نوآوری، نابرابری و حقوق مالکیت معنوی

We examine whether income inequality explains innovation across countries. The mechanism could be twofold: first, a large middle class could have an impact on institutions, including intellectual property rights (IPRs), which could in turn affect innovation. Second, evidence from US economic history suggests direct linkages between middle class share and innovation via supply and demand effects. Using IVE to address endogeneity, we find that middle class share explains patterns of resident patenting, while non-resident patenting is driven more by exogenous factors and global integration. Our results illustrate an additional channel through which income inequality can impact long run growth.

Most modern econometric analyzes have failed to find any robust relationship between income inequality and economic growth over the short or medium run (see, e.g., Banerjee and Duflo, 2000 and Lundberg and Squire, 2003). However, there is an emerging literature on the determinants of cross country living standards that suggests that the presence of a strong middle class could be an important determinant of long run growth, through its impact on the evolution of institutional quality. Theoretical models linking income inequality to growth through its effect on the quality of institutions include Alesina and Rodrik, 1994, Bénabou, 2000 and Persson and Tabellini, 1994. Historical research in the United States (see Khan and Sokoloff, 2001 and Sokoloff and Khan, 1990) has also suggested that inequality could have played a direct role in the evolution of innovation. Combining these two literatures, this paper asks whether income inequality could help explain current differential levels of innovation across countries. In particular, we are interested in the relationship between inequality, innovation and intellectual property rights (IPRs). A fair number of previous studies of the relationship between income inequality and innovation have dealt with the effects of technological change on wage inequality within an economy (see, e.g., Acemoglu, 2002, Aghion, 2002, Huw, 1999 and Mendez, 2002). Relatively fewer studies examine the reverse relationship of the effect of inequality on innovation, although there is a literature that looks at the market structure (i.e., concentration) of industry and its relation to innovation.1 For example, Agell, 2001 and Agell and Lommerud, 1997 focus on developed countries’ industrial relations and consider the linkages between wage inequality and the incentives to invest in physical and human capital. Bénabou (2005) theoretically explores how the diffusion of technology affects the abilities of individual nations to maintain their own redistributive institutions and social structures. However, all these studies are focused on different questions than we are here; in the former cases the emphases are on different forms of industrial organization, while in the latter the theoretical chain of causality under consideration runs from innovation to inequality. In this paper, on the other hand, we motivate our research by considering the theory and evidence regarding the long run, cross country relationships between inequality, institutional quality, and innovation. Recent historical and empirical literature on the relationship between initial natural endowments, European settlement patterns, and the subsequent evolution of both income inequality and quality of institutions has emphasized the close linkages between these factors, and their importance in explaining income differentials across countries. Engerman et al., 2000 and Engerman and Sokoloff, 1997 hypothesize that variation in initial climactic and natural resource endowments leads to differences in the ability to establish European settlements as well as agricultural and extractive activities that differ in their ease of expropriation by political and economic elites. Areas that were naturally amenable to large-scale cash crops and/or extractive activities based on low-skilled slave or native labor also tended to have (for natural or political reasons) relatively low rates of European immigration. Acemoglu, Johnson, and Robinson (2001) argues that these initial inequalities in wealth and human capital between the European elite and the working class were exacerbated by the evolution of institutions designed to protect the status of the upper class and facilitate the extraction of rents. In contrast, in areas where there was significant European settlement and agriculture was smaller scale and more focused on staple foods such as wheat, broader political participation resulted in institutions to support private property and check the power of the State. Acemoglu et al., 2001, Engerman and Sokoloff, 1997 and Engerman et al., 2000 studies all suggest that inequality is intimately linked with the evolution of institutions, which in turn is an important determinant of income. Chong and Gradstein, 2004 and Easterly, 2007 make this link more explicitly. Easterly (2007) uses instrumental variables to show that a broad middle share does have a positive effect on development, and that the mechanism through which this effect takes place is through the effect inequality has on institutions. Chong and Gradstein (2004) use granger causality analysis to empirically demonstrate dual causality between income distribution and a variety of institutional quality measures. While most of the cross country studies cited above focus on measures of property rights (risk of expropriation) as proxies for general institutional quality, this paper extends this literature to examine the relationship between income inequality and another potentially important dimension institutional quality: the protection of intellectual property rights (IPR). It is well established from both theory and empirical evidence that IPR can have an impact on rates of innovation at the level of the firm or industry, or (occasionally) across countries. 2 If the fundamental linkages between income inequality and institutional design described in the Acemoglu et al., 2001 and Easterly, 2007 (among others) analyzes are robust, then we might expect to see the impact of a large middle class on IPRs and/or innovation as well. Furthermore, there are several existing hypothesized mechanisms through which these possible effects might operate. In particular, the literature on early US industrialization suggests relative income equality in that country could have affected innovation via several channels. From the demand side, Engerman and Sokoloff, 1997, Khan and Sokoloff, 2001, Sokoloff and Khan, 1990 and Sokoloff and Khan, 2000 argue that the presence of a large middle class helped spur innovation via their high demand for relatively inexpensive, mass produced manufactured goods. While low income people could not afford much non-essential consumption and high income individuals tended to demand customized products and services, the middle classes were oriented towards more standardized manufactures. In some sense this is a similar argument to the demand structure story of Zweimuller (2000), but is more nuanced with a focus on the incentives for innovation spawned by standardization. Sokoloff argues that much of the most fundamental advances in technology during the nineteenth century were concerned with the production of standardized manufacturing products. On the supply side there are two channels through which a large middle class could matter for innovation. Khan and Sokoloff, 2001, Sokoloff and Khan, 1990 and Sokoloff and Khan, 2000 note that in the Unites States, innovative activities were widely dispersed across a population with a high degree of market participation. They cite evidence from United States agriculture and manufacturing that suggest this broad based market activity in turn led to wide participation in the patenting of advances, which is based largely on incremental improvements on older technologies and organizations. They comment on the “wide range of industries to which American inventors had made technological contributions, the extraordinary creativity displayed in lowering the costs of producing standardized goods, and the broad spectrum of the population involved in inventive activity.” (Sokoloff & Khan, 2000, p. 2)

This paper examined whether the presence of a well-developed middle class and effective IPRs are as relevant today as factors spurring domestic innovation across countries as they were during the early stages of industrialization in the United States. While there has been a considerable literature reviewing the latter question from both theoretical and empirical perspectives, we have taken a slightly different approach from (to these authors’ knowledge) other papers. In particular, for the case of IPRs and innovation, we are concerned whether the positive observed correlation between strong IPRs and innovation might be due to either endogeneity or omitted variable bias. In the first case it could be that particularly innovative countries (for whatever reason) have a bigger incentive to enact stronger IPRs. In the second case perhaps the correlation is simply picking up the effects of good quality institutions generally, of which IPRs are just one component. We address these concerns by both instrumenting for strength of IPR and controlling for general institutional quality. We find some evidence that exogenous variation in IPR has predictive power for innovation and thus at least some of the causality in that relationship runs from IPRs to innovation. However, when we control for middle class share, the significance of this relationship is greatly reduced (p-value = 0.17), casting some doubt on our ability to adequately instrument for all these variables with such a small data set and so few uniquely varying instrumental variables. The results are thus suggestive and call for further empirical work to look into this relationship. In the case of income distribution, this is the first paper to the authors’ knowledge that empirically introduces middle class share as a possible explanatory factor in the variation in innovation across countries today. To date most, if not all, of the evidence that suggests inequality might play a role comes from primarily from US economic history and secondarily from the institutions and growth literature. These suggest that inequality could have an effect on innovation either indirectly via its effects on institutional quality (including IPRs) or directly via market participation and demand effects (the US historical story). Thus, while there are several plausible mechanisms through which inequality could be related to innovation, we would expect that the two more direct mechanisms (participation and structure of market demand) would have differential effects on resident and non-resident innovation and patenting activity. In particular, if the channel is through increased market participation, then we would expect this to increase domestic innovation, but not necessarily non-resident patenting. On the other hand, if the channel is through demand structure then there could be incentives for both resident and non-resident innovators to file patents domestically. Our results indicate that the size of the middle class, and to some extent IPRs, explain resident patterns of patenting, but not non-resident patterns of patenting in a modern cross section of countries. These results are robust to controlling for quality of institutions and instrumenting for endogeneity. Thus our results are consistent with the general Sokoloff story from US economic history that income inequality is plays a direct role in domestic innovation via a broader market participation of the population. This has important policy implications: given the links in the literature between innovation, productivity and growth, our results lend support to the idea that governmental polices relating to education, health, etc., that help increase the size of middle class in a country are important for future growth. On the other hand, the evidence suggests that non-resident patenting patterns are driven by relatively exogenous features and degree of global integration rather than by the kinds of internal, institutional factors that drive domestic innovation. Finally, given the outcome of our analysis, we can revisit the question of the use of patent counts as a proxy for innovation. If it were the case that IPRs increase the propensity to patent, but not necessarily the propensity to innovate, then given that these incentives should hit domestic and foreign interests alike, one would expect to find similar results across both domestic and foreign patents counts. Again, the fact that our results are dramatically different for these two groups is consistent with the Sokoloff theory but not consistent with these alternative explanations. In addition, the robustness check using patents filed in the United States by foreigners, in which all players are exposed to the same underlying institutional features of US law, similarly suggests that our results are not driven only by spurious correlation. The analysis presented here is based on a basic cross country regression design and the sample of countries is inherently limited by the data available; certainly more research along these lines is necessary. Given these constraints, however, there seem to be significant opportunities for future studies using different data and methodologies to explore the role of a broad middle class in spurring innovation while carefully controlling for possible endogeneity in the relationships between growth, innovation, IPR institutions and income distribution.