Panel econometrics · 2026

Money, Science & the Clean Energy Transition

Does public R&D money drive clean-energy research, or just follow it? A 50-year, 34-country, 9-technology panel study.

View on GitHub ↗

The question

When governments fund clean-energy R&D, does the money lead to research output, or does research pull the money in? And does the answer change with how mature the technology is?

What I did

I linked IEA public R&D budgets (1974–2023) to OpenAlex publication counts across 34 OECD countries and 9 energy technologies, then ran first-differenced panel Granger causality with country fixed effects, backed by an eight-test robustness battery (leave-one-out, GDP normalization, a 2015 structural break, HAC errors, lag variation, and a 500-iteration placebo).

What I found

• A maturity gradient in causality: hydrogen shows robust bidirectional causality (8/8 stress tests); growing tech (wind, ocean, CO₂ capture) shows one-way money → research; mature tech (nuclear, hydro) shows no robust link.

• A 60× efficiency gap: hydro ≈ 24 publications per $M, nuclear ≈ 0.4.

• Portfolio rotation: every major OECD country diversified away from nuclear since 1980.

• A carbon-management spillover cluster: hydrogen spending predicts CO₂-capture output (F = 20.9, p < 1e-8).

Selected figures

Fig. 1 · Panel-Granger F-statistics and causality typology across 9 technologies

Fig. 1 · Panel-Granger F-statistics and causality typology across 9 technologies

Fig. 2 · Research efficiency over time, publications per $M, by technology

Fig. 2 · Research efficiency over time, publications per $M, by technology

Fig. 3 · R&D portfolio concentration (HHI) over time, top-10 spending countries

Fig. 3 · R&D portfolio concentration (HHI) over time, top-10 spending countries