Skip to main content Accessibility help
×
Hostname: page-component-8448b6f56d-gtxcr Total loading time: 0 Render date: 2024-04-23T08:44:11.191Z Has data issue: false hasContentIssue false

4 - Measuring the International Mobility of Inventors

A New Database

from Part I - The International Mobility of Inventors

Published online by Cambridge University Press:  08 June 2017

Carsten Fink
Affiliation:
World Intellectual Property Organization
Ernest Miguelez
Affiliation:
GREThA UMR CNRS 5113, Université de Bordeaux
Type
Chapter
Information
The International Mobility of Talent and Innovation
New Evidence and Policy Implications
, pp. 114 - 161
Publisher: Cambridge University Press
Print publication year: 2017

4.1 Introduction

The international mobility of knowledge workers and the associated brain-drain/brain-gain phenomena have gained prominence in public policy discussions on innovation and economic growth – in both developed and developing economies. Many governments have made efforts to attract skilled migrants from abroad – inciting what may be colloquially called a global competition for talent.

This chapter focuses on a special set of knowledge workers, namely, inventors. In particular, we introduce a new database that maps migratory patterns of inventors, extracted from information contained in patent applications filed under the Patent Cooperation Treaty (PCT). In addition to describing this newly constructed database, we provide a descriptive overview of inventor migration patterns around the world.

As described in Chapters 1 and 2, the economic importance of high-skilled migration has long been recognized in the literature, even if empirical research on the topic is of more recent vintage. Indeed, advances in our understanding of the effects of skilled worker migration to a significant extent have been due to new data becoming available over the last fifteen years. In particular, the pioneering study by Carrington and Detragiache (Reference Carrington and Detragiache1998) represents the first systematic attempt to construct a comprehensive data set on emigration rates by educational attainment. Their study provides 1990 emigration rates for sixty-one sending countries to countries of the Organization for Economic Cooperation and Development (OECD). They estimate skill levels by extrapolating the schooling levels of US immigrants by origin country to other receiving countries. Since then, other macro approaches have followed, including that of Docquier and Marfouk (Reference Docquier, Marfouk, Özden and Schiff2006), who estimate immigrant stocks in thirty OECD countries for 174 origin countries for 1990 and 2000, and Defoort (Reference Defoort2008), who extends this work by providing immigrant stocks by schooling level for five-year intervals from 1975 to 2000, but only to six OECD destination countries. Docquier et al. (Reference Docquier, Lowell and Marfouk2009) provide a gender breakdown, and Beine et al. (Reference Beine, Docquier and Rapoport2007) provide data broken down by the entry age of immigrants.

Çağlar Özden and Christopher Parsons provide a detailed overview in Chapter 2 of this book of the different data sets available from census records and describe in detail their own data work, which offers, to date, the largest available census-based data set – including numerous sending and receiving countries by gender, age, and educational attainment (see also Artuç et al. Reference Artuç, Docquier, Özden and Parsons2015). Özden and Parsons also review some of the main drawbacks of census-based data sets. Among them, it is worth highlighting two. First is the fact that the way to define educational attainment differs across OECD countries, complicating comparability, which is exacerbated when the sample includes non-OECD countries. Second, skill levels still differ markedly among skilled workers. Census-based data sets provide a skills breakdown based on three schooling levels, which offers only a rough differentiation of skills. In particular, tertiary education may include nonuniversity tertiary degrees, undergraduate university degrees, postgraduate degrees, and doctoral degrees. However, migration rates in certain skill-intensive professions – for instance, Ph.D. holders – tend to be higher than the general population. Likewise, their contribution to science and innovation in both sending and receiving countries will differ substantially from that of other tertiary-educated workers.

Recent research has shown that skilled migration, and especially that of scientists and engineers (S&Es), is the most dynamic component of total migration worldwide (Freeman Reference Freeman2010). Among them, inventors arguably are both a representative sample of high-skilled workers and a special category among them. Focusing on inventor migration as captured in patent applications constitutes an interesting and underexploited alternative to the use of more common migration stock data retrieved from censuses. It captures one specific class of high-skilled workers that is bound to be more homogeneous than the group of tertiary-educated workers as a whole. In addition, inventors have special economic importance because they create knowledge that is at the forefront of technological innovation and ultimately the genesis of technological and industrial transformation.

Patent and inventor data are increasingly exploited for migration research, as witnessed in other chapters of this book and the related literature. In particular, Agrawal et al. (Reference Agrawal, Kapur, McHale and Oettl2011) and Kerr (Reference Kerr2008) look at the relation between ethnic inventors in the United States and knowledge flows back to the ethnic inventors’ country of origin, finding relatively weak evidence of a positive relation between the two – stronger for the most valuable innovations and for certain technological fields and particular ethnic groups. At the same time, Foley and Kerr (Reference Foley and Kerr2013), Kerr and Kerr (Reference Kerr and Kerr2015), and Miguelez (Reference Miguelez2016) find stronger effects on the relationship between inventor diasporas and the formation of international co-inventorship teams – all these contributions using the same type of data source we embrace here.

Yet empirical evidence is still scarce and generally focused on a limited number of sending and receiving countries. This lack of evidence is especially serious considering the economic importance of migrant inventors, as well as the possibilities made available to researchers with patent and inventor information. For instance, inventor information can be exploited together with patent citations and information on co-inventors, thereby tracking, respectively, knowledge flows and social networks either within the same destination country or reaching back to inventors’ country of origin. After careful disambiguation of inventors’ names (see Chapter 3), it is also possible to track returnee inventors and thus explore their impact on origin countries. Foreign and native inventors can further be grouped across regions and metropolitan areas, technological sectors, and firms (especially multinationals), increasing in this way our understanding of the spatial distribution of skilled immigrants across regions, immigrants’ specialization in certain technologies, and the role of the firm in the migration process, including the business consequences of recruiting foreign talent.

Most of the inventor migration research has sought to identify the likely cultural origin of inventor names disclosed in patent data (see again Chapter 3). This approach has produced important insights. However, the cultural origin of inventor names may not always indicate recent migratory background. For example, the migration history of certain ethnicities spans more than one generation – think of Indian and Chinese immigrants in the United States or Turkish immigrants in Germany. Conversely, one may overlook immigrant inventors with names sharing the same cultural origins as the host country – think of Australian or British immigrants in the United States.

In this chapter we describe a new data set on the international mobility of inventors that overcomes many of the data limitations described so far. In particular, we make use of information on both the residence and the nationality of inventors contained in patent applications filed under the PCT. This approach offers several benefits. First, we directly rely on migratory background information revealed by inventors rather than indirectly inferring a possible migration history through the cultural origin of names. Second, patent applications filed under the PCT are less influenced by the peculiarities of national patent systems, and the underlying inventions are likely to have a larger economic value than the average national patent application. Third, PCT filing data cover a large number of countries and a long time span (from 1978 to 2012). Of course, our database shares some of the drawbacks associated with existing migration databases, and relying on patent information has drawbacks on its own, to which we will return.

The rest of this chapter is organized as follows. Section 4.2 describes the PCT system underlying our new database, and we outline, in particular, what types of information patent applications record. Section 4.3 describes the main features of our inventor migration database. In Section 4.4, we provide a descriptive analysis of inventor migration patterns as they emerge from our newly constructed database. Section 4.5 offers concluding remarks.

4.2 The PCT System as a Source of Inventor Migration Data

4.2.1 Patents and the PCT System

We derive information on the migratory background of inventors from patent applications filed under the PCT. Accordingly, we first provide some background on the patent system and especially on the PCT system, which facilitates the process of seeking patent protection in multiple jurisdictions.

A patent is the legal right of an inventor to exclude others from using a particular invention for a limited number of years. To obtain a patent right, individuals, firms, or other entities must file an application that discloses the invention to the patent office and eventually to the public. In most cases, a patent office then examines the application, evaluating whether the underlying invention is novel, involves an inventive step, and is capable of industrial application. Economic researchers have long used patent applications as a measure of inventive activity. The attraction of patent data relies on such data being available for a wide range of countries and years and for detailed technology classes (Hall Reference Hall2007). In addition, patent documents contain information on the application’s first filing date and on the applicants and inventors, including their geographic origin – down to the level of street addresses. Studies have made use of patent data to investigate the innovative behavior of firms (Griliches Reference Griliches1979; Hausman et al. Reference Hausman, Hall and Griliches1984), localized knowledge spillovers (Jaffe et al. Reference Jaffe, Trajtenberg and Henderson1993), international knowledge flows (Peri Reference Peri2005), networks of co-inventors (Breschi and Lissoni Reference Breschi and Lissoni2009; Singh Reference Singh2005), and inventor mobility (Almeida and Kogut Reference Almeida and Kogut1999; Breschi and Lissoni Reference Breschi and Lissoni2009; Miguelez and Moreno Reference Miguelez and Moreno2015).

The PCT is an international treaty administered by the World Intellectual Property Organization (WIPO) offering patent applicants an advantageous route for seeking patent protection internationally. The treaty came into force in 1978; starting with only eighteen members back then, there were 148 PCT contracting states in 2015.1

The key to understanding the PCT system’s rationale is to realize that patent rights are territorial in nature, meaning that they apply only in the jurisdiction of the patent office that grants the right. A patent applicant seeking to protect an invention in more than one country has two options. He or she can file applications directly at the patent offices in the jurisdictions in which the applicant wishes to pursue a patent – this approach is referred to as the Paris route toward international protection.2 Alternatively, the applicant can file an application under the PCT. Choosing the PCT route benefits the applicant in two main ways. First, he or she gains additional time – typically eighteen months – to decide whether to continue to seek patent protection for the invention in question and, if so, in which jurisdictions. Second, an International Searching Authority issues a report on the patent application that offers information on the potential patentability of the invention; this information can assist the applicant in deciding on whether and where to pursue the patent.3

Note that under the PCT system, the applicant still has to file applications in all jurisdictions in which he or she eventually seeks protection. An international patent right as such does not exist; the ultimate granting decision remains the prerogative of national and regional patent offices. However, the additional time gained and the first opinion on the invention’s patentability can be valuable for applicants at a relatively early stage of the patenting process, at which the commercial significance of an invention is still uncertain.4 Accordingly, applicants have opted for the PCT route for a significant share of international patent applications (see below).

For the purpose of economic analysis – including migration analysis – the PCT system has two key attractions. First, the system applies one set of procedural rules to applicants from around the world and collects information based on uniform filing standards. This reduces potential biases that would arise if one were to collect similar information from different national sources applying different procedural rules and filing standards. Working with only a single national source may be a viable alternative for studying inventor immigration behavior for a particular country, but this approach could not reliably track migrating inventors on a global basis. In any case, as will be further explained later, national patent data records generally do not offer information on both the residence and nationality of inventors.

Second, PCT patent applications are likely to capture the commercially most valuable inventions. Patenting is a costly process, and the larger the number of jurisdictions in which a patent is sought, the greater are the patenting cost. An applicant therefore will only seek a patent internationally if the underlying invention generates a sufficiently high return – higher than for patents that are only filed domestically.5 Turning to the migration angle, one may hypothesize that the most valuable patent applications emanate from the most skilled inventors, so while the focus on PCT patent applications clearly does not capture all patenting inventors, it is likely to capture the more important ones.

Before turning to how we extracted migratory background information from PCT filing data, we review a number of characteristics of the PCT system that are important to take into account when using these data for economic analysis. As already mentioned at the outset, not all countries are members of the PCT. Fortunately, the countries that have accounted for the great majority of patent filings over the past three decades – especially China, France, Germany, Japan, the Republic of Korea, the United Kingdom, and the United States – have either been founding members or joined the system before experiencing rapid patenting growth. Nonetheless, incomplete membership should be taken into account when interpreting data for different filing origins and especially when performing regression analysis.

In 2010, around 54 percent of all international patent applications went through the PCT system. The PCT share has continuously risen over the past two decades; in 1995, it stood at only 25.4 percent of all international patents (WIPO 2012a). In February 2011, the two millionth application was filed under the PCT system. However, the system has seen uneven growth since its inception in 1978. In particular, it took twenty-six years to reach the first million but only seven years to reach the second million (WIPO 2012a). Over the 1978–2011 period, the United States accounted for most filings (35.1 percent of all applications), followed by Japan (15.1 percent), Germany (11.9 percent), the United Kingdom (4.5 percent), France (4.4 percent), the Republic of Korea (3.2 percent), and China (2.9 percent).

Note that the total number of patent applications filed worldwide – at 2.14 million in 2011 – is considerably larger than the number of PCT filings – at 181,900 in the same year (WIPO 2012b). Two considerations account for this difference. First, for the majority of patents – around two-thirds in 2011 – applicants seek only domestic protection and do not apply for protection abroad. Second, each PCT filing may result in several national patent filings depending on the number of jurisdictions in which the applicant seeks protection.

While the PCT thus captures a sizable and important share of patent activity worldwide, there are considerable differences in how residents of different countries use the system. First, the propensity of patent applicants to seek protection beyond their national jurisdiction differs markedly. For instance, in 2011, residents of China filed fewer than 20,000 applications outside of China, or only 4.54 percent of all the applications by Chinese residents worldwide. In contrast, this share is considerably higher for the Republic of Korea (26.4 percent), Japan (39.1 percent), the United States (42.7 percent), Germany (57.6 percent), the United Kingdom (59.7 percent), France (62.8 percent), the Netherlands (74.7 percent), and Switzerland (78.6 percent).6

Countries also differ in the extent to which they rely on the PCT system – rather than the direct Paris route – for their international filings. Recall that in 2010 the PCT share of international filings for the world stood at around 54 percent. However, we see substantial variation around this average: the PCT share was between two-thirds and three-quarters for Finland, France, the Netherlands, Sweden, and the United States; it was between one-half and two-thirds for Australia, Germany, Russia, Switzerland, and the United Kingdom; and it was between one-quarter and one-half for Canada, China, Japan, and the Republic of Korea.

4.2.2 Information on Inventor Nationality and Residence in PCT Applications

Similar to other patent documents, PCT patent applications contain information on the names and addresses of the patent applicant(s) (generally, the owner) but also the names and addresses of the inventor(s) listed in the patent application. What is unique about PCT applications is that in the majority of cases they record both the residence and the nationality of the inventor. This has to do with the requirement under the PCT that only nationals or residents of a PCT contracting state can file PCT applications. To verify that applicants meet at least one of the two eligibility criteria, the PCT application form asks for both nationality and residence.

In principle, the PCT system only records residence and nationality information for applicants and not inventors. However, it turns out that US patent application procedures until recently required all inventors in PCT applications to also be listed as applicants. Thus, if a given PCT application included the United States as a country in which the applicant considered pursuing a patent – a so-called designated state in the application – all inventors were listed as applicants, and their residence and nationality are, in principle, available. Indeed, this is the case for the majority of PCT applications, reflecting the popularity of the United States as the world’s largest market. In addition – and fortunately for our purposes – a change to PCT rules in 2004 provided that all PCT applications automatically include all PCT member states as designated states, including the United States.

Unfortunately – for our purposes – the United States enacted changes to its patent laws under the Leahy-Smith America Invents Act (AIA) that effectively removed the requirement that inventors be also named as applicants. Starting on September 16, 2012, PCT applicants (automatically) designating the United States became free to list inventors without facing the requirement of indicating their nationality and residence – and, indeed, many applications quickly made use of this freedom. 7

In a nutshell, this means that we have good coverage of inventors’ residence and nationality information before 2004, excellent coverage from 2004 to 2011, and deteriorating coverage starting in 2012. Section 4.3 explains this in greater detail.

4.3 Data Coverage

By December 31, 2012, the total number of PCT applications stood at 2,361,455. Incorporating all the entities taking part in a PCT patent application, this figure translates into 10,725,384 records – unique combinations of patent numbers and names. This includes, for each patent application, the names of the applicants, agents, inventors, common representatives, special addresses for correspondence, and so-called applicant-inventors. Given our interest in studying the migratory background of inventors, we focus our attention only on inventor and applicant-inventor records. This subgroup accounts for exactly 6,112,608 records.

Ideally, we would like to group these 6,112,608 records along uniquely identified inventors and applicant-inventors in order to describe their migration patterns. However, the database does not provide for a single identifier for each inventor or applicant-inventor. The prior literature has disambiguated individual inventors through their names and surnames, as well as other information contained in patent documents.8 However, these approaches are far from perfect (see Raffo and Lhuillery Reference Raffo and Lhuillery2009), and the raw records on inventors and applicant-inventors already enable meaningful analysis at the aggregate (country) level or at the patent level. In particular, we can calculate immigration and emigration rates across countries and map bilateral inventor flows, whereby aggregate indicators are weighted by the productivity of inventors in terms of their number of patents. Clearly, name disambiguation would add important value to our database, though the best disambiguation approach may depend in part on the research question at hand. Indeed, we encourage other researchers to apply their own disambiguation methods to our database. In what follows, our unit of analysis will be the inventor/applicant-inventor name–patent number pair.

We observe both nationality and residence information for 4,928,076 of the 6,112,608 records, a coverage rate of 80.6 percent. The main reason for the less than complete coverage was already pointed out in Section 4.2.2: even though nationality and residence information is a compulsory field for applicants and applicant-inventors, it is not required for inventors who are not at the same time applicants. However, we observe other reasons for incomplete coverage. For some records, either the nationality field or the residence field is missing; in selected cases, both are missing. This could be due to the applicant omitting these fields in the original application or to errors in transferring information from the original patent application to the electronic filing system.9

Of the 1,184,532 records that do not offer complete nationality and residence information, 970,336 records – or 81.9 percent – relate to inventors who are not applicants; the remaining 214,196 records – or 18.1 percent – show missing or misrecorded information.

Figure 4.1 shows the availability of nationality and residence information for all inventor and applicant-inventor records from 1978 to 2012. It shows that we observe this information for the majority of records throughout the PCT system’s history. However, the coverage varies over time, standing between 60 and 67 percent during the 1990s and between 70 and 92 percent during the 2000s. It increases markedly after 2004, reflecting the PCT rule change described earlier. Unfortunately, we already observe a marked decline in the availability of nationality and residence information in 2012. As described earlier, following implementation of the AIA, PCT applications did not have to list all inventors as applicants any more as of September 16, 2012. Indeed, the incentive to not list inventors as applicants is strong because it facilitates the subsequent management of the patent; in particular, decisions such as withdrawal or reassignment of the patent only require the consent of a smaller number of parties – indeed, in most cases, there will only be a single applicant. As a consequence, the coverage of inventor nationality and residence information is bound to decline dramatically in 2013.

Figure 4.1 Coverage of nationality and residence information in PCT patents

Table 4.1 shows how the coverage of nationality and residence information differs across countries. It includes origins that account for most filings under the PCT. For the majority of countries shown, coverage lies above 90 percent, and for most others, it is above 80 percent. US applications stand out as showing the lowest coverage, of around 66 percent. This has to do with the special US filing rule discussed earlier. Before 2012, non-US PCT applications needed to list inventors as applicant-inventors if they indicated the United States as a designated state. However, US applicants generally file their applications at the US Patent and Trademark Office before submitting a PCT filing; thus, before 2004, they did not need to list the United States as a designated state. The same reason likely explains the low coverage of nationality and residence information for Canada and the Netherlands. Due to their geographic proximity, many Canadian applicants first file an application at the US Patent and Trademark Office before filing under the PCT. In the case of the Netherlands, a relatively small number of applicants account for a large share of PCT filings, and those applicants appear to have a long-standing tradition to first apply directly at the US Patent and Trademark Office.

Table 4.1 Total Records and Coverage of Nationality and Residence Information (Selected Countries)

Country/territory name Total records Records with information Records of inventors only Coverage (percent)
Austria 40,411 37,755 1,773 93.43
Australia 70,720 67,621 2,491 95.62
Belgium 46,488 41,743 4,200 89.79
Brazil 14,116 12,983 947 91.97
Canada 112,627 91,166 20,399 80.95
Switzerland 84,521 78,600 4,847 92.99
China 233,506 213,837 18,684 91.58
Germany 751,509 712,426 35,547 94.80
Denmark 46,493 42,097 4,115 90.54
Spain 51,020 48,440 2,085 94.94
Finland 64,450 59,677 4,464 92.59
France 248,541 233,372 13,030 93.90
UK 257,266 236,760 15,807 92.03
Israel 63,644 58,599 4,682 92.07
India 50,777 45,552 4,656 89.71
Italy 95,691 90,309 4,726 94.38
Japan 909,360 854,176 42,204 93.93
Netherlands 128,236 94,616 22,773 73.78
Norway 24,294 23,139 978 95.25
New Zealand 11,806 11,258 433 95.36
Russia 39,865 35,590 3,869 89.28
Sweden 114,614 101,894 12,134 88.90
Singapore 18,053 16,270 1,469 90.12
US 2,130,268 1,402,203 703,389 65.82
South Africa 10,594 10,015 502 94.53

Similar to Figure 4.1, Figure 4A.1 in Appendix 4A shows the evolution of inventor nationality and residence information for a selection of countries accounting for substantial filing shares under the PCT. Importantly, it shows that the relatively low coverage for Canada, the Netherlands, and the United States is due to pre-2004 records. From 2004 to 2011, these three countries equally show high coverage shares. In addition, all countries show a marked decline in coverage in 2012, reflecting the procedural change introduced by the AIA.

In sum, PCT records generally offer good coverage of inventor nationality and residence information and, as such, represent a promising data source for migration research. Coverage is high for all countries between 2004 and 2011. Before 2004, it is high for most countries except Canada, the Netherlands, and the United States. Unfortunately, as of September 16, 2012, the ability of PCT records to provide information on inventors’ migratory background appears seriously undermined.

4.4 Descriptive Overview

This section presents a descriptive overview of the database introduced in Section 4.3. It focuses on inventor immigration and emigration stocks and rates (see Box 4.1) in different parts of the world and for a selection of countries. It also identifies the most important bilateral migration corridors. Further, the overview looks at differences across technologies, subnational regions, and the largest applicants in each receiving country. Finally, it tests the hypotheses of outstanding contribution of migrant inventors in receiving economies, as well as whether migrant inventors in frontier knowledge economies engage with their homelands in the production of new ideas.

Box 4.1 Metrics Used in This Chapter

In this study, the stock of immigrants is defined as the number of individuals with foreign nationality residing in a given country i in a given year or period of time. For the case of this chapter, this will be the stock of immigrant inventors.

The stock of emigrants is defined as the number of people of a given nationality i residing abroad in a given year or period of time. Again, this chapter refers to the stock of emigrant inventors.

The immigration rate of a given country i in a given year is defined as the share of the foreign population over all residents of that country

IMi=immigrantsiresidentsi

The emigration rate of a given country i in a given year is defined as the share of the native population residing abroad over all nationals of that country i. To make the figures comparable to tertiary-educated emigration rates, the denominator also includes immigrant inventors residing in country i

EMi=emigrantsi(emigrantsi+residentsi)

In the migration literature, when the emigration rate is computed for tertiary-educated individuals, the resulting ratio is often termed the brain-drain rate.

4.4.1 Receiving Countries

We find exceptionally high migration rates for inventors. Recall that the prior literature has estimated a global migration rate in 2000 for the population of age twenty-five and older of 1.8 percent. It has also established that the migration rate increases with migrants’ skills; in particular, estimates suggest a 1.1 percent migration rate for the unskilled population, a 1.8 percent rate for the population with secondary education, and a 5.4 percent rate for the population with tertiary education.10 Our data, in turn, point to an inventor migration rate of 8.62 percent in 2000 – taking the skills bias in the propensity to migrate one step further.

Figure 4.2 shows the evolution of the share of inventors in PCT patent applications with migratory background for the world as a whole and for selected continents. As can be seen, the share of migrant inventors has increased steadily over time. North America stands out as seeing the highest shares of immigrant inventors relative to the continent’s population of resident inventors, followed by Oceania and the Pacific and Europe. These patterns and trends are in line with those observed for high-skilled migration more generally, whereby countries such as the United States, Canada, Australia, and New Zealand stand out as exhibiting the largest shares of immigrant workers, whereas European economies are lagging behind in attracting talent.11

Figure 4.2 Share of immigrant inventors, 1985–2010

Figure 4.3 shows the same inventor immigration shares for selected countries and confirms this point. In particular, Australia, Canada, and especially the United States stand out as the primary receiving countries relative to their population of inventors. While at the forefront of technological innovation, Germany and France have consistently seen lower inventor immigration rates. Of special interest is the United Kingdom, which has experienced a substantial increase in its share of immigrant inventors. Japan, in turn, remains the only country in this chart with an inventor immigration rate of less than 2 percent.

Figure 4.3 Share of immigrant inventors, 1990–2010

Figure 4.4 includes additional high-income economies and shows the immigration rates of inventors for the two separate time windows. The chart shows that relatively small countries see even larger immigration rates than the United States – notably, Belgium (19 percent), Ireland (20 percent), Luxembourg (35 percent), and Switzerland (38 percent). Moreover, countries such as Switzerland, Luxembourg, the Netherlands, Austria, and the United Kingdom, as well as the Scandinavian economies, have considerably increased their immigration rates in the 2000s versus their figures for the 1990s.

Figure 4.4 Immigration rates of inventors, 1991–2000 and 2001–10

Table 4.2 lists the same immigration rates as shown in Figure 4.4 and compares them with immigration rates of college graduates using Census 2000 data. It shows, first of all, a US immigration rate of college graduates far more in line with those of other large OECD countries, suggesting that the popularity of the United States is somewhat unique to inventors. More generally, it is instructive to compute the ratio between inventor immigration rates and the immigration rate of college graduates. This ratio indicates to what extent inventor and tertiary-educated immigration figures differ. The first thing to notice is that with the exception of Finland (ratio 3.88 in favor of inventors), the ratios range from 0.34 (Australia) to 1.75 (Belgium). This suggests that for the majority of countries, the estimated inventor immigration rates emerging from the PCT data are broadly consistent with census data. At the same time, smaller countries, similar to the United States, seem to be disproportionately popular among inventors compared to college graduates (ratio larger than 1.25). This is the case for Belgium, Denmark, Switzerland, and especially Finland.

Table 4.2 Immigration Rates of Inventors and College Graduates

Country Immigration rate, 1991–2000 Immigration rate, 2001–2010 Immigration rate college Ratio (a)/(c) Ratio (b)/(c)
(a) (b) (c) (d) (e)
Australia 10.89 11.20 33.17 0.33 0.34
Austria 8.80 12.45 14.33 0.61 0.87
Belgium 16.89 18.56 10.61 1.59 1.75
Canada 11.16 11.03 25.84 0.43 0.43
Denmark 5.07 9.98 8.00 0.63 1.25
Finland 2.93 8.74 2.25 1.30 3.88
France 5.12 6.32 12.38 0.41 0.51
Germany 3.76 5.54 11.39 0.33 0.49
Ireland 17.38 19.89 18.07 0.96 1.10
Italy 3.88 3.27 6.11 0.64 0.54
Japan 0.87 1.15 1.05 0.83 1.09
Luxembourg 23.14 35.42 49.04 0.47 0.72
Netherlands 7.80 13.77 11.36 0.69 1.21
New Zealand 14.72 16.60 24.85 0.59 0.67
Norway 4.96 9.17 8.09 0.61 1.13
R. of Korea 0.59 0.90 0.88 0.67 1.02
Spain 5.95 6.72 6.38 0.93 1.05
Sweden 4.61 8.44 14.26 0.32 0.59
Switzerland 28.45 38.41 28.38 1.00 1.35
UK 7.17 11.62 16.00 0.45 0.73
US 16.07 18.18 13.86 1.16 1.31

4.4.2 Which Are the Largest Sending Countries?

We next turn to inventor emigration patterns and trends. Recall that the prior literature has estimated a 5.4 percent global migration rate for tertiary-educated workers. However, this figure hides considerable variation in emigration propensities across continents: in high-income countries, the emigration rate stood at 3.6 percent versus 7.3 percent for low- and middle-income countries. It was much higher for least developed countries (13.1 percent) and for small island developing states (42.4 percent).12

These differences turn out to be even more marked when looking at inventor data. The global share of inventors with migratory backgrounds stood at 7.46 percent from 1991 to 2000 and at 9.94 percent from 2001 to 2010. However, the emigration rate of high-income countries for these two time periods stood at only 4.99 and 5.92 percent, respectively.13 It was much higher for low- and middle-income countries – standing at 41.73 and 36.40 percent, respectively.14

Table 4.3 provides top-thirty lists of immigrant and emigrant counts for the time period 2001–10, respectively. Unsurprisingly, the top-thirty immigrant list consists mostly of high-income economies, probably reflecting the attractive employment, education, research, and entrepreneurship opportunities offered by these economies. Interestingly, most high-income countries also show sizable diasporas abroad, although China and India come out as the top two inventor-sending countries.

Table 4.3 Immigrants, Emigrants, and Emigration Rates, Time Window 2001–10

Country/territory Immigrants Nationals Country/territory Emigrants Residents
US 194,609 875,962 China 53,610 141,902
Germany 25,341 432,136 India 40,097 38,486
Switzerland 20,416 32,737 Germany 32,158 457,477
UK 15,758 119,824 UK 27,746 135,582
Netherlands 9,665 60,513 Canada 21,315 65,808
France 9,540 141,413 France 19,123 150,953
Canada 7,257 58,551 US 11,131 1,070,571
Singapore 6,720 6,311 Italy 9,820 62,973
Japan 6,715 578,101 Netherlands 9,132 70,178
Belgium 5,042 22,122 R. of Korea 9,127 164,078
Sweden 4,832 52,451 Russia 7,878 20,561
Australia 4,427 35,088 Japan 6,986 584,816
China 4,251 137,651 Australia 5,631 39,515
Austria 3,113 21,896 Spain 5,154 35,786
Finland 3,095 32,314 Austria 5,122 25,009
Denmark 2,589 23,364 Sweden 4,025 57,283
Spain 2,406 33,380 Israel 3,668 42,001
Italy 2,060 60,913 Belgium 3,567 27,164
Ireland 1,689 6,803 Greece 3,209 2,025
R. of Korea 1,472 162,606 Turkey 3,119 6,202
N. Zealand 1,249 6,277 Switzerland 3,005 53,153
Norway 1,245 12,327 Ireland 2,686 8,492
Israel 694 41,307 Malaysia 2,682 4,154
S. Arabia 569 524 Romania 2,589 771
India 532 37,954 Poland 2,537 4,559
Malaysia 524 3,630 Denmark 2,411 25,953
South Africa 426 6,355 Iran 2,253 76
Brazil 376 9,050 Ukraine 1,911 2,464
Luxembourg 322 587 Brazil 1,859 9,426
UAE 273 54 N. Zealand 1,839 7,526

Note: The last column shows the emigration rates only if the country has at least ten nationals (both abroad and residents).

It is also worth looking at the net balance of immigrant and emigrant inventors for selected countries. Figure 4.5 shows for the 2001–10 period the number of immigrant and emigrant inventors and order countries according to their net immigration position. Again, the United States stands out in showing by far the largest immigration surplus; indeed, there are more than fifteen times as many immigrant inventors in the United States as there are US inventors residing abroad. By contrast, Canada and the three largest European economies – France, Germany, and the United Kingdom – see negative net immigration positions. The cases of Germany and the United Kingdom are especially interesting because they host considerable numbers of immigrant inventors, but even greater numbers of German and UK inventors reside abroad.

Figure 4.5 Net migration position, 2001–10

When looking at relative emigration rates – which take into account the size of the local inventor endowments – low- and middle-income countries dominate the top-thirty list, especially small and African economies. Figure 4.6 shows emigration rates – or brain-drain rates – in a map for the same time period. The map confirms that low- and middle-income countries and especially African economies are the most severely affected by inventor brain drain.

Figure 4.6 Brain-drain rates, 2001–10

4.4.3 Identifying the Largest Migration Corridors

Due to the bilateral nature of our data, we can identify the main inventor migration corridors. The left-hand side of Table 4.4 lists the thirty most important corridors for the 2001–10 period. These thirty corridors account for only 0.08 percent of country pairs in our data set. However, they represent 58.70 percent of overall migration counts for the whole period. In other words, inventor migration is a phenomenon that is highly concentrated among a relatively small number of countries. The United States appears most frequently in this list as a destination country.15

Table 4.4 Largest Inventor Migration Corridors, 2001–10

Largest inventor migration corridors Largest inventor migration corridors, limited to non-OECD sending countriesa
Origin Destination Counts Origin Destination Counts
China US 44,452 China US 44,452
India US 35,621 India US 35,621
Canada US 18,734 Russia US 4,339
UK US 14,893 China Japan 2,510
Germany US 10,297 China Singapore 1,923
Germany Switzerland 8,198 Turkey US 1,922
R. of Korea US 7,267 Iran US 1,438
France US 6,543 Romania US 1,220
Japan US 5,045 Russia Germany 1,207
Russia US 4,339 Mexico US 1,161
Australia US 3,241 Brazil US 1,115
Israel US 2,966 Malaysia Singapore 1,090
France Switzerland 2,747 Ukraine US 977
Netherlands US 2,698 China UK 920
Austria Germany 2,672 China Germany 892
France Germany 2,607 India Singapore 847
China Japan 2,510 Argentina US 820
Italy US 2,501 Singapore US 775
Germany Netherlands 2,285 Malaysia US 729
Netherlands Germany 2,138 South Africa US 719
France UK 2,044 Egypt US 667
UK Germany 2,043 China Canada 652
China Singapore 1,923 Bulgaria US 626
Turkey US 1,922 Pakistan US 626
Germany Austria 1,829 Turkey Germany 601
Germany UK 1,612 India UK 556
Germany France 1,609 India Germany 542
Spain US 1,559 Colombia US 532
UK Switzerland 1,555 Thailand US 494
Italy Switzerland 1,536 Philippines US 450

a We include Mexico and Chile – as the only middle-income OECD countries – among the sending countries.

The right-hand columns in Table 4.4 list the thirty most important corridors for which the sending country is not an OECD member. This allows us to look more carefully at south-north migration and possibly also south-south migration. The United States emerges by far as the most frequently listed destination country in both periods. Germany is the only continental European country appearing on this list, confirming the earlier finding that European countries lag behind in attracting inventors from non-OECD countries (Docquier and Rapoport Reference Docquier and Rapoport2009). Interestingly, Singapore – despite its relatively small size – appears several times as a destination country on this list, with China, India, and Malaysia as the most important inventor origins.

Table 4.5 lists all the bilateral country pairs where the ratio of the flow from origin to destination over the reverse flow is between 0.5 and 2; it orders pairs by the sum of the two flows for two different time windows – 1991–2000 and 2001–10. The corridors listed can be considered as having fairly balanced inventor migration flows. The resulting flows appear to reflect in large part the establishment of a single labor market in Europe.16 Aside from EU corridors, other interesting corridors that feature in the top-thirty list include United States-Israel (1991–2000), Switzerland-United States, China-Germany, and Singapore-United States. Interestingly, China features in several of these corridors in the second period, witnessing the rise of the country not only as a source of inventors for other countries but also as a host for inventors from many other economies – especially other Asian and European economies.

Table 4.5 Largest Bilateral Migration Corridors, 1991–2000 and 2001–10

Largest dual-direction migration corridors, 1991–2000 Largest dual-direction migration corridors, 2001–10
Origin (A) Destination (B) A → B B → A Origin (A) Destination (B) A → B B → A
UK Germany 780 476 Austria Germany 2,672 1,829
France UK 513 435 Germany Netherlands 2,285 2,138
Germany France 432 403 France Germany 2,607 1,609
Israel US 522 273 UK Germany 2,043 1,612
Belgium France 373 330 France UK 2,044 1,121
Netherlands Germany 384 296 Switzerland US 1,348 734
Ireland UK 419 210 UK Australia 977 609
UK Netherlands 304 205 Netherlands Belgium 890 535
Germany Belgium 290 147 Ireland UK 808 568
Italy UK 225 146 China Germany 892 468
UK N. Zealand 180 98 Singapore US 775 518
Italy France 177 100 Netherlands France 644 580
UK Sweden 164 84 Germany Belgium 694 406
Denmark UK 120 102 China Canada 652 387
France Netherlands 98 86 Japan Germany 502 280
Japan Germany 83 81 UK N. Zealand 418 342
Norway Sweden 75 56 Spain France 420 304
Singapore US 65 52 Germany Denmark 402 292
Japan UK 73 39 Sweden Denmark 377 250
Ireland Germany 54 53 UK Sweden 363 251
Netherlands Sweden 67 39 UK Denmark 367 214
Sweden France 58 40 Australia China 327 246
Finland UK 50 47 Finland Sweden 317 182
Germany S. Africa 54 42 Germany Finland 264 188
Canada Japan 61 33 Japan UK 255 175
Australia Canada 54 39 France China 211 183
UK Singapore 54 39 Sweden Norway 196 179
Germany Finland 48 42 UK Norway 238 119
Israel UK 57 31 S. Africa UK 172 128
Canada Switzerland 54 31 Ireland Germany 149 141

4.4.4 Do Migrant Inventors Differ across Technological Fields?

This section explores differences in inventor migration patterns across technology domains. This is partly motivated by previous research that has found that immigrants’ contribution to their host countries’ productivity is mainly driven by those specializing in specific sectors that happen to be more productive – the so-called composition effect (Hunt and Gauthier-Loiselle Reference Hunt and Gauthier-Loiselle2010). In light of these claims, this section provides some initial insights into differences in inventor mobility patterns across different technology sectors. It follows Schmoch’s (Reference Schmoch2008) classification of International Patent Classification (IPC) codes into thirty-five technology fields and groups them into five broad sectors – namely, electrical engineering, instruments, chemistry, mechanical engineering, and others (see Table 4A.1 in Appendix 4A).17

Figure 4.7 looks at the migration rate of inventors across sectors over time. As is apparent from this figure, immigrant inventors’ contribution to patenting differs markedly across technology fields. Electrical engineering and chemistry emerge as the most important technology fields. The case of electrical engineering – audiovisual technology, telecommunications, digital communications, computer technology, IT methods, semiconductors, and so on – is especially remarkable, showing a sudden jump in its migration rate around 2003–4.18

Figure 4.7 Inventor immigration rates over time by field of technology: three-year moving averages

Figure 4.8 reports inventor immigration rates for selected technology fields for a number of countries.19 Generally, countries such as Switzerland, the Netherlands, and the United States had high inventor immigration rates in all the reported fields for the 2006–10 period. In contrast, China, India, and Japan reported low inventor immigration rates for the same period. However, across countries and technology fields, there were considerable variations in inventor immigration rates.

Figure 4.8 Inventor immigration rates, selected fields and countries, 2006–10

4.4.5 Which Regions Attract Knowledge Workers?

One striking aspect of immigration, and particularly skilled immigration, is that migrants tend to concentrate in specific geographic areas within countries. In particular, immigrant inventors appear to cluster in metropolitan areas, thus contributing to the spatial concentration of inventive activity. This issue is analyzed by matching PCT applications with OECD’s REGPAT database (Maraut et al. Reference Maraut, Dernis, Webb, Spiezia and Guellec2008; refer to Miguelez and Raffo Reference Miguelez and Raffo2013 for details of the matching procedure).20 By linking inventor nationality information with REGPAT, it is possible to study the settlement patterns of immigrant inventors within countries beyond the settlement patterns of native inventors.

Table 4.6 ranks the top thirty European NUTS2 regions in terms of stocks of migrant inventors from 2001 to 2010 (left-side columns).21 As can be seen, in absolute numbers, regions of the core of Europe attract large numbers of inventors from other countries. However, this is partially related to their size and their innovative capacity. The right-side columns normalize these numbers using the number of resident inventors in each region. As is shown, some regions, particularly Swiss regions, are high in both rankings. Interestingly, the Swiss region of Nordwestschweiz leads both rankings. Recall that Switzerland was the country with the largest share of foreign inventors among the OECD countries. Luxembourg, Ireland, and Belgium also ranked high, which regional figures also reflect – the regions of these four countries dominate the ranking. Other important poles of attraction are London, Wien, and the Dutch region of Noord-Brabant, where the Phillips facilities are located.

Table 4.6 Top Thirty European NUTS2 Regions by Immigration Stocks and Rates, 2001–10

Country NUTS2 region Immigrant count Country NUTS2 region Immigration rate
CH Nordwestschweiz 6,733 CH Nordwestschweiz 0.470
NL Noord-Brabant 6,014 CH Rég. Lémanique 0.460
CH Rég. Lémanique 4,219 BE Bruxelles 0.409
FR Île de France 3,895 CH Zurich 0.391
CH Zurich 3,777 LU Luxembourg 0.339
DE Oberbayern 3,049 CH Zentralschweiz 0.334
DE Karlsruhe 2,734 CH Ostschweiz 0.299
DE Köln 2,473 GB Inner London 0.261
SE Stockholm 2,331 BE Brabant Wallon 0.248
GB East Anglia 2,286 CH Ticino 0.239
DE Darmstadt 2,275 IE Southern and Eastern 0.206
GB Inner London 2,264 BE Prov. Antwerpen 0.194
DE Rheinhessen-Pfalz 2,181 CH Espace Mittelland 0.194
FI Etelä-Suomi 2,037 BE Vlaams-Brabant 0.190
DE Düsseldorf 1,920 NL Noord-Brabant 0.179
GB Berkshire, Buckinghamshire and Oxfordshire 1,913 BE Prov. Hainaut 0.171
DK Hovedstaden 1,707 GB Outer London 0.171
DE Stuttgart 1,688 AT Wien 0.170
FR Rhône-Alpes 1,685 BE Prov. Luxembourg 0.167
CH Espace Mittelland 1,460 GB East Anglia 0.163
CH Ostschweiz 1,398 DK Nordjylland 0.158
NL Zuid-Holland 1,287 GB Eastern Scotland 0.153
IE Southern and Eastern 1,122 AT Tirol 0.151
ES Cataluña 1,098 BE Prov. Liège 0.148
BE Bruxelles 1,085 AT Kärnten 0.148
DE Berlin 1,051 ES Illes Balears 0.145
BE Prov. Antwerpen 1,012 IE Border, Midland And Western 0.144
CH Zentralschweiz 1,000 GB Berkshire, Buckinghamshire and Oxfordshire 0.144
NL Noord-Holland 957 GB Northern Ireland 0.142
AT Wien 921 NL Noord-Holland 0.141

Note: For calculation of regional immigration rates, regions with fewer than thirty resident inventors for the period 2001–10 are not displayed.

Table 4.7 repeats the same exercise as before but for the case of Metropolitan Statistical Areas (MSAs) of the United States. In terms of immigrant inventor counts, MSAs are generally larger than European NUTS2 regions – as they are in terms of total inventor counts. Leading the ranking we see some of the biggest and most innovative MSAs, as expected – San Diego, San Jose, New York, San Francisco, and Boston. When one looks at the ratio of immigrant inventors in MSAs, San Diego and San Jose still rank quite high – San Diego leads both rankings. In comparison with Table 4.6, one can see that the top four European regions attract more talented individuals (in relative terms) than San Diego. However, while the share of immigrant inventors in NUTS2 European regions drops rapidly over the ranking, a large number of MSAs show immigration ratios over 20 percent. That is to say, immigrant inventors’ settlement in European regions seems to be more skewed than in the case of the United States.

Table 4.7 Top Thirty US MSAs by Immigration Stocks and Rates, 2001–10

MSA Immigrant counts MSA Immigration rate
San Diego 20,752 San Diego 0.351
San Jose-Santa Clara 20,386 Evansville 0.321
New York 17,396 San Jose-Santa Clara 0.307
San Francisco 15,246 Stockton 0.303
Boston 14,753 Trenton 0.296
Los Angeles 6,500 Champaign-Urbana 0.294
Philadelphia 6,167 New Haven 0.285
Chicago 6,001 Albany 0.282
Houston 5,742 Lansing-East Lansing 0.263
Dallas 3,593 Ithaca 0.262
Washington 3,523 Ann Arbor 0.255
Minneapolis 2,921 Gainesville 0.255
New Haven 2,608 Athens 0.249
Seattle 2,514 College Station-Bryan 0.248
Trenton 2,248 Columbus 0.246
Portland 2,231 Santa Barbara 0.238
Atlanta 2,013 New York 0.237
Detroit 1,776 Dallas 0.226
Albany 1,755 San Francisco 0.224
Austin 1,722 Boston 0.223
Raleigh-Cary 1,598 Greensboro-High Point 0.221
Durham-Chapel Hill 1,565 Ames 0.219
Phoenix-Mesa-Glendale 1,556 State College 0.212
Ann Arbor 1,467 Portland-Vancouver-Hillsboro 0.210
Baltimore-Towson 1,399 Columbia 0.209
Hartford-West Hartford-East Hartford 1,224 Lafayette 0.206
Cincinnati-Middletown 1,189 Lexington-Fayette 0.198
Bridgeport-Stamford-Norwalk 1,161 Fayetteville-Springdale-Rogers 0.196
Indianapolis-Carmel 1,106 Sacramento–Arden-Arcade–Roseville 0.194
Worcester 1,099 Little Rock-North Little Rock-Conway 0.192

Note: For calculation of regional immigration rates, the areas with fewer than 300 resident inventors for the period 2001–10 are not displayed.

4.4.6 Migrant Inventors and the Role of Firms

Inventor immigration rates differ not only across countries and regions but also across different types of applicants. For example, Table 4.8 lists the immigration rates for the top ten PCT applicants – based on the residence of the first-named applicant for the 2006–10 period for a selection of countries. It shows that the distribution of immigrant inventors was very uneven across applicants, even between enterprises of a relatively similar size. In France, for example, France Telecom’s rate of immigrant inventors was between four and five times greater than that of Peugeot-Citroen – an imbalance that cannot be attributed solely to differences across technology fields. Peugeot-Citroen had an immigration rate that was more than ten times greater than that of Renault SAS. One interesting aspect of the data highlighted in Table 4.8 is the role played by universities and public research centers in the recruitment of talent from abroad. The top patenting universities and public research centers feature some of the highest inventor immigration rates among the top PCT applicants. This is the case for the University of California in the United States, for example, and also for Cambridge University, Imperial Innovations (Imperial College London), and Isis Innovation (Oxford University) in the United Kingdom, among others.

Table 4.8 Inventor Immigration Rates for Top Ten Applicants, Selected Countries, 2006–10

Applicant name Immigration rate Patents Inventors Applicant name Immigration rate Patents Inventors
United States Germany
Qualcomm Incorporated 50.8 6,528 19,907 Robert Bosch Corporation 2.8 6,480 17,484
Microsoft Corporation 57.4 3,020 11,297 Siemens Aktiengesellschaft 6.4 4,555 11,753
3 M Innovative Properties Company 11.0 2,577 8,852 Basf Se 14.4 3,562 15,427
Hewlett-Packard Development Company, L.P. 18.6 2,360 6,114 Bosch-Siemens Hausgerate Gmbh 3.2 1,679 4,575
E.I. Dupont De Nemours and Company 17.0 2,118 5,916 Fraunhofer-Gesellschaft Zur Forderung Der Angewandten Forschung E.V. 5.4 1,532 5,521
International Business Machines Corporation 21.4 2,006 6,854 Continental Automotive Gmbh 8.6 1,337 3,447
University of California 28.2 1,754 5,598 Henkel Kommanditgesellschaft Auf Aktien 6.4 1,210 4,420
Motorola, Inc. 23.4 1,573 4,488 Daimler Ag 3.8 1,196 3,601
Procter & Gamble Company 10.2 1,540 4,953 Evonik Degussa Gmbh 5.6 974 4,103
Baker Hughes Incorporated 12.8 1,461 3,552 Zf Friedrichshafen Ag 2.4 958 2,702
Switzerland United Kingdom
Nestec S.A. 56.4 619 1,781 Unilever Plc 10.4 594 1,536
F. Hoffmann-La Roche Ag 46.6 564 1,385 Glaxo Group Limited 12.6 409 1,590
Novartis Ag 62.6 489 1,179 British Telecommunications Public Limited Company 20.2 389 861
Syngenta Participations Ag 66.6 308 972 Bae Systems Plc 3.2 305 644
Actelion Pharmaceuticals Ltd 30.2 272 879 Imperial Innovations Ltd. 29.8 246 648
Alstom Technology Ltd 67.6 212 506 Isis Innovation Limited 29.8 242 618
Abb Research Ltd 65.0 201 529 Dyson Technology Limited 10.4 237 579
Swiss Federal Institute of Technology 49.2 186 534 Astrazeneca UK Limited 8.2 210 640
Sika Technology Ag 30.4 179 426 Cambridge University 36.6 205 572
Inventio Ag 23.6 174 338 Qinetiq Limited 2.2 185 458
Singapore France
Agency of Science, Technology and Research 62.2 791 2,690 Centre National De La Recherche Scientifique (CNRS) 8.0 1,892 7,002
National University of Singapore 57.6 213 735 Commissariat A L’Energie Atomique Et Aux Energies Alternatives 2.6 1,514 4,240
Nanyang TechnologicaL University 61.4 148 474 Renault S.A.S. 0.2 1,065 2,357
Creative Technology Ltd 21.6 88 217 France Telecom 11.6 963 2,188
Nanyang Polytechnic 23.0 74 166 L’oreal 1.8 849 1,730
Singapore Health Services Pte Ltd 37.4 35 160 Peugeot Citroen Automobiles Sa 2.4 772 1,502
Temasek Life Sciences Laboratory Limited 70.6 28 78 Thales Ultrasonics Sas 0.4 626 1,473
Razer (Asia-Pacific) Pte Ltd 4.6 27 44 Institut National De La Sante Et De La recherche Medicale (INSERM) 9.2 517 1,633
Siemens Medical Instruments Pte. Ltd. 25.0 27 76 Arkema 3.4 506 1,279
S*Bio Pte Ltd 77.6 17 49 L’air Liquide Société Anonyme Pour l’etude Et L’exploitation Des Procedes Georges Claude 5.0 471 1,332

4.4.7 Testing the Outstanding Contribution of Foreign Inventors

PCT-based inventor immigration data can offer a perspective on an ongoing debate in both the academic literature and journalistic discussions on the extent of foreign researchers’ contribution to scientific advancement and innovation. In the United States, some scholars remain skeptical about immigrants’ contribution to overall economic performance (Borjas Reference Borjas1999). Others have found strong evidence for a positive and important role played by skilled immigrants on receiving countries’ economic development.

In order to investigate the contribution of immigrants in their host country economy, it is insightful to explore the number of citations received by PCT applications with and without migrating inventors. The economic literature has used the number of citations as a measure of a patent’s underlying quality. Table 4.9 presents the share of all patents with at least one listed inventor with migratory background residing in the top twenty largest receiving countries – for all the years – and compares it with the share of inventors with migratory background listed in breakthrough patents – defined as the top 5 percent of patents in terms of forward citations received, by priority year and technology (five IPC broad technologies).

Table 4.9 Share of Immigrants in Highly Cited Patents, All Years

Country Percent foreigners in all patents Percent foreigners in most-cited patents z
US 30.9 41.3 46.5***
Germany 10.9 14.4 14.9***
Switzerland 47.6 54.5 9.1***
UK 16.1 20.1 12.2***
France 10.8 14.7 10.1***
Netherlands 20.3 23.2 5.1***
Canada 19.8 23.8 6.3***
Japan 2.7 3.5 7.2***
Singapore 66.3 73.5 3.9***
Australia 17.6 20.3 4.1***
Belgium 28.6 34.2 5.8***
Sweden 10.9 16.3 12.1***
China 6.7 16.8 15.2***
Austria 16.4 21.3 5.0***
Finland 11.1 16.3 8.7***
Denmark 14.3 17.0 4.1***
Spain 13.4 19.0 5.4***
Italy 5.7 7.5 4.7***
Ireland 32.7 37.6 2.4**
R. of Korea 2.4 2.7 1.0

*** p < 0.1; ** p < 0.5; * p < 0.10.

As can be seen, the proportion of immigrants is systematically larger among breakthrough inventions than among the whole universe of PCT patents. This supports the idea that immigrants disproportionately contribute to their host country productivity – measured here by citations received, even after controlling for time and technology differences. Note that the differences are statistically significant in most cases (see the last column in Table 4.9) except for the Republic of Korea.

4.4.8 Do Foreign Inventor Diasporas Engage with Their Homelands?

Despite the adverse consequences of the brain drain of high-skilled people on a country’s development potential, it is also well recognized that emigrants do not necessarily sever their ties with their homelands, and as diasporas, they may constitute a valuable resource in terms of accessing foreign knowledge and technologies. One way to obtain insight into such diaspora-homeland links is to analyze how extensively immigrant inventors collaborate with their conational colleagues at home. To explore this empirically, we assemble all PCT applications for which one inventor resides in the United States and another inventor resides outside the United States regardless of inventors’ nationality. We refer to this set of patents as global collaborative patents (Kerr and Kerr Reference Kerr and Kerr2015). We focus on the United States, which is arguably the world’s most technologically advanced country.

Focusing on the 2001–10 period, we look at global collaborative patents in two ways. First, we identify the nationality of the inventor(s) residing in the United States and calculate the shares attributable to the main origin nationalities, as shown in column 1 of Table 4.10. Thus 13.49 percent of global collaborative patents include US-resident inventors of Chinese nationality, 10.37 percent include US-resident inventors of Indian nationality, 15.44 percent include US-resident inventors of Canadian nationality, and so on. Note that these patents often include inventors of multiple nationalities; therefore, adding up all the percentages – including those not listed in Table 4.10 – would result in a value greater than 100. In addition, the US nationality – not shown in Table 4.10 – is represented in 81.65 percent of these global patents.

Table 4.10 Share of International Copatents Including Conationals, 2001–10

Origin country (1) (2) (3)
Global copatents with foreign inventors (percent) Bilateral copatents with foreign inventors (percent) z
China 13.49 24.20 16.20***
India 10.37 30.65 26.60***
Canada 15.44 12.79 −5.15***
UK 18.33 13.70 −8.99***
Germany 19.70 19.90 0.40
R. of Korea 3.36 24.30 33.13***
France 10.87 17.36 11.71***
Japan 7.10 22.67 29.93***
Russia 2.77 21.38 27.77***
Australia 3.71 13.42 17.31***
Israel 3.97 25.06 38.51***
Netherlands 5.44 6.07 1.17
Italy 4.17 6.04 3.18***
Turkey 0.79 6.03 6.33***
Spain 2.16 8.81 11.10***
Sweden 2.98 9.06 11.61***
Switzerland 3.21 2.48 −1.86*

*** p < 0.1; ** p < 0.5; *p < 0.10.

Second, we identify cases whereby the nationality of the US-residing inventor coincides with the country of residence of the inventor outside the United States and calculate the share of those patents in all bilateral collaborative patents that involve the United States and the origin country in question. If foreign inventors in the United States were not especially engaged with their homelands, we would expect the resulting shares to be similar to the ones shown in column 1 of Table 4.10. However, if they are more inclined to collaborate with inventors in their homelands, they would be overrepresented in bilateral collaborative patents, and we should observe a higher share. Indeed, column 2 of Table 4.10 reveals higher shares for the majority of nationalities, and in most cases, the differences are statistically significant based on the test of proportions. For example, while 13.49 percent of all global patents between the United States and other countries include US-resident inventors of Chinese nationality, this proportion almost doubles to 24.20 percent when we focus only on collaborative patents between the United States and China. Only US-residing inventors with Canadian, German, Dutch, Swiss, and UK nationalities do not show any special engagement with their respective homelands; other linkages above and beyond high-skilled migration may explain this result – notably, cultural linkages as well as the role of multinational corporations (Breschi et al. Reference Breschi, Lissoni and Miguelez2017).

4.5 Conclusion

This chapter describes a new global data set on migrant inventors that we built using information on inventor nationality and residence available in PCT applications. By using patent data to map the migratory patterns of high-skilled workers, we can overcome some of the limitations faced by existing data sets on the world’s migrant population.

In particular, our database covers a long time period, provides information on an annual basis, and includes a large number of sending and receiving countries. By focusing on inventors, we capture a group of high-skilled workers of special economic importance and with more homogeneous skills than tertiary-educated workers as a whole. Our data set relies on the PCT system, which applies a uniform set of procedural rules worldwide and has close to universal coverage – promoting the cross-country comparability of our data. In addition, patents filed under the PCT system are likely to include the most valuable inventions, as revealed in the willingness of applicants to potentially bear the patenting costs in multiple jurisdictions.

Of course, using patent data for economic analysis does not come without limitations. One important caveat is that we only observe inventors when they seek patents. However, not all inventions are patented; indeed, the propensity to patent for each dollar invested in R&D differs considerably across industries.22 In addition, there is no one-for-one correspondence between the number of patent applications filed and the commercial value of the underlying inventions or their contribution to technological progress. Studies have documented a skewed distribution of patent values, with relatively few patents yielding high economic returns.23 Similarly, as this chapter has pointed out, the propensity to patent abroad – and in particular through the PCT route – differs across countries, affecting the selection of inventors included in our data set.

As is the case for most other migration data sets, we can only identify inventors with migratory background, but we do not know where those inventors were educated. Anecdotal evidence suggests, for example, that many immigrant inventors in the United States received a scientific degree from US universities – although such cases may still involve a “drain of brains.” Another limitation is that our data set misses inventors with migratory backgrounds who have become nationals of their host countries. To the extent that it is easier to gain citizenship in some countries than in others, this introduces a bias in our data. A related bias stems from the possibility that migrants of some origins may be more inclined to adopt the host country’s nationality than migrants from other origins. Unfortunately, our data do not allow us to assess the severity of these biases. Researchers using our data should be aware of these limitations, especially when drawing policy conclusions.

Notwithstanding these caveats, we believe that our new database meaningfully captures a phenomenon of growing importance. Indeed, the descriptive overview presented in this chapter suggests that our database is consistent with migratory patterns and trends as they emerge from census data. At the same time, our database opens new avenues for research, promising to generate fresh empirical insights that can inform both innovation policy and migration policy.

Footnotes

We are indebted to Matthew Bryan, Julio Raffo, and participants at the WIPO Experts Meeting on Intellectual Property, the International Mobility of Knowledge Workers and the Brain Drain (Geneva, April 29–30, 2013) for valuable feedback and suggestions. However, any mistakes or omissions remain our own. The views expressed in this chapter are those of the authors and do not necessarily reflect the views of the World Intellectual Property Organization or its member states. All the data used and described in this chapter can be downloaded from the WIPO website at www.wipo.int/econ_stat/en/economics/publications.html (accessed June 14, 2016).

References

Agrawal, A., Kapur, D., McHale, J., and Oettl, A. (2011), “Brain drain or brain bank? The impact of skilled emigration on poor-country innovation,” Journal of Urban Economics, 69(1): 4355.Google Scholar
Almeida, P., and Kogut, B. (1999), “Localization of knowledge and the mobility of engineers in regional networks,” Management Science, 45(7): 905–17.Google Scholar
Artuç, E., Docquier, F., Özden, Ç., and Parsons, C. (2015), “A global assessment of human capital mobility: the role of non-OECD destinations,” World Development, 65: 626.Google Scholar
Beine, M., Docquier, F., and Rapoport, H. (2007), “Measuring international skilled migration: a new database controlling for age of entry,” World Bank Economic Review, 21(2): 249–54.CrossRefGoogle Scholar
Bertoli, S., Brücker, H., Facchini, G., Mayda, A. M., and Peri, G. (2012), “Understanding highly skilled migration in developed countries: the upcoming battle for brains,” in Boeri, T., Brücker, H., Docquier, F., and Rapoport, H. (eds.), Brain Drain and Brain Gain: The Global Competition to Attract High-Skilled Migrants, Oxford University Press, pp. 15198.Google Scholar
Borjas, G. J. (1999), “The economic analysis of immigration,” in Handbook of Labor Economics, New York, Elsevier, pp. 1697–760.Google Scholar
Breschi, S., and Lissoni, F. (2009), “Mobility of skilled workers and co-invention networks: an anatomy of localized knowledge flows,” Journal of Economic Geography, 9(4): 439–68.CrossRefGoogle Scholar
Breschi, S., Lissoni, F., and Miguelez, E. (2017), “Foreign-origin inventors in the USA: testing for diaspora and brain gain effects,” Journal of Economic Geography (in press). doi:10.1093/jeg/lbw044CrossRefGoogle Scholar
Carrington, W., and Detragiache, E. (1998), “How big is the brain drain?,” Brussels, International Monetary Fund.Google Scholar
Defoort, C. (2008), “Tendances de long terme des migrations internationales : analyse à partir des six principaux pays receveurs,” Population, 63(2): 317–51.Google Scholar
Docquier, F., Lowell, B. L., and Marfouk, A. (2009), “A gendered assessment of highly skilled emigration,” Population and Development Review, 35(2): 297321.Google Scholar
Docquier, F., and Marfouk, A. (2006), “International migration by education attainment (1990–2000),” in Özden, Ç. and Schiff, M. (eds.), International Migration, Remittances and the Brain Drain, London, Palgrave Macmillan, pp. 151–99.Google Scholar
Docquier, F., and Rapoport, H. (2009), “Documenting the brain drain of ‘la crème de la crème,’Journal of Economics and Statistics (Jahrbuecher Fuer Nationaloekonomie Und Statistik), 229(6): 679705.Google Scholar
Docquier, F., and Rapoport, H. (2012), “Globalization, brain drain, and development,” Journal of Economic Literature, 50(3): 681730.CrossRefGoogle Scholar
Foley, C. F., and Kerr, W. R. (2013), “Ethnic innovation and U.S. multinational firm activity,” Management Science, 59(7): 1529–44.Google Scholar
Freeman, R. B. (2010), “Globalization of scientific and engineering talent: international mobility of students, workers, and ideas and the world economy,” Economics of Innovation and New Technology, 19(5): 393406.Google Scholar
Griliches, Z. (1979), “Issues in assessing the contribution of research and development to productivity growth,” Bell Journal of Economics, 10(1): 92116.Google Scholar
Guellec, D., and Van Pottelsberghe de la Potterie, B. (2002), “The value of patents and patenting strategies: countries and technology areas patterns,” Economics of Innovation and New Technology, 11(2): 133–48.CrossRefGoogle Scholar
Hall, B. H. (2007), “Patents and patent policy,” Oxford Review of Economic Policy, 23(4): 568–87.CrossRefGoogle Scholar
Hall, B. H., Jaffe, A., and Trajtenberg, M. (2005), “Market value and patent citations,” RAND Journal of Economics, 36(1): 1638.Google Scholar
Hall, B. H., and Ziedonis, R. H. (2001), “The patent paradox revisited: an empirical study of patenting in the U.S. semiconductor industry, 1979–1995,” RAND Journal of Economics, 32(1): 101–28.Google Scholar
Hausman, J. A., Hall, B. H., and Griliches, Z. (1984), “Econometric models for count data with an application to the patents-R&D relationship,” Working Paper No. 17, National Bureau of Economic Research, Cambridge, MA, available at www.nber.org/papers/t0017 (accessed September 7, 2014).Google Scholar
Hunt, J., and Gauthier-Loiselle, M. (2010), “How much does immigration boost innovation?,” American Economic Journal: Macroeconomics, 2(2): 3156.Google Scholar
Jaffe, A. B., Trajtenberg, M., and Henderson, R. (1993), “Geographic localization of knowledge spillovers as evidenced by patent citations,” Quarterly Journal of Economics, 108(3): 577–98.Google Scholar
Kerr, S. P., and Kerr, W. R. (2015), “Global collaborative patents,” Working Paper No. 21735, National Bureau of Economic Research, Cambridge, MA, available at www.nber.org/papers/w21735 (accessed March 7, 2016).Google Scholar
Kerr, W. R. (2008), “Ethnic scientific communities and international technology diffusion,” Review of Economics and Statistics, 90(3): 518–37.CrossRefGoogle Scholar
Lissoni, F., Sanditov, B., and Tarasconi, G. (2006), “The Keins database on academic inventors: methodology and contents,” KITeS Working Paper No. 181, KITeS, Centre for Knowledge, Internationalization and Technology Studies, Universita’ Bocconi, Milan, Italy, available at http://ideas.repec.org/p/cri/cespri/wp181.html (accessed September 9, 2013).Google Scholar
Maraut, S., Dernis, H., Webb, C., Spiezia, V., and Guellec, D. (2008), “The OECD REGPAT Database,” OECD Science, Technology and Industry Working Papers, Organization for Economic Cooperation and Development, Paris, available at www.oecd-ilibrary.org/content/workingpaper/241437144144 (accessed September 2, 2013).Google Scholar
Miguelez, E. (2016), “Inventor diasporas and the internationalization of technology,” World Bank Economic Review, (in press), pp. 128. doi:10.1093/wber/lhw013Google Scholar
Miguelez, E., and Moreno, R. (2015), “Knowledge flows and the absorptive capacity of regions,” Research Policy, 44(4): 833–48.Google Scholar
Miguelez, E., and Raffo, J. (2013), “The spatial distribution of migrant inventors,” WIPO Economic Research Working Paper.Google Scholar
Peri, G. (2005), “Determinants of knowledge flows and their effect on innovation,” Review of Economics and Statistics, 87(2): 308–22.Google Scholar
Raffo, J., and Lhuillery, S. (2009), “How to play the ‘names game’: patent retrieval comparing different heuristics,” Research Policy, 38(10): 1617–27.Google Scholar
Schmoch, U. (2008), “Concept of a technology classification for country comparisons,” Final Report to the World Intellectual Property Organization (WIPO), Fraunhofer Institute for Systems and Innovation Research, Karlsruhe.Google Scholar
Singh, J. (2005), “Collaborative networks as determinants of knowledge diffusion patterns,” Management Science, 51(5): 756–70.Google Scholar
Trajtenberg, M., Shiff, G., and Melamed, R. (2006), “The ‘names game’: harnessing inventors’ patent data for economic research,” Working Paper No. 12479, National Bureau of Economic Research, Cambridge, MA, available at www.nber.org/papers/w12479 (accessed September 9, 2013).CrossRefGoogle Scholar
WIPO. (2011), World Intellectual Property Indicators, 2011 Edition, WIPO Economics & Statistics Series, World Intellectual Property Organization – Economics and Statistics Division, available at http://ideas.repec.org/b/wip/report/2011941.html (accessed September 2, 2013).Google Scholar
WIPO. (2012a), PCT Yearly Review: The International Patent System, 2012 Edition, WIPO Economics & Statistics Series, World Intellectual Property Organization – Economics and Statistics Division, available at http://ideas.repec.org/b/wip/report/2012901.html (accessed September 9, 2013).Google Scholar
WIPO. (2012b), World Intellectual Property Indicators, 2012 Edition, WIPO Economics & Statistics Series, World Intellectual Property Organization – Economics and Statistics Division, available at http://ideas.repec.org/b/wip/report/2012941.html (accessed September 9, 2013).Google Scholar
van Zeebroeck, N., and van Pottelsberghe de la Potterie, B. (2011), “Filing strategies and patent value,” Economics of Innovation and New Technology, 20(6): 539–61.Google Scholar
van Zeebroeck, N., van Pottelsberghe de la Potterie, B., and Guellec, D. (2009), “Claiming more: the increased voluminosity of patent applications and its determinants,” Research Policy, 38(6): 1006–20.CrossRefGoogle Scholar
Figure 0

Figure 4.1 Coverage of nationality and residence information in PCT patents

Figure 1

Table 4.1 Total Records and Coverage of Nationality and Residence Information (Selected Countries)

Figure 2

Figure 4.2 Share of immigrant inventors, 1985–2010

Figure 3

Figure 4.3 Share of immigrant inventors, 1990–2010

Figure 4

Figure 4.4 Immigration rates of inventors, 1991–2000 and 2001–10

Figure 5

Table 4.2 Immigration Rates of Inventors and College Graduates

Figure 6

Table 4.3 Immigrants, Emigrants, and Emigration Rates, Time Window 2001–10

Figure 7

Figure 4.5 Net migration position, 2001–10

Figure 8

Figure 4.6 Brain-drain rates, 2001–10

Figure 9

Table 4.4 Largest Inventor Migration Corridors, 2001–10

Figure 10

Table 4.5 Largest Bilateral Migration Corridors, 1991–2000 and 2001–10

Figure 11

Figure 4.7 Inventor immigration rates over time by field of technology: three-year moving averages

Figure 12

Figure 4.8 Inventor immigration rates, selected fields and countries, 2006–10

Figure 13

Table 4.6 Top Thirty European NUTS2 Regions by Immigration Stocks and Rates, 2001–10

Figure 14

Table 4.7 Top Thirty US MSAs by Immigration Stocks and Rates, 2001–10

Figure 15

Table 4.8 Inventor Immigration Rates for Top Ten Applicants, Selected Countries, 2006–10

Figure 16

Table 4.9 Share of Immigrants in Highly Cited Patents, All Years

Figure 17

Table 4.10 Share of International Copatents Including Conationals, 2001–10

Save book to Kindle

To save this book to your Kindle, first ensure coreplatform@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about saving to your Kindle.

Note you can select to save to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be saved to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Available formats
×

Save book to Dropbox

To save content items to your account, please confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your account. Find out more about saving content to Dropbox.

Available formats
×

Save book to Google Drive

To save content items to your account, please confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your account. Find out more about saving content to Google Drive.

Available formats
×