The Reason For The Globalisation Of Tech R&D

US firms have begun shifting R&D investment towards non-traditional destinations such as China, India, and Israel. The column argues that this is a response to a shortage in software and IT-related human capital within the US. When US multinationals are able to import talent or export R&D work, this reinforces US technological leadership. Conversely, politically engineered constraints on this response will undermine the competitiveness of US-based firms.

Despite rising globalisation after WWII, corporate R&D spending remained highly concentrated in the same small group of advanced industrial countries that dominated it for decades – until recently. Since the 1990s, the distribution of US multinational R&D investment across countries and industries has shifted dramatically toward non-traditional R&D destinations like China, India, and Israel (Kerr and Kerr 2018). Today’s leading US multinationals have developed a global innovation system that increasingly relies on emerging market talent to propel innovation for the global frontier.

Why these emerging markets – and why now? In new research, we argue that the rising importance of software and information technology as drivers of innovation and new product development across a wide range of industries led to a shortage in software/IT-related human capital within the US (Branstetter et al. 2018a). This has driven US multinationals abroad in a search for talent.

To support this argument, we show:

• the extent of globalisation of R&D by US MNCs,
• the growing importance of software and IT in firm innovation across industries,
• the rise of new R&D hubs, and
• the differences in the type of activity done there.

We also document that IT-intensive and software-intensive firms were more likely to conduct R&D abroad, and that foreign R&D is most pronounced in IT-intensive and software-intensive countries.

The increase in demand for an IT and software workforce

Software and IT patents have been growing in importance since the 1990s; Figure 1 shows the share of all USPTO patents that are based on software. The share grew from 5% in 1990 to nearly 40% by 2015, and from 10% to 45% for IT. This growth in the IT intensity of invention was explored by Arora et al. (2013) and Branstetter et al. (2018b), who interpret the rise in IT intensity as the emergence of a ‘general purpose technology’ in new production development that applies across manufacturing industries.

The advent of powerful microprocessors, memory chips, sensors, and digital control systems has enabled new generations of devices to become smarter and more responsive to their environment. Improvements in product functionality can often be achieved through better software alone. This has made IT, and especially software engineering, more central to success in innovation and new product development, increasing demand for IT and software engineering talent.¹

Figure 1 The growing IT/software intensiveness of US MNC invention

Importing talent from abroad

According to the National Survey of College Graduates, the IT/software workforce – made up of programmers, computer scientists, and electrical engineers – grew by 112% between 1993 and 2010, while the overall workforce grew by 70%. The foreign share of IT workers grew from 16% in 1993 to 32% by 2010. This phenomenon has been documented by Bound et al. (2015). These changes suggest an extremely large increase in demand that was partially met by importing talent from abroad, through mechanisms like the H-1B programme.

Wage comparisons provide evidence that the foreign talent supply did not meet demand in the software and IT sectors. Figure 2 shows average compensation per worker for US multinationals across different countries, using publicly available Bureau of Economic Analysis data [https://www.bea.gov/iTable/index_MNC.cfm]. 

Figure 2 Average compensation per employee at US MNC foreign affiliates

Source: Bureau of Economic Analysis.

For average compensation per employee in aggregate, the US is in the middle of the country distribution. If we consider IT-specific industries² like Electrical Equipment Appliances and Components, or Computers and Electronic Products, however, the average compensation per employee at US headquarters was much higher than pay for employees at a foreign affiliate. Average compensation represents many functions within US parents and affiliates, the wages of skilled R&D personnel are likely to be relatively high in both the US and in other countries. Nevertheless, these numbers are clearly consistent with the view that:

• demand has outpaced supply of IT and software workers in the US, and
• raw engineering talent of high quality is available in large quantity and at relatively low prices in emerging markets – especially India and China.

Interviews with the R&D managers of leading US multinationals both inside and outside the US supported the perception that there is a global shortage of IT and software talent. We also confirmed the need to move abroad to gain necessary access to large foreign supplies of skilled engineers.

New R&D destinations have an abundant supply of human capital

The supply of technically skilled workers is abundant in many of the same countries in which we found an increase in US MNC foreign R&D activity – notably India and China. Applications for Indian and Chinese high-skilled workers made up 85% of H-1B visa applications in 2017 (US Department of Homeland Security 2016), and Indian and Chinese students combined made up 18% of doctorates in science and engineering from US universities in 2016.³ This share was even larger in some key disciplines.

If we view the large number of Indian and Chinese students pursuing graduate education at American research universities as the extreme right tail of a distribution of science and engineering talent, most of which remained at home, then this suggests a massive amount of software- and IT-trained human capital available in China and India. Indeed, Arora and Gambardella (2005a and 2005b) record an abundant supply of engineering and technology graduates in emerging economies.

The types of activity done in new R&D destinations like China, India, and Israel suggests why they have been chosen as R&D hubs. Bureau of Economic Analysis data shows that R&D-performing affiliates in China, India, and Israel are concentrated in computer and electronic production manufacturing and professional, scientific, and technical services. In more traditional destinations like Germany, Japan, Canada, the UK, and France, R&D is concentrated in traditional manufacturing.

This suggests the need for human capital to meet the demand of software- and IT-intensive US multinationals has motivated US MNC decisions to do R&D in these locations. Figure 3 shows that patenting and the R&D investment of US multinationals outside the US has grown disproportionately in those regions where IT and software skills are well developed.

Figure 3 US patenting and R&D investment growth is increasingly concentrated in regions specialising in IT and software

In the two graphs, the horizontal axis measures the degree to which local inventors in a country tend to specialise in IT and software invention, as measured by their USPTO patent grants. This obviously tends to be places in which local IT and software skills are well developed. Regression analysis also implies this positive relationship, which is robust to the inclusion of control variables.

Conclusions and implications

Our analysis suggests that the increasing reliance on IT and software in innovation, and the growing endowments of specialised human capital in countries like India and China, have induced US MNCs to conduct more R&D in these locations. This has important implications:

• It suggests that there is a constraint on the supply of IT and software human capital in the US, and that these human resource constraints limit the invention possibilities for US-based multinational firms, even for those firms in which innovative activity and technological opportunity seem to be at the highest levels.⁴
• Global flows of investment, people, and ideas can help relax these constraints through open immigration policies and liberal trade and FDI policies. When successful, these flows raise growth, productivity, and consumption possibilities around the world. When US multinationals are able to import talent or export R&D work to the regions in which talent resides, this reinforces US technological leadership. Conversely, politically engineered constraints on this response clearly undermines the competitiveness of US-based firms.

We do not directly explore the impact of immigration policy in our paper, but existing evidence suggests that the openness of the US’ labour market to immigrants in the 1990s allowed US-based firms to quickly adapt to the software-biased shift in technological opportunity. This created an unexpected and sharp increase in demand for software engineers, met at the height of the internet boom by importing more software engineers than the US was training in its own universities.⁵

Since the early 2000s, however, the US labour market has become more closed to immigration. Caps on high-skilled visas like the H-1B visa programme have grown more restrictive, and evidence from Glennon (2018) shows that these restrictive high-skilled immigration caps drove US MNCs to shift some high-skilled activity abroad in an effort to address these constraints.

Relatively liberal trade and FDI policies have allowed MNCs to address their human resource constraints by sourcing from abroad, but an open immigration regime for highly skilled workers would further ease this constraint.

• Finally, in addition to open immigration and liberal trade and FDI policies, the constraint on the supply of IT and software human capital in the US could be addressed with education policies that expand the supply of domestic IT and software human capital.

Authors’ note: We gratefully acknowledge financial support from the National Science Foundation through two grants: 1360165 and 1360170. The statistical analysis of firm-level data on US multinational companies was conducted at the Bureau of Economic Analysis (BEA), United States Department of Commerce under arrangements that maintain legal confidentiality requirements. The views expressed do not reflect official positions of the US Department of Commerce or the NSF.