Does financial development matter for innovation in renewable energy?

ABSTRACT

This letter investigates a missing link in the literature – whether financial development matters for renewable energy innovation. Our analysis of 22 OECD countries between 1990 and 2014 suggests the crucial role of financial development in the development of both biomass and non-biomass renewable technologies. Additionally, the impact of financial development varies with countries’ carbon intensity and innovation growth rate.

KEYWORDS:

• Financial development
• innovation
• renewable energy
• carbon intensity growth
• innovation growth

JEL CLASSIFICATION:

• O30
• Q42

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ Our results still hold when we consider the ratios of stock market traded value and all private credit to GDP as proxies for financial development.

² The environmental policy stringency index captures the stringency of countries’ environmental policies such as taxes, trading schemes, fit-in-tariffs, emission standards, and R&D subsidies. Additionally, our conclusions still hold when we consider GDP and GDP per capita as proxies for market size and living standards.

³ Specifically, the past knowledge stock of country $i$ during year $t$ is defined as $K_{it}={\sum }_{s=0}^{t}PA{T}_{is}$, where $PA{T}_{is}$ is the number of renewable patents in country $i$ during year $s$. As a robustness check, we also consider alternative measures of past knowledge stock. Specifically, we calculate past knowledge stock using the perpetual inventory method, i.e. $K_{it}={\sum }_{s=0}^t(1-\delta {)}^{t-s}PA{T}_{is}$, where $\delta$ is the discount rate. We set $\delta =0.2$, following Aghion et al. (2016). Finally, we consider a global knowledge stock which varies over time, but not across countries. Overall, our conclusions are qualitatively similar under these alternative measures.

⁴ Our results are robust when we exclude the United States and Japan; however, we find smaller coefficients on the financial development variables when we exclude these countries from our data.

⁵ Our results are robust when we consider the first, third and fourth lags of the independent variables.

⁶ Our results are robust when we use bootstrapped standard errors and consider alternative models for count data such as the Poisson GMM model or the negative binomial model. We choose these specifications instead of a linear regression model because our dependent variable, the number of renewable patents, only takes non-negative values, therefore, it does not follow the distributional assumptions of a linear regression model. Additionally, log-transforming the number of patents will lead to the loss of data as zero-value patent counts will become undefined after the transformation.