The Location Decision of the Highly Educated: A Statewide Analysis

1
The Location Decision of the Highly Educated A
Statewide Analysis

• Sarah Wakefield

2
Motivation

• New Location Theory
• Less focus on business, more on people
• Impact on State Economies
• Observe positive impacts
• Develop Strategies to Attract Talented
  Individuals
• Talented Bachelors degree and higher

3
My Contribution

• Analysis at a statewide level
• Inclusion of explanatory variables such as park
  visitors and number of libraries
• Analyze the effect of net migration with patent
  data

4
Research Questions

• What factors attract talent at the state level?
• Does the relevance of each factor change as
  education level increases?
• Is there an observable positive impact on state
  economies?

5
Literature Review

• Kodrzycki (2001)
• location decisions of the college educated
• Positive net migration of recent college
  graduates into South Atlantic, Mountain,
  Pacific regions
• New England, East West North Central, and East
  South Central lost the highest shares of college
  graduates

6
Literature Review

• Florida (2002)
• Key role of diversity
• Diversity attracts better than climate,
  recreational, and cultural amenities
• Talent and high tech industries lead to higher
  regional incomes

7
Literature Review

• Glaeser (2005)
• Measure the effects of higher education on growth
  rates
• An area with twice as many colleges in 1940 faced
  4 faster growth rate after 1970
• Positive effect on wages for high and low skilled
  workers

8
Data By State

• Dependent Variables
• Net Migration Bachelors
• Net Migration Masters
• Net Migration Doctorate
• U.S. Census Bureau 1995-2000
• Number of Patents
• U.S. Patent and Trademark Office 2005
• Economic Growth
• Bureau of Economic Analysis 2004-2005

• Independent Variables
• Per Capita Income
• Bureau of Economic Analysis 2000
• Violent Crime Rate
• Bureau of Justice Statistics 2000
• Climate
• U.S. Dept. of Agriculture 1941-1970
• Number of Libraries Per Capita
• National Center for Education Statistics 2000
• Unemployment Rate
• Bureau of Labor Statistics 2000
• Restaurants
• National Restaurant Association 2000
• Percent Foreign Speaking
• U.S. Census 2000
• Park Visitors/Acreage
• U.S. Census 1997

9
Methodology

• Multiple Linear Regressions
• Dependent variable net migration of bachelors
  degrees, masters degrees, and doctorate degrees

10
Methodology

• Simple Linear Regressions
• Run regression with net migration by master and
  doctorate degrees

11
Overview

• Location choice of talent can positively impact a
  state
• Understanding which factors draw in talent can
  help states develop positive economic policies
• Regression results will indicate which factors
  are significant, and if talent has a positive
  impact on state economies

12
Descriptive Statistics

Variables N min max mean std
Libraries 50 0.078 20.658 4.95556 5.231233
Economic Growth 51 -1 8.7 3.423529 2.077555
Patents 48 35 17989 1588.583 2749.328
Foreign Speaking 51 2.5 39.3 12.04706 8.948304
Per Capita Income 50 21005 41489 28334.78 4413.04
Unemployment Rate 50 2.3 6.2 3.8440 0.90535
Violent Crime Rate 51 81.4 1507.9 443.3294 241.1401
January Sun 51 22 248 148.9862 46.6818
January Temperature 51 9 195 36.86379 26.0762
July RH 51 14.9 79 56.13847 17.19921
July Temperature 51 61.8 91.15 74.84787 5.664157
Park Acreage 50 9 3289 249.74 491.7086
Net Migration Bachelors and Above 50 -34642 191070 10701.24 33345.37
Net Migration Bachelors 50 -32331 140588 6948 25054.24
Net Migration Masters 50 -2179 39299 2796.44 6583.165
Net Migration Doctorate/Professional 50 -1159 11183 955.86 2294.873
Bachelors Above/Population 50 8.109E-05 0.003976 -0.01075 0.008209
Bachelors/Population 50 -0.0002072 0.003268 -0.00905 0.00672
Masters/Population 50 0.0002143 0.000598 -0.00131 0.001314
Doctorates/Population 50 7.403E-05 0.00024 -0.00054 0.000557
13
Tests of Multicollinearity and Heteroskedasticity

• For all four regressions of net migration,
  variance inflation factor lt 10, so
  multicollinearity is not an issue
• Heteroskedasticity Tests
• Cannot reject null
• No presence of
• heteroskedasticity

 Model Chi-Square P Value
Bachelors Above 47.90 0.6357
Bachelors 49.95 0.5549
Masters 49.51 0.5332
Doctorate 49.86 0.5910
14
ANOVA and Adjusted R-Squared Results

• ANOVA results
• Overall, each model is significant
• Adjusted R2
• Each model explains about 58 of the variation in
  net migration

Model  F Value P Value
Bachelors Above 6.89 lt.0001
Bachelors 6.25 lt.0001
Masters 6.93 lt.0001
Doctorate 7.95 lt.0001
Model  Adj. R2
Bachelors Above 0.5460
Bachelors 0.5174
Masters 0.5474
Doctorate 0.5864
15
Results from Bachelors and Above Regression
    Parameter Standard    
Variable Label Estimate Error t Value P Value
           
Intercept Intercept -0.01202 0.00974 -1.23 0.2245
JanSun January Sun -3.8E-06 1.57E-05 -0.24 0.8087
JanTemp January Temperature 1.88E-05 2.06E-05 0.91 0.3683
JulyRH July Relative Humidity -5.9E-06 2.65E-05 -0.22 0.8237
JulyTemp July Temperature -1.61E-05 9.62E-05 -0.17 0.8676
vcr Violent Crime Rate 4.97E-06 2.96E-06 1.68 0.1006
fl Percent Speaking Foreign Language 4.41E-05 7.57E-05 0.58 0.5637
inc Per Capita Income 3.992368E-7 3.76E-08 2.9 0.0061
lib Libraries per Capita -0.00022 8.13E-05 -2.76 0.0088
perpark Park Visitors/Acreage -1.3E-06 4.52E-06 -0.3 0.7679
UR Unemployment Rate 0.000254 0.000723 0.35 0.7275
16
Interpretation of Results

• If the violent crime rate increased by 1 (per
  100,000) net migration would increase by .497
  (per 100,000)
• If per capita libraries increase by 1, net
  migration would decrease by 22 people (per
  100,000)
• A 1000 increase in income increases net
  migration by 40 people (per 100,000)

17
Results from Bachelors Regression
Variable Label Parameter Standard t Value P Value
    Estimate Error    
Intercept Intercept -0.00937 0.00825 -1.14 0.2631
JanSun January Sun -5.66472E-07 0.00001328 -0.04 0.9662
JanTemp January Temperature 0.00001696 0.00001749 0.97 0.3381
JulyRH July Relative Humidity -6.03E-06 0.00002241 -0.27 0.7894
JulyTemp July Temperature -0.00001887 0.00008156 -0.23 0.8182
vcr Violent Crime Rate 4.32E-06 0.00000251 1.72 0.0928
fl Percent Speaking Foreign Language 0.00002126 0.00006417 0.33 0.7422
inc Per Capita Income 3.02E-07 1.16597E-07 2.59 0.0133
lib Libraries per Capita -0.0001973 0.00006891 -2.86 0.0067
perpark Park Visitors/Acreage -9.42927E-07 0.00000383 -0.25 0.8068
UR Unemployment Rate 0.00024233 0.00061261 0.4 0.6946
18
Interpretation of Results

• If violent crime rate increases by 1 (per
  100,000), net migration will increase by .432
  people (per 100,000)
• If per capita libraries increase by 1, net
  migration decreases by 19.73 people (per 100,000)
• A 1000 increase in income increases net
  migration by 30.2 people (per 100,000)

19
Results from Masters Regression
Variable Label Parameter Standard t Value P Value
    Estimate Error    
Intercept Intercept -0.00232 0.00146 -1.58 0.1211
JanSun January Sun -0.00000306 0.00000235 -1.3 0.2009
JanTemp January Temperature -3.57395E-07 0.0000031 -0.12 0.9087
JulyRH July Relative Humidity 1.30E-07 0.00000397 0.03 0.974
JulyTemp July Temperature 0.00000909 0.00001444 0.63 0.5326
vcr Violent Crime Rate 4.14E-07 4.43759E-07 0.93 0.3566
fl Percent Speaking Foreign Language 0.00001664 0.00001136 1.46 0.1511
inc Per Capita Income 7.61E-08 2.06477E-08 3.69 0.0007
lib Libraries per Capita -0.00001791 0.0000122 -1.47 0.1501
perpark Park Visitors/Acreage -2.50172E-07 6.78293E-07 -0.37 0.7143
UR Unemployment Rate -0.00002377 0.00010848 -0.22 0.8277
20
Interpretation of Results

• Only per capita income is significant
• A 1000 increase in income increases net
  migration of those with a masters by 7.61 people
  (per 100,000)

21
Results from Doctorate Regression
Variable Label Parameter Standard t Value P Value
    Estimate Error    
Intercept Intercept -0.00033218 0.00056193 -0.59 0.5578
JanSun January Sun -1.94955E-07 9.04631E-07 -0.22 0.8305
JanTemp January Temperature 0.00000218 0.00000119 1.83 0.0746
JulyRH July Relative Humidity -3.45E-08 0.00000153 -0.02 0.9821
JulyTemp July Temperature -0.00000636 0.00000555 -1.15 0.2588
vcr Violent Crime Rate 2.41E-07 1.70648E-07 1.41 0.1651
fl Percent Speaking Foreign Language 0.0000062 0.00000437 1.42 0.1641
inc Per Capita Income 2.08E-08 7.94009E-09 2.62 0.0125
lib Libraries per Capita -0.00000903 0.00000469 -1.93 0.0615
perpark Park Visitors/Acreage -1.50386E-07 2.60839E-07 -0.58 0.5676
UR Unemployment Rate 0.00003514 0.00004172 0.84 0.4047
22
Interpretation of Results

• 10 degree increase in January temperature
  increases doctorate migration by 2.18 (per
  100,000)
• Increase of 1 in per capita libraries decreases
  doctorate migration by .903 (per 100,000)
• 1000 increase in income increases migration of
  doctorate holders by 2.08 (per 100,000)

23
Positive Effects of Migration on a State

• Tests of Heteroskedasticity
• - Results of chi-square tests show no presence
  of heteroskedasticity

Patents Chi-Square P Value
Bachelors above 1.08 .5840
Bachelors 0.97 .6146
Masters 1.52 .4670
Doctorate 1.83 .4007
Economic Growth Chi-Square P Value
Bachelors above 0.33 .8498
Bachelors 0.71 .7000
Masters 1.77 .4133
Doctorate 0.09 .9541
24
Regression Results - Patents
 Model Intercept a R2
Bachelors and Above .00024638 (lt.0001) .01089 (.1407)  .0253
Bachelors .00024962 (lt.0001) .01135 (.2095) .0127 
Masters .00022218 (lt.0001) .11895 (.0130) .1057
Doctorate .00023711 (lt.0001) .14584 (.2320) .0096
25
Interpretation of Results

• Only masters degree holders have a significant
  effect on patents
• Increase of 10 masters degree holders in 2000
  increases patents in 2005 by 1.1895
• 11 of the variation in patents issued is
  explained by net migration of masters degree
  holders

26
Regression Results Economic Growth
 Model Intercept ? R2
Bachelors and Above 3.37953 (lt.0001) 129.0867 (.0903)  .0390
Bachelors 3.42122 (lt.0001) 150.65378 (.1047) .0342 
Masters 3.22745 (lt.0001) 758.51091 (.136) .0258 
Doctorate 3.18379 (lt.0001) 2785.64766 (.026) .0810 
27
Interpretation of Results

• Increase of 1 with bachelors degree and higher
  increases growth by .129
• Increase of 1 with bachelors degree increases
  growth by .151
• Masters degree holders no significant impact
• Increase of 1 with a doctorate increases growth
  by 2.79

28
Conclusion

• Per capita income is positive and relevant in the
  attraction of talent at all education levels
• Violent crime rate positive effect on
  bachelors degree holders
• Metropolitan areas
• Negative effect of per capita libraries on
  migration of bachelors and doctorate holders
• Educational resources not a strong priority
  (sample only includes ages 22-29)

29
Implications

• Preferences do vary based on education level
• Positive effects of net migration exist and are
  significant
• Policy Initiatives
• Restructure tax codes
• Attract businesses to increase salary
• Emphasize metropolitan areas
• To attract doctorate holders, promote climate

30
Bibliography

• Florida, Richard. The Economic Geography of
  Talent. Annals of the Association of American
  Geographers, 92(4), p. 743-755 (2002).
• Glaeser, Edward. Smart Growth Education,
  Skilled Workers, and the Growth of Cold-Weather
  Cities. Harvard University, April 2005.
• Gee, Wilson. The Drag of Talent out of the
  South. Social Forces, Vol. 15, No. 3, p.
  343-346 (1937).
• Harden, Brian. Brain-Gain Cities Attract
  Educated Youth. The Washington Post, 2004.
• Kodrzycki, Yolanda. Migration of Recent College
  Graduates Evidence from the National
  Longitudinal Survey of Youth. New England
  Economic Review (2001).
• Schwartz, Aba. Migration, Age, and Education.
  The Journal of Political Economy, Vol. 84, Issue
  4, Part 1, p. 710-720 (1976)