Presentation is loading. Please wait.

Presentation is loading. Please wait.

How Economics Shapes Science

Similar presentations


Presentation on theme: "How Economics Shapes Science"— Presentation transcript:

1 How Economics Shapes Science
Paula Stephan Georgia State University and NBER Melbourne, Australia August 16, 2017

2 Paula Stephan Georgia State University & NBER
Overview Economics is about incentives and costs Talk about how incentives and costs shape practice of science at research institutions, especially universities Do so by providing Examples of how incentives and costs affect practice of science Perspective on some of the unintended consequences of the system that have evolved Paula Stephan Georgia State University & NBER

3 Examples of How Incentives and Costs Affect Practices in Science
Paula Stephan Georgia State University & NBER

4 Incentives for Publishing in Top Journals
Paula Stephan Georgia State University & NBER

5 Where Research Is Submitted
Number of countries have adopted policies to encourage researchers to publish in top journals Divide countries into ones that have adopted policies where Institution is rewarded Individual is rewarded with cash Individual is rewarded with career advancement No policy change Paula Stephan Georgia State University & NBER

6 Paula Stephan Georgia State University & NBER
Response to Incentives to Publish in Top Journals: Submissions by country policies to Paula Stephan Georgia State University & NBER Source: Franzoni, Scellato, Stephan, Science (2011)

7 Acceptances and Success Rates
Also find that certain incentives are positively correlated with number of articles published. Incentives that matter: related to career advancement; neither institution-based incentives nor cash incentives to individuals are significantly associated with publications. Results also suggest that acceptance rates are negatively correlated to cash bonuses. Acceptance rates are not significantly related to other kinds of incentives. Paula Stephan Georgia State University & NBER

8 Example of Chinese Cash Awards: 100 Universities
Average Amount of Cash Awards Increased 43% Wei Quan at Wuhan University, Bikun Chen at Nanjing University of Science and Technology, and Fei Shu at McGill University in Montreal. Paula Stephan Georgia State University &

9 Example of Chinese Cash Awards: 100 Universities
Average Amount of Cash Awards Increased 43% Wei Quan at Wuhan University, Bikun Chen at Nanjing University of Science and Technology, and Fei Shu at McGill University in Montreal. Paula Stephan Georgia State University &

10 Paula Stephan Georgia State University & NBER
Put in Perspective Average annual salary of university professors is around 8,600 USD; average basic salary of new hired professors is about 3,100 USD (2012 data) Reward value for paper in PNAS equals more than a single year’s salary for newly hired professor; cash award for Nature or Science is up to 20 times university professors’ average annual salary. Some extreme rewards: $168,000 found in survey Paula Stephan Georgia State University & NBER

11 Faculty Salaries: Rising Inequality
Paula Stephan Georgia State University & NBER

12 Paula Stephan Georgia State University & NBER
Salary Faculty salaries in US depend upon institution, field, productivity and professorial rank Faculty know there are wide differences and behave accordingly, often moving in response to salary offer or getting institution to match offer Given concentration of resources at elite, private institutions, leads to greater inequality in earnings See in terms of rising Gini coefficients Paula Stephan Georgia State University & NBER

13 Gini Coefficient by Field, Rank and Year
Paula Stephan Georgia State University & NBER

14 Paula Stephan Georgia State University & NBER
Conclude With but few exceptions, Gini coefficient, while under .25, has more than doubled in 33-year period in academe Paula Stephan Georgia State University & NBER

15 Paula Stephan Georgia State University & NBER
Compared to US By comparison, for full time male earners during same period in US, Gini grew by 35%, going from around .314 to .424. Conclude: income inequality is greater in the larger society but it has been growing faster in academe. Paula Stephan Georgia State University & NBER

16 Paula Stephan Georgia State University & NBER
US Not Alone Data are bit thin but every indication that income inequality is growing among faculty in other countries Example: Income inequality arguably has grown in UK as a result of the Research Assessment Exercise and heavy emphasis on publications Study by De Fraja and colleagues finds positive relationship between inequality of pay and score received on the RAE, especially strong for Russell group Paula Stephan Georgia State University & NBER

17 Costs Affect Use of Mammals-- Mostly Mice
Of Mice and Women: Wald and Wu 2010; data collected by Zucker and Beery Paula Stephan Georgia State University & NBER

18 Paula Stephan Georgia State University & NBER
Of Mice and Gender Why are male mice more commonly studied than female mice? Only in reproductive studies and immunology is ratio of female subjects to male subjects greater than one. Costs are a factor: four-day ovarian cycle of female mice means researchers must monitor females daily in experiments where hormones may play a role. / Paula Stephan Georgia State University & NBER

19 Paula Stephan Georgia State University & NBER
Zebra Fish Tremendous growth in use of zebra fish in labs: Cost is one factor: Zebra fish goes from egg to fully formed embryo in less than 24 hours, more than ten times faster than a mouse Embryos are also see through, making it easier to track development Paula Stephan Georgia State University & NBER

20 Costs Affect When Equipment Used
Large Hadron Collider Near Geneva Cost of electricity in Switzerland depends upon use--highest in winter Means LHC generally is not run in winter Winter shutdown provides window for maintenance and upgrade Paula Stephan Georgia State University & NBER

21 Incentives and Costs Affect Staffing of Labs at Universities

22 Paula Stephan Georgia State University & NBER
Incentives Increased importance of Specialization in research Funding for research Publications as necessary condition for funding and continuation of funding Paula Stephan Georgia State University & NBER

23 Paula Stephan Georgia State University & NBER
Specialization Sole author is dinosaur when it comes to research—fewer than 15% of papers are now sole authored Reflects specialization occurring in scientific research Paula Stephan Georgia State University & NBER

24 External Funding Important to Universities High-end Shopping Mall Model
Business of building state-of-the art facilities and reputation that attracts good students, good faculty and resources “Lease” facilities to faculty in form of indirect costs Faculty “pay” for opportunity of working at university, by writing salary off grant Those in soft money positions receive no guarantee of income if they fail to bring in a grant

25 Paula Stephan Georgia State University & NBER
Faculty and Funding Faculty increasingly under pressure to bring in funding for research University provides start up package at time of hire to provide funds for lab, including funds for graduate students and postdocs After three to four years faculty member on own to cover expenses plus write-off part of salary on grant Funding absolutely essential for promotion and tenure Pressure to bring in funding even more acute for faculty in soft money positions— “funding or famine” to quote Stephen Quake Tenure not equivalent to salary guarantee at majority of medical schools Tenured faculty at Harvard Medical School write off 95% of their salary on grants At same time, funding is in short supply and success rates are declining Paula Stephan Georgia State University & NBER

26 NIH and NSF Success Rates Available Years
NSF rates for are for the Division of Biological and Medical Sciences

27 Paula Stephan Georgia State University & NBER
Key Role of Funding Raises importance of having other people to work in lab—PI’s time is diverted to grant preparation/grant administration Estimate that PIs on Federal grants spend 42% of research time on in grant-related administration (Kean) Raises importance of publications given key role publications play in grant review and grant success Paula Stephan Georgia State University & NBER

28 Paula Stephan Georgia State University & NBER
Staffing of Labs Forces of specialization, funding and importance of publications lead PIs to seek clever individuals to staff their labs and help in production of research Three groups to choose from: graduate students postdocs staff scientists This is where costs play large role Paula Stephan Georgia State University & NBER

29 Cost of a Graduate Student
Stipend between $24,000 to $36,000. National average $26,078* Can cost an additional $16,000 or more once tuition is included, depending upon limits set by funding agency and policies of university GRAs work approximately 1600 to 2500 hours per year, depending upon year and discipline Hourly rate is around $19.50, assuming cost of $42,000 and hours of 2150* Hourly rate can be as high as $27.50 on some other campuses before fringes *

30 Cost of Postdoctoral Scholars
NIH stipulated rate for FY 2016 was $43,692 for NRSA first-year postdoctoral scholar; up from $42,840 in 2015; $39,264 in 2013. Many institutions follow this rate for other postdocs Average postdoc reported working 2650 hours a year in life and physical sciences; 2550 in engineering and 2500 in math and computer sciences Hourly rate before fringes is currently about $16.50 in the biomedical sciences Paula Stephan Georgia State University & NBER

31 Cost of Staff Scientist
Start at approximately $60,000 Fringe benefits are significantly higher than those for a postdoc because they are treated as employees by university Hourly rate of approximately $30.00 before fringes Paula Stephan Georgia State University & NBER

32 Which Would You Choose? Staff scientist: $30.00 per hour minimum
Postdoc: $16.50 to $17.50 per hour Graduate student: $19.50 to about $27.50 depending upon tuition; hourly rate may be lower for more advanced students who work more hours

33 Stephen Quake’s Lab: Stanford University
Paula Stephan Georgia State University & NBER

34 Paula Stephan Georgia State University & NBER
First Authors: N=137 Paula Stephan Georgia State University & NBER

35 Unintended Consequences
Paula Stephan Georgia State University & NBER

36 Unintended Consequences
Outcomes that result from system that has evolved often have unintended consequences Talk about three and inefficiencies associated with them Over Training Risk Aversion Over Building Paula Stephan Georgia State University & NBER

37 Paula Stephan Georgia State University & NBER
Overtraining Paula Stephan Georgia State University & NBER

38 Paula Stephan Georgia State University & NBER
Overtraining: Supply of PhDs trained to do research & who want to do research is greater than demand Paula Stephan Georgia State University & NBER

39 Evidence of Overtraining Relative to Demand
Decline in definite commitments of new PhDs Significant percent of those with definite commitments take postdoctoral position Growth in non-research jobs Paula Stephan Georgia State University & NBER

40 Many New PhDs Lack Definite Commitments at Time They Finish
Paula Stephan Georgia State University & NBER

41 Definite commitments at doctorate award, by science and engineering fields of study: 1994–2014

42 Definite Commitments Overtime
Field 2004 2009 2014 All 70.0% 69.5% 61.4% Life sciences 71.2% 66.8% 57.9% Physical sciences 71.5% 72.1% 63.8% Social sciences 71.3% 72.9% 68.8% Engineering 63.6% 57.0% Education 74.6% 71.6% 64.6% Humanities 63.4% 63.3% 54.3% Paula Stephan Georgia State University & NBER

43 Many with Definite Commitments Will Go to Postdoctoral Training
Paula Stephan Georgia State University & NBER

44 U.S. postdoc rate, by field of study: 1994–2014

45 Paula Stephan Georgia State University & NBER
Why a Postdoc? Other positions not available Among recent physics PhDs 1 in 4 say “Could not obtain a suitable permanent position” is very influential in why they took a postdoc position Next logical step Paula Stephan Georgia State University & NBER

46 Paula Stephan Georgia State University & NBER
People Who Aspire to Faculty Position Are Likely to Take a Postdoctoral Position Paula Stephan Georgia State University & NBER

47 Majority of Postdocs Have Preference for Job in Academe
Paula Stephan Georgia State University & NBER Sauermann & Roach: “Why Pursue a Postdoc?”

48 Positions are Scare: Academe is the “alternative” career track now
Paula Stephan Georgia State University & NBER

49 Tenure and Tenure-track Positions 3-5 Years Since PhD
10.6% biological, agricultural and environmental sciences; (17.3%) 14.3% physical sciences; (18.8%) 14.6% engineering; (22.7%) 13.8% computer and information sciences; (55.7%) 29.6% math and statistics; (54.9%) Red is 2013; Blue is 1993; Table Indicators. Paula Stephan Georgia State University & NBER

50 Employment Outcomes by Cohort Biomedical Sciences

51 Paula Stephan Georgia State University & NBER
What’s Inefficient? Training has become less about future supply and more about getting research and teaching done now; in process we train more PhDs than will be able to use their knowledge and skills in research in workplace; much of this training at expense of the public Paula Stephan Georgia State University & NBER

52 Inefficiencies in the System
Overtraining Risk Aversion Overbuilding Paula Stephan Georgia State University & NBER

53 Risk Aversion Governments today support researchers at universities and institutes throughout much of world As funding model has grown and spread concern has arisen regarding whether funders, the universities where they work, and researchers are taking sufficient risk

54 Evidence? Recent research by Pérez and coauthors finds UK’s Engineering and Physical Sciences Research Council to be less likely to support research characterized as “radical” Research by Boudreau et al. find evaluators give lower scores to highly novel proposals—proposals bringing together “distant” pieces of knowledge—that could be viewed as “risky” Research by Rzhetsky and coauthors finds most researchers focus on molecules that are known to be important and closely related to each other. Over time, their choice of molecules to study becomes more conservative Considerable anecdotal evidence that funding agencies and universities are not inclined to support risky research today

55 Decline in Risk Taking Perception among many scientists and policy makers that appetite for risk taking among public funding agencies has declined or is sorely absent. To quote Nobel laureate Roger Kornberg, “If the work that you propose to do isn’t virtually certain of success, then it won’t be funded.”

56 James Rothman, 2013 Nobel Laureate in Physiology or Medicine, Comments on Risk
Rothman told interviewer that “he was grateful started work in the early 1970s when the federal government was willing to take much bigger risks in handing out funding to young scientists” “I had five years of failure, really, before I had the first initial sign of success. And I’d like to think that that kind of support existed today, but I think there’s less of it. And it’s actually becoming a pressing national issue, if not an international issue.” Nobel Laureate, Physiology or Medicine, 2013 Interview on NPR

57 PNAS Article Recent article by Bruce Alberts, Marc Kirscher, Shirley Tilghman and Harold Varmus in PNAS criticized the biomedical research in the US as being overly risk averse. Recommended that “Science agencies should significantly increase the numbers and kinds of awards that emphasize originality and risk-taking…”

58 “Goodbye, Columbus” Gregory A. Petsko Cornell Weill Medical School
“Explored” reasons why Columbus’s proposal “Finding a New Route to the Indies by Sailing West” was (hypothetically) rejected Genome Biology :155

59 Greg Petsko’s Take on Columbus’s “Hypothetical Rejection”
Too ambitious—suggest he go to Portugal, instead. Lack of preliminary data Failure would be disastrous for funder-- “think of how it would look if we funded something that didn’t pan out.” Poor fit for reviewers: Experts (da Gamma and Magellan) too busy to review proposal Limited funds Funds are used for data collection (“Grape Vine Sequencing”) rather than hypothesis testing—data collection projects are “guaranteed to work”

60 Not Unique to Biomedical Research
A researcher at a major US institution recently stated, speaking of NSF funding: “Programs are not very adventurous.” “And what I experienced was that I couldn’t get any new idea or anything I was really excited about funded by NSF. It never worked…the feedback is ‘well this is too new: we don’t know whether its going to work.’” DARPA, which once boasted that “it took on impossible problems and wasn’t interested in the merely difficult,” has according to critics, increasingly shifted to funding research that is more near-term and less risky

61 Not Unique to US Expressed concern in UK
ERC sufficiently concerned to create an ad hoc committee to make recommendations regarding ways to enhance risk taking

62 Why Should We Care? What’s Inefficient about This?
Major reason governments supports research is due to what economists refer to as “market failure” Market failure discourages (most) firms from investing in risky projects—too difficult to capture the financial benefits Yet it is risky projects that have potential to shift knowledge frontier and eventually contribute to economic growth

63 Causes for Risk Aversion?
Incentives and Structure of Funding System Promote Risk Aversion

64 Structure of Grant Short term nature of grants—3 to 4 years; hard to recover from “failure.” Focus on projects rather than people Heavy emphasis on preliminary data: “No Crystal No Grant” Ability to continue line of research Practice of relying on multiple grants to fund PIs Comparative Sequence and Structural Analyses of G-Protein-Coupled Receptor Crystal Structures and Implications for Molecular Models; PLOS One Sept

65 Review Process Reviewers appear inherently risk averse
Boudreau and coauthors find reviewers scores negatively correlated with novel combinations of medical subject terms Reviewers underappreciate multidisciplinary research Review by consensus Focus on bibliometric measures Some funding agencies require bibliometric measures; others do not discourage it Reviewers often rely on bibliometric measures

66 Relationship to Bibliometrics and Risk Aversion
Wang, Veugelers and Stephan create a measure of novelty of research that is arguably related to risk taking Find short-term bibliometric measures biased against novelty.

67 Novel Research Takes Time to Be Recognized
Paula Stephan Georgia State University & NBER

68 Journal Impact Factor Bias agains Novelty
JIF Poisson Moderately novel *** *** *** Highly novel *** *** *** Journal age < 4 *** Journal age (ln) 0.2211***

69 Novel Research is Appreciated Most Outside Field
Paula Stephan Georgia State University & NBER

70 Foreign vs Home Field

71 Trans disciplinary impact over time

72 Incentives Encourage PIs to be Risk Averse
Importance of having funding encourages submitting less risky research. University only supports lab for 3 to 4 years Need to establish research agenda soon after getting faculty position (or before) discourages risk taking; once funded continue in same path Risk aversion fostered by fact that faculty need funding to support salary—mandatory for those on soft money positions or in tenured positions where tenure does not come with salary guarantee

73 Emphasis on Accountability Discourages Risk Taking
At funder level Fear of failure Pressure to demonstrate to government success of investments—go for short term, tangible results Invest in translational research At university level Pressure to demonstrate contribution to economic growth in quest for ever more resources Precarious--in looking for resources universities have tendency to oversell short run benefits of investing in research and thus make a Faustian bargain by failing to remind public that link between research & growth can take a long time Paula Stephan Georgia State University & NBER

74 Not Limited to Public Funders of Research
Private foundations and philanthropists often focus on promoting research with translational goals; want research institutes to “stand-on-their own” in near term Paula Stephan Georgia State University & NBER

75 A Caveat Excellence and risk taking are not the same thing
Excellent research does not necessarily require risk taking Risk taking does not guarantee excellent research outcomes

76 Need a Portfolio Does not mean that there is not a substantial role for public support of “normal” research or what physicists in the 1980s referred to as “ditch digging.” Need normal research to explore and exploit knowledge that is advancing the frontier; however, need to remember that “normal” research is what firms excel at

77 Inefficiencies in the System
Overtraining Risk Aversion Overbuilding Paula Stephan Georgia State University & NBER

78 Overbuilding Doubling of NIH budget encouraged universities to build new biomedical research facilities—assumed NIH funding would continue to grow Many universities borrowed to do so Encouraged by government accounting rules which make debt an accounting asset Interest university pays for debt service can be included in calculating indirect Building binge further fuelled by competition among universities to recruit senior, high-performing faculty

79 University Response System provided incentives to train more individuals; especially because costs of using graduate students and postdocs to staff labs were low relative to staff scientists Incentives to create research programs, built on soft money and prospects of charging salary off grants

80 Debt Buildings often built with debt AAMC survey found
average annual debt service for buildings in 2003 was $3.5 million per medical school grew to $6.9 million in 2008 Where will the money come from?

81 Summary Incentives and costs affect the practice of science
They can lead to unintended consequences with inefficiencies Overtraining Risk aversion Overbuilding

82 Drawn from Recently published in paper
The Endless Frontier: Reaping What Bush Sowed?

83 Paula Stephan Georgia State University & NBER
Comments in Nature Research Efficiency: Perverse Incentives Blinkered by Bibliometrics Paula Stephan Georgia State University & NBER

84 Questions/Comments Paula Stephan


Download ppt "How Economics Shapes Science"

Similar presentations


Ads by Google