📈 economics

Determinants of Innovation Through Patents: An Analysis Using Panel Data

Using a panel dataset of 58 countries from 2000 to 2021, this study finds that R&D expenditure is the most robust determinant of patent activity, while per capita income shows a positive association and other factors like education and government effectiveness yield inconclusive results.

Original authors: Miguel Ángel Echarte Fernández, Sergio Luis Náñez Alonso, Beatriz Sastre Hernández, Ricardo Francisco Reier Forradellas

Published 2026-09-05

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Miguel Ángel Echarte Fernández, Sergio Luis Náñez Alonso, Beatriz Sastre Hernández, Ricardo Francisco Reier Forradellas

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the modern world, a nation's ability to invent new things is often seen as the engine of its prosperity. When a country produces more technology, it tends to grow richer, create better jobs, and handle global crises with greater ease. But measuring this invisible engine is difficult. How do you count the act of creating something new? Economists and researchers have long looked for a reliable way to track this progress. One of the most common tools they use is the patent. A patent is a legal document that grants an inventor the right to their invention, serving as a concrete record that knowledge has been turned into something tangible. While not every new idea gets patented, the number of patents a country generates is widely accepted as a strong signal of its technological health. The big question remains: what actually drives a country to produce more of these patents? Is it simply having more money, or does it depend on how well a government functions, how open a country is to trade, or how much it spends on schools?

A team of researchers from the Catholic University of Ávila in Spain set out to answer this question by looking at the big picture across the globe. They gathered data from fifty-eight different countries, tracking annual changes over a twenty-two-year period from 2000 to 2021. This massive collection of information allowed them to see patterns that might be hidden if they only looked at a single country or a single year. They focused on five main factors that might influence the number of patents a country produces: how much money is spent on research and development, how much is spent on education, how open the economy is to international trade, the average income of its people, and how effective the government is at its job. By using a statistical method that compares countries to themselves over time, they could isolate the specific impact of each factor, filtering out the unique quirks of each nation and the global events that affect everyone at once.

The results of this extensive analysis point to one clear, dominant driver. The study found that spending on research and development is the most reliable factor for increasing the number of patents. When a country invests more money into research, the number of patents rises in a consistent and predictable way. This relationship held true no matter how the researchers adjusted their calculations or which specific years they looked at. It suggests that direct financial support for creating new knowledge is the most effective way to boost a nation's innovative output. The data also showed a positive link between a country's average income and its patent production, but only when looking at the long-term trends within specific countries. This implies that as nations become wealthier, they tend to have more resources to dedicate to innovation, creating a supportive environment for new ideas to flourish.

However, the study also revealed that other factors often assumed to be crucial are not as straightforward as they seem. The amount of money a country spends on education did not show a clear, direct connection to the number of patents in their models. While education is undeniably important for building a skilled workforce, the researchers found that simply spending more on schools does not automatically translate into more patents in the short term. The effect might be too slow to measure in this specific way, or it might depend on how efficiently that money is used, rather than just the total amount. Similarly, the study found that a country's economic openness—how much it trades with the rest of the world—did not guarantee more innovation. In fact, the data showed a slight negative trend, suggesting that being open to trade does not automatically force a country to invent more; in some cases, it might even make it easier to import technology rather than develop it locally.

The quality of government institutions, measured by how effectively a public sector operates, also showed a complex picture. While good governance is generally seen as helpful for business, the study did not find a single, decisive rule linking it directly to patent numbers across all the countries examined. The researchers concluded that while a stable and effective government is likely a necessary background condition for innovation, it is not a magic switch that instantly increases patent counts on its own. The most important takeaway from this global review is that if a country wants to strengthen its ability to invent and innovate, the most certain path is to sustain investment in research and development. While other factors like income levels and institutional quality play a role, the direct funding of scientific and technological work remains the most powerful lever for turning a nation's potential into measurable, real-world progress.

Technical Summary: Determinants of Innovation Through Patents: An Analysis Using Panel Data

Problem Statement
While innovation is widely recognized as a fundamental driver of economic growth and competitiveness, the academic literature has not yet reached a consensus on how to measure and explain the specific variables that determine a country's degree of innovation. Traditional studies often rely on isolated indicators or fixed indices, potentially excluding critical factors that influence varying degrees of innovation across different nations. Furthermore, there is a need to move beyond a unidimensional view that links innovation solely to direct research and development (R&D) efforts, acknowledging instead that innovative capacity results from the interaction of technological, economic, human, and institutional factors. This study addresses the challenge of identifying a robust multivariate framework to explain the determinants of innovation using a large, heterogeneous sample of countries over an extended period.

Methodology
The study employs a panel data approach analyzing 58 countries over the period 2000–2021, resulting in 1,207 annual observations sourced from the World Bank. The dataset is unbalanced due to missing values in variables such as patent counts, R&D spending, and education spending, though per capita income data is complete.

Dependent Variable: The number of patents (absolute count).
Explanatory Variables: R&D expenditure (% of GDP), education expenditure (% of GDP), economic openness, per capita income, and government effectiveness.
Econometric Strategy: The analysis follows a rigorous selection process:
1. Baseline: A clustered Ordinary Least Squares (OLS) model serves as the initial specification.
2. Diagnostic Tests: Multicollinearity (Variance Inflation Factors), RESET specification tests, and the Breusch-Pagan LM test are applied.
3. Model Selection: The Hausman test is used to choose between fixed and random effects. The test results indicate a correlation between country-specific unobserved effects and explanatory variables, justifying the adoption of a Fixed Effects (FE) model.
4. Final Specification: A two-way fixed-effects model (incorporating both country-specific and year-specific effects) is estimated. To address potential heteroscedasticity and cross-sectional dependence, the final inference utilizes robust standard errors clustered by country.

The final estimated equation is:
$Patents_{it} = \alpha_i + \lambda_t + \beta_1(R\&D_{it}) + \beta_2(Education_{it}) + \beta_3(Openness_{it}) + \beta_4(Income_{it}) + \beta_5(GovEff_{it}) + \varepsilon_{it}$
Where $\alpha_i$ represents country fixed effects and $\lambda_t$ represents time fixed effects.

Key Results
The empirical analysis yields distinct findings across the explanatory variables:

R&D Expenditure: This is identified as the most robust and consistent determinant of patent activity. It exhibits a positive and statistically significant effect across all model specifications, including Pooled OLS, Fixed Effects (FE), and FE models with time effects. The results confirm that higher levels of R&D spending are directly correlated with a greater number of patents.
Per Capita Income: A positive association with innovation is observed, but this result is statistically significant only in the fixed-effects models (specifically models 3 and 4). The OLS model showed an inverse relationship, suggesting that unobserved heterogeneity biases the pooled estimates.
Education Expenditure: The results are inconclusive. While some models show a weak positive coefficient, the effect is not robust across specifications and loses significance in the final fixed-effects model without time effects. The data suggests a lack of a strong, direct linear relationship between education spending and patent counts in the short-to-medium term.
Economic Openness: This variable displays a negative coefficient across all models, though it generally lacks strong statistical significance. The data does not support a definitive conclusion that trade openness drives patent generation; in some contexts, it may even correlate negatively, possibly due to the ease of importing technology versus developing it domestically.
Government Effectiveness: The results are mixed, showing a change in sign between the OLS and FE models. While positive in fixed-effects specifications, the relationship is not statistically robust in the final clustered model.

The final model (Robust FE with time effects) achieves an $R^2$ of 0.175, a value consistent with the FE models with time effects (models 3 and 4), indicating adequate explanatory power for panel data of this nature, though the complexity of innovation implies the presence of unobserved factors.

Significance and Contributions
The paper contributes to the literature on innovation and economic growth by providing a multivariate framework that integrates technological, educational, economic, and institutional dimensions using exclusively official, internationally comparable data.

Empirical Validation: The study reinforces the consensus that sustained investment in R&D is the primary pillar of national innovative capacity.
Methodological Rigor: By utilizing a two-way fixed-effects model with clustered robust errors, the study isolates within-country variations over time, offering more reliable estimates than cross-sectional or pooled OLS approaches which may suffer from omitted variable bias.
Policy Implications: The findings suggest that while economic development (per capita income) supports innovation, policy efforts aimed at strengthening national innovative capacity should prioritize direct investment in R&D. The study cautions that other factors, such as education spending and economic openness, may require more nuanced, indirect, or long-term policy interventions to yield measurable impacts on patent output, as their effects are not immediately captured by simple linear correlations.

The authors conclude that while innovation is a multifaceted phenomenon, the evidence strongly supports the hypothesis that the generation of applicable knowledge depends fundamentally on the allocation of sufficient resources to knowledge-intensive activities, specifically R&D.

Technical Summary: Determinants of Innovation Through Patents: An Analysis Using Panel Data

More like this