ECONOMETRIC ANALYSIS OF PUBLIC PARKING PRICE ELASTICITY IN EUGENE, OREGON

ECONOMETRIC  ANALYSIS  OF  PUBLIC  PARKING  PRICE   ELASTICITY  IN  EUGENE,  OREGON                 by   Moshe  Farber  &  Erin  Weld   University  o...
Author: Horace Charles
46 downloads 1 Views 640KB Size
ECONOMETRIC  ANALYSIS  OF  PUBLIC  PARKING  PRICE   ELASTICITY  IN  EUGENE,  OREGON                

by   Moshe  Farber  &  Erin  Weld   University  of  Oregon   June  13th,  2013               ABSTRACT  

This  paper  explores  the  demand  sensitivity  to  price  changes  of  on  campus  municipally   operated  parking  infrastructure  in  Eugene,  Oregon  ex  ante  and  ex  post  of  a  13%  campus-­‐wide   price  increase  in  2012.  We  find  that  elasticity  of  demand  is  relatively  inelastic  at  -­‐0.3,  and  that  a   further  17%  increase  of  price  will  net  the  city  an  additional  $137,230.40  per  year  across  all   campus  zones.  This  paper  additionally  looks  at  the  effects  of  distance,  time,  and  various  other   demand  shifters  on  total  revenue.      

               

                Presented  to  the  Department  of  Economics,  University  of  Oregon,     in  partial  fulfillment  of  the  requirements  of  honors  in  Economics    

Farber  &  Weld    

  [This  Page  is  Intentionally  Left  Blank]  

1  

Farber  &  Weld    

 

2  

Table  of  Contents:    

I.  Introduction..................................................................................................................................4   II.  Literature  Review.....................................................................................................................5     III.  Methodology...........................................................................................................................10     A.  Research  Questions..................................................................................................10     B.  Empirical  Model.........................................................................................................11     C.  Variable  Description.................................................................................................12     D.  Data  Description........................................................................................................13     E.  Model  Description.....................................................................................................16   IV.  Econometric  Results............................................................................................................18     A.  Price.................................................................................................................................18     B.  Distance..........................................................................................................................19     C.  Distance  Interactions...............................................................................................21     D.  Time.................................................................................................................................23     E.  Student  Enrollment...................................................................................................25     F.  Further  Analysis.........................................................................................................26     G.  Conditional  Forecast................................................................................................27   V.  Conclusion.................................................................................................................................28   VI.  Works  Cited.............................................................................................................................30   Appendix  A.....................................................................................................................................32   Appendix  B.....................................................................................................................................35  

   

                                   

Farber  &  Weld    

  [This  Page  is  Intentionally  Left  Blank]  

 

3  

Farber  &  Weld    

 

4  

I.  Introduction    

University  parking  is  an  issue  that  affects  college  campuses  and  cities  across  the   nation.  Whether  it  is  sufficient  parking  availability,  enforcement  of  parking  rules  and   regulations,  or  the  nominal  price  of  parking,  many  different  factors  influence  the  efficiency   and  accessibility  of  university  parking.  Because  of  the  higher  than  normal  concentration  of   students  and  faculty  residing  on  and  around  campus,  as  well  as  other  various  commercial   operations  focused  on  serving  the  university  population,  parking  on  and  around  university   campuses  is  typically  in  higher  demand  in  comparison  to  many  other  locations.1  Although   there  are  grounds  to  argue  parking  as  either  a  public  good  or  a  marketable  commodity,  it  is   important  not  to  overlook  the  immense  value  they  hold.2  However,  due  to  the  many   different  factors  that  affect  parking  on  a  university  campus,  determining  the  specific  value   of  an  individual  parking  structure  or  space  is  often  a  difficult  task.  This  paper  will  explore   the  value  of  on-­‐campus  public  parking  in  Eugene,  Oregon—more  specifically  the  sensitivity   of  demand  to  price  changes  and  its  effect  on  total  revenue.   Providing  the  necessary  resources  for  commuters  to  conveniently  park  at  their   desired  destinations  typically  falls  under  the  purview  of  the  city,  with  the  aim  to  stimulate   local  business  and  generate  additional  income.  In  the  case  of  the  University  of  Oregon,  the   city  of  Eugene  and  the  University  of  Oregon  through  their  Department  of  Parking  and   Transportation  have  implemented  a  network  of  parking  structures  in  and  around  campus   in  an  effort  to  provide  these  necessary  resources.    However,  for  the  purposes  of  this  paper  

                                                                                                                1  (Dagget  &  Gutkowski,  1)   2  (Epstein,  2002)  

Farber  &  Weld    

 

5  

we  will  focus  solely  on  the  parking  infrastructure  under  the  domain  of  City  of  Eugene   Parking.     Eugene  City  Parking  is  faced  with  two  primary  tasks.  Firstly,  like  most  other   government  entities,  they  have  the  responsibility  of  serving  the  public,  which  in  this  case   translates  to  affordable  and  accessible  public  parking.  However,  Eugene  City  Parking  is  also   responsible  for  returning  a  certain  amount  of  revenue  earned  back  to  the  city  each  year.   For  this  reason,  Eugene  City  Parking  is  forced  to  balance  two  opposing  tasks:  maximize   revenue,  while  minimizing  extraneous  costs  to  consumers.  Focusing  on  revenue   maximization,  it  is  imperative  for  the  city  parking  department  to  fully  understand  the   willingness  of  its  customers  to  pay  for  the  service  they  provide.  By  analyzing  the  changes  in   total  revenue  ex  ante  and  ex  post  of  a  13.33%  increase  in  price  from  $1.50/hr.  to  $1.70/hr.   implemented  in  the  fall  of  2012,  we  are  able  to  provide  Eugene  City  Parking  with  a   comprehensive  view  of  on-­‐campus  parking’s  sensitivity  to  price  changes  and  its   implications  for  total  revenue.  Additionally  this  paper  will  examine  different  aspects  of   elasticities  as  applied  to  parking  on  campus,  such  as  distance,  time,  and  population  effects,   etc.  Finally,  this  paper  will  investigate  the  effect  a  simulated  price  increase  would  have  on   total  revenue.    

II.  Literature  Review    

There  are  numerous  studies  available  that  look  at  the  issue  of  city  parking.  Although  

none  of  them  specifically  address  price  elasticity  at  a  university  campus,  they  do  offer   insight  into  the  important  features  that  contribute  to  public  parking  demand.  Litman   (2013)  discusses  transport  demand  elasticities  as  a  whole,  including  the  effects  of  parking   fees  on  demand,  by  using  case  studies  from  around  the  world.  The  report  analyzes  various  

Farber  &  Weld    

 

6  

studies  of  parking  price  elasticity  and  concludes  that  nominal  increases  in  the  price  of   parking  generally  decreases  parking  demand  due  to  its  direct  cost  to  motorists.  It   particularly  looks  at  the  effects  of  price  changes  in  parking  on  drivers  behavior,  i.e.   switching  to  public  transportation  or  forgoing  solo-­‐travel  in  favor  of  carpooling.  However,   the  study  is  complied  by  an  urban  planner  with  no  formal  economic  background,  and   therefore  takes  the  issue  on  in  a  more  public  policy/urban  planning  setting  with  little   emphasis  on  economic  implications  or  methodology.     Stepping  away  from  the  public  policy  view,  Clinch  &  Kelly  (2003)  looked  specifically   at  parking  elasticities  in  downtown  Dublin,  Ireland  ex  ante  and  ex  post  of  a  citywide  50%   increase  in  parking  prices.  It  specifically  looks  at  the  price  elasticity  of  demand  based  on   average  parking  levels  for  specific  periods  of  time  during  5-­‐  and  6-­‐day  intervals.  The   authors  found  an  aggregate  4.18%  drop  in  parking  activity  and  16.5%  percent  drop  in   duration  of  stay  as  a  result  of  the  price  increase.  Nevertheless,  they  also  found  that  the   price  increase  raised  revenues  by  21%,  and  that  even  after  the  increase,  parking  prices   were  still  not  maximizing  revenue.  They  further  showed  that  the  price  elasticity  of  demand   averages  -­‐0.29,  pointing  to  relative  inelasticity.       A  similar  study,  Miller  &  Everett  (1982),  examined  the  effects  of  a  price  increase  and   parking-­‐subsidy  decrease  in  the  Washington  D.C.  metro  area  in  1982.  The  paper  uses  a   before  and  after  control  group  structure  to  show  that  price  increases  lead  to  a  sharp   increase  in  higher-­‐occupancy  modes  of  transportation,  though  the  results  were  not  fully   uniform.  Though  this  and  Clinch  &  Kelly  (2003)  provide  important  information  about  city   parking,  both  were  focused  on  larger  cities,  which  may  lead  to  results  that  vary  drastically   from  a  smaller  urban  region  such  as  Eugene.    

Farber  &  Weld    

 

7  

At  the  aggregate  level,  Concas  &  Nayak  (2012)  analyzed  the  variation  in  results   across  past  literature  regarding  parking  elasticities.  They  found  that  results  vary  across   different  regions,  data  collection  methods,  and  techniques  of  estimation,  also  stating  that   elasticities  are  site-­‐specific  and  vary  based  on  policy.  They  finally,  arrived  at  a  baseline   elasticity  of  -­‐0.39,  consistent  with  previous  estimates.   Specifically  examining  the  University  of  Oregon,  Barrar  (1978)  applies  a   multinomial  logit  model  to  explain  an  individual’s  transportation  choices  when  they   commute  to  and  from  campus.  The  study  focuses  on  the  utility  gained  by  the  different   modes  of  transportation,  as  individuals  act  to  maximize  their  utility  at  the  margin,  stating   that  individuals,  “derive  utility  from  the  attributes  of  each  commodity  rather  than  from  the   commodity  itself.“3  The  reason  the  majority  of  individuals  choose  to  use  a  personal  vehicle   is  due  to  the  comfort  and  convenience  they  provide,  therefore  deriving  the  most  utility   from  personal  transportation.  Barrar  proposed  a  utility  function  that  he  believed  explained   an  individuals  decision  making  process  when  traveling,  which  includes  such  variables  as   purpose  of  travel,  mode  of  travel,  frequency  of  travel,  time  of  day,  destination  of  travel,   route  of  travel,  vehicle  ownership,  work  and  residential  location.  The  multinomial  logit   model  describes  an  individual’s  choice  probability  for  exclusive  alternatives.  However,   “...because  alternatives  are  mutually  exclusive,  the  choice  of  one  alternative  precludes  the   selection  of  any  other.  Alternatives  are  ‘competitive’  in  this  sense…within  the  MNL   framework  there  is  no  pattern  of  differential  substitution  among  the  various  [competitive]   alternatives;  there  is  always  equal  proportional  substitution.”3  It  continues  to  show  that   this  creates  a  bias  because  alternatives  are  viewed  as  ‘similar’  instead  of  opposite,  and,  “if                                                                                                                   3  (Barrar,  30-­‐31)  

Farber  &  Weld    

 

8  

alternatives  are  perceived  as  ‘similar’,  the  definition  of  the  number  of  available  alternatives   becomes  obscured  resulting  in  the  observed  bias.”4    

Again  looking  at  the  university  level,  Lipscomb  &  Kofor  (2011)  directly  explores  

university  parking  from  the  perspective  of  a  university  operated  parking  program.  The   study  clearly  differentiates  students  from  faculty  and  staff,  finding  that  the  two  groups  have   different  willingness’  to  pay  for  on-­‐campus  parking  passes.  The  authors  were  able  to   determine  the  mean  willingness  to  pay  and  found  that  students  have  a  higher  willingness   to  pay  than  faculty/staff,  and  that  students  also  have  a  higher  price  change  acceptance  rate,   for  reasons  other  than  income.  These  findings  are  useful  when  analyzing  the  effects  parking   price  increases  have  on  the  separate  groups,  faculty  and  students.     Integrated  Urban  Traffic  Management  (1978)  explores  the  different  characteristics   of  urban  traffic  management,  breaking  them  down  into  eight  separate  groups:  time/place   of  trip,  choice  of  mode  of  transportation,  choice  of  route,  street  usage  allocation,  junction   control  (signals),  parking  control,  safety,  and  environmental  problems.5  The  element  most   applicable  to  our  research  is  parking  control,  consisting  of  parking  supply,  fee  schedules,   and  time  restrictions.  The  paper  outlines  that  efficient  parking  is  two-­‐fold:  It  first  implies   uncongested  roads  and  parking  structures.  Second,  it  means  those  who  are  willing  to  pay   receive  the  parking.  Furthermore,  “The  motivation  for  road  pricing  is  rooted  in  the   economic  idea  of  efficiency.  The  transport  system  will  be  used  efficiently  only  if  the   potential  trip  maker  perceives  the  full  cost  of  his  journey”.6  The  paper  also  discusses  how  

                                                                                                                4  (Barrar,  39)   5  (Integrated  Urban  Traffic  Management,  15)   6  (Integrated  Urban  Traffic  Management,  31)  

Farber  &  Weld    

 

9  

choices  of  mode  of  transportation  also  incorporate  alternatives  and  attractiveness  of   particular  methods,  consistent  with  the  findings  in  Barrar  (1978).       The  Transit  Cooperative  Research  Program  conducted  by  Federal  Transit   Administration  and  Transportation  Research  Board  (2005)  also  provided  useful  insights.   Parking  Prices  and  Fees  section  provides  invaluable  information  on  elasticity  rates  based  on   rate  changes,  location,  and  available  substitutions.    It  analyzes  the  effect  that  price   increases  have  on  individuals’  parking  frequency,  duration,  and  willingness  to  seek   alternative  modes  of  transportation  (carpool,  public  transportation,  walk,  etc.).    The  study   also  takes  into  consideration  an  individual’s  income  and  value  of  their  time,  finding  that  as   a  person’s  income  rises,  their  sensitivity  to  price  changes  decreases,  which  may  offer  useful   insight  for  University  of  Oregon  parking.    The  general  consensus  of  this  study  is  that  the   elasticity  rate  falls  between  -­‐.1  and  -­‐.6  depending  on  multiple  factors,  but  the  average  rate   tends  to  be  approximately  -­‐.3,  denoting  inelastic  demand.  Though  not  specifically  directed   toward  university  parking,  this  is  a  useful  source  that  provides  a  helpful  starting  point  for   our  research.     The  key  to  determining  whether  these  previous  case  studies  can  be  compared  to  the   University  of  Oregon  accurately  is  through  a  comprehensive  understanding  of  the  unique   aspects  of  the  University  of  Oregon  parking  system.  In  order  to  present  credible  predictions   for  the  future,  we  must  understand  the  trends  of  this  specific  scenario.  The  Role  of   Economic  Studies  in  Urban  Transportation  Planning  by  the  U.S.  Bureau  of  Public  Roads   differentiates  between  “projecting”  and  “forecasting”  for  future  changes  in  transportation   demand.7  Projecting  implies  simply  plugging  in  known  numbers  into  an  equation.                                                                                                                   7  Meck  (1965)  

Farber  &  Weld    

 

10  

Forecasting,  on  the  other  hand,  is  using  outside  information  to  create  a  more  realistic  look   into  the  future.  In  order  to  correctly  forecast,  one  must  be  knowledgeable  of  the  trends  of   transportation,  and  separate  correlation  between  variables  and  true  cause  and  effect.8   Applying  this  method  to  our  forecast,  we  can  provide  a  more  accurate  view  of  the  parking   landscape      

After  surveying  the  existing  literature  four  things  are  clearly  evident:  First,  parking  

demand  is  in  general  relatively  price  inelastic,  and  although  increasing  the  price  of  parking   may  negatively  affect  quantity  of  demand,  parking  revenue  tends  to  increase,  though  the   extent  of  these  depend  greatly  on  the  situation.  Secondly,  an  individual’s  utility  function   directly  determines  what  mode  of  transportation  they  will  choose  to  take.  Third,  there  are   multiple  substitutes  of  driving,  and  increases  in  parking  prices  tend  to  lead  to  a  higher  rate   of  substitution.  Lastly,  university  faculty/staff  members  and  students  respond  differently  to   price  changes  on  campus,  which  may  in  turn  influence  the  migration  to  alternative  parking   options,  in  this  case  city  maintained  meter  parking.  Utilizing  the  information  and  findings   provided  in  these  studies  as  a  building  block  and  applying  it  to  our  own  data,  we  hope  to   provide  Eugene  City  Parking  with  a  comprehensive  model  showing  the  price  elasticity  of   University  of  Oregon  campus  area  parking.    

III.  Methodology   A. Research  Questions   Through  this  paper,  we  wish  to  examine  three  main  questions  regarding  the  price   elasticity  of  public  parking  on  campus:  

                                                                                                                8  (Meck,  20)  

Farber  &  Weld     •

 

11  

Determine  the  price  elasticity  of  demand  for  public  parking  on  and  around  the   University  of  Oregon  campus.  Focus  on  the  application  to  total  revenue,  specifying   effects  price  changes  have  on  total  revenue  while  controlling  for  exogenous   variables.    



Examine  marginal  effects  of  distance,  time,  and  population  effects,  as  well  as  other   possible  demand  shifters.    



Perform  conditional  forecast  of  price  increase  from  current  price  of  $1.70/hr.  to   $2.00/hr.   B. Empirical  Model      

The  value  of  a  parking  zone  i  at  time  t  is  defined  by  the  relationship:  

 

𝑇𝑜𝑡𝑎𝑙𝑅𝑒𝑣!" = 𝛽! + 𝛽! 𝑃𝑟𝑖𝑐𝑒!" + 𝛽! 𝑀𝑎𝑗𝐸𝑣𝑒𝑛𝑡! + 𝛽! 𝑍𝑜𝑛𝑒! + 𝛽! 𝑇𝑖𝑚𝑒! + 𝛽! 𝑆𝑒𝑎𝑠𝑜𝑛! + 𝛽! 𝑆𝑡𝑢𝑑𝑒𝑛𝑡𝐸𝑛𝑟𝑜𝑙𝑙𝑚𝑒𝑛𝑡! + 𝛽! 𝐷𝑎𝑦𝑜𝑓𝑊𝑒𝑒𝑘! + 𝛽! 𝐹𝑖𝑛𝑎𝑙𝑠! + 𝛽!" 𝐵𝑟𝑒𝑎𝑘! + 𝜀!  

where  𝜀!  is  our  error  term,  i  represents  a  particular  zone,  and  t  represents  a  one-­‐hour   period.  The  following  table  gives  a  description  for  all  variables:   Table  1.   Variable   Priceit   MajEventi   Zonei   Timet   StudentEnrollmentt   Seasont   DayofWeekt   Finalst  

Description  of  Variables   Explanation   Price  of  parking  per  hour  at  time  t   Dummy  variable  indicating  occurrence  of  major   event  on  campus   (=1  if  yes,  =0  if  otherwise)   Zone  observation  took  place  in   Variable  denoting  time  of  day  revenue  was   recorded   University  population  at  time  t     Dummy  variable  indicating  observation  was   recorded  during  fall   (=1  if  yes,  =0  if  otherwise)   Dummy  variable  indicating  if  observation   occurred  on  weekday,  Saturday,  or  Sunday   (=1  if  yes,  =0  otherwise)   Dummy  variable  indicating  observation  occurred   during  finals  week   (=1  if  yes,  =0  otherwise)  

Farber  &  Weld     Breakt  

 

 

12  

Dummy  variable  indicating  observation  occurred   during  school  break   (=1  if  yes,  =0  otherwise)  

C. Variable  Description   Based  on  our  hypothesis  and  previous  research,  we  predict  certain  outcomes  for  

each  individual  explanatory  variable,  controlling  for  distance,  time  and  population  factors;   demand  shifters;  and  monetary  costs  associated  with  parking  described  below:   Price:   As  Litman  (2005)  and  Concas  &  Nyak  (2012)  found,  we  expect  price  to  have  an   inverse  relationship  with  the  demand  for  a  given  parking  spot,  with  most  literature   pointing  to  an  average  price  elasticity  hovering  around  -­‐0.3.   Major  Event:   We  expect  that  the  observation  of  a  major  event,  such  as  a  track  meet,  football  game,   or  basketball  game  on  a  given  day  will  share  a  direct  relationship  with  demand  for  on   campus  parking,  increasing  the  demand  for  parking.     Zone:   Representing  the  zone  which  an  observation  took  place,  we  are  able  to  control  for   distance  effects.  We  expect  that  as  the  distance  increases,  the  demand  for  that  zone  will   decrease  with  it.     Time  of  Day:   Based  on  the  results  of  Kelly  &  Clinch  (2005),  we  expect  a  similar  result  where   parking  demand  is  highest  in  the  morning,  as  students  and  faculty  arrive  for  morning   classes  or  to  take  advantage  of  the  flat  fee  for  all-­‐day  parking,  tapering  off  throughout  the   afternoon  and  evening.    

Farber  &  Weld    

 

13  

University  Population:   We  expect  demand  for  parking  to  be  highly  correlated  with  university  population,  as   increases  in  student  populations  brings  increases  in  cars,  therefore  increasing  the  demand   for  parking.   Season  (Fall,  Winter,  Spring,  Summer):   We  expect  demand  to  be  at  its  highest  during  fall,  winter  and  spring,  before  tapering   off  in  summer  as  many  students  return  home  or  engage  in  other  activates  outside  of  the   Eugene  area.     Day  of  Week:   We  expect  demand  to  be  highest  during  the  week,  when  class  is  in  session,  and   substantially  lower  on  Saturday  and  Sundays  as  the  traffic  to  and  from  campus  decreases.   Finals:   We  expect  demand  to  decrease  during  finals  week,  as  there  is  no  class,  and  students   gradually  leave  the  Eugene  area  to  return  home  for  breaks  throughout  the  week,  causing   the  majority  of  demand  for  on-­‐campus  parking  to  temporarily  dissipate.   Break:    

Again,  we  expect  that  due  to  the  significant  decrease  in  local  population  and  need  to  

be  on  campus,  we  expect  total  revenue  to  fall  during  breaks  from  school  (spring,  summer,   and  winter  breaks).  

D. Data  Description:   The  main  source  of  data  is  the  city  of  Eugene’s  IPS  system,  from  which  total  revenue   is  aggregated  into  hourly  revenue  summaries  containing  317,856  individual  observations   across  a  32-­‐month  period  from  September  2010  to  April  2013.  It  is  important  to  note  that  

Farber  &  Weld    

 

14   the  meters  only  operate  for  13  hours  a  

Figure 1 Total Revenue Across Time 30

day  from  07:00  to  20:00,  and  therefore  

Mean of Total Revenue 10 20

observations  outside  the  hours  of   operation  take  on  zero-­‐values,  which  will   be  discussed  further  in  next  section.  It  is  

0: 0 1: 0 0 2: 0 0 3: 0 0 4: 0 0 5: 0 0 6: 0 0 7: 0 0 8: 0 0 9: 0 10 00 : 11 00 : 12 0 0 : 13 00 : 14 00 : 15 00 : 16 00 : 17 00 : 18 00 : 19 00 : 20 00 : 21 00 : 22 00 : 23 00 :0 0

0

also  of  note  that  all  meters  on-­‐campus   Hour of Day

Figure 2

operated  by  the  city  accept  both  coins   and  credit  cards,  therefore  mitigating  a  

40

Total Revenue Across Zones

Mean of Total Revenue 10 20 30

reliance  on  those  who  carry  change  for   revenue,  and  making  parking  much  more   accessible.  Figure  1  shows  the  

th

ily

ar Ki d nc a O id rc ha Pa rd tte rs on Vi W llar d al nu tS tr

Zone

H

th

E. 18

th

E. 17

th

E. 15

th

E. 14

th

E. 13

r

E. 12

de Al

Ag

at e

0

distribution  of  total  revenue  across  the   24  hours  and  Table  2  further  details  total   revenue  over  time:     Table  2.   Summary  Statistics  (Time)   Time   1:00   2:00   3:00   4:00   5:00   6:00   7:00   8:00   9:00   10:00   11:00   12:00   13:00   14:00   15:00   16:00  

Obs.   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244   13244  

Mean   0.0028   0.0011   0.0004   0.0006   0.0814   1.5829   16.5569   27.5157   26.5339   20.0357   24.0352   21.9127   23.3585   16.8509   17.1826   12.5706  

Std.  Dev.   Min.  Revenue/Hour   Max.  Revenue/Hour   0   2.25   0.0482   0   1.35   0.0269   0   1.25   0.0189   0   1.25   0.0212   0   35   0.9303   0   94.1   4.8545   0   317.95   32.5260   0   283.4   46.0031   0   405.35   41.3788   0   287.35   29.4135   0   293.9   34.6601   0   267.35   29.5912   0   241.4   33.2191   0   283.7   24.0312   0   264.1   26.6566   0   245.05   19.1682  

Farber  &  Weld     17:00   18:00   19:00   20:00   21:00   22:00   23:00  

  13244   12.2999   19.3306   13244   7.8215   12.8225   13244   2.3970   3.9319   13244   0.0406   0.1732   13244   0.0182   0.1129   13244   0.0101   0.0892   13244   0.0069   0.0716  

15   0   0   0   0   0   0   0  

177.85   132.7   48.35   4.1   2.8   3.25   1.8  

Observations  are  additionally  separated  into  districts  and  zones.  This  paper  will  focus   specifically  on  the  campus  district,  which  is  further  partitioned  into  14  individual  zones:   Agate,  Alder,  E.  12th,  E.  13th,  E.  14th,  E.  15th,  E.  17th,  E.  18th,  Hilyard,  Kincaid,  Orchard,   Patterson,  Villard,  and  Walnut  St.  Figure  2  above  shows  the  distribution  of  revenue  over   zones  and  Table  3  below  summarizes  the  data  by  zone:     Table  3.   Summary  Statistics  (Zones)   Zone   Mean   Std.  Dev.   Min.   Max.   Num.  of   Revenue/Hour   Revenue/Hour   Meters   Agate  St.   13.00745   24.0705   0.00   293.40   75   Alder  St.   4.95858   12.00307   0.00   104.80   10   E.  12th  Ave.   8.391354   12.377443   0.00   71.40   22   E.  13th  Ave.   16.18966   24.29274   0.00   272.30   52   E.  14th  Ave.   14.06947   21.58082   0.00   165.40   37   E.  15th  Ave.   23.00794   40.05337   0.00   405.35   109   E.  17th  Ave.   2.174943   7.237046   0.00   161.35   25   E.  18th  Ave.   10.96401   22.88102   0.00   303.65   71   Hilyard  St.   1.521644   2.560688   0.00   15.80   9   Kincaid  St.   33.2851   49.3523   0.00   317.95   102   Orchard  St.     1.952621   7.49574   0.00   163.40   29   Patterson  St.   0.318   0.8893352   0.00   9.40   4   Walnut  St.   4.346738   10.57665   0.00   120.90   37   Villard  St.   0.457622   4.260945   0.00   263.35   21   For  the  purposes  of  this  paper,  we  assume  that  meters  within  the  same  zone  are  valued   equally,  and  therefore  total  revenue  data  is  aggregated  by  zone  instead  of  examining   meters  at  the  individual  level.  The  remaining  exogenous  variables  were  sourced  externally.   Student  population  and  events  for  major  events  at  the  university  were  taken  from  the   publically  available  data  posted  on  the  university  and  athletic  department  websites.  Finally,  

Farber  &  Weld    

 

16  

distance  values  were  calculated  by  hand  using  midpoints  of  each  zone  from  maps  and  data   provided  by  the  city  of  Eugene.     The  data  exhibits  some  minor  problems  that  required  correction.  During  the   summer  of  2011,  a  major  construction  overhaul  altered  the  parking  infrastructure  of  13th   and  Alder,  a  frequented  parking  destination  on  the  west  edge  of  campus.  This  included   changing  one  block  of  the  E.  13th  zone  from  a  typical  individual  coin  meter  system  and   installing  a  multi-­‐space  pay  station.  For  this  reason,  we  have  opted  to  exclude  the  block   from  our  data,  but  due  to  this  area  being  only  a  small  portion  of  total  revenue  for  the   campus  district  as  a  whole,  we  do  not  expect  the  exclusion  to  alter  our  results.  With   respects  to  distance,  three  zones,  E.  13th,  E.  15th,  and  E.  18th  streets  operate  on  both  east   and  west  sides  of  campus,  leading  to  a  zone-­‐midpoint  that  is  in  a  deceptively  central   location,  and  generating  distance  values  that  are  downward  biased.  Because  of  the   unreliability  of  these  distance  values,  the  three  zones  have  been  combined  together  and   utilized  as  the  reference  category  not  included  in  the  regression.  This  phenomenon  will  be   further  discussed  in  later  sections.  Finally,  within  the  date  range  of  our  study,  only  one   price  increase  was  implemented,  amounting  to  only  a  $0.20  increase  from  $1.50/hr.  to   $1.70/hr.  Therefore,  the  specific  demand  curve  for  parking  on  campus  and  its  associated   elasticity  will  be  difficult  to  determine.   E. Model  Description   Building  on  the  empirical  model  described  in  section  B  above,  we  found  it  necessary   to  alter  the  structure  and  implementation  of  our  model  to  better  fit  our  data.  Noting  that  of   the  317,856  observations  for  total  revenue,  62.25%  are  zero-­‐value  observations,  we  found   that  this  high  of  a  zero-­‐value  proportion  was  unsuitable  for  OLS  regression.  Figure  3  shows  

Farber  &  Weld    

 

17   the  distribution  of  actual  total  revenue  

Figure 3 Distribtuion of Total Revenue

.08

.1

values,  and  zero-­‐values  are  again  seen  to  

Density .06

occupy  a  large  proportion  of  the  data.  

.04

For  this  reason,  we  opted  to  use  the  

0

.02

Tobit  model,  designed  to  mitigate  large   0

100

200 Total Revenue

300

400

Actual Values

zero-­‐biases  outlined  in  Tobin  (1958),  in   lieu  of  the  unreliable  OLS  model.  The  

Figure 4

Tobit  model  assumes  that  both  a  latent  

.008

Distribution of Total Revenue

.006

and  observed  variable  exist,  which  are  

Density .004

equal  for  positive  values,  but  the  

.002

observed  variable  takes  a  zero  for  all  

0

negative  values,  referred  to  as  censoring   -300

-200

-100 Total Revenue

0

100

from  the  left.  After  applying  the  Tobit  

Fitted Values, Tobit Model

model  to  our  original  regression  model,   we  find  that  the  fitted  values  are  distributed  much  more  normally,  as  seen  above  in  Figure   4.      Further  examination  shows  that,  according  to  the  findings  in  Amemiya  (1984),  if  the   Tobit  model  is  subject  to  heteroskedasticity  or  has  a  non-­‐normally  distributed  residual,  it   yields  inconsistent    estimates.  Using  the  tobcm  command  in  Stata  as  detailed  in  Drukker   (2002),  a  test  for  normally  distributed  residuals  in  Tobit  regression  models,  we  find  that   𝑃 > 𝜒 ! = 0.00,  unfortunately  rejecting  the  null  hypothesis  of  normally  distributed   residuals.  However,  using  the  comparison  model  shown  in  Greene  (1981),  where  OLS   models  converge  to  𝜌𝛽!"#$% ,  where  𝜌  is  the  percentage  of  uncensored  data  within  the   sample,  we  find  that  the  results  from  our  Tobit  model  are  consistent  with  the  OLS  model,  

Farber  &  Weld    

 

18   and  further  confirming  the  normality  of   our  residuals  with  the  Q-­‐Q  residual  plot   shown  in  Figure  5  above.  Finally,  to   correct  for  heteroskedasticity,  all   regressions  were  run  using  White’s   robust  standard  errors  detailed  in  White   (1980).    

IV.  Econometric  Results   A. Price     First  applying  the  Tobit  model  to  our  regression  we  find  that  price  increases  of   $1.00  would  bring  in  a  total  $11.71/hr./zone,  or  a  total  of  $163.94/hr.  across  all  zones.  This   is  a  significant  increase  in  light  of  current  total  revenue  per  hour  is  only  $134.68  on   average  across  all  zones.  Comparing  results  from  both  Tobit  and  OLS  models  seen  in  Table   4  below,  we  find  that  OLS  results  in  only  a  mere  $5.33/hr./zone,  totaling  in  only  $74.62/hr.   across  all  zones.  

Variable   Price   Constant   Observations  

Table  4.   Regression  Results   Tobit  Coef.   11.706   (11.71)**   -­‐120.831   317,856  

OLS  Coef.   5.330   (10.46)**   -­‐4.257   317,856  

Robust  Standard  Errors  in  Parentheses   p  |t|   [95%  Conf.  Interval]   DistW   0.0031569   0.000201   15.71   0   0.002763   0.0035508   DistC   -­‐0.019082   0.0003169   -­‐60.21   0   -­‐0.019703   -­‐0.01846   DistE   0.0069074   0.0001398   49.41   0   0.0066334   0.0071814   Standard  Errors  are  Robust  

Farber  &  Weld    

 

21  

It  is  crucial  to  note  that  distance  is  not  the  only  determining  factor  of  a  zone’s  revenue.   Each  zone  possesses  a  varying  quantity  of  meters;  those  with  a  higher  quantity  of  meters   are  in  turn  able  to  accommodate  more  vehicles,  and  therefore  generate  more  revenue  per   hour  than  other  zones  with  fewer  spots  [See  Table  2].    While  this  is  controlled  for  by  zone   fixed  effects  in  the  above  regressions  using  Zonenumi,  it  is  not  accounted  for  in  the  distance   specific  regression.  For  this  reason,  we  felt  it  necessary  to  include  a  variable  to  account  for   the  quantity  of  meters  in  a  given  zone,  Metersi,  therefore  controlling  for  the  variation   caused.  The  coefficients  presented  in  the  above  table  represent  the  change  in  hourly   revenue  per  zone  as  they  increase  one  foot  away  from  the  points  in  east,  west,  and  central   campus.  These  results  curiously  show  that  as  a  zone  increases  in  distance  from  east  and   west  campus,  their  hourly  revenue  increases.  Conversely,  increasing  the  distance  from   central  campus  causes  revenue  to  decrease.    These  results  also  suggest  that  the  smaller  the   distance  between  a  zone  and  central  campus,  the  higher  its  hourly  revenue.  In  terms  of   daily  parking  on  days  there  is  no  major  event  on  campus,  the  results  align  with  the  notion   that  commuters  prefer  to  park  closer  to  central  campus,  which  captures  the  majority  of   educational  buildings  on  campus.    

C. Distance-­‐Event  Interactions   Distance  from  particular  campus  locations  is  not  solely  a  concern  for  the  daily   commuter.  It  is  also  a  major  factor  for  people  driving  to  the  university  district  for  particular   events,  more  specifically  sporting  events.  To  test  the  effect  distance  had  on  hourly  revenue   during  days  with  sporting  events,  we  created  interactive  variables  for  basketball  games,   football  games,  and  track  meets  with  the  three  distance  variables—east,  west,  and  central.   The  effects  are  shown  below  in  Table  7:  

Farber  &  Weld    

Total  Rev.   DistW   DistC   DistE   DistEFootball   DistWFootball   DistCFootball   DistEBasketball   DistWBasketball   DistCBasketball   DistCTrackmeet   DistWTrackmeet   DistETrackmeet  

 

Coef.   0.0030361   -­‐0.0195583   0.0072427   0.0000452   0.005648   -­‐0.0064784   -­‐0.0011992   0.0033146   0.002385   0.0149529   -­‐0.0049228   -­‐0.0079638  

Table  7.   Regression  Results   Std.  Error   t   0.0002096   14.49   0.0003313   -­‐59.04   0.0001448   50   0.0009323   0.05   0.0015064   3.75   0.0023418   -­‐2.77   0.0006656   -­‐1.8   0.0010284   3.22   0.0016583   1.44   0.0017761   8.42   0.001119   -­‐4.4   0.0006902   -­‐11.54  

P>  |t|   0.000   0.000   0.000   0.961   0.000   0.006   0.072   0.001   0.150   0.000   0.000   0.000   Standard  Errors  are  Robust  

22  

[95%  Conf.  Interval]   0.0026254   0.0034469   -­‐0.0202076   -­‐0.018909   0.0069588   0.0075265   -­‐0.0017822   0.0018725   0.0026954   0.0086005   -­‐0.0110683   -­‐0.0018885   -­‐0.0025037   0.0001054   0.001299   0.0053302   -­‐0.0008652   0.0056352   0.0114717   0.018434   -­‐0.007116   -­‐0.0027296   -­‐0.0093166   -­‐0.0066111  

The  results  show  that  all  interactive  variables  were  statistically  significant  with  the   exception  of  those  representing  the  eastern  campus  variable  interacted  with  football   games,  central  campus  interacted  with  basketball  games,  and  central  campus  interacted   with  basketball  games.  This  shows  that  distance  from  east  campus  is  not  a  deciding  factor   for  football  and  basketball  game  commuters,  and  distance  from  central  campus  is  not  a   factor  for  basketball  game  commuters.  This  can  be  attributed  to  drivers’  higher  willingness   to  accept  larger  distances  when  parking  for  such  events  as  sporting  events,  especially  since   on-­‐campus  parking  is  usually  experiencing  busier  than  average  use  at  those  times.  The   most  curious  result  from  this  is  the  variable  denoting  the  interaction  between  east  campus   and  basketball  games.  With  Matthew  Knight  Arena,  the  location  of  the  basketball  games,   acting  as  the  point  of  interest  for  east  campus,  the  expectation  would  be  that  the  distance  of   a  zone  to  where  the  basketball  game  is  held  would  have  a  significant  influence  on  demand   and  hourly  revenue,  but  the  results  suggest  otherwise.  

Farber  &  Weld    

 

23  

Other  noteworthy  results  from  this  regression  include  the  track  meet  variables.   Hayward  Field,  where  University  of  Oregon  track  meets  are  held,  is  located  roughly  a  block   away  from  both  the  central  and  east  campus  point.  Our  results  show  that  as  a  zone’s   distance  from  east  and  west  campus  increases  on  track  meet  days,  its  hourly  revenue   decreases.  On  the  other  hand,  distance  from  central  campus  has  a  significantly  positive   effect  on  hourly  revenue.  This  can  be  interpreted  as  commuters  not  wanting  to  park  near   central  campus,  opting  instead  to  park  on  the  outskirts  of  University  District  closer  to   Hayward  Field.    

D. Time   Considering  hourly  revenue  is  determined  by  the  demand  for  parking  at  a  given   time,  it  is  important  to  understand  how  time  relates  to  demand,  and  in  turn  how  total   revenue  changes  over  time.  Because  commuters  to  the  campus  district  are  primarily   students  and  faculty,  we  expect  there  is  a  reliable  pattern  to  traffic  patterns  to  and  from   campus  that  affect  parking  use.  We  anticipate  that  hourly  revenue  would  be  at  its  highest  in   the  early  to  mid-­‐morning  period,  as  people  arrive  early  to  take  advantage  either  the   $8.50/5  hour  or  $12.00/10  hour  deals  offered,  before  tapering  off  as  the  day  goes  on.  The   installation  of  meters  that  accept  both  credit  cards  and  coins  makes  these  discounted  deals   much  more  accessible  to  the  general  population,  as  large  quantities  of  coins  are  no  longer   necessary  to  feed  the  meter.  By  generating  dummy  variables  for  each  hour,  we  see  that   hourly  revenue  is  at  its  highest  between  8:00am  to  1:00pm,  consistent  with  our  hypothesis.   Furthermore,  as  the  day  progresses,  revenue  per  hour/zone  decreases,  with  a  distinctive   decrease  at  8:00pm  when  meters  end  operation.  Table  8  below  displays  the  regression   results:  

Farber  &  Weld    

 

Time   1:00   2:00   3:00   4:00   5:00   6:00   7:00   8:00   9:00   10:00   11:00   12:00   13:00   14:00   15:00   16:00   17:00   18:00   19:00   20:00   21:00   22:00   23:00  

Table  8.   Regression  results   Coef.   Std.  Error   t   -­‐4.648737   2.019054   -­‐2.3   -­‐15.78419   2.520904   -­‐6.26   -­‐28.67553   3.627735   -­‐7.9   -­‐24.73217   3.122558   -­‐7.92   6.89041   1.631757   4.22   48.46292   1.373204   35.29   89.4661   1.411544   63.38   106.4348   1.446544   73.58   106.707   1.421794   75.05   100.0869   1.385921   72.22   105.8886   1.397073   75.79   103.8784   1.383962   75.06   105.7321   1.392731   75.92   97.83315   1.376169   71.09   97.96724   1.382226   70.88   91.75185   1.370575   66.94   90.77356   1.372107   66.16   82.89803   1.366684   60.66   73.15335   1.362034   53.71   32.36169   1.41678   22.84   19.26443   1.489658   12.93   10.63926   1.581744   6.73   4.611357   1.692632   2.72   Standard  Errors  are  Robust  

24  

P>t   0.021   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.006  

Of  course,  we  expect  that  commuters  travel  to  campus  for  non-­‐class  reasons  as  well.  To  test   whether  time  influenced  hourly  revenue  differently  during  days  when  class  is  in  session   and  days  when  class  is  not  in  session,  we  used  variables  to  interact  time  with  weekday,   time  with  Saturday,  and  time  with  Sunday.  As  the  regression  results  in  Table  9  below  show,   hourly  revenue  decreases  at  a  faster  rate  as  time  passes  on  a  weekday  rather  than  during   weekends,  suggesting  that  hourly  revenue  is  higher  earlier  in  the  day  on  weekdays  than  on   weekends:   Table  9.   Regression  Results   Total  Rev.   Coef.   Std.  Error   t   P>  |t|   [95%  Conf.  Interval]   Timeweekday   -­‐2.988949   0.1580454   -­‐18.91   0.000   -­‐3.298714   -­‐2.679185   Timesaturday   -­‐1.118112   0.1621213   -­‐6.9   0.000   -­‐1.435865   -­‐0.8003589   Standard  Errors  are  Robust  

Farber  &  Weld    

 

25  

We  also  expect  that  the  time  of  day  affects  hourly  revenue  differently  based  on  a  zone’s   location.  Using  the  regression  containing  distance  variables,  and  controlling  for  the  number   of  meters  per  zone,  Time*Distance  variables  were  generated  to  test  this  hypothesis.  The   results  show  that  distance  does  have  a  significant  impact  on  how  a  zone’s  hourly  revenue   changes  with  the  time  of  day,  and  that  the  hourly  revenue  of  zones  closer  to  central  campus   are  influenced  more  by  time  than  zones  further  away  shown  in  Table  10:     Table  10.   Regression  Results   Total  Rev.   Coef.   Std.  Error   t   P>  |t|   [95%  Conf.  Interval]   TimeDistW   0.0002181   0.0000558   3.91   0   0.0001088   0.0003274   TimeDistC   0.0008421   0.0000895   9.41   0   0.0006666   0.0010175   TimeDistE   0.0001568   0.0000319   4.92   0   0.0000943   0.0002193   Standard  Errors  are  Robust  

E. Student  Enrollment     Another  influential  factor  in  accurately  forecasting  future  revenue  is  the  changes  in   student  population.  Originally  hypothesizing  that  since  University  of  Oregon  students  make   up  a  large  percentage  of  the  total  demand  for  university  district  parking  services,  a  rise  in   the  number  of  students  may  cause  an  initial  increase  for  demand,  the  data  reveals  the   opposite.  In  fact,  it  appears  that  each  additional  student  attending  the  University  results  in   a  $0.005  decrease  in  total  parking  revenue  per  hour  per  zone.  While  this  effect  may  be   small,  we  expect  the  it  to  increase  when  magnified  over  time  by  the  forecasted  trends   student  population  increases,  shown  in  Figure  6  &  Table  12  in  Appendix  B.  Additionally,   this  inverse  relationship  can  be  explained  by  the  traditional  “Tragedy  of  the  Commons”   dilemma;  as  the  number  of  students  commuting  to  campus  rises,  the  supply  of  parking,   held  constant,  must  be  allocated  amongst  more  people.  Commuters  may  then  become   discouraged  by  the  increased  congestion  and  inability  to  find  parking,  and  resort  to  taking  

Farber  &  Weld    

 

26   substitute  modes  of  transportation,  i.e.  

Figure 6 Student Enrollment

biking,  walking,  public  transportation.9  It   Student Enrollment 20000 22000 24000 26000

is  imperative  to  control  for  this  effect,  as   we  cannot  credibly  predict  the  affect  of   future  price  changes  on  revenue  without   2000

2002

2004

2006

2008

2010 Year

2012

2014

2016

2018

2020

Actual & Forecast

also  accounting  for  the  predicted   increase  in  student  population.  

According  to  Oregon  University  System  (OUS)  2012  Factbook,  the  number  of  students,  and   thusly  the  necessary  faculty  and  support  staff,  is  set  to  steadily  increase  for  the  next  nine   years  as  shown  in  Figure  6.    

F. Further  Analysis    It  is  clear  from  our  regression  that  demand  is  not  constant  across  zones;  those   directly  near  campus,  particularly  Kincaid  St.  and  15th  Ave,  have  a  higher  concentration  of   people  wanting  to  park  in  comparison  to  zones  further  away  from  campus.    The  large   percentage  of  observations  that  have  zero  hourly  revenue,  denoting  no  earned  revenue  for   that  period,  even  during  hours  of  operation  data,  also  shows  that  parking  is  not  being   utilized  to  its  maximum  capacity.  Therefore  it  can  be  inferred  that  revenue  is  also  not  being   maximized.  Wilson  (1992)  researches  the  utilization  adjusted  break-­‐even  fee;  the  concept   that  each  parking  space  has  a  certain  price  that,  when  aggregated  across  all  parking  spaces,   would  result  in  break-­‐even  profit.  However,  this  only  functions  under  the  assumption  that   all  spaces  are  100%  occupied.  If  this  is  not  the  case,  as  is  in  Eugene,  the  break-­‐even  price  

                                                                                                                9  (Integrated  Urban  Traffic  Management,  61)  

Farber  &  Weld    

 

27  

must  increase  for  spaces  that  are  being  used  to  compensate  for  those  that  are  left  empty.10   If  this  is  not  done,  vital  revenue  is  being  wasted.     The  key  to  handling  increasing  population  effects  is  either  increasing  parking   availability  in  areas  of  highest  demand,  or  developing  a  way  to  incentivize  commuters  to   park  in  the  zones  further  away  from  campus.  The  latter  option  would  allow  for  the   utilization  of  a  greater  quantity  of  meters  varied  across  all  zones,  and  also  has  the  potential   to  offer  drivers  a  lower  overall  cost  to  park  in  select  zones.  One  possible  method  is  to   implement  a  varied  fee  schedule,  decreasing  hourly  fee  rates  for  the  underutilized  areas,   thus  incentivizing  drivers  to  park  in  underutilized  zones.  In  our  opinion,  with  regards  to   those  zones  that  are  being  underutilized,  creating  even  a  small  amount  of  hourly  revenue  is   still  better  than  making  zero  hourly  revenue.  Doing  so  would  also  open  up  space  in  higher-­‐ demand  zones  for  those  with  a  larger  willingness-­‐to-­‐pay,  simultaneously  alleviating   congestion  issues  and  increasing  parking  efficiency  and  effectiveness.    

G. Conditional  Forecast   Utilizing  the  results  found  through  our  research,  we  are  able  to  construct  a   rudimentary  forecast  for  the  city  of  Eugene  in  the  event  of  a  planning  increase  in  parking   prices  from  the  current  $1.70  to  $2.00  at  the  end  of  2013.  Taking  into  account  this  potential   17.64%  increase  in  the  price  of  parking,  we  were  able  to  simulate  multiple  scenarios  based   on  the  price  coefficient  and  its  respective  confidence  interval.     Beginning  with  a  simple  simulation,  we  found  that,  ceteris  paribus,  a  $0.30  increase   in  parking  price  per  hour  will  result  in  a  baseline  $2.06/hour  increase  in  total  revenue  for   each  respective  zone.  This  results  in  an  aggregate  increase  of  $137,230.40  annually  across                                                                                                                   10  (Wilson,  25)  

Farber  &  Weld    

 

28  

all  zones.  Compared  with  total  revenue  for  the  campus  district  in  2012:  $1,198,186.70,  the   potential  increase  in  price  results  in  an  11.45%  increase  in  total  revenue.  Further   examining  the  effects  of  a  $0.30  price  increase,  we  are  able  to  utilize  the  confidence  interval   associated  with  our  price  coefficient  to  construct  a  “best/worst  case  scenario”  simulation.   We  found  that  along  with  the  previously  stated  price  increase,  the  city  of  Eugene  can  expect   increases  in  total  revenue  ranging  from  $1.72  to  $2.41  per  hour  for  a  given  zone,  or  a  total   of  $114,216.95  to  $160,196.97  annually  across  all  zones.   From  these  results,  we  can  calculate  both  the  price  elasticity  of  demand,  as  well  as   the  elasticity  of  total  revenue.  We  find  that  the  elasticity  of  demand  for  public  parking  on   campus  is  approximately  -­‐0.3,  pointing  to  relative  inelasticity.  This  leads  to  a  revenue   elasticity  of  0.65,  again  confirming  our  original  hypothesis.  These  results  are  consistent   with  previous  studies  such  as  Clinch  &  Kelly  (2003)  and  Concas  &  Nayak  (2012),  with  the   general  consensus  centering  on  a  price  elasticity  of  -­‐0.3.  

V.  Conclusion   From  our  results  we  have  explained  many  valuable  aspects  to  public  parking  in   Eugene,  Oregon.  Firstly,  we  found  that  increases  in  price  and  the  subsequent  increase  in   total  revenue  point  to  relatively  inelastic  demand  averaging  -­‐0.3.  Furthermore,  when   looking  at  distance  in  relation  to  changes  in  total  revenue,  we  found  that  increases  in   distance  per  feet  from  key  locations  on  campus  for  a  given  zone  result  in  a  reduction  of   total  revenue  for  the  central  campus  hotspot,  but  showed  slight  increases  for  the  east  and   western  campus  hotspots.  We  next  looked  the  distribution  of  total  revenue  across  times  of   day,  and  found  revenue  at  its  peak  during  the  period  from  8:00am  to  1:00pm.  We   additionally  looked  at  the  interactions  between  distance  and  major  events  on  campus,  as  

Farber  &  Weld    

 

29  

well  as  time  of  day  and  day  of  week,  with  mixed  results.  Finally,  we  looked  at  the  effect  of  a   future  price  change  of  $0.30  on  total  revenue,  and  found  that  revenue  increased  $2.06  for  a   given  zone  every  hour,  totaling  in  $137,230.40  annually.     This  paper  outlines  a  comprehensive  framework  for  the  study  of  price  elasticity  of   parking  demand  in  Eugene,  Oregon.  Our  hope  is  that  the  City  of  Eugene  and  other   organizations  can  utilize  the  results  of  this  paper  to  improve  the  effectiveness  of  their   parking  policies.  Additionally,  we  hope  that  future  inquiries  into  the  nature  of  demand   elasticity  for  campus-­‐area  parking  can  apply  the  basic  techniques  and  methodologies  used   in  this  paper  as  a  jumping  off  point.            

               

Farber  &  Weld    

 

30  

VI.  Works  Cited   Amemiya,  Takeshi,  1984.  "Tobit  models:  A  survey,"  Journal  of  Econometrics,  Elsevier,  vol.   24(1-­‐2),  pages  3-­‐61.       Barrar,  Jack  E.  A  Multinomial  Logit  Model  of  the  Journey  to  School:  A  Study  of  Individual   Travel  Demand.  ,  1978.  Print.     Clinch,  Peter  J.,  Kelly,  Andrew  (2003),  Temporal  Variance  Of  Revealed  Preference  On-­‐Street   Parking  Price  Elasticity,  Department  of  Environmental  Studies,  University  College  Dublin   (www.environmentaleconomics.net).     Concas,  SisinnioNagesh  Nayak  (2012),  A  Meta-­‐Analysis  of  Parking  Price  Elasticity,   presented  at  the  Transportation  Research  Board  Annual  Meeting  (www.trb.org);   http://amonline.trb.org/1spbip/1spbip/1.     Dagget,  j.  and  R.  Gutkowski.  University  Transportation  Survey:  Transportation  in   University  Communities.  City  of  Fort  Collins,  CO.:  Mount-­‐Plains  Consortium,  Colorado  State   University,  2002.     Drukker,  David  M.  "Bootstrapping  a  Conditional  Moments  Test  for  Normality  After  Tobit   Estimation."  The  Stata  Journal  2.2  (2002):  125-­‐39.    

Epstein,  Richard  A.  "The  Allocation  of  the  Commons:  Parking  on  Public  Roads."  The  Journal   of  Legal  Studies  31.S2  (2002):  S515-­‐544.  JSTOR.  Web.  10  June  2013.   .       Federal  Transit  Administration.  Transportation  Research  Board  of  the  National  Academies.   (2005)  Parking  Pricing  and  Fees:  Traveler  Response  to  Transportation  System  Changes.   Retrieved  from  http://onlinepubs.trb.org/onlinepubs/tcrp/tcrp_rpt_95c13.pdf     Greene,  William  H,  1981.  "On  the  Asymptotic  Bias  of  the  Ordinary  Least  Squares  Estimator   of  the  Tobit  Model,"  Econometrica,  Econometric  Society,  vol.  49(2),  pages  505-­‐13,  March.       Integrated  Urban  Traffic  Management:  A  Report.  Paris:  Organisation  for  Economic  Co-­‐ operation  and  Development,  1978.  Print.     Johansson,  B,  and  L.-­‐G  Mattsson.  Road  Pricing:  Theory,  Empirical  Assessment,  and  Policy.   Boston:  Kluwer  Academic  Publishers,  1995.  Print.         Lipscomb,  Clifford,  and  Brandon  Kofor.  "Conservative  Dichotomous  Choice  Responses  in   the  Active  Policy  Setting:  The  Case  of  University  Parking."  Applied  Economic  Letters.   (2010):  1-­‐4.  Web.    March.  2013.    

Farber  &  Weld    

 

31  

Litman,  Todd.  "Understanding  Transport  Demands  and  Elasticities."  Http://www.vtpi.org.   Victoria  Transport  Policy  Institute,  12  Mar.  2013.  Web.  19  Mar.  2013.     Meck,  J.  P.,  &  United  States.  (1965).  The  role  of  economic  studies  in  urban  transportation   planning.  Washington:  U.S.  Dept.  of  Commerce,  Bureau  of  Public  Roads.     Miller,  G,  &  Everett,  C,  (1982).  Raising  commuter  parking  prices—an  empirical  study.   Transportation  11:  105-­‐-­‐  129     White,  Halbert,  1980.  "A  Heteroskedasticity-­‐Consistent  Covariance  Matrix  Estimator  and  a   Direct  Test  for  Heteroskedasticity,"  Econometrica,  Econometric  Society,  vol.  48(4),  pages   817-­‐38,  May.       Willson,  Richard  W.  Suburban  Parking  Economics  and  Policy:  Case  Studies  of  Office   Worksites  in  Southern  California.  Washington,  D.C:  Federal  Transit  Administration,   University  Research  and  Training  Program,  1992.  Print.                                                          

 

Farber  &  Weld    

 

32  

Appendix  A:  Regression  Results    

Table  11.   Regression  Results   Variable  

Tobit  

Price     1:00     2:00     3:00     4:00     5:00     6:00     7:00     8:00     9:00     10:00     11:00     12:00     13:00     14:00     15:00     16:00     17:00     18:00     19:00    

11.706   (11.71)**   -­‐3.992   (1.99)*   -­‐15.237   (6.09)**   -­‐27.658   (7.71)**   -­‐23.998   (7.74)**   6.819   (4.25)**   46.487   (34.56)**   87.667   (63.96)**   104.892   (74.72)**   105.048   (75.97)**   98.169   (72.86)**   104.003   (76.71)**   101.938   (75.75)**   103.854   (76.75)**   95.774   (71.43)**   95.880   (71.32)**   89.429   (66.91)**   88.403   (66.01)**   80.333   (60.07)**   70.149   (52.58)**  

OLS   5.330   (10.46)**   -­‐0.001   (0.01)   -­‐0.003   (0.02)   -­‐0.004   (0.03)   -­‐0.004   (0.03)   0.077   (0.60)   1.579   (12.70)**   16.553   (66.41)**   27.511   (77.78)**   26.530   (83.53)**   20.031   (90.91)**   24.031   (93.37)**   21.908   (100.11)**   23.354   (94.97)**   16.847   (93.20)**   17.178   (86.45)**   12.566   (82.45)**   12.296   (77.08)**   7.817   (58.30)**   2.393   (20.22)**  

Farber  &  Weld    

  20:00     21:00     22:00     23:00     Spring     Fall     Winter     Student  Enrollment     Alder     E.  12th     E.  14th     E  17th     Hilyard     Kincaid     Orchard     Patterson     Villard     Walnut     Agate     Weekday     Saturday     Finals     Winter  Break    

33   31.820   (22.80)**   19.123   (13.00)**   10.426   (6.69)**   4.663   (2.79)**   6.548   (13.23)**   4.103   (7.22)**   4.657   (8.49)**   -­‐0.001   (4.63)**   -­‐30.680   (103.78)**   -­‐16.270   (75.51)**   -­‐4.637   (20.24)**   -­‐44.211   (138.45)**   -­‐32.495   (123.18)**   27.912   (79.23)**   -­‐36.721   (122.57)**   -­‐52.911   (168.71)**   -­‐28.517   (100.34)**   -­‐72.371   (124.05)**   -­‐7.527   (27.53)**   56.227   (150.90)**   42.060   (103.26)**   -­‐2.006   (6.38)**   -­‐22.332   (81.09)**  

0.036   (0.28)   0.014   (0.11)   0.006   (0.05)   0.003   (0.02)   1.591   (7.76)**   0.819   (3.37)**   1.081   (4.56)**   -­‐0.000   (5.17)**   -­‐11.762   (98.07)**   -­‐8.329   (76.51)**   -­‐2.651   (19.63)**   -­‐14.546   (122.94)**   -­‐15.199   (132.08)**   16.565   (57.73)**   -­‐14.768   (119.80)**   -­‐16.403   (138.29)**   -­‐12.374   (100.59)**   -­‐16.263   (133.55)**   -­‐3.713   (23.91)**   11.786   (150.82)**   6.108   (57.84)**   -­‐1.037   (5.54)**   -­‐7.425   (74.24)**  

Farber  &  Weld    

  Spring  Break     Summer  Break     Basketball  Game     Football  Game     Tack  Meet     Constant     Sigma     Observations   R2  

-­‐17.276   -­‐6.555   (39.89)**   (42.69)**   -­‐13.637   -­‐6.834   (9.59)**   (11.43)**   6.357   1.985   (15.36)**   (10.00)**   10.475   4.161   (17.44)**   (14.26)**   7.545   2.507   (17.98)**   (12.69)**   -­‐120.831   –   (41.36)**   –   26.271   -­‐4.257   (218.07)**   (3.72)**                  317,856                  317,856     0.37  

Robust  Standard  Errors  in  Parentheses   *  p

Suggest Documents