Transcript Trends in the Production of Scientific Knowledge

Trends in the Production of Scientific
Knowledge
Paula Stephan
Georgia State University and NBER
[email protected]
Pecs
July 2009
Overview
• Focus will be on production of scientific
research in the university sector
• Draws from updated article “Economics of
Science”
Role of Academic Sector
• Academic sector plays important role in
knowledge production.
• In US, 74% of all articles (fractional counts)
produced in academe; academe and PROs play
similar role in Europe
• Role of academe in patenting is increasing, but
considerably smaller
– 4.5%--grown from 1.5% in U.S.
– Similar order of magnitude in Europe but more
difficult to measure
Domain of Academe
• Academe historically been more focused on
basic research
• Evidence that individuals who value
independence in choosing research agendas
are more likely to work in academe than they
are in industry.
• Individuals working in industry generally place
higher value on monetary rewards.
Examples of Research
•
•
•
•
Bertozzi’s lab (Berkely )has 20 PhD students,
10 postdoctoral fellows & 10 undergrad
students. Senior staff scientist & research
associate also work in lab as do 3
administrative staff & a biosafety facility
director.
High Performance Networking Group at
Stanford, led by Nick Mckeown, includes 12
PhD students, two masters students, an
administrative assistant, three visitors, three
associates, and a research engineer.
Fluid physicist David Quéré, (on the faculty of
the Ecole Superieure de Physique et Chimie
Industrielles of France) & research director at
CNRS leads a CNRS research group composed
of a researcher, seven graduate students and
one postdoc.
Research of hydrologist Elizabeth Screaton
(University of Florida), which “investigates
the interrelationship of fluid flow and
deformation in subduction zones,” combines
field work—on board drilling vessels—with
lab work and numerical modeling.
Other examples
•
•
•
Caltech Observational Cosmology Group is
composed of 17 individuals: One professor
(Andrew Lange), an administrative assistant,
an electronics engineer, 6 postdocs, 5
graduate students, 1 undergrad student and
two visiting associates. Group’s focus is
development of novel instruments to “study
the birth and evolution of the universe.” It
has designed instruments that collect data at
South Pole Viper telescope as well as at
other locations.
Susan Lindquist’s lab at MIT, which studies
protein folding (and which we discussed in
Chapter 3) has 37 members: 20 postdocs, 7
graduate students, 1 visiting scientist, 1 staff
scientist, 3 technicians, 4 administrators and
Lindquist.
Zhong Lin (ZL) Wang’s Nano Research Group
in the College of Engineering at the Georgia
Institute of Technology includes two
postdocs, two visiting scientists, six research
scientists and 14 graduate students.
Commonalities and Differences
• All are doing science and engineering
• All share certain common characteristics
• But environments in which they work, the
importance of equipment in the research that
they do and way in which their work is
structured and supported varies considerably.
Production Function Approach
• No one model of production fits all of science and engineering.
• Mathematicians, chemists, biologists, high energy physicists, engineers,
and oceanographers share certain common characteristics in terms of
production.
– All require time and cognitive inputs.
– In other dimensions there is considerable variability.
– Way in which research is organized is case in point.
• Mathematicians & theoretical physicists rarely work in labs (although they may identify
with a group and work with coauthors) while most chemists, life scientists, engineers and
many experimental physicists do.
• Role of equipment provides another dimension. In some fields, equipment required to
do research is fairly minimal, as in the case of certain areas of math, chemistry and fluid
physics. In others, research is almost entirely organized and defined by equipment, as in
the case of astronomy and high energy experimental physics. Materials also play a role.
In vivo experiments require access to living organisms. For many biomedical researchers
this means having—and taking care of—large numbers of mice, and, in recent years,
zebra fish.
Production Functions
• Long tradition in economics of studying production
processes—or functions. When auto and steel plants
were important components of economies there were
studies of the productivity of the industry and
production processes within the firms.
• But when economists study science rarely think of how
science is produced.
• Instead—like sociologists-- economists focus on people
as unit of observation. Not surprising. People are
faces—and brains—behind science.
• But important to think of science as having multiple
inputs…not just inputs brought by people
How Science Is Produced
• K=f(Cg, R, t, e)
– K is knowledge being produced
– Cg= cognitive resources
– R=other resources, such as equipment, materials, lab
assistants
– t=time of researchers
– e is some error term, encompassing among other things
serendipity and uncertainty.
Scientist(s)
• Effort
– Science takes time; common observation is that
scientists work exceptionally long hours (52.6
hours per week in U.S.)
– Also requires motivation. “Informed observers
have long described high-producing scientists as
driving and indefatigable workers.” (Fox.)
Persistence
• > 50% of physicists chose persistence from list
of 25 adjectives of what it takes to be
successful. No other quality came close.
• Many examples
– Judah Folkman
– Lorenz
Dimensions of Cognitive Resources
• Ability: studies document that as a group scientists
have above average IQs.
• Knowledge base: Important in choosing and solving
problems.
– Education;
– Does scientist keep up?
– Raises possibility of obsolescence and related vintage
effects.
• Public nature of knowledge intensifies races in
discovery.
Embodied or Disembodied
Knowledge?
• Different types of research rely more heavily
on one than the other.
• Nuclear physicist Leo Szilard, who left physics
to work in biology, told the biologist Sydney
Brenner that he could never have a
comfortable bath after he left physics. “When
he was a physicist he could lie in the bath and think for hours,
but in biology he was always having to get up to look up
another fact.”
Too Much Knowledge?
• One can be encumbered by “too” much
knowledge
• One reason young may have an edge
Importance of Tacit Knowledge
• Difference between codified and tacit
knowledge
– Only way to acquire tacit knowledge is to work
with someone with the knowledge
– Lab rotations as a mechanism
– Visiting other labs
– Transgenic mice as an example—need to have
“magic hands”
Collaboration
• Research rarely done in isolation
• Often done in labs—common for individuals
to specialize
• Staffing of labs varies across countries
– U.S. model relies on “temporary workers”—
postdocs & doctoral students;
– European model: permanent staff—employees of
CNRS, Max Planck, etc.
Responsibility for Funding
• U.S. faculty has responsibility for funding
graduate students & most postdocs. Also
faculty member’s time.
– Grad student: $28,000 stipend plus $25,000
tuition.
– Postdoc: $38,000
• Europe: permanent staff generally employees
of state or PRO.
– Graduate students receive stipend from state
Biological and Medical Sciences Postdocs by Source of
Support
35,000
Research Grants
Non-Federal Sources
30,000
25,000
Number
Traineeships
Fellowships
20,000
15,000
10,000
5,000
0
Source: http://www.nsf.gov/statistics/gradpostdoc/
19
Full Time Biological and Medical Sciences Graduate
Students in Doctorate Granting Departments by
Mechanism of Support
90,000
80,000
Research Assistantships
Other (including self)
70,000
Teaching Assistantships
Number
60,000
50,000
40,000
Traineeships
Fellowships
30,000
20,000
10,000
0
Source: http://www.nsf.gov/statistics/gradpostdoc/
20
Labs “belong” to faculty in U.S.
• Most have web pages
• Lab is named for PI
• Sometimes lab members are referred to using
PI’s name as in “Sharpies” for Philip Sharp’s
students at MIT
Lab Structure: Example
• 415 labs affiliated with a nanotech center
• Average lab has 12 technical staff, excluding PI
• 50% are graduate students; 16% are postdocs
and 10% are undergraduates; rest are staff
scientists, etc.
Team Behind Science’s 2008 Breakthrough of the
Year: University of Wisconsin
James Thomson Lab
J. Yu—first author
Back View
Amon Lab: Whitehead Institute
• Amon, HHMI investigator, works on cell
division, focusing on how “cells make sure
their chromosomes separate in the right way.”
Christine White and Group: U. of
Illinois, Chemistry
Interface & Company
ESCPI
Quéré with group
Example from Science
Three-Dimensional Super-Resolution Imaging by Stochastic
Optical Reconstruction Microscopy Bo Huang,1,2 Wenqin
Wang,3 Mark Bates,4 Xiaowei Zhuang1,2,3*
Science, February 8, 2008
1
Howard Hughes Medical Institute, Harvard University, Cambridge, MA 02138, USA.
Department of Chemistry and Chemical Biology, Harvard University, Cambridge, MA 02138,
USA.
3 Department of Physics, Harvard University, Cambridge, MA 02138, USA.
4 School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138,
USA.
2
Members of team
• Xiaowei Zhuang, Prof of Chemistry & Chemical Biology, Prof of
Physics, HHMI, Harvard
• Bo Huang, post-doctoral fellow in Zhuang lab.
• Wenqin Wang, graduate student, Dept of Physics, Harvard; member
Zhuang lab
• Mark Bates, graduate student, Division of Engineering and Applied
Sciences, Harvard; member of Zhuang lab
Birth origin of PI Matters in terms of
lab staffing—at least in U.S.
• Korean-directed labs have 29% more Koreans than
labs directed by U.S.-born PIs
• Chinese-directed labs have 38% more Chinese
students than labs directed by U.S.-born PIs
• Indian-directed labs have 27% more Indians than in
labs directed by U.S.-born PIs
• Turkish-directed labs have 36% more Turkish
students than in labs directed by U.S.-born Pis.
Why?
• Networks and role PI has in staffing lab
• Efficient: language
Collaboration
• Common and growing in science
• Within labs and across labs
• Several ways of seeing trends
Evidence Concerning Teams
Figure 4--Mean Number of Authors per Paper, for Papers
With at Least One Author In the Top 110 U.S. Universities, 1981-1999:
Adams et al 2002
Authors Per Paper
4.40
4.00
3.60
3.20
2.80
2.40
81
84
87
90
Year
93
96
99
Source: Adams et al
Wuchty, Jones & Uzzi
• Analysis of approximately 13 million published
papers in S&E over the 45 year period 1955 to
2000 found team size to increase in virtually
every one of the 172 subfields studied.
• On average team size nearly doubled, going from
1.9 to 3.5 authors per paper (Wuchty, Jones &
Uzzi, 2006).
• Team size even increased in mathematics-- seen
as the domain of individuals working alone and
field least dependent on capital equipment:
Teams Increasingly Have Members
from Another Institution
• Jones, Wutchy and Uzzi
– Study 662 U.S. institutions which have received NSF
funding.
– Find collaboration in S&E across these institutions,
which was rare in 1975, grew in each and every year
between 1975-2005, reaching approximately 40
percent by 2005.
• My own work with Wolfgang Glänzel, Katholieke
Universiteit Leuven, Steunpunt O&O finds similar
results using Thomson Reuters ISI data for 1300
plus four year institutions in the U.S.
Percent of papers with U.S. author at
another U.S. institution by tier
Pub Intra Percent
60.00
55.00
50.00
Tier 1, 2 & TopLib
Research 1
45.00
Rest of Tier 1
Tier 2
40.00
Top Liberal Arts
Rest of Tier 3
35.00
30.00
Work is joint with Wolfgang Glänzel, Katholieke Universiteit Leuven,
Percent of U.S. papers with an
international author
Pub Inter Percent
30.00
25.00
20.00
Tier 1, 2 & TopLib
Research 1
15.00
Rest of Tier 1
Tier 2
10.00
Top Liberal Arts
Rest of Tier 3
5.00
0.00
Work is joint with Wolfgang Glänzel, Katholieke Universiteit Leuven, Steunpunt
Why Increase?
 Importance of interdisciplinary research
Systems biology is case in point
 Researchers are arguably acquiring narrower expertise
and thus have more to benefit through collaboration
 Vast amount of data that has become available—
Human Genome project; PubChem
 Increased complexity of equipment—accelerators and
telescopes are a case in point. CERN’s four colliders
have combined team size of just under 6,000.
 Rapid spread of connectivity decreases cost of
collaboration
Spread of Connectivity: Examples
from U.S.
• Twenty five years ago, only way to work with
someone at another institution was to talk
with them by phone, visit in person, or fax
them material
– Phone calls & travel were expensive. Cheapest
trip to Europe cost around 1800 in today’s dollars.
• Internet, as we know it, did not exist; e-mail
not a possibility.
• This changed with inauguration of BITNET.
BITNET
• Conceptualized by the Vice Chancellor of University
Systems at the City University of New York (CUNY)
• BITNET’s first adopters were CUNY and Yale in May
1981 (Bitnet history).
• At its peak in 1991-1992, BITNET connected about
1,400 organizations (almost 700 academic institutions)
in 49 countries (CREN).
• By the mid-1990s BITNET was eclipsed by Internet as
we know it today and began to fade away.
• We have collected information on date of adoption of
BITNET for 1300 four-year institutions in U.S.
Oct-90
Jul-90
Apr-90
Jan-90
Oct-89
Jul-89
Apr-89
Jan-89
Oct-88
Jul-88
Apr-88
Jan-88
Oct-87
Jul-87
Apr-87
Jan-87
Oct-86
Jul-86
Apr-86
Jan-86
Oct-85
Jul-85
Apr-85
Jan-85
Oct-84
Jul-84
Apr-84
Jan-84
Oct-83
Jul-83
Apr-83
Jan-83
Oct-82
Jul-82
70%
Apr-82
80%
Jan-82
Oct-81
Jul-81
Apr-81
Jan-81
Adoption of BITNET by Tier
100%
90%
Tier 1
Tier 2
Tier 3
60%
50%
40%
30%
20%
10%
0%
BITNET replaced by Internet as we
know it today
• Key requirement for efficient communication
on internet was development of domain name
system—such as gsu.edu.
• We have collected information on date that
almost every 4-year institution in U.S. took a
domain name.
40%
30%
Apr-85
Oct-85
Apr-86
Oct-86
Apr-87
Oct-87
Apr-88
Oct-88
Apr-89
Oct-89
Apr-90
Oct-90
Apr-91
Oct-91
Apr-92
Oct-92
Apr-93
Oct-93
Apr-94
Oct-94
Apr-95
Oct-95
Apr-96
Oct-96
Apr-97
Oct-97
Apr-98
Oct-98
Apr-99
Oct-99
Apr-00
Oct-00
Apr-01
Oct-01
Apr-02
Oct-02
Apr-03
Oct-03
Apr-04
Oct-04
Apr-05
Oct-05
Apr-06
Adoption of DNS by Tier
100%
90%
80%
70%
60%
50%
Tier 1
Tier 2
Tier 3
20%
10%
0%
Concurrently, increased incentives to publish
encourages collaboration
• Occurs at both the system level and at the individual
level
• Budgets of universities and departments in certain
countries depend heavily on publication and citation
counts.
• Funding for research of individual scientists depends
increasingly on publication track record.
• Bonus payments based on publications
Examples
•
•
•
•
•
UK—ranking of departments and allocation of funds based in part on
publications and citations. (Research Assessment Exercise).
Australia—funding of departments based in part on
publications/citations.
Flemish Science Foundation makes research awards based in part on
reputation of faculty as established through publication.
NIH in U.S. (with $29 billion budget) places considerable emphasis on
publication record of grant applicants.
Chinese researchers who place in top half of colleagues in terms of
bibliometric measures can earn three to four times salaries of coworkers. Some institutes pay cash bonus for publishing in Science,
Nature or Cell.
Increased emphasis on networking encourages
collaboration
• Government agencies have bought heavily
into the importance of networks
• “Networks of excellence” funding in EU
• Network funding at NIH through “glue” grants
and P01s.
Which grows faster: Lab size or
collaboration across labs?
• Number of names on an article has increased
by 50%
• Number of addresses has increased by 37%.
• Suggests lab size growing slightly faster than
institutional collaboration
Equipment
• Science heavily influenced by availability of technology
• Exceptions exist but
• Increasingly science requires access to complex equipment
– In genetics: DNA gene sequencer and synthesizer, protein synthesizer
& sequencer comprise the technological foundation for contemporary
molecular biology. Super Computers
– tunneling microscopy—key in nanotechnology
– Accelerators
– Cell lines
– Mice—90% of all mammals used in research are mice—13,000
published
Equipment changes output of lab
• 1990 best-equipped lab could sequence 1000 base pairs a day
• January 2000 the 20 labs mapping human genome were
collectively sequencing 1000 base pairs a second, 24/7
• Measured in base pairs sequenced per person per day, for
researchers operating multiple machines, productivity
increased more than 20,000 fold from early 1990s to 2007,
doubling approximately every 12 months.
• Costs per finished base pair fell from $10.00 in 1990 to
roughly $.01 in 2007
Just beginning…
• New technology for sequencing emerged
recently
• Does work of 100 earlier sequencing machines
• Ads
– “A billion a day, soon a billion an hour. “ (A billion an hour is what it
would take to do the human genome for $1000).
– “More applications lead to more publications”
– “length really matters”
Consequences of changing role of
equipment
• Increasing sophistication of research tools suggests that
capital-labor ratio is changing. (Under researched area)
– Broad Institute and Vetner Institute fired staff working in sequencing
late 2008/2009.
• Cost considerations (discussion to follow)
• Also substantially changed nature of dissertation work.
– Example: in chemistry, nuclear magnetic resonance combined with xray crystallography and advanced computing power allows protein
structures to be elucidated more rapidly. Result: a PhD thesis used to
be focused on defining structure of a single protein domain; now a
thesis in a similar field might examine and compare dozens of
structures.
Re-emphasizes importance of the nonlinear model
• Importance of equipment is one reason to
stress non-linearity of scientific discovery
• Not just that science affects technology
• Technology very much affects science:
– The history of science is the history of how
important resources and equipment are to
discovery. Theme in research of Nathan
Rosenberg; Joel Mokyr.
Protein Structure Initiative
• Funded by NIH. To date, $765,447,000
• Aim: (1) to increase number of sequence families;
(2) continue technology development, (3) facilitate
use of structure by broad scientific community
• Assessment report published in December 2007
Concludes
•
“The PSI centers have matured many new technologies, and the activity
around the PSIs has led to impressive advances that have a broad impact
and are much appreciated by the structural community.”
• Technologies developed are “increasingly used by the broader research
community.”
• Specific advances include “construction of the pipeline that enables the
entire process of going from sequence to solved crystal structure to be
almost fully automated and capable of working at high throughput for
amenable proteins.”
• Cost per structure is nearly $100,000.
Leroy Hood as an example
• Author of more than 500 papers
• Winner of 1987 Lasker Award for Basic Medical Research
• Winner of 2002 Kyoto Prize for Advanced Technology in
recognition of his inventions, including the automated DNA
sequencer and an automated tool for synthesizing DNA
• Winner of 2003 Lemelson-MIT Prize for inventing “four
instruments that have unlocked much of the mystery of
human biology…”
Mentor played role
• Hood’s mentor at Cal Tech, William Dreyer,
reportedly told Hood when he was a student,
“If you want to practice biology, do it on the
leading edge and if you want to be on the
leading edge, invent new tools for deciphering
biological information.”
Zhuang article as example
Abstract reads:
“Recent advances in far-field fluorescence microscopy have led to substantial
improvements in image resolution, achieving a near-molecular resolution of 20 to
30 nanometers in the two lateral dimensions. Three-dimensional (3D) nanoscaleresolution imagining, however, remains a challenge. We demonstrate 3D
stochastic optical reconstruction microscopy (STORM) by using optical
astigmatism to determine both axial and lateral positions of individual
fluorophores with nanometer accuracy.”
Cost
• Equipment is usually expensive
• Extreme: New LHC accelerator--$8 billion
• Examples: Sequencer such as Applied Biosystems’ 3730 model
is $300,000; tunneling microscope $1 million plus;
• New generation sequencers--$5 billion world market
• Mice are expensive:
– Off shelf mouse is $50;
– Transgenic mice can be much more--$2,000 and
some carry tag of $15,000.
Exceptions
• Some experiments inexpensive
• Quéré lab as example
– IKEA tape measures
– Plastic dishes from retail store—costs 30 times
more from supplier
– Paper clips
– Toy guns
– Sling shot
– But cameras were expensive…
Mouse upkeep
• Mice are expensive to keep
– Mouse upkeep (per diem) is high: $.05 to $.10 per day
– Irving Weissman reports he was spending $800,000 to $1
million a year at Stanford to keep his mice.
– Immune deficient mice cost more to keep
– Mouse packages play a role in recruitment. Researcher
recruited from one institution to another when offered a
mouse package that translated into $.036 per mouse per
day.
Role of materials and equipment means exchange plays
increasingly important role
• 75% of academics in the Walsh, Cho and Cohen
sample made at least one request for materials in a
two-year period (7 to academics; 2 to industry).
• Note: not everyone agreed to share. 19% of
material requests made by sample were denied.
Competition among researchers played a major role
in refusal, as did cost of providing the material.
• Patents “signal to other scientists that you [are ] a
valuable exchange partner…” (Murray)
Policies to Encourage Exchange
• Deposit banks for materials
• Government agencies taking role in
encouraging availability
– MOU from NIH for oncomouse as example
– Requirement that if government funded, data be
made available
Who Benefits?
• Individuals at non-elite institutions
• Similar findings found with regard to who
benefited from availability of IT
Factor Change
Results for Productivity
Effect of IT on Research Productivity
BITNET-Rank
DNS-Experience
BITNET-Experience
1.2
1.17
1.16
1.16
1.15
1.1
1.06
1.05
1.0
DNS-Rank
0.95
1.04
1.01
1.02 1.02
1.02
0.95
0.93
0.9
0.83
0.82
0.8
0.7
Effect of IT on Research Quality
BITNET-Experience
1.2
DNS-Experience
BITNET-Rank
DNS-Rank
Factor Change
0.95
0.95
0.93
0.9
0.83
0.82
0.8
Results for Research Quality
0.7
Effect of IT on Research Quality
BITNET-Experience
DNS-Experience
BITNET-Rank
DNS-Rank
1.2
1.14
1.1
1.11
1.11
1.11
1.04
1.04
1.0
0.97
0.96
0.97
0.99
0.9
0.96
0.94
0.95
0.90
0.88
0.83
0.8
Effect of IT on Research Collaboration
1.4
BITNET-Experience
1.3
1.2
1.23
DNS-Experience
BITNET-Rank
DNS-Rank
Factor Change
0.97
0.96
0.97
0.99
0.9
0.96
0.94
0.95
0.90
0.88
0.83
Results for Collaboration
0.8
Effect of IT on Research Collaboration
BITNET-Experience
1.4
DNS-Experience
BITNET-Rank
DNS-Rank
1.3
1.2
1.23
1.14 1.15
1.12
1.1
1.10
1.13
1.07
1.04
1.07
1.04
1.0
0.99
0.95
0.94
0.9
0.87
0.83
0.8
0.75
0.7
1-4
5-8
9-14 15-20 21-26
1-4
5-8
Experience(Years since Ph.D.)
9-14 15-20 21-26
Top25 26-50 Outside50
Top25 26-50 Outside50
Employer Ranking
Space Matters
•
•
•
•
•
Labs take space at universities
Expensive space!
How is physical size of lab determined?
Is lab space ever taken away?
Under researched questions
Serendipity
• Like the prince of Serendip (a legendary ruler of
Ceylon, known for knack of chance discovery),
researchers often find different, sometimes greater,
riches than the ones they are seeking.
• Research often provides answers to questions not
yet posed…
– Examples: the tetrafluoresthylene cyclinder that gave rise to Teflon
was meant to be used in the preparation of new refrigerants.
– Anti-AIDS druz AZT was designed as a remedy for cancer.” Eliel 1992
– Patel and her colleagues, set out to study glucocorticoid effects on 15hydroxyprostaglandin dehydrogenase (15-OH-PGDH) and inadvertently
produced evidence for the first mineralocorticoid receptor (MR)
responsive gene.
http://jcem.endojournals.org/cgi/reprint/84/2/393.pdf
Not all luck
• “Chance favors only the prepared mind”—Pasteur
• Researchers admit to role of chance:
• He (Richardson) obtained his PhD degree from Duke in 1966. His
thesis advisor was Professor Horst Meyer. In the Fall of 1966 he
began work at Cornell University in the laboratory of David Lee.
Their Research goal was to observe the nuclear magnetic phase
transition in solid 3He that could be predicted from Richardson’s
thesis work with Horst Meyer at Duke. In collaboration with
Douglas Osheroff, a student who joined the group in 1967, they
worked on cooling techniques and NMR instrumentation for
studying low temperature helium liquids and solids. In the fall of
1971, they made the accidental discovery that liquid 3He undergoes
a pairing transition similar to that of superconductors. The three
were awarded the Nobel Prize for that work in 1996.”
Access to resources necessary condition for
doing research
• In U.S. access comes first through a start-up package provided
by the dean
• Thereafter equipment and funds to hire students and
postdocs become responsibility of scientist; must apply by
writing research proposals to funding agencies.
• No grant no lab
• Increasingly, no grant no job
• Emphasis on individual scientist to generate resources has not
been as strong in other countries, where researchers are hired
into government funded and government run laboratories
such as CNRS in France.
Grants take time to write and
administer
• 2006 survey of U.S. university researchers
• Report spending 42% of “research time” on administrative
tasks
– Split between pre-grant (22%) and postgrant (20%).
• Most time consuming
– Filling out grant progress reports
– Hiring personnel
– Managing laboratory finances
• Recent changes have increased time requirement
– Health privacy laws
– Institutional review boards
– Accounting for “select agents” after 9/11
Decker: Northwestern University
Changes in funding
• Mix of funding in U.S. is for life sciences
• Success rates on grants has been declining in
U.S. in recent years
• Proportion going to young researchers also
declining
• Europe has begun to shift from the institute
approach to the U.S. PI-driven approach
Why Biomedical Sciences—NIH—
So Favored?
• Age distribution of U.S. Senate (100 members)
– Median age is 62
– Oldest is 91; youngest is 42
– 41 over 65
– Only 2 under 45
NIH Funding Billions of dollars
$30
$25
$20
$15
$10
$5
$0
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
Constant Dollars
Current Dollars
OER: NIH Budget over time
Number of NIH Competing R01 Equivalent*
Applications, Awards and Percent Funded
(Success Rate)
35%
25%
20
20%
15
15%
Percent Funded
30%
25
(in Thousands)
10
10%
5
5%
20
05
20
04
20
03
20
02
20
01
20
00
19
99
19
98
19
97
0%
19
96
-
19
95
Number of Applications
30
Fiscal Year
Review ed
Aw arded
Success Rate
NIH, OER: “Investment…”
R01 Equivalent* Includes R01, R23, R29 and R37
Number of New and Established Investigators Receiving
Competing and R01 and R01 Equivalent Grants to 1962 to 2004
40%
Established Investigators
New Investigators
Percent New Investigators
35%
6,000
30%
5,000
25%
4,000
20%
3,000
15%
2,000
10%
1,000
5%
0
0%
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
20
02
20
04
Number of Grants
7,000
Fiscal Year
NIH, OER for AIRI
Percent Grants to New Investigators
8,000
DISTRIBUTION OF INVESTIGATOR AGES
NIH Competing R01 Equivalent Aw ardees
14.5%
Percent of Total
14.6%
20.8%
24.9%
15.8%
15.2%
20.3%
23.9%
18.4%
19.0%
6.2%
1995
17.0%
14.6%
20.3%
24.6%
17.1%
17.9%
15.7%
20.5%
23.1%
16.5%
19.6%
16.4%
21.4%
22.4%
15.6%
21.7%
22.7%
23.9%
24.9%
27.5%
27.6%
16.6%
16.3%
16.0%
17.4%
18.1%
18.2%
21.1%
21.1%
22.2%
20.0%
20.1%
19.6%
22.1%
21.4%
21.7%
21.7%
21.7%
20.9%
14.1%
13.8%
13.0%
13.1%
12.9%
12.0%
3.4%
6.4%
6.4%
6.2%
4.8%
4.0%
3.8%
4.5%
3.5%
3.8%
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Fiscal Year
35 an d Yo un ger
36 - 40
41 - 45
46 - 50
51 - 55
Over 55
NIH, OER for AIRI
Other Countries Experience
Stop-and-go Funding
• Italy and France had major hiring waves in the
1980s (France 1985; Italy 1980)
• Lissoni, Mairesse, Montobbio, and Pezzoni &
find that individuals hired into a “wave” are
less productive throughout their career.
Demand for scientists
• Demand depends on
– K/L ratio for efficient production— depends on
technology & relative costs of inputs
– But will also depend on “scale” of scientific operation.
Substitution of equipment for labor is occurring but as
scale of science continues to grow because costs of
discovery are decreasing, could result in employing more
scientists in labs.
– Also depends on relative prices but
– Science may not be “efficient.” If there are funds for
students and not equipment may not be on efficient
production path
Who will staff labs in U.S.?
• Staffing labs with doctoral students and postdocs provides a
ready flow of “new” ideas and “temporary” workers.
• Produces more than “absorptive” capacity of university;
• Movement of scientists from academe to industry is a major
way in which knowledge is transferred from the public to
private sector.
• But if industry and academe cannot readily absorb the
production of new PhS there can be a problem of over supply.
• Can this staffing model persist?
Closing thoughts
• Need to study labs: economists almost always approach
productivity issues by studying individuals scientists rather
than the labs in which scientists work
• Need to think of production function for science
• Once shift to study of labs, numerous questions invite
exploration:
– Need to learn more about production function of the lab, degree of
substitution between capital and labor; whether capital-labor ratio has
changed over time and scale effects
– How this affects labor market for scientists and engineers
– How lab size is determined and to what extent economic factors come
into play?
– How outcomes relate to funding
• Networking
• Size of labs
Some under-researched
questions/issues
• How is lab size determined in terms of numbers of
people?
• Efficiency of multiple grants
• Will research opportunities change as amount of
data increases: Jeremy Berg hypothesis?
• How is capital labor ratio changing in labs?
• How will this affect labor market for scientists and
engineers?
• Cross country comparisons of production function—
input prices vary across countries.
Final Thought…
• “Know more about how to organize a car
factory than how to organize science”
• And recently it doesn’t appear that we know
all that much about cars!
Comments/Questions
• [email protected]