New highways and land use change: Results from a quasi-experimental research design

of 23

Please download to get full document.

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
23 pages
0 downs
New highways and land use change: Results from a quasi-experimental research design
  New highways and land use change: Results from aquasi-experimental research design Richard G. Funderburg a, * , Hilary Nixon b,1 , Marlon G. Boarnet c,2 , Gavin Ferguson d a Graduate Program in Urban and Regional Planning, The University of Iowa, 338 Jessup Hall, Iowa City, IA 52242-1316, USA b Department of Urban and Regional Planning, San Jose State University, One Washington Square, San Jose, CA 95192-0185, USA c Departments of Planning, Policy, and Design and of Economics, University of California, Irvine, Irvine, CA 92697-7075, USA d Department of Planning, Policy, and Design, University of California, Irvine, Irvine, CA 92697-7075, USA a r t i c l e i n f o  Article history: Received 29 October 2008Received in revised form 14 October 2009Accepted 10 November 2009 Keywords: HighwaysQuasi-experimental designPropensity score matching a b s t r a c t Understanding links from new highway construction or capacity expansion to regionalgrowth patterns is crucial for transportation planners and policy makers. In this paper,we incorporate a lagged adjustment regional growth model into a quasi-experimentalresearch design to examine the association between new highway investments and landusechangeinthreeCaliforniacounties.Ourstudyareasprovideamixofurban,smalltown,and exurban highway projects in order to explore the different effects across project typesandgeographic contexts. Thecentral findingof thisresearchisthatwhileimprovementsinsurfacetransportationinfrastructurecanhavelargeimpactsongrowthpatterns,thenatureof the effect depends on the context of the highway investment.   2009 Elsevier Ltd. All rights reserved. 1. Introduction The link between highway construction and urban development patterns is one of the enduring mysteries of modern ur-banplanning.NewevidencesuggeststhattheUSInterstateHighwaySystemcontributedimportantlytothedecentralizationof the country’s metropolitan areas (Baum-Snow, 2007), but the question of the impact of specific highway links on urbangrowth, especially roads built after the maturation of the national highway system, remains unclear. A fundamental debatein urban and regional planning is whether new highway infrastructure induces growth or merely follows the path of devel-opment to service regions that would have grown with or without the new investment. In this research, we focus on thequestionof causality. Weincorporatealaggedadjustmentregionalgrowthmodelintoaquasi-experimental researchdesignto examinethe associationbetweennewhighwayinvestmentsin the Californiacounties of Merced, Orange, and Santa Claraand changes in population and employment location, while controlling for no-build counterfactuals. We study this mix of urban, small town, and exurban highway projects to examine the possibility of differential effects across varying projecttypes and geographic contexts.The central finding of the paper is that while improvements in surface transportation infrastructure tend to have largeimpactsongrowthpatterns,thenatureoftheeffectdependsonthecontextofthehighwayinvestment.Ourmodelsestimatethat, on average, 13–9410 new Orange County jobs occurred within a typical census tract in the formerly exurban regionafter gaining highway access when compared to no-build counterfactuals. At the lower end, this gain is not economically 0965-8564/$ - see front matter   2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.tra.2009.11.003 *  Corresponding author. Tel.: +1 319 335 0036; fax: +1 319 335 3330. E-mail addresses: Funderburg), Nixon), Boarnet),fergusog@uci. edu (G. Ferguson). 1 Tel.: +1 408 924 5852. 2 Tel.: +1 949 824 7695. Transportation Research Part A 44 (2010) 76–98 Contents lists available at ScienceDirect Transportation Research Part A journal homepage:  significant, although the upper end of the range and the impact identified by a differences-in-differences point estimate of 3948 new jobs is economically significant. Our quasi-experimental regression model does not predict a statistically signif-icanteffectasaresultofahighwaybypassbuiltoutsidethesmalltownofLivingstoninMercedCounty;however,wedofindan economically significant loss of 36 jobs per square kilometer within a two mile distance of Livingston’s new bypassthrough our differences-in-differences point estimate. This represents a 33% slower growth rate compared to 1990 levels.We find no statistically significant effects on population or employment growth that can be attributed to the new highwayinvestment near the urban center of Santa Clara County. The different effects from highway investments in the three con-texts illustrate the importance of choosing appropriate comparison groups in forecasts of population and employmentgrowth for build and no-build scenarios. 2. Literature review The literature on the growth impacts of highways is vast, but key points can be summarized succinctly. Giuliano (2004),in a comprehensive review of the literature on highways and metropolitan growth patterns as of the late 1980s, concludedthat highways might have had important growth impacts in the early years of Interstate Highway construction (the 1950sand 1960s), but any growth impacts had become small or zero by the time the system had matured in the 1980s. Giulianoarguedthatbythe1980sadditionstotheinterstatesystemweremarginaladditionstoalreadycompletednetworks,andthenetwork accessibility and hence the land use impact of those additions was small. Different literatures, focused on popula-tion and employment location modeling or related topics, have typically found that highways are a statistically significantpredictor of the cross-sectional pattern of residential and firm location (e.g., Boarnet, 1996; Bollinger and Ihlanfeldt, 2003;Cervero, 2003; Holl, 2004; Luce, 1994), but that finding sometimes leaves open the question of whether the association re-flects recent movements or location choices from prior decades.Baum-Snow (2007) provides an econometric test of the impact of the US Interstate Highway System on metropolitandecentralization. He finds that building the first new highway through a central city reduces central city population growthby 17%, while increasing suburban population growth. Nationally, Baum-Snow(2007) estimates that building the InterstateHighway System resulted in central city population growth that was 8% lower than what would have otherwise occurred,again shifting growth to the suburbs. In a similar vein, Holl (2004) studied firm location effects in the first decade afterimprovements to Portugal’s highway network and found statistically significant impacts on firm location. The magnitudeof the effect, for both Baum-Snow and Holl, is economically important, but the question of whether the impact is drivenby the initial years of national highway system construction (as Giuliano, 2004 would argue) still lingers. In addition, testsof entire highway systems give average effects, while the pertinent policy question in many instances involves the growthimpactsofspecifichighwayprojects(ForkenbrockandFoster,1990)–apointthatwasvividlyillustratedbyalegalchallengeto toll road projects in the state of Illinois.In 1997, a US District Court Judge ruled that the Environmental Impact Statement for a proposed Illinois toll road wasdeficient because the growth forecasts were the same in the build and no-build scenario (Sierra Club v. United StatesDOT,1997).ThisestablishesalegalprecedentthatEnvironmentalImpactStatements,undertheNationalEnvironmentalPol-icy Act (NEPA), require some understanding of how growth patterns will differ depending on whether or not a highway isbuilt. It is this backdrop specifically, and the continuing concern about metropolitan decentralization and urban growth ef-fects of highways more generally, that lead us to study the relationship between population and employment growth andthree highway projects, all of which were built in the context of already mature highway systems.Ourmethodsdrawfromtwoliteratures–(1)researchonquasi-experimentalmatchingtechniquesanddifferences-in-dif-ferences estimation and (2) lagged adjustment urban growth models. The quasi-experimental approach has a long traditioninthestudyofhighwayimpacts.Examplesofmatchedpairanalysesofgrowthinlocationsnearandmoredistantfromhigh-waysincludeWheat(1969),HumphreyandSell(1975),andRephannandIsserman(1994).Withthepassageoftime,studieshave typically used more sophisticated matching techniques to choose control groups and more abundant data, often madeavailable in part by the development of geographic information system technology. 3 Inour study, wefollowontherecent literaturethat uses propensityscorematchingtechniques toconstructquasi-exper-imental evaluations of interventions or programs by selecting control groups that are as similar as possible to the experi-mental group based on variables selected by the researcher (see, e.g., Dehejia and Wahba, 1999, 2002; O’Keefe, 2004;Rosenbaum and Rubin, 1983; Smith and Todd, 2005). We first perform a differences-in-differences test for populationand employment growth across experimental groups near the new highways and control groups further away, in the tradi-tion of, e.g., Card and Krueger (1994) or Smith and Todd (2005).As an additional test, we then nest a full regression analysis of population and employment changes within the matchedexperimental-control quasi-experimental framework. The regression model draws on the now long tradition of simulta-neouspopulationandemploymentlocationmodels(BradfordandKelejian,1973;CarlinoandMills,1987;SteinnesandFish-er,1974).WeusethespatialeconometricversionofthelaggedadjustmentmodeldevelopedbyBoarnet(1992,1994), which 3 In addition, when early studies, such as Wheat (1969) or Humphrey and Sell (1975) found associations between highways and growth patterns, that was in the context of the construction of the initial parts of the network, and so does not necessarily inform our question about growth impacts of highwayconstruction in a mature network. R.G. Funderburg et al./Transportation Research Part A 44 (2010) 76–98  77  in various forms has been used to examine a broad range of intra-metropolitan policy questions, including growth impactsrelatedtotransportationinfrastructure(e.g., Boarnet,1996; Bollinger andIhlanfeldt, 1997; Henryet al., 1997, 1999; Schmittand Henry, 2000) and related metropolitan or rural development questions (Boarnet et al., 2005; Carruthers and Mulligan,2007, 2008; CarruthersandVias, 2005; ClarkandMurphy,1996; Delleret al., 2001; Duffy-Deno, 1998; Edmiston, 2004; Gla-vac et al., 1999; Henry et al., 2001; Leichenko, 2001; Mulligan et al., 1999; Poppert and Herzog, 2003; Vias, 1999; Vias andMulligan, 1999).Within the context of the literature, our study of growth impacts near recent highway investments makes several con-tributions. We study the impacts of specific highwayinvestments, movingfromquestions of overall (e.g., national) averagesto the question of impacts of particular highway investments. This project-specific approach will be needed to respond toquestions such as those raised by the legal challenge to the Illinois toll roads under NEPA in the case of  Sierra Club v. UnitedStatesDOT(1997). Westudyinvestmentsinalreadymaturehighwaynetworkstoaddressthelingeringquestionof whethergrowth impacts from highways are absent or minimal in the modern day context. We use state-of-the-art propensity scorematching techniques and quasi-experimental research design that is then nested within a lagged adjustment regressionanalysis to provide a counterfactual that can be linked to required ‘‘build/no-build” scenarios used in project and environ-mental analyses. Last, we test highway projects of different magnitudes in different geographic areas to shed light onwhetherandhowthegrowthimpactsofhighwaysvarybasedoncontext.Overall,weviewoureffortprimarilyasanattemptto‘‘concepttest”theapplicationofquasi-experimentalmethodstothequestionofgrowthimpactsofmodernhighways.Ourresults give important insights, both into the application of quasi-experimental methods and the growth impacts of high-ways in different contexts. We suggest that future research using similar methods applied to other highway projects andcontexts would be a useful way to build a knowledge base that can inform project and environmental analysis.Theremainderofthepaperisstructuredasfollows.Thenextsectiondescribesthethreestudycountiesandthenewhigh-way investments constructed in each county during the mid-1990s that constitute the policy interventions. Section 4 de-scribes the research design in broad terms. Section 5 describes the data. In Section 6 we give details about implementing the propensity score matching methods used to select a control for each spatial unit that received access to new highwayinfrastructure in the mid-1990s. Section 6 also includes a description of our differences-in-differences test for highway Fig. 1.  Locations of study counties.78  R.G. Funderburg et al./Transportation Research Part A 44 (2010) 76–98  impact and a more elaborate regression-based lagged adjustment model test. In Section 7 we provide our findings and weconclude the paper with implications and suggestions for further research in the last section. 3. Study regions and highways We selected three sets of highway projects constructed in California during the mid-1990s. The projects are located inSanta Clara County, a predominantly urbanized region, the small town of Livingston in Merced County, a rural region of the state, and southern Orange County, which at the time was part of the growing exurban fringe of the Los Angeles metro-politanregion.ThethreecountiesandtheirlocationinCaliforniaareshowninFig.1. Thenewhighwaysandthepre-existinghighway network in each county are shown in Figs. 2–4.  3.1. Santa Clara County Santa Clara County is an urban county on the southern end of the San Francisco Bay Area’s Silicon Valley. The highwayprojects studied in Santa Clara County are the extension of State Route 87, also called the Guadalupe Freeway, from Inter-state 280 near downtown San José to State Route 85 plus the constructionof an interchange north of downtown with High-way101andNorthFirstStreet,anextensionofStateRoute85,whichlinksI-280inCupertinotothenorthandjoinsHighway101 in the southern part of San José, and the upgrading of State Route 237 from surface streets to a grade-separated, con-trolled-access freeway in northern San Jose connecting I-880 to Highway 101.  3.1.1. State Route 87  Asearlyasthe1960s, theGuadalupeParkwayprovidedaconnectionfor driversfromHighway101inthenorthtodown-town San José. Construction to convert the parkway to a grade-separated, controlled-access freeway began on a northern Fig. 2.  Santa Clara County highway network. R.G. Funderburg et al./Transportation Research Part A 44 (2010) 76–98  79  portion in the 1970s. The intervention studied in this paper is the five new centerline miles that comprise the highway’ssouthern extension, a project that was completed in 1993, and the interchange connecting Highway 87 to Highway 101north of downtown.  3.1.2. State Route 85 The srcinal plans to construct Highway 85 date back to 1957 when the California Divisionof Highways signed an agree-ment with cities in the south Bay Area to build Highway 85 (Richards, 1994). The first seven miles of the route, fromCuper-tino to Mountain View (I-280 north to Highway 101) opened in 1971. No new construction occurred along the designatedroute for the next decade and a half, and during the 1970s and 1980s some building construction was allowed within theHighway 85 right-of-way – structures that later had to be demolished to make way for the freeway. Construction of theremaining portions of State Route 85 became a possibility in 1984 with the passage of Measure A, the first local sales taxin California for transportation. The intervention modeled in this paper includes all 19 new centerline miles of Highway85 that connect Interstate 280 South to Highway 101. Construction to extend Highway 85 from I-280 south to Highway101 began in the late 1980s and the first two-mile section, from Cottle Road to Santa Teresa Boulevard, opened in 1991.Ayearlater,threeadditionalmilesopened,fromSantaTeresaBoulevardtoAlmadenExpressway.Thelast14miles,complet-ing the route, opened October 19, 1994 (Richards, 1994).  3.1.3. State Route 237  Theprimaryimpetusfor MeasureAwas State Route237. Inthe1980s, thispopularconnectionforhigh-techworkersliv-inginFremontorMilpitastogettojobsinMountainViewandPaloAltowasdescribedasa‘‘parkinglot”duringthemorningand afternoon commutes (Richards, 1997). Construction began in 1990 and the final traffic lights were removed in 1997. Fig. 3.  Merced County highway network.80  R.G. Funderburg et al./Transportation Research Part A 44 (2010) 76–98
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks