Americas leaning ivory tower the measurement of and response to concentration of federal funding for academic research

A few federal agencies such as the National Science Foundation and theNational Institute of Health receive billions of appropriated funds each year tosupport academic research, and they

SPRINGER BRIEFS IN POLITICAL SCIENCE

Yonghong Wu

America’s Leaning Ivory Tower

SpringerBriefs in Political Science

Yonghong Wu

Tower

The Measurement of and Response

to Concentration of Federal Funding for Academic Research

123

Department of Public Administration

University of Illinois at Chicago

Chicago, IL, USA

SpringerBriefs in Political Science

ISBN 978-3-030-18703-3 ISBN 978-3-030-18704-0 (eBook)

https://doi.org/10.1007/978-3-030-18704-0

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2020

This work is subject to copyright All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, speci fically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on micro films or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar

or dissimilar methodology now known or hereafter developed.

The use of general descriptive names, registered names, trademarks, service marks, etc in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made The publisher remains neutral with regard

to jurisdictional claims in published maps and institutional af filiations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

As a scholar in thefield of public budgeting and finance, I have a genuine interest inthe allocation of government resources Government allocation decisions affect theoperation of various public, private, and not-for-profit organizations and, ultimately,the lives of people in this country Because elected representatives make budgetarydecisions at all levels of government, government resource allocation is basically anoutcome of the political process However, federal resource allocation does not endwhen Congress passes appropriation bills and the president signs them.Government agencies often need to further allocate appropriated funds to otherinstitutions and individuals While public budgeting andfinance scholars focus onthe politics of government budgeting, the post-appropriation allocation has not beenstudied enough

A few federal agencies such as the National Science Foundation and theNational Institute of Health receive billions of appropriated funds each year tosupport academic research, and they allocate the funds to higher education insti-tutions and researchers via a competitive selection process With my bachelor’sdegree in applied physics and doctoral training in science and technology policy, Iunderstand the rationales of government funding of academic research and the needfor merit-based competition in funding scientific research However, a competition-based allocation mechanism has led to a substantial concentration of federal funding

in high-capacity states, while low-capacity states are largely left underfunded Theuneven distribution of federal funding for academic research has been an importantpublic policy issue for decades

Under the mandate of Congress, some federal agencies launched special grams to address funding concentration issue in the 1980s and 1990s Although theprograms have shown some modest effects on research-capacity building throughinvesting in collaboration development and infrastructure improvement, insufficientattention to the institutional environment limits the progress toward a more equi-table distribution of federal research funding across jurisdictions I feel the obli-gation to thoroughly investigate the issue of uneven distribution of federal resourcesand recommend ways to improve the effectiveness of government efforts in thisarena

pro-v

The book is the product of my long-term focus and research on this topic.

I sincerely hope that this work will have a positive impact on policy-making andprogram implementation aimed at addressing this funding concentration in theUnited States

I am indebted to several people who assisted my research on this topic JuliaMelkers was the co-author of my first paper on EPSCoR, which inspired mydecade-long interest in the topic Eric W Welch generously shared materials anddata that are helpful to this work I also want to thank the external reviewers fortheir thoughtful comments

Finally, this book is dedicated to my wife, Karen, and my children, Jerry andCindy This book would not have been possible without their love and support

1 The Funding of Academic Research in the U.S 1

1.1 Fiscal Federalism in Financing Academic Research 2

1.2 Trends of Fiscal Federalism in Funding Academic Research 3

1.3 Substantial Disparity in Federal Funding of Academic S&E Research 5

References 10

2 Geographical Concentration of Funding of Academic Research 11

2.1 Measurement of Concentration of Federal Funding of Academic Research in the U.S 12

2.2 Causes and Consequences of Uneven Distribution of Federal Funding of Academic R&D 23

References 27

3 Public Policy Response to Concentration of Academic Research 29

3.1 History of NSF’s EPSCoR and Similar Programs 29

3.2 State-Level EPSCoR Coordination and Heterogeneity 33

3.3 An Evaluative Framework on EPSCoR 38

References 40

4 Assessment of Scientists’ Research Capacity 43

4.1 Empirical Test of the Determinants of Individual Research Capacity 44

4.2 Comparison of Research Capacity Between EPSCoR and Non-EPSCoR States 48

References 55

5 Multi-level Assessment on EPSCoR 57

5.1 The Changing Top 100 Recipients of Federal Academic R&D Support 58

5.2 Macro-level Assessment of Concentration of Federal Funding for Academic Research 59

vii

5.3 State-Level Assessment of NSF EPSCoR and NIH IDeA 64

References 72

6 EPSCoR Programs and Research Facilities 73

6.1 Size, Funding and Density of Academic Research Facilities 74

6.2 The Impact of EPSCoR on Funding of Research Facilities 81

References 86

7 The Future of EPSCoR 89

7.1 Evolving Political Support of EPSCoR 89

7.2 Strategic Shift: From Infrastructure Improvement to Institutional Innovation 92

References 95

Appendix 97

List of Figures

expenditures for all higher education institutions (in billions

constant dollars) The gray solid and dashed lines represent

expenditures for public higher education institutions

(in billions constant dollars) 3

Fig 1.2 Shares of federally and state/local financed higher education R&D expenditures.Note The black solid and dashed lines represent the shares of federally and state/localfinanced R&D expenditures for all higher education institutions (in %) The gray solid and dashed lines represent the shares of federally and state/local financed R&D expenditures for public higher education institutions (in %) 4

Fig 1.3 State’s share of federal academic R&D support in 2015 6

Fig 2.1 State’s share of federal academic R&D support in 1975 12

Fig 2.2 State’s share of doctorate recipients in engineering in 2016 26

Fig 2.3 State’s share of doctorate recipients in Physical Sciences, Life Sciences, Math and Computer Sciences and Geosciences in 2016 27

Fig 3.1 Framework of research capacity and competitiveness 39

Fig 6.1 State’s share of total space for S&E research in 2003 75

Fig 6.2 State’s share of total space for S&E research in 2015 76

Fig 6.3 Funds for new construction of research facility by source—all academic institutions Note The solid, dashed, and dotted lines represent funds from institutions or other sources, state/local governments, and federal government (in billions constant dollars) 77

ix

Fig 6.4 Funds for repair/renovation of research facility by source—all

represent funds from institutions or other sources, state/local

governments, and federal government

dashed, and dotted lines represent funds from institutions

or other sources, state/local governments, and federal

dashed, and dotted lines represent funds from institutions

or other sources, state/local governments, and federal

List of Tables

R&D support by year 13

Table 2.2 Descriptive statistics for state’s share of NSF academic R&D support by year 14

Table 2.3 Concentration index for 50 states 17

Table 2.4 Concentration index for groups of 5-state (G5) 19

Table 2.5 Concentration index for groups of 10-state (G10) 20

Table 3.1 EPSCoR jurisdictions and their years of entry 31

Table 3.2 NSF EPSCoR funding by year 32

Table 3.3 Major initial NSF EPSCoR awards infive states 36

Table 4.1 Regression analysis of scientists’ research capacity 47

Table 4.2 Characteristics of respondents from EPSCoR versus non-EPSCoR states 50

Table 4.3 Comparison of collaborative networks (1) 50

Table 4.4 Comparison of collaborative networks (2) 52

Table 4.5 Comparison of satisfaction with work environment 53

Table 4.6 Comparison of grant-seeking performance 54

Table 5.1 Distribution of top 100 academic institutions receiving federal R&D support 60

Table 5.2 Macro analysis of NSF EPSCoR 62

Table 5.3 Macro analysis of NIH IDeA 63

Table 5.4 State-level analysis of NSF EPSCoR 67

Table 5.5 State-level analysis of NIH IDeA 68

Table 5.6 NSF EPSCoR effects by state 70

Table 5.7 NIH IDeA effects by state 71

Table 6.1 Comparison of research density—all academic institutions 80

Table 6.2 Comparison of research density—public academic institutions 80

Table 6.3 Regression analysis of funding of both new construction and repair/renovation projects 84

xi

Table 6.4 Regression analysis of funding of new construction

of academic research This book describes the funding disparity aspect of highereducation in the U.S and assesses the effectiveness of federal programs tackling thisissue.

The main goal of this chapter is to introduce readers to the broad context ofgovernment funding of academic research In the American science policy arena,there have been continuous debates on peer-review versus equity-based approaches

to the allocation of federal research funding The peer-review system has been theprimary mechanism for distributing federal government funding for research amonguniversities since shortly after World War II Peer review ensures the production ofthe best science by funding the most capable researchers in the country As a result,federal research funding has been concentrated in “high-capacity” states where many

of the most capable researchers reside, and a large number of “low-capacity” stateshave received substantially less research funding from federal agencies

In fiscal year 2016, all higher education institutions in the U.S spent a total of

$67.7 billion in their conduct of research and development (R&D) in science and neering (S&E) Public institutions spent $44.2 billion, about 65.4% of total academicR&D expenditures in that year The money spent by higher education institutionscame from different sources As the primary sponsor of academic research, the fed-eral government provided $37.7 billion or about 55.7% of total R&D spending withinuniversities in 2016 ($23.1 billion of which went to public universities) State gov-ernments contributed $3.7 billion, about 5.5% of total academic R&D expenditures

engi-in 2016 ($3.4 billion to universities under their control) Other sources engi-included the

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2020

Y Wu, America’s Leaning Ivory Tower, SpringerBriefs in Political Science,

https://doi.org/10.1007/978-3-030-18704-0_1

1

business sector, higher education institutions, and other organizations, accountingfor 6.0%, 23.5%, and 9.3%, respectively.

The most recent data on academic R&D expenditures reveal important features

of financing the conduct of S&E research in American colleges and universities.Scientists and engineers primarily rely on government funding for research, andthe federal government plays a dominant role by providing a much larger share

of financial support than state governments Federal dominance in this area is aresult of investment that began shortly after World War II, when Vannevar Bushrecommended that the federal government take the lead in “promoting the flow ofnew scientific knowledge and the development of scientific talent” (Bush, 1945,

p 4) The involvement of state governments came later, as state governments startedenhancing their efforts to assert a greater role in the formulation and administration

of national science and technology policies beginning in the early 1980s (Feller,1997)

1.1 Fiscal Federalism in Financing Academic Research

The term federalism is used to describe a system of government in which the power togovern is constitutionally divided between a central (federal) governing authority andconstituent political units (like states) Scholars have developed different theoreticalmodels (dual, cooperative, and coercive federalism) that are dominant in differenttimes and applicable to different policy arenas For instance, the cooperative fed-eralism model posits that federal and state governments interact cooperatively andcollectively to solve common problems

Federalism in funding academic research refers to the shared funding bility for S&E research by both federal and state governments Federalism in sciencepolicy has been cooperative over time Research funding comes from several sources,including federal and state governments The federal government never told state gov-ernments to support or not support any particular type of research Funding decisions

responsi-at the stresponsi-ate level are independent of federal government, except for the cost-sharingrequirement of federal research grants For instance, while the Bush administrationforbade the use of federal funds for research involving the destruction or creation ofembryos, some states could still step in to advance this vital research without federalpreemptions

Fiscal federalism is based on the idea that a public service should be financed insuch a way that the benefits are confined to the jurisdiction financing the service.This no-spillover arrangement is supposed to achieve and maintain efficient decision-making at different levels of government The application of this theory in scientificresearch means that federal and state governments share responsibility in proportion

to the expected scope of benefit from specific research projects Federal governmentshould support research projects that benefit the entire country (like basic research)

or meet national needs (like mission-driven research) On the other hand, it is thestates’ responsibility to support research projects with outputs benefiting primarilytheir jurisdictions

1.2 Trends of Fiscal Federalism in Funding Academic Research 3

Fig 1.1 Amounts of federally and state/local financed higher education R&D expenditures Note

The black solid and dashed lines represent the amounts of federally and state/local financed R&D expenditures for all higher education institutions (in billions constant dollars) The gray solid and dashed lines represent the amounts of federally and state/local financed R&D expenditures for public higher education institutions (in billions constant dollars)

1.2 Trends of Fiscal Federalism in Funding Academic

Research

In order to understand how fiscal federalism works, I examine the pattern and trends

of academic research funding over time in terms of annual shares of funding foracademic research by federal and state governments The data source is the National

Science Foundation (NSF)’s Survey of Research and Development Expenditures at

Universities and Colleges The survey has been collecting data since 1972 on

sepa-rately budgeted R&D expenditures within academic institutions by source of funds,including federal government, state/local governments, businesses, higher-educationinstitutions, and other sources (NSF,2011) The data are collected from universitiesdirectly, using consistent, uniform definitions and collection techniques

Figures 1.1 and 1.2 show the main pattern and time trends of federally andstate/local financed higher education R&D expenditures for all or public higher edu-cation institutions, respectively The actual amounts (converted to constant dollars)1are presented in Fig 1.1and the shares as percentage of total expenditures are inFig.1.2

The two figures show that the federal government financed increasing amounts

of academic R&D during 1972–2016, whereas state government support was fairlystable The gap between federally financed R&D expenditures for all and public insti-tutions indicates that federal agencies also provide substantial amounts of financial

1 I use the GDP implicit price deflator with base year 2009 as of July 2017.

Fig 1.2 Shares of federally and state/local financed higher education R&D expenditures Note

The black solid and dashed lines represent the shares of federally and state/local financed R&D expenditures for all higher education institutions (in %) The gray solid and dashed lines represent the shares of federally and state/local financed R&D expenditures for public higher education institutions (in %)

support to private universities The ratio of federally financed R&D expenditures ofpublic to private institutions has been quite constant, moving from 1.3 in the 1970sand 1980s to 1.5–1.6 in the 1990s and afterward Meanwhile, state/local financialsupport overwhelmingly flows to public institutions

The two figures also demonstrate that (1) government funding of academic R&Dhas been dominant among all sources, with government’s share of total academicR&D expenditures between 61% and 79% for all higher education institutions,60–78% for public institutions only; (2) the government funding of academic R&Dhas been primarily from the federal government, with federal share between 56%and 69% for all higher education institutions, and 52–63% for public institutionsonly; (3) the state governments play a relatively minor role in this regard, only pro-viding 5–10% and 7–15% of academic R&D expenditures for all and public highereducation institutions, respectively

Federal dominance has been fairly stable over the years At the start of this period,the federal shares of academic R&D expenditures were about 68 and 63% for all andpublic higher education institutions The two shares declined slightly over time to 58and 52% in the year 2000; increased modestly during 2000–2005 and 2009–2011 The

1.2 Trends of Fiscal Federalism in Funding Academic Research 5state share of academic R&D expenditures for all higher education institutions hasdecreased almost continuously from slightly above 10% in the early 1970s to about5.5% at the end of the period Although state/local financed R&D expenditures aremostly taken by public institutions, the state share of academic R&D expendituresfor public institutions has declined substantially, dropping from 14% in the early1970s to 7.7% in 2016.

1.3 Substantial Disparity in Federal Funding of Academic S&E Research

In addition to the overall pattern of shared funding by federal and state governments,

it is important to examine the distribution of federal funding of academic R&D Thepursuit of efficiency in the conduct of academic research mandates the use of peerreview in allocating federal support in this arena However, the dominance of gov-ernment, particularly federal government, legitimatizes the concern for inequitabledistribution of federal resources across jurisdictions

Figure1.3presents each state’s share of federal R&D support to academic tions in 2015 The federal R&D support refers to the federal obligations for academicR&D in S&E fields It covers all direct, indirect, incidental, or related costs resultingfrom or necessary to the performance of R&D by private individuals and organi-zations under grant, contract, or cooperative agreement, as well as demonstrationprojects and research equipment (NSF,2015) The data source is the NSF’s Survey

institu-of Federal Science and Engineering Support to Universities, Colleges, and Nonprinstitu-ofit Institutions The survey includes all academic institutions that receive funding from

federal agencies that finance federal R&D obligations to the academic sector Thedata are collected from federal agencies directly

The disparity in federal academic R&D funding is substantial The top ten statesreceived an aggregation of over 60% of total federal R&D support to academicinstitutions in 2015, while the bottom ten states received less than 2% of the totalfederal support in that year The disparity in per capita terms becomes less impressive

as the top ten states account for about 48% of the total U.S population, and onlyabout 4.5% reside in the bottom ten states.2 However, the disparity remains quitesubstantial when I compare individual states The state of Maryland, with about1.9% of the U.S population, received 6.4% of federal R&D support; the state ofWest Virginia, with about 0.9% of U.S population, received only 0.2% of federalR&D support

The federal pursuit of efficiency in the conduct of scientific research coupledwith uneven distribution of research capacity plays a leading role in this fundingdisparity Competition for federal S&E research funding is primarily merit-basedvia peer review The concentration of academic research in a few states is a naturalresult of agglomeration of prestigious research universities in those states Aside from

2 I use 2010 Census data from https://www.census.gov/2010census/data/

Fig 1.3 State’s share of federal academic R&D support in 2015

getting more or less federal research dollars, education and training opportunities forcollege students in S&E fields are affected, and spillover benefits from federallyfunded research projects such as patents with potential commercial prospects andthe incubation of new industries and products become similarly concentrated Federaldominance coupled with the uneven distribution of federal funds raises legitimateconcerns about how federal academic R&D dollars should be distributed amongindividual states

The substantially uneven distribution of federal funding for academic researchhas been an important public policy issue for decades Under the mandates ofCongress, NSF launched the first Experimental Program to Stimulate CompetitiveResearch (EPSCoR) to support low-capacity jurisdictions in 1979, and severalother federal agencies established similar funding programs in the 1990s Thenumber of EPSCoR-eligible jurisdictions increased from 5 in 1980 to 31 in 2012.Decades of federal EPSCoR efforts have not delivered the expected results Asthe America COMPETES Reauthorization Act of 2010 states, “National ScienceFoundation funding remains highly concentrated, with 27 states and two territories,taken together, receiving only about 10% of all NSF research funding.” The book isintended to provide a comprehensive assessment on the effectiveness of the EPSCoR

1.3 Substantial Disparity in Federal Funding of Academic S&E Research 7programs in mitigating undue concentration of academic research funding acrossstates Although the policy goals have been expanded over time, the improvement ofresearch capacity and enhancement of research competitiveness have remained theprimary goals of EPSCoR The assessment focuses on scientists’ research capacity

as measured by their grant-seeking performance, and jurisdictional research petiveness as measured by the success in winning federal academic R&D support.There has been an increasing academic and non-academic literature on issues offunding academic research and related policy initiatives, most of which is in journalarticles and reports of think-tanks, consulting firms, and organizations such as theNational Academies and the American Association for the Advancement of Science(AAAS) A National Academies’ 2013 report, based on a comprehensive examina-tion of EPSCoR’s evolving mission, program operations, and program evaluation,makes a number of recommendations to improve the effectiveness of EPSCoR pro-grams (National Academies, 2013) Two years later, the Science and TechnologyPolicy Institute published another report on EPSCoR (Zuckerman et al.,2015) Theresearch team collected a lot of data from a variety of relevant sources, and conducteddescriptive comparisons between EPSCoR and non-EPSCoR jurisdictions

com-The two reports provide valuable operational and management details of theEPSCoR programs However, they are not program evaluations by academic stan-dard With a focus on the issue of inequity, this book takes a broader view of gov-ernment funding of academic research by quantifying the degree of concentration inthis area, and contributing a multi-level, vigorous assessment on government effortstackling this issue While EPSCoR efforts have shown some effects on researchcapacity-building through investing in collaboration development and infrastructureimprovement, insufficient attention to institutional environment limits the progresstoward a more equitable distribution of federal research funding In addition, thesize of research facilities relative to the academic R&D expenditures is significantlylarger in EPSCoR states, indicating over-investment in physical research infrastruc-ture and inefficiency in the conduct of research in EPSCoR institutions Our analyticalresults indicate that it is the time to shift EPSCoR focus from research infrastruc-ture improvement and collaboration development to innovating institutional envi-ronments to recruit and motivate scientists

This book incorporates extensive quantitative description and regression ses Data from authoritative sources and graphical tools are employed to illustratethe extent of concentration of federal funding of academic research in the U.S Panelregression techniques are used to test the hypothesis about how EPSCoR programshave affected various measures of research capacity and competitiveness of EPSCoRjurisdictions Since research universities in the EPSCoR states are predominatelypublic universities, the book also examines state government investments in the con-struction and renovation of physical research infrastructure in EPSCoR jurisdictions.After the introduction of the broad context of government funding of academicresearch in America, Chap.2focuses on the measurement of jurisdictional concen-tration of federal funding of academic research in the U.S Beyond the comparison ofstates’ shares of total federal obligations for academic R&D, conventional descriptivestatistics such as mean and standard deviation and a newly developed concentration

analy-index are used to describe the extent of concentration of academic research ing from several federal agencies In particular, several group-based concentrationindices are introduced to succinctly summarize the level of jurisdictional concentra-tion of federal obligation for academic R&D The group-based concentration indexhas the advantage to avoid false indication of policy effect The chapter concludeswith a brief discussion of equity implications of the uneven distribution of federalresearch funding by showing that academic R&D funding is closely tied to a state’seducational opportunities and economic growth.

fund-Chapter3describes federal government response to the uneven distribution of demic research funding The chapter briefly reviews the history of federal EPSCoRprograms, particularly NSF’s EPSCoR and National Institutes of Health (NIH)’sInstitutional Development Award (IDeA), and describes evolving policy goals andprogrammatic features of the programs and capacity-building activities in highereducation institutions in the eligible states State-level EPSCoR coordination andheterogeneity are discussed as well I also develop an evaluative framework ofresearch capacity and competiveness as a conceptual guide to the subsequent multi-level assessment of EPSCoR effects on research capacity at the individual level andresearch competitiveness at the jurisdiction level The framework encompasses tal-ent, collaboration, support, and motivation as four key determinants of individualresearch capacity, because the ability to conduct scientific research relies not only

aca-on the scientific and collaborative abilities of the researchers but also aca-on their access

to necessary facilities and equipment and encouraging institutional and work ronments

envi-The first part of Chap.4is an empirical test of the evaluative framework trated in Chap.3 Using a recent data set of a sample of academic scientists, I developmeasures of talent, collaboration, support, and motivation, and examine how thesemeasures affect scientists’ research capacity as demonstrated by their grant-seekingperformance The focus on scientists’ grant-seeking performance is closely related

illus-to the primary goal of EPSCoR in the pursuit of a more equitable distribution offederal research funding After the evaluative framework is empirically validated, Iconduct an assessment of EPSCoR efforts in building scientists’ research capacity

in the eligible jurisdictions by comparing the mean values of the four key nants of individual research capacity between scientists in EPSCoR states and those

determi-in other states The analysis of variance results suggest that determi-individual scientists determi-inEPSCoR states do not show significant weakness in research talent, collaboration,and motivation, and they seem to perform equally well in grant-seeking as theircounterparts in non-EPSCoR states But the results also reveal important frustra-tions among scientists in EPSCoR states that EPSCoR initiatives might address andmitigate

Chapter5provides a comprehensive and updated assessment on the effectiveness

of the EPSCoR beyond the individual level It begins with a descriptive analysis

of the mobility of the top 100 academic institutions in receipt of federal R&Dsupport from 1975 to 2015 This institution-level analysis reveals the dominance of

1.3 Substantial Disparity in Federal Funding of Academic S&E Research 9non-EPSCoR institutions among the top competitors for federal funding of academicR&D and a modest gain by academic institutions in EPSCoR states It is followed

by a macro-level assessment showing that the two largest EPSCoR programs—NSFEPSCoR and NIH IDeA—have been effective in reducing the concentration index

of the respective agency support to academic research, but the magnitude of theeffects is small Two additional state-level assessments present quite modest effects

of NSF EPSCoR and NIH IDeA on a state’s shares of NSF and NIH obligations foracademic R&D, respectively In consideration of the heterogeneity of state EPSCoRprograms, supplemental analysis is also performed on the share of NSF or NIHfunding of academic R&D for each state to identify the varying effects of EPSCoRacross the eligible states These assessments are complementary to each other, andcollectively provide solid empirical evidence on the effects of EPSCoR on variousmeasures of research capacity and competiveness

Chapter6focuses on the construction and renovation of research infrastructure inhigher education institutions Research infrastructure is a critical pillar of academicresearch capacity and has been a primary focus of EPSCoR since the early 2000s Ifirst develop a measure of research density by comparing the R&D expenditures made

by academic institutions within a state with the size of its academic research ties The analysis shows that EPSCoR states have a larger size of research facilitiesrelative to their academic R&D expenditures than non-EPSCoR states, indicatingthat EPSCoR institutions have likely over-invested resources in physical researchinfrastructure and do not utilize research facilities as efficiently as their counterparts

facili-in non-EPSCoR states The chapter also demonstrates that state governments havebeen playing a more important role than federal government in funding of researchfacilities The empirical evidence furthermore shows that EPSCoR state governments

do not invest significantly more funds in research facilities than non-EPSCoR states

I conclude in Chap.7with a synthesis of the analyses and a discussion of the cations for the future of EPSCoR and similar efforts to address the concentration offederal funding for academic research Although EPSCoR efforts have been effective

impli-in buildimpli-ing scientists’ research capacity, the limited effects at impli-institutional, state andnational levels indicate the need for program improvement Our empirical evidencesuggests that scientists are significantly dissatisfied with institutional environments

in the EPSCoR states, and this may limit the progress toward a more equitable bution of federal research funding at the institution or state level I also find evidence

distri-of redundancy and inefficiency in the construction and utilization distri-of research ities in the EPSCoR states The book therefore calls for a shift in EPSCoR strategyfrom research collaboration and infrastructure to innovating and improving institu-tional environments that help recruitment, retention, and motivation of S&E researchtalents The chapter also describes evolving political support for EPSCoR, and makesadditional recommendations to improve EPSCoR’s effectiveness

Bush, V (1945) Science, the endless frontier: A report to the President Washington, DC: U.S.

Government Printing Office.

Feller, I (1997) Federal and state government roles in science and technology Economic

Devel-opment Quarterly, 11(4), 283–295.

National Academies (2013) The experimental program to stimulate competitive research

Wash-ington, DC: The National Academies Press.

National Science Foundation (2011) Academic research and development expenditures: Fiscal

year 2009 (NSF 11-313) Arlington, VA: National Science Foundation.

National Science Foundation (2015) Federal science and engineering support to universities,

colleges, and nonprofit institutions: FY 2013 (NSF 15-327) Arlington, VA: National Science

Foundation.

Zuckerman, B L., et al., (2015) Evaluation of the National Science Foundation’s Experimental

Program to Stimulate Competitive Research (EPSCoR): Final report IDA Paper P-522 Science

and Technology Policy Institute.

concentra-This uneven pattern of distribution has persisted for a long time Similar to Fig.1.3,Fig.2.1shows that the shares of federal R&D funding received by academic institu-tions in the 50 states differed substantially in 1975 Together, the figures show thatfederal support of academic R&D is concentrated in a few states, and a large number

of the 50 states have received minimal proportions The top ten and the bottom tenstates largely overlap, even with 40 years between the data in the figures California,New York, Maryland, Pennsylvania, Massachusetts, Texas, Illinois, and Michiganare among the top ten states in both 1975 and 2015 Washington and Ohio only appear

in the top-ten list of 1975, whereas North Carolina and Georgia make the list in 2015.Although four states get in and out of the top-ten list over the period 1975–2015, thechange is not dramatic For instance, Washington and Ohio dropped from 8th and10th in 1975 to 13th and 11th in 2015 North Carolina and Georgia moved up from12th and 19th in 1975 to 7th and 10th in 2015 In other words, the top winners offederal academic R&D funding are virtually the same in 1975 and 2015

Conversely, North Dakota, Nevada, Arkansas, Idaho, South Dakota, West ginia, Wyoming, and Maine are in the bottom ten states in both 1975 and 2015.Montana and Delaware were among the bottom ten states in 1975 (43rd and 44th)and moved up to 39th and 40th in 2015 Alaska and Vermont, on the other hand,were 35th and 38th in 1975, but dropped to 41st and 42nd in 2015 It seems thatthe winners and losers in receiving federal academic R&D dollars in the U.S havestayed essentially the same for 40 years

Vir-© The Author(s), under exclusive license to Springer Nature Switzerland AG 2020

Y Wu, America’s Leaning Ivory Tower, SpringerBriefs in Political Science,

https://doi.org/10.1007/978-3-030-18704-0_2

11

Fig 2.1 State’s share of federal academic R&D support in 1975

2.1 Measurement of Concentration of Federal Funding

of Academic Research in the U.S.

We can use one graph to show the disparity of federal funding of academic R&Dwithin a particular year, but the 45 figures for the years 1971–2015 are too complex toreveal the pattern of change from year to year To present a clearer picture of fundingover time, I develop some common descriptive statistics such as minimum, maxi-mum, mean and standard deviation of state’s share of federal funding for academicR&D in each year Table2.1presents the statistics for the share of federal academicR&D support by year from 1971 to 2015 Because NSF has been an important spon-sor of academic research, I also present the statistics for the share of NSF academicR&D support by year in Table2.2

2.1 Measurement of Concentration of Federal Funding … 13

Table 2.1 Descriptive statistics for state’s share of federal academic R&D support by year

Year Number of

states

Mean (%)

Standard deviation (%)

Table 2.1 (continued)

Year Number of

states

Mean (%)

Standard deviation (%)

Standard deviation (%)

2.1 Measurement of Concentration of Federal Funding … 15

Table 2.2 (continued)

Year Number of

states

Mean (%)

Standard deviation (%)

at quite a slow pace

Table2.2shows a similar decline of dispersion in state’s share of NSF fundingfor academic research The standard deviation was at its peak in 1976 (3.58%) and

dropped to 2.43% in 2009, then rose briefly and declined again to 2.41% after 2012.Similar to total federal support of academic R&D, the mean shares of NSF support arefairly constant in this period with the exception of some minor sliding in 2009–2014.The combination of mean and standard deviation is helpful to describe the dis-tribution of an important variable The standard deviation is a common measure

of overall deviations of individual observations from the mean A smaller standarddeviation indicates that the individual observations converge to the mean However,

it is not particularly useful in describing the degree of geographic concentration offederal funding because the computation of standard deviation relies on the mean ofthe data In other words, the mean should be constant for standard deviation to be acompatible measure of dispersion over time

I introduce an alternate measure that is not dependent upon the mean value ineach year Rather than calculating the sum of squared deviations from the mean (theformula of standard deviation), I simply square each state’s share of federal academicresearch funding and sum them up In one year, 50 squared shares are aggregated toget what I would call the concentration index in that year

The concentration index has several desirable features First, its range is from1/50 (equal distribution of federal funding) to 1 (maximum concentration of federalfunding) The index takes the value 1/50 only when all 50 states get equal shares offederal academic research funding In the scenario of maximum concentration, allfederal funding goes to one state and the other 49 states receive nothing Second,when the index moves from the minimum (1/50) to the maximum (1), the degree

of concentration escalates In other words, the index values close to 1 mean highlevels of concentration, whereas the values close to 1/50 represent low levels ofconcentration

The concentration index provides an opportunity to concisely summarize with

a single numeric value the degree of geographical concentration of federal support

of academic research This index can be used as a thermostat for policy response,indicating when government should take action as the level of concentration goesbeyond a certain benchmark It also makes it easier to keep track of change ofconcentration of federal funding from year to year The index could be used in policyassessment and program evolution because a significant decrement is likely a sign

of success for policy initiatives aimed at reducing geographic concentration

I calculate the concentration index using the state shares of federal support ofacademic R&D for years from 1971 to 2015.1 I also calculate the concentrationindex for support provided by major federal sponsors of academic research such asDepartment of Defense (DOD), Department of Energy (DOE), NIH and NSF Thedetails are presented in Table2.3

1 I use the total federal support of academic R&D in 50 states rather than in the entire U.S as the denominator in the calculation of a state’s share of federal support The two share measures only differ slightly I make this choice to ensure that the concentration index is exactly 1 in the scenario

of maximum degree of concentration.

2.1 Measurement of Concentration of Federal Funding … 17

Table 2.3 (continued) Year Federal NSF DOD DOE NIH

The concentration index values are relatively small because they are fairly close

to the theoretical minimum value of 0.02 (or 1/50) This is due to another feature

of the concentration index—the index values become smaller when the number ofobservations is larger In this case, the number of observations is 50, which is solarge that each state gets a quite small share of federal funding of academic R&D.Because the squared shares are even smaller than the shares, the sum of the squaredshares gives a small concentration index

To demonstrate the impact of the number of observations, I develop differentversions of the concentration index by dividing the 50 states into groups of 25, 10, 5,and 2 states For demonstration purpose, I only present the index values for groups of

5 states (G5 index) and 10 states (G10 index) For the G5 index, all 50 states are firstranked by the share of federal academic R&D support in a particular year Then thetop five states will form one group, the second five states will form another group,until the bottom five states form the last group We end up with 10 five-state groups.For every group, I calculate their aggregate federal academic R&D support aspercentage of total federal support in that year Each group’s percent share is squaredand the ten squared group shares are summed up to calculate the G5 concentrationindex The G10 concentration index can be calculated in a similar way, except thateach group now has ten states instead of five Tables2.4and2.5present the G5 andG10 concentration indices for each year

Unlike the original concentration index (the G1 index), the group concentrationindices possess two additional features First, because the grouping is solely based

on relative rankings of individual states and state rankings may change from year

to year, one particular group (i.e., top-five group or top-ten group) may not includethe same states in different years For instance, Washington and Ohio are amongthe top-ten group in 1975 but belong to the group next to the top ten in 2015 NorthCarolina and Georgia are included in the second-ten group in 1975 but move up to thetop-ten group in 2015 It should be noted that the change of state rankings is gradual,and there is no significant change in state rankings during two adjacent years.Second, the group indices do not imply geographic agglomeration or clustering

in any way The states in the same group may spread from the east coast (New York

2.1 Measurement of Concentration of Federal Funding … 19

Table 2.4 (continued) Year Federal NSF DOD DOE NIH

2.1 Measurement of Concentration of Federal Funding … 21

Compared with the original G1 index, the G5 and G10 indices are substantiallylarger For instance, the G5 and G10 indices based on state’s shares of federal aca-demic research funding ranges 0.23–0.28 and 0.42–0.47 as compared to the originalrange of 0.05–0.06 The concentration index as constructed has a ground value theindex can never go below and a ceiling value it can never surpass in any circum-stance As discussed before, the ceiling value (or theoretical maximum value) of anyconcentration index is 1, and the ground value (or theoretical minimum value) ofthe index equals the reciprocal of the number of observations So the ground values

of G1, G5, and G10 indices are the reciprocals of 50, 10, and 5—0.02, 0.1 and 0.2,respectively

The concentration index provides a common ground to compare geographic centration of federal support of academic R&D across agencies Among the eightfederal agencies,2 the overall concentration of DOD and NASA funding for aca-

con-2 The eight agencies include DOD, DOE, NIH, NSF, Department of Agriculture (DOA), Department

of Commerce (DOC), Environmental Protection Agency (EPA), and National Aeronautics and Space Administration (NASA) They are selected because they are major federal R&D agencies.

demic research is higher than others The 45-year average G5 indices are 0.57 and0.52 for DOD and NASA respectively, which are larger than the average indices ofthe funding from other agencies This means that the funding from those two agencies

is more geographically concentrated than funding from the other six major federalsponsors of academic research The concentration of DOD and NASA is probablymore acceptable, given their mission-driven research agenda The other agencies arenot far behind (the 45-year average G5 concentration indices are between 0.46 and0.51 for five of the six agencies), but the DOA’s 45-year average G5 concentrationindex is only 0.26

If public and private academic institutions are considered separately, federal demic research funding is much more concentrated among private institutions thanpublic ones The average G5 concentration index is 0.37 for public institutions and0.75 for private institutions with the distribution of total federal support of academicR&D during 1971–2015 The concentration index is even above 0.85 for DOA, DOC,DOD, EPA, and NASA funding of academic research in private institutions Thisindicates that the distribution of academic research capacity is highly uneven amongprivate colleges and universities The establishment of public research colleges anduniversities helps alleviate the degree of concentration in this arena

aca-A value of the index closer to the ceiling means higher level of concentration.Larger values of G5 and G10 indices seem to indicate more serious concentrationissues However, it is meaningless and even misleading to compare the concentrationlevels across different versions of the index because the number of observations alsomatters to the calculation Policy assessment should focus on the change of oneparticular concentration index after the launch of a policy or program compared tothe level before the policy or program

One potential issue with the original concentration index G1 (50 states are rately considered) is that a certain change of the index could be the result of differentscenarios of redistribution of federal research funding across states, some of whichare policy relevant and others are not For instance, suppose that California, the toprecipient of federal academic R&D funding, received 10% less funding in a yearcompared to the prior year This could happen if the other top five states get what

sepa-is lost by academic institutions in California, or the bottom ten states are able toincrease their receipt of federal research funding by 80% each.3The latter scenario

is really an indication of the success of the EPSCoR-type program However, theformer scenario does not indicate any significant progress in raising the shares ofthe “low-capacity” states The two scenarios produce the same percentage decline

of the G1 concentration index, but the change itself may not necessarily indicate theexpected policy effect and does not help policy assessment in this regard

Group-based concentration indices such as the G5 and G10 avoid the false tion of policy effect Because the calculation of the G5 or G10 indices is based on the

indica-3 In 2015, California received 14.2% of total federal support of academic R&D, and the sum of the funding received by the bottom ten states was about 1.8% A 10% loss of California’s funding is roughly equivalent to about 80% increase of the total federal academic R&D funding won by the bottom ten states.

2.1 Measurement of Concentration of Federal Funding … 23shares of federal funding at group level, any redistribution within each group doesnot affect the value of the indices This effectively addresses the issue of possibleinternal redistribution on the concentration index that may be falsely viewed as theeffect of EPSCoR-type programs.

I would suggest that either G5 or G10 index be used to describe the concentration

of federal resources Like the G1 index, group indices with few states included

in each group (such as the G2 index) can potentially falsify actual policy effect.Moreover, given relative large number of observations, the magnitude of the indices

is relatively small and likely under-represents the extent to which federal funding

is concentrated in a few states On the other hand, the concentration indices with alarge number of states included in each group (such as the G25 index) likely inflatemagnitude and may artificially exaggerate the issue of undue concentration.Although different versions of the concentration index produce quite differentmagnitudes, they show almost identical time trends Taking the shares of total federalsupport of academic R&D as an example, the original G1, G5, and G10 indicespeaked in 1983, declined continuously to the minimum in 2009, and rose after thatyear The shape of the three trends is highly similar, with correlation coefficients at0.96 between the G1 and G10 indices, 0.95 between the G5 and G10 indices, and 0.84between the G1 and G5 indices The three correlation coefficients are statisticallysignificant at the 1% level For the shares of NSF support, the resemblance of thethree trends is even stronger, with correlation coefficients at 0.98 between the G1and G10 indices, 0.99 between the G5 and G10 indices, and 0.97 between the G1and G5 indices The three correlation coefficients are also statistically significant atthe 1% level

Similarly constructed indices have been used in other fields For instance, theHerfindahl-Hirschman index (HHI) in economics research indicates the level ofcompetition among individual firms in one industry by integrating the relative size

of individual firms in relation to the whole industry (Hirschman,1964) The sure is also widely used by federal agencies as an indicator of market concentration(Brown & Warren-Boulton, 1988) The HHI is calculated by squaring the marketshare of each firm competing in the market and then summing all the squared terms

mea-It approaches zero when a market is occupied by a large number of firms of relativelyequal size and reaches its maximum of one when a single firm controls a market.The HHI increases both as the number of firms in the market decreases and as thedisparity in size between those firms increases

2.2 Causes and Consequences of Uneven Distribution

of Federal Funding of Academic R&D

Federal research awards and grants are made primarily through merit-based petition Therefore, the uneven distribution of federal academic R&D support issimply a result of substantial disparity in academic institutions’ research capacityand competitiveness across states The concentration of prestigious research univer-

com-sities and highly capable scientists in such states as California and Massachusettsmakes them much more competitive for federal R&D funding than some other stateswith a handful of research institutions and just a few good university scientists.

In addition to research capacity, the distribution of federal funding of academicR&D is also determined by states’ requests for earmarked projects that get into theappropriation bills through the influence of their representatives in the appropriationscommittees As Savage observed, the House and Senate appropriations committeesplay a decisive role in federal earmarked spending on academic research (Savage,1999) Since the early 1980s, it has been common for members of the appropriationscommittees to bring academic earmarks back to their congressional districts or states(Savage,1999) Academic earmarks totaled less than $17 million in 1980 and rose

to about $17 billion in 2001, representing about 10% of total federal funding ofacademic research (De Figueiredo & Silverman,2006)

There are a handful of empirical studies on the distribution of federal researchfunding to universities Payne (2003) focuses on the effects of congressional appro-priation committee membership Using a panel of 220 universities over the period1973–1999, she presents mixed estimates with regard to the effects of congressionalappropriation committee membership on federal research funding to universities Inanother study, De Figueiredo and Silverman (2006) examine the effects of universitylobbying on academic earmarks Their statistical results show that having a mem-ber on either the House’s Appropriations Committee or the Senate’s AppropriationsCommittee increases academic earmarks for a university by a significant amount.Wu’s 2013 study expands the empirical literature that has provided the mixedempirical evidence on the distribution of federal academic R&D funding Unlike thetwo prior studies, he focuses on individual states rather than universities Using apanel of 50 states over the 28-year period from 1979 to 2006, Wu (2013) finds that astate’s research capacity, as measured by the annual doctorate recipients in science-related and engineering disciplines, is a statistically significant determinant of howmuch federal research funding the universities within its jurisdiction can receive A10% rise in a state’s number of doctoral recipients in science-related disciplines mayincrease its receipt of federal academic R&D funding by about 6% In addition, a10% rise in a state’s number of doctoral recipients in engineering disciplines mayincrease its receipt of federal academic R&D funding by about 2%

Like Payne (2003), Wu (2013) also reports mixed evidence about the effects of astate’s membership in congressional appropriations committees The estimates of astate’s representation in the House’s Appropriations Committee tend to be insignif-icant, while the estimates of a state’s representatives as percent of majority partymembers in the U.S Senate’s Appropriations Committee are statistically significant.The estimated effect is quite small: having one more senator of the majority party onthe appropriations committee could lead to an increase of the state’s annual federalresearch funding by about $92,000 (in 2000 constant dollars), all else being equal

In addition, Wu (2013) finds some spillover effects from a strong federal researchpresence within a state’s boundaries The estimates of the annual federal funds forfederal intramural R&D are positive and statistically significant A 10% increase

in federal funds for federal intramural R&D in a state may lead to about a 0.8%

2.2 Causes and Consequences of Uneven Distribution … 25rise in federal academic research funding to the state’s universities The findingindicates that universities may take advantage of spillover effects from local federalresearch, and this effect could be better utilized if more effective measures are taken

to strengthen the contact and collaboration between researchers in academia andfederal laboratories

Some may argue that scientific research is just a game of talented intellectualsand does not relate to the lives of ordinary citizens, so it does not matter much ifsome institutions and researchers receive more support than others It is not necessary

to view this issue through the lens of inequality However, federal academic R&Dsupport may bring multiple benefits to the recipient states In addition to the inflow offederal funds, funded research projects may produce scientific knowledge that couldlead to new products and start-ups The substantial disparities in federal researchfunding do have equity implications within the broader context of higher educationand regional economic growth

In higher education it is well recognized that the involvement of S&E uate and graduate students in quality research projects is important and perhaps nec-essary to cultivate future scientists and engineers Disparity in funded research leads

undergrad-to the lack of training opportunities for S&E students in low-capacity states, limitingtheir opportunities and exposure to educational resources critical to their profes-sional growth The shortage of educational and training resources and opportunities

is reflected in relatively low shares of doctorate recipients in science and engineeringdisciplines Using NSF survey data of earned doctorates by academic discipline, Icalculate each state’s share of doctorate recipients in engineering and science-relateddisciplines in the year of 2016 and present the results in Figs.2.2and2.3

The pattern is evident: Academic institutions in the EPSCoR states produced fewerdoctorates in science and engineering disciplines compared with their counterparts

in non-EPSCoR states The exceptions are quite scarce For the share of doctoraterecipients in engineering in 2016, Connecticut and Oregon are the only two non-EPSCoR states falling behind the top 25 states, while Tennessee is the only EPSCoRstate that barely makes the top-20 list For the share of doctorate recipients in PhysicalSciences, Life Sciences, Math and Computer Sciences, and Geosciences in 2016,Oregon is the only non-EPSCoR state ranked below the 25th of the list and Missouri

is the only EPSCoR state that barely makes the top-20 list

The distribution of academic R&D funding is also likely tied to regional nomic growth and prosperity The evolution of endogenous growth theory is primar-ily based on the spillover effect of technological knowledge generated from researchand development activities Romer (1990) points out that the distinguishing feature

eco-of technological knowledge as an input eco-of production is that it is neither a tional good nor a public good; it is a non-rival, partially excludable good More thanone firm or industry can use knowledge concurrently, without the use by one entityprohibiting the use by other entities (non-rivalry), and other entities than the entitythat developed the knowledge can often not be excluded from using the knowledge(non-excludability) The spillover of knowledge can be beneficial to the economicoutput of one firm or institution, because it can take advantage of both internal andexternal technological resources to strengthen its capacity of R&D and enhance theperformance of economic activities

conven-Fig 2.2 State’s share of doctorate recipients in engineering in 2016

As one major producer of technological knowledge, academic institutions, cially major research universities, play a key role in national and regional economicdevelopment One important mechanism through which academic institutions con-tribute to regional economic growth is by converting scientific inventions to innova-tion through patenting and licensing of research outputs The Bayh-Dole Act of 1980particularly enables universities to patent publicly funded research and engage withindustries in technology transfer and research commercialization By 1998, everyCarnegie I or II research university had established a Technology Transfer Office

espe-to facilitate patenting and commercialization of university research (Bercovitz &Feldman,2007), and university patenting activity has increased since the passage ofthe Bayh-Dole Act (Henderson, Jaffe & Trajtenberg,1998; Mowery, Nelson, Sampat,

& Ziedonis,2001; Mowery, Sampat & Ziedonis,2002; Mowery & Ziedonis,2002;Shane,2004) The University of California is one of the most profitable because ofits commercialization of patented research (Mowery & Ziedonis,2002)

References 27

Fig 2.3 State’s share of doctorate recipients in Physical Sciences, Life Sciences, Math and

Com-puter Sciences and Geosciences in 2016

References

Bercovitz, J., & Feldman, M (2007) Academic entrepreneurs and technology transfer: Who

par-ticipates and why? In F Malerba & S Brusoni (Eds.), Perspectives on innovation (pp 381–398).

Cambridge, MA: Cambridge University Press.

Brown, D M., & Warren-Boulton, F R (1988) Testing the structure-competition relationship on

cross-sectional firm data Discussion Paper 88-6 Economic Analysis Group, U.S Department

of Justice.

De Figueiredo, J M., & Silverman, B S (2006) Academic earmarks and the returns to lobbying.

The Journal of Law and Economics, 49(2), 597–625.

Henderson, R., Jaffe, A B., & Trajtenberg, M (1998) Universities as a source of commercial

technology: A detailed analysis of university patenting, 1965–1988 The Review of Economics

and Statistics, 80(1), 119–127.

Hirschman, A O (1964) The paternity of an index The American Economic Review, 54(5),

761–762.