Migrating Talent: The Location Decisions of Science and Engineering Ph.D.s

1
Migrating Talent The Location Decisions of
Science and Engineering Ph.D.s

• Albert Sumell
• Youngstown State University

2
Acknowledgements

• Science Resource Statistics, NSF
• Paula Stephan, Georgia State University

3
Major questions/objectives

• Examine degree to which new Ph.D.s headed to
  industry remain in the state or MSA where they
  received their training.
• Examine what draws new Ph.D.s to an area
• Relationship between location choice, individual
  characteristics, amenity quality, and expected
  income levels.
• Use Ph.D. city location decisions to estimate how
  they value of location-specific amenities.
• What can policymakers do to retain or attract
  more highly educated people to their region?

4
Why do we care?

• Creating a highly skilled work force is one way
  universities contribute to economic growth.
• Universities use the economic development
  argument as a lever for state funds.
• Location decisions of the highly educated affect
  human capital levels, and a regions
  productivity.
• Ph.D.s are integral to the creation and diffusion
  of knowledge and the development of new products
  and processes.
• Generally know little about their migration
  decisions.
• Mobility increases with education.
• Ph.D.s are an extremely mobile cohort, especially
  right after graduate school.
• Amenities play an important role in location
  decisions.
• Not all amenities can be changed.
• Natural and publicly-provided amenities.

5
Overview of Presentation

• Data
• Summary of migration trends by region, state, MSA
• Framework for empirical analysis
• What determines whether they leave?
• What determines where they go?
• Results
• Conclusions

6
Data

• Primary data comes from the 1997-1999 Survey of
  Earned Doctorates.
• Administered by Science Resources Statistics,
  National Science Foundation
• Over 92 response rate
• Coded U.S. city choices of all newly minted
  Ph.D.s with definite plans to work in industry,
  academe, and academic postdocs.
• City location has never been coded for those
  going to industry previously.
• Data misses individuals who have not finalized
  their work plans at time questionnaire is filled
  out.
• Use Survey of Doctorate Recipients to impute
  expected wages in each city.

7
Summary of Data

• 65,427 Ph.D.s trained in SE field
• 41,670 with definite plans
• 37,395 stay in U.S.
• 27 industry, 14 academic faculty, 46 academic
  postdocs
• 25,827 have known city of employment

8
Placement of New SE PhDs by Field of
Study
9
Findings with regards to retention

• Regions retain 46.1
• States retain 37.2
• MSAs retain 19.1
• Substantial variation
• Low compared to other degrees
• Law school graduates 57 stay in state
• Bachelors and Masters in science 64
• Bachelors and Masters in engineering 62

10
Percent of New PhDs Who Stay In State
11
Percent Gain/Loss of New PhDs By State
12
Percent Gain/Loss of New PhDs By Region
13
Top 25 Producing MSAs
14
Top 25 Employing MSAs
15
Considerable overlap

• Eighteen metropolitan areas are in the top 25 in
  producing and employing
• Areas that import more Ph.D.s from other areas
  generally retain more of their own.
• High geographic concentration
• Top 20 production cities produce 50 of all new
  Ph.D.s.
• Top 20 destination cities attract 47 of all new
  Ph.D.s

16
Industrial Ph.D.s

• 40 of industrial Ph.D.s work for top 200 RD
  firm
• Expend more than 70 of all RD in U.S.
• Top 20 cities employ 60 of new industrial PhDs.
• Not as geographically concentrated as patent or
  SBIR counts.
• Top five destination cities.
• San Jose 1878
• Boston 1015
• New York 937
• Washington DC 758
• Chicago 669
• Atlanta ranks 14th, with 150 industrial Ph.D.s,
  almost 50 of whom were trained in Atlanta.
• Many new PhDs work for consulting firms and
  financial services, not manufacturing firms.
• Innovative activity occurs in processes, not just
  products.

17
Certain areas stand out

• Major brain drain from Midwest
• Produces 19 of new PhDs employ only 13.
• Retains only 37 of industrial Ph.D.s
• Indiana retains 19 of all Ph.D.s (12 of
  industrial)
• Lafayette, Indiana2.9 retention,
  Urbana-Champaign3.2 retention, State College,
  Pa3.3 retention
• Pacific retains 69 of industrial Ph.D.s, saw 39
  net gain in industrial Ph.D. employment.
• California plays a special role produces more,
  retains more and import more from any other
  state.
• Has 5 of top 20 destination cities.
• South Atlantic retains 44 of industrial Ph.D.s,
  experienced 29 loss (only 6 loss if consider
  all Ph.D.s)
• Atlanta trained 700 new Ph.D.s overall, employs
  only 332.
• Surprisingly, New York loses more on net than any
  other state.

18
International Destinations

• Five percent of industrial Ph.D.s have plans to
  work for industry outside U.S.
• Korea--250
• Germany--96
• Japan--93
• Canada--66
• Taiwan55
• Approximately 60 are headed to China, India and
  Thailand

19
Empirical Models

• View migration as an investment decision
• Individual will move if present value of stream
  of benefits resulting from move is greater than
  cost of moving. Will locate in area that offers
  them highest utility level.
• First analysis focuses on whether the PhD leaves
  area from which they receive degree. State,
  CMSA, PMSA unit of observation.
• Second analysis focuses on where PhDs go to
  which MSA attributes are they most drawn.

20
Whether to stay or go

• Binary choice models
• Logit equations estimated at state and MSA
  levels, marginal effects reported.
• Focus on industrial Ph.D.s with definite plans
  and known location (N10,000)
• 3 sets of variables
• Variables that reflect attributes of state and
  local area
• Variables that reflect individual characteristics
• Variables that reflect field differences,
  institutional characteristics and RD status of
  hiring firm

21
Findings

• Most demographic factors affect mobility in
  expected way.
• Older, married, with dependents more likely to
  stay.
• Nonwhites, temporary residents more likely to
  leave.
• Quality of program matters.
• Graduates from top programs in engineering,
  chemistry, earth science, math are more likely to
  move out of state.
• Where you went to high school and college
  matters.
• Networks matter - what you did last year in
  graduate school
• Those who worked more likely to stay, if on
  fellowship more likely to leave.
• Earth scientists, agricultural scientists, and
  chemists most likely to leave.
• Debtors more likely to leave.

22
Some effects are quite substantial

• Temporary residents 6 to 7 more likely to leave
  state or MSA. Permanent residents 3 more likely
  to stay in state.
• Those part time employed final year of program
  19 more likely to stay in state, 14 in MSA.
  Those full time employed 9 more likely to stay
  in state, 5 in MSA.
• Those who earn PhD in same state they went to
  college are 9 more likely to stay in state 5
  more likely to stay in MSA.
• Those who earned PhD in same state they went to
  high school are 6 more likely to stay in state.
• Ph.D.s from top ranked medicine and biology
  programs 11 and 9 more likely to leave state
  respectively.

23
Effect of Location Attributes

• Technological infrastructure matters
• More likely to stay in state the higher the state
  industrial RD expenditures.
• More likely to stay in MSA more RD labs there
  are more patents that are issued higher milken
  index.
• Education levels, employment opportunities, and
  per capita income
• Effect of employment opportunities (relative
  absorptive capacity of the area) is substantial.
• Percent highly educated is surprisingly
  insignificant.
• P.C. income is positive and significant at state
  level, not at MSA level.

24
What characteristics attract new Ph.D.s?

• Random Utility Model
• Compare attributes of chosen (utility maximizing
  cities) to potential alternatives.
• How much are new Ph.D.s willing to pay to live in
  city with higher quality amenities?
• Variables include
• Natural amenities
• Summer and winter temps, humidity, hours of
  January sunlight, coastal access.
• Publicly provided amenities
• Crime rate, air quality, superfund sites, park
  acres, commute time, entertainment quality, pupil
  teacher ratio.
• Other characteristics
• Density, size, demographic variables, percent
  highly educated, percent democratic, number of
  patents.
• Estimate expected income and housing
  expenditures.

25
Main Findings

• Effects are generally substantially larger in
  magnitude for natural amenities than publicly
  provided amenities.
• Indicates new Ph.D.s put more weight on climate
  than crime.
• Willingness to pay is largest for ocean
  proximity, lower summer humidity, and lower
  summer temperature.
• What publicly provided amenities matter?
• Ph.D.s are more likely to locate in cities with
  higher air quality, less traffic, lower crime
  rates, and greater levels of diversity (percent
  foreign born or nonwhite).
• Other characteristics
• Coefficients on number of higher education
  institutions, number of patents, and percent
  highly educated all positive and significant.
• Percent democratic positive but generally
  insignificant.

26
Surprising results

• Superfund sites, student expenditures, hours of
  sunlight take on counterintuitive signs.
• Superfund sites correlated with urban
  amenities?
• Student expenditures suggest school
  inefficiency?
• Hours of sunlight - Pacific Northwest effect?
• Art and entertainment quality and acres of
  parkland do not have an impact.
• Willingness to pay estimates are sometimes larger
  than expected.
• 10,000 for coast, 6,000 and 3,000 for 10
  reduction in July temp and humidity, respectively.

27
Individual heterogeneity

• Preferences vary according to
• Age, marital status
• Younger, single Ph.D.s are move greater distances
  than older, married Ph.D.s
• Whether they have children
• Parents care much more about violent crime rates
  and pupil teacher ratios
• Citizenship
• Non-citizens are more likely to locate in cities
  with greater amounts of foreign born residents
• Employment sector
• Number of patents have large affect industrial
  Ph.D.s choice.
• Postdocs care more about art and entertainment
  quality, less about commute times

28
Summary and Policy Questions

• Some results are consistent with Floridas
  Creative Class, some are not.
• Ph.D.s are different, and have unique
  preferences.
• Will Central states continue to produce and
  invest in Ph.D.s that immediately head out of
  state?
• Could argue their investment benefits the
  national economy more than their local economy.
• Are state legislators unaware of migration flows?
• Will the Federal government need to provide
  financial assistance to maintain highly trained
  SE workforce?

29
Policy Issues

• Are new Ph.D.s a poor investment?
• Migration rates are correlated with but not equal
  to returns.
• Universities benefit from doctoral students
  relatively cheap labor for the classroom and
  labs.
• Composition of workforce would likely be much
  worse without role of Universities.
• Purdue is still major employer of industrial SE
  Ph.D.s in Indiana.
• How much does the migration trend reflect the
  decline of industrial prowess of Midwest in
  recent years, and how much does it contribute to
  it?
• Self reinforcing effects work in both directions.

30
Policy Issues Continued.

• So what are low quality amenity cities or regions
  to do?
• Education. Those educated in the area are most
  likely to locate in an area.
• Consider cost effectiveness of improving
  reproducible amenities.
• No magic bullet, but taken together, effects
  could be substantial.
• Dont only focus on post-graduate school policies
• Target potential students with ties to the area,
  and/or the amenities (cold-blooded).
• Consider tying funds to retainment rates.
• Build on networks and search for niches.
• Consider direct and indirect financial incentives
  of firms and workers.
• Look at Minneapolis.

31
Additional Considerations

• Migration decisions are complicated, and
  sometimes intangible.
• Look at employee choice, not employer choice.
• Analysis does not include new Ph.D.s who did not
  have definite plans or seasoned Ph.D.s (post
  postdocs).
• Analysis does not consider foreign trained
  Ph.D.s.
• Labor market conditions likely changed following
  precipitous decline of the tech industry.
• Not all highly educated are created equal.
• Whats attractive to one cohort of Ph.D.s is not
  necessarily attractive to all highly educated.

32
Questions/Comments?

• e-mail ajsumell_at_ysu.edu