Regional Policy and Balanced Development: Spatial Evidence from Inner Mongolia of China

Abstract

Unfavorable initial conditions and limited factor endowments often constrain economic development in border regions, whereas regional development policies can alter locational disadvantages and promote balanced regional development. Based on nighttime light data from 1999 to 2022, this paper employs a spatial regression discontinuity design (spatial RDD) to examine economic differences between border and non-border areas in Inner Mongolia, China, and to assess the effects of regional development policies in underdeveloped regions. The results show that, after controlling for initial endowments and economic characteristics, the Program of Border Areas Revitalization and Poverty Alleviation significantly enhances economic vitality in border regions and generates persistent growth effects; these findings remain robust after excluding potential confounding factors and conducting a series of robustness checks. Heterogeneity analysis indicates that the policy effects are more pronounced in eastern Inner Mongolia, in more densely populated areas, and in regions with higher market vitality. Mechanism analysis suggests that the Program of Border Areas Revitalization and Poverty Alleviation alters locational disadvantages by strengthening the provision of public goods, thereby enhancing economic linkages across regions; the development of characteristic and comparative advantage industries promotes industrial structure upgrading and drives economic growth in border areas. Further analysis finds that the Program of Border Areas Revitalization and Poverty Alleviation reduces, to some extent, intra-regional economic disparities within border areas and promotes sustainable economic development while improving the ecological environment, further indicating that there exists a compatible pathway between regional development and ecological sustainability. Overall, this study provides spatially explicit micro-level evidence on how regional policies can reshape geographical constraints and foster balanced development in underdeveloped border regions.

1. Introduction

How underdeveloped regions can address diverse regional challenges and sustain economic growth has long been a central concern of social scientists and policymakers [1,2,3]. Tackling the problem can help remove institutional barriers to growth and achieve balanced regional development [4].

In the existing literature, scholars often investigate the institutional impact on regional economies through geographical and historical factors. For example, Acemoglu et al. (2001, 2005) argue that geography played a crucial role in shaping early institutions, with long-lasting consequences for national growth [5,6]. Dell et al. (2010) examine Peru’s Mita system and find that, despite its abolition, the legacy of Mita continues to influence resource distribution, political power, and trust [7]. Regional development policies, meanwhile, show heterogeneous effects across contexts: Brazil’s Amazon development plan [8], Italy’s Southern development scheme [9], and Indonesia’s transmigration program [10] all spurred short-term growth but failed to sustain productivity improvements, while also facing limited educational investment and weak policy continuity. In contrast, China’s regional policies emphasize comprehensive systems: the central government formulates strategic plans and sets overarching objectives, while local governments devise concrete action schemes with detailed targets [11]. Such policies have mitigated locational disadvantages and inadequate endowments, enabling economic takeoff [12,13]. Notably, to accelerate border development, China formally launched the Program of Border Areas Revitalization and Poverty Alleviation in 2000, which built on regional endowments and implemented strategic plans in economic growth, income distribution, public goods, characteristic industries, and fiscal-financial support. By promoting tourism, green agriculture and animal husbandry, and cross-border trade, the program has delivered sustained economic growth in border areas.

In underdeveloped regions, public goods provision is often regarded as a priority [14]. Fiscal expenditure improves governance quality and helps achieve development goals [15]. Yet fiscal governance systems vary across countries, and decentralization affects the supply of public goods [16,17]. It results in inconsistent fiscal powers [18], inefficient resource allocation, and cost overruns [19,20,21]. By contrast, Inner Mongolia provides a suitable case: local governments hold comparable decision-making authority in economic, administrative, personnel, and fiscal affairs, enabling clearer identification of policy impacts. Moreover, border and non-border regions share similar natural features and demographic structures and exhibited parallel economic trends before policy implementation, offering a favorable quasi-natural experiment consistent with counterfactual requirements. Using a spatial regression discontinuity approach and building on Henderson et al. (2012) on the elasticity of nighttime lights to income, this paper finds that increased light intensity boosted border GDP growth rates by 1.03%, with declining intra-border disparities and more balanced regional activity [22]. Furthermore, government expenditure and infrastructure construction alleviated financing constraints, improved locational conditions, and enhanced interregional connectivity [23,24]. Robustness checks accounting for cross-regional flows confirm these results.

Theoretically, this study argues that regional policies promote border economic development mainly through the pathway of “public goods provision—development of characteristic industries—improvement of locational conditions.” Fiscal spending expands public goods supply, easing capital constraints while enhancing human capital and factor mobility. Regional policies encourage local governments to leverage comparative advantages and endowments, supporting green agriculture, ecotourism, and renewable energy industries, which drive industrial upgrading and endogenous growth. Infrastructure and cross-border cooperation reduce factor mobility costs, shifting border areas from “locational disadvantage” toward an “open frontier.” Importantly, public goods provision supports industry development, characteristic industries stimulate fiscal revenues, and locational improvements magnify spillover effects. Together, these dynamics form the foundation of border development and provide an institutionalized path toward green development and ecological civilization.

Despite previous research examining the impacts of regional policies [25,26,27], a consensus regarding impoverished regions continues to be unattainable. The majority of research has concentrated on developed regions [28,29,30,31,32], neglecting border economies. Border regions are frequently seen as unique functional areas [33], where locational limitations impede factor mobility and restrict growth [34,35]. Notably, underdeveloped regions face the dilemma of balancing industrialization and sustainable development while pursuing economic growth [36]. Meanwhile, to achieve economic development, these regions often accelerate industrialization, which may exacerbate environmental pressures such as carbon footprints. However, enhancing industrial competitiveness and relying on renewable energy are expected to achieve compatibility between emission reduction and development, thereby realizing sustainable development [37]. Furthermore, the majority of research employs a macro perspective, with less micro-level investigation. This paper makes three contributions: first, it introduces a spatial regression discontinuity design to identify micro-level policy effects, thereby enhancing causal inference; second, by examining Inner Mongolia—a representative case regarding endowments, location, and policy support—the study offers micro-level evidence for border development and insights for underdeveloped regions; and third, it reveals the mechanisms by which regional policies modify initial constraints in the provision of public goods, development of characteristic industries, locational enhancements, and mitigation of disparities.

The remainder of this paper is organized as follows: Section 2 introduces the policy and regional background, Section 3 presents data and empirical strategy, Section 4 provides empirical analysis and robustness checks, Section 5 examines mechanisms of policy-driven growth, Section 6 explores intra-regional disparities and sustainability, and Section 7 concludes.

2. Background

2.1. Policy Background of the Program of Border Areas Revitalization and Poverty Alleviation

The Program of Border Areas Revitalization and Poverty Alleviation was designed to address the problems of lagging economic growth and unbalanced development in China’s border regions. Since the launch of reform and opening-up, the southeastern coastal areas have experienced rapid economic development, whereas the border regions have remained relatively underdeveloped [38,39]. To stimulate development in these regions, China officially launched the Program of Border Areas Revitalization and Poverty Alleviation in 2000.

As a supporting project of the Western Development Strategy, the initiative aimed to narrow regional development gaps and achieve the dual strategic goals of “revitalizing border areas” and “enriching residents.” The initiative unfolded in three main phases. The first was the pilot exploration stage (2000–2008), during which the policy coverage expanded from the initial nine counties (including banners and districts, hereafter collectively referred to as counties) to all border counties. Fiscal transfers and special funds were directed toward infrastructure construction in pilot areas, addressing deficiencies in initial endowments. The second was the institutional consolidation stage (2009–2015), when the program was fully implemented. In addition to strengthening infrastructure, the focus shifted toward developing characteristic industries, with successive “Five-Year Plans” increasing support. Transfers, poverty alleviation funds, and infrastructure investment provided essential material capital for economic takeoff in border (The central government has provided direct financial support to border regions through channels such as fiscal transfers and earmarked funds. For instance, since the launch of the 14th Five-Year Plan, the central government has allocated a cumulative RMB 25.2 billion in rural revitalization subsidies to border provinces and autonomous regions, with a particular focus on promoting the development of characteristic industries such as specialized animal husbandry and handicrafts. In 2025, the central government continued to prioritize border areas in fiscal allocations; for example, Alxa League received a special subsidy of RMB 30.9 million to support rural construction and innovation in handicraft industries). The third phase, quality and efficiency enhancement (2016–present), marked a shift from a “blood transfusion” approach to one that fosters “self-sustaining” endogenous growth. Governance logic moved from filling shortfalls to strengthening growth capacity. Meanwhile, the initiative granted greater administrative authority to local governments, while central authorities monitored local decision-making through performance evaluations, gradually advancing the modernization of governance systems and capacity.

Although initially implemented on a pilot basis, the effects of the initiative extended far beyond pilot areas, generating significant spillover and demonstration impacts across the entire border region. Within China’s vertical performance evaluation system, economic growth is a key metric in assessing local officials’ performance and directly influences promotion prospects. This competitive environment strengthened local governments’ incentives to promote economic development [40]. Furthermore, improved infrastructure and expanding market scale accelerated factor mobility [41]. Pilot areas also offered successful experiences in developing characteristic industries and cross-border trade, which encouraged imitation and policy learning in other border counties, thereby producing positive demonstration effects.

2.2. Location Conditions of Border Areas in Inner Mongolia

Conventional economic theory suggests that disadvantaged geographic conditions constrain factor mobility and economic activity, often trapping regions in a low-growth equilibrium and hindering economic takeoff [42]. However, the Revitalizing Border Areas and Enriching Residents initiative has spurred economic growth in Inner Mongolia’s border regions by enhancing infrastructure construction and fostering characteristic industries, thereby strengthening endogenous development dynamics and helping these areas escape the low-growth trap. Among these measures, transportation infrastructure has been particularly critical in reshaping locational conditions. Improved transportation networks reduced logistics costs, accelerated factor flows, and facilitated optimal resource allocation, transforming internal locational disadvantages. Moreover, Inner Mongolia’s border areas have increasingly become China’s frontier for opening up to the north. Growing cross-border cooperation and trade with Russia and Mongolia have strengthened their role as multifunctional gateways to the outside world [43], further improving external locational conditions.

The transformation of locational conditions has lowered market entry barriers—particularly for small and medium-sized enterprises—and created a more favorable business environment for regional development. On the one hand, these areas have leveraged natural resource endowments to develop characteristic industries such as green animal husbandry and border tourism, with emerging industries becoming new engines of economic growth. On the other hand, border port trade has strengthened connectivity with neighboring countries. The establishment of cross-border trade zones and industrial demonstration parks has further enhanced economic cooperation and commercial exchanges. For instance, cross-border economic activities in port cities such as Manzhouli and Erenhot have grown significantly, establishing new growth poles for regional development [44].

2.3. Theoretical Analysis

From the perspective of regional economic development theory, underdeveloped regions are constrained by unfavorable conditions such as insufficient initial endowments and locational disadvantages, making it difficult to achieve economic take-off. Regional development policies can provide inclusive institutions for underdeveloped regions and promote balanced regional development. In particular, in terms of public goods supply and the development of characteristic and comparative advantage industries, they can help underdeveloped regions achieve corner overtaking, break away from initial unfavorable conditions, and narrow regional development gaps. Such gaps exist not only between border regions and non-border regions but also within border regions themselves.

On the one hand, according to public goods theory, public goods such as infrastructure and public services are the core prerequisites for regional economic development. The Program of Border Areas Revitalization and Poverty Alleviation can form innovative production factors, promote the free flow and efficient combination of production factors, and realize the full allocation of resources. In addition, regional development policies can strengthen the construction of infrastructure such as transportation and energy, reduce the cost of inter-regional economic linkages, and break away from locational disadvantages, thereby enhancing economic vitality and marketization level and narrowing regional development gaps. On the other hand, based on comparative advantage theory, developing characteristic and comparative advantage industries is an important path for underdeveloped regions to achieve economic catch-up [45,46]. The Program of Border Areas Revitalization and Poverty Alleviation can provide policy support and resource inclinations for border regions, fully tap into resource endowment advantages, cultivate characteristic industries (such as tourism and modern agriculture), promote the optimization and upgrading of industrial structure, and further activate economic growth momentum, thus realizing economic growth. See Figure 1.

Based on the above analysis, this study proposes the following two research hypotheses:

Hypothesis 1:

The “Action to Prosper Border Areas and Enrich the People” has a significant positive impact on the economic development of border regions.

Hypothesis 2:

The “Action to Prosper Border Areas and Enrich the People” narrows regional development gaps by strengthening public goods supply and promoting the development of characteristic and comparative advantage industries.

3. Data and Empirical Strategy

3.1. Data

This study primarily relies on the following datasets.

First, nighttime light remote sensing data are used to measure economic activity in border areas. Since township-level economic growth data in China are not publicly available, we employ DMSP-OLS nighttime light data, with reference to the study of Wu et al. (2021) [47]. Specifically, the DMSP-OLS data for the 1999–2013 period were calibrated using a quadratic model anchored in the “pseudo-invariant pixel” method before being fitted with the 2022 administrative boundaries of Inner Mongolia using ArcGIS 10.8. For the purpose of ensuring data completeness and consistency across the entire study period, we further applied an exponential smoothing model to predict and fill in missing values for the monthly SNPP-VIIRS data covering 2013–2022 and conducted a rigorous screening process to eliminate outliers and noise from the annual aggregated data. On this basis, a sigmoid model was employed to produce simulated DMSP-OLS (SDMSP-OLS) data for the 2013–2022 period, which was then merged with the calibrated 1999–2013 DMSP-OLS data. After projection adjustment and continuity correction for the integrated dataset, we finally obtained annual light intensity values for the full 1999–2022 timeframe. Following Henderson et al. (2012) [22], we use the elasticity between nighttime light growth and GDP growth (1% corresponding to 0.3%) to measure the relationship between light intensity and economic performance. Given that this coefficient is based on global samples and may not be directly applicable to Inner Mongolia, we further estimate a localized conversion coefficient by regressing prefecture-level GDP data against light intensity and compare the results with Henderson et al.’s benchmark. The two estimates are highly consistent, indicating that our conclusions do not rely on a specific elasticity assumption. In addition, we calculate the Theil index of nighttime light to capture intra-border regional disparities.

Second, data on public goods and distinctive industries are used to examine locational and endowment conditions in border areas. Fiscal expenditure, serving as the financial source for public goods provision, is measured using the share of fiscal spending, with data obtained from the China County and Township Statistical Yearbook (county-level volume). Transportation infrastructure, a crucial public good for economic development in border areas, is proxied by the mileage of township roads, county roads, and total road length within the region. These data are drawn from the global open-source mapping project OpenStreetMap (OSM) and matched with Inner Mongolia’s administrative boundaries using ArcGIS 10.8. (The cross-sectional data used in this study are from the year 2022, due to the unavailability of data prior to the policy implementation. Although this limitation prevents an examination of the full dynamic effects of the policy, it still allows for an assessment of its prospective impacts). We also incorporate indicators of public services, including the number of schools for basic and adult education, as well as the number of hospitals, clinics, and pharmacies. These data are collected from Baidu Maps and Gaode Maps. To assess the development of distinctive industries, we use the sectoral shares of primary and tertiary industries, also from the China County and Township Statistical Yearbook (county-level volume). Furthermore, we employ 2022 Points of Interest (POI) data to capture industry development, including the number of industrial parks, homestays, and travel agencies as proxies for modern industries and border tourism. Agricultural development is measured using the China Agricultural Cropland Dataset (CACD), a 30-m resolution dataset generated on the Google Earth Engine cloud computing platform.

Third, ecological and environmental indicators are employed to assess environmental protection and green development in border areas, which are also core tasks of the Revitalizing Border Areas and Enriching the People Initiative. Specifically, we use the degree of desertification, vegetation coverage, and PM2.5 concentration to represent land productivity, vegetation cover, and air quality, respectively. Data on desertification and vegetation coverage are mainly derived from MODIS (Moderate Resolution Imaging Spectroradiometer), with desertification indicators constructed through information overlay, classification, and discrimination analysis. PM2.5 concentration data come from global gridded datasets provided by Washington University in St. Louis. These environmental data are integrated with Inner Mongolia’s administrative boundaries using ArcGIS 10.8.

Fourth, geographic variables and population density are included as control variables. Geographic variables include slope, aspect, and elevation data, sourced from high-resolution topographic surveys jointly conducted by the National Geospatial-Intelligence Agency (NGA) and the National Imagery and Mapping Agency (NIMA). These indicators reflect the natural geographic features of the region. Population distribution data are derived from the LandScan program initiated by Oak Ridge National Laboratory (ORNL). We use 30-m resolution gridded population density data from 2000 to 2022 to capture the spatial distribution of population.

After excluding observations with missing values, the total sample consists of 1022 observations. Within a bandwidth of 50 km, the sample size is 323, and within a bandwidth of 30 km, the sample size is 214. Descriptive statistics are reported in

Table 1. Summary statistics.

Variable	Obs	Mean	Std. Dev
Panel A: town-level variables
Nighttime light mean in 1999	1022	5.30	13.97
Nighttime light sum in 1999	1022	121.48	241.99
Nighttime light mean in 2022	1022	12.63	17.84
Nighttime light sum in 2022	1022	2102.97	2140.52
Light intensity growth from 1999–2022	1022	1.26	0.70
Light sum growth from 1999 to 2022	1022	4.65	2.72
Border	1022	0.17	0.38
Longitude	1022	115.41	5.63
Latitude	1022	43.14	3.17
Mean_high	1022	902.96	446.44
Slope	1022	3.29	3.07
Aspect	1022	172.40	19.80
Road_all	1022	80.73	90.45
Panel B: City-level variables
Fiscal Expenditure Proportion	1932	0.22	0.12
Fiscal pressure	1932	0.17	0.12
Proportion of primary industry	1932	0.23	0.14
Proportion of tertiary industry	1932	0.34	0.12

Notes: This table gives the summary statistics of main variables.

.

3.2. Empirical Strategy

Given the comparable geographic and socio-economic characteristics of Inner Mongolia’s border and non-border regions, we adopt a regression discontinuity (RD) design to analyze economic activities and development disparities. Following Jia et al. (2021) [48], the model is specified as follows:

$${\mathit{Y}}_{\mathit{i}}={\mathit{\beta }}_0+{\mathit{\beta }}_1{\mathit{Border}}_{\mathit{i}}+\mathit{f}\left(\mathit{geographic} {\mathit{location}}_{\mathit{i}}\right)+{\mathit{X}}_{\mathit{i}}+{\mathit{ϵ}}_{\mathit{i}}$$

(1)

Here, ${\mathit{Y}}_{\mathit{i}}$ denotes the outcome variable for the administrative unit $\mathit{i}$, measured as the growth rate of nighttime light intensity. Since some areas report zero light intensity, we log-transform the light value after adding 1:

$${\mathit{Y}}_{\mathit{i}}\equiv \mathit{ln}\left(1+{\mathit{LightIntensity}}_{\mathit{i},\mathbf{2022}}\right)-\mathit{ln}\left(1+{\mathit{LightIntensity}}_{\mathit{i},\mathbf{1999}}\right)$$

(2)

Equation (2) measures the cumulative growth rate of nighttime light intensity of the i-th administrative unit over the study period, with its design rationale and variable definitions specified as follows:

Zero-value adjustment: We add 1 to the light intensity values prior to logarithmic transformation to avoid the undefined issue of logarithmic operation on zero light intensity values. This treatment effectively mitigates heteroscedasticity and renders the estimated coefficients interpretable.

Variable definitions: ${\mathit{LightIntensity}}_{\mathit{i},\mathbf{2022}}$ and ${\mathit{LightIntensity}}_{\mathit{i},\mathbf{1999}}$ represent the raw nighttime light intensity values of the i-th administrative unit in 2022 (the terminal year of the study period) and 1999 (the base year of the study period, prior to the full implementation of the Program of Border Areas Revitalization and Poverty Alleviation), respectively. The difference between the two logarithmic terms directly reflects the long-term growth trend of regional economic activities proxied by nighttime light intensity.

Linkage to the identification strategy: The ${\mathit{Y}}_{\mathit{i}}$ constructed by Equation (2) serves as the outcome variable in Equation (1), acting as a valid measure of regional economic growth. By focusing on the cumulative growth rate between the pre-policy base year and the terminal year, we eliminate the interference of short-term economic fluctuations and thus accurately capture the long-term impact of the policy.

${\mathit{Border}}_{\mathit{i}}$ is a binary indicator for townships, where 1 indicates a border region and 0 otherwise. ${\mathit{\beta }}_1$ captures the economic effect of the Prosperous Border Program. The function $\mathit{f}\left(\mathit{geographic} {\mathit{location}}_{\mathit{i}}\right)$ controls for spatial factors via a bivariate polynomial of longitude and latitude. We compute the geographic centroid of each unit and its distance to the border boundary (our running variable). A positive value indicates distance from the border inside border townships, and a negative value indicates distance from non-border townships. ${\mathit{X}}_{\mathit{i}}$ includes control variables such as slope, aspect, elevation, and population density, while ${\mathit{ϵ}}_{\mathit{i}}$ is the error term. We use robust standard errors to mitigate the influence of random shocks.

Equation (1) can be estimated using two approaches: the local linear method and the global polynomial method. This paper primarily employs local linear regression, applying a narrow bandwidth around the cutoff and controlling for longitude–latitude polynomials and other covariates, while using global polynomial regression as a robustness check.

Given the absence of a universally optimal bandwidth [49], the bandwidth is set at 30 km for township-level analysis and 50 km for county-level analysis. The rationale is threefold: first, these ranges broadly correspond to the actual spatial extent of regional economic activities. Second, overly narrow bandwidths may lead to insufficient sample size and overly sensitive estimates, while excessively wide bandwidths risk incorporating confounding factors that dilute policy effects. The chosen 30 km and 50 km settings strike a balance between “locality” and “sample adequacy,” thereby capturing border policy effects more accurately. Third, further supported as well as robustness checks under various kernel functions, the results remain consistent across bandwidths ranging from 20 to 80 km, confirming the validity of the bandwidth choice. To further ensure reliability, this study also reports results across different bandwidths, which reflect the spatial trend of policy impacts, i.e., the way policy effects vary with distance from the border.

3.3. Balance Tests

The validity of regression results depends on discontinuity in the outcome variable and continuity in the covariates. If nighttime light intensity shows a discontinuity between border and non-border regions, it indicates spatial differences in economic activity; otherwise, no such differences exist. Hahn et al. (2001) argue that including covariates does not affect the validity of regression discontinuity estimates [50]. To this end, a bandwidth of 100 km is adopted, applying local linear regression with a triangular kernel to test for continuity and discontinuity in both the dependent and control variables. This approach also helps mitigate potential manipulation of the cutoff. Figure 2 presents the first- to fourth-order nighttime light growth rates, showing a distinct jump at the cutoff. This indicates significant differences in nighttime light intensity between border and non-border regions, thereby satisfying the requirement that the outcome variable exhibits a discontinuity at the threshold.

In

Table 2. Controls of the RD results.

Sample Within	Dependent Variable: Control Variables
	Mean_high	Aspect	Slope	ln_sum_pol
	(1)	(2)	(3)	(4)
Border	−78.541	−1.625	1.071	−0.549
	(177.841)	(9.099)	(0.759)	(0.696)
Distance	<30 km	<30 km	<30 km	<30 km
Controls	Yes	Yes	Yes	Yes
Observations	214	214	214	214

Notes: All regressions include two-dimensional geographic controls. The county-level clustered standard errors are reported in parentheses.

, elevation, slope, aspect, and population density are regressed as dependent variables. Results reveal no significant discontinuities, suggesting that natural and demographic characteristics remain smooth across the border. These variables are further included in subsequent regressions, and the coefficients remain stable, confirming that geographic and demographic factors are not the main drivers of economic growth in border regions.

4. Main Results

4.1. Baseline Regression

Table 3. Baseline RD results.

Sample Within	Dependent Variable: Light Intensity Growth from 1999 to 2022
	Local Linear		Local Quadratic		Local Linear		Local Linear
	<30 km	<50 km	<30 km	<50 km	<30 km	<50 km	<30 km	<50 km
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Border	0.768 ***	0.440 ***	0.778 ***	0.442 ***	0.741 ***	0.459 ***	1.343 ***	1.128 ***
	(0.215)	(0.165)	(0.211)	(0.163)	(0.210)	(0.156)	(0.342)	(0.264)
Polynomial	linear	linear	quadratic	quadratic	linear	linear	linear	linear
Controls	No	No	Yes	Yes	Yes	Yes	Yes	Yes
Observations	214	323	214	323	214	323	214	323

Notes: The dependent variable in (1) to (6) is $\ln \left(1+{\mathit{LightIntensity}}_{i,2022}\right)-\ln \left(1+{\mathit{LightIntensity}}_{i,1999}\right)$. The dependent variable in (7) to (8) is $\ln \left(1+{\mathit{Lightsum}}_{i,2022}\right)-\ln \left(1+{\mathit{Lightsum}}_{i,1999}\right)$. All regressions include two-dimensional geographic controls. The county-level clustered standard errors are reported in parentheses. *** indicate statistical significance at 1% levels, respectively.

reports the regression results for economic growth in the border areas of Inner Mongolia. Column (1) shows that the growth rate of nighttime light intensity in the border areas increased by approximately 76.8%. According to Henderson et al. (2012) [22], a 1% increase in nighttime light intensity corresponds to roughly a 0.3% increase in GDP. Based on this estimation method, we find that for every 1% increase in the elasticity of nighttime light intensity, the GDP of county-level cities in Inner Mongolia increases by approximately 0.38%, which is slightly higher than the estimate of Henderson et al. (2012) [22]. Comparatively, the annual GDP growth rate in the border areas rose by 0.84%, indicating a substantial economic growth effect for underdeveloped regions with relatively backward infrastructure and limited factor endowments. Column (2) sets the bandwidth at 50 km, further supporting this conclusion. To address potential endogeneity, Columns (3) and (4) include quadratic polynomials of longitude and latitude. Columns (5) and (6) use first-order polynomials of longitude and latitude, adding population density, elevation, slope, and aspect as control variables. Across all specifications, the regression coefficients are positive and statistically significant at the 1% level, with no substantial changes, indicating that geographic and demographic characteristics are sufficiently smooth at the cutoff and have minimal impact on the results. For additional robustness checks, columns (7) and (8) replace the dependent variable with total nighttime light intensity, and the results remain positive and significant at the 1% level. Overall, these findings consistently demonstrate that the Action for Prosperity and Enrichment along the Border has effectively promoted economic growth in the border areas. The conclusion is consistent with the findings of Zhang et al. (2023) [51].

4.2. Robustness Checks

The selection of different bandwidths and estimation methods may affect the estimation results of this study. An overly narrow bandwidth leads to excessively sensitive results, while an overly wide bandwidth dilutes local characteristics. The existing literature generally adopts 30 km and 50 km as standard bandwidths, but such selections are somewhat subjective and non-random, which may compromise the reliability of the conclusions to a certain extent. To address this problem, this paper sets the bandwidth range from 20 km to 80 km in Figure 3 and constructs a multi-method estimation framework to conduct rigorous sensitivity tests. Specifically, Panel A of Figure 3 incorporates a second-order polynomial of longitude and latitude into the regression to control for spatial heterogeneity; Panels B to D report the estimation results using three methods, namely the Triangular kernel, Epanechnikov kernel, and Uniform kernel, respectively.

As shown in Figure 3, across the entire bandwidth range of 20–80 km, regardless of the estimation method adopted, the regression coefficients of the economic effects of the Program of Border Areas Revitalization and Poverty Alleviation remain consistent in sign, significance level, and magnitude, with no obvious fluctuations. This indicates that the significant promoting effect of the Program of Border Areas Revitalization and Poverty Alleviation on the economic growth of border areas does not depend on the selection of a specific bandwidth or estimation strategy. In other words, the policy effect identified in this study is not a spurious result caused by subjective bandwidth selection or methodological bias, but a consistent and reliable causal relationship. Therefore, the analysis in Figure 3 effectively rules out the interference of bandwidth selection and estimation methods on the research results, further verifying the robustness of the main conclusions.

4.3. Placebo Tests

Although the smoothness of the predetermined variables at the cutoff has been verified above, potential confounding factors may still affect the estimation results. To further examine the robustness of the regression findings, this study artificially shifted the boundary between border and non-border areas. Specifically, the boundary was moved 30 km and 50 km to the south and north, respectively. The regression results in

Table 4. Placebo test for cutoff point shift.

Sample Within	Dependent Variable: Light Intensity Growth from 1999 to 2022
	Move the True Boundary 30 km		Move the True Boundary 50 km
	<30 km	<30 km	<50 km	<50 km
	(1)	(2)	(3)	(4)
Border	0.205	0.036	0.037	−0.093
	(0.190)	(0.206)	(0.194)	(0.302)
Polynomial	linear	linear	linear	linear
Direction	South	North	South	North
Controls	Yes	Yes	Yes	Yes
Observations	214	214	323	323

Notes: The dependent variable is $\ln \left(1+{\mathit{LightIntensity}}_{i,2022}\right)-\ln \left(1+{\mathit{LightIntensity}}_{i,1999}\right)$. We set 30 km and 50 km bandwidth. The county-level clustered standard errors are reported in parentheses.

indicate that, regardless of the direction of the boundary shift, the coefficients for the border areas are statistically insignificant. There is no discernible difference in economic growth trends on either side of the artificially set boundaries, suggesting that the economic growth in border areas is not driven by chance. In other words, factors outside the policy are not the primary drivers of economic development in these areas.

4.4. Sample Selection Sensitivity

The rationality of sample selection must take into account the impacts of non-random sample missing and administrative division adjustments; otherwise, it will lead to selection bias. On the one hand, data statistics are missing in some regions, which may result in incomplete sample information. On the other hand, some regions underwent administrative division adjustments (such as county mergers and administrative boundary changes) during the research period, leading to inconsistent matching of samples in adjacent areas and thus affecting the accuracy of estimation results. To effectively mitigate the above problems, this paper conducted unified calibration of administrative divisions and excluded regions with frequent administrative division adjustments. In addition, the manipulation of samples near the regression discontinuity is likely to cause selection bias and reduce the reliability of regression results. For this reason, this paper conducted sensitivity tests by gradually excluding samples near the boundary line. Specifically, we sequentially excluded 2%, 4%, 6%, 8%, and 10% of the samples adjacent to the boundary line and visualized the regression coefficients in Figure 4 to demonstrate the correlation characteristics between the proportion of sample exclusion and estimation results.

As shown in Figure 4, after excluding the samples with the above proportions, the regression results remained significant at the 5% confidence level, and the coefficient values did not experience drastic fluctuations or abnormal changes. This indicates that there was no manipulation of the samples near the discontinuity, thus ruling out the interference of selection bias on the estimation results. Furthermore, it proves that the promoting effect of the Program of Border Areas Revitalization and Poverty Alleviation on border areas does not depend on changes in sample size, and the policy effect is highly stable. This sensitivity analysis further verifies the reliability of the research conclusions.

4.5. Heterogeneity Analysis

The Program of Border Areas Revitalization and Poverty Alleviation exhibits heterogeneity in its effects across regions, population density, and market vitality. Therefore, this study conducts subgroup regressions for the eastern and western parts. Additionally, to mitigate potential endogeneity issues, we divide population density and market vitality (proxied by the number of newly registered enterprises) into high-value and low-value groups using the median as a cutoff, strictly based on the pre-policy (1999) data of these two indicators, so as to examine the heterogeneity in population mobility and market environment, as reported in

Table 5. Heterogeneous effects analysis.

Sample Within	Dependent Variable: Light Intensity Growth from 1999 to 2022
	<30 km	<30 km	<30 km	<30 km	<30 km	<30 km
	(1)	(2)	(3)	(4)	(5)	(6)
Border	0.478 **	1.110 ***	1.335 ***	0.039	0.523 **	3.001 ***
	(0.234)	(0.337)	(0.322)	(0.219)	(0.205)	(0.446)
Polynomial	linear	linear	linear	linear	linear	linear
Direction	West	East	High_pop	Low_pop	High_com	Low_com
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	117	97	112	102	161	53

Notes: The east boundary includes Hulunbeier, Xinganmeng, Tongliao, Chifeng, and Xilinguole; the west boundary includes other areas. Columns (3) and (4) do not include population density as a control variable. The market vitality data is sourced from Tianyancha. The county-level clustered standard errors are reported in parentheses. **, and*** indicate statistical significance at 5% and 1% levels, respectively.

.

First, the study considers the impact of spatial differences on policy effects. The eastern region of Inner Mongolia has a higher share of the primary and tertiary sectors, with its comparative advantages concentrated in animal husbandry, agricultural processing, and tourism. In contrast, the western region relies on abundant coal and mineral resources, with a higher share of heavy industry. The urban agglomerations centered around Hohhot, Baotou, and Ordos have a significant driving effect on the development of the western region. Columns 1 and 2 show that the regression coefficients are positive and significant in both regions, indicating that the program has a favorable impact. However, a further comparison reveals that the annual growth rate of nighttime lights in the eastern region is approximately 2.56% higher than in the western region. This suggests that the western region, with relatively advanced infrastructure, human capital, and industrial clusters, exhibits diminishing marginal returns to program investment, whereas the eastern region has more “development gaps,” making program inputs more effective in alleviating growth constraints. Furthermore, the eastern region aligns better with the program’s development philosophy of “building ecological civilization and promoting green development,” while the western region, constrained by industrial structure and resource dependence, requires investment in new infrastructure and green industrial transformation. Accordingly, the Program of Border Areas Revitalization and Poverty Alleviation should emphasize place-based approaches: promoting green industrial upgrading in the eastern region and strengthening new infrastructure and industrial optimization in the western region.

Second, the effect of population density on program outcomes is examined. Previous analysis indicated that there is no local discontinuity effect between border and non-border areas, but very low population density can weaken agglomeration of economic activities and firm innovation capacity [52]. Columns 3 and 4 show that the program is more effective in regions with higher population density. This implies that population size is not only a necessary condition for labor supply but also determines the efficiency of public goods provision and the scale of market demand, amplifying the program’s external effects. In sparsely populated areas, insufficient labor supply hinders the formation of human capital and efficient division of labor, limiting program effectiveness. Therefore, the Program of Border Areas Revitalization and Poverty Alleviation should ensure comprehensive public services and social security in sparsely populated regions to improve the foundational conditions for long-term development.

Finally, the impact of market vitality is assessed. Regions with low market vitality suffer from limited factor mobility and an unfavorable business environment, making industrial upgrading more dependent on program support. Using the number of newly registered enterprises at the county level as a proxy for market vitality, Columns 5 and 6 show that the program promotes economic growth in both low- and high-market-vitality areas, but its effect is stronger in regions with lower market vitality. By improving the business environment and reducing institutional transaction costs, the Program of Border Areas Revitalization and Poverty Alleviation enhances the endogenous momentum of economic development. The program not only fosters overall growth but also emphasizes support for disadvantaged areas, reflecting the inclusive nature of regional coordinated development.

4.6. Controlling for Confounding Policies

Although the series of robustness checks presented above demonstrate the effectiveness of the Program of Border Areas Revitalization and Poverty Alleviation, other regional policies may potentially confound the regression results. To address this, the study examines major policies related to the program, removes the relevant samples, and re-estimates the regressions. The empirical results reported in

Table 6. Eliminate the impact of confusing policies.

Sample Within	Dependent Variable: Light Intensity Growth from 1999 to 2022
	Local Linear		Local Linear		Local Linear
	<30 km	<30 km	<30 km	<30 km	<30 km	<30 km
	(1)	(2)	(3)	(4)	(5)	(6)
Border	0.818 ***	0.563 ***	0.967 ***	0.922 ***	0.891 ***	0.729 ***
	(0.236)	(0.214)	(0.243)	(0.227)	(0.210)	(0.221)
Polynomial	linear	linear	linear	linear	linear	linear
Excluding strategy	The Belt and Road	Rural revitalization	Fight against poverty	Ecological function area	City clusters	Coal producing area
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	180	182	161	135	193	193

Notes: The dependent variable is $ln\left(1+{ \mathit{LightIntensity}}_{i,2022}\right)-ln\left(1+ { \mathit{LightIntensity}}_{i,1999}\right)$. *** indicate statistical significance at 1% levels, respectively.

show that, after excluding the influence of these policy shocks, the regression coefficients remain positive and statistically significant at the 1% level, confirming the robustness of the results.

Impact of the Belt and Road Initiative (BRI): Inner Mongolia’s border regions are adjacent to Russia and Mongolia. The BRI may affect economic outcomes through cross-border trade, tourism, and economic cooperation zones, potentially confounding the effects of the Program of Border Areas Revitalization and Poverty Alleviation. To address this, the study excludes samples from border port cities and re-estimates the regression. Column (1) shows that the regression coefficient remains positive and significant at the 1% level, indicating a clear policy effect of the Program of Border Areas Revitalization and Poverty Alleviation.

Impact of the Rural Revitalization Strategy: The Rural Revitalization Strategy aims to promote agricultural and rural modernization, which partially overlaps with the strategic objectives of the Program of Border Areas Revitalization and Poverty Alleviation. After excluding the key implementation areas of the Rural Revitalization Strategy, Column (2) shows that the regression coefficient remains positive and significant at the 1% level, indicating that the strategy does not confound the estimated effects.

Impact of the Poverty Alleviation Strategy: The Poverty Alleviation Strategy targets the production and living difficulties in impoverished areas, playing a critical role in promoting economic development. To prevent exogenous shocks from this strategy, the study excludes key poverty alleviation areas. Column (3) indicates that the regression coefficient for border regions remains positive and significant at the 1% level, confirming that the strategy does not bias the estimation results.

Impact of Ecological Civilization Demonstration Zones: These zones are subject to environmental regulations on energy conservation and emissions reduction, which may limit resource-based and heavy industries. At the same time, these zones benefit from ecological compensation and fiscal transfers, promoting green and low-carbon industries. After removing samples from these zones, Column (4) shows that the regression coefficient remains positive and significant at the 1% level, suggesting that the policy effect of the Program of Border Areas Revitalization and Poverty Alleviation remains robust.

Impact of National Urban Agglomerations: Urban agglomerations centered around Hohhot, Baotou, and Ordos benefit from national-level strategic support, and the resulting polarization effect can reshape spatial development patterns. To account for this, the study excludes samples from national urban agglomerations. Column (5) indicates that the program’s policy effect is unaffected.

Impact of Coal-Dominated Regions: The industrial structure in coal-dominated areas primarily relies on traditional industries such as coal chemical and power sectors, and economic growth is more resource-driven than policy-driven. After excluding coal-dominated regions, Column (6) shows that the regression coefficient remains positive and significant at the 1% level, indicating that the policy effect of the Program of Border Areas Revitalization and Poverty Alleviation is not influenced by economic activities in coal-producing areas.

5. Mechanism Analysis

5.1. Public Goods

Public goods are relatively scarce in border regions and require priority in supply. Transportation infrastructure, as a critical type of public good, serves as a material foundation for interregional connectivity and improves the mobility and allocation efficiency of production factors [53]. In this study, transportation infrastructure is used as a proxy for public goods, and the total road mileage, township road mileage, and county road mileage are log-transformed for regression analysis. As shown in

Table 7. Public goods RD results.

Sample Within	Road_All	Village_Road	City_Road	Hospital	School
	<30 km	<30 km	<30 km	<30 km	<30 km
	(1)	(2)	(3)	(4)	(5)
Border	0.920 **	1.488 ***	1.966 ***	0.516	0.652
	(0.420)	(0.441)	(0.517)	(0.482)	(0.434)
Polynomial	linear	linear	linear	linear	linear
Controls	Yes	Yes	Yes	Yes	Yes
Observations	214	214	214	214	214

Notes: All regressions include two-dimensional geographic controls. The county-level clustered standard errors are reported in parentheses. ** and *** indicate statistical significance at 5%, and 1% levels, respectively.

, columns (1) to (3) show that all coefficients are significantly positive, highlighting the key role of roads in promoting economic growth. This suggests that, under the constraints of fiscal pressure, priority should be given to ensuring transportation infrastructure to sustain economic growth. Further analysis reveals that the coefficients for township roads and county roads are approximately 56.8% and 104.6% higher than that of total road mileage, respectively, indicating that the Program of Border Areas Revitalization and Poverty Alleviation has more substantially promoted the construction of township and county roads, facilitating inter-urban and inter-rural connectivity and cooperation, thereby sustaining economic growth in border areas.

Columns (4) and (5) further examine the effects of public services. Schools and hospitals are used as proxies for public services, log-transformed, and included in the regression. The results, although positive, are not statistically significant. This can be attributed to the equitable and inclusive nature of public services in border regions, which primarily achieve the equalization of access rather than directly driving economic growth, resulting in a limited growth effect.

5.2. Specialized Advantage Industries

The development of specialized advantage industries can reduce excessive reliance on resource-based sectors and, through technological innovation, promote the growth of green agriculture and animal husbandry, ecological tourism, cultural industries, and other distinctive sectors. This contributes to industrial structure optimization and sustains economic growth in border regions.

Table 8. Special advantage industry RD results.

Sample Within	Cultivated Area	Travel Agency	Homestay	Industrial Park
	<30 km	<30 km	<30 km	<30 km
	(1)	(2)	(3)	(4)
Border	0.616	1.918 **	1.568 ***	0.403 *
	(0.413)	(0.962)	(0.513)	(0.207)
Polynomial	linear	linear	linear	linear
Controls	Yes	Yes	Yes	Yes
Observations	183	44	54	38

Notes: All regressions include two-dimensional geographic controls. The county-level clustered standard errors are reported in parentheses. *, **, and *** indicate statistical significance at 10%, 5%, and 1% levels, respectively.

reports the regression results for the development of specialized advantage industries. Inner Mongolia Autonomous Region possesses approximately 170 million mu of arable land, accounting for 9.1% of China’s total, providing a comparative advantage in land resources suitable for large-scale and mechanized modern agriculture rather than traditional agriculture. To further verify this, Column (1) uses arable land area as an alternative proxy for specialized agriculture. Although the coefficient is not statistically significant (p = 0.13), expanding the bandwidth to 50 km renders the coefficient positive and significant at the 10% level, providing some evidence that specialized agriculture contributes to economic growth. Columns (2) and (3) use the number of travel agencies and homestays as proxies for the tourism sector. Travel agencies, as the main operators in the tourism market, accurately reflect the scale of the local tourism industry, while homestays exclude business or official travel, more precisely capturing actual tourism demand. The regression results are significantly positive, indicating that border tourism has become a new growth point, driving ecological tourism and cultural industries, and promoting economic growth in border regions. Finally, column (4) examines the impact of modern industrial parks on economic growth. The results indicate that the Program of Border Areas Revitalization and Poverty Alleviation has effectively promoted the construction of modern industrial parks, advancing industrial structure optimization and further supporting the development of border areas.

6. Further Analysis

6.1. Regional Development Gaps

The Program of Border Areas Revitalization and Poverty Alleviation promotes regional economic growth by increasing public goods provision and developing specialized advantage industries, thereby narrowing the economic gap between border and non-border areas. However, it remains unclear whether regional economic growth has exacerbated internal disparities within border areas, potentially leading to structural imbalances. To address this, this study examines the impact of the Program of Border Areas Revitalization and Poverty Alleviation on regional development gaps.

The Theil index is an important indicator for measuring economic disparities, effectively capturing the spatial distribution of economic activities. Following Zhou et al. (2015) [54], this study calculates the Theil index based on county-level nighttime light intensity, with the results reported in

Table 9. Difference of economic development RD results.

Sample Within	Dependent Variable: Theil from 1992 to 2022
	Local Linear		Local Linear
	<30 km	<50 km	<80 km	All
	(1)	(2)	(3)	(4)
Border	−0.007	−0.025	−0.044	−0.041 *
	(0.138)	(0.059)	(0.049)	(0.024)
Polynomial	linear	linear	linear	linear
Controls	Yes	Yes	Yes	Yes
Observations	16	26	40	83

Notes: The dependent variable is the Theil index. The time dimension in column (1) to column (4) is from 1999 to 2022. We sampled data at 5-year intervals. Considering the sample size, the dependent variable has not been differentiated. All regressions include two-dimensional geographic controls. The county-level clustered standard errors are reported in parentheses. * indicates statistical significance at 1% levels, respectively.

across different bandwidths and for the full sample. Columns (1) to (3) show that the regression coefficients are negative but not statistically significant, providing no evidence that the Program of Border Areas Revitalization and Poverty Alleviation has widened internal disparities within border areas. Further, the full-sample regression results reported in Column (4) indicate a negative coefficient significant at the 10% level. This suggests that while the Program of Border Areas Revitalization and Poverty Alleviation fosters overall regional economic growth, it does not significantly enlarge internal disparities and even shows some convergence trends. These findings align with the policy objective of “development with equity,” implying that border areas have achieved balanced development while sustaining continuous economic growth.

6.2. Sustainable Development

The Inner Mongolia Autonomous Region possesses the largest and most ecologically diverse functional areas in northern China. A healthy ecological environment enhances residents’ quality of life, providing necessary conditions for sustainable development [55]. It also aligns with the objectives of building ecological civilization and promoting green development, which are essential for the green transformation of economic growth.

Table 10. Ecological environment RD results.

Sample Within	Desert	NDVI	PM2.5
	<30 km	<30 km	<30 km
	(1)	(2)	(3)
Border	−1.443 *	1.691 ***	−0.235 **
	(0.759)	(0.362)	(0.101)
Polynomial	linear	linear	linear
Controls	Yes	Yes	Yes
Observations	214	214	183

Notes: Some regions did not provide complete data in column (3), resulting in certain missing values. All regressions include two-dimensional geographic controls. The county-level clustered standard errors are reported in parentheses. *, **, and *** indicate statistical significance at 10%, 5%, and 1% levels, respectively.

reports the empirical results on sustainable development, using desertification degree, vegetation coverage, and fine particulate matter (PM2.5) as proxy variables. Historically, rapid industrialization in Inner Mongolia caused significant resource depletion and ecological degradation, leading to land desertification, grassland deterioration, and pollution. Columns (1) and (2) show that the Program of Border Areas Revitalization and Poverty Alleviation significantly reduced desertification and increased vegetation coverage in border areas, indicating that the policy facilitated economic growth while promoting ecological restoration, reflecting a synergy between economic and environmental benefits. This is consistent with the conclusions of Yan et al. (2024) and Lai et al. (2025) [56,57].

Column (3) uses PM2.5 concentration as an indicator of air quality. Although Inner Mongolia’s energy structure relies heavily on coal and other traditional energy sources, and urbanization may exacerbate air pollution, the empirical results show that the policy significantly reduced PM2.5 concentrations in border areas (negative coefficient is significant at the 1% level). This demonstrates that the Program of Border Areas Revitalization and Poverty Alleviation plays an active role in promoting green industries and strengthening environmental governance. With intensified pollution control measures, PM2.5 concentrations in 2025 are projected to decrease by 7.1% relative to 2020, and days of severe pollution will be kept below 0.5%, directly improving air quality. These outcomes fully reflect the local government’s commitment to advancing ecological civilization and achieving comprehensive green transformation for sustainable development. These findings further corroborate the conclusion presented in Column (4) of Table 6, namely that the policy effect of the Program of Border Areas Revitalization and Poverty Alleviation remains highly significant even after excluding the samples subject to national-level “Ecological Function Area” policies.

7. Conclusions

How underdeveloped regions can overcome initial disadvantages and achieve economic take-off has long been a central issue in development economics and regional policy research. This paper takes the border areas of Inner Mongolia as a case study to examine how the Program of Border Areas Revitalization and Poverty Alleviation promotes economic growth and fosters regional coordinated development through the provision of public goods and the development of specialty industries. The results show that fiscal expenditure alleviates funding constraints in border areas, transportation infrastructure improves locational conditions, and specialty industries make effective use of resource endowments while driving industrial upgrading, thereby mitigating initial disadvantages in economic development. At the same time, the Program of Border Areas Revitalization and Poverty Alleviation narrows internal regional development gaps and improves the ecological environment, achieving a certain degree of balance between equitable development and green development.

The experience of this study indicates that effective regional policies can help underdeveloped areas overcome development bottlenecks and achieve economic takeoff. The policy implications are as follows: First, infrastructure construction should shift from a “scale-oriented” to an “efficiency-oriented” approach, emphasizing inter-county transportation and port connectivity. Second, industrial layout should follow regional comparative advantages, promoting green and ecological industries, advancing low-carbon transformation of resource-based industries in the western regions, and cultivating emerging industries. Third, equalization of public services is crucial for low population-density areas, requiring institutional innovations to improve access to education and healthcare. Fourth, expanding tax incentives and innovation subsidies for small and medium-sized enterprises and reducing entry barriers and transaction costs through institutional innovation can attract more market participants, foster healthy competition and industrial vitality, and stimulate endogenous economic growth. These recommendations are not only supported by empirical findings but also align with China’s strategic goals of high-quality economic development and ecological civilization construction.

This study has several limitations. First, it focuses exclusively on the Inner Mongolia Autonomous Region, and the generalizability of the findings requires validation in other countries and regions. While China’s regional development policies demonstrate high implementation capacity and continuity, Latin American countries such as Brazil face political instability and fiscal constraints, resulting in limited policy effectiveness [58]. Similarly, Indonesia is constrained by governance challenges, including political institutions [59] and official corruption [60,61]. Differences in local administrative capacity and management, insufficient long-term planning, and the absence of clear evaluation standards further complicate policy outcomes. Second, the Program of Border Areas Revitalization and Poverty Alleviation relies heavily on large-scale fiscal investment, raising concerns about the sustainability of long-term fiscal inputs, especially amid rising local government debt risks [62]. Finally, this study primarily focuses on the causal identification of policy effects, with less attention given to the role of market mechanisms in resource allocation. Regional development policies face the dual challenge of fiscal sustainability and institutional efficiency. Indeed, sustained economic growth depends on a balanced interaction between the “visible hand” of government and the “invisible hand” of the market, a balance that remains an important area for future research.