the opposite. Moreover, the coefficients of the interaction terms further show that engaging in digital occupations has the effect of narrowing the wage gap, while the effects of the other two types of occupations are not significant enough, indicating that the narrowing of the gender wage gap is mainly achieved through occupations related to the digital economy.

Table 9. Mechanism analysis: the impact of the digital economy on housework.

	(1)	(2)
Model	Tobit	Probit
Dependent Variable	Housework	Nonagri
DE	−0.049 ***
	[0.013]
Female	7.871 ***	−0.042
	[0.226]	[0.041]
Housework		−0.017 ***
		[0.001]
Female × Housework		−0.007 ***
		[0.002]
Observations	19,308	19,308
Pseudo R2	0.023	0.355

Notes: The figures in parentheses are robust standard errors. *** represents significance at the 1 percent level.

examines whether the digital economy can narrow the gender gap in non-agricultural employment by reducing the time individuals spend on housework. Column (1) analyzes the impact of digitalization on the time individuals spend on doing housework. It can be seen that the time women spend on doing housework is significantly higher than that of men. Rural women do indeed undertake more housework, and the digital economy can reduce the time individuals spend on housework. In column (2), the interaction term between the duration of housework and gender is negative, which means that housework has a greater negative impact on women’s non-agricultural employment. However, the digital economy has liberated a large number of female laborers from housework, achieving the effect of narrowing the gender gap.

Table 10. The impact of the digital economy on the cost of information search.

	(1)	(2)
Model	Probit	Probit
Dependent Variable	Internet_Inf	Nonagri
DE	0.003 **
	[0.001]
Female	−0.060 **	−0.217 ***
	[0.024]	[0.029]
Internet_Inf		0.106 ***
		[0.035]
Female × Internet_Inf		0.189 ***
		[0.051]
Observations	19,308	19,308
Pseudo R2	0.142	0.354

Notes: The figures in parentheses are robust standard errors. *** and ** represent significance at the 1 and 5 percent levels respectively.

examines the mediating variable of the individual’s ability to collect information on the Internet. The results in column (1) show that the development of the digital economy can enhance the individual’s information acquisition ability. Meanwhile, the coefficient of Female is significantly negative, and this result is consistent with the fact of the digital gender divide. According to the report of the International Telecommunication Union, 52% of women globally have not used the Internet yet. Only in 8% of countries do women have a higher Internet usage rate than men, and the gender gap is larger in developing countries. In column (2), the coefficient of the interaction term is significantly positive, indicating that the improvement of the individual’s information acquisition ability has a greater promoting effect on women’s non-agricultural employment. Thus, it can be proven that the digital economy reduces the individual’s information collection cost and improves their information acquisition efficiency, thereby narrowing the gender gap in employment.

5. Conclusions

After comprehensively examining the structural challenges in the rural female non-agricultural employment market, this study reveals that the digital economy, as an emerging force, is gradually bridging the traditional gender gap and facilitating access to non-agricultural employment for rural women. This paper has conducted an in-depth analysis of the impact of the digital economy on the gender gap in rural non-agricultural employment. Using the Digital Inclusive Finance Index and data from the China Family Panel Studies, a series of empirical studies were carried out. The results show that the digital economy has not only significantly increased rural women’s non-agricultural employment opportunities but has also raised their income levels, thus effectively narrowing the gender gap. This transformation benefits from the digital economy’s reshaping of the skill requirements in the labor market, lowering the threshold for information acquisition, and at the same time reducing women’s household burden, creating a more equitable employment ecosystem for them.

To further explore the underlying mechanisms, this paper used the occupational skill evaluation system of the O*NET database of the U.S. Department of Labor to divide occupations into three types, digital, physical, and social. It was found that the digital economy shows a significant effect of narrowing the gender gap within these three types of occupations. Behind this phenomenon lies the innovation of work patterns and production methods brought about by digital technology, which reduces the dependence on physical strength and increases the value of non-physical skills, thereby effectively reducing the gender differences in non-agricultural employment and remuneration. Further, this paper also elaborates on the additional assistance provided by the digital economy to female workers from the perspectives of reducing the burden of housework and lowering the cost of information search. With the development of the digital economy, the social division of household affairs has alleviated women’s household pressure, and at the same time, the convenience of information acquisition has lowered the entry threshold of the job market, opening up a broader path for female job seekers and further narrowing the gender gap in the employment field. This study further reveals the differentiated effects of the digital economy among groups with different levels of human capital, ages, marital statuses, and number of children in the family through heterogeneity analysis. It is worth noting that the positive role of the digital economy is particularly significant among female groups with medium to high levels of human capital, in the younger age group, and those who are married with children, indicating its inclusiveness and universality for the socio-economic disadvantaged groups.

While our study effectively links digital inclusive finance to gender gaps in rural non-agricultural employment, it does not fully unpack the heterogeneous impacts of specific digital tools (e.g., e-commerce platforms, gig economy apps). Future research could explore these nuances to enhance the applicability of the findings and deepen the understanding of how digital economy mechanisms interact with local institutional environments. Nonetheless, this limitation does not undermine the study’s core contributions to documenting the transformative role of digitalization in promoting rural gender equality, which remains a critical reference for policy and academic discourse in similar contexts. The findings of this study may hold certain implications for researchers that could potentially contribute to their work in relevant areas. These results open up new research directions. Future studies could explore how different digital technologies, such as artificial intelligence and blockchain, impact rural women’s employment in more detail. There is also a need for research on the long-term effects of the digital economy on rural women’s career development and economic status. Additionally, cross-country comparisons could be conducted to understand how the findings in the Chinese context can be applied or adapted in other developing countries.

Based on this, this paper puts forward the following policy recommendations, aiming to accelerate the pace of rural women’s integration into the digital economy and promote gender equality in non-agricultural employment: First, digital skills education and popularization should be strengthened. Strengthened digital skills training in rural areas ensure that female groups have equal access to digital education resources and bridge the digital gender gap. Second, importance should be attached to the construction of digital infrastructure. The construction and optimization of rural digital infrastructure should be accelerated, the coverage and quality of network services should be improved, and usage costs should be reduced. Third, gender-neutral employment policies should be introduced. Gender-neutral employment policies should be formulated and implemented, gender discrimination in the job market should be eliminated, and it should be ensured that women enjoy equal employment rights. Fourth, a monitoring and evaluation system for gender equality should be established and improved. The changing trends of the gender gap in non-agricultural employment in rural areas should be dynamically tracked and evaluated, and a scientific basis for policy adjustments should be provided, so as to jointly promote rural women to achieve fuller and higher-quality employment in the digital economy era.