Economics of Transportation

A review of the economics of parking

Eren Inci n

Sabanci University, Faculty of Arts and Social Sciences, Orhanli/Tuzla, 34956 Istanbul, Turkey

a r t i c l e i n f o

Article history:Received 27 May 2014Received in revised form30 September 2014Accepted 15 November 2014

Keywords:CongestionParkingParking policyPricingTraffic

a b s t r a c t

This paper reviews the literature on parking with an emphasis on economic issues. Parking is not justone of the most important intermediate goods in the economy; it is also a vast use of land. Manytheoretical and empirical papers analyze the quantity and pricing of parking by concentrating onparticular aspects of the issue. The aspects covered in this review are cruising for parking, spatialcompetition, (minimum and maximum) parking requirements, parking pricing and road pricing in thebottleneck model, and temporal-spatial pricing. Various forms of parking, including residential parking,shopping mall parking, and employer-provided parking, are also reviewed before identifying under-studied topics that should be on the research agenda.

& 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Economists come up with ideas to deal with imperfect markets,but the market for ideas in economics can be as imperfect as anyother market. What is believed to be of general interest to econo-mists is sometimes not important in real-world transactions; andwhat can have a great impact on the welfare of the many does notalways attract the level of attention it deserves. The economics ofparking is an example of the latter imperfection. Few economistsdevote full-time effort to analyzing parking markets, even thoughthe economics of parking has a lot to say about how to improve thequality of urban life. There is also a dearth of parking studies intransportation science, transportation engineering, and urban plan-ning. Despite the facts that cars are parked 95% of the time (Shoup,2005b, p. 624) and that vast amounts of land are used for parking(Jakle and Sculle, 2004), more ink has been spilled trying to dealwith the problems caused by cars when they are in motion thanwhen they are parked. In fact, cars create perhaps less visible butequally serious problems when parked, as Shoup (2005b, p. 625)points out in his landmark book The High Cost of Free Parking.

Most transportation activities are initiated by getting into aparked vehicle and terminated by parking it again. This makesparking one of the most important intermediate goods in themodern market economy, after money and credit cards (Haskerand Inci, 2014). Realizing this should be sufficient to help usunderstand why an economic analysis of parking is vitally impor-tant. As Arnott and Inci (2006) state, early work on the problem

treated parking only as a cost added on at the end of a trip, which, asa fixed cost, does not really affect decisions at the margin. Later workon the pricing of parking, however, has repeatedly shown that thisapproximation limits generality (see, e.g., Glazer and Niskanen, 1992;Anderson and de Palma, 2004, 2007; Arnott and Inci, 2006, 2010). Inaddition to being an important intermediate good, parking is also amajor use of land in any country, city, or town. All vehicles, whetherparked or in traffic, occupy space. It is eye-opening to visualize thetotal amount of land that is taken up by parking. In the United States,it is at least as large as the total land area of the state ofMassachusetts, and in Europe, it is at least one-half of the entireland area of Belgium. Now consider how the mispricing of parkingcan distort land use, car usage, and the pricing of other goods.

Only 79 years have passed since a driver fed a parking meter forthe first time.1 This review aims to increase awareness about thehigh potential for work on parking to immediately improve city-dwellers' welfare.2 The existing work in economics, transportationscience, transportation engineering, and urban planning looks atparking from various angles. In this review, I concentrate only onits economic aspects. These aspects address mainly the issues

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ecotra

Economics of Transportation

http://dx.doi.org/10.1016/j.ecotra.2014.11.0012212-0122/& 2014 Elsevier Ltd. All rights reserved.

n Tel.: þ90 216 483 9340; fax: þ90 216 483 9250.E-mail address: [email protected]

1 The first parking meter was installed in Oklahoma City on July 16, 1935. Theoriginal motivation of Carl Magee, the inventor of the parking meter, was toencourage parking turnover, not to collect revenue.

2 Arnott (2011) provides another discussion of the economics of parking inwhich he lists some empirical regularities and briefly reviews the existingliterature. He also applies the standard transportation microeconomic theory toparking and touches on selected issues in parking policy, one of which is parkingfreeze (i.e., maintaining the supply of parking at the same level as it were prior to aspecified date), which I do not cover in this review. Other than that, I provide amore extensive and recent literature review.

Please cite this article as: Inci, E., A review of the economics of parking. Economics of Transportation (2014), http://dx.doi.org/10.1016/j.ecotra.2014.11.001i

Economics of Transportation ∎ (∎∎∎∎) ∎∎∎–∎∎∎

www.sciencedirect.com/science/journal/22120122

www.elsevier.com/locate/ecotra

http://dx.doi.org/10.1016/j.ecotra.2014.11.001



mailto:[email protected]






related to the quantity and price of parking. Various distortions inthe parking market need to be taken into account to determineoptimal parking quantity and price. These distortions naturallydetermine the scope of this review.

If curbside (i.e., on-street) parking demand exceeds curbsideparking supply, some drivers cannot immediately find a vacantparking space; thus cruising for parking will emerge, whichimposes external costs on all drivers by increasing congestion.Section 2 reviews the papers that analyze this important phenom-enon. Another distortion is the parking garages' market power,which stems from the fact that they are discretely spaced through-out the city. Section 3 reviews the papers focusing on spatialcompetition in the parking market and parking garages' exercise ofmarket power. In most cities, zoning regulations specify howmuchparking has to be supplied by each land use. Because theseregulations are usually adopted ad hoc from city to city, theysignificantly distort parking supply and thus land use. Section 4reviews the papers on such zoning regulations. Yet another sourceof distortion is underpriced (in fact, for most cities, unpriced)congestion. Section 5 reviews the literature linking parking pricesto road prices by embedding parking into bottleneck models.Vickrey's (1954) wisdom was to charge different parking pricesacross time and space, which was less feasible given the technol-ogy of that time. Today we see a general movement toward suchtemporal-spatial pricing. Section 6 outlines the efforts in thatdirection. Underpriced parking is particularly an issue for somespecial forms of parking. Employers often provide parking toemployees at no cost, shopping malls typically provide parkingto their customers for free, and cities provide parking to residentsat nominal prices lower than market prices. Section 7 brieflyreviews the work on these parking forms. Section 8 identifiessome under-researched topics.

2. Cruising for parking

One of the most studied topics in the economics of parking is thephenomenon of cruising for parking, which is a typical example of thetragedy of the commons. Parking spaces are overdemanded if they areunderpriced (or free), and no one cares about his contribution toothers' travel time while cruising for parking slowly around the block.Shoup (2005b, Ch. 14) recites from his field study in the 1980s thatdrivers lose about 100,000 hours (over 11 years) while cruising forparking in a given year on a 15-block business district near the UCLAcampus. Shoup (2006) reports the findings of 16 different studiesdone between 1927 and 2001 in congested downtown areas fromaround the world. According to these studies, between 8 and 74(on average 30) percent of all cars in traffic are cruising for parking,and they spend between 3.5 and 14 (on average 8.1) minutes on thatactivity. Cruising for parking is an inefficient transport activity. Carsslow down traffic while they are cruising for parking and thuscontribute to traffic congestion disproportionately more than cars intransit. Cars cruising for parking increase fuel consumption andcontribute to air pollution via carbon emissions. They may evenincrease the probability of traffic accidents. How to decrease cruisingfor parking has been at the top of the agenda in the literature.

Researchers have constructed a series of parking models toanalyze the economic effects of cruising for parking. It is in fact asearch externality. If there are no available parking spots near adriver's destination, he will search for one. There are many ways tosearch for a parking spot.3 However the driver searches, this

activity involves at least some time costs. Search time is anincreasing function of how many others are searching and howlong they park (see, e.g., Glazer and Niskanen, 1992; Inci andLindsey, 2014). The presence of cruising for parking shows thatdrivers' willingness to pay exceeds the price of parking; thus astandard rationing problem arises, which may result in substantialwelfare losses. The drivers cruising for parking usually drive moreslowly than cars in transit; thus they slow down traffic, whichimposes external costs on all drivers. Any efficient parking pricingscheme should internalize this externality.

Arnott and Inci (2006) developed the first “bathtub model”4 ofdowntown parking, via which they analyze the effects of cruisingon traffic and provide parking pricing recommendations toremedy the problems it causes. They envisage a spatially homo-geneous downtown area, which simplifies the analysis by makingthe density of traffic uniform over the space. One can imagine agrid network of streets like that of Manhattan. Curbside parking isthe only option and drivers are identical. A driver enters thedowntown area, drives to his destination, and immediately parksthere if there is an empty parking spot. If there are no emptyparking spots, he cruises around the destination block until hefinds one.

In this bathtub model, there are three pools at any time. Arnottand Inci (2006) analyze the steady state of the model. Cars firstenter into the pool of cars in transit, represented by T per unit area,by driving to their destinations. The inflow rate into the in-transittravel pool is given by demand function D per unit area. Becausethe in-transit trip length is m and the travel time per mile is t, theoutflow rate from the in-transit travel pool is T/mt. Those who exitthe in-transit travel pool enter into the pool of cars cruising forparking, represented by C per unit area. Thus, the outflow ratefrom the in-transit pool is also the inflow rate into the pool of carscruising for parking. Because there are P parking spaces per unitarea and each car parks for a fixed visit of l hours, the exit ratefrom the pool of cars cruising for parking is P/l. Finally, cars exitingthe pool of cars cruising for parking find a parking spot and thusenter the pool of parked cars. They stay in the parking space for lhours and then exit the downtown area. The exit rate per unit areais also P/l.

Demand D is a function of expected full price of a trip F, whichequals in-transit travel time costs, given by ρmt, where ρ is thevalue of time, plus cruising-for-parking time costs, given byρðCl=PÞ, plus the total parking fee, given by fl, where f is the hourlyparking fee:

F ¼ ρmtþρClPþ fl: ð1Þ

In-transit travel time per mile, t – in other words the congestionfunction – depends on the number of cars in transit per unit area, T,the number of cars cruising for parking per unit area, C, and thenumber of parking spaces per unit area, P: t ¼ tðT ;C; PÞ. It is assumed

3 One may drive around the destination block until a parking spot becomesavailable; one may drive farther away from the destination to locations where theparking demand is relatively lower and then walk to his destination; one may evenwait at the destination until someone leaves a nearby parking spot. Guo et al.

(footnote continued)(2013) analyze the parking search behavior from a behavioral economics perspec-tive. They compare the performance of a static game-theoretical model wheredrivers are completely rational with that of a model with rational and irrationaltypes in terms of parking search behavior. In particular, they take into accountoptimistic and pessimistic behavior in parking search. They calibrate their model byusing a genetic algorithm on video observations from some parking lots on auniversity campus and use this model to predict behavior on other parking lots. Itturns out that the behavioral model performs more accurately than the rationalgame-theoretical model. There is also an extensive operations research literatureon parking search (see, e.g., Teodorovic and Lucic, 2006 and the references therein).

4 See Arnott (2013) for a description of bathtub models. The “bathtub” analogywas coined by William Vickrey in an unpublished draft found after his death.Inspired by the hydrodynamic models, cars entering the traffic network aremodeled as water flowing into a bathtub, cars exiting from it as water flowingout of a bathtub, and the density as the height of the water in the bathtub.

E. Inci / Economics of Transportation ∎ (∎∎∎∎) ∎∎∎–∎∎∎2






that a car cruising for parking contributes to traffic congestion at leastas much as a car in transit.

In the steady state of this model, the inflow into the pool of carsin transit equals the outflow from that pool,

DðFÞ ¼ Tmt

; ð2Þ

and the inflow into the pool of cars cruising for parking (whichequals the outflow from the cars in transit) equals the outflowfrom the cars cruising for parking,

Tmt

¼ Pl: ð3Þ

Substituting for tðT ;C; PÞ and F yields two equations in T and C:

D ρmtþρClPþ fl

� �¼ TmtðT ;C; PÞ ð4Þ

TmtðT ;C; PÞ ¼

Pl: ð5Þ

Given P, the first equation gives the locus of T and C such thatdemand inflow equals parking outflow, so that the size of thein-transit pool remains steady. Arnott and Inci (2006) call this thedemand locus. The second equation gives the locus of T and C suchthat the outflow from the in-transit pool equals the outflow fromthe parking pool, so that the size of the cruising-for-parking poolremains steady. Arnott and Inci (2006) call this the cruising-for-parking locus. The intersection of the two loci gives the steady-state equilibrium.

One can easily transform the problem into a supply-demandspace, which is more relevant to this review. Fig. 1 does the job,which is drawn for a fixed curbside parking capacity. The counter-part of quantity in this analysis is throughput, which is defined tobe T/mt and is shown on the x-axis. Since cars cruising for parkingjust circle around the destination block, they do not contribute tothroughput (but they do contribute to flow). Trip price is shown on

the y-axis. As explained in detail by Arnott and Inci (2010), just asin the well-known traffic flow analysis without parking, the supplycurve here is also backward bending.5 However, throughputcannot be higher than parking turnover rate, and hence the regionof the supply curve above this constraint, which is called theparking capacity constraint, is replaced by the constraint itself. As aresult, we get a modified backward-bending supply curve, shownin bold in the figure. The intersection of the demand curve withthe (modified) supply curve gives various equilibria; in this caseE1, E2, and E3, where E3 corresponds to gridlock equilibrium withzero throughput and infinite trip price and thus cannot bedepicted. E4 would have been an equilibrium if the parkingcapacity were binding beyond the throughput level associatedwith E4, but in the current figure it is not. Arnott and Inci (2010)provide a detailed stability analysis of these equilibria along withtheir dynamics.

We can now show how cruising for parking can distort theparking market. Consider panel (a) of Fig. 2, which undertakes awelfare analysis of equilibrium E1 shown in Fig. 1. If the parkingcapacity constraint was not binding, the equilibrium would occurat the intersection of the user cost curve and the demand curve.But it does bind, and thus the equilibrium occurs at the intersec-tion of the parking capacity constraint and the demand curve.With priced (but underpriced) parking, the user cost shifts upwardbut the equilibrium remains at E1. As a result, some cruising forparking will occur, although some parking revenue is collected.Thus, the parking pricing shown in this panel is still associatedwith welfare losses. As shown in panel (b) of Fig. 2, if thegovernment increases the parking fee such that the user costcurve shifts upward to cut the demand curve at its intersectionwith the parking capacity constraint, the equilibrium price and theconsumer surplus remain unchanged but the parking revenueincreases. Moreover, cruising for parking is just eliminated at thispoint, which is where the optimal parking fee is obtained. So, byincreasing the parking fee to the point where cruising is elimi-nated, the government converts welfare losses dollar for dollar toparking fee revenue and collects revenue with no burden at all.6

This line of reasoning is extended by Arnott and Rowse (2009) andArnott and Inci (2010) to show how double or even tripledividends emerge from increasing parking fees.7

Arnott and Inci's (2006) model contains only curbside parking.Calthrop and Proost (2006) concentrate on the optimal regulationof curbside parking spaces when off-street garage parking is alsoavailable as a perfect substitute. They underline the importance ofthe dependence of optimal regulation on the pricing of the rivalfacility, in this case garage parking. If the garage parking fee ishigher than the curbside parking fee, there will be wasteful searchfor curbside parking spaces, which is a form of a dissipativeactivity to equilibrate full prices of both modes of parking. If, onthe other hand, the garage parking fee is lower, the curbside willbe underused and thus there will be excessive supply costs. As aresult, with constant returns to scale and a competitive parkinggarage market, the optimal regulation equalizes the full prices.Calthrop and Proost (2006) also show that a particular time

Parking CapacityConstraint

Trip Price

Throughput

DemandSupply

0

E1

E2

E4

Fig. 1. Demand and supply analysis when there is curbside parking.Note: An equilibrium type that cannot be shown on the diagram is E3 with zerothroughput and infinite trip price.

5 The easiest way to understand why the supply curve is backward bending isto think about the extreme case of zero throughput. It can either correspond to thecase in which there are no cars on the road or there is a gridlock. In a similarfashion, there are two velocities associated with a given level of throughput for allother levels of throughput, except for the capacity throughput. Hence, the supplycurve is backward bending.

6 At this point, the reader should also see Shoup (2004), which reviews variousother arguments about why parking is “the ideal source of local public revenue.”

7 These models treat parking search as a deterministic process. In reality,drivers face uncertainty about unoccupied parking spaces, which results in somelevel of cruising even in a social optimum. Such concerns are considered toward theend of this section.

E. Inci / Economics of Transportation ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 3






restriction policy, according to which one can park for free up to afixed amount of time and then has to depart, is inferior to optimalprice regulation in terms of social welfare. They also analyze aStackelberg model where the government is the market leader andshow that the result of matching the curbside parking fee nolonger holds.

Arnott and Rowse (2009) extend the base model in Arnott andInci (2006) by introducing garage parking, which can be usedwithout contributing to traffic congestion. The garage operatorsare assumed to be operating competitively, and thus the garageparking fee, c, is equal to the garage parking resource cost.Moreover, by assumption, the curbside parking fee is lower thanthe garage parking fee (i.e., f oc). As in Calthrop and Proost (2006),the excess demand for curbside parking is rationed throughcruising for parking until the full prices of the two parking modesare equalized to each other:

cl¼ flþρClP

; ð6Þ

where the left-hand side is the total parking fee paid by a driverwho parks in the parking garage for a duration of l hours and theright-hand side is the total curbside parking fee paid by a driverwho parks curbside for the same duration plus his time costs ofcruising for parking. One immediate implication of this equaliza-tion is that there will be cruising for parking in equilibrium as longas there is a differential between curbside and garage parking fees.

This model fits nicely into US cities where garage parking feesare considerably higher than curbside parking fees. Nevertheless,the distortion here is caused only in the curbside parking marketas garage operators are assumed to be exercising marginal costpricing. If the market power of parking garages is incorporatedinto the model, the garage parking fees will probably be higher,and thus the differential between them and curbside parking feeswill also be higher. As a result, the levels of cruising for parkingwill increase even more. I review the work analyzing the marketpower of garage operators in Section 3.

To simplify the analysis, Arnott and Rowse (2009) assume thattrip demand is perfectly inelastic. Arnott et al. (2013) go one step

further by allowing trip demand to be elastic. Their extensionfurther modifies the backward-bending supply curve, and thusseveral different types of equilibria emerge. Arnott et al. (2013)work on the welfare properties of these parking equilibria andequilibrium dynamics. More importantly, they analyze curbsideparking capacity choice in detail when garage parking supple-ments curbside parking and when it does not, and when curbsideparking is priced efficiently and when it is not.

Arnott–Inci–Rowse type bathtub models provide some generallessons. First of all, cruising for parking can create substantialwelfare losses that can be diminished, if not eliminated, byeffective parking pricing. Second, if curbside parking is under-priced, parking fees should be increased until cruising for parkingis eliminated. Third, pricing is not the only tool that local govern-ments have to address failures in the parking market, especially inthe long run. Arnott et al. (2013) show how to deal with them byadjusting parking capacities. It is also evident from these modelsthat downtown travel and parking will have various equilibria. Butthe traffic will never be in a steady state due to stochastic shocks,regular daily fluctuations in parking demand, etc. Nevertheless,this does not render steady-state analysis useless. By looking atsteady states, we can understand the tendency of traffic towardany particular traffic equilibria. Using comprehensive stabilityanalysis, Arnott and Inci (2010) and Arnott et al. (2013) showhow such tendencies and transitions occur. In that sense, theyexplain why traffic can result in a traffic jam today but notyesterday, even though traffic conditions on both days are verysimilar.

The Arnott–Inci–Rowse line of models ignore the contiguity ofparking spaces by assuming that the downtown area is spatiallyhomogeneous, in which downtown parking is either completelysaturated or completely unsaturated. As Martens et al. (2010)argue in detail, in reality, there is a gradual transition fromcompletely unsaturated parking to completely saturated parking,during which some streets may become saturated before others.Cruising for parking will emerge on those streets that havebecome completely saturated, while parking will remain unsatu-rated on the other streets. Levy et al. (2012) show that spatial


Trip Price

Throughput

Demand

0

User Cost

User Cost + Parking FeeE1

ConsumerSurplus

ConsumerSurplus

In-transit TravelCost

In-transit TravelCost

ParkingRevenue

Efficiency LossDue to Cruising


Trip Price

Throughput

Demand

0

User Cost

User Cost + Optimal Parking Fee

E1

ParkingRevenue

Fig. 2. Efficiency loss due to cruising for parking and optimal pricing of parking.







heterogeneities become very important for occupancy levels above92% by comparing the analytical simulation model PARKANALYST(Levy et al., 2012) with the agent-based geosimulation modelPARKAGENT (Benenson et al., 2008), which is able to deal withspatial heterogeneities in the downtown area and can describe theparking dynamics in full. Gallo et al. (2011) offer some other waysto simulate cruising for parking in an assignment model of urbanroad networks.

The theory on cruising for parking largely assumes underpricedcurbside parking. As discussed before, cruising for parking willexist as long as there is a differential between the full prices ofcurbside and garage parking. The optimal policy in those cases isto increase the curbside parking fee until cruising for parking isjust eliminated. Nevertheless, in some countries, such as Belgiumand the Netherlands, curbside and garage parking fees are veryclose. Cruising for parking occurs even in this case because ofdrivers' uncertainty about unoccupied parking spaces.8 Never-theless, one should expect lower levels of cruising for parking insuch circumstances. To quantify cruising for parking in suchsituations, van Ommeren et al. (2012) look at a nationwide Dutchdataset. They find that about 30% of all car trips involve some levelof cruising. The average trip time is 20 min in their dataset, and onaverage only 36 s of that trip time are taken up by cruising. Theyalso concentrated on the determinants of cruising for parking, andtheir results are completely in line with the cruising theoryoutlined above. Their results show that, although the current levelof cruising for parking is low in the Netherlands, better parkingpricing could decrease it further. By making use of the capitaliza-tion of outside private parking spaces into the house prices andthe waiting lists for curbside parking permits, van Ommeren et al.(2011) estimate the cost of cruising per resident to be about 1 Europer day in Amsterdam.

3. Spatial competition in the parking sector

In urban areas, parking garages compete spatially with eachother and with curbside parking spaces. Several empirical papersexamine competition in the parking market (Froeb et al., 2003;Chone and Linnemer, 2012; De Nijs, 2012; Kobus et al., 2013; Linand Wang, forthcoming).9 One of the robust observations of thesestudies is that drivers do not want to walk more than a few blocksfrom where they park to their destination. This gives parkinggarages local monopoly power. Another finding is that, although

exceptions exist, parking fees are generally concave in parkingduration. That is, hourly parking fees are lower the higher theparking duration, which hints at price discrimination in thesector.10 As a matter of fact, third-degree price discrimination islikely to be operational when travelers differ substantially in theirdesired parking durations, and second-degree price discriminationis likely to be operational for individuals choosing between“adjacent” time durations.

A number of theoretical papers deal with various spatialaspects of parking markets. Arnott (2006) is probably the mostcomprehensive. It analyzes downtown parking policy by modelingspatial competition between parking garages. The paper firstunderlines that horizontal economies of scale in garage parkingstem from the fact that the central ramp in a parking garageentails a fixed cost. Due to these horizontal economies of scale,parking garages are discretely spaced. A driver cares about hiswalking time costs, and thus he is willing to pay a premium topark in a closer parking garage. As a result of parking garages' localmarket power, garage parking fees will be inefficient, just like thespacing between them. Arnott (2006) mentions that optimalparking pricing should take such effects into account. The basemodel of the paper applies only to garage parking but it is laterextended to include curbside parking. In that extension, the fullprice of garage parking gets equalized to the full price of curbsideparking via adjustments in the level of cruising for curbsideparking. As in Arnott and Inci (2006), increasing curbside parkingfees until cruising for parking is just eliminated (without parkinggetting unsaturated) has the benefit of decreasing overall trafficcongestion. The paper also discusses how availability of masstransportation affects its main insights.

Arnott and Rowse (1999) is one of the influential papers in theliterature dealing with stochasticity of parking vacancy, whichresults in cruising for parking. In their model, the city is locatedaround a Salop circle. The number of parking spaces is fixed perunit distance. Individuals wait at home for trip opportunities,which exogenously arrive according to a stochastic time-invariantprocess. An individual may or may not accept a trip opportunity; ifhe accepts it, he decides whether to walk or drive to thedestination. If he drives, he decides how far from his destinationto start cruising for parking. Once he finds a vacant parking spot,he immediately parks there and walks to his destination. Theexpected walking distance depends on the average density ofvacant parking spaces, which is endogenously derived in themodel. The market outcome is inefficient because no one caresabout his own contribution to this average density. Arnott andRowse (1999) conclude that optimal parking fees may not achieveoptimality as a result of multiple equilibria. This is an importantinsight underlining the difficulties associated with implementingparking policy in practice.11

Anderson and de Palma (2004) present another spatial modelof parking with search. In their model, a central business district(CBD) is the most desirable parking location, and shoppers arelocated at the other end of a Hotelling line. Parking is provided onside streets that are perpendicular to the main street. It gets moredifficult to find a parking vacancy if there are more parkers. The

8 Parking information systems can be very useful in mitigating this kind ofparking search activity. There is an extensive literature developing smart parkingguidance algorithms (to name just a few: Caicedo, 2009, 2010 and Shin and Jun,2014).

9 Froeb et al. (2003) provide numerical experiments showing how capacityconstraints in merging and non-merging parking lots influence merger effects andpropose a demand estimator that can recover model parameters from the data.Chone and Linnemer (2012) apply their treatment and control group selectionmethodology for retrospective merger evaluation to the merger that took placebetween two large French conglomerates. The merged firm became the leadingparking operator in France. They find that the merger led city-owned parking lotsthat are exposed to the merger to increase their hourly parking fees by about 3%relative to the others that are unexposed. In some sense, they underlined thespatial interdependence in the parking sector even when two parking lots are notdirectly competing with each other (Inci and Lindsey , 2014, theoretically underlinethis indirect competition effect as well as the direct competition effect). Byconcentrating on the same merger and using the same methodology in selectingthe parking lots exposed to the merger, De Nijs (2012) shows that the level of pricediscrimination in the market increased in response to the merger in the sense thatlow-end prices in the parking price schedule increased proportionally more thanthe high-end prices. By making use of the difference between the parking searchbehavior of short- and long-term parkers, Lin and Wang (forthcoming) show asymmetric result in their analysis of parking garages in Manhattan. In particular,they show that when competition increases, the low-end prices decrease propor-tionally more than high-end prices so that the price schedule becomes less curved.

10 In addition to the price discrimination due to (spatial or non-spatial) marketpower in the parking sector, the discreteness in parking prices can be due in part todifficulties for the parking staff or the parking meters to charge for shorterdurations. In many cities, parking fees are set for hourly increments (1 h, 2 h, 3 h,and so on). This discreteness in and of itself distorts parking durations. Caicedo(2012) analyzes what happens if parking fees are charged by blocks of minutesrather than blocks of hours from the perspective of parkers, nearby merchants, andparking operators.

11 Another well-known spatial treatment of parking fees is provided by Arnottet al. (1991), which I review in Section 5 because it mainly deals with tripgeneration in the famous bottleneck model.







most desirable parking spaces are the ones closer to the CBD andthus these are overused, while the spaces further away from theCBD are underused. As a result, optimal parking prices are higherthe closer one parks to the CBD. The socially optimal parking feecorresponds to the parking fee chosen by private parking opera-tors that monopolistically compete with each other, as would bethe case in any Coasian environment. This result, of course,requires the parking operators to be “small” enough so that theytake the utility of drivers as given, but “big” enough to internalizelocal externalities due to parking congestion. Thus, privatizationyields the same outcome as the socially optimal outcome. The onlyexternality that cruising for parking creates is on parkers in thisbase model. Anderson and de Palma (2004) show that this resultno longer holds when cruising for parking imposes externalitieson through traffic as well as on parkers. In a companion paper,Anderson and de Palma (2007) embed their spatial setting into amonocentric city model with commuters and endogenous landuse. In this setting's equilibrium, individuals who live closer to theCBD prefer to walk to the CBD, while those who live further awaydrive to a parking lot and then walk to the CBD. Provided thatcruising for parking does not impose externalities on throughtraffic, in this extension, too, the social optimum is reached ifparking garage operators set prices in a monopolistically compe-titive fashion.

Most theoretical papers analyze parking markets by assuminghourly uniform parking fees. However, spatial competition in theparking market results in parking garage operators chargingdifferentiated parking prices (remember that empirical papersusually find parking fees to be concave in parking duration).Parking garages are in spatial competition not only with eachother but also with curbside parking spaces. They are mostlyprivately operated, while curbside parking spaces are mostlypublicly operated.12 Even the curbside parking fees are differen-tiated in some cities (for example, Istanbul). Inci and Lindsey(2014) try to move toward realism by incorporating these featuresinto their model. In their model, two different types of driverswish to park for different durations. Unlike in standard Salop circlespatial competition models, the utility derived from curbsideparking (which is the outside option to garage parking) isdecreasing in the number of drivers who use curbside parking,and thus it is endogenously determined.13 Inci and Lindsey (2014)show that, with inelastic parking demand, charging differentiatedcurbside parking fees achieves the social optimum without need-ing to regulate garage parking fees. In cases with partial segrega-tion of driver types in terms of parking duration, uniform hourlycurbside parking fees can also achieve full efficiency. Inci andLindsey (2014) also considered direct regulation of parking gar-ages, in which case the social optimum can be achieved as long asan appropriate differential between curbside and garage parking

fees is maintained. In general, the optimal parking policy is mostlydependent on local parking conditions, and thus each city shouldset its parking policies based on city-specific analyses.

The discussion of non-linear pricing of parking should alsocover the trade-off between parking duration and parking fees.Glazer and Niskanen (1992) highlight this trade-off clearly. Ahigher parking fee induces drivers to park for shorter durations,which increases parking turnover. Thus, it is possible for trafficcongestion to increase as parking fees increase simply becausemore people will be using parking spaces. Glazer and Niskanen(1992) find that if traffic flow is suboptimally priced, lump-sumparking fees can increase welfare while per-unit parking fees donot. This suggests that social welfare can also be improved byimposing curbside parking time limits. After all, one can occasion-ally achieve equivalent outcomes by changing quantity rather thanprice. Quantity of parking (per driver) can be controlled by limitingparking capacity or parking time. Arnott and Rowse (2013) con-centrate on the latter and show that curbside parking time limitsmay eliminate cruising for parking by discouraging drivers whowish to park for longer durations. So, setting efficient parking timelimits may be effective when parking has to be underpriced. Twoextensions would be useful in the Arnott and Rowse (2013) settingto sharpen its policy conclusions: allowing parkers to feed theparking meter, and more importantly, allowing drivers to changevisit durations in response to policy changes.

Kobus et al. (2013) estimate the effects of parking fees on thechoice between curbside and garage parking in downtown Almerein the Netherlands, where all parking spaces are regulated by thecity. They use data for all paid parking transactions duringshopping hours. They find the curbside parking demand in thiscity to be extremely elastic, and thus even tiny changes in curbsideparking fees may dramatically alter the stock of occupied curbsideparking spaces. This result is driven mainly by long-term parkers,who are very sensitive to parking prices and have a large effect onthe occupancy rate. Because curbside parking spaces are onaverage closer to drivers' final destination in the area, drivers arewilling to pay a premium for curbside parking. Kobus et al. (2013)estimate this premium to be in the range of 0.37–0.60 Euros.Moreover, when the parking duration is 1 h, the price elasticity ofthe share of curbside parking is about �5.5, while it is muchsmaller for shorter durations. One notable feature of the Dutchcities is that curbside parking is usually at least as expensive asregulated garage parking. This is the opposite of many NorthAmerican cities, where garage parking is not regulated and isconsiderably more expensive than curbside parking.

4. Minimum and maximum parking requirements

A large part of the literature is devoted to getting parking pricesright. It is true that parking pricing is extremely important, butquantity issues are just as important as pricing. In Section 2, Ireviewed some papers trying to answer the question of how manyparking spaces should be supplied in downtown areas. In thissection, I review papers evaluating (minimum and maximum)parking requirements, which determine how many parking spacesshould be supplied by each land use.

Minimum parking requirements, which exist in almost all citiesand towns, determine the minimum amount of parking that eachland use must provide. They are set because property developershave little incentive to provide enough parking spaces to cover thedemand generated by the properties they construct if convenientparking spaces are available nearby. After all, there are alternativeuses of the land for more shops and residences. Moreover, asexplained in Section 2, unpriced (or underpriced) parking suffersfrom the tragedy of the commons, which creates an additional

12 Tsai and Chu (2006) provide a (non-spatial) model with both public andprivate management of parking spaces. In their model, first the government acts asa Stackelberg leader by allocating the fixed amount of parking supply betweenpublic and private management. Then, both the government and the privatecompany set their parking fees. Finally, drivers choose whether to park, and if so,whether to park in a public or private parking space. It is reasonable to assume thatprivate management is more efficient because the private firm has high incentive torecord the parking time very quickly to be able to maximize its profits. In the end,the government forms a mixed market where some spaces are privately managedwith franchise-type contracts while others are publicly managed.

13 This results in various nonconvexities. So, if a parking garage increases theparking fee aimed for a long-term parker, some long-term parkers switch toparking curbside. This means that there will be more cruising for curbside parking,which in turn increases the full price of curbside parking for short-term parkers. Asa result, some short-term parkers switch to parking in a parking garage, whichallows garage operators to increase parking fees for short-term parkers, too. Thiseffect confers extra market power on parking operators, and thus policy makersshould take it into account.







motivation for setting parking minimums to prevent spilloverparking around properties.14

There is empirical evidence that minimum parking require-ments affect the amount of parking supplied on a parcel. Cutterand Franco (2012) concentrate on Los Angeles and analyzeempirically whether minimum parking requirements are bindingfor suburban land uses by using data for commercial, industrial,and retail property sales. Their indirect test, looking at the gapbetween the marginal cost of an additional square foot of parkingarea and its marginal value (measured by the increase in the salesprice from an additional square foot of parking area), shows thatthe minimum parking requirements bind for the majority of thesix different land-use categories they investigate. This means thatproperty developers would supply less parking if there were nominimum parking requirements. They also look at the differencebetween the average parking supply ratio and the estimatedrequired parking ratio. With this direct test, they show thatminimum parking requirements for service retail propertiesstrongly bind. Those for offices and warehouses do not bind asstrongly as they do for other properties.

Although parking minimums are set to ensure that “enough”parking spaces are provided by property developers, the word“enough” turns out to be tricky. As Shoup (1999) points out, thezoning and building codes determine parking capacity for eachland use so as to satisfy peak demand for free parking. Whenparking is free, everyone, not just the drivers, bears the costs ofproviding parking spaces, because those costs are embeddedpretty much in the prices of everything else in the city, especiallyin the property prices and rents. Manville (2013) summarizesnicely how this acts in the housing market:

When local governments require onsite parking for new hous-ing, the cost of housing rises and the price of driving falls. Thecost of parking, which drivers should arguably pay at the end oftheir trips, is instead paid by developers at the start of theirprojects. The terminal cost of driving becomes an up-front costof property development. (Manville, 2013, p. 49).

In his back-of-the-envelope calculations, Shoup (1999) findsthat construction cost of four above-ground (below-ground) park-ing spaces per 1000 square feet of office space increases the cost ofthe office space by 27% (67%). In a study of six neighborhoods inSan Francisco, Jia and Wachs (1999) find that the prices of single-family houses and condominiums are more than 10% higher whenthey include off-street parking than when they do not. Manville(2013) estimates that there will be less housing and less variety inhousing (bundled or unbundled with parking) in response to thezoning laws on parking. Manville et al. (2013) show that parkingrequirements have strong effects on housing, vehicle, and popula-tion densities. They compare Los Angeles with New York to showhow residential minimum parking requirements decrease conges-tion. It turns out that, when parking requirements decreasecongestion, they do so by decreasing housing and populationdensity around the location, not by reducing the number of cars.In fact, they find that a 10% increase in residential minimumparking requirements in New York increases vehicles per square

mile by 5%, vehicles per person by 4%, and decreases housing andpopulation density by 6%.

There are very specific parking requirements for each land use,including exterminators, asylums, and slaughterhouses, to name just afew. The zoning and building codes determine them ad hoc. However,the implicit message of many parking papers is to tailor parkingpolicies according to local conditions (see, e.g., Inci and Lindsey, 2014).In fact, the SFpark experiment, reviewed in Section 6, revealed thateven the differences in adjacent blocks are very important to optimalpolicy (Pierce and Shoup, 2013). Despite these facts, unfortunately,neither theory nor data plays a significant role in determining parkingrequirements in current practice. Willson (1996) reports from hissurvey with 144 planning directors that inspecting nearby cities andconsulting the handbooks of the Institute of Transportation Engineersare the most frequently used methods in determining parkingrequirements. Jakle and Sculle (2004) provide a historical review ofhow these requirements spread from one city to another. No oneknows if copying a city's parking requirement will be optimal foranother city. The requirements in the handbooks are based on anunstated assumption of free parking, because most parking was freewhen they were published.

Most of the recent work on urban planning has recommendedeliminating parking requirements completely and replacing themwith a more market-based approach, in which costs and benefitsbecome the main determinants of parking supply. Willson (2013)provides detailed prescriptions on how to reform parking require-ments. Shoup (1999, 2005b) provides convincing arguments onwhy minimum parking requirements can also excessively increasecar ownership and usage. Weinberger (2012) gives some evidence.He concentrates on private, on-site, residential parking in NewYork City and shows that those who have guaranteed on-siteparking spaces at their origin have a greater propensity to use cars.This holds even between origins and destinations that are wellconnected by public transportation. Because more driving meansincreased traffic congestion, Weinberger's (2012) results implythat drivers trade off ease in parking for higher travel time andhigher oil consumption. One practical and successful implementa-tion of reducing minimum parking requirements is California'sparking cash-out legislation from the mid-1990s (see Section 7.3).

It is unclear if we should eliminate minimum parking require-ments completely or tailor them to local conditions (with theassumption that the market alone is unable to do so). In fact, in hisreview of The High Cost of Free Parking, Levinson (2005) argues thatsuch requirements do not appear to be problematic in low-densitysuburban areas. As Barter (2010) states, even if we want toeliminate them, we do not know if the market will be ideal afterelimination or if we will need to regulate the market further(for example, by instating maximum parking requirements thatdetermine the maximum amount of parking each land use mustprovide).15 The planning literature does not say much about howthe transition will happen or about what will happen to theexisting parking market during the transition. Barter (2010) offerssuggestions on how to put together parking policy in the absenceof parking requirements. He believes that, for well-functioningparking markets, we need to actively foster and regulate them.

Guo and Ren (2013) analyze the 2004 parking reform in Londonthat replaced minimum parking requirements with maximumparking requirements. They analyze garage parking in new hous-ing properties in 22 boroughs from 2004 to 2010. They find thatthe parking supply in residential developments decreased by 40%after the parking reform. The removal of the minimums explains

14 Whether increasing parking capacity in fact helps in solving the spilloverparking problem is not certain. Suppose that some of the total parking demand issatisfied by subsidized (or free) parking. Merriman (1997) shows that, if the parkingdemand is price elastic, total parking demand increases disproportionately more inresponse to an increase in subsidized parking capacity. That is, an additionalsubsidized parking space attracts more than one additional parking demand. Thus,spillover parking around subsidized parking areas can be paradoxically higher if thesubsidized parking spaces are expanded. Lai (2006) extends this result to anendogenous parking pricing setting.

15 Arnott (2006), reviewed in Section 3, discusses the effects of imposingminimum and maximum parking requirements on a downtown area. Hasker andInci (2014), reviewed in Section 7.2, provide a foundation for imposing minimumparking requirements in shopping malls.







97.8% of the change, while imposing maximums explains only2.2%. Without maximum parking requirements, more parking issupplied in the unregulated parking market in high-density areaswith abundant public transportation possibilities. An unregulatedmarket ignores the high social cost of driving. Thus, maximumparking requirements appear to be an important policy tool whenin addition curbside parking is controlled. The parking reform inLondon is a valuable natural experiment because it is a once-and-for-all, large-scale change. So, the results could be viewed as moregeneral than those from small-scale case studies.

5. Bottleneck model, road pricing, and parking pricing

A large body of literature concentrates on the optimal pricing ofroads in bottleneck models. Vickrey (1969) is the father of themost widely used model in the literature, which was laterformalized by Arnott et al. (1990) and elaborated by a series ofother papers by the same coauthors. In these types of models,rush-hour traffic dynamics are modeled as a queue that formsbehind a bottleneck. Despite the facts that optimal road pricing isknown to be very effective and that it has been proposed byprominent transportation economists for many years, cities sel-dom use it in practice. London, Singapore, and Stockholm are thelarger cities known for their success in implementing it. Roadpricing, or congestion tolling, is not applied frequently in practice,given the political risks and large implementation costs it involves.A less (though still) risky, cheaper, and somewhat neglected optionis parking pricing. After all, road prices and parking fees cansubstitute each other to some extent, and parking fees requiresimpler and cheaper investments. Thus, a part of the literaturecompares the two pricing methods from a welfare perspective.

Arnott et al. (1991) explore the effects of parking fees onmorning rush-hour traffic congestion in a deterministic bottleneckmodel. They show that spatially varied parking fees can signifi-cantly decrease travel costs in urban areas. In their model, theworkplace is at the CBD, and parking spaces are between the CBDand the bottleneck. Commuters choose their departure time andparking location, and they have to walk to the workplace afterparking. Their aim is to minimize the full price of a trip, which iscomposed of in-transit travel time, walking time costs, scheduledelay costs, tolls, and parking fees. Parking search time isneglected, and the parking fee is not dependent on parkingduration. When roads and parking are free, commuters park inorder of increasing distance from the CBD. As a result, scheduledelay costs are higher in the aggregate. These costs can beminimized by competitively set parking fees, but this does noteliminate congestion. So, the welfare gains from competitively setparking fees alone are limited and sometimes even negative.Arnott et al. (1991) analyze two types of optimal pricing schemes.A time-varying toll can eliminate queuing, but drivers still park inorder of increasing distance from the CBD. However, although alocation-dependent pricing does not eliminate queuing, it reversesthe order of parking spots being taken. As a result, arrival timescome close to the work start time; hence, this pricing scheme maybe superior. Thus, the optimal parking fee system can be at least asefficient as the optimal time-varying tolls. The social optimum canbe achieved by a jointly optimal congestion toll and location-dependent parking fee scheme.

Like Arnott et al. (1991), Qian et al. (2011) concentrate on themorning commute in the bottleneck model, but they makedifferent assumptions about the parking market setting. First,rather than continuously provided, parking spaces in their modelare in finitely many parking areas composed of parking lots. Eachparking lot in a parking area charges the same parking fee, and theaccess times from any of these parking lots to the CBD are the

same. There is one central and one peripheral parking area in theirmodel. Second, the capacity of the parking areas and access timesare endogenously determined. Third, parking lots are privatelyowned and in competition with each other (see Qian et al., 2012,for the version of the model in which all lots are publicly owned).Qian et al. (2011) analyze the implications of three regulationschemes: price-ceiling regulation, capacity-floor or capacity-ceiling regulation, and quantity tax/subsidy regulation.

Zhang et al. (2011) analyze the effectiveness of parking permitdistribution in eliminating the external costs of parking search inthe morning commute. They compare the case in which all drivershave reserved parking with the case in which all drivers face abinding parking capacity constraint. They show that commutingcosts are lower in the former, since the parking space reservationsystem can completely eliminate inefficient competition for park-ing spots. Yang et al. (2013) extend this model and allow onlysome drivers to have reserved parking, and they compare thismixed-parking supply case with the two extremes considered byZhang et al. (2011). They show that the mixed parking supply caseis more efficient when the total number of parking spots exceeds acertain threshold.

Zhang et al. (2008) use the setting of Arnott et al. (1991) andcombine the morning and evening commutes to derive an overallcommuting pattern for the day. Fosgerau and de Palma (2013) alsoconcentrate on the overall commuting pattern in a different citysetting with parking locations at the CBD and a bottleneck on theway. They find that only a small share of the efficiency gains fromoptimal tolls can be gathered using a time-varying parking fee. Atthe social optimum, the optimal parking fee divides the work tripsinto two time intervals. In the first time interval, parking ischarged at a zero rate and queuing takes place, while in thesecond interval parking is charged at a time-varying rate and thebottleneck capacity is just utilized without any queues. The timeintervals are reversed for the evening commute. In a similarfashion to Zhang et al. (2005), Fosgerau and de Palma (2013) alsoshow that the parking policy can make use of the interactionbetween the morning and evening commutes via the choice oftime spent at work. They show that a combined morning andevening parking fee is more efficient than a fee only in themorning or a fee only in the evening. In fact, the combined feecan eliminate congestion.

Verhoef et al. (1995), too, analyze parking fees in comparisonwith congestion tolls. They compare the effects of parking supplyrestrictions vis-a-vis pricing instruments. They emphasize thatparking fees are only second-best policy instruments for addres-sing congestion because they are levied at the end of a trip andthus cannot be differentiated with respect to trip length or roadstraveled. They find time-varying parking fees to be superior toparking supply restrictions for three reasons. First, an informationproblem arises in supply restrictions in that drivers have to knowbefore making their trip decisions if there will be a parking spaceavailable to them. Otherwise, the reduction in trips may not berealized. Second, there is temporal inefficiency in supply restric-tions because they do not guarantee that the one who values aparking spot the most gets that parking spot. Most of the time, it isfirst come, first served. Third, there is also an intertemporalinefficiency. Early parkers do not incur the costs of searchexternalities that they imposed on late parkers, which distortsparking occupancy over the course of the day. This in turn distortsthe distribution of traffic over the course of the day. Verhoef et al.(1995) also discuss the merits of location-dependent parking feesin replacing road prices.

Glazer and Niskanen (1992) discuss parking pricing in thepresence and absence of road pricing. They note that parking feesare substitutes for tolls when higher parking fees also mean highercosts of the trip. But this does not always hold, because drivers can







vary their parking durations or some of them may be just through-traffic drivers who do not intend to park at all. As a result, a higherparking fee results in lower parking durations and thereby inhigher parking turnover, which in turn increases congestion.Calthrop et al. (2000) discuss parking policy with and withoutsingle-ring cordon pricing using the TRENEN static urban modelcalibrated for Brussels. They underline the interdependencebetween parking and cordon prices. If parking prices are far belowtheir efficient levels, the optimal cordon price increases; and ifcordon pricing is introduced in the city, the optimally set parkingfee decreases. So, as in Arnott et al. (1991), the most efficient policyis a combination of cordon pricing and parking pricing. They alsoshow how powerful the parking prices are: in their simulations,70% of the welfare gains are gathered by imposing parking feeseven if cordon pricing is not implemented.

On the empirical side, some papers compare the performanceof road pricing with parking pricing in regulating traffic. By usingopinion surveys, Baldassare et al. (1998) find that drivers are moresensitive to parking pricing. Shiftan and Golani (2005) use astated-preference survey on higher parking costs and the intro-duction of congestion tolling in central Tel Aviv. In line with thetheory, the majority of workers and nonworkers stated that theywould respond to the changes by changing their travel mode, andin the case of congestion tolling, also by rescheduling their triptimes. By using another stated-preference survey with peoplelinked to Technion, Albert and Mahalel (2006) find travel demandto be highly elastic with respect to both pricing schemes: �1.8with respect to congestion tolls and �1.2 with respect to parkingfees. In this survey, 54% of the respondents stated that they wouldprefer to use other options to avoid charges if a parking fee isintroduced, and 72% of them would do the same in response to acongestion toll. By concentrating on a stated-preference survey inthe city of Mashhad, Azari et al. (2013) find that drivers are moresensitive to parking charges than to cordon tolls.

Jansson (2010) focuses on a different aspect of congestion tollsin comparison with parking fees. He provides strong evidence thatcongestion tolling has such high running costs that it might bebeneficial for the city to replace congestion tolling with parkingpricing, which is usually considered to be second-best option inmitigating traffic congestion. He offers two policy tools. One isfringe benefit taxation of employer-provided parking; the other isa two-part-tariff kind of parking pricing scheme whose fixed partis for making the trip into the CBD and the other part is the usualhourly parking charge. Bonsall and Young (2010) consider thepolitical acceptability of road user charges and offer a partialsolution. According to their proposal, the city could introducecongestion tolling, but at the same time it would make publicparking free. Under some circumstances this scheme is shown toincrease public revenue. However, we know from the theory that acombined policy with congestion tolling and parking pricingshould perform the best, and a priori there is no reason why freeparking would achieve the social optimum (see, e.g., Calthropet al., 2000).

6. Toward temporal-spatial pricing of parking

In his statement to the Joint Committee on WashingtonMetropolitan Problems back in the 1950s, Vickrey underlined thatparking should be priced at its social marginal cost just like anyother commodity. He argued that parking prices should bedemand responsive and thus temporally and spatially varied.Vickrey (1954) goes through his arguments in minute detail.Temporal-spatial parking pricing, or Vickrey parking pricing, wasdifficult to implement when it was introduced, given traditionalparking meter technology and the available data. However,

nowadays cities are able to employ “smart” parking systems andcollect comprehensive data, both of which allow for more sophis-ticated pricing techniques. The practical implication of Vickrey'swisdom has been defended by Shoup for a long period of time:achieving 85% occupancy on each and every block, which mostlymeans one or two empty parking spots on each street so thatcruising for parking becomes negligible.

Temporal-spatial pricing of parking is already in place in differentparts of the world. The city of Rotterdam optimizes its parking feesstreet by street to achieve less cruising. In Istanbul, parking fees aredifferentiated across time and space and the parking fee for theshortest duration (i.e., the entry fee) is kept high so as to controlexcess demand (just as advised in Jansson, 2010).16 One experiment inthe United States has been attracting much attention in the media. InApril 2011, San Francisco Municipal Transportation Authority launcheda plot study, called the SFpark Experiment, with a $18-million federalgrant from the US Department of Transportation. As Pierce and Shoup(2013) put it, the idea of the experiment is to “get the prices right” fora given occupancy level. That is, the parking fees are adjusted fromperiod to period until the “right” occupancy rate (i.e., the targetoccupancy rate) is achieved. In this section, I review the SFparkexperiment.17 My review is mostly based on Pierce and Shoup (2013).

The SFpark experiment aims to change the parking fees blockby block until there are only one or two empty parking spots perblock. In that way, a driver is always able to find an empty parkingspot, and thus cruising for parking is minimized. This is in linewith the theoretical finding of Arnott and Inci (2006) reviewed inSection 2: increase the parking fee until parking becomes justsaturated. In a world of uncertainties, it is tricky to find the “right”occupancy rate. The experiment requires increasing the parkingfees on busy high-demand blocks while decreasing them on low-demand ones. However, there are always stochastic variations inparking demand, and thus guaranteeing one or two empty parkingspots at each block without making them underused for longperiods of time requires great sophistication. The experimentovercomes this problem by coming up with a target occupancyband. The parking fee on a street is decreased if the occupancy rateis below 60% and increased if it is above 80%. Although there issome rationale for this band, it is determined only intuitively. In arecent paper, Arnott (2014) goes one step further to figure out theoptimal target occupancy rate in a world of stochasticities.

The experiment includes 7000 metered parking spaces on 254blocks, 14 publicly operated parking garages, and a surface lot. Theminimum price on any block was 25 cents and the maximum priceon any block was $6. Prices are changed in response to theoccupancy rates once every 6 weeks. The price is decreased by50 cents an hour if the occupancy rate in the previous period wasbelow 30%, decreased by 25 cents an hour if occupancy rate wasbetween 30% and 60%, kept unchanged if the occupancy rate wasbetween 60% and 80%, and increased by 25 cents an hour if theoccupancy rate was above 80%. There are three time blocks(opening time to noon, noon to 3 pm, and 3 pm to closing time)and two day times (week and weekend). Pierce and Shoup (2013)report that there were 5294 price changes in the first year of the

16 Van Ommeren and Russo (2014) empirically show the significance of thegains associated with varying parking fees with respect to time. In particular, theyshow that the gain from shifting from free parking to time-varying parking fees at ahospital in the Netherlands is on the order of 10% of overall parking costs of thehospital.

17 A comprehensive description of the SFpark experiment can be found in SanFrancisco Municipal Transportation Authority (2011) and its evaluations in SanFrancisco Municipal Transportation Authority (2014). The evaluations show thataverage parking fees became lower; parking availability improved; parking searchand parking fee payment became easier; double parking, peak period congestion,traffic volume, greenhouse gas emissions, and vehicle miles traveled decreased, andtraffic speed and road safety increased.







experiment. Prices were increased in 32% of the cases, decreasedin 31%, and kept unchanged in 37%. As a result, the average pricefell by about 1% overall.

The beauty of the experiment is that all decisions are based oneconomic criteria and data, and all changes are announced pub-licly. Thus, political influence is minimized and transparency ismaximized. Moreover, revenue generation is not among the mainaims. Basically, demand determines parking fees. As argued indetail by Pierce and Shoup (2013), in response to the pricechanges, short-time parkers, carpoolers, those who have difficultywalking, and those who attach high value on saving time areexpected to park disproportionately in convenient parking spots,while long-term parkers, solo drivers, those who love walking, andthose who attach low value on saving time are expected to movetoward distant parking spaces. Pierce and Shoup (2013) calculatethe simple mid-point price elasticity of occupancy rates. Althoughthey do not control for the endogeneity of the price changes,18

their estimates show how elasticities vary by time and location,which highlights once again why the city should move towardtemporal-spatial parking pricing.

The SFpark experiment shows us that by fine-tuning parkingfees, cities can let drivers save time via reduced traffic congestion.Seeing this potential, other cities have also started similar experi-ments. The LA ExpressPark is one of them. Milan and Mexico Cityhave been considering similar experiments. One dimension thatshould be added to these experiments is the ability to predict.These experiments are reactive in that they adjust prices afterdemand realizations and continue to iterate (and reiterate) fromthere until the target is achieved. This is costly. After all, if a priceof a good is changed too frequently, there will be menu costs,frustration among customers, and the experiment may not achieveits desired result simply because people cannot follow the currentprices and react optimally to them. A less costly way of designingparking policy is to turn to demand-prediction methods based ontheoretical models, microsimulations, and empirical estimations.In that way, at least the number of iterations can be lessened.

Millard-Ball et al. (2014b) move toward this direction byoffering simulations based on the SFpark data. They evaluate thefirst two years of the experiment and find that the program slowlyachieved the 60–80% average occupancy rate goal and diminishedcruising for parking by 50%. Although the short-term impact of anysingle price adjustment appears to be small, the long-term impactof a series of small adjustments on occupancy and cruising issignificant. Chatman and Manville (2014) also evaluate the firsttwo years of the SFpark experiment. They show that despite thefact that the price increases lowered the block-level parkingoccupancy as intended, parking availability (defined to be theshare of time at least one space on the block face is vacant)improved only very modestly. This is because a block can beentirely full for many hours even though the monthly averageparking occupancy falls in the 60–80% interval.

Another performance-based parking pricing scheme wasimplemented by the city of Seattle in the first four months of2011. The target occupancy rate was chosen to be 71–86% in thisexperiment. Ottosson et al. (2013) provide the main findings of theexperiment. They obtain price elasticity of parking demand bytime of the day and neighborhood characteristics. Again, in linewith Vickrey's vision, elasticities change with time and space. They

also analyze the effects of parking fee changes on parking dura-tion, parking turnover, and total revenue. As in the SFpark experi-ment, short-run parking demand is found to be inelastic, but withpersistent implementation, a significant change in variables ofinterest occurred in the long run. In particular, drivers ended upparking for a shorter time on those blocks where prices increasedand for a longer time on those blocks where prices decreased.19

7. Various forms of parking

Various forms of parking have attracted particular interest in theliterature. In this section, three specific parking forms are reviewed.Section 7.1 reviews the work on residential parking, Section 7.2 thework on shopping mall parking, and finally Section 7.3 the work onemployer-provided parking.

7.1. Residential parking

Residential parking is usually provided curbside to addressspillover parking around shopping malls, office buildings, sta-diums, etc., for the benefit of residents living nearby. In mostcountries, residents have to pay for residential parking rights (orparking permits). Residential parking also includes on-site garageparking available in residential properties, whose supply is usuallydetermined by parking zoning regulations (reviewed in Section 4).Residential parking may also include driveway parking.

The only existing theoretical work on residential parking todate is that by Molenda and Sieg (2013). They analyze the trade-offbetween the convenience of living close to the amenities of thecity and the difficulty of finding a parking space in central areas.They determine when it is desirable to allocate residential curb-side parking from both the resident's and the social planner'sperspective.

Using a household survey in New York City, Guo (2013a, 2013b)finds that availability of residential parking has a large effect on carownership. Usually, the cost of residential parking is embedded inhousing costs. Making use of house prices in Amsterdam, vanOmmeren et al. (2011) estimate residents' daily willingness to payfor parking permits to be about 10 Euros. They argue that thisamount is higher than what residents actually pay for parkingpermits but lower than what nonresidents pay for parking there.Thus, parking permits are an inefficient use of curbside parkingspaces. In a similar vein, van Ommeren et al. (2014) concentrate onthe relationship between parking supply and residentialparking permits. They estimate that parking permits impose about275 Euros welfare loss per permit.

7.2. Shopping mall parking

Shopping malls allocate enormous amounts of land for parking,and most of them provide parking for free. According to a surveyof the International Council of Shopping Centers and the UrbanLand Institute (2003), 94% of the malls in the US provide parkingfor free, and they allocate 4–6 parking spaces per 1000 square feetof gross leasable area. This translates into an amount of land largerthan the amount they allocate for shops. It is not just that land isexpensive; allocating more space for parking on that land has thehigh opportunity cost of making it unavailable for shops.

Hasker and Inci (2014) provide a rationale for these observa-tions in shopping mall parking. Their explanation is based oncustomer search. Suppose for simplicity that the mall sells onegood, and consider a risk-averse customer. When the customer

18 By employing a regression discontinuity analysis, Millard-Ball et al. (2014a)find much lower (short-run) price elasticities than those calculated by Pierce andShoup (2013). They argue that there is an endogeneity problem in Pierce andShoup's (2013) mid-point elasticity calculations since even random fluctuations indemand could trigger price changes under the price-adjustment rule used by theSFpark experiment. In fact, they find that the impact of parking fee adjustments ondrivers' behavior is statistically insignificant in the short run. 19 Similar ideas are entertained in the usage-sensitive road pricing literature.







visits the mall, there are two possible outcomes. He may find outthat the good is his ideal good,20 in which case he buys it andleaves the mall happily. If he came by car and if the parking fee ispositive, he pays the parking fee. Alternatively, he may find outthat the good is not his ideal good, in which case he leaves the mallempty-handed. Nevertheless, he still has to pay for parking. Thus,having a positive parking fee at the mall is no different frompunishing the customer for searching for the good at the mall. Thisis the last thing that a mall would like to do, and thus it providesparking for free. Hasker and Inci (2014) show that this argumentholds whether the mall has monopoly power or prices competi-tively; it holds if there is a trade-off between shopping andparking spaces. They also show that a parking validation programdoes not change the result. Although the mall in their model sells asingle good, their argument can be generalized to the case inwhich the mall sells many goods.

The fact that parking is free does not mean that there is no costof parking for the customer. In fact, the mall embeds the parkingcosts in the prices of the goods it sells. Thus, a welfare analysis of afree parking provision at the mall is vital. Hasker and Inci (2014)show that the free provision of parking is also the second-bestsocial optimum. That is, a benevolent social planner would alsoimplement free parking at the mall. According to the InternationalCouncil of Shopping Centers and the Urban Land Institute (2003),when the parking fee is positive, it is mostly so in urban malls,where the parker is not necessarily a customer. Hasker and Inci(2014) show that, if a sufficient number of parkers park at the malllot because they have other errands to run than shopping at themall, the mall starts charging them since it is the only way ofmaking a profit out of them.

Hasker and Inci (2014) find that the socially optimal lot size isalways larger than the lot size chosen by a profit-maximizing mall,because the planner cares about the welfare of those who visit themall but are unable to find their ideal good. This provides arationale for applying minimum parking requirements at shoppingmalls. However, they obtain this result when there is no trade-offbetween parking lot size and shop area, which is more of an issuefor urban malls. In an extension of their model, they show thatminimum parking requirements might be replaced by maximumparking requirements when this trade-off is sufficiently important.In fact, in addition to minimum parking requirements, a few citiesimpose maximum parking requirements for all residential proper-ties. However, the cap is so high that they almost never bind.

In an early paper, Sutherland (1959) discusses parking pro-blems in shopping malls. More recent papers concentrate onparking around shopping areas. For example, Lindsey and West(1997) analyze the use of parking coupons by downtown retailers,and Lindsey and West (1998) empirically evaluate the perfor-mance of such a price discrimination program in Edmonton,Canada. Lan and Kanafani (1993) concentrate on a similar pro-gram, a park-and-shop program, in a non-spatial model withexogenous prices. The attractiveness of the downtown as ashopping place is also a concern in determining parking fees.Rietveld et al. (2002) undertake an evaluation of 30 municipalitiesin the Netherlands and find no significant effect of pricing andcapacity measures on shopping behavior. Bacon (1993) analyzesthe effects of changes in the parking capacity on travel forshopping and finds that an enlarged parking capacity dispropor-tionately increases traffic congestion downtown. Mingardo andvan Meerkerk (2012) test this hypothesis by concentrating on 80shopping areas in the Netherlands. They find that higher parking

fees increase parking turnover but that parking turnover isunrelated to parking capacity for all shopping areas, excludingthe regional ones.

7.3. Employer-provided parking

Firms tend to provide parking for their employers, and often-times for free. Small and Verhoef (2007) note that commuters payat most 2.5% of their parking costs at the workplace. The work onemployer-provided parking usually looks at the effects of employ-ers' parking supply on employees' mode of travel. Willson andShoup (1990) find that solo-driven trips are cut on average by 41%in response to eliminating free parking at the workplace, which isargued to have positive effects on congestion and air pollution.Willson (1992) finds that auto travel to work decreases by 25–34%if employees have to pay for parking.

Van Ommeren and Wentink (2012) estimate the welfare losscaused by free employer-provided parking. Free employer-provided parking is not taxed as a benefit in kind. Hence, in asufficiently competitive labor market, employers tend to providefree parking rather than paying higher wages. Moreover, manycities impose minimum parking requirements. If not chosenwisely, these requirements make parking a fixed cost, which thenresults in underpriced parking. Both of these result in deadweightloss. Van Ommeren and Wentink (2012) estimate that the taxpolicy causes a welfare loss on the order of 10% of the resource costof the employer-provided parking spaces, while minimum parkingrequirements for employers create an additional welfare loss onthe order of 18% of the resource cost of the employer-providedparking spaces.

Parking “cash out” policies have been offered as a powerfulmethod of regulating downtown parking demand. Under such apolicy scheme, employers are required to offer employees theoption to take the cash amount of the employer-provided parkingspaces. This approach had in fact been in the air since Roth (1965)and extends up until Shoup (1995, 1997, 1999, 2005a, 2005b).Shoup (1997, 2005a) carried out extensive work on the parkingcash-out policy in California, which applied only to employers whorented spaces from a third party. He finds that solo driversdecreased from 76% to 63%, carpoolers increased from 14% to23%, public transportation users increased from 6% to 9%, numberof workers walking increased from 2% to 3%, and number ofbicyclists increased from 0.8% to 0.9%.

Ison and Wall (2002) and Rye and Ison (2005) look at the UKexperience in implementing workplace parking charges. Theyemphasize how to overcome the pitfalls in practical implementa-tion. Watters et al. (2006) look at behavioral reactions to imposingwork place parking charges by various age and income segmentsin Dublin. They find that 22.3% of individuals would continue tocommute to work by car if a 5-Euro charge is imposed per day forparking at the workplace. By administering some questionnairesand focus groups with employees, Aldridge et al. (2006) explorethe possibility of introducing paid-employee parking at airports.21

In sum, whether to have employer-provided parking dependscrucially on the level of road congestion. However, in busy down-town areas, it should be seen as a form of underpricing of parkingand thus should be limited.

20 For example, he may be looking for a particular brand of an LCD TV, and hesaw a partial description of it in advertisements. Upon seeing it in the mall, he mayfeel that this is what he is looking for.

21 Another special parking form is university campus parking, where not justthe employees (staff and academics) but also students and other universitycommunity members can use the parking spaces. Barata et al. (2011) concentrateon parking problems at the University of Coimbra in Portugal and find that 45% ofthe parking spaces are subject to no parking regulations, thus parking is seriouslyunderpriced. Tezcan (2012) tries to identify what happens if paid parking isintroduced at the Ayazaga Campus of Istanbul Technical University.







8. Directions for future research

Parking policy has traditionally been supply oriented. Withrecent work, the emphasis has shifted to a more market-oriented,demand-focused approach. Although exceptions exist, for mostforms of parking, cashing out free or subsidized parking andeliminating cruising for parking as much as possible are the mostimportant messages of the literature. The topics covered in thisreview paper are topics that have been studied extensively. I finishthis review by identifying (some of) the understudied topics thatmay require further elaboration for improvements to parkingpolicy.

The performance of any parking policy (or, as a matter of fact,any policy) depends crucially on the level of enforcement.Kladeftiras and Antoniou (2013) use microsimulations to analyzethe effects of illegal parking on traffic congestion and show thataverage traffic speed can be increased by at least 10–15% bylimiting double parking, and that it can be increased by up to44% by completely eliminating double parking. They also showthat even higher rates of improvements can be obtained in delaysand stopped time. However, most parking papers, with theexception of Elliott and Wright (1982), Cullinane (1993),Thomson and Richardson (1998), and Petiot (2004), assumeperfect enforcement. There is thus an immediate need for theore-tical, empirical, or case-study-based analyses of illegal/informalparking and parking enforcement.

Another urgent need in the literature is further elaboration onthe political economy of parking. Russo (2013) provides anevaluation of political concerns in determining road charges andparking prices at both the regional and city level. Button (2006)touches on some political economy concerns in parking. Marsden(2006) tries to determine the evidence base for parking restraintsin practice. Many papers underline the influence of lobbying bymerchants on free parking in downtown areas in order not to losetheir competitive advantage to suburban malls, which mostlyprovide free parking. Other papers highlight that an effectiveway to increase the acceptability of higher parking fees is to usethe parking revenue to improve the public space around thestreets whose parking fees are increased while making theseimprovements as explicit as possible. However, structural exam-inations of the role of politics in the determination of parkingprices and quantities require further work. A nice piece towardthat direction is provided by Rye et al. (2008). They consider theparking policy changes in Edinburgh partly in response to publicand business opinion surveys in the city and compare them withparking policies that would be based solely on travel trends. Inthat sense, they demonstrate the influence of the political processon parking policy formulation.

For better policy making, cities need to better understand theinteraction between parked cars and cars in transit. The studies oncruising for parking partially contribute to this by analyzing theeffects of cruising cars on the travel time of all cars. But, anothersuch interaction – the relationship between parking and modechoice – is somewhat understudied, especially theoretically. Gillen(1977) provides an early empirical assessment of the effects ofparking costs on mode choice and finds parking fee elasticities tobe low, since higher parking fees trigger not only modal shift butalso parking relocation. Sabir et al. (2013) recommend coordinat-ing mode choice by appropriate pricing of parking at the beach.Merriman (1998) estimates that increasing parking capacity byone unit at parking-capacity-restrained rail stations in Chicagoincreases the number of passengers boarding by between 0.3 and2.2. Voith (1998) looks at the effects of parking taxes (and publictransportation subsidies) on community size, land value, andmode choice in a general equilibrium setting. He finds that parkingfees have non-monotonic effects on these interest variables. To set

the baseline for parking policy formation, empirical work shouldfocus on obtaining relevant sensitivities for as many locations andtimes of interest as possible.

Perhaps the most important and practical input for parkingpolicy formation is parking demand elasticity. Many studiesconcentrate on estimating this variable. The price elasticity ofparking demand is found to vary across time, space, and groups ofpeople.22 However, given the current state of administrative data,in estimating elasticities, many existing papers neglect cruising forparking, which creates latent demand. The literature should focuson incorporating parking search into estimation procedures inorder to measure cruising-for-parking time costs more precisely,which can then be used to estimate elasticities based on general-ized prices of trips (see Madsen et al., 2013, for an attempt in thatdirection). Because many interest variables in parking are specificto time, location, and groups of people, I should also emphasizeagain the importance of microsimulations on real road networks.For that matter, engineers, who develop traffic simulation models,and economists, who specialize on price and quantity determina-tion, should interact more to improve the existing models anddevelop better ones. Last but not least, we need “bridge” studies totranslate scientific insights into detailed policy prescriptions thatcities can realistically implement.

Acknowledgments

I would like to acknowledge financial support from the Scientificand Technological Research Council of Turkey (TUBITAK Career Grant111K051) and the Turkish Academy of Sciences (Outstanding YoungScientist Award – TUBA-GEBIP), and the recognition by the ScienceAcademy (Turkey) via their Young Scientist Award (BAGEP). I wouldlike to thank Ken Small and participants at the TIDE Training andExchange Workshop in Milan (2014) and “Excellence in Science: ERCEnables Young Researchers” Symposium (jointly organized by theERC, TUBA, and TUBITAK) for very helpful comments and GunerVelioglu and Ozde Ozkaya for excellent research assistance. All errorsare my responsibility.

References

Albert, G., Mahalel, D., 2006. Congestion tolls and parking fees: a comparison of thepotential effect on travel behavior. Transp. Policy 13, 496–502.

Aldridge, K., Carreno, M., Ison, S., Rye, T., Straker, I., 2006. Car parking managementat airports: a special case? Transp. Policy 13, 511–521.

Anderson, S.P., de Palma, A., 2004. The economics of pricing parking. J. Urban Econ.55, 1–20.

Anderson, S.P., de Palma, A., 2007. Parking in the city. Pap. Reg. Sci. 86, 621–632.Arnott, R., 2006. Spatial competition between parking garages and downtown

parking policy. Transp. Policy 13, 458–469.

22 Feeney (1989) reports 20 different studies on parking demand. Vaca andKuzmyak (2005) report elasticities between �0.1 and �0.6, where �0.3 is themost mentioned value. Using a survey of 1007 curbside parkers in Dublin, Kelly andClinch (2006) find little difference between demand for business and non-businesstrips in response to small increases in parking prices. However, for large increases,as expected, the demand for business trips is much more inelastic than that fornon-business trips. Kelly and Clinch (2009) estimate demand elasticities by makinguse of the 50% city-wide increase in hourly parking prices in Dublin. In response tothe price change, the average parking duration drops by 16.5%, implying anincrease in parking turnover rate and thus parking availability. The average priceelasticity is �0.29, which is inelastic. They find that parking demand is most priceelastic around 9 pm and particularly on weekdays; most inelastic around 12 pm; inbetween the two around 3 pm. Gillen (1977) provides demand elasticity estimateswith respect to parking fees for various income classes in Toronto. As expected,elasticity drops significantly as income increases: demand is pretty inelastic(�0.169) for the income class that earns $10,000 or more, while it is elastic forthe income class earning between $4000 and $6000 (�1.69). See also Axhausenand Polak (1991) and Hensher and King (2001).



http://refhub.elsevier.com/S2212-0122(14)00056-2/sbref1













Arnott, R., 2011. Parking economics. In: de Palma, A., Lindsey, C.R., Quinet, E.,Vickerman, R. (Eds.), A Handbook of Transportation Economics. Edward ElgarPublishing Limited, UK, pp. 726–743.

Arnott, R., 2013. A bathtub model of downtown traffic congestion. J. Urban Econ. 76,110–121.

Arnott, R., 2014. On the optimal target curbside parking occupancy rate. Econ.Transp. 3, 133–144.

Arnott, R., de Palma, A., Lindsey, C.R., 1990. Economics of a bottleneck. J. UrbanEcon. 27, 111–130.

Arnott, R., de Palma, A., Lindsey, C.R., 1991. A temporal and spatial equilibriumanalysis of commuter parking. J. Public Econ. 45, 301–335.

Arnott, R., Inci, E., 2006. An integrated model of downtown parking and trafficcongestion. J. Urban Econ. 60, 418–442.

Arnott, R., Inci, E., 2010. The stability of downtown parking and traffic congestion.J. Urban Econ. 68, 260–276.

Arnott, R., Inci, E., Rowse, J., 2013. Downtown Curbside Parking Capacity. CESifoWorking Paper Series 4085, CESifo Group Munich.

Arnott, R., Rowse, J., 1999. Modeling parking. J. Urban Econ. 45, 97–124.Arnott, R., Rowse, J., 2009. Downtown parking in auto city. Reg. Sci. Urban Econ. 39,

1–14.Arnott, R., Rowse, J., 2013. Curbside parking time limits. Transp. Res. Part A 55,

89–110.Axhausen, K.W., Polak, J.W., 1991. Choice of parking: stated preference approach.

Transportation 18, 59–81.Azari, K.A., Sulistyo, A., Hamid, H., Rahmat, R.A.O.K., 2013. Modelling demand under

parking and cordon pricing policy. Transp. Policy 25, 1–9.Bacon, R.W., 1993. A model of travelling to shop with congestion costs. J. Transp.

Econ. Policy 27, 277–289.Baldassare, M., Ryan, S., Katz, C., 1998. Suburban attitudes toward policies aimed at

reducing solo driving. Transportation 25, 99–117.Barata, E., Cruz, L., Ferreira, J., 2011. Parking at the UC campus: problems and

solutions. Cities 28, 406–413.Barter, P.A., 2010. Off-street parking policy without parking requirements: a need

for market fostering and regulation. Transp. Rev. 30, 571–588.Benenson, I., Martens, K., Birfir, S., 2008. PARKAGENT: an agent-based model of

parking in the city. Comput. Environ. Urban Syst. 32, 431–439.Bonsall, P., Young, W., 2010. Is there a case for replacing parking charges by road

user charges? Transp. Policy 17, 323–334.Button, K., 2006. The political economy of parking charges in “first” and “second-

best” worlds. Transp. Policy 13, 470–478.Caicedo, F., 2009. The use of space availability information in PARC systems to

reduce search times in parking facilities. Transp. Res. Part C 17, 56–68.Caicedo, F., 2010. Real-time parking information management to reduce search

time, vehicle displacement and emissions. Transp. Res. Part D 15, 228–234.Caicedo, F., 2012. Charging parking by the minute: what to expect from this parking

pricing policy? Transp. Policy 19, 63–68.Calthrop, E., Proost, S., 2006. Regulating on-street parking. Reg. Sci. Urban Econ. 36,

29–48.Calthrop, E., Proost, S., van Dender, K., 2000. Parking policies and road pricing.

Urban Stud. 37, 63–76.Chatman, D.G., Manville, M., 2014. Theory versus implementation in congestion-

priced parking: an evaluation of SFpark, 2011–2012. Res. Transp. Econ. 44,52–60.

Chone, P., Linnemer, L., 2012. A treatment effect method for merger analysis with anapplication to parking prices in Paris. J. Ind. Econ. 60, 631–656.

Cullinane, K., 1993. An aggregate dynamic model of the parking compliancedecision. Int. J. Transp. Econ. 20, 27–50.

Cutter, W.B., Franco, S.F., 2012. Do parking requirements significantly increase thearea dedicated to parking? A test of the effect of parking requirements values inLos Angeles County. Transp. Res. Part A 46, 901–925.

De Nijs, R., 2012. The price discrimination effect of a large merger of parkinggarages. Econ. Lett. 117, 928–931.

Elliott, J.R., Wright, C.C., 1982. The collapse of parking enforcement in large towns:some causes and solutions. Traffic Eng. Control 23, 304–310.

Feeney, B.P., 1989. A review of the impact of parking policy measures on traveldemand. Transp. Plan. Technol. 13, 229–244.

Fosgerau, M., de Palma, A., 2013. The dynamics of urban traffic congestion and theprice of parking. J. Public Econ. 105, 106–115.

Froeb, L., Tschantz, S., Crooke, P., 2003. Bertrand competition with capacityconstraints: mergers among parking lots. J. Econom. 113, 49–67.

Gallo, M., D'Acierno, L., Montella, B., 2011. A multilayer model to simulate cruisingfor parking in urban areas. Transp. Policy 18, 735–744.

Gillen, D.W., 1977. Estimation and specification of the effects of parking costs onurban transport mode choice. J. Urban Econ. 4, 186–199.

Glazer, A., Niskanen, E., 1992. Parking fees and congestion. Reg. Sci. Urban Econ. 22,123–132.

Guo, L., Huang, S., Zhuang, J., Sadek, A.W., 2013. Modeling parking behavior underuncertainty: a static game theoretic versus a sequential neo-additive capacitymodeling approach. Netw. Spat. Econ. 13, 327–350.

Guo, Z., 2013a. Does residential parking supply affect household car ownership?The case of New York City. J. Transp. Geogr. 26, 18–28.

Guo, Z., 2013b. Residential street parking and car ownership. J. Am. Plan. Assoc. 79,32–48.

Guo, Z., Ren, S., 2013. From minimum to maximum: impact of the London parkingreform on residential parking supply from 2004 to 2010?. Urban Stud. 50,1183–1200.

Hasker, K., Inci, E., 2014. Free parking for all in shopping malls. Int. Econ. Rev. 55,1281–1304.

Hensher, D.A., King, J., 2001. Parking demand and responsiveness to supply, pricingand location in the Sydney central business district. Transp. Res. Part A 35,177–196.

Inci, E., Lindsey, C.R., 2014. Garage and Curbside Parking Competition with SearchCongestion. Working Paper, ⟨http://www.ieb.ub.edu/files/Inci,%20Eren.pdf⟩.

International Council of Shopping Centers, Urban Land Institute, 2003. ParkingRequirements for Shopping Centers: Summary Recommendations and ResearchStudy Report. Urban Land Institute, Washington.

Ison, S.G., Wall, S., 2002. Attitudes to traffic-related issues in urban areas of the UKand the role of workplace parking charges. J. Transp. Geogr. 10, 21–28.

Jakle, J., Sculle, K., 2004. Lots of Parking: Land Use in a Car Culture. University ofVirginia Press, Charlottesville.

Jansson, J.O., 2010. Road pricing and parking policy. Res. Transp. Econ. 29, 346–353.Jia, W., Wachs, M., 1999. Parking requirements and housing affordability: a case

study of San Francisco. Transp. Res. Rec. 1685, 156–160.Kelly, J.A., Clinch, J.P., 2006. Influence of varied parking tariffs on parking occupancy

levels by trip purpose. Transp. Policy 13, 487–495.Kelly, J.A., Clinch, J.P., 2009. Temporal variance of revealed preference on-street

parking price elasticity. Transp. Policy 16, 193–199.Kladeftiras, M., Antoniou, C., 2013. Simulation-based assessment of double-parking

impacts on traffic and environmental conditions. Transp. Res. Rec. 2390,121–130.

Kobus, M.B.W., Gutierrez-i-Puigarnau, E., Rietveld, P., van Ommeren, J.N., 2013. Theon-street parking premium and car drivers' choice between street and garageparking. Reg. Sci. Urban Econ. 43, 395–403.

Lai, F., 2006. Subsidized parking and neighborhood nuisances: a supplement.Interdiscip. Inf. Sci. 12, 125.

Lan, L.W., Kanafani, A., 1993. Economics of park-and-shop discounts: a case ofbundled pricing strategy. J. Transp. Econ. Policy 27, 291–303.

Levinson, D., 2005. Review of the high cost of free parking by Donald Shoup. J. Am.Plan. Assoc. 71, 459.

Levy, N., Martens, K., Benenson, I., 2012. Exploring cruising using agent-based andanalytical models of parking. Transportmetrica 9, 773–797.

Lin, H., Wang, Y., Competition and price discrimination: evidence from the parkinggarage industry. J. Ind. Econ., forthcoming.

Lindsey, C.R., West, D.S., 1997. Spatial price discrimination: the use of parkingcoupons by downtown retailers. Rev. Ind. Organ. 12, 417–438.

Lindsey, C.R., West, D.S., 1998. An empirical analysis of retailer participationdecisions in a downtown parking coupon program. J. Reg. Sci. 38, 1–22.

Madsen, E., Mulalic, I., Pilegaard, N., 2013. A Model for Estimation of the Demandfor On-street Parking. MPRA Paper 52361, University Library of Munich,Germany.

Manville, M., 2013. Parking requirements and housing development. J. Am. Plan.Assoc. 79, 49–66.

Manville, M., Beata, A., Shoup, D.C., 2013. Turning housing into driving: parkingrequirements and density in Los Angeles and New York. Hous. Policy Debate 23,350–375.

Marsden, G., 2006. The evidence base for parking policies—a review. Transp. Policy13, 447–457.

Martens, K., Benenson, I., Levy, N., 2010. The dilemma of on-street parking policy:exploring cruising for parking using an agent-based model. In: Jiang, B., Yao, X.(Eds.), Geospatial Analysis and Modelling of Urban Structure and Dynamics.GeoJournal Library, Vol. 99. Springer Science Business Media B.V., pp.121–138.

Merriman, D., 1997. Subsidized parking and neighborhood nuisances. J. Urban Econ.41, 198–201.

Merriman, D., 1998. How many parking spaces does it take to create one additionaltransit passenger? Reg. Sci. Urban Econ. 28, 565–584.

Millard-Ball, A., Weinberger, R.R., Hampshire, R.C., 2014a. Comment on Pierce andShoup: evaluating the impacts of performance-based parking. J. Am. Plan.Assoc. 79, 330–336.

Millard-Ball, A., Weinberger, R.R., Hampshire, R.C., 2014b. Is the curb 80% full or 20%empty? Assessing the impacts of San Francisco's parking pricing experiment.Transp. Res. Part A 63, 76–92.

Mingardo, G., van Meerkerk, J., 2012. Is parking supply related to turnover ofshopping areas? The case of the Netherlands. J. Retail. Consum. Serv. 19,195–201.

Molenda, I., Sieg, G., 2013. Residential parking in vibrant city districts. Econ. Transp.2, 131–139.

Ottosson, D.B., Chen, C., Wang, T., Lin, H., 2013. The sensitivity of on-street parkingdemand in response to price changes: a case study in Seattle, WA. Transp.Policy 25, 222–232.

Petiot, R., 2004. Parking enforcement and travel demand management. Transp.Policy 11, 399–411.

Pierce, G., Shoup, D.C., 2013. Getting the prices right. J. Am. Plan. Assoc. 79, 67–81.Qian, Z., Xiao, F., Zhang, H.M., 2011. The economics of parking provision for the

morning commute. Transp. Res. Part A 45, 861–879.Qian, Z., Xiao, F., Zhang, H.M., 2012. Managing morning commute traffic with

parking. Transp. Res. Part B 46, 894–916.Rietveld, P., Koetse, M., Woudenberg, S., 2002. Parking in cities: how essential is

parking for healthy central shopping areas? Neth. Geogr. Stud., 29–46.Roth, G., 1965. Paying for parking. Hobart Papers. Institute of Economic Affairs,

London.Russo, A., 2013. Voting on road congestion policy. Reg. Sci. Urban Econ. 43, 707–724.




























































































http://www.ieb.ub.edu/files/Inci,%20Eren.pdf












































































Rye, T., Hunton, K., Ison, S.G., Kocak, N., 2008. The role of market research andconsultation in developing parking policy. Transp. Policy 15, 387–394.

Rye, T., Ison, S.G., 2005. Overcoming barriers to the implementation of car parkingcharges at UK workplaces. Transp. Policy 12, 57–64.

Sabir, M., van Ommeren, J., Rietveld, P., 2013. Weather to travel to the beach. Transp.Res. Part A 58, 79–86.

San Francisco Municipal Transportation Authority, 2011. SFpark: Putting Theory intoPractice. Retrieved from ⟨http://sfpark.org/wp-content/uploads/2011/09/�sfpark_aug2011projsummary_print-2.pdf⟩.

San Francisco Municipal Transportation Authority, 2014. SFpark: Pilot ProjectEvaluation, The SFMTA's Evaluation of the Benefits of the SFpark Pilot Project.Retrieved from ⟨http://direct.sfpark.org/wp-content/uploads/eval/SFpark_Pilot_Project_Evaluation.pdf⟩.

Shiftan, Y., Golani, A., 2005. Effect of auto restraint policies on travel behavior.Transp. Res. Rec. 1932, 156–163.

Shin, J., Jun, H., 2014. A study on smart parking guidance algorithm. Transp. Res.Part C 44, 299–317.

Shoup, D.C., 1995. An opportunity to reduce minimum parking requirements. J. Am.Plan. Assoc. 61, 14–28.

Shoup, D.C., 1997. Evaluating the effects of cashing out employer-paid parking:eight case studies. Transp. Policy 4, 201–216.

Shoup, D.C., 1999. The trouble with minimum parking requirements. Transp. Res.Part A 33, 549–574.

Shoup, D.C., 2004. The ideal source of local public revenue. Reg. Sci. Urban Econ. 34,753–784.

Shoup, D.C., 2005a. Parking Cash Out. American Planning Association, Chicago.Shoup, D.C., 2005b. The High Cost of Free Parking. American Planning Association,

Chicago.Shoup, D.C., 2006. Cruising for parking. Transp. Policy 13, 479–486.Small, K., Verhoef, E., 2007. The Economics of Urban Transportation. Routledge,

Abingdon, UK.Sutherland, S.M., 1959. Shopping center parking problems. Proc. Okla. Acad. Sci. 39,

179–182.Teodorovic, D., Lucic, P., 2006. Intelligent parking systems. Eur. J. Oper. Res. 175,

1666–1681.Tezcan, H.O., 2012. Using parking pricing as a travel demand management tool at a

university campus: an example for Istanbul Technical University. Transp. Lett.:Int. J. Transp. Res. 4, 181–192.

Thomson, R.G., Richardson, A.J., 1998. A parking search model. Transp. Res. Part A32, 159–170.

Tsai, J., Chu, C., 2006. Economics analysis of collecting parking fees by a private firm.Transp. Res. Part A 40, 690–697.

Vaca, E., Kuzmyak, J.R., 2005. Parking Pricing and Fees. TCRP Report 95, Chapter 13.Transit Cooperative Research Program, Transportation Research Board.Retrieved from ⟨www.trb.org/publications/tcrp/tcrp_rpt_95c13.pdf⟩.

van Ommeren, J.N., de Groote, J., Mingardo, G., 2014. Residential parking permitsand parking supply. Reg. Sci. Urban Econ. 45, 33–44.

van Ommeren, J.N., Russo, G., 2014. Time-varying parking prices. Econ. Transp. 3,166–174.

van Ommeren, J.N., Wentink, D., 2012. The (hidden) cost of employer parkingpolicies. Int. Econ. Rev. 53, 965–977.

van Ommeren, J.N., Wentink, D., Dekkers, J., 2011. The real price of parking policy.J. Urban Econ. 70, 25–31.

van Ommeren, J.N., Wentink, D., Rietveld, P., 2012. Empirical evidence on cruisingfor parking. Transp. Res. Part A 46, 123–130.

Verhoef, E., Nijkamp, P., Rietveld, P., 1995. The economics of regulatory parkingpolicies: The (im)possibilities of parking policies in traffic regulation. Transp.Res. Part A 29, 141–156.

Vickrey, W.S., 1954. The economizing of curb parking space. Traffic Eng., 62–67(Reprinted in J. Urban Econ. 36, 1994, 42–65.).

Vickrey, W.S., 1969. Congestion theory and transport investment. Am. Econ. Rev. 59,251–260.

Voith, R., 1998. Parking, transit, and employment in a central business district.J. Urban Econ. 44, 43–58.

Watters, P., O'Mahony, M., Caulfield, B., 2006. Response to cash outs for work placeparking and work place parking charges. Transp. Policy 13, 503–510.

Weinberger, R., 2012. Death by a thousand curb-cuts: evidence on the effect ofminimum parking requirements on the choice to drive. Transp. Policy 20,93–102.

Willson, R.W., 1992. Estimating the travel and parking demand effects of employer-paid parking. Reg. Sci. Urban Econ. 22, 133–145.

Willson, R.W., 1996. Local Jurisdiction Parking Requirements: a Survey of Policiesand Attitudes. Parking Survey Working Paper, California State PolytechnicUniversity, Department of Urban and Regional Planning.

Willson, R.W., 2013. Parking Reform Made Easy. Island Press, Washington, DC.Willson, R.W., Shoup, D.C., 1990. Parking subsidies and travel choices: assessing the

evidence. Transportation 17, 141–157.Yang, H., Liu, W., Wang, X., Zhang, X., 2013. On the morning commute problemwith

bottleneck congestion and parking space constraints. Transp. Res. Part B 58,106–118.

Zhang, X., Huang, H.-J., Zhang, H., 2008. Integrated daily commuting patterns andoptimal road rolls and parking fees in a linear city. Transp. Res. Part B 42,38–56.

Zhang, X., Yang, H., Huang, H.-J., 2011. Improving travel efficiency by parkingpermits distribution and trading. Transp. Res. Part B 45, 1018–1034.

Zhang, X., Yang, H., Huang, H.-J., Zhang, H.M., 2005. Integrated scheduling of dailywork activities and morning–evening commutes with bottleneck congestion.Transp. Res. Part A 39, 41–60.









http://www.sfpark.org/wp-content/uploads/2011/09/~sfpark_aug2011projsummary_print-2.pdf



http://www.direct.sfpark.org/wp-content/uploads/eval/SFpark_Pilot_Project_Evaluation.pdf

http://www.direct.sfpark.org/wp-content/uploads/eval/SFpark_Pilot_Project_Evaluation.pdf






























http://www.trb.org/publications/tcrp/tcrp_rpt_95c13.pdf













































Documents

Economics of Transportation