Land use externalities, open space preservation, and urban sprawlIntroductionInteraction effects and the optimal timing of developmentHazard model of land developmentDataIdentifying neighborhood interactions and the indirect policy effectEmpirical resultsPredicting the effects of preserved open space on sprawlConclusions: unintended consequences and sprawlAcknowledgementsReferencesLand use externalities, open space preservation,and urban sprawlElena G. Irwina,*, Nancy E. BockstaelbaDepartment of Agricultural, Environmental, and Development Economics, Ohio State University,2120 Fyffe Road, Columbus, OH 43210, USAbDepartment of Agricultural and Resource Economics, University of Maryland, USAAccepted 2 March 2004Available online 18 May 2004AbstractParcel data on residential land conversion are used to investigate how land use externalitiesinfluence the rate of development and modify policies designed to manage urban growth andpreserve open space. Several ‘‘smart growth’’ policies are found to significantly influence landconversion, including a development clustering policy that concentrates development and generatespreserved open space. In addition to directly affecting a parcel’s hazard rate of conversion, thispolicy is found to affect neighboring parcels’ conversion by generating a positive open spaceexternality that hastens their development. The implication that the clustering policy could generate amore sprawled pattern of development is explored using spatial simulation.D 2004 Elsevier B.V. All rights reserved.JEL classification: R14; R52Keywords: Land development; Urban land use patterns; Smart growth1. IntroductionSprawl, a term often used to describe non-compact features of urban land usepatterns, is a regional-level phenomenon driven by individual choices over location andland use that are influenced by a myriad of factors, including land features, infrastruc-ture, policies, and individual characteristics. Because it is the cumulative result ofindividual actions, an understanding of sprawl requires an understanding of individual0166-0462/$ - see front matter D 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.regsciurbeco.2004.03.002* Corresponding author. Tel.: +1-614-292-6449; fax: +1-614-292-0078.E-mail address: [email protected] (E.G. Irwin).www.elsevier.com/locate/econbaseRegional Science and Urban Economics 34 (2004) 705– 725decision making and how these decisions ‘‘aggregate up’’ over time and space.However, because of a lack of spatial data on individual-level land use and locationchoices, most empirical studies have defined sprawl at an aggregate level, e.g., censustract or county level, and have measured it using aggregate values of population,employment, or urban development (e.g., Fulton et al., 2001; Galster et al., 2001). Atthis more aggregate scale of analysis, causal explanations of sprawl that link underlyingindividual behaviors to observed pattern s are difficult to come by and empiricalanalyses of sprawl are most often limited to identifying characteristics that areassociated with, but do not necessarily cause, sprawl. On the other hand, parcel andplot-level models of individual land use conversion decisions have been useful inidentifying factors that drive changes at a disaggregate level (e.g., Bockstael, 1996;Irwin et al., 2003; Landis and Zhang, 1998; McMillen, 1989 ), but this approach offersan incomplete picture for understanding sprawl, which is the cumulative result of manyindividual decisions.This paper uses parcel-level data on land conversion and a Geographical Informa-tion System (GI S) to explore the pattern of development as the result of several typesof influences: (1) spatially varying policy variables, including several ‘‘smart growth’’policies and a clustering regulation that requires developers to preserve varyingamounts of open space by clustering development on a parcel; (2) other spatiallyheterogeneous features of land parcels, including accessibility to urban centers, soilquality, and size; and (3) interactions among neigh borin g landowners ’ land usedecisions due to land use externalities. We do so by building on the optimal timingof development model developed in Irwin and Bockstael (2002), which incorporatesthe e ffects of land use externalities and is estimated using a hazard model ofdevelopment. Our analysis extends the previous work by incorporating a much moreextensive set of policy variables, including a variety of ‘‘smart growth’’ variables thatare hypothesized to influence development timing , and by focusing on how thepresence of land use externalities may moderate the effectiveness of these policies.In particular, we focus on whether land use externalities alter the effectiveness of aclustering policy designed to protect open space by creating a posit ive amenityassociated with the preserved open space that, under certain conditions, may attractdevelopment and exacerbate sprawl.To structure our analysis, we argue that policies have both direct and indirect effects.Direct policy effects are those that influence a landowner’s decision by directly influencingeither the costs or returns to development in some way, e.g., by constraining land usechoices, limiting the allowable density of development, or improving the level of publicservices associated with a parcel. Indirect effects, on the other hand, arise via land useexter nalities from neighb oring land parcels and can either increa se or decrea se thedevelopment v alue of a parcel. An indirect policy effect arises if a policy induce s a landuse change on one parcel that in turn affects a neighboring parcel’s development value.The specific externality that we hypothesize in the current analysis is a positive amenityeffect that arises from designating neighboring land as preserved open space and whichmay be associated with a scenic view, increased privacy, or guarant ee of no neighboringfuture development. If present, such indirect effects would certainly complicat e the task ofpredicting the likely impacts of a policy on land use patterns and, in some cases, mayE.G. Irwin, N.E. Bockstael / Regional Science and Urban Economics 34 (2004) 705–725706contr ibute to unintended consequences such as exacerbation of a sprawl pattern ofdevelopment.The rest of the paper is organized as follows. First, the influences of land useexternalities and direct and indirect policy effects on the optimal timing of landdevelopment are