Technical Summary: Institutionalized R&D, Patent Output and the Domestic Core–Periphery of Innovation in Türkiye

1. Problem Statement and Research Context

The paper investigates the spatial dynamics of innovation in Türkiye, specifically addressing how the country's domestic core–periphery structure influences the conversion of institutionalized Research and Development (R&D) capacity into regional patent output. While Türkiye has expanded formal innovation policy instruments—such as R&D centers, Technology Development Zones (TDZs), and patent incentives—over the last two decades, these investments have not resulted in a uniform diffusion of technological capability across regions.

The central problem is whether these institutional inputs generate inventive output equally across all NUTS-2 regions or if their effectiveness is conditioned by regional welfare, absorptive capacity, and technology specialization. The study posits that without accounting for the cumulative nature of regional advantages, innovation policies may inadvertently reinforce existing spatial inequalities, creating a "spatial version of the middle-income trap" where national R&D expenditure rises while regional divergence persists. The research asks: How does Türkiye's domestic core–periphery structure shape the conversion of institutionalized R&D capacity into regional patent output across NUTS-2 regions?

2. Methodology and Data

The study employs an integrated empirical framework at the NUTS-2 regional level, combining descriptive spatial analysis with econometric modeling.

Data Sources and Variables:

• Unit of Analysis: 26 NUTS-2 regions in Türkiye.
• Time Period: Data spans from 2009 to 2025 (with specific cross-sectional snapshots for concentration analysis).
• Dependent Variable: Regional patent applications (treated as count data).
• Key Independent Variables:
  • Institutionalized R&D Capacity: Number of R&D centers.
  • Innovation Infrastructure: Number of Technology Development Zones (TDZs).
  • Regional Welfare: Per capita GDP (and the ratio of regional to national per capita GDP).
  • Human Capital & Investment: R&D personnel intensity and R&D expenditure.
  • Technological Specialization: Distribution of R&D centers across 13 substantive technology groups (e.g., defense, aerospace, digital technologies, automotive).
• Composite Core Index: A constructed index standardizing seven indicators (per capita GDP, relative GDP, patents, R&D centers, TDZs, R&D intensity, and personnel intensity) to classify regions into Core (top 5), Semi-periphery, and Periphery.

Empirical Strategy:

1. Spatial Concentration Analysis: The study calculates concentration ratios (CR5), Herfindahl-Hirschman Index (HHI), Gini coefficients, and coefficients of variation to map the distribution of innovation assets.
2. Count-Data Regression: Given the nature of patent data (non-negative integers with overdispersion), the study utilizes Poisson and Negative Binomial models. The Negative Binomial model is preferred due to significant overdispersion (variance >> mean).
3. Interaction Effects: The models test whether the relationship between R&D inputs and patent output is moderated by regional core status, welfare levels, and technology specialization.
4. Spatial Diagnostics: Global Moran's I statistics are used to assess spatial autocorrelation in patent outputs and model residuals to ensure robustness against omitted spatial dependence.

3. Key Results

Spatial Concentration and Core–Periphery Structure:
The analysis reveals a highly concentrated domestic innovation geography. The top five regions (the "Core") account for:

• 76.2% of all patent applications.
• 76.3% of R&D centers.
• 79.7% of R&D expenditure.
• 71.0% of R&D personnel.
  This concentration is sharper than that of per capita GDP alone, indicating that technological capability is more unevenly distributed than income. The Core regions (Ankara, İstanbul, Kocaeli–Sakarya–Düzce–Bolu–Yalova, İzmir, and Bursa–Eskişehir–Bilecik) exhibit a cumulative advantage where welfare, institutional density, and inventive output reinforce one another.

Regression Findings:

• R&D and TDZ Impact: Both lagged R&D centers and TDZ infrastructure show a positive and statistically significant association with regional patent output. TDZs, in particular, demonstrate a strong incidence-rate ratio, suggesting they effectively support inventive output.
• Welfare as a Predictor: The per capita GDP ratio is a robust predictor of regional inventive performance, confirming the link between regional welfare and technological output.
• Core Status Effect: The "Core" dummy variable is positive and significant, indicating that core regions produce substantially higher patent output (approximately three times that of non-core regions) even after controlling for R&D inputs and infrastructure.
• Interaction Effects: While the main effects of R&D and welfare are strong, the interaction terms (e.g., R&D centers × Core status or R&D centers × GDP) were not statistically significant in all specifications. This suggests that the core advantage is structural and cumulative rather than solely a result of a multiplicative interaction where R&D inputs yield disproportionately higher returns only in the core.

The Case of Ankara:
A striking finding is the ranking of Ankara as the top region in the composite core index (2.60), surpassing İstanbul (2.10). Despite lacking coastal market access or port advantages, Ankara leads due to:

• The highest R&D expenditure-to-GDP ratio (3.88%) among core regions.
• The highest R&D personnel intensity (3.48%).
• A dominant specialization in defense, aerospace, and weapon systems.
• High institutional density and public-university-industry linkages.

4. Key Contributions

The paper makes four specific contributions to the literature on regional innovation systems and development economics:

1. Integrated Empirical Framework: It provides a unified NUTS-2 level analysis that simultaneously examines R&D institutions, patent outputs, welfare indicators, human capital, and technology specialization, moving beyond single-indicator assessments of regional innovation.
2. Domestic Core–Periphery Interpretation: It demonstrates that in an emerging economy, technological capacity and regional welfare are jointly and unevenly concentrated. The study validates that the R&D-to-patent conversion is not a uniform national mechanism but is conditioned by a domestic hierarchy.
3. Identification of a Mission-Oriented Core: The paper challenges the assumption that innovation centrality is solely a function of market size or global connectivity. It identifies Ankara as a distinct "mission-oriented" science and technology core, driven by public demand, defense-industrial specialization, and state-linked R&D, contrasting with İstanbul's market-scale/digital hub and İzmir's industrial-coastal profile.
4. Nuanced View of National Innovation Systems: It argues that national innovation cores are not homogeneous. Instead, emerging economies may develop multi-centered structures where different regions specialize in different technological functions (e.g., defense, digital, automotive), contributing to a more complex understanding of uneven development.

5. Significance and Claims

The paper claims to demonstrate that Türkiye's innovation geography is shaped by cumulative technology-welfare mechanisms rather than isolated innovation inputs. The conversion of institutionalized R&D capacity into inventive output is deeply embedded in a broader regional hierarchy.

The study concludes that while R&D centers and TDZs are positively associated with patent output, their effectiveness is situated within a spatially uneven structure where core regions accumulate advantages. The findings suggest that policy interventions focusing solely on increasing the number of R&D centers in peripheral regions may be insufficient without parallel investments in absorptive capacity, skilled labor, and institutional depth.

The paper modestly notes its limitations, acknowledging that patent counts do not measure quality or commercialization success, and that the R&D center variable does not capture the internal quality or budget of individual centers. However, it asserts that the evidence robustly supports the existence of a domestic core–periphery structure where Ankara, driven by mission-oriented public demand and strategic specialization, emerges as the country's leading science and technology center, distinct from the economic dominance of İstanbul.