Can institutions explain cross country differences in innovative activity?

Highlights

• We find a positive effect of institutions on R&D intensity by using 2SLS.

• This effect is robust to alternative measures of variables and estimation techniques.

• Human capital accumulation and financial development enhance this effect.

• Trade openness can neither enhance nor undermine the effect of institutions on R&D.

Abstract

Motivated by theoretical arguments (see e.g. Romer, 2010, Mokyr, 2008) that assert a positive impact of institutions on R&D, this paper aims to provide some empirical analysis on the relationship between the two variables. In particular, using a core sample of 98 countries over the period 1996–2009, this paper has found a significant direct effect of institutions on R&D intensity. Countries with better institutions qualities as captured by the World Banks’ Worldwide Governance Indicators (WGI) tend to attract more scientists and engineers into the research field and to spend more on R&D as well. This paper has also found evidence that the effect of institutions varies in different economies characterized by different levels of financial development and human capital accumulation, but stays relatively unchanged across countries with different levels of trade openness.

Introduction

Average R&D intensity levels in our core sample of 98 countries (measured by the average share of labor in the R&D sector over the period 2005–2009) in the world’s most and least research intensive regions differ by a factor of more than 33. The Sub-Saharan region, which consists of 20 countries in our sample, has on average less than 4 R&D workers per 10,000 total employees, compared to the Western Europe region, which consists of 16 countries in our sample and has on average more than 100 R&D workers per 10,000 total employees. What accounts for these differences, and what (if anything) can we do to reduce them?

This has become an interesting question since the realization of the uttermost importance of R&D in economic growth. There is little doubt today that R&D can affect growth. It is been theorized and empirically tested many times that R&D is the key variable that determines technological change, which then is the root of economic growth in the R&D based endogenous growth framework. (see, e.g. Romer (1990), Aghion and Howitt, 2006, Aghion and Howitt, 2009, Ulku, 2007, Ha and Howitt, 2007, Madsen, 2008, Madsen and Ang, 2011). This tells us that to drive up economic growth we need to increase R&D intensity, but what shapes the incentives to do R&D and what are the determinants of R&D in the first place?

In current literatures, R&D effort is thought to be mainly influenced by human capital, development of capital market and openness to trade. (See e.g. Romer, 1992, World Bank, 1998 and Dahlman (1994).) These determinants of R&D are also believed to have some degree of influence on growth too as pointed out by the above cited literature. The intuition here is that R&D determinants could either indirectly affect growth through their effect on R&D, or they could have direct effect on growth in the first place. Which is the exact truth has been debated among literatures at this stage. Nevertheless it is worth checking whether or not many established growth determinants could be potential R&D determinants as well. There are voluminous empirical literatures asserting a close link between institutions and growth (see e.g. Acemoglu et al., 2001, Acemoglu et al., 2005, Easterly et al., 2006, North, 1990, North and Thomas, 1973). However, to my best knowledge, very few of them (see e.g. Varsakelis, 2006, Tebaldi and Elmslie, 2013 and Jaumotte and Pain (2005)) examine the potential impact of institutions on R&D activities despite the fact that some theoretical arguments on this matter are already in place.

The main differences between our study and the above cited empirical works are: (1) our work focus mainly on the innovative activity by looking at both the input (labor, expenditure) and output (patents) of R&D. R&D activity is associated with great risk. Huge amount of money and manpower may be put into action, but successful results are not guaranteed. Therefore defining innovative activity only as patent applications or grants is a bit biased in the sense that countries can have the same level of innovative activities in terms of inputs (R&D personnel, expenditure) but achieve very different output results (patents). This may be caused by the fact that the quality of R&D labor differs greatly among countries, but the incentives to do R&D is measured purely by the inputs of R&D. Hence, including both inputs and outputs of R&D as measures of innovative activity is a more balanced approach as one single measure of R&D intensity is usually not ideal (see e.g. Acs et al. (2002)). (2) Our study focus on the general picture of political institutions or governance qualities rather than a specific type of institution such as educational institutions. (3) The coverage of countries in our sample includes most developed and developing countries across the globe. (4) We also analyse this relationship depending on the absorptive capacity of countries.

Employing both a two-stage least squares estimation technique which uses latitude and ethnolinguistic diversity as instruments for institutions and the system-GMM (Blundell and Bond (1998)) approach, we found a significant direct effect of institutions on R&D even after controlling for religion, legal origin and geography. The results are also robust to alternative measures of R&D and institutions as well as additional instruments and estimation techniques. We also find that the effect of institutions varies among different types of economies. In particular, economies with higher human capital accumulation and financial development seem to benefit more from better institution qualities, whereas trade openness shows no such trait.

The rest of this paper is structured as follows: Section 2 reviews theoretical and empirical arguments concerning R&D and institutions to provide some basis as to why we are interested in institutions as a potential determinant for R&D. Section 3 will discuss the data we use to test this relationship as well as underlying endogeneity issues and solutions, benchmark results will be presented. Section 4 provides sensitivity analysis by checking the benchmark results against alternative IVs, different measures of R&D and institutions and estimation techniques. Section 5 examines whether or not this effect is universal across different types of economies characterized by level of human capital accumulation, financial development and trade-openness. Section 6 concludes.