The neasy case for lower Parking Standards W. Bowman Ctter 1, Sofia F. Franco 2 Janary 2012 (First Draft) Minimm parking reqirements are the norm for rban and sbrban development in the United States (Davidson and Dolnick (2002)). The jstification for parking space reqirements is that overflow parking will occpy nearby street or off-street parking. Shop (1999) and Willson (1995) provides cases where there is reason to believe that parking space reqirements have forced parcel developers to place more parking than they wold in the absence of parking reqirements. If the effect of parking minimms is to significantly increase the land area devoted to parking, then the increase in impervios srfaces wold likely case water qality degradation, increased flooding, and decreased grondwater recharge. However, to or knowledge the existing literatre does not test the effect of parking minimms on the amont of lot space devoted to parking beyond a few case stdies. This paper tests the hypothesis that parking space reqirements case an overspply of parking by examining the implicit marginal vale of land allocated to parking spaces. This is an indirect test of the effects of parking reqirements that is similar to Glaeser and Gyorko (2003). A simple theoretical model shows that the marginal vale of additional parking to the sale price shold be eqal to the cost of land pls the cost of parking constrction. We estimate the marginal vales of parking and lot area with spatial methods sing a large data set from the Los Angeles area non-residential property sales and find that for most of the property types the marginal vale of parking is significantly below that of the parcel area. This evidence spports the contention that minimm parking reqirements significantly increase the amont of parcel area devoted to parking. JEL Codes: R52, H23. Keywords: Parking; Land Use; Sprawl 1 Pomona College, Department of Economics, Claremont, CA 91711; Telephone 909-607-8187; Email Bowman.Ctter@Pomona.ed 2 Corresponding athor. Nova School of Bsiness and Economics (Nova SBE), Universidade Nova de Lisboa, Camps de Campolide P-1099-032, Lisboa, Portgal. Email: sfranco@novasbe.pt. Phone: +351-21-380-1600 Ext: 1353. And UECE-Research Unit on Complexity and Economics, Lisboa, Portgal. 1
1. Introdction Most cities in the US have parking standards which reqire developers to provide a minimm amont of off-street parking per sqare foot of floor space. 3 These minimm parking reqirements are sally set by city planners from standardized transportation planning manals, which typically measre parking and trip generation rates at peak periods with ample free parking and no pblic transit. The goal of Minimm Parking Reqirements (MPRs) is to ensre adeqate parking at a low price in order to limit local congestion and to stimlate local bsiness (Shop (1999)). Minimm parking reqirements have been criticized de to their land and transportation market distortions. Opponents of MPRs (Shop (1999, 2005), Willson (1995)) arge that these parking standards create an overspply of parking in most rban areas which decreases the cost (direct and time) of parking and therefore encorages more atomobile trips (Shop (1999, 2005), Shop and Pickrell (1978)). As the nmber of cars increase and mass transit se decreases, air qality decreases. Decreased air qality creates and/or aggravates health problems for rban residents. Individals who sffer from plmonary diseases sch as asthma and bronchitis sffer as the increased nmber cars decrease air qality (EPA (1999, 2008), Shop (1999)). In addition, critics allege that minimm parking reqirements force developers to se more land space per sqare foot of bilding constrction and make development in areas where land has a high vale mch more expensive and less profitable (Willson (1995)). As a reslt, minimm parking reqirements inflence the location of new development, make infill projects and historic bilding retrofits less attractive and feasible (Shop and Pickrell (1978)) and contribte to the sprawling of impervios parking srface at the expense of the environment (Feitelson and Rotem (2004)) and rban design (Mkhija and Shop (2006)). However, the previos debate does not consider other factors that may interact with MPRs sch as Floor-Area- Ratio (FAR) restrictions. A FAR restriction is a density reglation that imposes a restriction on a bilding s FAR, which eqals the total floor area in the bilding divided by the lot size. A limit on FAR ths prevents a developer from constrcting tall bildings. 3 For example, the zoning ordinance for Wilshire Bolevard in Los Angeles reqires three parking spaces per 1,000 sqare feet of gross floor area for office bildings and other commercial ses (Shop (1999)). 2
The prposes of this paper are twofold. First, we develop an analytical model of bilding constrction that incldes MPRs, FAR restrictions and endogenos decision-making over srface verss below-grond parking. This theoretical model allows s to formally examine the impacts of minimm parking reqirements on development density, parking external costs and amont of parking spplied, adding to a small analytical literatre on this reglation. And second, we test the hypothesis that parking reqirements case an overspply of parking sing data on commercial, indstrial and retail property sales from Los Angeles Conty. Even thogh parking reqirements are intensely debated in rban and transport planning arenas, little effort has been devoted to the theoretical analysis of this instrment. The only analytical stdies are by Shop and Pickrell (1978) and Feitelson and Rotem (2004). Both stdies se graphical analysis to discss the impacts of minimm parking reqirements on parking price and qantity. Shop and Pickrell (1978) focs their graphical discssion on how minimm parking reqirements may affect the development of parking sbmarkets. They arge that if parking reqirements are niform throghot a jrisdiction while demand and spply vary by location, the reqirements may force an increase in the total amont of parking provided bt also an inefficient allocation of parking across space. Feitelson and Rotem (2004) focs their graphical discssion on the external costs of srface parking. The athors arge that even withot minimm parking reqirements, developers will overspply parking becase of the direct environmental negative externalities and the indirect sprawl-indcement externalities. There are also some theoretical analyses on the impacts of density zoning on the rban spatial strctre and on the rban welfare (Arnott and Mackinnon (1977), Bertad and Breckner (2005)). For example, Bertad and Breckner (2005) show theoretically that FAR restrictions encorage rban sprawl and increase commting costs for edge residents of a city. However, we are not aware of any analyses in the zoning literatre that also take into accont the interaction of FAR restrictions with transportation policies sch as MPRs. Both FAR and MPR are land se reglations with impacts on strctral densities. FAR restrictions are designed to deal with neighborhood crowding or congestion and reglate how bildings impede each other s view and access to snlight. MPRs on the other hand are designed to prevent traffic congestion effects and spillover parking effects - parking from a development that overflows into srronding areasfrom new development. However, becase MPRs increase the costs of new development, these policies not only increase the cost of hosing bt may also redce density at which developers 3
wold otherwise be able to bild new hosing, potentially restricting the spply of new nits. Additionally, these minimm standards may also conteract other local policies designed to encorage downtown density to increase agglomeration economies or policies designed to promote development in areas easily accessible by pblic transit. One of the advantages of constrcting a theoretical model that captres the essence of the problems associated with minimm parking reqirements is that sch a model can develop and spport hypothesis that we can test empirically. Moreover, it also may provide sefl insights beyond those provided by earlier papers. For example, parking reqirements can inflence bilding density, bt may not always be the largest barrier to density. Zoning conditions, in particlar FAR restrictions may be a larger barrier in certain cases. Ths, in contrast to Shop and Pickrell (1978) and Feitelson and Rotem (2004), we develop a theoretical framework where we model separately the behavior of city center developers and sbrban developers and where parking and floor space are bndled and rented as a package to tenants of a bilding. Both types of developers maximize profits. We extend previos analyses by considering two types of parking strctres: ndergrond parking or srface parking. 4 In or model srface parking also generates negative external environmental costs. Another featre of or model is the presence of a floor-to-area (FAR) restriction. Within this model we examine not only the impact of MPRs on total parking spplied bt also on the spply of different types of parking (srface and ndergrond), bilding sqare footage and level of parking externality. Or analytical reslts show that srface parking is more efficient if the price of land is relatively low. For a sfficiently high price of land, the developer provides ndergrond parking instead of srface parking. Becase parking space is capitalized into rents, this will encorage developers to volntarily spply parking space whenever the reslting revene will cover its costs, even in the absence of parking reqirements. Becase developers do not take into accont parking external costs, developers will overspply parking in an nreglated market. The main effects of MPRs are that parking spaces will be priced below the cost of providing them and the external costs associated with srface parking will be exacerbated becase MPRs exacerbate the market overspply of parking. In addition, MPRs constitte an indirect tax on bilding sqare 4 We realize that there is also above-grond strctre parking that is an intermediate choice between ndergrond and srface parking, bt we believe that the ndergrond and srface parking captres the essential elements of the developer s parking choice problem. 4
footage which creates a disincentive to high-density development. In particlar, minimm parking reqirements may drive the total sqare footage allowed and potentially inhibit density below what a FAR limit permits. 5 However, in areas with stringent limits on bilding height and where parking can feasibly be provided, or analytic reslts sggest that FAR limits may be a more important constraint on densities. Parking lots occpy a significant proportion of the bilt cover in many rban and sbrban areas. Fergson (2005) estimates that in mlti-family areas parking lots coise abot 30% of the bilt cover and in commercial areas parking lots coise abot 60% of the bilt cover. Increases in impermeable srfaces sch as parking have important environmental conseqences becase impermeable srfaces are thoght to case a variety of environmental, mainly waterrelated externalities (Arnold and Gibbons (1996)). If the effect of parking minimms is to significantly increase the land area devoted to parking, the increase in impervios srfaces wold likely case water qality degradation, increased flooding, and decreased grondwater recharge. Ths, knowledge of whether or not MPRs are binding is important information. Shop (1999) and Willson (1995) provide cases where there is reason to believe that parking reqirements have forced developers to place more parking than they wold in the absence of this reglation. However, we are not aware of any broad based empirical evidence that tests whether it appears that developers are placing significantly more parking on their land than they wold in the absence of parking reqirements. In addition, to or knowledge there is also no stdy that examines which land se categories are most affected by minimm parking reqirements. In this paper we test the hypothesis that parking space reqirements case an overspply of parking sing both direct and indirect approaches. Or direct test compares actal verss reqired parking for a sbsample of or data where we are able to approximate MPRs for the property. Or indirect approach ses the gap between the marginal parking space costs and the marginal vale of an additional parking space to the sale price to measre the extent of overspply of parking. Based on or analytical reslts, this gap shold be zero whenever MPRs are not binding. Therefore, if MPRs bind this wedge shold be positive. Or indirect test is ths very similar to Glaeser and Gyorko (2003) who se the gap between real estate prices and the 5 This cold occr whenever the amont of parking is constrained for example de to site geometry (size and shape) and/or inconvenient site topography (slopes and poor soil conditions) which sbstantially increase parking costs. 5
costs of prodcing a marginal apartment to measre the distortions in the hosing market cased by zoning restrictions on new constrction. 6 We estimate the marginal vales of parking and lot area with spatial and non-spatial methods from Los Angeles area non-residential property sales. The data encompass a wide variety of indstrial, service, shopping and general retail properties, and office properties from 1997-2005. We se a spatial hedonic approach that incldes property and locational characteristics and varios controls for spatial dependence. Significant work has been done sing hedonic methods to examine how varios attribtes affect the sale price of non-residential parcels. Most of this work relates to the determinants of commercial property sale indices. 7 This literatre considers a variety of locational, neighborhood, bilding and parcel characteristics, bt we are only aware of one previos stdy that considers the amont of parking on the property (Ctter and DeWoody (2010)). In addition, other work attempts to vale environmental disamenities by examining whether property prices change in response to the listing or de-listing of hazardos waste sites near a property (Ihlanfeldt and Taylor (2004)). We se a spatial hedonic approach where parking area, a characteristic that has only been inclded in Ctter et al. (2010), is one of the characteristics of the property. Or empirical reslts find that the marginal vale of parking on a lot appears to be significantly less than the marginal vale of parcel area in several land-se categories in all specifications. The difference between the marginal parking and marginal property area vales is significant and spports the contention that minimm parking space reqirements sbstantially increase the amont of parcel area allocated to parking by developers. Or direct test of whether 6 The athors provide evidence that zoning is in fact constraining the spply of hosing in a nmber of hosing markets across the United States. Glaeser and Gyorko (2005) emphasize that there is no reason why hosing prices cannot fall below constrction costs in declining areas. In those areas, there will be no constrction and any reglation that place limits on residential bilding wold not bind. 7 Some examples are Hodgson et al. (2006) and Colwell et al. (1998). Other empirical stdies have examined demand side and spply side inflences on office-commercial rents. Clapp (1980) tested the rent-accessibility tradeoff sing 105 office bildings data in Los Angeles and fond significant evidence to spport the negative rental fnction with respects to the distance from the CBD and the commting time. His reslts also spported the importance of face-to-face interaction in the CBD. Using a more recent set of office rental data in Greater Los Angeles from the same sorce, Sivitanido (1995) again fond that the accessibility factors (distance to CBD, distance to airport and nmber of interstate freeways) are significantly reflected in variations of the office rental fnction. However, the athor fond that the standard bid-rent fnction is incomplete in explaining office bid-rent relationships nless other variables like worker amenities, zoning and local instittional control are inclded in the model. Sivitanido (1996) shows that office-commercial firms vale access to service centers within Los Angeles PMSA (Primary Metropolitan Statistical Area). None of these previos stdies examine how parking area inflences office-commercial bid rent fnctions. 6
MPRs bind, which ses a sbsample of office properties, shows also that properties tend to have jst the minimm parking reqirement or somewhat less parking than reqired. The plan of the paper is as follows. Section 2 provides the reglatory backgrond for Minimm Parking Reqirements and describes the environmental impacts associated with conventional parking lots. Section 3 otlines the theoretical model and examines the impacts of minimm parking reqirements on the spply of srface parking, spply of ndergrond parking, bilding density and level of parking externalities. Section 4 develops the empirical model. Section 5 describes or data and variables. Section 6 discsses or parking reglation tests. Section 7 presents the empirical reslts and discsses their implications and finally, Section 8 offers conclsions. 2. Reglatory and Environmental Backgrond 2.1. Minimm Parking Reqirements The minimm parking reqirements for cities in the Los Angeles area are Byzantine in their complexity. 8 Parking reqirements can differ significantly across property types that wold seem similar to the ninitiated. The basis also can differ. Many reqirements are expressed as nmbers of parking spaces per sqare foot of gross bilding floor area, bt other depend on adjsted gross area, nmber of employees, nmber of seats in a restarant, and other more complex ratios. In addition, cities often have different zones where different parking reqirements apply. Finally, projects can be granted variances from general MPRs. For all these reasons, a direct approach to estimating the effect of parking minimms on the amont of parking space, sch as regressing parking area on the parking minimm and controls, faces serios obstacles becase it wold be very difficlt to know what MPR applies to a property. This point is reinforced by or data analysis. We analyze data for a sbsample of properties (office bildings in certain cities) where the crrent MPR reqirements were only based on sqare footage of the bilding and examined the ratio of actal to reqired parking spaces. We fond the mean ratio is.97, indicating that these bildings sally have close bt slightly less parking than reqired. However, many bildings have many fewer than the reqired nmber of spaces in crrent reglation. The 25 th percentile of actal to reqired spaces is.72, so 25% of 8 For instance, the Santa Monica reqirement for health clbs is 1 space per 80 sq. ft. of exercise area, 1 space per each 300 sq. ft. of locker room/sana/ shower area, pls applicable code reqirement for other ses. 7
properties have less than 72% of the reqired spaces. This cold either be the reslt of not having the exact parking reqirements at the time of constrction or widespread exceptions (variances) allowing less than the MPR. If widespread variances exist, then it cold be that, while in theory MPRs prodce inefficient otcomes, in practice local athorities consider the costs and benefits for MPRs and grant variances where costs of compliance are particlarly high and by doing this they redce the social cost of MPRs. Shop (2005) states that many mnicipalities rely on the Institte of Transportation Engineers (ITE) parking generation rate stdies, and in particlar their maximm generation rate 9, to set minimm parking standards. However, or data raises the possibility that the actal reqirements for projects can be lower than the reglatory standards criticized by Shop (2005). This is similar to the common finding in the environmental reglatory literatre that local reglators adjst reglations for the costs and benefits of polltion control at a particlar facility (Gray and Deily (1991)). Or indirect method avoids the pitfall of assming that MPRs apply niformly and tests whether parking reqirements, as they are actally applied, reslt in economic distortions. 2.2. Environmental Costs of Parking Parking lots degrade air and water qality (Akbari et al. (2001)), decrease the amont of greenspaces, increase rban heat island effects, and can decrease water spply (Ctter et al. (2008)). These negative environmental externalities will differ in their magnitde from place to place. The water-related externalities are generated becase parking lots are impervios to water and moistre and therefore case more rainfall to flow off the srface as rnoff. This decreases water spplies as grondwater aqifers are not recharged and decreases srface water qality as the rnoff tends to accmlate oil, grease, and other polltants from the pavement. Ctter et al. (2008) estimates that an additional acre of impervios srface can decrease annal infiltration 9 These figres are based on stdies where a land se category is observed over several days to several weeks and the maximm observed parking spaces over that period of time is designated as the maximm parking generation rate. I.e. if a shopping center were observed over five weekdays and the maximm occpied parking spaces over that period was 300, that wold be the maximm parking generation rate for that observation. Then the average of those maximm parking generation rates (sally per sqare foot) is the maximm parking generation rate for that land-se category. 8
and recharge to grondwater spplies by 1.1 acre-feet per year with a present vale loss of $16,400 in water vale. Polltant air emissions also occr throgh the lifecycle of a parking lot (EPA (2008)). Asphalt cement plants emit particlate matter, nitrogen oxides, slfr oxides, carbon monoxide and carbon dioxide dring the manfactring process. The activities associated with constrction and maintenance of parking lots also generate emissions in the form of dst, fmes and eqipment and vehicle exhast. The after effects of parking lot constrction, sch as fewer trees and less vegetation de to clearing also lead to higher amonts of carbon dioxide in the air. In addition, parking is generally constrcted of dark asphalt that, together with other bilt p dark srfaces generates an rban heat island effect. Akbari et al. (2001) estimates that the rban heat island effect in Los Angeles increases power consmption by 1-1.5 GW at a cost of $100 million per year. Also, the higher temperatres de to the rban heat island effect increases ozone (smog) prodction. Taha (2008a,b) shows that replacement of low-albedo srfaces sch as parking with vegetation can reslt in significant decreases in ozone levels 3. The Analytical Model This section describes the featres of the analytical model we will se to examine the impacts of minimm parking reqirements on strctral density, amont of land developed and type of parking spplied. We also provide and interpret key eqations associated with the developers problem in the absence and presence of minimm parking reqirements. Complete derivations are provided in Appendix A. 3.1. Model Assmptions Office-Commercial Rents Sppose that the office-commercial-space bid rent in a given location is represented by: 10 10 The office-commercial bid rent represents the maximm willingness-to-pay by a firm for commercial-office space in a bilding at particlar location within the city. In this paper we focs on the behavior of developers and therefore, we do not derive analytically eqation (1). However, a demand model for office space, see for example Sivitanido and Wheaton (1992), wold sggest that the qantity demanded is a fnction of rent, a firm s otpt, and the amont of office space it ses per worker. If the otpt market is competitive, in the long-rn a firm s profit is zero. From this latter condition we can ths determine the eqilibrim rent per nit of floor space (1). Moreover, 9
B = f ( N, A) (1) where, B is the office-commercial rent per nit of floor space, N is total parking spaces and A represents a vector of amenities associated with the location. Eqation (1) ths sggests that parking spaces are capitalized into bid rents 11. Becase we focs on a single location the vale of A is fixed. We assme that (1) is concave in its argments. Parking space can take two forms: srface parking and ndergrond parking. Srface parking refers to lots directly on land and ndergrond parking consists of strctred parking nder mlti-story bildings. 12 Both forms are assmed to be perfect sbstittes from the tenants perspective and therefore, total parking spaces are represented by: N = N s + N (2) where spaces. N s is nmber of srface parking spaces and N is nmber of ndergrond parking Bilding Technology Office-commercial floor space is prodced according to a strictly concave, constant-retrns prodction fnction, H = f ( K, L), where K is capital sed to prodce floor space and L is the amont of land physically covered by K (referred to sbseqently as covered land ). The intensive form of this prodction fnction is written as h (S), where S is capital per nit of covered land or strctral density and h satisfies h > 0 and h 0. h (S) represents officecommercial total floor space per nit of covered land. We assme that L is fixed. // < for simplicity, we assme that bid rents reflect prevailing market prices for office floor space, so that the two are synonymos. Ths, the price of an office bilding wold be the prodct of the bid-rent and total office floor space. 11 Bilding characteristics can inflence the bid rent for office-commercial space at a particlar location. Rents are higher for bildings with greater total sqare footage, more floors and parking space. A higher total sqare footage may be also indicative of bilding amenities (e.g. restarants), face-to-face agglomeration economies or shopping externalities. Office-commercial bid rents are also inflenced by locational factors sch access to the central bsiness district, access to srface and air transportation (freeways and major airports), access to good schools and level of crime. For empirical evidence on the determinants of office-commercial bid rents see Sivitanido (1995) and Bollinger et al. (1998). 12 Other types of parking facilities inclde on-street parking, off-street parking and strctred parking in mltistory bildings. On-street parking consists of parking lanes provided within pblic road rights-of-way. Off-street parking are parking facilities on their own land, not on road rights-of-way. Strctred parking (also called parkades or ramps) are parking facilities in or nder mltistory bildings. It is not ncommon to find strctred parking in downtown areas since land costs are very high. For simplicity we only analyze the cases of srface and ndergrond parking (strctred parking nder mlti-story bildings). However, or reslts on ndergrond parking can also be extended to other types of strctred parking. 10
Parking Costs Parking costs differ depending on the type of parking facility provided. Constrction costs (exclding land) per space of srface parking are lower than the constrction costs per space for ndergrond parking (Hnnictt (1982)). 13 However, srface parking precldes alternative ses of land and hence its total costs are the sm of total constrction costs and land costs ( C N ) ): s ( s C ( N ) = N ( p K p l ) (3) s s s k + where p l and l p k are the exogenos prices of land and capital, K is the fixed amont of capital per srface parking space and l is the fixed amont of land per srface parking space. We assme that no additional land is necessary for ndergrond parking since it will be bilt below the office-commercial bilding. Therefore, total costs for ndergrond parking reflects mainly its constrction costs: C ( N, S) = N pk K( N, S) (4) where K ( N ) is the capital cost reqirement per ndergrond parking space and is assmed 2 K( N, S) S to be a convex fnction with > 0. The reason is becase as more ndergrond parking is added more nits of capital are necessary to fortify the bilding strctre and to provide vertical-transportation reqirements. Srface Parking Externalities Srface parking generates mltiple environmental externalities. Let L s be the amont of impervios land de to srface parking and E L ) denote the external costs associated with impervios srfaces. We assme that E L ) is linear in the amont of land allocated to srface parking: ( s ( s E ( L s ) = el s (5) where e > 0 is the nit of land external cost and Ls = N sl. Parking Reqirements 13 Constrction costs (exclding land) average abot $1,600 per space for srface parking and $20,000 or more per space for ndergrond parking (Hnnictt (1982)). 11
The city government imposes a minimm parking reqirement expressed as nmbers of parking spaces per sqare foot of gross bilding floor area: N alh(s) (6) where 0 < a < 1 is a parameter imposed by the city government. 14 Floor-Area-Ratio (FAR) restrictions There is also an pper limit on the sqare footage of office-commercial space per nit of land sch that: h( S) hˆ( Sˆ) (7) where ĥ is the FAR limit per nit of land and Ŝ is the total strctral density associated with ĥ. 3.2 Type of parking provided Differentiating (1) with respect to N and N, yields the same marginal benefit for both types of parking spaces: B( N, A) B( N, A) = s N s As a reslt, the developer determines which type of parking space will be provided based on the marginal cost. The private marginal cost per space of ndergrond parking is given by: p K + p N k k K( N, S) The first component of (9) is the marginal cost for the additional ndergrond parking space, which is the same for all parking spaces in the strctre. The second component of (9) is the inframarginal cost associated with the additional ndergrond space. Note that the marginal cost of ndergrond parking increases with the height of the bilding becase of the costs of providing vertical transport and spporting a heavier bilding. (8) (9) 14 Note that H(K,L)=Lh(S). Since H(K,L) is concave and homogenos of degree one, it follows that H(K/L,1)= h(s). Also, the vale of a can be lower, eqal or higher than the vale of a that wold exist in an nconstrained market. However, for the prpose of exploring the effects of minimm parking reqirements we consider the case where a is sch that the constraint is always binding and ths, affects the market eqilibrim. If for example, a =1/200sft, it means that developers are reqired to provide one parking space for each 200sft of gross floor area. 12
The private marginal cost per space of srface parking is given by: pk K + pll (10) The first component of (10) is the marginal cost for the additional srface parking space and the second component is the marginal cost of land. In contrast to (9), the marginal cost of srface parking is constant and ths, it does not vary with the amont of parking space. Comparing (9) and (10), the marginal cost of ndergrond parking is greater than the marginal cost of srface parking if the cost of land is small relative to the degree of diminishing marginal retrns: K( N, S) pk N > p l l Eqation (11) sggests that srface parking is more efficient if the price of land is relatively low. 15 For a sfficiently high price of land, the developer provides ndergrond parking instead of srface parking. 16 Given that land prices are typically very high in downtown areas, it is not srprising that most parking bndled with office-commercial development in Central Bsiness Districts (CBD) is strctred parking. In contrast, low-density office-commercial strctres with large srface parking lots sch as shopping malls are mostly fond in sbrban areas where the price of land is lower. Next we specify the developer s problem separately for the sbrbs and the central city taking into accont the FAR restriction and Minimm Parking Reqirements (MPRs). 17 Given (11) 15 Strctred parking typically becomes cost effective when land prices exceed abot $1 million per acre. 16 If the marginal benefit of ndergrond parking exceeds the marginal benefit of srface parking becase of differences in the costs and speeds of walking and elevator travel, then the developer may provide ndergrond parking even if the price of land is relatively low: B K( N, S) h( S) > pk N pl l s If is large enogh to offset the diminishing marginal retrns from ndergrond parking, then s ndergrond parking will be provided. Note also that a developer cold choose a combination of srface and ndergrond parking spaces that minimizes total costs of providing parking spaces. However, in this paper we focs only on bondary soltions and we assme that a developer can only provide ndergrond or srface parking spaces, bt not both types. 17 In addition to these land se reglations, other parking and zoning policies may be in place that may inflence the spplies of parking and office-commercial floor-area. Sch additional parking policies inclde shared parking, sitespecific redctions and cash in lie. Among non-far limitations on bilding size and form are set-back reqirements, open space reqirements, height limits and blk reglations, which reqire minimm sky exposre to the street. Becase an analysis of the interactions between all these policies and FAR and MPR reglations wold be very cmbersome withot adding to the discssion, we do not address them in or analysis. We hope to revisit some of these limitations in ftre research. 13
the preceding discssion, we assme that CBD developers provide ndergrond parking and sbrban developers provide srface parking. 3.3. Central Bsiness District (CBD) The developer s problem in the central city is to choose the level of strctral density and nmber of ndergrond parking spaces that maximize his profits per nit of covered land taking into accont the FAR restriction and MPRs: 18 Max B( N, A) h( S) pk S p N, S h( S) h( Sˆ) s. t. N alh( S) l p N K( N k, S) For the sake of expository convenience, we represent the Khn-Tcker conditions for an interior soltion for problem (12) in Table 1, where λ 1 is the shadow price associated with the FAR constraint and λ 2 is the shadow price associated with the MPR constraint. 19 All the firstorder conditions are evalated at the optimm levels. For fll details see Appendix A. There are three main conclsions that we can take pon examining table 1. First, becase parking is capitalized into office-commercial rents, this will encorage developers to volntarily spply parking whenever the reslting revene will cover its costs, even in the absence of MPRs (( λ 1 = 0, λ 2 = 0 ) and ( λ 1 > 0, λ 2 = 0 * (12) )). If the price of additional ndergrond parking B( N, A) * C ( N, S ) ( h( S ) ) is at or above the marginal cost of providing it ( ), there is no * * 18 Eqation (12) also implies that total floor space and parking space are bndled and rented as a package to the tenants of a bilding. For example, in nearly all bildings in Los Angeles today, parking is inclded in the price or rent of the nit. Tenants do not have the option of nbndling the cost of parking from their prchase or rent. The main exception is in the Downtown area where some bildings do not inclde parking in their rental rates. 19 This shadow price (the Lagrange mltiplier) provides a measre of how a relaxation in the constraint will affect the developer s profit per nit of covered land. Ths, a high vale of λ i indicates that the profit per nit of land cold be increased sbstantially by relaxing the constraint. In contrast, a low vale of λ i indicates that there is not mch to be gained by relaxing the constraint. When λ = 0, the constraint is not binding. i 14
reason why developers wold not provide it on their own in downtown areas ( N 0 ). Ths, even in the absence of MPRs, developers can offer a bndle of (parking spaces, floor space) as a strategy to maximize profits. 20 The main effects of MPRs (see both cases where λ 2 > 0 ) are that parking spaces will be priced below the cost of providing them (for example C ( N, S ) B( N, A) > h( S ) ) and total spply of parking will be above its market determined eqilibrim level ( N > N ). * Second, parking reqirements may also case serios problems in the office-commercial floor space market. When MPRs bind ( λ 2 > 0 ), the excess ndergrond parking reslts in a deficit for the developer of a new bilding. This indced deficit constittes an indirect tax on λ2l h( S ) bilding sqare footage ( > 0 ). As a reslt, this creates a disincentive to high- N S density development ( * S > S * > ) becase it imposes an extra wedge between the marginal h( S ) revene gain from additional bilding sqare footage ( B( N, q, A) ) and the marginal S C ( N, Sˆ) constrction costs ( pk + ). Since the marginal cost of providing more parking spaces S at a site sally increases dramatically for ndergrond strctres 2 C ( N, S) C ( N, S) ( > 0, > 0 ), this parking tax is also higher for larger bildings. 2 S 2 Finally, FAR restrictions also constitte a tax on bilding sqare footage which leads to * FAR bilding heights smaller than those in nconstrained markets ( S > S ). Based on crrent constrction costs, a developer might want to bild a taller strctre than allowed by density controls and provide the reqisite parking in order to maximize retrns, even in the absence of parking reqirements ( λ 1 > 0, λ 2 = 0 ). Minimm parking reqirements may nevertheless drive the total sqare footage allowed and potentially inhibit density below what the FAR limit permits, in particlar when the amont of 20 In this paper, decisions to bndle parking into rents are freely-made private decisions, not forced by excessive parking reqirements. Ths, even in the absence of MPRs, bndling of parking remains persistent. A possible reason that may prompt developers to bndle parking is the absence of priced parking nearby. Examining nderlying cases of persisting bndling is otside the scope of this stdy. 15
parking is constrained de to site geometry (size and shape) and site topography (slopes and poor soil conditions). Site geometry and site topology may make the reqired parking not physically FAR h( Sˆ ) fit on to a site and increase sbstantially parking costs and ths, λ 2. In this sitation λ 1 S may be smaller than λ 2 N L h( S ) and ths, S S ˆ > S. However, in areas with stringent limits on bilding height and where parking can feasibly be provided ndergrond, parking reqirements may not be the greatest constraint on densities. When both reglatory restrictions bind ( λ 1 = 0, λ 2 = 0 ), it may be the case that the FAR limit pshes densities frther down than MPR. Or reslts ths show that minimm parking reqirements have conterprodctive reslts in downtown areas becase they try to solve a problem in the transportation market that is only indirectly related to the office-commercial floor space market. Becase minimm parking reqirements increase the costs of new development, these minimm standards tend to decrease the potential office-commercial density of new projects which is conter the objective of most cities to promote downtown density to increase agglomeration economies and control for rban sprawl. In addition, minimm parking reqirements may also conteract other local policies designed to encorage development in areas easily accessible by pblic transit as well as coomise the feasibility of mass-transit investments in certain downtown areas. 3.4. Sbrban Areas The developer s problem in the sbrbs is to choose the level of strctral density and nmber of srface parking spaces that maximizes his profits per nit of land covered taking into accont the FAR restriction and Minimm Parking Reqirements (MPRs): Max B( N NS, S s. t. S, A) h( S) p h( S) h( Sˆ) N S alh( S) k S p l N s ( pll + pk K ) L (13) Since the capital-to-land ratio tends to decrease with distance from a CBD, it follows that a FAR restriction will bind in the central part of a city, where the capital-to-land ratio wold normally be high, being nonbinding farther from the center (Bertad and Breckner (2005)). Ths, we focs or discssion on the cases where only the MPR restriction is binding. The 16
Khn-Tcker conditions for an interior soltion for problem (13) when the FAR restriction is not binding are presented in Table 2. Again, λ 1 represents the shadow price associated with the FAR constraint and λ 2 is the shadow price associated with the MPR constraint. All first order conditions are evalated at the optimm. For fll details see Appendix A. Parking spply imposes the negative externalities discssed in Section 2. Upon examining Table 2, it is clear that whenever private decisions do not take into accont these negative externalities, too mch srface parking is provided even in the absence of MPR. 21 Notice that the socially optimal nmber of srface parking spaces maximizes π en s l s + L, where π s is profit per nit of land covered. The first order condition for the social optimal srface parking spaces, N o s, is given by: B( N, A) ( pl + e) l h( S) = L s + s p k K (14) Comparing (14) and the first order condition in absence of reglatory constraints ( λ 1 = 0 and λ 2 = 0 ) one notices that becase of the negative externalities associated with impervios srface, the social marginal cost of srface parking ( ( pl + e) l + p L k K ) is higher than the private marginal cost ( ( o N s p l pk K L l + ). As a reslt, the socially efficient amont of land in srface parking * l ) is less than the privately optimal amont ( l ). Ths, if left to the market the spply of srface parking is likely to be excessive. 22 Not only are socially beneficial ses of srface N s 21 Or argment ignores the free rider congestion effect. The free rider problem with respect to private downtown parking provision is a potentially serios one and is the jstification for MPRs in the first place. Therefore, one wold need to consider this explicitly before claiming that too mch land is allocated to srface parking. However, to the extent that most parking in downtown areas is ndergrond and or srface parking analysis is related to sbrban areas where this problem is not so severe, we have decided not to model this effect in or sbrban developer problem since it wold make the analysis cmbersome withot additional insights. Moreover, Ctter and DeWoody (2010) indicates that, to the extent congestion effects are capitalized into real estate vales, the externalities are only significant for commercial parking garages and are not large for on-site private parking. 22 Government can encorage the social amont of both srface and ndergrond parking throgh the imposition of a Pigovian tax, where the magnitde of the tax is set eqal to the marginal external effect at the efficient allocation. This wold reqire imposition of a per nit tax on srface parking, where the magnitde of the tax is set eqal to the dollar vale of the environmental external costs. Alternatively, the government can also set a maximm parking reqirement for srface parking, where the reqirement wold be eqal to the social optimm. However, even if markets shold be able to provide abot the right amont of parking, there are inadeqate economic incentives for qality, althogh planners cold address some of the concerns over the qality of parking by reqiring better design of lots and parking strctres (Mkhija and Shop (2006)). 17
parking land foregone, bt becase land is paved, it also increases storm water rnoff and has other negative environmental impacts. In the context of the monocentric city model, the spatial area of the city can be fond by adding total land in office-commercial se and total land in srface parking. Given that o Ns ln * s l < the market eqilibrim is characterized by inefficient spatial expansion of the rban area, providing a basis also for criticism of rban sprawl. 23 Like in downtown areas, MPRs also enforce an overspply of parking in sbrban areas * ( S N S N > ) which, intensifies the external costs associated with impervios srface coverage: s e ln > eln > eln *. Becase minimm parking reqirements increase the cost of development, s o s densities are also lower in sbrban areas compared to the nconstraint market otcome * ( S > S ). Note that in eqilibrim, the shadow price associated with the MPRs satisfies: B( N S, A) 2 pll + pk K Lh( S ) a h( S ) = λ2 (15) S From (15) land ses that have higher parking reqirements sch as retail verss warehoses and ths have a higher vale of a, are expected to have also a higher shadow price, λ 2. Moreover, if MPRs do not bind ( λ = 2 0 ), then the marginal vale of parking shold be eqal to the marginal vale of additional land pls marginal parking constrction costs. If parking constraints bind, the marginal parking se vale shold be less than the land vale pls constrction costs ( λ > 2 0 ). 3.5. Testable Hypothesis The predicted theoretical relationships between marginal parking se vale, marginal land vale, and marginal bilding area vale give s the testable hypotheses for the empirical portion of the paper. With proper data, we can estimate the shadow price associated with MPR and make inferences abot the nderlying eqilibrim. For example, in the case of srface parking, the shadow price can be calclated with eqation (15). Calclating the extensive margin vale of land is simple once we have the constrction cost data and the parking reqirements data. However, the additional vale of a property from adding an additional parking space cannot be 23 Other cases of rban sprawl can be fond in Breckner (2000) and Bento, Franco and Kaffine (2006). 18
calclated explicitly. It mst be inferred sing hedonic regression techniqes to estimate the marginal contribtion of each type of parking space to the price of the office-commercial bilding. Or model also implies that the marginal revene from additional bilding area shold eqal marginal constrction costs in the absence of binding parking reqirements, bt be greater than marginal constrction costs if parking constraints bind. We do not test this second hypothesis in the present version of this paper, bt may examine it in the ftre. 4. The Empirical Model The empirical part of this paper focses on the office-commercial-indstrial property market within sbrban areas of Los Angeles and on srface parking lots. The sbrban market was chosen becase, as or analytical reslts sggest, in downtown areas FAR restrictions are likely to bind which in trn can inflence development density and the amont of land allocated to srface parking and parking density per acre. 24 Moreover, srface parking in downtown LA also reslts from a speclative decision process. 25 Therefore, isolating the effect of MPR on the amont of land allocated to parking in the bilt-p areas of Los Angeles may be tricky. In frther research we hope to investigate the claims by critics that parking restrictions in fact prevent developers from attaining the maximm FAR. In addition, or analytical reslts also sggest that srface parking is more efficient if the price of land is relatively low. Given that land prices are typically lower in the sbrbs compared to downtown areas it is not srprising that most srface parking occrs in the otskirts of the city. 24 In some cases, zoning reglations inclde bons programs that permit developers to exceed the base maximm FAR if they inclde certain amenities sch as affordable hosing. 25 Landowners may face ncertainty that increases the vale of waiting to develop the parcel or may anticipate higher profits from developing at a later date. If ftre needs are ncertain and land se is not easily adaptable (becase the developed strctre is drable), it may maximize the landowner s retrn to keep land ndeveloped or nderdeveloped (for example developed with a cheap land se sch as srface parking that allows land to be pt to some crrent se withot reqiring significant demolition costs in the ftre) and reserve it for the more highly valed ftre se than to develop it today. The landowner who keeps land or development capacity off the market is ths taking an option on ftre development. It has been shown that ncertainty reslts in a price premim for vacant or ndersed land (after acconting for demolition costs) and/or that it redces crrent development, sggesting that investors recognize the vale of real options in real estate development (Sivitanido and Sivitanides (2000), Cnningham (2006), Grenadier (1996), Holland et al. (2000)). It is also possible that ndersed land in downtown LA is de to instittional or reglatory barriers and not jst landowners calclation of expected retrns. Zoning ordinances, bilding codes and historic preservation obligations may keep land from being developed. 19
We estimate the marginal vales of parking and lot area with spatial and non-spatial methods from Los Angeles area non-residential property sales. In order to simlate the marginal on-site parking and parcel vales we estimate a spatial error model for different land-se categories becase parking reqirements often differ across land se types (ITE 1985.) 4.1. Hedonic Price Model The bid-rent fnction given by (1) is the starting point for or hedonic price fnction specification. This eqation implies that office-commercial bildings can largely be considered as bndles of attribtes that cannot easily be repackaged to sit individal preferences. The attribtes typically evalated by byers in the hosing market inclde not only strctral characteristics of the properties, bt also neighborhood attribtes linked to the properties. 26 We se specifications of the following form in or hedonic price regression: J J LP it = β 0 + β j LX ijt + δ j D ij + α t Y t + ε i (16) j=1 j=1 T t=1 In (16) LP it represents the log of the sale price of property i at time t, t =1997 2005. X ijt is a vector containing j continos property characteristics of property i in time t, and D, are j i j binary property characteristics of property i. 27 property was last sold. β 0 is the intercept regression coefficient and Y t is a dmmy variable indicating the year the β j, δ j and α t represent the regression coefficients associated with the explanatory variables. The error term isε i. Note that or logarithmic form is an approximation to the nonlinearities sally involved in the soltion of models sch as the one presented in section 3. 4.2. General Spatial Model Spatial econometric techniqes are now common in estimating the determinants of property prices becase of the likelihood of nobserved spatial relationships. This is becase nearby properties are likely to have similar nobservable characteristics (Bell and Bockstael (2000), 26 Neighborhood attribtes may inclde physical characteristics of the neighborhood, the socio-economic characteristics of the local residents, pblic service provisions, and environmental amenities. 27 The log transformation of the continos property characteristics is consistent with past hedonic literatre on hosing and commercial/retail/indstrial properties (Hodgson et al. (2006)). In addition, Ctter and DeWoody (2010) shows that this specification is sperior for a similar data set. 20