Spatial heterogeneity analysis of biased land resource supply policies on housing prices and innovation efficiency

Land and resource supply involves government allocation of land resources via acquisition and distribution, with distribution affecting regional economies and societies. Internationally, governments influence land tenure, supply systems, and housing prices through policies (e.g., Egypt, Israel, UK). In China, multiple national plans and laws (e.g., Thirteenth Five-Year Plan, Land Management Law revisions) guide land use and prioritize major projects, while emphasizing housing’s residential function. Since around 2003, China implemented spatially biased land supply policies favoring central and western regions to balance development, increasing their land supply share. Land supply controls influence housing prices; shifts in supply have significantly impacted price fluctuations. Rising housing prices affect innovation by increasing factor costs and crowding out R&D and talent. As China pursues high-quality, innovation-driven development, local competition increasingly centers on innovation. Given disparities in innovation capacity and systemic obstacles, the paper asks whether land supply policies favoring central and western regions influence housing prices and innovation, and whether these effects are spatially heterogeneous. This study integrates land supply, housing prices, and innovation within one framework, constructs a theoretical model using instrumental variables (regional land supply as IV) to address endogeneity, measures innovation efficiency, and applies TSLS with robustness checks. The main result is that policy-driven land supply bias raises housing prices faster in the eastern region, which in turn suppresses innovation efficiency there. The study provides evidence and policy implications for optimizing land supply policies, aligning them with population migration, stabilizing real estate, and supporting innovation.

Pathways from land resource supply to housing prices: The land supply policy comprises quantity, structure, method, and pricing. Four channels link land supply to housing prices: (1) Production function channel—land quantity affects housing supply and prices; increased supply can aid destocking though mismatches may persist. (2) Expectations channel—policy shifts alter expectations and supply-demand dynamics, influencing prices. (3) Supply scale—differences in scale affect housing supply elasticity and thus prices. (4) Supply structure—planning and infrastructure (e.g., HSR) reconfigure land use and differentially impact prices across regions. International cases highlight varied planning regimes and their influence on housing markets (UK, France, Sweden). In China, post-2003 regulation reduced residential land supply and elasticity, constraining new housing. Residential, industrial, and commercial land prices rose sharply by 2012 vs. 2007. Many studies find increasing land supply curbs housing prices and note spatial/structural mismatches: land supply increases in central-west and decreases in east raise eastern land and housing prices; wages may rise with prices. Some argue limited or no direct causal relationship, or even positive correlation in high-cost regions. Housing prices and regional innovation: Predominant view—rising housing prices inhibit innovation via (1) crowding-out of innovation financing as capital flows to real estate; (2) crowding-out of innovative talent and reduced mobility; (3) blocking effect on entrepreneurial spirit and innovation vitality. The relationship can be U-shaped over time; many Chinese regions appear past the turning point where further price increases suppress innovation. Evidence shows higher prices deter entrepreneurship, reduce consumption slack, and discourage innovation inputs/outputs; spillovers occur across city clusters. Alternative perspectives note cases where housing appreciation correlates with higher innovation quality or spurs entrepreneurial vitality in some contexts. Overall, effects vary across time and space. Research gap: Prior research largely examines dyads (land supply–prices or prices–innovation). Few integrate all three. This paper unifies land supply, housing prices, and innovation, using panel data for 31 provinces (2003–2018) to test whether biased land supply policies generate spatially heterogeneous effects on prices and innovation efficiency.

The study unifies theoretical modeling, measurement of innovation efficiency, econometric identification with instrumental variables, and spatial-temporal data analysis. Theoretical model: Building on new economic geography (Helpman, Ottaviano, Krugman), the model considers two regions (a: central-west; b: east) with freely mobile labor. Consumers maximize utility over tradable industrial goods and non-tradable housing with utility Uq = Cqm^μ Cqh^(1-μ), budget Pqm Cqm + Pqh Cqh = Wq, leading to indirect utility Vq ∝ Wq − θ Pqh (θ = 1 − μ). Incorporating idiosyncratic regional preferences eaj, with free labor mobility implies equalized utilities across regions and determines labor distribution. On production, a CRS production function yields lq = [wq(1−β)]/(1−α−β) + C under unrestricted capital mobility (α + β < 1). Housing market: each worker consumes 1 unit housing; pq denotes housing supply elasticity parameter (larger pq implies lower elasticity), with lq = pq hq. Combining relationships yields equilibrium housing price expressions. Comparative statics show that increasing pq (lower supply elasticity in the east due to biased land supply toward central-west) raises house prices in both regions and accelerates price increases in the east relative to central-west. Hypothesis 1: Land supply policies favoring central-west raise housing prices nationwide, with faster growth in the east. Hypothesis 2: Faster eastern price growth induces stronger inhibition of innovation in the east vs. central-west. Innovation efficiency measurement: Stochastic Frontier Analysis (SFA) using a Translog production frontier for innovation output: ln(Inn_product_it) = ln f(x_it) + β1 ln Inn_labor_it + β2 ln Inn_capital_it + β3 ln Inn_labor_it ln Inn_capital_it + β4 ln^2 Inn_labor_it + β5 ln^2 Inn_capital_it + v_it + μ_it, where v_it ~ N(0, σ^2), μ_it ~ N+(μ, σ_μ), μ_i time-varying via μ_i = μ_i exp[−η(t−T)]. Innovation efficiency Inn_efficiency_it = exp(−μ_it) ∈ [0,1]. Frontier4.1 is used to estimate parameters; σ^2 and γ are significantly positive and one-sided LR test is significant, validating SFA. Econometric model: To estimate the effect of housing prices on innovation efficiency: Inn_efficiency_it = α0 + α1 Hou_price_it + α2 Z_{i,t−1} + δ_i + ω_t + ε_it, where Z are controls (lagged): Per_gdp, Hum_capital (university density), Ind_structure (secondary/tertiary output ratio), Imp_volume, Exp_volume, Tax_revenue. Province/region and year fixed effects are included. Endogeneity & IV: Innovation may affect housing prices (income/expansion effects), and omitted variables may influence both. The instrumental variable is the lagged state-owned construction land transfer area (Land_supply_{i,t−1}). Relevance: land supply affects housing prices via four channels (quantity, expectations, scale/elasticity, structure). Exogeneity: land supply is tightly constrained by the centrally determined annual land use plans; using the lagged value reduces reverse causality from prices to land allocation. TSLS is employed, with strong first-stage F statistics. Data and spatial analysis: Panel data for 31 mainland Chinese provinces, 2003–2018, classified into eastern and central-western regions per China’s Seventh Five-Year Plan delineation. Variables: Inn_product (invention patent grants), Inn_labor (R&D FTE), Inn_capital (internal R&D expenditure), Hou_price (average sales price of commercial housing), Land_supply (state-owned construction land transfer area, lagged), controls as above. Descriptive stats (N=465) and Pearson correlations are presented; Land_supply and Hou_price correlation is −0.0527. Spatial-temporal plots show post-2013 increases in central-west land supply, faster housing price increases in the east, and divergent trends in innovation capability (declining in many eastern provinces, slight increases in central-west). Robustness checks include replacing the explained variable with an index from Peking University’s China Regional Innovation and Entrepreneurship Index and estimating fixed-effects models.

• SFA validity and innovation efficiency trends:
  • SFA parameters indicate good fit: γ = 0.7962 (p<0.01), η = 0.0483 (p<0.01), log-likelihood = −205.4521; one-sided LR test significant (reported values indicate strong adequacy). Innovation efficiency rises nationally over time but remains low; east > central-west, with widening gap.
• TSLS first-stage (IV relevance: Land_supply_{t−1} → Hou_price):
  • Entire country: Land_supply coefficient −0.1300 (SE 0.0165), F = 772.74.
  • Eastern region: −0.1476 (0.0278), F = 391.20.
  • Central-West: −0.0792 (0.0207), F = 320.00.
  • Interpretation: Smaller prior-year land transfer areas are associated with higher current housing prices. The effect is stronger in the east, consistent with biased land supply favoring central-west reducing eastern supply elasticity.
• TSLS second-stage (Hou_price → Inn_efficiency):
  • Entire country: −0.0757 (0.0184), p<0.01.
  • Eastern region: −0.0924 (0.0190), p<0.01.
  • Central-West: −0.0160 (0.0066), p<0.05.
  • Interpretation: Rising housing prices significantly inhibit innovation efficiency nationwide, with a much stronger inhibitory effect in the east.
• Robustness: Replacement of explained variable (innovation efficiency index) yields consistent negative effects of Hou_price:
  • Entire country: −0.2622 (0.1178), p<0.05; East: −0.3290 (0.1714), p<0.05; Central-West: −0.2563 (0.2089), p<0.10. Direction and regional pattern align with benchmark.
• Robustness: Fixed-effects model indicates small but significant negative effect of Hou_price on Inn_efficiency (−0.0002424, p<0.05) and a weakly positive effect of Land_supply (0.0000707, p=0.048), corroborating that higher housing prices reduce innovation efficiency while land supply can enhance it.
• Additional descriptive insights:
  • Pearson correlation between Land_supply and Hou_price is −0.0527, suggesting that decreased land supply is associated with higher housing prices.
  • Spatial analysis shows post-2013 central-west land supply expansion; housing prices rise faster in the east; innovation capability tends to decline in the east and slightly rise in central-west.
• Hypotheses supported:
  • H1: Land supply favoring central-west raises prices in both regions, faster in the east.
  • H2: Faster eastern price growth yields stronger inhibition of innovation than in central-west.

The study links land resource supply policies to housing market dynamics and innovation outcomes in a unified framework. By leveraging exogenous variation from centrally controlled, lagged land supply as an instrumental variable, the analysis isolates the causal impact of housing prices on innovation efficiency while accounting for endogeneity. Findings show that the policy of favoring land supply to central-west regions reduces supply elasticity in the east, accelerating eastern housing price growth. Elevated housing prices then suppress innovation efficiency through higher factor costs, crowding out R&D investment and innovative talent, and dampening entrepreneurial vitality. Spatial heterogeneity is central: the east’s tighter land supply and more rapid price increases produce a markedly stronger inhibitory effect on innovation than in the central-west. These results highlight a spatial misallocation arising from administrative land allocation diverging from market-driven population flows, particularly in the east where continuing population inflows met constrained land supply, intensifying housing demand and costs and stifling innovation. The findings address the research questions by demonstrating both the nationwide and regionally differentiated pathways from land policy to prices to innovation, reinforcing the need for policy coordination that aligns land allocation with migration and innovation goals.

This paper contributes a unified theoretical and empirical framework connecting biased land resource supply policies, housing prices, and regional innovation efficiency. Theoretical modeling shows that shifting land supply toward central-west reduces eastern housing supply elasticity, increasing prices especially in the east. Empirically, using 31 provinces (2003–2018), SFA-based innovation efficiency, and TSLS with lagged land supply as an instrument, the study finds: (1) biased land supply policies elevate housing prices nationwide, with a stronger effect in the east; (2) higher housing prices significantly inhibit innovation efficiency, particularly in the east. Robustness checks (alternative innovation measure, fixed effects) corroborate these results. Policy recommendations include: tailoring land supply to local economic conditions, aligning construction land indicators with population migration, strengthening land use planning and structure, and enhancing interagency coordination to balance regional development and sustain innovation vitality. Future research should explore inter-provincial spillovers among land supply, housing prices, and innovation; unpack specific mechanisms behind regional differences (e.g., governance, market structures); and complement IV estimation with case comparisons to deepen causal understanding.

• Spatial interactions not modeled: The study focuses on east vs. central-west differences but does not analyze inter-provincial spillovers or spatial dependence among land supply, housing prices, and innovation.
• Mechanism granularity: While channels are theorized, the empirical analysis does not fully disentangle specific mechanisms (e.g., labor mobility constraints, firm financing structures) driving regional heterogeneity.
• Instrumental variable limitations: Although lagged land supply under centralized planning supports exogeneity, unobserved policy shocks or anticipatory behavior may still partially correlate with innovation environments.
• Measurement constraints: Innovation efficiency relies on SFA and proxies (patent grants, R&D inputs). Alternative innovation quality measures and firm-level data could refine results.
• Time and scope: Data end in 2018; subsequent policy changes and market cycles are not captured.