27
THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang [email protected] April 9, 2004 ABSTRACT The unique impact of international science and engineering graduate students on US innovation has never been systematically studied. Understanding the contributions of these students has become particularly important for proper review of post-9/11 legislation that severely limited student visa issuance, leading to the first decline in international student enrollment in more than a decade. Additionally, understanding such contributions is key to analyzing the rapid growth of India and China, who currently lead in terms of numbers of students sent. Using an idea generation model, I construct an econometric specification consisting of three groups of four equations, where utility patent applications and grants in universities and other institutions are dependent variables. Results indicate that foreign science and engineering graduate students play a significant, positive role in sharpening US innovation, but that this role is not straightforward, and contrasts in important ways with native student contributions.

THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang [email protected]

  • Upload
    vukien

  • View
    216

  • Download
    1

Embed Size (px)

Citation preview

Page 1: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING

GRADUATE STUDENTS ON US INNOVATION

Xiao Yu Wang [email protected]

April 9, 2004

ABSTRACT

The unique impact of international science and engineering graduate students on US innovation has never been systematically studied. Understanding the contributions of these students has become particularly important for proper review of post-9/11 legislation that severely limited student visa issuance, leading to the first decline in international student enrollment in more than a decade. Additionally, understanding such contributions is key to analyzing the rapid growth of India and China, who currently lead in terms of numbers of students sent. Using an idea generation model, I construct an econometric specification consisting of three groups of four equations, where utility patent applications and grants in universities and other institutions are dependent variables. Results indicate that foreign science and engineering graduate students play a significant, positive role in sharpening US innovation, but that this role is not straightforward, and contrasts in important ways with native student contributions.

Page 2: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

1

1. Introduction

The international presence on US university campuses, both within faculty and the student body,

experienced rapid growth during the 90s. America’s extra-strong economy and expanding educational programs

made her especially attractive to foreign students. The number of international graduate students in particular shot

up as the last of the old origins quotas were repealed and American doors were finally fully opened to Eastern

Europe and Asia. Foreign graduate enrollment in the sciences, engineering, and economics rose across the board as

the economy, and technology-based industries in particular, took off.

After graduation, many of these students chose to stay in the United States, finding employment in

engineering companies, research firms, the technology industry, or academia. Their success encouraged further

inflows of international skill, strengthening doctorate programs across the country in numbers and in research

output.

Today, foreign graduates continue to play major roles in the nation’s academic spheres. For example, 37%

of all MIT graduate students are international, with China, South Korea, and India as the leading countries of origin

(www.mit.edu). Unsurprisingly, in 2005, India, China, and South Korea were the overall leaders in sending students

to the United States. Engineering (23.8%), business (14.9%), and physical/life sciences (13.3%) were the top fields

of graduate study (Open Doors 2005).

Post-9/11 visa restrictions have corresponded with lower applications and enrollment from abroad, an

unsurprising relationship given that 87.4% of all international graduate students travel with F-1 visas, the type

affected by recent legislation (Open Doors 2005). Moreover, other countries are increasingly seen as academically

attractive; Australia in particular has been welcoming larger and larger populations of Asian university students

wishing to study abroad. The loss of international skill has led some to worry that such restrictions will have a

negative effect on US growth, especially since native US students have been less likely to concentrate in science and

engineering fields in the past, where research in such fields is thought to be a major contributing factor to

technological innovation and economic growth.

This paper seeks to analyze the current state of international student enrollment in US science and

engineering (S&E) graduate programs. More specifically, it seeks to evaluate the determinants of US innovation and

growth, to explore whether international students significantly affect these determinants or are determinants

themselves, and to analyze current student visa policy to answer the following questions: does visa policy affect

Page 3: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

2

international graduate student enrollment (as a whole and by field), and does policy that lowers the population of

international S&E graduate students within US universities affect (positively or negatively) US growth? If so, is

such policy optimal or are there superior alternatives?

I find that a key distinction needs to be made between utility patent applications and grants. That is,

international students in some fields actually had a small negative impact on total patent applications, while native

students largely positively impacted both patents and grants. However, international students had a stronger positive

impact on total patent grants, and especially on total university patent grants, while native students had an especially

strong positive impact on non-university patent grants. This indicates that while international students may

ambiguously affect patent applications, the applications they submit are often considered new and useful by the

patenting office, and are particularly significant for academic research. In general, I find that graduate students in

science and engineering, and university research as a whole are key to US innovation.

The framework of the paper will be as such: part (a) of the following section will discuss historical

international student flows in and out of the United States and the characteristics of these students, and part (b) will

discuss the US patenting process and the history of university research. Part three will review past literature. Part

four will detail and justify methodology, develop a model, and elaborate the specifications. Part five will describe

results. Part six will conclude.

2a. Background: International Students

The 2004-2005 academic year saw a total international graduate student enrollment of 264,410, down from

the previous year’s 274,310, the first decline since 1994 (Open Doors 2005). Many have postulated that this decline

is a result of the harsher and more inhibiting visa and student immigration policies passed in light of the Sept. 11

events. In general, total numbers of international graduate students have increased substantially in the past half-

century (see Appendix 1).

The composition of the international student body in terms of country of origin, however, has been far less

consistent. Today’s mix of Eastern European, Asian, and Latino students is very different from the Western

European-dominated populations of just half a decade ago. This evolution is largely the consequence of a variety of

legislation that has been passed since public sentiment shifted and the US government began to encourage a

representative international student presence.

Page 4: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

3

The post-9/11 measures are certainly not the only legislation to have affected international student flows.

Previous efforts have been made to limit or at least to monitor the inflows of international students. The past century

in particular was characterized by important legislation varying the restrictiveness and leniency of visa issuance,

employment policies, and applicant screening.

The first major 20th century relaxation of immigration laws was the Hart-Cellar Immigration Act of 1965,

which removed national origins quotas (www.historicaldocuments.com). The Simpson-Mazzoli Immigration

Reform and Control Act of 1982 then criminalized the hiring of an illegal immigrant, while the Immigration and

Nationality Act Amendments of 1986 repealed various documentation requirements and reduced the requisite pre-

naturalization period. In a similar vein, the Immigration Act of 1990 increased the number of legal immigrants

allowed into the US each year, especially from formerly Communist regions, such as Eastern Europe and Asia, two

major contributors of international S&E students. However, the Illegal Immigration Reform and Immigrant

Responsibility Act, passed in 1996, increased penalties for immigration-related offenses, expanded grounds for

deportation, and restricted benefits and sponsorship by relatives (thomas.loc.gov). Finally, the Patriot Act of 2001

required the tracking and fingerprinting of all students and other visa-holding non-immigrant visitors, and the

Enhanced Border Security and Reform Act of 2002 established a foreign student monitoring system and restricted

issuance of student visas (see Appendix 2 for restrictive and lenient periods between 1980 and 2005). (Open Doors

2005)

While lenient legislation may still be in place at the same time a new restrictive bill is passed, the selected

policies are such that their impacts overwhelm the lingering effects of old legislation. For example, while the

Immigration Act of 1990 further increased permitted immigration flows, the PATRIOT Act, passed 11 years later,

was restrictive enough, and the Immigration Act was old enough, such that the restrictive nature of the PATRIOT

Act was enough to transform the nature of its period into one more prohibitive for incoming international students.

Figures from the National Science Foundation (2003) indicate that many of these international graduates

choose to stay in the US—in 1973, 13,600 foreign doctorates were employed in academia alone; by 2003, that

number had risen to 60,400. According to Aslanbeigui and Montecinos (1998), 15% of international students from

developing countries intend permanent residence, with about half indicating a desire to stay at least temporarily.

Thus, it is not unreasonable to expect that US training of these international students will have long-term returns, as

Page 5: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

4

students contribute to industrial productivity and academic research, and add to the overall innovative capacity of

the nation.

b. The US Patenting Process and University Research

The website of the United States Patent and Trademark Office declares, “The patent laws of the United

States make no discrimination with respect to the citizenship of the inventor. Any inventor, regardless of his/her

citizenship, may apply for a patent on the same basis as a U.S. citizen.” (uspto.gov) In other words, the US patenting

process for citizens and non-citizens is virtually the same, so differences between foreign and native student

patenting behavior should not strictly result from differences in patenting procedure.

Unsurprisingly, there are numerous costs associated with applying for, receiving, and maintaining a patent,

as well as the perpetual possibility of legal battles. These costs include the hiring of a patent attorney or agent, filing

fees, copy fees, postal fees, post-allowance fees, fees associated with reissue, examination and re-examination fees,

publication fees, post-issuance fees, patent search fees, maintenance fees and a host of others (uspto.gov). All these

costs imply that those who complete the patenting process strongly believe in the utility of their inventions, and that

universities will tend to apply for relatively fewer patents compared to the private industrial sector, as a result of

smaller financial endowments (in general). However, the expected importance of these patents is a separate matter,

and can be gauged by examining percentage of successful applications.

Given that the length of stay of foreign graduate students is indeterminate, interrupted by visits home, and

dependent on ease of visa renewal, financial limitations, and numerous other considerations unique to studying

abroad, it is possible that international students may be given second preference by professors involved in patent-

producing research opportunities. On the other hand, such students are almost always legally constrained in the jobs

they are allowed to seek; specifically, these students are normally only permitted to work within the university,

meaning a large percentage of international graduate students end up working in university research institutions.

Furthermore, most international students arrive from math- and science-intensive countries, so they are likely to be

extremely productive in terms of pure research, despite weaker linguistic abilities.

Professors may wish to avoid working with students with limited language skills, but the growing

population of foreign academics may nullify such propensities. A scarcity of fellowships, low salaries, and a lack of

a network base could further limit total population of international graduate students, but the relative prevalence of

Page 6: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

5

foreign graduate students in the nation’s top universities (Columbia, the University of Wisconsin, the University of

Pennsylvania) indicates that, with the best research facilities and faculty at their disposal, they are especially

equipped to contribute to US innovation (Open Doors 2005). This prevalence is unsurprising—first, because top

S&E and top research universities tend to have bigger and better international student programs, since foreign

students are reputed to be exceptionally able and productive in scientific fields, and second, because such

institutions are relatively more famous abroad and receive more applications from international students.

All this intuition suggests that the effect of international S&E graduate students on patent applications will

be ambiguous, as on one hand, foreign students are disadvantaged practically in terms of language, network, and

visa problems, and on the other, S&E foreign graduates have very strong backgrounds in their fields, are

concentrated in top research universities, and are potentially more involved with academic research. Further

intuition suggests that the former characteristics are more important for quantity of patent applications (language,

network, continuous work time), while the latter characteristics may be more important in the long run, i.e. for

quantity of patent awards (strong math and science backgrounds, top research facilities, longer and more productive

work hours). In other words, we should expect to see more patent applications from native graduate students, and

potentially more patent grants for foreign graduate students. The magnitude of the impact of graduate students on

US innovation, however, depends largely on the significance of university research.

While the importance of university research to general innovation is not known, it is certainly true that

much more of it occurs today than pre-1980, when the Bayh-Dole Act was passed, allowing universities to

commercialize their previously unprofitable research findings. This study takes place entirely in the post-Bayh-Dole

period in an attempt to concentrate solely on the impact of international graduate students on patenting, regardless of

the effects of this Act. In other words, given the passage of Bayh-Dole, what are the effects of S&E graduate

students? This may lead to a slight bias in the estimated R&D effect, despite my framing of R&D university

expenditure in terms of federally-financed R&D expenditures in universities as opposed to spending by universities

themselves, but this bias is negligible and shouldn’t significantly affect the study.

Interestingly, in the last century, all kinds of R&D expenditures have skyrocketed (see Appendix 3)—

notably during the two World Wars, the Cold War, and the 90s with the birth of the technology and software

industries (nsf.gov). Lately, there has been talk of falling behind again in terms of scientific innovation, prompting

Senators John Ensign and Joseph Lieberman to introduce, as recently as December 2005, a bipartisan “National

Page 7: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

6

Innovation Act” that would nearly double federal research funding (www.technet.org). The introduction of this bill

hints at another wave of national concern about math and science programs, perhaps fueled by the recent growth of

China and India (nations that traditionally emphasize science and engineering), making analysis of the innovative

contribution of international students particularly salient. In other words, even if international students are found to

have a significant positive impact on US research output, should we encourage their presence? Or should we

“protect” native students from this kind of “foreign competition”? Does our policy approach depend on whether or

not these students choose to stay in the United States?

Previous authors have attempted to comment on one or more of these questions in various manners using

various methodologies. While these questions have been by no means resolved, it is important to look to past

literature to understand some of the disagreement and provide a basis for the systematic evaluation I use.

3. Literature review

There is not a very large body of existing literature dealing specifically with the innovative and economic

impacts of international graduate students, but there are several important studies that motivate and lay the

groundwork for this one.

Borjas (2004) argues that foreign students at a specific university crowd out enrollment of native white

males in the graduate programs at that university, even considering the expansion of those programs, and that this

crowding out is most apparent in private, top-tier institutions. (He defines this crowding out “…in the sense that

native enrollment would have risen faster if the university had not increased its supply of foreign students.”) (Borjas

2004) He further claims that while foreign students generate tuition revenue, such revenue is more than offset by the

cost of taxpayer support of public universities and of the limitations on educational opportunities for native white

males (and, marginally, native white females) the international inflow imposes (Borjas 2004).

This paper, then, implies that foreign students are preferred over their equally qualified native counterparts,

since Borjas claims that foreign students are not just fueling expansion of graduate programs, but are actually

replacing native students. However, a study done by Attiyeh and Attiyeh (1996) shows that in most cases, US

students are given preference. Additionally, the inferior performance of native students on international math and

science tests relative to foreign students indicates that the pool of native students wishing to pursue a master’s or

doctoral degree in any science or engineering field may be smaller than the foreign pool. This is particularly possible

Page 8: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

7

since Borjas (2004) uses figures for all graduate students at some institution, and does not distinguish these students

by field. Strangely, Borjas finds an absolute lack of substitution among native minorities, and a very small potential

replacement of native white women. Given that Borjas dismisses the possibility of better qualification, this implies

that all minorities, women, and international students are given significant unfair preference over native white

males, across the board and particularly in Ivy League institutions. At best, such a situation seems unlikely.

Chellaraj, Maskus, and Mattoo (2005), on the other hand, find that international graduate students and

skilled immigrants have a significant positive impact on US innovation, where innovation is measured in patents

applied for and patents granted (to universities and other institutions). Furthermore, they find that native students

actually have a significant negative effect on innovative output. Additionally, they find that rigorous enforcement of

visa restrictions has a significant negative effect on innovation, and they apply these results to the post-9/11

situation, concluding that Section 214(b) of the Patriot Act threatens US economic growth.

However, these economists were also unable to distinguish science and engineering graduate students from

students of other fields. It’s possible, therefore, that while the post-9/11 restrictions may have negatively impacted

the body of foreign graduate students as a whole, the subpopulation of international engineering graduates may have

remained unaffected. It could be that it is this (large) subpopulation that drives the positive effect that the entire

foreign pool was found to have on US patent applications and awards.

Moreover, the authors never define “international student” so it is unclear which students fall into this

category. For instance, this population could be defined simply as graduate students who are citizens of other

countries, or as students who are in the United States on temporary visas, such as the F-1 (which is how I have

defined it)., where the latter is a subset of the former group.

Further, Chellaraj, Maskus, and Mattoo (2005) examine total amounts of US patent applications and

awards. Utility patents, however, comprise the majority of total patent applications and awards by far, and chiefly

contribute to technological innovation (compare design and plant patents, where the nature of the former is cosmetic

and the nature of the latter is almost completely non-technological). Because technological innovation is widely

acknowledged to be most critical for economic growth (Basu, Fernald, and Shapiro 2001), I examine figures for

utility patent applications and awards alone.

Additionally, the study accounts for visa policy by measuring periods of strict enforcement of visa

regulations with a dummy variable reflecting high and low unemployment. This method is workable because

Page 9: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

8

Chellaraj et. al. analyze foreign students and skilled immigrants; however, my paper examines only international

graduate students, so this method is not as appropriate (high unemployment is more likely to be correlated with

stricter immigration laws rather than tighter student visa issuance). Therefore, I use a dummy variable based on

restrictiveness or leniency of legislation passed between 1980 and 2005 that affected international students (see

Section 2 for details). The model will not be able to account for post 1999 legislation (namely, the PATRIOT Act

and the Enhanced Border and Security Reform Act), due to the 5- and 7-year lags used to account for time periods

between patent research, application, and grant, but I will be able to comment on expected levels of patent

applications and grants given changes in the size of the foreign student body.

Finally, Chellaraj et. al. incorporate a measure of “non-university real R&D expenditures” into their

regression analysis of non-university patent grants. I modify this to be “industrial real R&D expenditures” for

several reasons: first, because industrial R&D expenditures have comprised about 65-70% of total real R&D

expenditures in the past decade, and second, because the majority of utility patents granted to non-university

institutions are granted to companies (nsf.gov, uspto.gov). This setup is more conducive to examining the

increasingly important link between universities and industries, since it isolates the effect of industries on patenting

and the effects of students on industries, so that we are able to gain some insight into the impact of students on non-

university patents and the impact of university and non-university research on general innovation.

For similar reasons, I measure the R&D expenditures specific to patents granted to universities as federally-

financed R&D at universities and not university R&D expenditure itself. This is largely because industries and

universities increasingly collaborate in research, and university funding of R&D would include federal and private

support, where some of that private support would certainly be double-counted in the “industrial real R&D

expenditures” variable, since industries currently fund a large portion of university research. Also, federally-

financed R&D at universities is more likely to be a consistent presence in and consistently distributed across

universities, unlike university R&D expenditures, which would be influenced by spontaneous alumni donations that

might be effective for a year or for one project in one field but would not necessarily enhance university research as

a whole. Finally, this characterization lends itself more readily to policy prescription as such funding is directly

within government control.

Clearly, there is a lot of disagreement within the limited realm of relevant previous literature. While the

importance of innovation (most notably technological advances) for economic growth is widely acknowledged, the

Page 10: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

9

overall impact international science and engineering students have on innovation, the contribution of university

research to general innovation, and even the true productivity of R&D expenditures are disputed. Consequently,

relevant policy recommendations are largely contradictory.

I concentrate solely on the impact international S&E graduate students have on US innovation and growth.

A brief examination of the correlations (see Appendix 4) indicates that a large presence of foreign graduate science

and engineering students is strongly correlated with high levels of patent applications and grants. The total numbers

of Ph.D. scientists and engineers in the country, R&D expenditure, and existing knowledge (cumulative total

patents—here, I consider all types of patents, not just cumulative utility patent numbers, to analyze the full “standing

on shoulders” effect) are also strongly positively correlated with high levels of research output. Thus it seems likely

that international graduate S&E students positively impact US innovation.

4. Methodology

Following Chellaraj et. al. I consider the “national ideas production function” used in a variety of

innovation studies (Porter and Stern, 2000):

Ạt=δΗA,t

λAtΦ (1)

where the dependent variable is rate of new ideas production, δ describes changes in US idea productivity over time,

ΗA,t denotes the allocation of resources to the R&D sector, λ is the productivity of those resources, At is the stock of

ideas already in existence, and Φ describes the ability of that stock to support new invention.

The general measure of new ideas production in this study is total utility patent applications and total utility

patent grants, where total utility grants are further broken down to distinguish between those granted to universities

and those granted to other institutions. While patents are not a perfect measure of national innovation, due to the

tremendous variation in the contributions they make (or do not make) to technological advancement, there are

several compelling reasons to use them. First, I consider utility patents only, the most technological of patent types,

second, the patenting process is not easygoing or cheap, so the inventor must truly feel that she has something new

and useful to contribute, third, the patent reviewers grant only the most innovative portion of the applications, and

fourth, extensive patent search databases and rigorous reviewing procedures eliminate most repetition and serve as a

strong foundation for further, new patenting.

Like Chellaraj et. al. I redefine ΗA,tλ , but in more detail, such that:

Page 11: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

10

ΗA,tλ = [(Hλr) RD, t][(Hλe)ENG, t][(Hλpl)PL, t][(Hλk)SK, t] (2)

where HRD represents total R&D expenditure and λr represents its productivity, [(Hλe)ENG, t]=[(Hλef)F, t][(Hλen)N,t]

and represents flows of foreign and native engineering graduate students, with λef, λen representing respective

productivities, [(Hλpl)PL, t]=[(Hλpf)F, t][(Hλpn)N,t] represents flows of foreign and native physical and life science

graduate students and their productivities, and HSK represents total Ph.D. scientists and engineers, with λk

representing that population’s productivity. (Physical and life sciences are defined to include all mathematical

sciences, health sciences, biological sciences, physical sciences, earth/atmospheric and oceanic sciences, and

agricultural sciences.)

To account for the AtΦ term, that is, to measure the “standing on shoulders” effect of existing knowledge

stock, I take cumulative numbers of total patent stock, total university patent stock, and total non-university patent

stock, where the 2000 level is the sum of the 1996-2000 inclusive annual levels. The δ parameter in equation (1)

measures changes in US idea productivity over time and is taken to be both a function of time and, more

importantly, of policy changes, such as legislation affecting international student inflow.

Several other factors must be considered before constructing an econometric specification. First, following

Chellaraj et. al. I have taken into consideration time lags between commencing research and patent application, and

between patent application and grant. After reevaluating lags of previous studies, I have determined that 5 years is

still the most appropriate average lag for the time between science and engineering research and patent application,

and 2 years is still most appropriate for the time between patent application and grant (Johnson, et. al. 2003). In

other words, resources employed in 1980, such as numbers of foreign graduate students and R&D expenditures,

affect the 1985 number of patent applications, and the 1987 number of patent grants. It is important to note that lag

time, especially between application and grant, has not remained constant over the years, and has in fact increased

post-1995 due to increases in patent application and to decisions reached in the GATT Uruguay Round (Johnson et.

al. 2003), so these figures should be treated as rough estimates. Additionally, lag time varies severely across

industry—pharmaceutical breakthroughs generally require much longer periods of research (NIST Status Report

2003)—and depends on length and number of legal battles as well, since patent legitimacy is solely upheld by the

courts.

Page 12: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

11

Preserving stationarity of the variables is key, given the time series analysis. Therefore, I have scaled all

variables with respect to annual US labor force. However, this alone may not be sufficient—it might be necessary to

examine the percent differences annually for each variable to truly preserve stationarity, since the percentage of US

labor force that is international science graduate students may have experienced a significant upward trend.

Nevertheless, the growth in US labor force should be sufficient to ensure minimal trend effects, so the following

econometric specification remains legitimate; the results may be negligibly affected but will provide an accurate

general idea. Finally, estimation technique is ordinary least squares (OLS).

A.

lnTPA(t+5)= K1A + (λ11

A)lnFORENG + (λ12A)lnNENG + (Φ1

A)lnTOTPSTOCK + (λ13A)lnSK + (λ14

A)lnRD + (δ1A)ED + η1t

A

lnTPG(t+7) = K2A + (λ21

A)lnFORENG + (λ22A)lnNENG + (Φ2

A)lnTOTPSTOCK + (λ23A)lnSK + (λ24

A)lnRD + (δ2A)ED + η2t

A

lnUIPG(t+7) = K3A + (λ31

A)lnFORENG + (λ32A)lnNENG + (Φ3

A)lnUPSTOCK + (λ33A)lnSK + (λ34

A)lnFRD + (δ3A)ED + η3t

A

lnOIPG(t+7) = K4A + (λ41

A)lnFORENG + (λ42A)lnNENG+ (Φ4

A)lnOPSTOCK + (λ43A)lnSK + (λ44

A)lnIRD + (δ4A)ED + η4t

A

B.

lnTPA(t+5)= K1B + (λ11

B)lnFORPL + (λ12B)lnNPL + (Φ1

B)lnTOTPSTOCK + (λ13B)lnSK + (λ14

B)lnRD + (δ1B)ED + η1t

B

lnTPG(t+7) = K2B + (λ21

B)lnFORPL + (λ22B)lnNPL + (Φ2

B)lnTOTPSTOCK + (λ23B)lnSK + (λ24

B)lnRD + (δ2B)ED + η2t

B

lnUIPG(t+7) = K3B + (λ31

B)lnFORPL + (λ32B)lnNPL + (Φ3

B)lnUPSTOCK + (λ33B)lnSK + (λ34

B)lnFRD + (δ3B)ED + η3t

B

lnOIPG(t+7) = K4B + (λ41

B)lnFORPL + (λ42B)lnNPL+ (Φ4

B)lnOPSTOCK + (λ43B)lnSK + (λ44

B)lnIRD + (δ4B)ED + η4t

B

C.

lnTPA(t+5) = K1C + ( λ11

C)lnFORSE + (λ12C)lnNSE + (Φ1

C)lnTOTPSTOCK + η1tC

lnTPG(t+7) = K2C + (λ21

C)lnFORSE + (λ22C)lnNSE + (Φ2

C)lnTOTPSTOCK + η2tC

lnUIPG(t+7) = K3C + (λ31

C)lnFORSE + (λ32C)lnNSE + (Φ3

C)lnUPSTOCK + (Φ3C’)lnOPSTOCK + η3t

C

lnOIPG(t+7) = K4C + (λ41

C)lnFORSE + (λ42C)lnNSE + (Φ4

C)lnUPSTOCK + (Φ4C’)lnOPSTOCK + η4t

C

Individual fields were regressed separately against each dependent variable. A logarithmic approach seemed most

appropriate in determining the effect of growth in research inputs on innovative growth. TPA(t+5) represents total

utility patent applications with a 5-year lag, TPG(t+7) represents total utility patent grants with a 7-year lag, UIPG(t+7)

represents total utility patents granted to universities with a 7-year lag, and OIPG(t+7) represents total utility patents

granted to other institutions with a 7-year lag. K is the constant term. The prefixes “FOR” and “N” denote “foreign”

Page 13: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

12

and “native,” respectively, SK refers to total Ph.D. scientists and engineers, TOTPSTOCK, UPSTOCK, OPSTOCK

refer to cumulative utility patent stock, cumulative university patent stock, and cumulative other institutional patent

stock, RD, FRD, IRD refer to total real R&D expenditure, total real federally-financed R&D expenditure at

universities, and total real industrial R&D expenditure, and ED is the enforcement dummy variable distinguishing

between restrictive and lenient periods of visa enforcement and issuance.

Foreign graduate students are defined as students in US university graduate programs holding temporary

visas, such as F-1 student visas. The enforcement dummy ED measures periods of restrictive visa issuance and

enforcement based on legislation passed since 1980. Thus, a negative coefficient on this variable would indicate that

today’s restrictive legislation, specifically Section 214(b) of the PATRIOT Act and the Enhanced Border Security

and Visa Entry Reform Act of 2002, has a significant negative impact on total utility patent applications and grants.

Because the restrictive legislation has a negative impact on the foreign student presence due to imposed limitations

on visa issuance, the implication is that foreign students have a significant positive impact on US innovation and

economic growth, implying that a lesser foreign student population is detrimental to US expansion.

The econometric specifications are divided by field to isolate the effect each subpopulation has on the

dependent variables. The subpopulations could not be combined into one regression due to insufficient number of

observations. The final grouping, (C), determines the general effect of foreign science graduate students, native S&E

graduate students, and cumulative patent stock on each of the dependent variables (new ideas production in terms of

patents).

This model was implemented using annual data between 1980 and 2003. The dataset was compiled using a

variety of sources. The Institute for International Education’s annual publication, Open Doors, and the National

Science Foundation’s Science and Engineering Indicators, and Historical Statistics provided R&D figures and data

on graduate student enrollment by citizenship and field. US Labor Force figures were taken from the Bureau of

Labor Statistics website. Data on Ph.D.-holding scientists and engineers were taken from Chellaraj et. al. (2005),

and all patent data were taken from the United States Trademark and Patenting Office website’s Patent Statistics.

5. Results and Discussion

Regression results are presented in Appendices 5-7. Each Appendix corresponds with specifications (A)-

(C) introduced in the previous section, where (A) and (B) focus on field and (C) focuses on citizenship.

Page 14: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

13

Examining the effects of citizenship on innovation, it may be seen that foreign students ambiguously

impact total utility patent applications but positively impact total patent grants, and have the most positive impact on

university patent grants. Native students, on the other hand, positively affect total patent applications and grants in

most cases, but the effect is smaller. Further, native students most positively affect other institution patent grants.

Examining foreign and native S&E graduate students in (C) shows that as a whole, foreign science students’

elasticities were 0.24, 0.65, 1.25, and 0.15 for total applications, grants, university grants, and other institution grants

(Appendix 7, row (ii)), whereas native science students’ elasticities were -1.51, 1.21, 0.85, and 1.53 for applications,

grants, university grants, and other institution grants, respectively (Appendix 7, row (iii)). All results were

significant at the 10% level or better. Note that Chellaraj et. al. (2005) similarly found foreign elasticities to be about

0.3 for total applications and grants, and 0.6 and 0.4 for university and other institution grants. However, that study

contradictorily found native students’ elasticities to be around -0.39, -0.42, -0.49, and -0.53 for total applications,

total grants, university grants, and other institution grants (Chellaraj et. al. 2005). (Recall that Chellaraj et. al. used

populations of all foreign and native graduate students without being able to differentiate by field, and considered

levels of total patents instead of the subpopulation of utility patents.)

To put these numbers into perspective, observe that a 10% increase in foreign S&E (social science

inclusive) graduate students, from its sample mean level of 83,615, would lead to 2,000 more utility patent

applications (0.9% of the sample mean level of utility patent applications) and 5,452 more utility patent grants (5%

of the sample mean level of utility patent grants). This effect is quite substantial.

In general, I find that foreign and native students have significant positive impacts on innovation, but that

they affect research output in different ways. While results indicate varying effects on total utility patent

applications, it is more important and more interesting to note the positive effect on patent grants, and, specifically,

the innovative differences between foreign and native S&E students in universities versus in other institutions. In

accordance with our intuition, the relatively stronger presence of foreign students in the nation’s (top) universities

(as a result of professional constraints and foreign student reputation in S&E fields), and the relatively stronger

background international students have in math and science, strengthen their effectiveness in academic innovation.

In contrast, native students, who are able to find employment outside of the university, have a very strong positive

effect on innovation in other institutions, which are, more often than not, industries. This relationship holds by field

as well—foreign engineering and physical life sciences students had elasticities of 1.244 and 1.822 on university

Page 15: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

14

patent grants (significantly different from 0 at the 1% level) (Appendix 5 and 6, row (ii), column 3), and native

engineering and physical life sciences students had elasticities of 0.87and 1.503 on other institution patent grants

(significantly different from 0 at the 1% and 5% levels) (Appendix 5 and 6, row (iii), column 4). But how important

are university and other institution innovations for general innovation?

Results are encouraging—examining Appendix 7, row (vi), we see that a 1% increase in cumulative patent

stock leads to a 0.813% increase in total utility patent grants, where cumulative patent stock is largely composed of

cumulative university and other institution patent stock. Further, when populations of engineering students are

accounted for, cumulative patent stock grows in importance, where a 1% increase in total patent stock now leads to a

1.189% increase in total patent grants. With physical and life sciences students, this elasticity becomes 0.583. In

particular, the spillover effect of university innovation on other institution innovation is 0.47 (Appendix 7, row (iv),

column (5)), while the spillover effect of other institution innovation on university innovation is a startling -1.1

(Appendix 7, row (v), column (4)).

In other words, cumulative university patent stock was important for university and other institution patent

grants, while cumulative other institution patent stock actually had a negative impact on both. This emphasizes the

importance of university research in particular for national innovation as a whole, magnifying the strong positive

impact international S&E graduate students have overall. In general, previous knowledge is exceptionally important

in motivating future research—this dramatic “standing on shoulders” effect is not unexpected, as the accessibility of

all previous patents makes it easy for an aspiring inventor to see “what’s been done,” to be inspired by gaps in

previous inventions, and to decide what could be done, knowing what has been tried, what has failed, and what has

succeeded.

This conclusion is supported by the Chellaraj et. al. (2005) paper, which also found cumulative patent stock

to be a key determinant of innovation, with an average elasticity of 0.14 at 5% significance. The paper also found

university patent stock to be important for non-university innovation, where the university patent stock spillover

elasticity fell between 0.19 and 0.26 (Chellaraj et. al. 2005).

Research and development expenditures also had consistent, significant, and positive effects on patent

applications. A 1% increase in total R&D expenditure increases total patent applications by 1.82% and total patent

grants by 0.18% in Appendix 5 (row (vii)). The physical and life sciences regression yields total R&D elasticities of

1.399 and -0.192 for applications and grants, respectively. This tells us that, while R&D significantly increases the

Page 16: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

15

number of research opportunities in the nation, so that we see many more patent applications as a result,

expenditures do not necessarily improve the quality of the research—few of these additional applications are

actually granted. Given that this is usually the tool government innovation policy uses (recall the proposed National

Innovation Act, the main feature of which would be to double federal R&D expenditure), we may want to reconsider

our policies in light of these results.

The elasticities of the population of Ph.D. scientists and engineers were generally positive, although

statistically insignificant. Surprisingly, they seemed to have the smallest effect on other institution patent grants,

while having a strong positive effect on total utility patent applications and university patent grants. Appendices 5

and 6, row (iv) show us that a 1% increase in Ph.D. scientists and engineers, holding all else constant, leads to an

average of a 0.5% increase in total utility patent applications, a negligible effect on total patent grants, a 0.27%

increase in university patent grants, and a slightly negative effect on other institution patent grants. However, given

the increasing collaboration between industries and universities, and given that this is the population of scientists

and engineers with Ph.D.s, it does not seem so implausible that they would have stronger impacts on university

patents rather than other institutions.

Restrictive enforcement seems to have very slight positive effects on the dependent variables, but the

coefficients were never statistically significant, and the observed effects were very small, so it is impossible to make

any conclusive statements based on the regressions. However, Chellaraj et. al. (2005), using an indicator variable for

employment, did find restrictive visa issuance policies to have a significant negative effect (around -0.22 elasticity)

on US innovation.

Knowing that international students face legal employment constraints, interrupted research time, risk of

visa loss, financial, linguistic, and cultural barriers, and the disadvantage of smaller networks, but benefit from

stronger elementary backgrounds in math and science, from being among the strongest students in their home

countries, and from being relatively concentrated in the top research universities, implies ambiguous effects on

patent applications but probable positive effects on patent grants (because the latter characteristics are more

important for research quality). Results confirm this intuition. Similarly, results confirm the intuition that native

students should have a stronger positive impact on other institution patent grants, because they tend to be employed

in these other institutions, since they are not legally obligated to stay within university research. In sum,

contradicting Borjas (2004), foreign S&E graduate students were found to have significant positive effects on US

Page 17: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

16

innovation, especially within academia; contradicting Chellaraj et. al. (2005), native S&E graduate students were

also found to have significant positive effects on US innovation, especially within industry.

Existing knowledge, in turn, is shown to be a significant contributor to general innovation, with university

patent stock proving to be a particularly significant factor. Research and development expenditures were seen to

definitively boost patent applications, but not patent grants, implying that other provisions need to be attached to

such funding for the dollars spent to truly augment national innovative capacity.

6. Conclusion

While the results show that, in answer to questions posed in the beginning of this paper, international

students significantly and positively affect US innovation, and strict visa enforcement and issuance have

indeterminate effects on university patenting (which is in turn important for general new idea production), the policy

questions remain difficult to answer. For instance, optimal policy would depend on national goals. If we were only

concerned with generating as much innovative output as possible, then current policies (Section 214(b), the

Enhanced Border Reform and Security Act) are detrimental to our goals because, by discouraging international

student inflows, they lower our potential for technological advancement and growth. However, if we are more

concerned with sheer numbers of native students in science and engineering fields, then the policies are not

necessarily bad.

The official goal of the recent policies is to maintain homeland security. However, to determine whether

the PATRIOT Act and other legislation are optimal for the country, we need to decide whether the value of potential

increased security is greater than or at least balanced by the loss in potential innovation, especially at our

universities. The policy leads to an overall decrease in numbers of international students, but also shifts the

composition of this international student body in terms of countries of origin. As per the study, international

engineering students negatively affected patent applications and less positively impacted patent grants, in

comparison with physical and life sciences students. Moreover, most S&E students are Asian or Middle Eastern. If

this is in fact true, should we shape future policy to target only those students who seem more likely to make

innovative contributions? Or is this, again, not our goal when we welcome international students into our schools?

Whatever our objectives may be, it is important that we rectify the inconsistencies in our international

student policies. US officials issue visas based on the probability that the student returns home—in other words, a

Page 18: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

17

student who indicates a desire to stay in the United States is much more likely to be rejected than a student who

swears to go home. Similarly, foreign countries are much more reluctant to issue visas to students who express

intentions of staying in the United States. Such policies mean that the students who do end up studying in the United

States are those most likely to return home. Yet conventional wisdom dictates that we should only educate those

students who stay in the country and make long term contributions to the US economy. Consequently, international

students who want to stay are punished and international students who wish to return home are disliked. Optimal

international student inflows cannot be achieved until these contradictions are resolved.

It is important to recognize the limitations of this study. Statistical significance of the estimates of the

coefficients was a consistent problem. Stationarity of data, too, is something that needs to be re-examined, where

perhaps differences from year to year are taken for each variable, instead of mere scaling with respect to labor force.

Additional observations would have improved the study, although no reliable data on international and native

graduate students by field exists pre-1980. Patent lags were exceptionally difficult to measure, given their variability

and inconstancy over time. This has implications for the results—for example, it could be that foreign students, for

whatever reason, work slower than native students (perhaps due to discontinuous residence in the US). Then the

negative effects we see on patent applications are perhaps not a “bad” thing—the negative elasticities would simply

indicate slower rate of research completion, which might partially explain international students’ higher percentages

of patent grants. However, this is just conjecture.

This study did not take into account software copyright as a further measure of new ideas production,

because no such data is available (also, software is a much more recent development). Furthermore, no good

measure of fellowships/grants/publications was able to be found. The study also does not consider foreign

academics or visiting scholars.

The enforcement variable requires much further experimentation—the measurement I employ is slightly

arbitrary in some sense, because it is difficult to determine when any piece of legislation truly takes effect—does the

period of restrictive enforcement begin when the bill is introduced, when it is passed, or a couple years after that,

when it has had time to really gain potency? If the last is true, how do we determine that particular lag? Other

methods of accounting for such enforcement are possible; determining the optimal measurement will be difficult.

The increasing number of international students in economics bears observation and is certainly interesting

when applied to students who return to their home countries. It could be asked how these international students, in

Page 19: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

18

perhaps applying their knowledge of the US economy, affect the economic policy of their countries. Alternatively,

one could examine the effects these students have on the US economy if they stay and combine their knowledge of

their home country with what they learn in this country. The idea is that international students can definitively shape

the economies of the US and their home countries without patenting and making technological contributions.

However, this was not addressed due to lack of data.

Finally, quality of patents was not captured by this study. For example, foreign students may significantly

affect total utility patents granted, but it is difficult to say anything about the usefulness of these awarded patents. A

possible method would be to count the citations these patents have received. Either way, it is entirely possible that,

while native and international students positively impact patent levels, these new patents are not very useful in terms

of economic growth or motivating other research.

Further research should improve the data in general by including more observations, by ensuring

stationarity of variables, by using time series analysis to make predictions about current policies that were unable to

be included in the study due to the lagged periods, and by accounting somehow for patent quality. It should also

improve the enforcement variable, and could improve measurement of new ideas production, if possible. However,

this area is underdeveloped and could benefit from more study in general. Detailed policy recommendations cannot

be made, despite solid indication that international graduate students do positively impact US innovation and

economic growth, until more is understood about the relationship between enrollment of foreign and native graduate

students, between student enrollment and patenting activity, and between patenting activity and innovation. We

may know that increased inflows of international students are beneficial for US innovation, but we still do not know

the best way to orchestrate such increased inflows. Nevertheless, it is clear that policies restricting international

intellect are detrimental to efforts to expand our knowledge stock, and will not lead to economic growth.

Page 20: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

19

References

1. Aslanbeigui, N. and V. Montecinos. 1998. “Foreign Students in U.S. Doctoral Programs.” Journal of Economic Perspectives 12: 171-182. 2. Attiyeh, G. and R. Attiyeh, 1996. “Testing for Bias in Graduate School Admissions.” The Journal of Human Resources 32: 524-548. 3. Basu, S., J.G. Fernald, and M.D. Shapiro. 2001. “Productivity Growth in the 1990s: Technology, Utilization, or Adjustment?” Working Paper, No. w8359, NBER. 4. Borjas, G.J. 2004. “Do Foreign Students Crowd out Native Students from Graduate Programs?” Working Paper No. 10349, NBER. 5. Chellaraj, G., K. Maskus, and A. Mattoo. 2005. “The Contribution of Skilled Immigration and International Graduate Students to US Innovation.” Working Paper No. 3588, World Bank. 6. Henderson, R. and A. Jaffe. 1998. “Universities as a Source of Commercial Technology: Detailed Analysis of University Patenting, 1965-1988.” Review of Economics and Statistics 80: 119-127.

7. Hock, Jim. 2005. “TechNet Hails Introduction of National Innovation Act of 2005.” http://www.technet.org/press/Press_Releases/?newsReleaseId=3996

8. “The Immigration and Naturalization Act of 1965.” Historical Documents. http://www.historicaldocuments.com/ImmigrationActof1965.htm

9. Institute for International Education, 1948-2004. Open Doors. 10. Johnson, D.K.N., T. Juhl, and B. Popp. 2003. “Time in Purgatory: Determinants of the Grant Lag for US Patent Applications.” Working Paper No. W9518, NBER. 11. Laursen, K. and A. Salter. 2004. “Searching High and Low: What Types of Firms use Universities as a Source of Innovation?” Research Policy 33: 1201-1215. 12. The Library of Congress: THOMAS. Bills, Resolutions. http://thomas.loc.gov/home/bills_res.html 13. MIT. MIT Facts 2006: “International Students and Scholars.” http://web.mit.edu/facts/international.shtml 14. National Science Foundation. 1993-2005. Science and Engineering Statistics. 15. Porter, M.E. and S. Stern. 2000. “Measuring the 'Ideas' Production Function: Evidence from International Patent Output.” Working Paper No. 7891, NBER. 16. US Census Bureau. 1958-2003. Statistical Abstract of the United States. 17. US Patent and Trademark Office. 1963-2004. Patent Statistics.

Page 21: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

20

Appendix 1: Science and Engineering (S&E) Graduate Student Enrollment in US Universities,

1980-2003

Science and Engineering (S&E) Graduate Students in US Unviersities, 1980-2003

0

50,000

100,000

150,000

200,000

250,000

300,000

350,000

400,000

450,000

500,000

1980 1983 1986 1989 1992 1995 1998 2001

Year

Num

ber

of S

tude

nts

Total S&E Graduate Students

Native S&E Graduate Students

S&E Graduate Students with TemporaryVisas

Data from National Science Foundation, Science and Engineering Statistics, 2003

Page 22: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

21

Appendix 2: Major Legislation Affecting International Student Inflows, 1965-2005

Time Period Legislation Effect

1. 1965-1982 Hart-Cellar Immigration Act of 1965 Lenient

2. 1982-1986 Simpson-Mazzoli Immigration Reform and Control Act of 1982 Restrictive

3. 1986-1995 Immigration and Nationality Act Amendments of 1986,

Immigration Act of 1990

Lenient

4. 1996-2000 Illegal Immigration Reform and Immigrant Responsibility Act Restrictive

5. 2001 PATRIOT Act of 2001 Restrictive

6. 2002 Enhanced Border Security and Reform Act of 2002 Restrictive

Page 23: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

22

Appendix 3: Real-Valued R&D Expenditure in the United States, 1953-2004

Data from National Science Foundation, Science and Engineering Statistics, 2005

Real-valued R&D Expenditure in the United States, 1953-2004

0

50000

100000

150000

200000

250000

300000

350000

1/1/00

Year

R&

D e

xpen

ditu

re, i

n m

illio

ns o

f rea

l dol

lars

Federally-financed R&DExpenditures at universit iesIndustrial R&D Expenditures

Total R&D Expenditures

1953

Page 24: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

23

Appendix 4: Basic Correlations

Notes: TPA is total utility patent applications, FORENG is population of international engineering graduate students, NENG is population of native engineering graduate students, FORPL is population of international physical and life science students, NPL is population of native physical and life science students, and SK is population of Ph.D. holding scientists and engineers employed in the United States. RD is total R&D expenditure in the US, FRD is total federally-funded R&D expenditure at US universities, and IRD is total industrial R&D expenditure. All these numbers are annual. UPSTOCK is cumulative university patent stock (where the 2000 figure is the sum of 1996-2000 levels of utility patents awarded to universities), OPSTOCK is cumulative other institution patent stock, and TOTPSTOCK is total cumulative patent stock.

Tpa Foreng Neng Forpl Npl Sk Rd Frd Ird Ed Upstock Opstock Totpstock

Tpa 1

Foreng .40 1

Neng .17 .77 1

Forpl .57 0.96 0.78 1

Npl .65 -.25 .24 -.13 1

Sk .90 .57 .60 .67 .45 1

Rd .82 0.82 .70 .92 -.01 .65 1

Frd .96 .67 .71 .80 .43 .90 .82 1

Ird .95 .73 .70 .85 .31 85 .92 .96 1

Ed -.05 -.40 -.14 -.45 -.10 -.52 -.33 -.59 -.44 1

Upstock .98 .42 .63 .65 .62 .90 .71 .97 .91 -.65 1

Opstock .92 .57 .51 .54 .65 .90 .54 .91 .80 -.55 .96 1

Totpstock .91 .43 .52 .65 .65 .90 .54 .91 .80 -.39 .96 1 1

Page 25: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

24

Appendix 5: Foreign and Native Engineering Graduate Students, Total Utility Patent Applications, Total Utility Patent Grants, University Patent Grants, and Non-University Patent Grants, 1980-1996 (1) (2) (3) (4) lnTPA lnTPG lnUIPG lnOIPG

(i) CONSTANT -15.810

(4.229)*

4.760

(3.924)

7.885

(6.704)

2.739

(4.402)

(ii) lnFORENG -.922

(.298)*

-.025

(.318)

1.244

(.358)*

-.109

(.298)

(iii) lnNENG .573

(.176)*

.862

(.280)**

.315

(.342)

.870

(.261)*

(iv) lnSK .583

(.392)

-.088

(.431)

.181

(.426)

-.180

(.418)

(v) lnFRD .091

(.426)

(vi) lnIRD .239

(.240)

(vii) lnRD 1.820

(.270)*

.180

(.431)

(viii) lnUPSTOCK .511

(.254)***

(ix) lnOPSTOCK 1.087

(.259)*

(x) lnTOTPSTOCK .739

(.186)*

1.189

(.198)*

(xi) ED -.006

(.024)

.067

(.037)***

.082

(.047)

.054

(.045)

R-squared 0.980 0.961 0.982 0.961

Notes: All variables (except ED, enforcement) are proportions of labor force. lnTPA is natural log of total utility patent applications, lnTPG is natural log of total utility patents granted, lnUIPG is natural log of utility patents granted to universities, and lnOIPG is natural log of those granted to other institutions. lnFORENG is natural log of foreign engineering graduate students, lnNSOC is natural log of native engineering graduate students, lnSK is natural log of total PhD scientists and engineers, lnRD is natural log of total real US R&D expenditure, lnFRD is natural log of real federally-financed R&D expenditure, and lnIRD is natural log of real industrial R&D expenditure. lnTOTPSTOCK is natural log of cumulative utility patents awarded, lnUPSTOCK is natural log of cumulative university utility patents awarded, and lnOPSTOCK is natural log of cumulative utility patents awarded to other institutions. ED is a dummy variable for restrictive and lenient student visa issuance and enforcement. Variables in the lnTPA equations are lagged five years; variables in the lnTPG, lnUIPG, and lnOIPG equations are lagged seven years. Figures in parentheses are the standard errors and are marked as significantly different from zero at the one-percent (*), five-percent (**), and ten-percent (***) levels. Estimation technique is ordinary least squares (OLS).

Page 26: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

25

Appendix 6: Foreign and Native Physical and Life Sciences Graduate Students, Total Utility Patent Applications, Total Utility Patent Grants, University Patent Grants, and Non-University Patent Grants, 1980-1996 (1) (2) (3) (4)

Notes: All variables (except ED, enforcement) are proportions of labor force. lnTPA is natural log of total utility patent applications, lnTPG is natural log of total utility patents granted, lnUIPG is natural log of utility patents granted to universities, and lnOIPG is natural log of those granted to other institutions. lnFORPL is natural log of foreign physical and life sciences graduate students, lnNSOC is natural log of native physical and life sciences graduate students, lnSK is natural log of total PhD scientists and engineers, lnRD is natural log of total real US R&D expenditure, lnFRD is natural log of real federally-financed R&D expenditure, and lnIRD is natural log of real industrial R&D expenditure. lnTOTPSTOCK is natural log of cumulative utility patents awarded, lnUPSTOCK is natural log of cumulative university utility patents awarded, and lnOPSTOCK is natural log of cumulative utility patents awarded to other institutions. ED is a dummy variable for restrictive and lenient student visa issuance and enforcement. Variables in the lnTPA equations are lagged five years; variables in the lnTPG, lnUIPG, and lnOIPG equations are lagged seven years. Figures in parentheses are the standard errors and are marked as significantly different from zero at the one-percent (*), five-percent (**), and ten-percent (***) levels. Estimation technique is ordinary least squares (OLS).

lnTPA lnTPG lnUIPG lnOIPG

(i) CONSTANT -3.221

(5.681)

16.239

(4.945)*

22.194

(11.326)***

12.461

(6.186)***

(ii) lnFORPL -.051

)243(.

.745

(.255)**

1.822

(.524)*

557.

(.277)***

(iii) lnNPL 1.183

(.588)***

1.535

(.611)**

.407

(1.019)

1.503

(.597)**

(iv) lnSK .426

(.469)

.176

(.503)

.362

(.478)

.208

(.429)

(v) lnFRD -.898

(.808)

(vi) lnIRD .070

(.384)

(vii) lnRD 1.399

(.351)*

-.192

(.423)

(viii) lnUPSTOCK .884

(.527)

(ix) lnOPSTOCK .531

(.442)

(x) lnTOTPSTOCK .473

(.303)

.583

(.360)

(xi) ED .003

(.037)

.047

(.053)

.052

(.056)

.0318

(.0645)

R-squared 0.96 0.946 0.9767 0.945

Page 27: THE IMPACT OF INTERNATIONAL SCIENCE AND …munia/590/Paperxiaowang.pdf · THE IMPACT OF INTERNATIONAL SCIENCE AND ENGINEERING GRADUATE STUDENTS ON US INNOVATION Xiao Yu Wang xywang@wisc.edu

26

Appendix 7: Foreign and Native Science and Engineering Students, Total Utility Patent Applications, Total Utility Patent Grants, University Patent Grants, and Non-University Patent Grants, 1980-1996 (1) (3) (4) (5) lnTPA lnTPG lnUIPG lnOIPG

(i) CONSTANT -5.497

(3.372)

10.833

(4.356)**

5.913

(3.980)

6.025

(2.693)**

(ii) lnFORSE .236

(.273)

.652

(.232)**

1.253

(.221)*

.154

(.187)**

(iii) lnNSE -1.509

(.422)*

1.290

(.706)***

.852

(.488)***

1.527

(.370)*

(iv) lnUPSTOCK .803

(.145)*

.469

(.115)*

(v) lnOPSTOCK -1.085

(.454)**

-.510

(.355)

(vi)lnTOTPSTOCK 1.598

(.188)*

.813

(.294)**

R-squared 0.866 0.892 0.973 0.948

Notes: All variables are proportions of labor force. lnTPA is natural log of total utility patent applications, lnTPG is natural log of total utility patents granted, lnUIPG is natural log of utility patents granted to universities, and lnOIPG is natural log of those granted to other institutions. lnFORSE is natural log of foreign science and engineering graduate students, lnNSE is natural log of native science and engineering graduate students. lnTOTPSTOCK is natural log of cumulative utility patents awarded, lnUPSTOCK is natural log of cumulative university utility patents awarded, and lnOPSTOCK is natural log of cumulative utility patents awarded to other institutions. Variables in the lnTPA equations are lagged five years; variables in the lnTPG, lnUIPG, and lnOIPG equations are lagged seven years. Figures in parentheses are the standard errors and are marked as significantly different from zero at the one-percent (*), five-percent (**), and ten-percent (***) levels. Estimation technique is ordinary least squares (OLS).