Does Digitalization Mean Equality? Digital Finance and Balanced Development Between Counties

Abstract

Digitalization has been remodeling the pattern of economic development. Digital finance, the important manifestation of digitalization in this financial field, has profound effect on reducing inequalities (Sustainable Development Goal 10). With data from 2014 to 2022 across 71 counties in the Wuling Mountain Area, our research reveals the non-linear role of digital finance in the balanced development between counties. Moreover, it explores the moderating effects of tourism development and educational level in reshaping this impact. It is found that there is a U-shaped relationship between digital finance and balanced development between counties. With the improvement of government support, the positive effect of digital finance becomes positive and stronger. We demonstrate that tourism development and educational level exert a negative moderating effect on the association between digital finance and balanced development between counties. This research has far-reaching implications for the policy of coordinated development between regions in China.

1. Introduction

The wealth inequality both within and between nations has continued to increase [1], considerably hindering the achievement of the Sustainable Development Goals (SDGs), especially for SDG 10 (social fairness). In the new century, increasingly more developing countries pay attention to economic growth while ignoring balanced development between nations. China, the largest developing country, presents a tremendous development gap between various areas [2]. Although the relative gap has been narrowed through the implementation of a series of policies such as “Targeted Poverty Elimination” and “Common Prosperity” [3], the absolute disparity has widened. Balanced development between areas means that the development gaps among areas are relatively smaller [4]. In particular, the cooperation and communication among various areas is frequent. As a consequence, accelerating economic growth while promoting balanced development between areas is an important challenge faced by China.

Digital finance (DF), a new financial service, integrates digital technologies [5]. In a broader sense, DF has permeated into various fields of socio-economic development [6]. Especially, extant literature has focused on the DF–social fairness nexus [7]. The influence of DF on urban–rural income gaps is one of the important research topics in this field. One viewpoint holds that DF contributes to narrowing urban–rural income disparity through increasing employment opportunities and building up business in rural areas [8,9]. Owing to the digital divide, it is found that DF aggravates the urban–rural income gap [10,11]. Although scholars have explored the effect of DF on poverty alleviation in less-developed areas [12,13], there is a dearth of direct literature on the association among DF and balanced development between areas, with a particular scarcity of studies at the county level in under-developed areas. Thus, what is the nature of the relationship between DF and balanced development between counties?

Given the influences of external factors, the relationship between DF and balanced development between counties may vary. For example, the tourism industry is the vital force driving economic growth in under-developed areas based on the tourism-led growth hypothesis (TLGH) [14]. Therefore, can tourism development positively moderate this effect of DF on balanced development between counties? Moreover, improving educational level is conducive to narrowing the digital divide [15,16]. Likewise, can educational level positively moderate this impact of DF on balanced development between counties? In order to answer the above questions, our research is conducted, which sets out to delve into the role of DF in balanced development between counties. Meanwhile, our research examines the moderation effects of both tourism development and educational level on the DF–balanced development between counties nexus.

With regard to theoretical contributions, this study enriches and broadens the research content on the SDGs, namely reducing inequality (SDG 10). Compared to the previous literature that concentrates on the direct effect of DF on income inequalities, this study introduces two moderating variables (tourism development and educational level) into model estimations so as to examine the effect of exogenous variables on the DF–balanced development nexus. This can provide a new research framework and methods for similar research. In terms of practical contributions, our research can provide references for policy-makers who expect to achieve coordinated development between areas, especially in under-developed areas.

The remaining sections are organized as follows. Section 2 reviews the relevant literature. Section 3 presents the supporting theories and provides hypotheses. Section 4 and Section 5 show the research design and empirical results, respectively. Section 6 discusses the empirical findings compared to the extant literature. Section 7 summarizes the main conclusions and provides recommendations as well as future prospects.

2. Literature Review

2.1. Digital Finance and Urban–Rural Income Gap

The urban–rural income gap represents a manifestation of the imbalanced development between urbans and rural areas [17]. DF can mitigate urban–rural income gaps, given high availability and low cost [18,19]. Furthermore, the influencing paths by which DF mitigates urban–rural income disparity have been delved into by researchers. Some scholars believe that DF can narrow urban–rural income gaps through economic growth [9], industrial upgrading [18,20], technological innovation [19], and employment improvement [21]. However, quite a few researchers unveil a disagreement that DF exacerbates urban–rural income gaps. More seriously, it enlarges urban–rural income gaps in the adjacent regions because of a lack of unified planning and tight cooperation [10]. Taking into account the dynamic characteristics, Xiao, Yin, Wang, and Xiang [11] demonstrate that there is an inverted U-shaped relationship between DF and urban–rural income gaps.

2.2. Digital Finance and Balanced Development Between Areas

There is a dearth of direct literature on the relationship between DF and balanced development between areas. Scholars mainly discuss the impact of DF on poverty reduction in rural areas [22]. For example, DF can accelerate poverty reduction through creating employment opportunities and promoting economic growth in the Middle East and Northern Africa [13]. Also, Dawood et al. [23] demonstrate that DF is able to offer fiscal support and employment positions, thereby reducing the poverty rate in Indonesia. In China, the link between DF and the poverty-returning risk has attracted great concerns. Extant literature manifests that DF is beneficial for overcoming the challenge of poverty-returning risk and thus achieving the goal of common prosperity in under-developed areas [12,24,25]. Moreover, some scholars have paid attention to the relationship between DF and income inequality [26]. For instance, de Moraes et al. [27] found that DF is more important in emerging economics.

Although prior literature considerably examined the effect of DF on urban–rural income gaps, fewer scholars have investigated the role of DF in balanced development between areas, with a particular dearth of studies at the county level in less-developed areas. Quite a few scholars have investigated the impact of DF on escaping from the poverty trap; however, more empirical studies need to be conducted to examine the non-linear relationship between DF and balanced development between counties. Tourism development is conducive to improving digital infrastructures. Meanwhile, education can narrow the digital divide. Thus, it is worthwhile to explore how the above two variables moderate the DF–balanced development between counties nexus.

3. Theoretical Analysis and Research Hypothesis

3.1. The Impact of Digital Finance on Balanced Development Between Counties

Metcalfe’s Law points out that the internet economy is characterized by high permeability and strong externality [28]. DF is a financial product of the internet economy. Analogously, the spillover effect (the impact of economic activities is not limited to direct beneficiaries but spreads to other individuals, groups, or regions) brought by DF becomes stronger when providing financial services for under-developed areas or for disadvantaged groups [29,30]. DF lowers the imbalance of resource allocation, improves income distribution, and decreases exclusiveness in under-developed areas [31,32]. Furthermore, neoclassical economics emphasizes that an inadequate market aggravates economic imbalance [33]. Fortunately, DF is conducive to overcoming market failure through promoting the interaction of production factors, thereby compensating the dearth of information [34,35]. In summary, DF can boost balanced development between counties.

However, given a dearth of financial knowledge and digital literacy in less-developed areas, DF may expand the development gap between counties through financial exclusion [36]. Above all, the issues of immature intermediaries and financial suppression caused by DF in early stages can widen development gaps between developed counties and undeveloped counties. Moreover, the digital divide can dilute inclusiveness of DF and hinder balanced development between counties [37,38]. In addition, poor digital infrastructures keep less-developed areas locked out of financial resources [39,40]. Therefore, because of the limitations of digital infrastructure and literacy, DF is not beneficial for balanced development between counties. Taking the above analysis into account, this study puts forward the following hypothesis:

H1: 

Digital finance has a non-linear effect on balanced development between counties.

3.2. Digital Finance, Tourism Development, and Balanced Development Between Counties

The TLGH stresses that tourism development can boost economic growth through increasing tax, creating jobs, stimulating consumption, and introducing investment [41,42,43]. Therefore, tourism development can accelerate economic growth in under-developed areas and thereby mitigate the development gap between developed areas [44,45]. In addition, there is no doubt that economic growth is conducive to the prosperity of DF by providing funds and improving digital infrastructure, particularly in undeveloped areas [46,47,48]. Thus, tourism development may serve as a catalyst to shrink development disparity between areas through digital finance. Therefore, this research raises the following hypothesis:

H2: 

Tourism development can promote the positive effect of digital finance on balanced development between counties.

3.3. Digital Finance, Educational Level, and Balanced Development Between Counties

Digital literacy and financial knowledge are two basic conditions needed for the development of DF [49,50]. Furthermore, financial exclusion caused by a dearth of digital literacy and financial literacy is not conducive to expanding the coverage and depth of DF in less-developed areas, thereby impeding balanced development between counties [16,51]. Digital literacy and financial knowledge are boosted by increased educational level in undeveloped areas [15]. Moreover, the improvement of digital literacy and financial knowledge sets the foundation for the far-ranging utilization of DF. Especially, it is found that education is beneficial for bridging the digital divide and thus promoting regional fairness [52,53]. Additionally, the enhancement of human capital through mutual promotion between educational level and DF contributes to lowering unemployment rates and boosting technological innovation [54,55]. Therefore, this study raises this hypothesis:

H3: 

The improvement of educational level can expand the positive effect of digital finance on balanced development between counties.

The diagram of the theoretical mechanism is as follows (Figure 1).

4. Research Design

4.1. Research Area

The Wuling Mountain Area (WMA) is located at the junction of four provinces (municipalities) in China, namely Hubei Province, Hunan Province, Guizhou Province, and Chongqing Municipality (Figure 2). The tremendous development gap between counties inevitably hinders high-quality development in the WMA. For instance, the GDP of Hecheng County was more than thirteen times that of Guzhang County in 2022. Therefore, promoting economic growth while narrowing the development disparity among counties is suggested in the WMA. DF is conducive to lowering financial thresholds and further providing financial resources for under-developed counties in the WMA. Above all, DF helps technological innovation and in building up business, which is beneficial for the achievement of common prosperity.

4.2. Model Setting

4.2.1. Benchmark Regression Model

This paper uses the two-way fixed effect model to examine the effect of DF on the balanced development between counties. The benchmark regression model is set as follows.

$${Bal}_{it}={\alpha }_0+\alpha {Dig}_{it}+{\displaystyle \sum {\beta }_j}{Control}_{jit}+{\mu }_i+{\varphi }_t+{\epsilon }_{it}$$

(1)

where Bal_it is the degree of balanced development of i county in t year. Dig_it is DF. Control_jit is a control variable. μ_i, φ_t, and ε_it are the individual fixed effect, the time fixed effect, and the random disturbance term, respectively.

4.2.2. Panel Threshold Regression Model

A panel threshold regression model is adopted to investigate the non-linear relationship between DF and balanced development between counties. The panel threshold regression model is as follows.

$${Bal}_{it}={\alpha }_0+{\alpha }_1{Dig}_{it}\times I\left({Dig}_{it}\le \gamma \right)+{\alpha }_2{Dig}_{it}\times I\left({Dig}_{it}>\gamma \right)+{\displaystyle \sum {\beta }_j}{ontrol}_{jit}+{C\epsilon }_{it}$$

(2)

where γ denotes a threshold variable; I (×) is an indicative function.

4.2.3. Moderating Effect Model

This study adopts the moderating effect model to reveal the moderating roles of tourism development as well as educational level. The model is as follows.

$${Bal}_{it}={\alpha }_0+{\alpha Dig}_{it}+{\alpha }_1{M}_{it}+{\alpha }_2{M}_{it}\times {Dig}_{it}+{\displaystyle \sum {\beta }_j}{Control}_{jit}+{\mu }_i+{\varphi }_t+{\epsilon }_{it}$$

(3)

where M_it represents a moderating variable.

4.3. Variable Measurement

4.3.1. Dependent Variable

The Theil index, Gini index, and coefficient of variation are able to assess the whole gap of economic development, while the index regarding balanced development can calculate the balanced development of each region in each year [56]. Moreover, the above three indexes evaluate the situation of balanced development between counties based on datasets at the town level. Unfortunately, there is a lack of datasets at the town level in the WMA. This study adopts the index regarding balanced development to analyze the balanced development between counties (Bal) in the WMA [4]. This formula is as follows.

$$Ba{l}_{it}={\left\{{\displaystyle \frac{\left|{rpcgdp}_{it}-rpcgd{p}_t^M\right|}{rpcgd{p}_t^M}}+1\right\}}^{-1}$$

(4)

where Bal_it is the degree of balanced development of i county in t year; rpcgdp_it denotes the per capita GDP; $rpcgd{p}_t^M$ represents the media of per capita GDP among all counties in the WMA. Bal_it ranges from 0 to 1. The larger Bal_it is, the smaller the development disparity between counties is.

4.3.2. Independent Variable

Because of the advantages of low threshold, low cost, and high convenience, DF can provide financial resources for under-developed counties [57]. Our study adopts the index regarding digital financial inclusion to evaluate the development level of DF (Dig). The index of digital financial inclusion includes three dimensions, namely, coverage breadth, usage depth, and digitalization level.

4.3.3. Moderating Variable

(1) Tourism development (Tour). Tourism development plays an indispensable role in accelerating economic growth in less-developed areas. The ratio of tourism income to the GDP is adopted to mirror the level of tourism development [43].

(2) Educational level (Edu). Given the fact that the WMA is an under-developed mountain area, the educational level is generally low [58]. The progress of primary education profoundly helps economic development and poverty alleviation. More importantly, the dataset about years of schooling among adult population, upper-secondary completion rates, and tertiary share is lacking. Referring to the practice of Gu and Shen [59], we use the number of primary and secondary students per 10,000 people to evaluate the educational level.

4.3.4. Control Variable

Referring to Li, Li, and Kong [4] and Ni, Liu, and Huang [56], we add some control variables to reduce biases of model estimations.

(1) Government support (Gov), assessed by the ratio of fiscal expenditure to GDP, can mitigate the development gap between counties through transfer payment.

(2) Transportation accessibility (Trans), measured by the total length of highways, exerts an effect on balanced development between counties by accelerating the flow of production factors.

(3) Urbanization (Urb) can break the urban–rural dichotomy and further narrow development gaps. However, the siphonic effect caused by urbanization may aggravate development disparity between counties. This study adopts the ratio of urban residents to total people to represent urbanization.

(4) The upgrading of industrial structure (Indus) is beneficial for industrial transfer among counties, which promotes economic growth in under-developed areas, thereby shrinking development disparity. The ratio of value-added of the tertiary sector to value-added of the secondary industry is employed to represent the industrial structure.

(5) County scale (Scal). Given the unsubstantial infrastructure, the expansion of county scale hinders the siphonic effect from neighboring counties. In contrast, the spillover effect of technological innovation caused by the expansion of the county scale can enlarge the trickle-down effect from neighboring regions. This paper uses the population density to mirror county scale.

(6) Consumption level (Cons). The flow of production factors caused by consumption exerts a positive effect on balanced development between counties. Moreover, consumption can stimulate industrial upgrading and promote economic growth in under-developed counties. The ratio of total retail sales to GDP is adopted to mirror the consumption level.

4.4. Data Sources

County, the basic unit of economic development in China, plays a vital role in the coordinated development between areas. Seventy-one counties in the WMA are chosen as the empirical object for this study. The original data is mainly taken from China Statistical Yearbook (County-level, 2015~2023). The dataset on tourism development comes from annual statistical bulletins of each county. Moreover, the original data of DF at the county level is from the Peking University Digital Financial Inclusion Index of China (PKU-DFIIC). The descriptive statistics of each variable can be seen in

Table 1. Descriptive statistics of variables.

Type	Variables	Mean	Std. Dev.	Min	Max	VIF
Dependent variable	Bal	0.761	0.153	0.293	0.999
Independent variable	lnDig	4.477	0.432	1.903	4.987	1.41
Moderating variable	lnTour	2.415	1.0223	−0.275	6.189	1.60
	lnEdu	10.880	0.610	8.931	12.274	1.06
Control variable	lnGov	−1.169	0.387	−2.879	−0.292	1.80
	lnTrans	7.956	0.666	5.704	9.221	1.31
	lnUrb	−0.855	0.262	−2.846	−0.069	1.62
	lnIndus	0.569	0.668	−1.139	4.296	1.26
	lnScal	5.175	0.500	4.109	6.894	1.58
	lnCons	−0.040	0.573	−1.490	1.562	1.24

. In addition, the variance inflation factor (VIF) of the independent variable, moderating variable, and control variable is less than 5, illustrating that there is no issue of multicollinearity. The value of the dependent variable ranges from 0 to 1; therefore, relevant variables are taken as logarithms so as to eliminate the effect of heteroscedasticity, except for the dependent variable.

5. Empirical Results

5.1. Baseline Regression Analysis

Our study adopts various methods to uncover the effect of DF on balanced development between counties (

Table 2. Baseline regression results.

Variables	lnBal
	(1)	(2)	(3)	(4)	(5)
lnDig	0.053 *** (0.019)	0.079 *** (0.013)	0.054 ** (0.021)	0.095 *** (0.028)
L.lnDig					0.058 *** (0.009)
Control variables	Yes	Yes	Yes	Yes	Yes
Constant	0.570 ** (0.239)	0.224 (0.182)	1.461 ** (0.666)	1.361 ** (0.668)	0.233 (0.170)
County FE	No	No	Yes	Yes	Yes
Year FE	No	Yes	No	Yes	Yes
Observations	639	639	639	639	639
R2	0.243	0.261	0.154	0.152	0.309

Note: The values of clustered robust standard errors are in parentheses; **, and *** represent 5%, and 1%, respectively.

). Columns (1) to (3) present estimation results of the OLS, the year fixed effect (FE), and the individual FE regressions, all of which demonstrate that DF can promote balanced development between counties through knowledge spillover, resource allocation, and economic cooperation. Column (4) shows that the lnDig is significantly positive (α = 0.095, p < 0.01), which implies that DF can mitigate development gaps between counties.

Furthermore, the regression coefficient of L.lnDig is significantly positive, showing that there is an accumulation effect of DF on balanced development between counties. In other words, it takes time for the achievement of narrowing development gaps among counties through DF. As a consequence, the long-term effect of the development of DF needs to be fully taken into consideration in less-developed areas.

5.2. Robustness Analysis

5.2.1. Substituting Explained Variable

This study re-calculates the degree of balanced development between counties through changing the calculating model. The formula of the new method is as follows.

$$Ba{l}_{it}=\left|\ln rpcgd{p}_{it}-\ln rpcgd{p}_{-mt}\right|/\left(\left(\ln rpcgd{p}_{it}+\ln rpcgd{p}_{-mt}\right)/2\right)$$

(5)

where rpcgdp_it is the per capita GDP; rpcgdp_m_t is the average value of per capita GDP in the WMA.

Column (1) of

Table 3. Robustness analysis results.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
lnDig	0.233 *** (0.079)	0.045 ** (0.022)	0.064 ** (0.024)	0.051 * (0.028)	0.048 *** (0.011)	0.054 *** (0.015)
Control variables	Yes	Yes	Yes	Yes	Yes	Yes
Constant	0.475 * (0.268)	0.923 (0.808)	2.520 ** (1.112)	0.567 (0.405)	−22.601 (18.383)	0.344 *** (0.128)
County FE	No	Yes	Yes		Yes	Yes
Year FE	No	No	Yes		Yes	Yes
County × Year FE	No	No	No		Yes	No
LM statistic						74.071 (0.000)
F statistic						6386.42 <16.38>
Observations	639	587	426	639	639	639
R2	0.131	0.123	0.047		0.882	0.304

Note: The values of clustered robust standard errors are in parentheses; *, **, and *** represent 10%, 5%, and 1%, respectively.

shows that the lnDig is positively correlated with lnBal (α = 0.233, p < 0.01), indicating that DF is beneficial for reducing development disparities between counties. Consequently, the results of this study are reliable.

5.2.2. Decreasing Particular Samples

Because economic development may be influenced by the administration faculty, this study eliminates six districts in the WMA and re-estimates the regression model. The estimation coefficient of the lnDig to the lnBal passes the significance test (α = 0.045, p < 0.05) in column (2), which further illustrates that the results of baseline regression are reliable.

5.2.3. Eliminating Unusual Years

COVID-19 exerts a negative effect on economic development across the world, which inevitably aggravates the development disparity between areas. Affected by COVID-19, the economy of some counties stopped increasing and even showed a decrease in the WMA. Consequently, this paper reduces the research samples from 2020 to 2022. The lnDig shows a positive influence on the lnBal in column (3) of Table 3. Therefore, the results of this study are stable.

5.2.4. Alternative Estimation Regression

We re-estimate the effect of DF on balanced development between counties by adopting the feasible generalized least squares model, which can modify the issue of heteroscedasticity. Column (4) illustrates that the lnDig shows a significantly positive relationship with lnBal, implying that the results of baseline regression are still convincing.

5.2.5. Controlling High-Dimensional Fixed Effect

External factors may change in the context of different counties or years. As exhibited in column (5) of Table 3, the impact of the lnDig is still positive at a 1% level, while controlling the County × Year FE. This suggests that the results of baseline regression are robust.

5.2.6. Endogeneity Test

This research establishes the instrumental variable and conducts a 2SLS model to solve the endogeneity issue. Referring to Zhang et al. [60], we adopt the spherical distance between the city of a county and Hangzhou City as the instrumental variable. The specific reasons for this selection are as follows. First, the development of DF is represented by Alipay that originated in Hangzhou City; it is obvious that the closer it is to Hangzhou City, the more prosperous the development of DF should be. As a consequence, the spherical distance between the city of a county and Hangzhou City meets the requirement of correlation. Moreover, the spherical distance between the city and Hangzhou City is natural data, and it has less impact on the balanced development between counties. Thus, this instrumental variable is in line with the requirement of exogeneity. Because the spherical distance between the city and Hangzhou City is static data, we use the interaction term of this spherical distance and DF as the final instrumental variable and further conduct the 2SLS model.

The LM statistic passes the significance test at the 1% level, which rejects the null hypothesis regarding the insufficient identification of the instrumental variable. Furthermore, the value of the F statistic is greater than the critical value of the Stock-Yogo 10% significance level, which rejects the null hypothesis of a weak instrumental variable. In short, the Bartik instrument variable is suitable. As shown in Table 3 column (6), the lnDig presents a positive influence on the lnBal (α = 0.054, p < 0.01), which is consistent with the baseline model that DF positively affects balanced development between counties.

5.3. Heterogenous Analysis

5.3.1. Formerly Poor Counties and Non-Formerly Poor Counties

Although all counties have got rid of absolute poverty in the WMA, there is the enormous development gap between formerly poor counties and non-formerly poor counties. Meanwhile, because of various resource endowment and location conditions, the balanced development between counties in the WMA cannot be ignored. As displayed in columns (1)–(2) of

Table 4. Effect heterogeneity across various types of counties.

Variables	Formerly Poor Counties	Non-Formerly Poor Counties	Interprovincial Border Counties	Non-Interprovincial Border Counties	2014~2020	2021~2022
	(1)	(2)	(3)	(4)	(5)	(6)
lnDig	0.088 ** (0.040)	0.105 ** (0.039)	0.095 * (0.053)	0.096 *** (0.032)	0.071 *** (0.024)	−0.069 (0.152)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes
Constant	1.355 (1.075)	1.035 (0.742)	0.593 (0.856)	2.358 ** (0.999)	2.549 *** (0.965)	3.664 ** (1.479)
County FE	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Observations	378	261	243	396	497	142
R2	0.102	0.074	0.246	0.151	0.071	0.023

Note: The values of clustered robust standard errors are in parentheses; *, **, and *** represent 10%, 5%, and 1%, respectively.

, the estimation coefficients of lnDig to lnBal are significantly positive in formerly poor counties and non-formerly poor counties. However, the influence of lnDig on lnBal in non-formerly poor counties is larger, which can be accounted for by the increased platform for exchanging resources. This strengthens the irradiation effect and facilitating effect of DF in relatively developed counties. Accordingly, the spillover effect of information, technology, and knowledge can further narrow the development gap among non-formerly poor counties.

5.3.2. Interprovincial Border Counties and Non-Interprovincial Border Counties

Given the limitation of terrain conditions, the flow of resource factors suffers from a series of obstacles in interprovincial border counties of the WMA. Therefore, there are differences in the positive impact of DF. Columns (3)~(4) of Table 4 indicate that the lnDig exerts a significantly positive effect on lnBal in interprovincial border counties and non-interprovincial border counties, demonstrating that DF can shrink the development gap between counties. Because of the limitation of niche conditions and transportation conditions, the flow of DF faces a series of challenges such as local protectionism and administrative barriers. Therefore, the effect of narrowing development disparities associated with DF in the interprovincial border counties is weaker.

5.3.3. Before and After Poverty Alleviation

China implemented the policy of Targeted Poverty Alleviation during the period of 2014~2020, which has eliminated the absolute poverty in the WMA and promoted balanced development between counties. However, given the decreased marginal effect, the absolute gap among counties in the WMA inevitably expanded. Two sub-group regressions are shown in columns (5)–(6). During the period of 2014~2020, the estimation coefficient of DF to balanced development between counties is positive at the 1% significance level. However, the coefficient of lnDig to lnBal is not statistically significant during the period of 2021~2022, indicating that DF does not exert an apparent effect on the balanced development between counties. Economic cooperation and communication driven by administrative orders were reduced, which results in the recession of the positive impact of DF on balanced development between counties during the period of 2021~2022.

5.4. Non-Linear Effect Analysis

Because of the digital divide, DF may exacerbate the development disparity among counties. As a consequence, this research further investigates whether the influence of lnDig on lnBal is non-linear. Taking into account the fact that government support is a crucial path to narrow the digital divide [61], this paper adopts the government support as the threshold variable. As shown in

Table 5. Estimation results of threshold values.

Threshold Model Type	F-Value	1%	5%	10%	Threshold Estimate Value	95% Confidence Interval
Single threshold model	44.387 ***	8.374	3.871	1.940	−1.580	[−1.625, −1.560]
Double threshold model	150.563 ***	25.184	14.349	10.721	−1.375	[−1.376, −1.373]
Triple threshold model	20.923 ***	11.269	7.332	5.904	−1.187	[−1.295, −1.078]

Note: *** represents 1%.

, the single threshold value (−1.580, p < 0.01), double threshold value (−1.375, p < 0.05), and triple threshold value (−1.187, p < 0.01) all pass the significance test, indicating that there is a triple threshold effect of the government support on the impact of lnDig on lnBal. Therefore, this study conducts the triple threshold model to examine the non-linear effect of DF on balanced development between counties. As depicted in Figure 3, the red dashed line shows the critical value of 7.35 at the significance level of 5%. The LR statistics of three threshold values all fall below the critical value, indicating that three threshold values are valid.

The results in column (1) of

Table 6. Non-linear effect and moderating effect analyses.

Variables	(1)	(2)	(3)
lnDig		0.372 *** (0.091)	0.498 * (0.294)
lnDig × I (TH ≤ r1)	−0.019 (0.021)
lnDig × I (r2 ≥ TH > r1)	0.026 (0.020)
lnDig × I (r3 ≥ TH > r2)	0.062 *** (0.019)
lnDig × I (TH > r3)	0.079 *** (0.019)
L.lnTour		0.347 *** (0.098)
lnEdu			0.171 (0.131)
lnDig×L.lnTour		−0.074 *** (0.020)
lnDig×lnEdu			−0.037 (0.027)
Control variables	Yes	Yes	Yes
Constant	0.749 (0.136)	0.004 (0.051)	−0.468 (1.551)
County FE		Yes	Yes
Year FE		Yes	Yes
Observations	639	639	639
R2	0.282	0.662	0.150

Note: The values of standard errors are in parentheses of column (1); The values of clustered robust standard errors are in parentheses of columns (2) and (3); *, and *** represent 10%, and 1%, respectively.

exhibit that, when the value of lnGov is lower than −1.580, the lnDig has a negative effect on the lnBal, indicating that DF widens development gaps through a polarization effect that refers to the phenomenon in economic development where developed regions accelerate their growth due to accumulated advantages, leading to widening economic disparities between regions. There is a lack of digital construction and digital literacy in under-developed counties, which hinders the rapid development of DF. Because of the existence of a digital divide, DF flows to relatively developed counties and thus generates the polarization effect. This contributes to economic growth and residents’ wellbeing in relatively developed counties. However, given a lack of macroeconomic regulatory, DF is not beneficial for the economic transition and technological progress. Eventually, the development gap between counties will further widen.

When the value of the lnGov ranges from −1.580 to −1.375, the coefficient of lnDig becomes positive but fails to pass the significance test. This shows that the weak degree of government support has no apparent effect on the high-quality development of DF. Especially, the construction of digital infrastructure in the WMA heavily depends on the investment from governments. When the value of lnGov ranges from −1.375 to −1.187, the regression coefficient of the lnDig to the lnBal is positive at a 1% level. Moreover, when the value of lnGov is more than −1.187, the effect of DF on the balanced development between counties gradually becomes larger. The results demonstrate that government support contributes to the construction of digital infrastructures, the improvement of digital literacy, and increased digital investment. In addition, the continuous development of DF helps under-developed counties to promote industrial transition and technological innovation, which eventually narrows development gaps between counties. Therefore, these findings substantiate H1 and reveal a U-shaped relationship among DF and balanced development between counties. By comparing this threshold with the actual levels of government support across counties, this study discovers that, in most counties, DF can effectively narrow the development gap between counties. For example, in 2022, DF in 56 counties plays a positive role in promoting coordination development with other counties.

5.5. Moderating Effect Analysis

5.5.1. The Moderating Effect of Tourism Development

In order to reduce the impact of the endogeneity issue, tourism development is a lagged treatment. As shown in column (2) of Table 6, the estimation coefficient of L.lnTour to lnBal is significantly positive, which implies that tourism development can help to narrow regional gaps in the WMA. However, the regression coefficient of lnDig × L.lnTour is significantly negative (α₂ = −0.074, p < 0.01), manifesting that tourism development negatively moderates the link among DF and balanced development between counties.

The WMA, an under-developed area, has a vast number of high-level attractions such as Wulingyuan, Enshi Grand Canyon, Fenghuang Ancient City, and so on. There is no doubt that tourism development plays an indispensable role in driving socio-economic development. Especially, some under-developed counties such as Fenghuang, Lichuang, and Jiangkou have alleviated absolute poverty through tourism development. Therefore, tourism development itself is conducive to narrowing development gaps between counties.

However, there are tremendous gaps in terms of tourism income in the WMA. For example, the tourism revenue in Enshi was ten times that of Guzhang in 2022. Counties with a higher level of tourism development tend to enhance digital investment and improve digital infrastructures, thus promoting the development of DF. This polarization effect generated by the gap of digital input can expand the developed gap between counties. In other words, because of the differences in the ability to obtain resources, tourism development can counteract the positive externalities associated with DF. Therefore, H2 is rejected.

5.5.2. The Moderating Effect of Educational Level

Column (3) of Table 6 shows that educational level contributes to balanced development between counties, demonstrating that improving educational level plays the role of a catalyst in poverty alleviation. However, the positive effect is not statistically significant, which can be explained by the fact that behindhand education struggles to roundly support technological innovation. Meanwhile, there is a dearth of cooperation of educational fields.

In addition, the interaction term of DF and educational level does not pass the significance test, with a coefficient of −0.037, showing that educational level has a negative moderating role in the relationship between DF and balanced development between counties. The specific reasons are as follows. First, the relatively lower level of education is not beneficial for narrowing digital divides. Generally speaking, the improvement of digital literacy depends more on higher education. In other words, it is difficult for the WMA to expand the coverage and depth of digital financial services. Second, concentrating on the improvement of educational infrastructure while ignoring the cultivation of digital literacy and financial knowledge reduces the inclusiveness of DF. Over the long term, the development gap among counties will be gradually enlarged. Thus, H3 is rejected.

6. Discussion

This study mainly analyzes the effect of DF on balanced development between counties and explores how tourism development and educational level moderate this effect. Moreover, the non-linear effect of DF on balanced development between counties is also examined.

It is found that DF can boost balanced development between counties. It is consistent with the previous viewpoint that the digital economy has significantly promoted coordinated development in China [62]. This positive impact can be explained through the fact that DF contributes to accelerating the flow of capital among areas [31]. Especially, with the promotion of credit quality, financial resources flow to less-developed counties through digital platforms. On the contrary, Zhang et al. [63] hold that financial development has a stronger positive effect on high-quality growth in developed regions. Unlike traditional financial products, DF, which is characterized by low thresholds and low costs, shows inclusiveness for less-developed counties [64].

Our research delves into the non-linear association between DF and balanced development between counties. Given a dearth of constraint and supervision, DF can exacerbate the development gap between counties through a siphonic effect and digital divide, which is in accordance with the finding that digital financial inclusion may cause issues of social inequality during the initial period [65]. For example, Shang [66] found that DF initially exacerbates educational inequality due to the digital divide. With the improvement of government support, DF effectively promotes balanced development between counties through boosting technological innovation and industrial upgrading [67]. As a consequence, it is worthwhile to discuss whether DF is inclusive or repellent [65].

This study demonstrates that tourism development has a negative moderating effect on DF’s contribution to balanced development between counties. This challenges the finding that tourism growth can achieve coordinated development among regions [68,69]. This can be attributed to the expansion of digital divides due to tourism growth, thus increasing digital exclusion and development gaps. Especially, the digital divide in the tourism industry can diffuse its positive effect on balanced development between counties [70]. Moreover, excessive reliance on the tourism economy may hinder the prosperity of other industries or sectors, which narrows the usage coverage of DF and expands the development gap between counties [71]. In addition, DF flows to the counties with a higher level of tourism development which would further widen the development gap between counties.

Meanwhile, our research underscores that educational level negatively moderates the association between DF and balanced development between counties, which is consistent with quite a few studies demonstrating a positive relationship between educational level and digital divide [72]. The possible reason is that the gap of educational level among various groups in less-developed counties aggravates the digital divide, thereby expanding the development disparity compared to relatively developed counties [66]. Meanwhile, primary education struggles to adequately boost digital literacy and narrow the digital divide between developed counties and under-developed counties [73]. Especially, the expansion of usage depth in terms of DF in counties with a higher educational level can contribute to economic growth and industrial transformation. Over time, the development gap will continue to grow [38].

7. Conclusions

7.1. Main Conclusions

Whether DF can facilitate balanced development between counties is a crucial research topic in the digital age. This study delves into the effect of DF on balanced development between counties and, more importantly, the moderating roles of tourism development and educational level in the relationship between the two. Our research overcomes some limitations of the extant literature that only explores the association between DF and urban–rural income gaps. Also, this paper can further complement the research content on the spillover effect of DF on social sustainability through adding moderating variables. We have several conclusions.

First, DF has a non-linear effect on balanced development between counties in the WMA. Because of market failure, the polarization effect can extend the development gap between counties. With the improvement of government support, the promotion effect of DF on balanced development between counties gradually becomes positive and stronger.

Second, tourism development cannot promote the positive effect of digital finance on balanced development between counties. Given the differences in resource endowment, counties with a higher level of tourism development can attract more digital resources, thus aggravating the development gap between developed counties and under-developed counties.

Third, the improvement of educational level cannot expand the positive effect of DF on balanced development between counties. Given the fact that primary education takes the lead in the WMA, the progress of educational level cannot adequately support the development of DF. Therefore, the moderating effect of the education level is not significant.

7.2. Theoretical Implications

First, our paper breaks through the literature on the effect of DF on urban–rural income disparity, adopts an unbalanced index, and delves into the impact of DF on balanced development between counties. Our study complements the literature on the association between DF and SDGs, especially SDG 10. Moreover, it advances the understanding of the positive externalities from DF in the digital era.

Second, we further investigate the non-linear effect of DF on balanced development between counties, which provides a comprehensive understanding of DF and coordinated development between areas. More importantly, it is helpful to offer references for governments to shrink the development gap through improving political leadership.

Third, this study introduces moderating variables such as tourism development as well as educational level and empirically reveals the moderating role of these variables in the DF–balanced development between counties nexus. On the one hand, our research contributes to broadening the research framework of DF and social fairness, especially coordinated development between areas. On the other hand, it is helpful to enrich the application scope of the TLGH in under-developed areas. Furthermore, our study boosts confidence in the execution of policies regarding coordinated development between areas.

7.3. Policy Implications

First, DF can promote balanced development between counties in the WMA; therefore, governments at all levels should increase the fiscal input of DF, which will contribute to the upgrading of industrial structure as well as technological innovation. Especially, the WMA, suffering from a lack of financial resources and digital infrastructures, needs to promote economic growth, improve residents’ wellbeing, and thus achieve high-quality development. Moreover, vicious competition and less communication in terms of the development of DF can also widen the development gap between counties. Thus, the cooperation mechanism of DF should be better constructed in the WMA.

Second, given a dearth of government support, DF may aggravate unbalanced development between counties. As a consequence, governments at the county level need to make policies regarding financial supervision to reduce financial repulsion and risk. Authorities should prevent the inclusiveness and fairness of DF from being destroyed. Additionally, governments can also make full use of digital platforms to reasonably allocate financial resources. Furthermore, the accurate digital services provided by financial technology can enhance the business value in less-developed counties, thereby narrowing development gaps among counties.

Third, tourism development and educational level exert a negative moderating effect on the relationship between DF and balanced development between counties. Thus, more attention needs to be paid to the digital divide brought by the imbalance of educational level and tourism economy in the WMA. Especially, primary education in the WMA cannot adequately support the positive effect of DF on balanced development between counties. Therefore, the popularization of financial knowledge and digital knowledge can be further boosted by governments and social organizations. Moreover, the dearth of cooperation and communication of tourism development not only directly expands the development gap but also affects the flow of DF to developed counties. Thus, tourism enterprises should enhance cooperation regarding resource utilization, management, and marketing to decrease the negative externalities. More importantly, the construction of digital infrastructures driven by tourism development is conducive to making full use of DF, thereby achieving the goal of common prosperity in the WMA.

7.4. Limitations and Future Prospects

Given the limitation of data unavailability, this study only analyzes balanced development at the economic level. However, it is worthwhile to further pay attention to social equality in under-developed areas around the world. Although this paper examines the moderating effect of tourism development and educational level on the association among DF and balanced development between counties, the influencing paths by which DF affects balanced development between counties need to be discussed in the future. Furthermore, as new datasets become available, the establishment of an evaluation system for balanced development, which integrates objective facts and subjective feelings, will be accurate and useful.