The study, published in Frontiers in Energy Research by Simon Baradziej of UiT The Arctic University of Norway, assembles one of the largest panels ever used to probe this question. It links the World Bank’s six Worldwide Governance Indicators—rule of law, regulatory quality, control of corruption, government effectiveness, political stability, and voice and accountability—to the renewable share of electricity generation for 195 countries across the years 2000 to 2023, yielding 4,359 country-year observations. The renewable data come from Our World in Data, compiled from Ember and the Energy Institute. Crucially, the design separates two things earlier work often blurred together: differences between countries, and changes within a single country over time.

The first surprise lies in the raw data. Unadjusted, the association between governance and renewables is actually negative. The reason is an artifact of history and hydrology: low-income countries average a 46 percent renewable electricity share, against just 25 percent in high-income countries, because hydropower—the oldest large-scale renewable—dominates the grids of poorer, well-watered economies while wealthy nations built their power systems on coal, gas, and nuclear. A naive cross-country comparison therefore suggests that weaker governance accompanies greener electricity, a conclusion nobody should draw.

Conditioning on income reverses the picture. Once log GDP per capita and urbanization are held constant, a one-unit rise in the governance index is associated with roughly 11 to 13 percentage points more renewable electricity—a large, statistically significant effect. But the study then applies harder tests. Adding natural-resource rents as a share of GDP trims the estimate, and region fixed effects, which absorb the geographic and endowment differences separating hydropower-rich Latin America from the fossil-fueled Gulf, cut it to 9.99 points with a confidence interval spanning zero. About a quarter of the apparent governance advantage, in other words, is regional composition; what remains is statistically indistinguishable from a null effect.

The within-country analysis is even more deflationary. In a two-way fixed-effects model—where identification comes only from each country’s own deviations around its mean, with common year shocks removed—the governance coefficient is essentially zero: minus 1.6 percentage points, with a p-value of 0.50. Countries whose governance scores improved over these two decades did not, on average, see their electricity mixes shift toward renewables as a result. Robustness checks reinforce the null: substituting low-carbon or wind-and-solar outcomes, lagging governance by one to three years, and dropping the 54 hydropower-dominated countries all leave the estimate small and insignificant. The result is a clean conceptual split: at a given income, better-governed countries tend to have greener grids, but the governance changes countries actually experienced did not move their grids.

Why can the study not identify which dimension of governance matters? Because the six indicators are nearly one construct. Pairwise correlations range from 0.69 to 0.94, and principal component analysis shows the first component absorbing 87 percent of their variance, with near-identical loadings across all six. Variance inflation factors reach 17.8, making any joint ranking meaningless. Estimated one at a time, five of the six dimensions carry positive, income-conditional associations, but their confidence intervals overlap so heavily that the apparent ordering—regulatory quality marginally ahead, political stability the only null—cannot be read as a genuine ranking. Any single-dimension coefficient simply stands in for the shared governance signal.

The most tantalizing result concerns time. In the baseline model, the within-country governance slope rises by about 0.39 percentage points per year, climbing from roughly minus 4 points early in the period to about plus 5 by 2023—a cumulative swing of some 9 percentage points. This pattern survives Driscoll–Kraay standard errors robust to cross-sectional dependence, three-year averaging of governance, and region-by-year and income-group-by-year effects. It is also stronger for wind and solar than for hydropower, a pattern consistent with the idea that governance matters most where deployment is policy-made rather than dictated by geography, since auctions, feed-in tariffs, and grid-access rules all run through the regulatory machinery that governance indicators measure.

But the temporal result collapses under the two most demanding specifications. When each country is allowed its own linear renewable trend—capturing the fact that transition trajectories diverge for reasons unrelated to institutions—the interaction drops to 0.005, with a p-value of 0.98. A first-difference specification returns a null as well. This is the decisive test: the apparent strengthening rests on comparing governance movements to a rising renewable trend, and that trend varies across countries for many reasons. Once each country’s own trajectory is removed, governance changes add nothing to explaining the timing of its transition. The pattern also appears in absolute quantities—renewable generation per capita—yet carries the same caveat.

What does this mean for policy? The message is deliberately measured. Governance reform is not an energy policy, and nothing in the data shows that improving a governance score will quickly lift a country’s renewable share. Nor do the results license prioritizing one institutional reform over another, given the collinearity of the indicators. At the same time, the study does not prove governance irrelevant. As the transition increasingly runs through competitive procurement and grid integration, governments capable of running credible auctions, enforcing contracts, and connecting projects to the grid may be better positioned to convert plummeting technology costs into actual deployment. The findings simply show that the observational evidence for this has been overstated.

The broader lesson is methodological as much as substantive. Many earlier studies, often confined to the European Union or the OECD and frequently predating the post-2010 collapse in wind and solar costs, relied on cross-sectional variation that conflated institutions with endowments. By mapping precisely where the governance–renewables link is robust—fragile between countries, null within countries on average, and not robust to differential trends in time—the study explains why the prior literature reached such conflicting conclusions. The estimates are associational, the design observational, and the author is candid that unmeasured time-varying confounders, reverse causation, and the perception-based nature of the governance data all remain live concerns. Whether a genuine governance effect emerges as the wind-and-solar era matures is a question that better-identified designs—event studies tied to specific policy adoptions, for instance—and further data will have to answer. For now, the careful verdict is this: the world’s most ambitious statistical case for governance as the engine of decarbonization does not, on close inspection, hold up.

Subject of Research: The relationship between national governance quality and the share of renewable electricity generation across 195 countries from 2000 to 2023

Article Title: Governance quality and the renewable-electricity transition: evidence from a 195-country panel, 2000–2023

Article References: Baradziej, S. (2026). Governance quality and the renewable-electricity transition: evidence from a 195-country panel, 2000–2023. Frontiers in Energy Research, 14, Article 1939944. https://doi.org/10.3389/fenrg.2026.1939944

Image Credits: AI Generated

DOI: 10.3389/fenrg.2026.1939944

Keywords: governance, renewable energy, electricity transition, Worldwide Governance Indicators, panel data, hydropower, wind and solar, institutions, economic development, corruption, energy policy, fixed-effects models