The research, published in Frontiers in Sustainable Food Systems, treats China’s national return-home entrepreneurship pilot as a quasi-natural experiment. Beginning in 2016, the National Development and Reform Commission designated 206 county-level areas in two cohorts, then added 135 more in 2017, for a total of 341 pilot counties, county-level cities, and urban districts. Author Wenyi Zhou of Xiangnan University assembled an unbalanced panel covering 2,097 county-level units from 1999 to 2023, a quarter-century of annual data that allows the study to compare what happened in designated counties against what happened in similar places that never received the designation.

The statistical core of the study is a staggered difference-in-differences design, a technique that exploits the fact that counties entered the pilot in different years. Because county fixed effects absorb every stable difference between places and year fixed effects absorb every shock shared nationally in a given year, the treatment coefficient is identified from within-county changes over time relative to contemporaneous changes in untreated counties. The outcome variable is a composite index of rural economic resilience, built by principal component analysis from eight indicators spanning economic activity, financial and fiscal buffers, investment, human capital, medical services, agricultural productivity, and mechanization. The first three components, whose eigenvalues exceed one, jointly explain 64.91 percent of total variance and are combined using their normalized shares of explained variance.

The headline result is positive and statistically precise. In the fully specified two-way fixed-effects model, pilot designation is associated with an increase of 0.0652 in the resilience index, which amounts to roughly 0.136 standard deviations relative to the spread of the index in nonpilot counties. That may sound modest, but for a policy that operates through the slow accumulation of firms, jobs, and knowledge rather than through direct transfers, it is a nontrivial effect. The estimate survives a battery of robustness checks, including propensity-score matching, alternative staggered-adoption estimators, placebo tests that shift the policy date backward by one to six years, and a sample restricted through 2021, the year the formal pilot ended.

Timing is one of the study’s most interesting findings. The effect does not appear immediately. Event-study coefficients are small in the year of designation and the following year, then rise and become statistically significant in later periods. The authors interpret this lag as economically sensible: entrepreneurs need time to establish operations, assemble resources, learn about local markets, and expand. The pattern also fits the policy’s own biography. The pilot grew out of a 2015 State Council directive that identified migrant workers, university graduates, and veterans as priority groups, and it matured through phases of experimentation, evaluation, and nationwide diffusion that resemble China’s broader tradition of policy learning through pilots.

What mechanisms might carry the effect? The study tests three candidate channels. The strongest evidence concerns industrial restructuring. Pilot designation is associated with a 0.891-percentage-point increase in the share of secondary- and tertiary-sector value added in county GDP after county and year fixed effects are included. The logic is straightforward: rural economies that depend heavily on agriculture are exposed to weather and commodity-price shocks, while returning entrepreneurs, armed with nonagricultural skills and market access, can extend agricultural value chains, draw manufacturing and services into county economies, and diversify local production so that a single sector’s collapse does not cascade through the whole economy.

A second channel, innovation, receives weaker but still suggestive support. The study uses the China Rural Agriculture-Related Innovation Index, a county-level measure that combines technological, brand, green, and digital innovation among agriculture-related entities, from patent applications per thousand firms to organic certifications and the share of Taobao villages. Pilot designation is associated with a positive change in this index, though the estimate is significant only at the 10 percent level once both county and year fixed effects are included. The finding is consistent with related evidence from the Yangtze River Economic Belt showing that the pilot reduces agricultural carbon emissions through technological innovation and knowledge spillovers, but the author is careful to describe it as channel-consistent rather than causally identified mediation.

The third channel fails the test. Digital financial inclusion, measured by the Peking University Digital Financial Inclusion Index, shows a positive association with pilot designation when year fixed effects are omitted, but the coefficient turns negative and statistically insignificant once both county and year fixed effects are included. The study does not find that the pilot measurably raised county-level digital finance after common time trends are removed. This is a notable negative result, because entrepreneurship raises demand for credit while policy support may expand supply, and the net effect on a broad access index could plausibly go either way. The author concludes that easing financing constraints is not established as a principal channel in these data.

Heterogeneity tells a more selective story. Formal between-group tests identify a significant difference between eastern and western counties, with the largest subgroup point estimate in the west, and a significant difference by historical poverty status, with formerly poor counties showing larger and more precise gains. But the tests do not establish significant differences between eastern and central counties, between major grain-producing provinces and others, or between major labor-exporting provinces and others. In fact, the labor-exporting subgroup estimate is smaller than the comparison-group estimate, contrary to the study’s directional hypothesis. The author concludes that this hypothesis is only partially supported, a refreshingly honest verdict in a literature where subgroup differences are often overinterpreted.

The study is candid about its limits. The resilience index is an ex ante capacity proxy, not a direct measure of realized recovery after a specific shock, and components like nighttime luminosity and fixed-asset investment may capture development as much as resistance. Reconstructing the index after excluding these development-oriented components separately and jointly yields positive and significant coefficients, which reduces but does not eliminate the concern. A joint test of pretreatment coefficients does not reject equality to zero at the 5 percent level, though the result is marginal at 10 percent, and a negative pooled long-lead estimate warrants caution. A complete-case regression using only original, non-interpolated observations retains about 70.1 percent of the baseline sample and yields a smaller but still significant coefficient. Pilot counties applied and were reviewed, so matching and fixed effects address only observed and time-invariant differences.

What emerges is a carefully hedged but substantively important conclusion: policies that encourage migrants to return home and start businesses appear to strengthen the structural conditions that allow rural economies to absorb disturbances, mobilize resources, and adjust. The effect operates through industrial diversification and, more tentatively, through agriculture-related innovation, not through broad digital finance. It builds with a lag, it is largest in the west and in formerly poor counties, and it persists beyond the formal pilot’s 2021 close, when the government shifted from designated pilots to nationwide promotion of the accumulated experience. For policymakers elsewhere grappling with rural decline, the lesson may be less about the number of start-ups subsidized and more about the quality of human capital that returns, and whether local institutions can convert individual experience into relationships, routines, and organizations that outlast temporary support.

Subject of Research: The effect of China's national return-home entrepreneurship pilot on county-level rural economic resilience

Article Title: Return-home entrepreneurship and rural economic resilience: evidence from China's National pilot program

Article References: Zhou, W. (2026). Return-home entrepreneurship and rural economic resilience: evidence from China's National pilot program. Frontiers in Sustainable Food Systems, 10, Article 1961226. https://doi.org/10.3389/fsufs.2026.1961226

Image Credits: AI Generated

DOI: 10.3389/fsufs.2026.1961226

Keywords: return-home entrepreneurship, rural economic resilience, China, difference-in-differences, industrial restructuring, rural innovation, digital financial inclusion, county-level panel, policy pilot, human capital return, rural revitalization, staggered adoption