Chicago Employment

Document Sample
Chicago Employment Powered By Docstoc
					Suburban Sub-centers and
 employment density in
  metropolitan Chicago


 Daniel P. McMillen (Tulane U)
John F. McDonald (U of Illinois)
  Journal of Urban Eco, 1998
               Introduction (1)
• A location well served by highways, rail lines and
 other transportation network may attract many
 firms even when the firms have no interest in locating near
 one another ---- suburban sub-center may form near
 transportation centers. Accessibility to these sub-centers
 lead to employment concentration. This implies that scale
 economies -- agglomeration        (transportation
 saving cost) can generate concentration of employment
 at certain locations within an urban area.
Scale Economies - transportation
           network
                • Transportation
                  network
           Theoretic Framework
• Bid-rent function is used to see effects of suburban sub-
  center on employment density. A bid-rent function
  represents the maximum amount a firm or an individual
  will pay for a unit of land. In standard mono-centric model,
  the bid-rent function is a simple function of distance
  from the city center -- because all economic
  activities is assumed to take place there. Here, sub-urban
  employment is influenced by access to expressway
  interchanges and other features of the
  transportation network. (cont…)
        Theoretic Framework (2)
• Such accessibility measures are represented by the vector
  A, sub-center access measures by vector S and
  idiosyncratic characteristics (clear, level land, swampy) by
  C, this affects construction cost.
• Hypothesis
• Employment probability in the sub-centers
• Their impact on employment density
                 Methodology
• Non-residential: lnR1 = 1X + 1         X = (A,S,C)
• Household:          lnR2 = 2X + 2
• Net employment density: ln(E/Le) = ln R1 + 1 = 1X
  +1+ 1
• Gross employment:lnD = Z +  ------ (1)
• Employment density is a function of the same variables
  that determine land rents.
• Employment density increases when non-residential land
  rent increases
               Methodology (2)
• Employment density increases when non-residential
  land rent increases
• Employment density decreases when residential land
  rent increases
•     Prob (I=1) = Prob ( Z +  > 0 )      ------- (2)
• This equation determines whether there is some
  employment in a zone or not.
• Correlation between  and  implies employment density
  functions are subject to selection bias.
          Estimation Procedure
       • Two-stage method: a) probit/logit b) OLS
       • Maximum-likelihood estimation:
• E (lnD/I=1) =  Z +u  ( Z) /  ( Z)
• Northeastern Illinois Planning Commission data
• Sub-center identification: A set of nearby tracts that
  each have at least 10 employee/acre in either 1980 or
  1990 and together have an average over the 2 sample years
  of at least 10,000 employees ---20 sub-centers
             Estimation Results
• Expect: Increasing distance from a suburban
  employment sub-center lowers non-residential bid
  rents if scale economies exist --- lower employment
  density -- negative coefficient on distance ---- lower
  employment
  Estimation Results (2)
 T able 4 – M arginal effects for expected log-e m ploym ent density

V ariables                w ithin 15 m iles      m ore than 15 m iles
                          O f o,hare             from o’hare
                          ------------------     ------------------------
                          1980          1990     1980          1990
                          ------        ------   -------       -------

D istance to O ’hare      -0.079       -0.072    -0.005       -0.017
A irport                  (3.705)      (4.750)   (1.123)      (4.550)

D istance to C B D        -0.055       -0.026    -0.007       -0.008
                          (2.604)      (1.372)   (2.270)      (2.481)

D istance to com m uter -0.239         -0.110    -0.250       -0.200
T rain station          (1.506)        (0.727)   (5.771 )     (5.324)

D istance to high w a y   -0.134       -0.131    -0.083       -0.138
Interchange               (2.032)      (2.068)   (2.734)      (5.045)
(absolute asym ptotic t-values are in parentheses)
   Estimation Results (3)
 T able 4 – M arginal effects for expected log-e m ploym ent density

V ariables             w ithin 15 m iles      m ore than 15 m iles
                       O f o,hare             from o’hare
                       ------------------     ------------------------
                       1980          1990     1980          1990
                       ------        ------   -------       -------

P roportion w ater     0.202        -0.870    -1.225       -1.184
                       (0.114)      (0.345)   (3.095)      (3.406)

P roportion rail       1.968        1.552     4.272        3.180
                       (2.898)      (2.033)   (7.288)      (5.450)

P roportion parks      -1.110       -1.113    -0.608       -0.482
A nd open space        (2.382)      (1.103)   (3.490)      (3.057)

D istance to nearest   -0.339       -0.351    -0.126       -0.117
S ubcenter             (6.219)      (4.415)   (7.902)      (7.783)
(absolute asym ptotic t-values are in parentheses)
                     Conclusion
• Correlation between errors of employment density and
  employment probability exists ---OLS estimates are
  subject to selection bias -- either two-stage method or
  maximum likelihood is appropriate.
• Transportation facilities are subject to economies of scale.
  Firms will cluster near transportation facilities even if there
  are no direct benefits of locating near one another. The
  empirical results show that the measures of access to the
  transportation system are highly statistically significant
  determinants of both employment probability and
  employment density.
          Thank you

• Thank you for listening