Why is Europe so efficient at producing scientific papers, and does this explain the European Paradox? Patricia Foland and R. D. Shelton WTEC 11th International Conference on S&T Indicators, Leiden, Sept. 10, 2010
Apr 02, 2015
Why is Europe so efficient at producing scientific papers, and does this explain the European Paradox?
Patricia Foland and R. D. SheltonWTEC
11th International Conference on S&T Indicators, Leiden, Sept. 10, 2010
NSF Science and Engineering Indicators--fractional counts.
Q: Why did the EU become world leader in mid-90s?
SCI and SSCI Papers
0
50,000
100,000
150,000
200,000
250,000
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
US
EU15
PRC
Relative Efficiency ki
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
US
EU15
PRC
Relative efficiency is this ratio normalized by OECDg values.
EU and US had the same ki in 1990, but curves diverged in 1990s. After 1998, curves are flat with EU 60% more efficient. Finding why 1990s changed will suggest why the EU is more efficient today.
A: EU passed the US because it sharply increased its ratio of papers/R&D €. Q: But, why?
Model of a national scientific enterprise
Indicators measure inputs like R&D investment (GERD) and outputs like patents.
Multiple linear regression can identify which inputs are most important
Resources InS&T Outputs
Statistical techniques for accounting for national publication changes
• Multiple regression shows that publication outputs are more highly correlated with R&D investment (GERD) than other inputs
• The Shelton Model predicts publication share mi from overall GERD share wi : mi = ki*wi
• ki is the “relative efficiency,” it is also papers/€, normalized by values for whole set
• Model works well after 2000, when ki is fairly constant, but not for 1990s EU advance
• Need to search for better model, by analyzing effect of input components on efficiency
Analysis of components of efficiency
• Numerator (papers) vs denominator (GERD)?
• Artifact of SCI: new journals favored EU?• GERD sources: Govt, Industry, Foreign,
Other?• GERD spending: HERD, BERD, Goverd,
Other?• Govt appropriations: military, civilian? • Labor vs capital: HR, GERD?• Countries of the EU?
Only a few of these analyses will be shown here.
EU shot ahead in efficiency ratio because it slowed rises in R&D $, but still sped rises in papers. Q: how?
GERD = gross expenditure on R&D. US and EU reduced rises in GERD; the EU more so. Paper change is more dramatic.
Annual Increases in GERD
0
1
2
3
4
5
6
EU15 US
Perc
ent 81-89
91-99
Numerator Denominator
Was this because the SCI added journals favoring the EU in the 90s?
In 1992 and 1994 the SCI added many journals that seem to favor the EU. (From NSI CD)
Increase Over Previous Year
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Per
cen
t
SCI
EU-US Paper Shares
World Paper Shares Fixed Database vs. Variable
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Per
cen
t
US Fixed
EU15 Fixed
US Variable
EU15 Variable
Until 2004 NSF used a fixed set of journals. With SEI2004 it used full SCI. Shelton and Foland (2008) confirm this finding with a different approach.
No, change is real and not an artifact of the SCI
A: The EU increased paper share in the 90s by sharpening focus on sectors that maximize papers
Government funding instead of industryUniversity instead of business R&D spendingCivilian instead of military R&D
Multiple regression shows these input components are more effective in producing paper outputs—first two much more so.
BUT, this allocation minimizes outputs like patents with more immediate economic benefits—thus the European Paradox
Regression analysis of which GERD components best account for paper outputs
Year = 1999, Constant $ PPP series usedDependent Variable (DV) = papers in SEI, fractional counts Independent Variables = two components of R&D funding (IV1, IV2)—several typesN = 39 countries in OECD Group, sometimes fewerP is significance probability of IV; if p < 0.05, variable is importantR2 > 96% always – IVs are very good predictors
Government vs. industrial funding of R&D
IV1 = government funded part of GERDIV2 = industry likewiseMuch smaller components omittedIV1: P = 0.000 (very significant)IV2: P = 0.838 (not significant)Regression equation for papers:DV = 2.73 IV1 – 0.031 IV2 + 1957
Part of GERD From Public vs. Private Funding
0
10
20
30
40
50
60
70
80
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
Per
cen
t
US Public
EU15 Public
US Private
EU15 Private
In 1990s, both shifted R&D funding from government to industry, but this change was much smaller in the EU
Government funding is much more likely to produce papers. Paper advantage: EU.
SCI and SSCI Papers
0
50,000
100,000
150,000
200,000
250,000
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
US
EU15
GERD from Govt Constant $ PPP
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
$ M
illi
on
s
US
EU
Patterns are almost identical with a small lag. Is this the smoking gun? (EU curve is shifted slightly above US because of another factor--HERD.)
Input
Output
University vs. business expenditure of R&D
IV1 = HERD, higher education part of GERD spendingIV2 = BERD, business part likewiseMuch smaller components omittedIV1: P = 0.009 (very significant)IV2: P = 0.000 (very significant, but coefficient is much smaller)Regression equation for papers:
DV = 2.53 IV1 +0.58 IV2 + 2136
A model that can account for EU passing US
DV here is paper shareIV1 = Government funding, shareIV2 = HERD, higher education spending shareMuch smaller components omittedBoth IV: P = 0.000 (very significant)Regression equation for paper share:
DV = 0.582.53 IV1 +0.271 IV2 + 0.249
HERD Constant$ PPP
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
50000
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
$ M
illi
on
s
US
EU15
Higher education R&D spending
Despite smaller overall GERD, the EU spends more on university R&D. Paper advantage: EU.
HERD Part of GERD
0
5
10
15
20
25
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
Per
cen
t
US
EU15
This shows EU sharpened focus on HERD in 1990s. “Over 1990s academic researchers contributed almost ¾ of the total [paper] output” SEI 1998. Advantage: EU.
Higher education part of R&D expenditure
BERD Constant$ PPP
0
50000
100000
150000
200000
250000
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
$Mil
lio
ns
US
EU15
Business R&D spending
This sector produces fewer papers, and US sharply increased this focus in the 1990s, while EU investment was fairly flat. Advantage: EU.
Defense vs. civilian government appropriations
IV1 = civilian part of GBAORD fundingIV2 = defense part likewiseN = 29 available IV1: P = 0.000 (very significant)IV2: P = 0.000 (very significant, coefficient is somewhat smaller)Regression equation for papers:
DV = 2.65 IV1 +2.19 IV2 + 2531
Government Funding of Military R&D
0
10
20
30
40
50
60
70
80
Pe
rce
nt
of
GB
AO
RD
US
EU15
More than 50% of US government R&D is for military; EU has about 10%
Both cut military R&D share after Cold War, but EU cut far more than the US. Paper advantage: EU.
Civilian R&D from Government
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
$ m
illl
ion
s C
urr
ent
PP
P
US
EU15
Despite smaller overall R&D funding, EU governments spend more on the civilian sector
Paper advantage: EU
Conclusions
At the end of the Cold War, the EU spent more of the peace dividend on R&D that produced papers, and the US spent more of it on R&D that did not.This sharply increased EU efficiency, causing it to become the world leader in papers.But, papers are only one output of the R&D enterprise. The European Paradox is the perception that Europe does not reap the full economic benefits of its leadership in papers.This analysis suggests that EU focus on investments that produce papers probably lowers outputs with more immediate economic benefits—patents, for example.Regression also shows that those outputs come more from the private, business, and perhaps military investments—the opposite of research papers.
Appendix Charts
Government Funding of Military R&D
0
10
20
30
40
50
60
70
80
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
Per
cen
t o
f G
BA
OR
D FR
DE
IT
UK
US
EU15
More than 50% of US government R&D is for military; EU has about 10%
All reduced military R&D share after Cold War, but EU cut far more than the US. Paper advantage: EU.
Average Annual Growth in Papers
-1
0
1
2
3
4
5
6
DE UK FR IT EU15 US
Per
cen
t
Papers 81-89
Papers 91-99
EU paper gains driven by large countries
Large EU countries greatly increased output. US decline is related. This is almost a zero sum game.