1.1.1 Statistical Overview

Begin with a simple look at primary data on within-country income distributions across a large sample of countries at varying levels of development. The most common measure is the Gini coefficient, constructed to capture how unequally income is distributed across households. The coefficient runs, in principle, from zero (which means that all households have an identical share of national income) to 100 (which means that just one household captures all the income). The higher the Gini coefficient, the less equal the income distribution.Footnote ⁶ Historically, such coefficients were infrequently and intermittently estimated for different countries, making statistical analysis difficult. However, a current project has compiled and published annual Gini coefficients for many countries to facilitate cross-country comparisons.Footnote ⁷

Table 1.1 shows the evolution of these measures from 1990 to 2015 for a selection of ninety-seven countries with data broken into income groups identified by the World Bank as of 1999. Two Gini coefficients are reported: the Gini Market, which is the coefficient based on incomes before taxes and transfers, and the Gini Disposable, which is the coefficient after such transfers are made. The difference between them indicates how much the income-transfer system in each country offsets (or reinforces) inequities established by the markets. Listed are simple averages within each country group and weighted averages, with the weights based on within-group populations. The countries are listed in Appendix Table A1.1.

These data suggest that there has been a steady but modest increase in inequality when all countries are averaged equally. Among the high-income countries (HICs), the average coefficient rose by two points using after-transfer income (around 7 percent) and by nearly four points using market incomes (almost 9 percent). The other income groups with rising coefficients were the lower-income countries (LICs) and lower-middle-income countries (LMICs). Interestingly, there was little net change in inequality in the upper-middle-income countries (UMICs) over the period. Presumably, this reflected a balance between higher inequality from technological change and lower inequality from the ability of lower-wage workers to gain from exporting labor-intensive goods, among other factors.

Unweighted averages can be misleading, however, as measures of inequality faced by large aggregations of households. The growth in the population-weighted average coefficients was considerably larger (though it declined in the UMICs), largely because of the significant increases in the United States (HI) and China (LMI). As shown, these two behemoths saw exceptionally large increases in both pretransfer and posttransfer household income inequality during this period of rapid globalization and technical change. In this context, those nations bear a significant share of the observed inequality.

It should be noted that the Gini coefficient may not be an accurate measure of rising inequality if that process is skewed toward higher incomes of the richest households. The way such coefficients are calculated gives small weights to households at the extremes (because there are not many in those ranges) and high weights to those in the middle ranges (where there are many). In consequence, if the top earners gain a disproportionate share of income within the distribution, the Gini may not go up by much, despite the rising disparities. Thus, Table 1.2 lists the pretax and pretransfer income shares (including capital gains) of the top 10 percent of households in a smaller sample of countries.Footnote ⁸ This measure offers readier evidence of rising inequality at the top, except for the LICs, where the highest households already captured more than 50 percent of income in 1990. The highest earners gained 3.7 points of income shares in the HICs, 7 points in the UMICs, and 4.6 points in the LMICs. Again, the United States and China were the largest contributors to these trends, with the top share in China rising by a remarkable 11 percentage points.

These figures show that there has been a marked rise in income disparities in the last thirty years, though some of that increase is concentrated in specific large countries. The question for this chapter is the degree to which stronger IP rights may be a factor. This is an exceptionally difficult question to answer statistically for many reasons that will be brought out in the text. For now, however, Table 1.3 demonstrates that there has been a marked increase in the strength of legalized patent rights over the same period. Listed are changes in the average Ginarte–Park (GP) index of patent strength across another subsample of our main sample.Footnote ⁹ This index notes whether a country has a law or regulation offering various elements of patent scope across five categories: coverage, membership in international conventions, patent duration, the possibility of losing patent rights, and enforcement. These component scores are added, and the final index ranges, in principle, from zero to five.Footnote ¹⁰

As seen in Table 1.3, the average index grew in all income groups during this period, reflecting a combination of unilateral reforms, adherence to standards required in the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) of the World Trade Organization (WTO), additional standards negotiated within regional trade agreements, and other factors. At the beginning of the period, the HICs already had strong protection on this measure and did not have to change their patent laws much over the period.Footnote ¹¹ The eye-opening feature is the substantial increase in protection in emerging and developing countries (EDCs). The index doubled in the UMICs and nearly tripled in the LMICs while growing sharply even in the LICs. As described elsewhere, the period since 1990 ushered in the most remarkable transformation of global IP policies in history (Maskus, Reference Maskus2012).

An obvious question for this chapter is, how well did the IP reforms track changes in income inequality? Did countries with relatively stronger increases in the patent index see greater increases in inequality, at least as measured by the Gini coefficients? The evidence provided in Table 1.4 suggests the answer is “not much.” The right-most panel depicts a growing but weakly positive correlation between the GP index and the market-based Gini coefficients in the HICs, consistent with higher inequality in more strongly protected jurisdictions. However, the correlation is negative in those countries using the posttransfer Gini measures, suggesting that whatever contribution IP may have made to inequality was more than offset by redistribution policies. A similar process may exist in the UMICs. Otherwise, there is little evidence of any relationship between measured patent rights and household income inequality.

1.1.2 Econometric Analyses

Computing simple correlations does little to sort out potential relationships and says nothing about any causal impact of IP protection on inequality. The standardized database of Gini coefficients offers some scope for econometric analysis, however, and two studies have estimated the impact of IP protection on inequality using such data. Because these studies paint a consistent picture, it is worth reviewing their findings before asking if they are reliable.

The first paper was by Adams (Reference Adams2008), who sought to fit IP rights empirically into a selection of determinants of inequality. The author acquired Gini coefficients from the World Bank for intermittent years between 1985 and 2001 for a sample of sixty-two developing countries and regressed them on various independent variables that were lagged to try to reduce reverse causality problems. Specifically, the model related the Gini coefficients to various macroeconomic globalization and policy variables, including trade openness, the degree of incoming foreign direct investment (FDI), the GP index as a measure of IP rights, secondary education rates, government consumption, a measure of institutional quality, and GDP per capita. The estimation found a consistently positive and significant effect of the GP index on subsequent inequality, with the primary coefficient suggesting that a one-unit increase (e.g., from two to three) in the patent index would increase the average Gini coefficient by around 1.2 points, in a scale between 0 and 100. A stronger institutional environment tends to reduce inequality, indicating that if a country plans to strengthen its patents and is concerned about impacts on income distribution, it may wish to complement that policy with a more secure set of rules governing investment.

A second study was by Saini and Mehra (Reference Saini and Mehra2018). The question they posed was whether strengthened IP rights in the post-TRIPS era had affected income inequality, using a sample of sixty-five developing and developed countries over the period 1995–2009. These authors used the Gini coefficients from the Standardized World Income Inequality Database (reviewed earlier) as the dependent variable in a panel regression highly similar to that in Adams (Reference Adams2008). Specifically, using five-year averages of most data, they regressed the posttransfer Gini measures on the GP index, openness to imports, inward FDI, GDP growth, and an indicator of political stability. The patent index interacted with the economic growth measure to assess whether faster-growing economies had a larger or smaller effect of IP on inequality. Similarly, it interacted with a dummy variable for developed countries to see if that relationship varied among poorer countries. They estimated a random-effects model because of the wide heterogeneity in country characteristics in conjunction with the higher dimensionality of countries than time periods.

Remarkably, the findings were completely at odds with Adam’s (Reference Adams2008) findings. In particular, the authors estimated that increases in the GP index tended to reduce the average Gini coefficient in developing countries, suggesting that stronger patent protection reduced income inequality. The authors speculated that this outcome reflected that stronger IP rights tend to attract more inward technology transfer, which could raise the relative wages of lower-skilled workers in labor-abundant countries. The coefficient on the interaction term of the patent index and GDP per capita was significantly positive, implying that the inequality reduction was lower in rich nations. Indeed, the relationship could be positive for countries above a threshold income level, implying higher inequality with strengthened patent rights in developed economies. They interpreted this outcome to suggest that stronger patent laws may induce innovation in the latter group of countries, with rents to that activity favoring those with more technical and managerial skills. Unfortunately, the authors did not attempt to subject these broad conclusions to further empirical testing.

These results are intriguing because, for now, they stand as the only cross-country estimates available of the potential impacts of IP protection on internal income distribution. However, they find distinctly opposite impacts, suggesting that the correlation between the legal determinants of patent scope and inequality, as measured by Gini coefficients, is ambiguous. Its estimation may depend on the data used and the specifications set out.

Moreover, such results are not likely to be reliably robust for further study, given the nature of the analysis. Using national macroeconomic data to assess complex relationships is fraught with risks and subject to fragile interpretations, as shown in prominent debates over the macroeconomic sources of economic growth.Footnote ¹² Both studies rely on reduced-form specifications that do not emerge from any particular theory other than what may seem to be common sense, and neither worries about general equilibrium and feedback effects among the different variables employed. Their regressions are subject to omitted variable bias, meaning that both inequality and IP rights could be driven higher by excluded variables, leading to spurious positive correlations. Similarly, the GP index is highly correlated with other measures of institutional quality and property rights, raising questions about whether it is really IP or something else that may raise inequality.

All of which suggest that further analysis is needed to sharpen predictions and improve confidence in such estimates. The remarkable element is that so little sound empirical analysis of the underlying questions exists. For that purpose, cross-national macroeconomic studies are unlikely to provide much utility. Rather, the emphasis must be placed on clear theoretical approaches that make detectable predictions about the microeconomic aspects of IP rights and inequality, measured perhaps through wage differentials between skilled and unskilled workers. A literature of this kind is now emerging, as described in the next section.

1.2.1 IP Protection and Effects on Inequality

These are reasonable propositions, even if the mechanisms involved are understudied. They find some explanation and support in scarce theoretical and empirical studies. For example, Chu (Reference Chu2010) studied the impact of stronger patent scope on economic growth and inequality in a single-economy model. The theoretical foundation is the “quality ladders” model, in which a continuous innovation process generates products of increasing quality that displace those previously at the top of the ladder.Footnote ¹³ The canonical model assumes individuals share identical and homothetic preferences, implying that income distribution does not matter for innovation and growth, nor is it affected by growth outcomes. Chu extended the model to an environment where households have different wealth profiles. How much to work in the labor market depends on savings and consumption decisions. His model is limited in assuming the distribution of assets among rich and poor households is unchanged in all steady-state equilibrium growth paths.

In this environment, a policy raising the scope of patents increases wealth and income inequality by raising R&D, which enhances the growth rate and raises the interest rate. In turn, asset returns rise, disproportionately favoring the rich, who own more assets. Furthermore (and somewhat counterintuitively), because the higher interest rate raises savings and reduces consumption, it induces households to spend more time at work. This factor also generates relatively higher income for the rich, further expanding income inequality. However, because consumption is reduced by relatively less among the poor, the outcome of stronger patents may be reduced consumption inequality. The author noted that these results are consistent with macroeconomic data in the United States in the 1980s and 1990s when there was a sharp rise in wealth and income inequality but a much smaller rise in consumption inequality. Because this model is highly stylized and does not fit detailed data, it is impossible to determine how much stronger patent scope in this era contributed to inequality via this mechanism. Still, it offers an interesting mechanism for further study.

The paper just described posits a heterogeneous and unchanging asset distribution between rich and poor, without explaining why households might vary that way while arguing that investment returns drive changes in inequality. It does not consider the skill bias of patent-induced innovation. A model doing so is set out by Pan et al. (Reference Pan, Li and Zou2015). The model is based on the idea of “directed technological change,” also a fundamental model in modern growth theory.Footnote ¹⁴ This approach argues that investments in new technology depend on numerous factors, including relative factor costs. In skill-abundant countries, such as the United States, such investments tend to reduce demand for low-skilled labor (which is expensive relative to other countries) and raise demand for high-skilled labor. This R&D-skill complementarity, found clearly in information technologies, is one basis for the predominance of SBTC.

In the model by Pan et al., R&D investments predominantly seek new technical solutions in higher-skilled industries, raising the demand for skills and, in turn, increasing wage inequality. Patents may be gained in new technologies complementary to either low-skilled or high-skilled workers. The key assumption is that skill-intensive industries increase returns to scale, lowering costs and raising productivity as output expands. Thus, in countries with abundant skills, it is optimal to scale a patent policy to favor the latter. The optimal patent policy then favors broader breadth to encourage scale economies. The result is a “skill-biased patent policy,” which raises wage inequality. It also imparts a clear bias between countries: Skill-abundant nations prefer broader patents, and labor-abundant nations prefer narrower protection. This paper offers no empirical analysis, so it cannot assess the validity of its key assumptions or the contribution of patent policy to growth and inequality. Again, however, it offers a useful perspective on which to build further analysis.

If theory suggests that strong IP protection supports wage inequality based on skills, entrepreneurship, or other worker characteristics, an empirical analysis should gravitate toward considering micro data on wage gaps across regions or within firms. This approach is natural because a patent exists at the firm level in specific locations, suggesting that carefully specified analysis could trace the impacts of patenting on wage inequality within the enterprise.

Two recent papers of note take this approach. First, Aghion et al. (Reference Aghion, Akcigit, Bergueard, Blundell and Hemous2019) analyzed state-level panel data in the United States to study whether “top income inequality” is caused to some degree by innovation. Top income inequality refers to increases in the income shares of the top 1 percent of households. The paper sets out a model of endogenous innovation by incumbents (who own patents and enjoy monopoly markups) and entrants (who innovate to gain patents). Innovation by either group raises the income shares of entrepreneurs and generates more top income inequality. But only investments by entrants increase social mobility, meaning the ability to enter the top income level. Entry may be blocked through high innovation costs, including extant patents. Such blockages slow mobility into the top tier. Regrettably, the model does not explicitly consider the role of stronger patent scope. Presumably, it has offsetting effects. On the one hand, patents should raise the returns on innovation and increase the top income shares. On the other, blocking entry should reduce the inequality associated with more rapid entry into entrepreneurship.

These predictions were tested using state-level panel data from 1975 to 2010. They gathered data on the top 1 percent and top 10 percent of income shares in all fifty states plus Washington, DC. Top income shares rose in every state, from an unweighted average of 8 percent in 1975 to a peak of 21 percent in 2007, before falling during the financial crisis. Moreover, there was increased variability across states, with the highest shares going toward states with stronger patent profiles. Additional data suggested that income from entrepreneurship (patent rents in the model) was disproportionately high in the top income group in such states.Footnote ¹⁵ The income figures were combined with state-level patenting data from the U.S. Patent and Trademark Office, including patent citations to construct quality measures. The cross-state variation in patenting was markedly high, generating scope for identification. The authors regressed the top income shares across states on lagged patents and patent quality, controlling for business conditions, the prevalence of the financial sector, state GDP, and population, plus state and year fixed effects. In ordinary least squares (OLS) regression, they found consistently positive and significant effects of patents and patent quality on the top 1 percent of incomes.

To control for the endogeneity of patents, they included each state’s representation on Congressional Appropriations Committees and other factors as instrumental variables. These regressions reported similar impacts of patents on top income shares. In particular, in the preferred specification, they found that a 1 percent rise in patents per capita raised a state’s top income share by 0.17 percent. That is, patenting alone could explain 17 percent of the rise in the top-level income proportion across states. This effect was even greater in high-patent states such as California. Put differently, if a state were to move from the bottom quartile of patents to the top quartile in 2000, the coefficients would imply an increase in its top income share of about 1.5 percentage points, a substantial increase. The authors argued that this calculation underestimated because it failed to account for the possibility that a successful inventor in a low-patent state would likely move to a high-patent state and other factors. Their results were robust to using other measures of top income inequality.

A second study of note is by Bhattacharya et al. (Reference Bhattacharya, Chakraborty and Chatterjee2022). Briefly, these authors took advantage of implementing a new Indian patent law between 2002 and 2005 to analyze whether the gap between manager wages and other wages within firms varied by whether those firms had patents before and after the legal implementation. They found strong evidence of an increase in these wage gaps, which was more pronounced in high-technology sectors. This evidence shows that firms do transfer patent rents disproportionately to skilled and managerial workers within firms, tending to raise wage inequality.

Studies using such microeconomic data are instructive and suggest that patenting and patent reforms can contribute to income and wage inequality through expected channels. The literature would benefit from considerably more such analyses, using other databases across countries, industries, and firms. It would be equally useful to quantify, with microdata, how patents and patent laws contribute to growing intra-industry market concentration and monopoly power within and across countries and how those rents have been distributed.

1.2.2 Alternative Perspectives

Even less studied by economists is the novel idea that inequality itself may be a determinant of innovation and the growth effects of IP protection. Put differently, while IP may generate higher growth and inequality, causation may also run from inequality to innovation and growth.Footnote ¹⁶ Mendez (Reference Mendez2002) noted that in an economy with a dual labor market, where some workers are paid efficient wages and others competitive wages, even neutral technological change can further increase wage inequality, as can globalization.

Newer literature is emerging and needs much fuller development. However, two papers illustrate how certain mechanisms could link inequality and innovation. First, Weinhold and Nair-Reichert (Reference Weinhold and Nair-Reichert2009) asked whether income inequality differences help explain innovation performance across countries.Footnote ¹⁷ There are two potential mechanisms identified. First, countries with large and growing middle classes (and therefore more limited inequality) should see political pressure raised for stronger governance institutions, including IP rights, which could increase innovation. Second, a larger middle class may offer more entrants into sufficiently skilled labor categories that raise the supply of innovation while increasing the demand for innovation through preferences for new goods. There are multiple relationships here to sort out through data analysis.

For this purpose, they used patenting data from the World Intellectual Property Organization (WIPO) database as measures of innovation, distinguishing between resident patent applications and foreign patent applications within each country.Footnote ¹⁸ They regress patenting on the size of each country’s middle class and the GP index as a measure of patent strength, instrumenting the latter with various structural and geographic factors. The argument is that if the income distribution affects innovation only through institutional reforms (the first channel), it should not significantly affect the regression when IP rights are included. However, if it influences innovation directly through supply and demand factors (the second channel), it should be independently significant. Further, its effects should vary between domestic patents (with a positive impact) and foreign patents (with little or no impact). They estimate a cross-section model, averaging national data from 1994 to 2000, leaving just fifty-three observations across countries. Most regressions found a positive and significant coefficient on the instrumented GP variable in explaining domestic patent applications, supporting the institutional channel. More importantly, they found consistently positive and significant estimates of the size of a country’s middle-income class on resident patenting, suggesting that more income equality is pro-innovation. In contrast, these variables had little effect on foreign patenting, which depended more on population, market size, and other macroeconomic factors.

This result is interesting but suffers from the usual concerns about cross-country econometric analysis with national data. The paper could be usefully extended, at least through industry patenting data, in which comparisons are made between the technological orientation of industries. It would also be useful to see if explicit measures of income inequality would demonstrate negative impacts on innovation through the identified channels.

A second paper, by Kiedaisch (Reference Kiedaisch2021), offers a theoretical model aimed at the related but deeper question of whether the impact of IP rights on economic growth depends on the level of economic inequality. The author studied this possibility in a “product variety” model of endogenous growth.Footnote ¹⁹ In this model, innovators seek to develop new varieties of existing products across industries, responding to the idea that consumers prefer diverse choices and will pay a premium for such decisions. The economy’s growth rate depends on the pace of new product development. Kiedaisch introduces inequality into the model by assuming that rich and poor consumers have different preferences. Specifically, richer households both consume a greater number of varieties and prefer more innovative or complex goods. Thus, in the model, the pace and nature of innovation in a country depend on income distribution. Along a balanced growth path, countries with more unequal distributions in the sense of relatively more rich people would have longer patent duration to incentivize more new products. In this sense, inequality increases expected growth.

The model has many more dimensions but, importantly for this chapter, establishes a theoretical mechanism that helps explain why countries with relatively larger rich groups (or more concentration of income and wealth at the top) prefer stronger IP protection, which in turn implies faster economic growth. The model has not yet been subjected to empirical analysis but doing so would be instructive. For example, while it seems intuitive and consistent with the casual observation that higher top incomes should push for stronger IP rights, the idea that such economies should grow faster clashes with the recent convergence in incomes between rich and emerging countries. Presumably, there is more to inequality across countries than can be explained in a single-country approach, as discussed in the following section.

1.3.1 Stylized Facts

To put this claim in perspective, consider the simple data in Table 1.5. These figures show, over certain subperiods and the full period, average annual growth in real GDP per capita, measured in 2017 U.S. dollars at purchasing power parity (PPP) exchange rates.Footnote ²⁰ Such exchange rates are appropriate for focusing on changes in actual living standards.Footnote ²¹ Again, the data are broken down into income groups, but now using the 2020 classifications established by the World Bank. In particular, China is now considered a UMIC, and India is an LMIC. This updated breakdown is more appropriate for analyzing income convergence. The comprehensive samples, rather than limited to those nations with available Gini coefficients as above, paint a complete picture. Finally, they are weighted by consistently measured GDP totals, capturing actual growth experiences in the income groupings more accurately.

It is readily seen in Table 1.5 that the HICs saw their growth rates fall from 2.3 percent in the 1990s to 1.3 percent in the following decade, before the stagnation following the financial crisis of 2008–2010. The 1990s were a period of slow growth for both the LICs and LMICs, though both saw substantial increases in growth in the 2000s. For both the LMICs and UMICs, the 2000s were a period of extremely rapid growth in real consumption standards. This was a period of great convergence in real incomes between EMCs and the HICs.Footnote ²² Over the entire period, the UMICs grew their consumption abilities sharply compared with the HICs, resulting in marked reductions in poverty and improvements in health and education status.Footnote ²³ In contrast, the LICs remained in relative stagnation.

There are, therefore, the following stylized facts. From 1990 to the present, the UMICs and LMICs considerably expanded their legal IP rights, as shown in Table 1.3, while experiencing significantly faster economic growth than the rich world. The LICs also adopted stronger laws, though these remained well behind those of the other groups while stagnating in relative terms. In short, the period was one of both extensive IP reforms and notable “conditional convergence” in living standards. Again, it would be difficult to argue and not credible to demonstrate with macroeconomic data that the former caused the latter. This is because many other factors could have driven both upward, with no necessary relationship between IP protection and growth.Footnote ²⁴

While that claim is largely correct, it is misleading in at least one important context. Economic logic and evidence both suggest that, at a microeconomic level, IP reforms in EDCs have attracted more technology flows, raising productivity. The next two subsections develop that argument. At this point, it is important to note that higher real incomes in EDCs do not necessarily imply more equal internal income distributions, as the gains may have been acquired largely by the already well-off. Put differently, the growth impacts of IP reforms may not trickle down into widespread income gains.

1.3.2 The Economics of IP Rights and Technology Development

How can IP reforms in EDCs raise domestic incomes, at least in the aggregate? The economics literature has emphasized three factors. First is the possibility that stronger home patent rights may directly raise domestic innovation.Footnote ²⁵ This question lies outside the scope of this chapter, and, in any case, the linkages from innovation to inequality remain unexplored in those countries.Footnote ²⁶ Briefly, the literature suggests that strengthening domestic IP rights can stimulate innovation, typically measured by future patent applications at home or abroad. This conclusion, however, must be strongly conditioned. It holds largely for higher-income emerging economies with, among other things, large domestic markets, adequate supplies of human capital, robust market competition, and sound policy governance. Countries lacking in such dimensions do not become markedly more innovative after patent reforms, highlighting the complexity of effective innovation systems. Specifically, there is little evidence that formal innovation is responsive to patent reforms in the poorest countries simply because the promise of domestic patents is a small incentive in weak economic environments. A corollary is that stronger patent rights likely do not diminish formal domestic innovation in poor countries, a frequently heard concern. Regrettably, there is no systematic evidence about how IP reforms may influence informal (unmeasured) innovation or competition through local imitation in poor countries. It is likely that domestic firms engaged in imitation and counterfeit production may be forced out of the market after significant reforms. Still, shortcomings in economic surveys have defied the systematic linkage of firm exits with IP policy. This remains a significant shortcoming in the economics literature.

A further important qualification is that even in dynamic emerging economies, undertaking extensive upgrades of IP protection may not be advisable in the medium term. The reason is that local firms in those countries, while undertaking R&D programs and exhibiting some innovativeness, may still rely on imitative technical change, often borrowing foreign technologies and making minor adaptations. In this regard, significant IP reforms may discourage local innovation and economic development by securing, often on behalf of foreign inventors, strong exclusionary rights that raise the costs of imitative and adaptive innovation (Kim et al., Reference Kim, Lee, Park and Choo2012).

1.3.3 Trade-Induced Innovation

Trade liberalization and skill-biased technological change are commonly thought to be the primary sources of income and wealth inequality across countries, including in the developing world. This is largely true regarding SBTC, as noted earlier. Reductions in trade and investment barriers, however, have complex impacts and may reduce inequality in developing countries. The standard reason is that integration with the global economy offers more export opportunities for industries that rely on labor-intensive production, tending to raise the real wages of low-skilled workers. At the same time, it encourages more imports of capital-intensive and skill-intensive goods, diminishing those incomes.Footnote ²⁷ Similarly, inward FDI into developing countries tends to be labor-seeking, raising local wages.

This tendency toward diminished inequality is far from certain, however. For example, current trade theory emphasizes that market opening pushes resources into the most efficient enterprises, which raises labor productivity and wages in general but could raise the wages of those with greater skills.Footnote ²⁸ Moreover, when global trade is liberalized, firms that start exporting generally must adopt globally efficient techniques, which tend to raise the relative wages of the most productive workers, who have higher skills even within occupational classes.Footnote ²⁹ This logic helps explain why exporting firms and multinational enterprises (MNEs) in manufacturing and services tend to pay significantly higher wages in EDCs than local firms, enhancing this form of cross-firm inequality.Footnote ³⁰

An exceptional form of this process emerged sharply in the 1990s and 2000s, the era of rapid growth in offshoring through vertical production networks (Baldwin, Reference Baldwin2016). The ability of multinational firms to fragment their production into lower-skilled activities (resource extraction, intermediate input production, and product assembly) and higher-skilled activities (design, marketing, R&D, and other headquarter services) and to transfer production of the former to developing markets was driven by cuts in trade costs and improvements in information and communication technologies. With lower transport costs and extremely low taxes on trade, parts and semi-finished goods could be shipped across borders multiple times in the production of final goods, a process exemplified by the proliferation of automobile production and trade within North America after the implementation of the North American Free Trade Agreement. The deployment of information and communication technologies permitted efficient inventory and shipping logistics management among nodes within production networks, greatly raising international trade in intermediate inputs.Footnote ³¹

This fragmentation led to the rapid offshoring of lower-skilled and medium-skilled jobs from the United States, Europe, and other high-wage nations to a limited group of “globalizers” among EDCs, especially China (Baldwin, Reference Baldwin2016). The harmful effects on low-skilled manufacturing and service jobs in these rich countries are widely identified as a source of inequality (Autor et al., Reference Autor, Dorn and Hanson2013). Less widely appreciated, however, is that offshoring also may serve to increase inequality in recipient countries. The reason is that the transferred jobs, while coming from the low and medium ends of the skill distribution in, say, the United States, tend to require skills and commitment to formal employment at higher ends in EDCs.Footnote ³² Thus, recent globalization, at least within vertical production networks, has been a source of growing inequality in the HICs and EDCs. Recall, however, the earlier discussion of how trade liberalization could raise the demand for low-skilled labor in EDCs through the comparative advantage channel. These counterbalancing impacts make the overall effects ambiguous.

Foreign direct investment and offshoring are important forms of trade-induced innovation because they generate new and more efficient forms of production processes in recipient countries. However, trade liberalization through tariff cuts and joining free trade agreements may also raise innovation on the part of domestic enterprises. As suggested earlier, firms must lower costs to compete with more efficient imports or develop new products to enter export markets.Footnote ³³ Both processes require investments in R&D, new capital goods, and better management techniques. Initial evidence for this spur to innovation in the wake of trade opening is in Bustos (Reference Bustos2011), who found that Argentinean firms experiencing relatively larger cuts in Brazilian tariffs after the foundation of MERCOSUR invested relatively more in upgraded technology.Footnote ³⁴ However, this happened primarily among firms at the upper reaches of productivity within Argentinean firms, not among lower-productivity and inefficient firms. A second important study is by Aghion et al. (Reference Aghion, Bergeaud, Lequien and Melitz2018), who developed a theoretical model in which greater access to export markets changes the incentives of domestic firms to innovate. Specifically, high-productivity firms have the resources to invest more in R&D and develop new products, while competition forces low-productivity enterprises to reduce innovation spending. These predictions were borne out using exporting and patenting data from French firms from 1994 to 2012.

The relationships between trade and investment liberalization and innovation are considerably more complex than suggested here. Much depends on local circumstances in each country. The broad view, however, posits that increasing global integration has encouraged more innovation, at least in developed and higher-income emerging economies. Because these innovation responses are concentrated in high-productivity enterprises, they likely have contributed to higher wage inequality across skill classes and across (and even within) firms within occupational groups. These issues will continue to attract scholarly attention for years.

It is worth noting an additional complication involving trade and IP rights. Emerging and developing countries that seek to export higher-technology products to developed countries may encounter import barriers. Many advanced economies have laws blocking imports that violate patents or trademarks in the importers. A well-known example is section 337 of the U.S. Tariff Act of 1930. Shin et al. (Reference Shin, Lee and Park2016) found evidence that the operation of such import restraints reduces the exports of more sophisticated products from emerging countries.

While important background for this chapter, none of the prior reviews directly implicated IP rights and how they interact with globalization to affect inequality. At this stage, the best guess is indirect: To the extent that stronger IP protection has causally attracted more trade, FDI, and offshoring, it presumably has contributed to those sources of inequality within nations. And here, the evidence is clear, as developed in the next subsection.

1.3.4 Technology Transfer and Income Convergence

Massive literature is now studying how IP reforms affect international flows of technical information through high-technology trade, FDI, offshoring, and licensing. Reviewing that literature is beyond the scope of this subsection.Footnote ³⁵ Rather, the intention here is to discuss three observations about how such flows could affect international inequality, defined as divergence or convergence in international living standards, as described empirically in Table 1.5. These issues can be subtle and counterintuitive, making it important to draw lessons from economic theory and empirical work.