Suppression or promotion: research on the impact of industrial structure upgrading on urban economic resilience

The study addresses how industrial structure upgrading (rationalization and advancement) influences urban economic resilience, especially under varying levels of globalization. Urban economic resilience—cities’ ability to resist and recover from shocks—has become critical given frequent crises (financial shocks, pandemics, disasters). China’s economy, transitioning from high-speed to high-quality growth, provides a pertinent context to examine whether upgrading the industrial structure enhances resilience. The authors posit that upgrading can both strengthen resilience (via diversification, high-tech orientation) and potentially weaken it (e.g., service-sector hollowing), implying nonlinearity. With globalization accelerating foreign direct investment (FDI) inflows and technological spillovers, the paper asks: Does industrial structure upgrading affect urban economic resilience nonlinearly across globalization stages? Are there threshold effects and regional heterogeneities? Two hypotheses are proposed: H1: Industrial structure rationalization positively affects urban economic resilience across globalization stages. H2: The effect of industrial structure advancement on resilience varies by globalization level (stage differences).

The literature on urban economic resilience spans measurement approaches (core variables vs. composite indices), determinants (fiscal gaps, geography, resource endowments, urbanization, innovation, agglomeration), and industrial impacts (sectoral influences and macro industrial transformation). Prior studies link globalization to industrial upgrading through division of labor, technological innovation, and factor allocation, but note cyclical, stage-dependent effects. Evidence on globalization’s impact on resilience is mixed (reducing, expanding, or dynamic effects). Industrial structure rationalization and advancement can buffer international market shocks by improving configuration and quality, facilitating adaptive adjustment. However, few works integrate globalization, industrial upgrading, and urban resilience in a unified framework or test nonlinear threshold effects. This study fills that gap by modeling globalization as a moderator (threshold) of the upgrading–resilience relationship.

Design: A panel threshold regression framework tests nonlinear effects of industrial structure upgrading on urban economic resilience using globalization as the threshold variable. Data comprise 267 prefecture-level-and-above Chinese cities from 2008–2021 (excluding Hong Kong, Macao, Taiwan, Xinjiang, Xizang), sourced from the Statistical Yearbook of Urban Construction in China and the Statistical Yearbook of Chinese Cities. Models: Two threshold panel models are estimated with city and time fixed effects: - UER_it = β1*RIS_it*I(GL≤r1) + β2*RIS_it*I(r1<GL≤r2) + β3*RIS_it*I(GL>r2) + αX_it + ν_it + ε_it (RIS model) - UER_it = α1*AIS_it*I(GL≤r1) + α2*AIS_it*I(r1<GL≤r2) + α3*AIS_it*I(GL>r2) + κX_it + η_it + ω_it (AIS model) Where UER is urban economic resilience; GL is globalization (threshold variable with thresholds r1, r2); RIS is industrial structure rationalization; AIS is industrial structure advancement; X_it are controls. Variables: - Dependent: Urban Economic Resilience (UER) measured by an output-based core variable method: UER = standardized GDP × standardized absolute change in GDP growth rate (adjacent years) × 100. - Core explanatory variables: • RIS (rationalization): a Theil-type measure of the coupling between sectoral output and labor across primary, secondary, tertiary sectors. • AIS (advancement): ratio of tertiary industry added value to secondary industry added value. - Threshold variable: Globalization (GL), proxied by per capita FDI. - Controls: industrial enterprise agglomeration (number of industrial enterprises per urban construction land area); local fiscal gap ((expenditure−revenue)/revenue in local fiscal budget); technology investment level (science & technology expenditure / local general public budget); population density (provincial permanent residents / area); infrastructure level (public transport route coverage / total population); economic development (GDP / population). Additional robustness controls: market size (retail sales/GDP), financial development (bank loans/GDP), cultural soft power (log book collection). Estimation and tests: - Endogeneity check via Durbin-Wu-Hausman (DWH): p=0.000 rejects exogeneity null; the benchmark specification adopts fixed effects after Hausman test (p=0.0000). - Threshold effect testing (Hansen’s panel threshold approach): F-tests for single, double, triple thresholds with bootstrap p-values; threshold values estimated by minimizing sum of squared residuals; confidence intervals via likelihood ratio (LR) statistics (critical value 7.35 at 5%). - Robustness: (i) drop extremes (Beijing 2019, Dingxi 2018); (ii) include time×individual interactions; (iii) include additional controls (market size, finance, culture). Regional heterogeneity assessed by splitting cities into eastern, central, northeastern, western regions and re-estimating threshold models.

National-level threshold existence and values: - Both RIS and AIS models exhibit significant double threshold effects with GL as threshold (Table 2): • RIS: single F=193.71 (p=0.0033), double F=55.85 (p=0.0700), triple not supported. • AIS: single F=514.98 (p=0.0000), double F=90.97 (p=0.0133), triple not supported. - Estimated GL thresholds (95% CI) (Table 3): • RIS: r1=6.8724 (6.8234, 6.9304); r2=7.5034 (7.4333, 7.6229). • AIS: r1=6.5514 (6.5056, 6.5928); r2=6.8724 (6.8234, 6.9304). - LR consistency checks confirm thresholds are valid (LR values below 7.35 critical value). Effects across globalization stages (Table 4, basic models): - RIS (rationalization) effect on UER: • Low GL: 0.023 (SE 0.005), significant at 1%. • Medium GL: 0.095 (0.006), 1%. • High GL: 0.037 (0.009), 1%. Interpretation: Rationalization consistently and significantly promotes resilience at all GL stages (supports H1). - AIS (advancement) effect on UER: • Low GL: −0.035 (0.040), not significant. • Medium GL: 0.313 (0.054), 1%. • High GL: 0.780 (0.049), 1%. Interpretation: Advancement shifts from non-significant/negative at low GL to significant positive at higher GL (supports H2). At high GL, the effect magnitude is large (0.780). Robustness: - Dropping extremes, adding time×individual interactions, and adding extra controls preserve signs and significance patterns of core results, with minor changes (e.g., AIS significance lowering to 5% in some robustness runs). Overall model remains robust and reliable. Regional heterogeneity: - Eastern region: Both RIS and AIS show double threshold effects. Coefficients increase with GL and indicate an inverted U-shaped relationship across stages. Reported coefficients include (narrative): • Low GL: RIS 0.056***; AIS 0.275**. • High GL: RIS 0.109***; AIS 1.546***. - Central region: Double threshold effects; inverted U-shaped pattern similar to eastern region. From Table 6: • RIS: Low 0.019***; Medium 0.085***; High 0.032***. • AIS: Low 0.023 (ns); Medium 0.708***; High 0.251***. - Northeastern region: Only a single threshold (RIS model). RIS effect is insignificant below threshold (~6.63) and becomes significantly negative above (−0.021***), indicating inhibitory influence; AIS shows no double threshold. - Western region: No threshold effect detected for RIS or AIS; effects on resilience are not significant, likely reflecting infrastructure gaps and limited FDI spillovers. Descriptive statistics highlights (Table 1): - UER mean 0.701 (SD 1.594), min 0.000 (Dingxi 2018), max 20.033 (Shanghai 2019). - GL ranges from −2.028 to 9.970, indicating substantial variation across cities and time.

The findings confirm that the relationship between industrial structure upgrading and urban economic resilience is nonlinear and stage-dependent on globalization. Rationalization of the industrial structure consistently enhances resilience by improving the match between inputs and outputs and fostering diversification that buffers shocks. Advancement of the industrial structure exhibits stage differences: at low globalization, the benefits of upgrading are muted or negative, possibly due to competitive pressures, early-stage hollowing, and limited absorptive capacity; as globalization deepens, FDI inflows, technology and talent spillovers, and transitions to higher value-added activities increase, turning advancement into a strong positive driver of resilience. Regional analyses show that coastal and more globally integrated regions (eastern, central) reap stronger benefits and exhibit clear threshold dynamics (inverted U patterns), while legacy industrial structures (northeast) or underdeveloped infrastructure and lower openness (west) constrain the resilience gains from upgrading. These results underscore the importance of aligning industrial policy with globalization stage and regional conditions to maximize resilience gains.

This study couples industrial structure upgrading, globalization, and urban economic resilience within a unified empirical threshold framework using panel data for 267 Chinese cities (2008–2021). It demonstrates significant double threshold effects of globalization on the upgrading–resilience nexus nationally: rationalization promotes resilience across all globalization stages, while advancement transitions from non-significant/negative to strongly positive as globalization rises. Robustness checks corroborate the findings. Regional heterogeneity is pronounced: eastern and central regions display double thresholds and inverted U-shaped effects; the northeast shows only a single threshold (with inhibitory RIS effect at higher GL); and the west shows no threshold effect. Policy implications include accelerating rationalization and targeted advancement of industrial structures, leveraging openness and innovation, and tailoring strategies to regional globalization levels and capacities. Future research should refine resilience and upgrading measures, consider cyclical dynamics, and address data gaps to further unpack mechanisms.

- Measurement of urban economic resilience relies on a core variable, output-based approach (standardized GDP and changes in GDP growth), which may omit multifactor interactions and periodicity; future work could incorporate comprehensive indices and temporal cycles. - Proxies for industrial structure advancement (tertiary-to-secondary ratio) and rationalization (Theil-type index) may not fully capture the multidimensional nature of upgrading; further classification and refined metrics are needed. - Data limitations include missing values for certain regions (especially western China). Expanding coverage to all prefecture-level cities and improving data completeness could affect conclusions and should be explored in future research.