POLICY INSIGHTS FROM ANALYSING EUROPEAN UNIVERSITIES
Andrea Bonaccorsi University of Pisa OECD Workshop Assessing the Impacts of Public Research Systems 2-3 May 2016
Outline of policy issues 1. Can we do something to improve the productivity and quality of research of universities in our country? 2. Should we concentrate research funding in a small number of large universities? Appendix Can we measure the impact of university research at regional level on • • •
Entrepreneurship Productivity Growth of companies Pagina 2
Research strategy • Construction of a census (= validation by National Statistical Authorities, NSAs) • Data integration from heterogeneous sources after substantial disambiguation work Data sources • ETER (European Tertiary Education Register) All European Higher Education Institutions (HEIs)- data on students and staff + institutional data. 2293 institutions. Data 2011-2012 available. Data 2013 and 2014 in progress.
• GRBS (Global Research Benchmarking System) Data on Scopus publications 2007-2010 and 2008-2011 disaggregated by 251 Subject categories for North America, Asia and Europe
• Eurostat- regional covariates Pagina 3
Distribution of publications per capita among European universities. Medicine
Publications per capita
35.00
30.00
25.00
20.00
15.00
10.00
There are large differences in productivity among universities
5.00
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 129 133 137 141 145 149 153 157 161 165 169 173
0.00 Universities
120.00
100.00
80.00
60.00
Citations per capita
140.00
Distribution of citations per capita among European universities. Medicine
There are large differences in research quality among universities
40.00
20.00
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101 105 109 113 117 121 125 129 133 137 141 145 149 153 157 161 165 169 173
0.00
Publications per capita
Distribution of publications per capita among European universities. Computer science 9.00 8.00 7.00 6.00
5.00 4.00
These differences are found across all scientific disciplines
3.00 2.00 1.00 0.00 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91
Universities
Distribution of citations per capita among European universities. Computer science
Citations per capita
35.00
30.00
Differences in research quality (citations per capita) tend to be larger than differences in productivity (publications per capita)
25.00
20.00
15.00
10.00
5.00
0.00 1
4
7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91
Universities
Can we do something to improve the scientific productivity of universities in our country? • ETER + GRBS data • Multilevel approach • Dependent variable = 4 indicators of scientific productivity and quality (% publications in, or citation from, top 10% or 25% SNIP journals) • Independent variables – University level – – – –
Age Size PhD intensity Internationalization
- Generalist vs specialist - Public vs private - Hospital - Student load
– Regional level – GDP per capita – GERD per capita
- No beds per 100,000 inhabitants - No. medical doctors per 100,000 inhabitants Pagina 8
• Main results in the Medicine sector - Size of university affects negatively - Overall research excellence + internationalization of PhD strong positive effect - Age of university no effect - Generalist universities better than specialist - Public vs private no effect - PhD intensity no effect - Overall student load at university level (surprising) positive effect - Strong context effects (GERD at regional level)
Pagina 9
Policy insights • The generalist model («Humboldtian model») is a dominant model, performing relatively well • There is complementarity between research and educationuniversities with a high student load are not necessarily performing badly in research • Age and governance (private/ public) do not have any systematic effect • Larger universities do not perform better
Pagina 10
Policy insights/2 Scientific productivity is influenced by the governance at university level – Importance of university autonomy – Policies should create incentives for competitive recruitment – Universities should adopt consistent quality criteria for recruitment, assessment and promotion Large difference between US and Continental Europe with respect to the scientific excellence model
Pagina 11
Agri
Bio
Chem
Comp
Earth
Econ
Eng
Env
Health
Mater
Math
Medic
Life
Harvard Michigan-Ann Arbor MIT UC Berkeley Toronto * ETH Zurich Nanyang Tech U Nat U Singapore Stanford Oxford UC Los Angeles Hong Kong U Georgia Tech Cambridge UC San Diego Maryland Minnesota Washington-Seattle Wisconsin-Madison Yale Hong Kong Polyt U Johns Hopkins McGill * Nat.l Cheng Kung U Nat.l Chiao Tung New York Northwestern
Pagina 12
Phys
Eindhoven Goteborg Hong Kong U S&T Jilin KLU King’s College Korea Adv Inst S&T Kyoto Kyushu Leiden Nagoya Oregon Peking PennState- U Park Pohang Radboud Njimegen Seoul National U Southeast Sungkyunkwan Swedish U Agr Scien TU Denmark Edinburgh Manchester Southampton Paris V Pierre Marie Curie UCL
Pagina 13
Should we concentrate research funding in a small number of large universities? Underlying rationale: economies of scale Clearly separate empirical issues: - Economies of scale in administrative activities (YES) -
Efficiency in the utilization of common infrastructures and services Administrative activities
- Economies of scale in higher education (YES, BUT..) -
Teaching in a large classroom requires the same effort than teaching in a small one But increasing the scale of interaction-intensive activities (tutoring, supervision) reduces quality
- Economies of scale in research (NO) -
Threshold for research teams at small level No systematic evidence of increasing returns at department or university level Pagina 14
Publications per capita by size of academic staff in European universities. Medicine
Publications per capita
35.00
30.00
25.00
20.00
Largest medical schools are among the least productive
15.00
10.00
5.00
0.00 0
500
1000
1500
2000
2500
3000
3500
Total academic staff F09 (Head Count)
4000
4500
Citations per capita by size of academic staff in European universities. Medicine
Citations per capita
140.00
120.00
Largely cited research articles are found more likely among small-tomedium-sized medical faculties
100.00
80.00
60.00
40.00
20.00
0.00 0
500
1000
1500
2000
2500
3000
3500
4000
Total academic staff F09 (Head Count)
4500
Publications per capita by size of academic staff in European universities. Computer science
Publications per capita
9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00 0
100
200
300
400
500
Total academic staff F06 (Head Count)
600
Citations per capita by size of academic staff in European universities. Computer science 35.00
30.00
Citations per capita
25.00
20.00
15.00
10.00
5.00
0.00
0
100
200
300
400
500
Total academic staff F06 (Head Count)
600
Average citations 500
6 Citations per paper by number of publications in World Asia Europe North America Oceania universities. Chemistry 5
4
1000
1500
2000
2500
3000
3500
4000
3
2
1
Largest volumes of publications are produced by universities above but mainly below the median level of citations per paper Number of publications Pagina 19
Percentage of publications in top 10% journals by volume of publications of World universities. Chemistry %Pubs in Top 10% SNIP
60 Asia
Europe
North America
Oceania
50
40
500
1000
1500
2000
2500
3000
3500
4000
30
20
10
0
Largest volumes of publications are produced by universities which publish in top journals less than the average Number of publications Pagina 20
Policy insights
• No evidence of increasing returns to scale in research activities • Policies of restructuring aimed at administrative rationalization- legitimate policies but no argument from gaining efficiency in research
• Research funding should be channeled directly to research teams according to their scientific productivity/ quality • Better policy is to push researchers to compete with world level frontier in their own field Pagina 21
Can we measure the impact of university research at regional level? Dependent variable: rate of creation of new firms at province level Italy data at province level
Independent variables (breakdown by Field of Science/ Field of Education) -
Academic staff PhD students
- Publications - Patents
Pagina 22
Impact of public research on entrepreneurship Main results
- Impact on entrepreneurship depends on the subject matter - Science and Social sciences and humanities (SSH) lowest impact - Engineering largest impact
- Impact greater in laggard regions - Impact follows different paths - Knowledge embedded in publications less important - Knowledge embedded in people (Academic staff; PhD students) more important - Geographic decay different for different channels (publications decay at 50km; academic staff decay at 100km) Pagina 23
Impact of public research on firm growth Dataset: number of new firms created in all European countries in 2010 (n> 500,000). Source: ORBIS Dependent variables -
Firm growth (economic and financial data) Productivity (value added)
Independent variables (breakdown by Field of Science/ Field of Education) at NUTS 2 and 3 level -
Academic staff PhD students Excellence indicators
- Publications - Citations - Financial endowment at regional level
Pagina 24
Policy insights • Multiple channels of impact of universities on regional economy • Importance of teaching • Spillovers are discipline- and industry-specific
• Avoid «monistic» policies (e.g. exclusive focus on technology transfer based on patents)- beyond commercialization of research only
Pagina 25
Policy insights/2 • Largest effect when there is complementarity between public research (= publications) and private R&D (= patents) in the same region- policies aimed at complementarities more effective • Density effects are important
• Co-specialization between research fields and industry specialization often missing • In laggard regions we do not see «excellent universities» but mainly pockets of excellence Pagina 26
Policy insights/3 • Further steps • Quality of research vs volume (= does high quality research produce more spillovers on entrepreneurship, productivity, and company growth?) • Social impact of public research can be measured • Value surveys • Social capital • Civic/ political participation • STI innovation model vs DUI (doing, using, interacting)
Pagina 27
Conclusions Integration of heterogeneous microdata with data from officially validated censuses is a promising strategy • comparative cross-country analysis • benchmarking • econometric exercises Many important (and controversial) policy issues can be addressed with an evidence-based approach The measurement of the impact of public research on economy and society is close to become a reality
Pagina 28
Productivity of university research and returns to scale Bonaccorsi A., Secondi L. (2016a) The determinants of research performance in European universities. A large scale multilevel analysis, Submitted for publication Bonaccorsi A., Secondi L. (2016b) Field of science differences in research performance. In preparation Bonaccorsi A., Secondi L. (2016c) Are there economies of scale in research? In preparation
Models of academic excellence Bonaccorsi A., Haddawy P., Cicero T., Saeed H. (2106) Explaining the transatlantic gap in scientific excellence, Submitted for publication Bonaccorsi A., Haddawy P., Cicero T., Saeed H. (2106) The solitude of stars. Academic excellence in European universities. In preparation Pagina 29
Impact of university research Bonaccorsi A., Colombo M.G., Guerini M., Rossi Lamastra C. (2014) The impact of local and external university knowledge on the creation of knowledge-intensive firms: evidence from the Italian case. Small Business Economics, DOI 10.1007/s11187-013-9536-2 Bonaccorsi A., Colombo M.G., Guerini M., Rossi Lamastra C. (2013) University specialization and new firm creation across industries. Small Business Economics DOI 10.1007/s11187-013-9509-5.
Bonaccorsi A., Colombo M.G., Guerini M., Rossi Lamastra C. (2016a) Estimating the impact of public research on entrepreneurship and firm growth. In preparation Bonaccorsi A., Colombo M.G., Guerini M., Rossi Lamastra C. (2016b) Quality of research and firm growth. In preparation
Pagina 30
Pockets of excellence Bonaccorsi A. (2016) Addressing the disenchantment. Universities and regional development. Journal of Economic Policy Reform, forthcoming.
Pagina 31