a game theory approach to the analysis of land and property development processes

Land Use Policy 27 (2010) 564578

Contents lists available at ScienceDirect

Land Use Policy

journa l homepage: www.e lsev ier .com/ l

A game adevelo

D. Ary A. eema Department o e Nethb Department o gen, T

a r t i c l

Article history:Received 4 NoReceived in reAccepted 24 Ju

Keywords:Land and property developmentGame theory

ssesviduaalyseper, ws to i

theoretical modelling for analysing and predicting the behaviour of actors in decision-making processeswith respect to the development of land and property. For that purpose, we have developed gamemodelsfor the case study of the development of a greeneld residential location in the Netherlands with respectto the implementation of new Dutch legislation on cost recovery.

Our study demonstrates that game theory could help us to identify the key strategic decisions of landand property development projects by showing the different payoffs for stakeholders of their chosen

1. Introdu

Land anseen as socgroups of inand Munto1988). To sto analyseand to anal

Decisionproperty dcesses (Alexgroups of aproperty dend users astrategies dthe utility fally based o

CorresponE-mail add

0264-8377/$ doi:10.1016/j.strategies and selecting the equilibrium in which all stakeholders involved are best of. We also foundmany limitations of using game theory in our case study especially regarding the assumptions underlyingthe model. However, we conclude that game theoretical modelling can be a useful decision support toolin spatial planning, because it provides a way to think about the complexity of strategic interaction and,in particular, about the conicting structure of collective decision-making processes.

2009 Elsevier Ltd. All rights reserved.

ction

d property development processes obviously can beial situations in which the interaction of individuals ordividuals is one of the essential elements (Goodchildn, 1985; Adams and May, 1991; Adams et al., 1985,tudy and understand social situations, it is importantthe decisions of actors and how they are interrelatedyse how those decisions result in outcomes.-making processes, especially with respect to land andevelopment, are, almost by denition, complex pro-ander, 1965; Byrne, 2003;DeRoo, 2004). First, differentctors or stakeholders are involved (e.g., landowners,evelopers, various municipal departments, investors,nd real estate agents) and they respond to each othersifferently and also with different rationalities. Second,unctions of most of the stakeholders involved are usu-n more than one goal. Financial aspects are of course

ding author. Tel.: +31 24 3616206; fax: +31 24 3611841.ress: [email protected] (D.A.A. Samsura).

very relevant to decision-making, but the same goes for abstractthemes like spatial quality and the wish to continue cooperationafter completion of the project. These utility functions probablyvary for each stakeholder. Third, many studies have demonstratedthe strong link between the institutional context and market pro-cesses which in the end may change the stakeholders strategies(Healey, 1992; Van der Krabben and Lambooy, 1993; Keogh andDArcy, 1999;Adamset al., 2005; Buitelaar, 2007). And fourth, thereis interdependency among stakeholders decisions in which thedecision of one stakeholder will inuence or be affected by thedecision of other stakeholders.

This complexity issue is taken in this paper as the starting pointfor the analysis of decision-making in land and property develop-ment. The interactions among stakeholders in land and propertydevelopment processes can be investigated by case study analysis.However, one of the disadvantages of this approach is the limitednumber of successful cases. Moreover, using case studies analy-sis could bear some difculties regarding the inuence of uniquefactors on the results and hence it would lead to the difculties ofgeneralizationof the insights inbehaviourof stakeholders.Analter-native research approach is to use more experimental settings foranalysis. This approach offers the opportunity to study behavioural

see front matter 2009 Elsevier Ltd. All rights reserved.landusepol.2009.07.012theory approach to the analysis of landpment processes

Samsuraa,, Erwin van der Krabbena, A.M.A. van Df Spatial Planning, Nijmegen School of Management, Radboud University Nijmegen, Thf Strategy and Marketing, Nijmegen School of Management, Radboud University Nijme

e i n f o

vember 2008vised form 22 July 2009ly 2009

a b s t r a c t

Land and property development proceaction of individuals or groups of indisocial situations, it is important to andecisions result in outcomes. In this pait. Hence, the objective of the paper iocate / landusepol

nd property

enb

erlandshe Netherlands

obviously can be seen as a social situation in which the inter-ls is one of the essential elements. To study and understandhow the decisions of actors are interrelated and how thosee propose a game theoretical modelling approach to analyse

nvestigate the usefulness as well as the limitations of game

D.A.A. Samsura et al. / Land Use Policy 27 (2010) 564578 565

aspects in relation to responses on controlled stimuli. Simulationand gaming tools have often been used in experiments with realplayers to analyse decision-making processes in land and propertydevelopme2007; Devigame theorpaper is, inas the limitpredictingwith respecgaming expthe advantathe equilibrtheory is abactors in lalead us todevelopmedevelopmeenvironmenthat many utions that c

To inveswith respecdevelopmecase studyin the Nethbehaviour oto the impl(the Land Dcase study otion that thactors invol(Korthals Awill thus bestrategies othe equilibr

The struan introducadvantagesplanning andiscusses th5 we use gaeld reside4 presents aresidentialtheir strateate. In Sectithe outcommentation o6 discussesproperty defurther ana

2. Why gam

Game thwhich the dand the outparty or acttheory (Myesion theoryplayers posis in the infocuses onplay a role,

The term game theory stems from the resemblance of col-lective decision-making situations to well-known parlour gameslike chess, poker, and monopoly (Aumann, 1989). Because of its

on conicting preferences, game theory is often denedeory of conict (Myerson, 1991; Luce and Raiffa, 1957).n has even proposed to speak of Interaction Decision instead of Game Theory, since the former namemore accu-describes the content and focus of the theory (Myerson,Modern game theory began in the 1920s with the work of1921) and Von Neumann (1928). But it was after the sem-ork of Von Neumann and Morgenstern (1944) that gametriggered an outburst of interests, especially among math-ians and economists, and started to ower as a eld ofSincer ter s, psyColmed ad onstudnd srmortandn-mticalmostprob1

h replicaan,nderplanues oe mun cuam,, mad are-spnd pen seorteopertheoral., 1e the, a ssom

realiatedons h

studyhe Logparticnaturintereal intessedbmma td. For1971,nt (see e.g., Devisch et al., 2005; Havel and Viitanen,sch, 2008). In this paper, alternatively, we propose aetical modelling approach. Hence, the objective of thethe above context, to investigate the usefulness as wellations of game theoretical modelling for analysing andthe behaviour of actors in decision-making processest to the development of land andproperty. Compared toeriments, game theoreticalmodelling potentially offersge to select the best possible outcomes of a game, i.e.iums. Moreover, we will argue in this paper that gamele to conceptualise the structure of relations betweennd and property development, which in the end maya better understanding about the land and propertynt process. This may support, for instance, the choice ofnt models in the context of the institutional-economict. On the other hand,we also know from earlier studiesnderlying assumptions in game theory contain limita-ould reduce the applicability of the results.tigate the usefulness of game theoretical modellingt to collective decision-making in land and propertynt processes, we have developed a game model for theof the development of a greeneld residential locationerlands. In this case study, we analyse the strategicf private developers and municipalities with respectementation of new Dutch legislation on cost recoveryevelopment Act, issued July 1st, 2008). The focus in then the Land Development Act follows from our assump-e new legislation not only affects the strategies of theved, but also the interrelationships between the actorsltes, 2006; Van Dinteren et al., 2008). The game modelemployed to analyse the effects of the actors changingn both the outcomes of each others strategies and oniums that can be achieved.cture of this paper is as follows. Section 2 providestion to game theory and a discussion of the supposedas well as limitations of applying game theory to urband land and property development studies. Section 3en how game theory actually works. In Sections 4 andme theory to construct a model with respect to green-ntial location development in the Netherlands. Sectionbrief description of the land development process for

use in the Netherlands, introducing the stakeholders,gies and the institutional context in which they oper-on 5 the modelling analysis is carried out. It will showes of the game model, both before and after the imple-f the new legislation on cost recovery. Finally, Sectionthe usefulness of game theory for modelling land andvelopment processes and suggests the next steps forlysis.

e theory?

eory is a theory of interdependent decision-making inecision-makers involved have conicting preferencescome of their decisions cannot be determined by oneor only. The roots of game theory derive from decisionrson, 1991). However, there is a clear distinction. Deci-usually analyses decision-making processes from oneint of view, while game theory emphasizes its analy-teraction among many players. Because game theorysituations in which interactions and interdependencyit can be seen as an extension of decision theory.

focusas a thAumanTheoryrately1991).Borel (inal wtheoryematicstudy.ing othcompupology1990;increasis basefor thetures aFurtheundersdecisiotheoreof theing of2000).

Witthe ap(Berkmhaps uspatialThe issbecomingly, iNeedharguedinvolvethe casland amay this suppand prgame t(Ball et

Gamexercisever, inof theapprecsituatistudy.

1 TheOlson: Tfocusesthere is aof thoseindividuAsdiscuers DileexplaineHardin,e then, game theory has been profoundly inuenc-elds in natural science, such as biology, physics, andcience, as well as social science, including anthro-chology, sociology, politics, and philosophy (Kreps,an, 1999; Hargreaves Heap and Varoufakis, 2004). Thettention to game theory especially in social sciencethe idea that it can provide solid micro-foundationsy of collective decision-making processes and struc-ocial change (Elster, 1982; Aumann and Hart, 1992).e, game theory has the potential to improve ouring of social interactions in general and interactiveaking in particular. One important example is the gamemodel of the prisoners dilemma which has been oneprominent analytical foundations for the understand-lems of collective action (Hardin, 1971, 1982; Ostrom,

spect to land and property development processes,tions of game theory so far are limited in number1965; Mu and Ma, 2007). The lack of attention is per-standable because land and property development andning are characteristically very much context-driven.f pluriformity, complexity and interdependency havech more important in urban societies and, accord-rrent planning literature (Healey, 1992; Taylor, 1998;2007). Modelling those processes, so it is sometimeskes no sense, because the interactions of the actorse undeniably too complex and so much dependent onecic circumstances. To argue in favour of modellingroperty development processes in simplied modelsem to be odd. However, we will argue otherwise. Thisd by a review of the analytical tools for studying landy development processes by Ball et al., which considersy as a promising methodology in this eld of research998; Ball, 1998).eoretical modelling indeed implies, like any modellingimplication and abstraction from the realworld.How-e disciplines, for instance in economics, the translationworld into a formal model is very much accepted and. Mathematical approaches to the analysis of complexave evolved through time as a very signicant eld of

of collective actions has been introduced in the classical work ofic of CollectiveAction (Olson, 1965).Olsons analysis of collective actionularly on the problem of public goods provision. According to Olson,al tendency for peoplewith shared interests to act together in pursuitsts. However, there is sometimes a problem of congruence betweenrests and group interestswhich is notwell explained in Olsonswork.yHardin (1971, 1982), using thegame theoreticalmodel of thePrison-he divergence of individual interest and group interest can be clearlymore information about collective action and game theory (see e.g.,1982; Ostrom, 2000, 1990; Olson, 1965; Oliver, 1993).

566 D.A.A. Samsura et al. / Land Use Policy 27 (2010) 564578

Beingawareof the limitations tomodellingexercisesof complexsituations, there are some good reasons to look at the possibilitiesof game theory as a decision support tool for the analysis of landand property development. First, one of the benets of using gametheory is that the constructionof amodel of complex landandprop-ertydevelopment requires todeneveryprecisely theassumptionsabout the complex reality underlying the model (see Section 5). Inline with that, as a formal model, game theory allows us to seeexactly howtions. Of couthe outcomprecisely. Inture for an aamodel alloto derive netion of a proin order todifferent wor land andof a formalother situatmake amorincrease theinternation

Second,off maximicomplete inbutions toincrease itsbeen capabels, includinemotions ainformationIn this resitage and idecision-mtheory (seetions of gathat. In thigame theothat it struclear way.ing of thebe renedmodel.

Third, reand validatelling. In th(van Deemeers involvedcan help toa more gen

3. How do

This secory actuallto investigathe analysi

2 Several auferability of pl1998; DolowiDolowitz and

the basic concepts of game theory are explained here, includ-ing games, players, strategies, outcomes and payoffs and solutionconcepts.3

3.1. Games

Any game consists of two parts, namely a descriptive partwhichdescribes t

es oThismaticns dion octivens. Tnothof en atmewameve foNeu(Sectn exttheposide s, thered bcal re1.es artreees reose fotheis arrutcoill hahisme.ith

sion.possic ford. Cof infic forbstrale, ites ining tin exdesct couil, thes aler cforssary

ral inrmichthe conclusions of a model follow from its assump-rse, the assumptions can and should be discussed, bute of game theoretical modelling can be described veryaddition, formalmodelling also creates a logical struc-rgument and its accumulation. The logical structure ofws amodeller to add to it or rub out something from itw conclusions and leads to new insights or the clarica-blem. Moreover, sensitivity analysis can be carried outexamine the robustness of the model. There are manyays to dene issues and arguments in urban planningproperty development. However, without the rigourmodel, we might fail to generalize such argument toions. And moving away from complexity by trying toe general structure of argumentsmay also contribute totransferability of planning anddevelopment strategiesally.2

many have criticized the notion of rational or pay-zing players underlying game theory, which assumesformation for each player. However, recent contri-game theory seem to offer new opportunities toapplicability. In the last decade game theory has

le of developing more realistic decision-making mod-gbounded rationalitymodels,models takingaccountofnd intuitive decision-making, models with incompleteand models with asymmetric information positions.

pect, game theory is leaving its mathematical her-s becoming more and more a behavioural theory ofaking and has lead to much experimentation in gamee.g., Camerer, 2003; Camerer et al., 2004). Applica-

me theory to spatial planning can take advantage ofs paper, however, we stick to the more traditionalretical methods. The advantage of this approach isctures the situations to be studied in a simple andAs a rst step, it therefore raises our understand-decision problems at stake. In a later stage, it mayand extended into a more sophisticated or realistic

sults of earlier studies show that it is possible to teste empirically the outcomes of game theoretical mod-is respect, we refer to good results of earlier researchn, 2006). Validating the preferences of the stakehold-and testing the outcomes of the games to be played

increase the applicability of the outcomes of games ineral way.

es game theory work?

tion provides a brief introduction to how game the-y works. Since the main objective of the paper iste the usefulness of game theoretical modelling fors of land and property development processes, only

thors have addressed the problem of the limited international trans-anning and development strategies and instruments (e.g., Dolowitz,tz and Marsh, 2000; Salskov-Iversen, 2006; Bulmer et al., 2007;Medearis, 2009).

describgame.matheequatioa soluta collesolutiodates asystemsolutiotive fraform, gextensiby Vonpapergame iture ofand theers decamovestructugraphiin Fig.

Trein thebranchto chowhichcomethese oally, wmakesthe gasions wsucces

It isstrategdetailea loss ostrategmore aprincipto gamincreasgamesway totimes iin detatal issua simpnienceis neceform.

3 Gene1957; Cahe game under scrutiny, and a solution part whichr predicts the outcomes given the description of thetwo-component approach can be compared with aal problem, e.g., in linear algebra.We then have a set ofescribing the problem and, subsequently, we try to ndf the system of equations. In game theory we describedecision-making situation as a game and try to nd itshis analogy of game theory with linear algebra eluci-er important point in game theory. In linear algebra, aquations may have one solution, many solutions or noall. The same is true for games. Basically, three descrip-orks for the games are distinguished: games in strategys in coalition form or characteristic form and games inrm. These three formats have already been constructedmann andMorgenstern (1944). In the case study in thision 5) we model the decision-making processes as aensive form. In thisway,wewill arrive at a detailed pic-structure of land and property development processessible outcomes of this structure. It is assumed that play-equentially, as in playing chess. The rst player makessecond one responds and so on. Thewhole game can bey means of a so-called game tree. This can be seen as apresentation of the strategic interactions of the players

e dened in terms of nodes and branches. The nodesrepresent the decisions made by a player, while thepresent the moves or alternative actions for a playerrom. The branches end in different decision nodes atr players can make a decision and so on until an out-ived at and the game ends. The nodes representingmes are called end nodes. Note that a game tree, usu-ve many end nodes. The node at with the rst playerdecision is called the initial node or also the root ofSo, a game in extensive form is a sequence of deci-respect to possible actions, made by the players in

ible to transformgames in extensive form into games inm. However, games in extensive form are usuallymorenversions into strategic form games therefore lead toormation which in most cases is difcult to repair. Them game, based on the game in extensive form, gains inction and hence in the loss of detailed information. Inis also possible to move from games in strategy formcoalition form. Again, information will be lost whenhe level of abstraction. According to Myerson (1991),tensive form usually offer the most richly structuredribe the structure of game situations, although some-ld seem too complicated. When the game is showedis may hinder the analysis, because the fundamen-re obscured. The strategic or normal form is offeringonstruction and, conceptually, it gives more conve-purposes of general analysis. Therefore, sometimes itto represent the game in both extensive and normal

troductions to game theory include (Aumann, 1989; Luce and Raiffa,ael, 2004; Dixit and Skeath, 2004).


3.2. Players

In a gamknows wheinformationstill estimaspeak of imthe moves mprimitive teto get any mare interpreproperty de

3.3. Strateg

A strategulates a prianother plaset of differgame theorstrategic ma variety oftions have iobjectives,players alsoa plan howtheory is thbut the oveonly partial

3.4. Outcom

In any gconcepts ofcal state whcomplete sdecision, ifan outcomeDifferent plconsequentcomes. Themeans of extheory, byvidual prefe

tion oreast (Luns thst ousic qnt co

lutio

re art is trequolech noy. Thategihe coual sbesBesd to. Maheors beFig. 1. Game tree.

e, players are the decision makers. If a player exactlyre he stands in the game tree, then we speak of perfect. If a player does not knowwhere he is in the tree, he cante the probability that he is at a certain node. We thenperfect information. In this situation, he does not knowade in a previous stage by other players. A player is a

rm that has to be linked to an empirical domain in ordereaning. In the case study in Sections 4 and 5, playersted as stakeholders involved in the process of land andvelopment in the Netherlands.

y

y can be dened as a contingent plan of actions. It stip-ori how the player will act when a move is made byyer. Players are supposed to be able to choose from aent strategies. There is a similarity between a strategy inetical terms and the business meaning of strategy. Theanagement literature (see e.g., Ansoff, 1979) containsdenitions of the concept strategy, but most deni-

n common that a strategy contains long-term goals andas well as a plan how to attain them. In game theory,aim at a goal, namely utility maximization, and adopt

the noof thisconicfunctioThe beThe badiffere

3.5. So

Theever, imost fas a prin whistrategthe strity of tindividtion ofplayer.not leauniquein the ttion hato achieve this in their strategy. The essence of gameat each player makes his own selection of a strategy,rall result depends on the choices of all. So each playerly controls the outcome of a game.

e and payoff

ame, a clear distinction has to be made between theoutcome and payoff. An outcome is a social or physi-ich may result from the behaviour of the players or a

et of strategic selections in the game. In fact, it is theany, arrived at by the players collectively. The payoff offor a player is the value of that outcome for the player.ayers will, in general, value outcomes differently and,ly, will have different preferences over the set of out-players preferences over outcomes are represented bypected utility functions or, as they are called in gamepayoff functions. These presumed differences in indi-rences and hence in the players payoff functionsmake

used renemtion is the nof an extenbe considerin the gamesive form gahence therefect. TherefAn SPE canmethod staof the tree i

4 The variatdifferent typesmost commoncooperative gaNeumannMorconcepts. ForCarmichael, 20f conict very important in game theory. It is becauseon that game theory sometimes is called a theory ofce and Raiffa, 1957; Jones, 2000). The individual payoffat assign values to outcomes vary across individuals.tcome for one player may be the worst for the other.uestion therefore is how to solve games given thesenicting payoff functions.

n concepts

e several solution concepts for game theory.4 How-he concept of Nash Equilibrium (NE) that is certainlyently applied (Aumann, 1985). An NE can be denedof selected strategies one strategy for each player player has an incentive to deviate from his selected

e strategy selected by any player is a best response toes chosen by all the other players. Note the complex-ncept: each player selects a best response to any othertrategy selected. Hence, Nash equilibrium is a situa-t responses toward each other, one strategy for eacht hereby means that deviating from the response willan increase in payoff. Nash equilibrium need not beny games have multiple Nash equilibriums. Therefore,y of games in strategic and extensive formmuch atten-en paid to renements of the NE concept. A frequently

ent for games in extensive formwith perfect informa-

otion of subgame perfect equilibrium (SPE). A subgamesive form game is a part of the game that as such caned as a game on its own. It corresponds with a sub-treetree. An SPE is an NE for every subgame in an exten-me. Clearly, an SPE is always an NE, but not conversely,may beNash equilibriumswhich are not subgame per-ore, an SPE is a stronger solution concept than an NE.be found by using the backward inductionmethod. Thisrts the searching process after equilibriums at the endn a sub-tree. Rolling back through the several sub-trees

ion of solution concepts for game theory mainly ensues based on theof game that are played in different situations.Nash Equilibrium is thesolution concept to be applied in Non-cooperative games. While inmes such concepts like the core, the stable set (also known as the vongenstern solution) and Shapley value are the most important solutionmore information (see e.g., Aumann, 1989; Luce and Raiffa, 1957;04; Dixit and Skeath, 2004).


to the root of the tree and analysing for equilibriums lead to a SPE.It is a backward reasoning method starting at the endnodes of thetree.

All the concepts that have been discussed in this section will beapplied in the case study in the next two sections.

4. Case study: residential land development in theNetherlands

The main goal of the case study is to investigate the potentialusefulness of game theoretical modelling in supporting decision-making processes regarding the strategies to choose for land andproperty deoretical mothe Netherlbrownelders is usuallMoreover, tto the impholders anddemonstratestimating

4.1. Municidevelopmen

First, wepalities to dthe Dutch cfor greeneplan is impthe land decalled activemodel, whicbe developeplots suitabthem to buNeedham, 2in situationthe locationopers usualwithout tryto the munserviced lanthey sold ththe buildingthe developlocation forNeedham a

An oftenland developublic privathe privateadvantage o

5 Note thatgoals. The resinvolved, due

6 For more ipolicies we ref2002).

7 It is hard fothat have alreathat they haveto do this. In th

is that they can share nancial risks and expertise. In this model,the public and private stakeholders involved establish a joint ven-ture land development company which takes over the role of themunicipality in the public development model. This joint venturecompany will come to an end, as soon as all building plots are sold.The shareholders in this joint venture generally agree not to makeany prots, but to accept that only the costs of land developmentwill be covered by the sale of building plots. In turn, the privatedevelopers will have the rst right to buy serviced building plots,against prices that may be below market value, depending on thenegotiations between the municipality and the private developer.To supportpany, the m

ssaryriatiabopmelandd devpmes invsing

gh theld rels ant stn devwithpmetivelypmeual dchooe, thpmepmethervelonot

rivatght till bted brticiuallyvicefor tpertie extionncialpateut thuatiots athe

municvelopntial pred br parelopmcompvelopment. Our case study concerns the game the-delling of greeneld residential land development inands. Compared to, for instance, the transformation oflocations into residential use, the number of stakehold-y limited, which helps us to use it as a clear study case.he choice for this case study allows us to pay attentionact of new legislation on the strategies of the stake-, consequently, on the outcome of the game. This wille the usefulness of game theoretical modelling, whenthe effects of changes in the institutional context.5

pal strategies for greeneld residential landt in the Netherlands

must dene the possible strategies for the munici-evelop a greeneld location for residential use withinontext.6 The municipality usually takes the initiativeld residential developments. To make sure that thelemented, municipalities often decide to take part invelopment process (Needham, 2006). This approach island policy and is referred to as the public developmenth means that the municipality acquires all the land tod, services the land, readjusts the parcels into buildingle for the desired development and after that releasesilders/developers and end users (Van der Krabben and008). This public developmentmodel is also often useds that private developers have already acquired land onwith the intention to build houses. The private devel-

ly agree to sell their unserviced land to themunicipalitying tomake prots. In return, they hold a building claimicipality that guarantees them the rst right to buy thed, against prices that have been agreed upon beforee unserviced land. This model is usually referred to asclaim model. This approach excludes competitors fromment and ensures the private developer a high-qualityprotable housing development (Korthals Altes, 2006;nd Verhage, 1998; Boelhouwer, 2005).7

used alternative development strategy to the publicpment model is a land development model based on ate partnership between the municipality and some ofdevelopers that have acquired land on the location. Thef this model for both the public and the private actors

the case study, in its present form, is primarily meant for scienticults are only of limited use for strategic advice to the stakeholdersto the empirical restrictions to the game (see Section 5).nformation about the Dutch residential land market and residentialer to, among others (Korthals Altes, 2006; Needham, 2007; Verhage,

r the municipality to refuse cooperation with the private developersdy acquired land. According to Dutch law, landowners that can provethe skills to carry out the planned development have the legal rightis situation they cannot be expropriated by the municipality.

if neceexprop

ThedeveloNetherthe landeveloholderfor houalthougreenor modernmelocatioopers,developro-acdevelothe actities topracticdevelodevelo

Anoland dethat doother plegal ripany wsupporoper pawill usthe servationthe exthan thnegotiaof naparticiment, bthis sitno protion in

8 Theon the dea substabe acquidevelopecient devventurethe development strategy of the joint venture com-unicipality usually agrees to use its land policy tools,, including the possibility of a pre-emption right andon powers.ve models of public or public private residential landnt usually work well for most municipalities in thes. However, the active involvement of municipalities inelopment process is notwithout risks. Sometimes landnt processes will result in nancial losses for the stake-olved, for instance because of a drop in the demandand, subsequently, the demand for land. Therefore,e development models above are commonly used foresidential developments, still alternative approachesre in use as well. One of them is known as a local gov-rategy based on facilitating land policy. In this model,elopment is carried out by one or more private devel-out any interference of the municipality in the landnt process. The instruments of land acquisition are not

used in this model. The municipality initiates thent by issuing a land use plan, but is not involved inevelopment of the location. One reason for municipal-se this strategy is to reduce land development risks. Inis development model is often used for ofce or retailnt in the Netherlands, but hardly ever for residentialnt.strategy for the municipality is to form a joint venturepment company with one or more private developersown any land on the location, despite the fact thate developers do own parts of the location and have theo develop it. The intention of this joint venture com-e then to acquire the remaining land on the location,y the municipal land policy tools. The private devel-pating in the joint venture land development companybe offered a building claim or the rst right to buy

d land that is owned by the joint venture. The moti-he municipality to choose this strategy might be thatse of the selected developer is believed to be betterpertise of the land-owning private developers; or, thes with the land-owning developers have failed becausereasons.8 The motivation for the private developer tois, again, not primarily the protability of land develop-e chance to buy the serviced land to develop houses. Inn, private developers would usually also accept low ort all from land development, as long as their participa-land development process will guarantee them to buy

ipality may consider this strategy in a situation that part of the landment location has been already acquired by private developers,whileart of the land is still in the hands of the original owners and has toy the joint venture development company. This will guarantee theticipating in the joint venture company that he will still have suf-ent opportunities on the locations that will be acquired by the joint

any.


Table 1Land development strategies.

Land development strategy Initial situation on theland market

Acquisition of agricultural land Servicing andreparcelling the land

Acquisition of building plots

Active land policy by the municipality(1) Public land development Original owners Municipality acquires all land Municipality Private developers buy numerous

(2) Building claim model Private developers withintentions to buildhouses

Municipality acquires all land

(3)Public private partnership model Original owners Joint venture companyrivate

rivate

Facilitating l(4) Private

building plobeforehand

A nal ston the locahousehold)this paper wnot often usdevelopmeate with printegrated dfor municipprivate devas we explaa choice, be

Table 1that are ava

4.2. The Neproperty de

Themaito the devepartnershipon the spatwhich theyities will aldevelopmeuations, howto acquire anot agree wprivate devthem. Somerelated infr2007). Theying the lega

9 This startewide policy to(the so-calledvate developeshown any intto the buildinmore informat2006).

costunicstmeheores didedt recthend Delopcosexibnd toeld ahenforcle, thesultconslandainedprivaals Aotheperscost(including land-owning pdeveloper)

Private developers withintentions to buildhouses

Joint venture company(excluding land-owning pdevelopers)

and policy by the municipalityland development model Original owners Private developers;

End users

ts against acceptable prices which is to be negotiated.rategy for the municipality might be to acquire all landtion and sell it directly to the end user (the individual, offering the end user to construct its own house. Ine leave aside this approach, because municipalities doe it for residential development (though it is the regularnt model for industrial estates). They prefer to cooper-ofessional private developers, because they aim at theevelopment of the location.Moreover, it ismuch easieralities to sell the serviced land to a limited number ofelopers instead of to thousands of individuals. Besides,ined before, in many cases municipalities do not havecause private developers acquired all land.summarises the different land development strategiesilable for municipalities.

w Land Development Act and its impact on land andvelopment

n reason for adopting an active land policy with respectlopment of greeneld sites (public or public privatemodel) is that municipalities can exert more inuenceial development process compared to the situation inadopt a facilitating land policy. Moreover, municipal-so have an opportunity to gain some prots from the

costs (somemlic invegame toutcomnet decfor cosvicingnew La(re)devrelatedmore ment agreenEvenwcan beprincipment r

Thein thehave gof the(Korthon thedevelorelatednt even though it is not their main purpose. In some sit-ever, it appeared to be impossible for themunicipality

ll land. Some of the land-owning private developers didith the public or publicprivate model and aimed forelopment of the part of the location that is owned byof them also refused to contribute to the costs of plan-astructure and public space development (Needham,are considered as free riders.9 Municipalitieswere lack-l tools to require a nancial contribution to plan-related

d in the 1990s, after the national government had launched a nation-develop a substantial number of large residential greeneld locationsVINEX locations). The unexpected effect of this policy was that pri-rs started to acquire land at the VINEX locations (while they had noterest in acquiring land before). The private developers strategies ledg claim model and, in some situations, to free rider problems. Forionabout thedevelopmentofVINEX locations (seee.g., KorthalsAltes,

quality of ththat will betowhat extefor the stak

5. Game co

5.1. Framew

For thisgame in extconcept forthebest strawill allow u

10 A short inton www.vrombuilding plotsEnd user buys single building plot

Municipality Private developers with building claimbuy building plots

Joint venturecompany

Private developers with building claimbuy building plots

Private developers;end users

End users buy building plots; end usersalready own building plots

recovery) from those free riders. This has resulted, foripalities, in seriousproblemswith cost recoveryofpub-nts in public land development processes. To put it inetical terms, the payoff for the municipality of someecreased substantially. For that reason, the Dutch Cabi-recently (per 1 July, 2008) to introduce new legislationovery of public investments, for example, costs of ser-land, road infrastructure and open space. Under theevelopment Act, all landowners that take part in thement can be forced, if necessary, to contribute to plan-ts. This Land Development Act is expected to provideility to municipalities with respect to location develop-make it easier for municipalities to (re)develop bothnd browneld locations (Van Dinteren et al., 2008).10

municipalities do not own land, the private landownersed to contribute to the costs of public investments. Ine new legislation guarantees them the same develop-, either with active or with facilitating land policy.equence of this is that the positions of the stakeholdersdevelopment game have changed. The municipalitiespower in the negotiation processes, while the power

te developers has decreased. Results of recent studiesltes, 2006; Van Dinteren et al., 2008; Needham, 2007),r hand, show that under the old regime most privateeven without legislation agreed to contribute to plan-s, because they expect to benet from it since a higher

e location leads to higher housing prices. In the gameconstructed in the next section wewill be able to shownt the changes in legislationmay result in otherpayoffseholders and other equilibriums.

nstruction

ork for the game

study, we have restricted ourselves to the analysis of aensive form or a game tree analysis. By using a solutionthegame, the steady-state situation canbe found tondtegy for all players, i.e. theequilibriumof thegame. Thiss to compare the outcome of complex decision-making

roduction in English to theDutch LandDevelopment Act can be found.nl/pagina.html?id=2706&sp=2&dn=7198.


in actual projects with the best outcome of the game for all players.Here we will discuss the basic elements to construct a game tree,including: players, strategies and payoffs.

5.1.1. PlayeIn linew

stakeholder

(1) the mun(2) a prope(3) a prope(4) and the

In realitinvolves mprovince, pcompetingbecause themore, in reis usually msimplicatithe four sta

Note thaventure comdevelopmesider this joprot (or, intool of thetheir utilitybetween prwho have nholders wil

5.1.2. StrateIn the ga

different lastakeholderin every gamof both connot only tryor reach agrtotal beneipality andventure comcosts of formit could be athe game, eshould be.

5.1.3. PayofIn the ga

offs at inteorder to doerty developayoffmaxiopment probe achievedimplementethese threemunicipalitimportant tthis is likeldifferent w(total weighwill make

Table 2Variables for outcomes assessment.

Players Variables

ipality (M) (1) Implemented plan(2) Giving building claim to PL(3) Public land development model(4) Property market conditions(5) Establishing a joint venture company(6) Buying agricultural land

ty developer (PD) (1) Opportunity to develop properties

ty dev

wner (

ub-gwillthe rub-gwn ie abt scores frrelevtelykeneredon othe sdix A).hin the context of this game, those variables are jointlysible for the way each player assesses the different out-. For example, with respect to variable 1, for M outcomesve led to the implementation of the plan will be assessedthan outcomes that have not led to the implementation ofn.With respect to variable 2, outcomes that are based on theg claim model will be assessed less than outcomes that arecauseMwill expect that in this building claimmodel PL willto negotiate a better nancial result. Regarding variable 3,esbasedonapublic landdevelopment strategyareassessedthan outcomes based on a private land development strat-ith respect to variable 4, property market conditions areed to inuence theoutcomeaswell. Outcomes that arebasedint venture model, as in variable 5, are assessed better thanes that are not based on a joint venture model. And, nally,spect to variable 6, it is assumed that outcomes that are

estimations are based on our own experiences in this eld and earlier(Van Dinteren et al., 2008). They are believedmore or less to represent theopinion about the strategies of the stakeholders. As has been mentioned inious section, the payoffs have not been validated. We believe this is accept-he context of this paper, considering the main purpose of the case studyto analyse the usefulness of this approach).rsith ourdiscussion in Section4,we identify fourdifferents, i.e. players for the game:

icipality (M),rty developer who has not acquired land (PD),rty developer who has acquired land (PL),original landowner (LO) which usually a farmer.

y, the process of land and property developmentore actors, such as the national government, theressure groups, landowners and/or developers in otherlocations, but we only consider those four in our game,y are the crucial stakeholders in this process. Further-al situations on large development locations thereore than one PD, LO and PL. However, for the sake ofon, we have assumed here a small location with onlykeholders.t, although in Section 4 we have mentioned the jointpany as one of the stakeholders in land and property

nt, we do not consider it as one of the players. We con-int venture company which is expected to have a zerogame theoretical perspective: zero payoff) just as a

municipality and the private developers to maximizefunctions. It also should be noted that we distinguishoperty developers who have acquired land and thoseot, because we assume that strategies of both stake-l be different.

giesmewe constructed, the strategies are derived from thend development strategies that are available for each(see Section 4.1). It is important to notice that almoste, strategic interactions of players consist of amixture

icts and common interests. It implies that players areing to compete with each other but may also cooperateeements to form a coalition in order to maximize theirts. In ourmodel, a coalition takes placewhen themunic-the property developers decide to implement a jointpany to service the land. However, we leave aside theing coalitions in the land development game, thoughconsideration for the players in assessing the payoff ofspecially when the joint venture is not functioning as it

fsme model presented here we have estimated the pay-rval level for every possible outcome in the game. Inthat, we rst conceptualised the goal of land and prop-pment process in three sub-goals that are in line withmization,which are (1) the nancial result of the devel-cess, (2) the spatial result or the spatial quality thatwill, and (3) the period of time in which the plan will bed (the sooner, the better). However, the relevance ofdifferent payoffs varies for the different players. For they, spatial quality and time are believed to be morehan nancial result, while for the property developersy to be vice versa. These differences are expressed ineights to these three sub-goal features for every playert of 100% for all three features together). Every player

an assessment to all outcomes with respect to those

Munic

Proper

Proper

Lando

three splayerreectthree sare sho

To bwe rsby scois verycomplebeen taconsidvaluatiture of(Appen

Witresponcomesthat habetterthe plabuildinnot, bebe ableoutcombetteregy. Wexpecton a jooutcomwith re

11 Ourresearchgeneralthe prevable in t(namely(2) Having building claim(3) Property market conditions(4) Selling serviced land(5) Buying serviced land(6) Buying agricultural land

eloper with land (PL) (1) Opportunity to develop properties(2) Having building claim(3) Property market conditions(4) Selling unserviced land(5) Selling serviced land(6) Risk of servicing the land(7) Buying serviced land

LO) (1) Land sold(2) Property market conditions(3) Land buyer(4) Cost of selling(5) Future occupancy of the land

oals based on its preferences. For this assessment, everyuse different variables based on their preferences andelevance of those variables with respect to each of theoals of land and property development. These variablesn Table 2.11

le to calculate the payoff of all outcomes for each player,red all variables above. The variables have been valuedom 0 to 10. Value 10 means that a particular variableant to achieve a particular sub-goal and value 0 meanstheopposite. In this valuation, negativevalueshavealsointo account to show that a particular variable may beto have a negative impact on one of the sub-goals. Thef each variable is multiplied by the weight of each fea-ub-goal, resulting in a weighted score to all indicators


based on thassessed beend score thave used a

We appldevelopmening systemthe implemnew legislanot know yHowever, thtions of thefrom the ex2008). We hwith respecdevelopmeone. Additiorium in accdeviate or nassume per

5.2. The gam

The struregime areFig. 2. Game in extensive formland and property development process for residen

e strategy of M to buy agricultural land from LO aretter than strategies to buy land from other players. Thehat results from this exercise is the payoff for M. Wesimilar line of argument for the other players.ied the game for the case of greeneld residential landnt within the institutional context of the Dutch plan-, more specically under the regime before and afterentation of the new Land Development Act. Since thistion has been implemented only very recently, we doet the stakeholders strategies under the new regime.ere is some empirical evidence supporting our estima-payoffs of the possible outcomes under the new regime-ante evaluation of the new law (Van Dinteren et al.,ave explored the game tree to analyse the differencest to the stakeholders strategic behaviour regarding thent process under the old planning regime and the newnally, we have predicted the subgame perfect equilib-ordance with both games and nd out whether theyot. The game we constructed is a static game and wefect information.

e models

ctures of the game under the old and the new planningpresented respectively in Figs. 2 and 3. The structures

of both gamstakeholderers perceptproperty deduction of texpected todifferenceswhich expr

In the gtiator of thtree, becausinitiative fo(Needham,ipality cana private lamodel eithor a publicdeveloper).constructedare within tbased on thchoose a delic land devland from Lestablish atial use in the Netherlands under old regulation.

es are identical, because the available strategies for alls are the same under both conditions. However, play-ions of the expected utility of the outcomes of land andvelopment processes are different, due to the intro-he new Land Development Act. The new legislation islead the stakeholders to choosedifferent strategies. Thebetween both games are visible in the payoff systemsess the players expected utility on the game outcomes.ames we constructed, the municipality (M) is the ini-e game, i.e. the rst mover at the root of the gamee in the Netherlands themunicipality usually takes ther land and property development in the Netherlands2007, 2006). As we explained in Section 3, the munic-choose between a public land development model andnd development model. The public land developmenter involves the municipality as public land developerprivate joint venture company (together with privateBased on this condition, ve different moves can befor M from the root of the game tree. Four of themhe public land development model and one strategy ise private land development model. When M decides tovelopment strategy for the location based on the pub-elopment model, two moves can be considered: to buyO or to buy land from PL. Other possible moves are tojoint venture company, either with PD or with PL. The


nal potentmodel, if Mbased on threspond anchooses to bchance to rethe land toM decides nwill end at tM, and thenPD. At this pbuilding ploend points oin this gameserviced lanhas decidedclaim modeserviced lan

PL and Pwith otheras a matterbilitywith rventure comual propertyFig. 3. Game in extensive formland and property development process for residen

ial move ofM is based on the private land developmentdecides to allow PL to develop the land. Sequentially,ese ve different moves by M, the other players willd decide to take their moves. For instance, suppose Muy land from LO in the rstmove, then LOwill have thespond. At its decision node, LO can choose either to sellM or not. Suppose LO decides not to sell its land (andot to make use of its expropriation powers), the gamehis point. And suppose that LO chooses to sell its land toM will service the land and sell the building plot(s) tooint, PD will have to respond: it may decide to buy thets or refuse to buy. Both decisions will lead to differentr terminal nodes as the outcome of the game. Note that, PD and PL do not have the choice to refuse buying thed, either fromM or the joint venture company, after Mto grant them a building claim, because the buildingl also includes the obligation for PD or PL to buy thed.D might decide to establish a joint venture companyproperty developers. However, this is considered onlyof strengthening or increasing their capacity and capa-espect to land development. The strategy of such a jointpanywill then be similar to the strategy of the individ-developer PL or PD. That iswhywe put this alternative

aside fromwe put asidLO or PL), besame behavof our modacquired its

5.3. Finding

In our mdifferent ououtcomes oproject resuthe economthis is that tmarket consame timewmarket conitable land dless protab

12 Land priceprice minus thtial use in the Netherlands under new regulation.

PDs and PLs choices of actions. The other alternativee from PDs strategy is to acquire land (most likely fromcause if PDwould like to do this, it will then follow theiour as PL. At the same time, it shows the boundariesel as we did not take into account how and when PLland.

s

odels, both under old and new regulation, we found 23tcomes at which the game could end. In specifying thef the game, we take into account that the nal nanciallts as well as the strategies of the players depend onic situation of the housing market. The explanation forhe increase of the demand for housingunder favourableditionswill result in higher housing prices, which at theill increase land prices aswell.12 Therefore, favourable

ditions on the housing market will result in more prof-evelopment,while badmarket conditionswill result inle land development. In the game we constructed, the

s for housing are generally calculated from the residual of the housinge construction costs.


housingmarket conditions are unknown or uncertain. For that rea-son, we incorporated nature (or: the economy) as a chance playerin the model and assigned a symmetric probability for good andbad market conditions to it.13 The payoff under favourable marketconditions is obviously higher than under poor market conditions.However, this probability leads touncertaintywhenweanalyse thebest strategy for the players, because themarket conditions cannotbe inuenced by the player. In order to reduce the uncertainty, wecalculated the expected value for the particular outcomes whichare under the probability condition.

The payoff structure for both games shows the preferences ofthe playersthat underforM, becauFig. 3 showsunder thenM, both und7.6 in both

The outcemerges frolish a jointthen buys lwill servicetion in thebe better coPD has no bto implemeof servicedat prot mmean thatthis will giresults of othat it willwith PD wcompany.

Under thpayoff for Mnew regimefor the stratafterwardsis that undePL to contrM can refusfor all costsinstance infareas outsiddecide to exand PL knoM, even uncan still beThough thihigh, it is unvery often,joint ventuunserviced

13 We assummany variablewe assign theemerging of baprobability. Ththat this typeincluded in ga

Furthermore, the lowest payoff to M emerges from a strategythat started in the same way as outcome 17. However, despite ofacquiring land from LO, the joint venture company that is estab-lished with PD acquires land from PL. Outcome 22 turns up whenPL refuses to sell its land to the joint venture company, but insteadto PD. If this situation would occur, Ms plan for land developmentwill not be implemented.Moreover, bothMand PDwill have to payfor the costs of establishing the joint venture company. Therefore,this outcomPD. Again, ibe a contra

enturontragimehestd themparfor Poment veergese jor parPL, thnt ouoff fs toboth.4). On toion thgularomto glly f, oututcoandle 3or eapaynt prbothict sigamebac

eachs strincenartedicesamesameoffsoneiouss besitions. It p

that,egimetheire.with respect to every outcome of the game. Fig. 2 showsthe old regime, outcome 17 is the most preferable oneseMwill have the highest payoff from it (payoff: 23.4)., however, that outcome10 is themost preferable forMewregime (payoff: 16). The leastpreferableoutcome forer the old and the new regime, is outcome 22 (payoff:situations).

omewith the highest payoff forMunder the old regimem a strategy that starts from Ms decision to estab-venture company with PD. The joint venture companyand from LO. Subsequently, the joint venture companythe land and sells the serviced land to PD. Ms posi-negotiations in a joint venture company with PD willmpared to a joint venture company with PL, becauseuilding claim. With this strategy, M will be able bothnt its plan and to maximize its prots from the saleland. Although we argued before that M does not aimaximization in providing building plots, it does notthey do not benet from the nancial result, becauseve M an opportunity to cover the negative nancialther projects. The other advantage of this strategy isgive M the opportunity to share development risks

ith respect to land development in the joint venture

e new regime, outcome 17 does not give the highestanymore. Instead, the highest payoff for M under theis given by outcome 10 that emerges when M decidesegy to establish a joint venture companywith PL, whilethe serviced land is sold to PD. The explanation for thisr the new regime Ms position in the negotiations withibute to the costs of land development has improved.e a joint venture with PL, but instead force PL to paythat are necessary to service the land (including forrastructure and the development costs of recreationale the development scheme). If PL refuses this, M maypropriate PL (under certain conditions). Since both M

w this can happen, PL might decide to cooperate withder less favourable conditions. In this strategy, therea nancial prot from the sale of serviced land to PD.s explains why (in this situation) the payoff for M islikely (but not impossible) that this strategy will occurbecause PL would usually refuse to participate in there company with M. Instead PL may decide to sell theland to any interested party.

e that the market conditions are uncertain because they depend ons; some of them could be exogenous. Because of its uncertainty,probability of 0.5 for the emerging of good market and 0.5 for thed market. The term of A symmetric probability refers to this equale issue of market conditions in ourmodelmainly serves to illustrateof exogenous factors (that inuence the players strategies) can beme theoretical modeling.

joint vIn c

new rethe higold anless copayoffto outcof a joi16 emfrom thanothe

Fordiffereest paychangefor PL,and 5decisioconditnew restarts fPL (and

Finaregime11.7). Othe old

Tabcome f

Thedifferegame,a conof theusing awhichplayerhasanthat stM servboth gof the gthe paythe old

Obvplayerno posholder

14 Notethe old rachieveold regime gives not only the lowest payoff for M, but also fort is unlikely that it occurs, because there will probablyctual agreement between M and PD, as partners in thee company, not to acquire land on their own.st, the highest payoff for PD both under the old and theis given by outcome 16. Although the outcome withand the lowest payoff do not change for PD under thenew regime, the payoffs under the new regime areed with the old regime. Under the old regime, the bestD is 15.4, while under the new regime it is 13.8. Similar17, outcome 16 also emerges from the establishmentnture company between M and PD. However, outcomefrom a situation in which PD buys the serviced land

int venture company, while in outcome 17 it is sold toty.e highest payoffs under the old and new regime are intcomes. Under the old regime, outcome 4 is the high-or PL (18.9); under the new regime the highest payoffoutcome 9 (13.2). Outcome 14means the lowest payoffunder the old and the new regime (respectively 6.4utcome 4 emerges from a strategy that starts fromMsbuy unserviced land fromPL and PL sells the landwith aat M grants a building claim to PL. However, under thetion, it is better for PL to respond to Ms strategy thatMs decision to establish a joint venture company withrant PL a building claim).or LO, it is found that, both under the old and the newcome 16 has the highest payoff (respectively 12.9 andme 18 gives the lowest payoff for LO (2.7 both underthe new regime).gives a summary of the maximum and minimum out-ch player.14

off structures of all players show that every player haseferences with respect to the possible outcome of theunder the old and the new regime. This could lead totuation. For the solution, we analysed the steady stateby looking for the subgame perfect equilibrium (SPE)kward induction method. The result is outcome 1 inplayers strategy is an optimal response to the other

ategies. Recall that any SPE is an NE, so that no playertive todeviate. Thisoutcomeemerges fromthestrategyfrom Ms decision to buy land from LO. Subsequently,

the land and sells it to PD. This equilibrium is found for, which means that there is no change in equilibriumafter the introduction of the new regulation. However,in the SPE under the new regime are lower than under.ly, the payoff structure at SPE is lower than everyt payoff. Particularly, PLs payoff is low, but PL holdsto negotiate for a better payoff with the other stake-roves that in this game we can nd more than one NE.

in general, the payoffs under the new regime are lower than under. It means that for the players, the importance of such strategies todevelopment goals under the new regulation is less than under the


Table 3Maximum and minimum payoffs for stakeholders.

Stakeholder New regime

Ma

Ou

M 10PD 16PL 9LO 16

Table 4Payoffs at SPE.

Player

MPDPLLO

However, tnot subgamnal solutiostakeholder

The gamconclusionequilibriumthe negotiahas improvwise for thestrategy asof the ex-anDinteren etnew legislanew legislamunicipalitriding privathe municip

6. Conclus

By makistrategic beTo investigdecision-mprocesses,greeneld rthe implemstudy has dstrategic deshows the dtheir choseations in waware thatlimitations.still be solvdecision su

First, asare not basedecision supvalidate theuse of statedgame can alor experts i

ond,te inptionthefortute iner pr straes olete ilayeplaymetarernatepeayersingly: stagamayindateyofftheeatest thergtionao reatherith

ies ofectlecenore andeOld regime

Maximum Minimum

Outcome number Payoff Outcome number Payoff

17 23.4 22 7.616 15.4 22 9.84 18.9 14 6.4

16 12.9 18 2.7

Payoff

Old regime New regime

20.7 15.213.8 11.80 0

10.8 10.8

hese NEs are not credible or, in other words, they aree perfect. Therefore they cannot be considered as then for the game. Table 4 summarises the payoffs for eachat SPE both under the old and the new regime.e theoretical modelling in this case study leads to thethat the new regulation will not result in an alternativeoutcome. Though the position of the municipality in

tion process, in contrast to the other players positions,ed under the new regulation, it would nevertheless bemunicipality to continue with the same developmentbefore. This is in line with the outcome of the resultste evaluation of the impact of the new regulation (Vanal., 2008). It is also in line with the objectives of thetion. As we explained in Section 4, the intention of thetion is not to change development practices, but to offeries a safety net in the rather exceptional cases that freete developers/landowners do not want to contribute toalitys land development costs.

ion

ng use of game theory, we have analysed stakeholdershaviour in land and property development processes.ate the usefulness of game theory for modellingaking processes in land and property developmentwe have built a game theoretical model of a typical

Seccompleassumknowsers. Uncomplethe othtion fooutcomincompmore ption (aor asymothers)An alteis to rthe plaaccordmationof theedly pland upand paEven iners, repto adjuFudenbthat ragame t

Anonality wunderlact perever, rpay ming bouesidential development in the Netherlands concerningentation of newDutch legislation on cost recovery. Ouremonstrated that game theory helps to identify the keycisions to bemade in this type of development projects,ifferent payoffs, in relative terms, for stakeholders of

n strategies and enables to select the equilibrium situ-hich all stakeholders are best of. However, we are alsothe case study, in its present form, still contains manyIn this nal section we discuss the problems that musted to increase the attractiveness of game theory forpport with respect to land and property development.long as the strategies of the stakeholders in the modeld on empirical data, the usefulness of the outcomes forport is only limited. However, it is certainly possible topreferences of the stakeholders, for instance bymakingpreference techniques.Moreover, the outcomes of the

so be tested by playing the gamewith real stakeholdersn a laboratory situation.

It can be usmodel.

Finally,complex dedevelopmeof reality ininvolved aninformal relin reality. Odependencymunicipalitsituation thably lead tocomplexityother modepresumed wexercises artools. In genximum Minimum

tcome number Payoff Outcome number Payoff

16 22 7.613.8 22 5.413.2 14 5.411.7 18 2.7

initially game theory assumes that the players possessformation about the strategies of the other players. Thisunderlies ourmodel aswell. It implies that each playerstrategies and the payoff functions of the other play-nately, in practice, stakeholders do usually not holdformation about the strategies and payoff functions oflayers. Moreover, stakeholders may withhold informa-tegic use. This reduces of course the usefulness of thef the model as presented here. However, games withnformation (the strategies or payoff functions of one orrs are unknown or partially known), imperfect informa-er does not know for sure where he stands in the game)ric information (some players are better informed thanmore and more studied, especially in microeconomics.ive approach which applies to games in strategy formt a game several times through time. This will givethe opportunity to learn and to adapt their strategies. This is alsomeaningful in the case of incomplete infor-rting with incomplete information in the beginninge, the players will collect information during repeat-g the game. They then will be able to learn rationallytheir conjectures about the other players strategiesfunctions by observing the other players behaviour.case of conicting optimal strategies of the stakehold-d games would give the stakeholders the opportunityeir strategies to the strategies of other stakeholders.and Levine (1993) and Kalai and Lehrer (1993) revealedl learning in repeated games eventually can lead thech the equilibrium.limitation regarding game theory is its notion of ratio-respect to the behaviour of the players. This notion alsour model. Of course, the assumption that individualsy rational may never match a real-life situation. How-t developments and experimentation in game theoryttention to behavioural aspects of the player, includ-d rationality, emotions and intuitive decision-making.ed in further research to increase the reliability of the

the application of game theoretical modelling tocision-making processes like in land and property

nt processes, involves, by denition, the simplicationthe model. There are much more coincidental eventsd much more linkages between types of actors (e.g.,ationsbetween stakeholders) andmixed typesof actorsne example of this simplication problem is the pathof the tree. In our model, we have assumed that they will start the tree, but it is also possible in a real-lifeat other stakeholderswill start the treewhichwill prob-different outcomes. Although we believe that morecan be built in, game theoretical models like anyl are always an abstraction of reality. However, theseeaknesses can also be seen as strength, as modellinge often used in other research elds as decision supporteral, models coerce the accuracy of argument. Models


force the modellers to be explicit in expressing the assumptions,and in arriving at conclusions bydeduction. Theyhave to showhowaparticular conclusionderives fromcertain assumptions. Thegamemodels in this paper are able to capture the logic of the process ofland and property development.

Furthermore, models help us to discipline and formalize intu-ition. Intuition, undoubtedly, is central to any understanding,including in modelling. However, intuition alone is not reliable.Although the results of many models agree with our intuition, notall intuitions can be supported by models. Game theoretical mod-elling can be one of the tools for exploring the strategic logic ofsituations. It forces us to be specic about the characteristics ofconicting situations.

The specic advantage of game theory in formal modelling isits focus on strategic interaction and conict in collective decision-making processes. It naturally leads us to consider the individualstrategic decisions and their interdependency in relation to collec-tive outcomes. In addition, game theory provides a way to think

about the complexity of strategic interaction and, in particular,about the conicting structure of collective decision-making pro-cesses. When the game is constructed, the choices of the playersand their coThat speciferent playA game treand structuferent setsand hence tvariation insocial constrongly beelling technand properthe complecan be valid

Appendix A

Old regulation Aspects of preference Relevance to achieve goals with respect to

Spatial quality Financial a

Score Weightedscore (50%)

Score

Municipality (M) Plans implemented 10 5 4Plans notimplemented

0 0 0

No building claim 4 2 7Building claim to PL 0 0 7Building claim to JVmember

0 0 0

Active policy 3 1.5 6Passive policy w/prop.devt

2 1 3

Passive policy withoutprop. devt

1 0.5 3

Good market 2 1 8Bad market in activepolicy

0 0 5

Bad market in passivepolicy

0 0 0

Serviced land unsold inactive policy

0 0 10

Serviced land unsold inpassive policy

1 0.5 0

Join service with PL 2 1 6Join service with PD 2 1 6Alone 1 0.5 3Land from LO 2 1 5Land from PL 2 1 3

Old regulatio with r

Fina

Scor

Property dev 100

100

108

1008

2

8n Aspects of preference Relevance to achieve goals

Spatial quality


eloper (PD) Develop property 3 0.9No property 0 0Have building claim 2 0.6No building claim 0 0Good market (with property) 2 0.6Good market (no property) 2 0.6Bad market (with property) 2 0.6Bad market (no property) 0 0Selling serviced land to PLwithout B.C.

0 0

Selling serviced land to PLwith B.C.

0 0

Serviced land unsold 0 0nsequences are specied as demonstrated in this study.cation is a representation of a conicting structure. Dif-ers want different things, opt for different outcomes.e is an expression of rules about how to play a gameres in this way the conict among the players. Dif-of rules will, ceteris paribus, lead to different gameso different structuring of social conict. This structuralgames may help us understand the consequences of

ict in collective decision-making processes. In sum, welieve that game theory, compared to alternative mod-iques, offers a promising approach for modelling landty development processes, because it takes account ofxity of the process and the assumptions in the modelated empirically.

.

Total score(weighted)

spects Time process

Weightedscore (30%)


1.2 4 0.8 70 0 0 0

2.1 4 0.8 4.92.1 4 0.8 2.90 1 0.2 0.2

1.8 6 1.2 4.50.9 0 0 1.9

0.9 1 0.2 1.2

2.4 4 0.8 4.21.5 0 0 1.5

0 0 0 0

3 0 0 3

0 0 0 0.5

1.8 7 1.4 4.21.8 5 1 3.8

0.9 0 0 0.41.5 4 0.8 3.3

0.9 1 0.2 0.1

espect to Total score(weighted)

ncial aspects Time process

e Weightedscore (50%)


5 4 0.8 6.70 0 0 05 4 0.8 6.40 0 0 05 6 1.2 6.84 4 0.8 5.4

5 6 1.2 6.80 0 0 04 0 0 4

1 0 0 1

4 0 0 4


Appendix A (Continued )

Old regulation Aspects of preference Relevance to achieve goals with respect to Total score(weighted)

Spatial quality Financial aspects Time process




Buy serviced land from M 1 0.3 6 3 2 0.4 3.7Buy serviced land from JV 1 0.3 3 1.5 2 0.4 2.2Buy serviced land from PL 0 0 3 1.5 2 0.4 1.1Buy raw land from LO 0 0 6 3 2 0.4 3.4Buy raw land from PL 0 0 0 0 0 0 0

Property developer withland (PL)

Develop property 3 0.9 10 5 4 0.8 6.7No property 0 0 0 0 0 0 0Have building claim 2 0.6 10 5 4 0.8 6.4No building claim 0 0 0 0 0 0 0Good market (with property) 2 0.6 10 5 6 1.2 6.8Good market (no property) 2 0.6 8 4 4 0.8 5.4Bad market (with property) 2 0.6 10 5 6 1.2 6.8Bad market (no property) 0 0 0 0 0 0 0Very bad market (servicedland unsold)

0 0 1 0.5 0 0 0.5

Selling unserviced land to PD 0 0 6 3 2 0.4 3.4Selling unserviced land to JV 0 0 5 2.5 2 0.4 2.9Selling unserviced land to JVw/b.c.

0 0 5 2.5 1 0.2 2.7

Selling unserviced land to M 0 0 3 1.5 2 0.4 1.9Selling serviced land 0 0 8 4 2 0.4 4.4Serviced land unsold 0 0 8 4 2 0.4 3.6No risk in servicing land 0 0 5 2.5 2 0.4 2.9Risk sharing in servicing land 0 0 0 0 1 0.2 0.2Risk taker in servicing land 0 0 5 2.5 1 0.2 2.3Buying serviced land from M 0 0 5 2.5 2 0.4 2.9Buying serviced land from JV 0 0 2 1 2 0.4 1.4Unsuccesful negotiation w/M 0 0 2 1 1 0.2 1.2Unsuccesful negotiation w/JV 0 0 4 2 1 0.2 2.2

Old regulation Aspects of preference Relevance to achieve goals with respect to Total score(weighted)





Landowner (LO) Land sold 0 0 10 6 0 0 6Good market 0 0 8 4.8 0 0 4.8Selling land to JVC 0 0 3 1.8 2 0.6 2.4Negotiation w/M 0 0 3 1.8 3 0.9 2.7Selling land to PD w/B.C. 0 0 3 1.8 0 0 1.8

New regulation Aspects of preference Relevance to achieve goals with respect to Total score(weighted)





Municipality (M) Plans implemented 10 5 4 1.2 4 0.8 7Plans not implemented 0 0 0 0 0 0 0No building claim 2 1 3 0.9 2 0.4 2.3Building claim to PL 0 0 2 0.6 1 0.2 0.8Building claim to JVmember

0 0 0 0 0 0 0

Active policy 2 1 2 0.6 2 0.4 2Passive policy w/prop.devt

1 0.5 2 0.6 0 0 1.1

Passive policy withoutprop. devt

0 0 2 0.6 1 0.2 0.4

Good market 2 1 8 2.4 4 0.8 4.2Bad market in activepolicy

0 0 3 0.9 0 0 0.9

Bad market in passivepolicy

0 0 0 0 0 0 0

Serviced land unsold inactive policy

0 0 10 3 0 0 3

Serviced land unsold inpassive policy

1 0.5 0 0 0 0 0.5

Join service with PL 2 1 3 0.9 3 0.6 2.5Join service with PD 2 1 3 0.9 1 0.2 2.1








Alone 1 0.5 2 0.6 0 0 0.1Land from LO 2 1 3 0.9 2 0.4 2.3Land from PL 2 1 1 0.3 1 0.2 0.5






Property developer (PD) Develop property 3 0.9 10 5 4 0.8 6.7No property 0 0 0 0 0 0 0Have building claim 2 0.6 4 2 2 0.4 3No building claim 0 0 0 0 0 0 0Good market (withproperty)

2 0.6 10 5 6 1.2 6.8

Good market (noproperty)

2 0.6 8 4 4 0.8 5.4

Bad market (withproperty)

2 0.6 10 5 6 1.2 6.8

Bad market (no property) 0 0 0 0 0 0 0Selling serviced land toPL without B.C.

0 0 7 3.5 0 0 3.5

Selling serviced land toPL with B.C.

0 0 1 0.5 0 0 0.5

Serviced land unsold 0 0 7 3.5 0 0 3.5Serviced land from M 1 0.3 4 2 2 0.4 2.7Serviced land from JV 1 0.3 2 1 2 0.4 1.7Serviced land from PL 0 0 1 0.5 0 0 0.5Buy raw land from LO 0 0 4 2 2 0.4 2.4Buy raw land from PL 0 0 0 0 0 0 0

Property developer with land (PL) Develop property 3 0.9 10 5 4 0.8 6.7No property 0 0 0 0 0 0 0Have building claim 2 0.6 4 2 4 0.8 3.4No building claim 0 0 0 0 0 0 0Good market (withproperty)

2 0.6 10 5 6 1.2 6.8

Good market (noproperty)

2 0.6 8 4 4 0.8 5.4

Bad market (withproperty)

2 0.6 10 5 6 1.2 6.8

Bad market (no property) 0 0 0 0 0 0 0Very bad market(serviced land unsold)

0 0 1 0.5 0 0 0.5

Selling unserviced landto PD

0 0 4 2 2 0.4 2.4

Selling unserviced landto JV

0 0 3 1.5 2 0.4 1.9

Selling unserviced landto JV w/b.c.

0 0 3 1.5 0 0 1.5

Selling unserviced landto M

0 0 1 0.5 2 0.4 0.9

Selling serviced land 0 0 6 3 2 0.4 3.4Serviced land unsold 0 0 6 3 2 0.4 2.6No risk in servicing land 0 0 2.5 1.25 1 0.2 1.45Risk sharing in servicingland

0 0 2.5 1.25 2 0.4 1.65

Risk taker in servicingland

0 0 5 2.5 1 0.2 2.3

Buying serviced landfrom M

0 0 2 1 2 0.4 1.4

Buying serviced landfrom JV

0 0 1 0.5 2 0.4 0.9

Unsuccessful negotiationw/M

0 1 0.5 1 0.2 0.7

Unsuccessful negotiationw/JV

0 0 2 1 1 0.2 1.2



New regulation Aspects of preference Relevance to achieve goals with respect to Total score (weighted)


Score Weighted score (10%) Score Weighted score (60%) Score Weighted score (30%)

Landowner (LO) Land sold 0 0 10 6 0 0 6Good market 0 0 8 4.8 0 0 4.8Selling land to JVC 0 0 3 1.8 2 0.6 2.4

References

Adams, D., MaStudies 28

Adams, D., Bauprice of va

Adams, D., Badevelopm

Adams, D., WaBlackwell,

Alexander, C.,Ansoff, H.I., 19Aumann, R.J.,

HonkapohAumann, R.J., 1

WestviewAumann, R.J., HBall, M., 1998.

(9), 1501Ball, M., Lizier

Markets. RBerkman, H.G.

41 (1), 11Boelhouwer, 2

explodingand the Bu

Borel, E., 1921de ses app

Buitelaar, E., 2nomics to

Bulmer, S., Dopean Unio

Byrne, D., 200Planning T

Camerer, C.F.,Princeton

Camerer, C.F.,nomics. Pr

Carmichael, F.,Colman, A., 19

Sciences. RDe Roo, G., 200

the searchlio, P.J., Grand EuropEnvironme

Devisch, O., 20SimCity an

Devisch,O., Archoice dynputers inU29 June1

Dixit, A., SkeatDolowitz, D.P.

of the BritDolowitz, D.P.

contempoDolowitz, D.P.

ties to forstudy of thmany. Envpublicatio

1982.482.rg,D.,ica 61ld, R.,on.., 197ce 16., 198

ves Heon..B., Vn devP., 19erty R,2000hesterLehre0191., DArn StuAltesh Pla), 97., 199rd.., Rair Puba, J., 2l Jour, R.B.,bridge, B., 2ledge, B., 2ether, B., Vporta.E., 197130., 1965E., 19usettsE., 200omicIverset. Inte., 199men,oud Urabbee DutKrabbsing tteren,atie W, R., 2ate, ANegotiation w/M 0 0 3 1.8Selling land to PD w/B.C. 0 0 1 0.6

y, H.G., 1991. Active and passive behaviour in landownership. Urban, 687705.m, A., MacGregor, B., 1985. The inuence of valuation practice on thecant inner city land. Land Development Studies 2, 157173.um, A., MacGregor, B., 1988. The availability of land for inner cityent: a case study of inner manchester. Urban Studies 25, 6276.tkins, C., White, M., 2005. Planning, Public Policy & Property Markets.Oxford.1965. A city is not a tree. Architectural Forum 122 (1), 5861.79. Strategic Management. Palgrave Macmillan, New York.1985. What is game theory trying to accomplish? In: Arrow, K.,ja, S. (Eds.), Frontiers of Economics. Basil Blackwell, Oxford.989. Lectures on Game Theory. Underground Classics in Economics.Press, Boulder.art, S. (Eds.), 1992. Handbook of Game Theory. Elsevier, Amsterdam.Institutions in British property research: a review. Urban Studies 351517.i, C., MacGregor, B., 1998. The Economics of Commercial Propertyoutledge, London., 1965. The game theory of land use determination. Land Economics19.005. The incomplete privatization of the Dutch housing market:house prices versus falling house-building output. Journal of Housingilt Environment 20 (4), 363378.. Applications aux jeux de hazard, Traite du calcul des probabilities etlications. Gauthier-Villars, Paris.007. The Cost of Land Use Decisions. Applying Transaction Cost Eco-Planning & Development. Blackwell, Oxford (UK)/Cambridge (USA).lowitz, D.P., Humphries, P., Padgett, S., 2007. Policy Transfer in Euro-n Governance. Routledge, London.3. Complexity theory and planning theory: a necessary encounter.heory 2 (3), 171178.2003. Behavioral Game Theory: Experiments in Strategic Interaction.University Press, Princeton.Loewenstein, G., Rabin, M. (Eds.), 2004. Advances in Behavioral Eco-inceton University Press, Princeton.2004.AGuide toGameTheory. Financial Times/PrenticeHall,Harlow.99. Game Theory and its Applications: In the Social and Biologicaloutledge, London.4. Copingwith the growing complexity of our physical environment:for new planning tools in the Netherlands. In: Sorensen, S., Marcotul-ant, J. (Eds.), Towards Sustainable Cities; East Asian, North Americanean Perspectives on Managing Urban Regions, Urban Planning andnt. Ashgate, Aldershot, pp. 161175.08. Should planners start playing computer games? Arguments fromd Second Life. Planning Theory & Practice 9 (2), 209226.entze, T., Timmermans,H. (2005). Anagent-basedmodel of residentialamics in non-stationary housing markets, Paper presented at Com-rbanPlanningandUrbanManagement (CUPUM),ConferenceLondon,July.h, S., 2004. Games of Strategy. W.W. Norton & Co, New York., 1998. Learning from America: Policy Transfer and the Developmentish Workfare State. Sussex Academic Press, Brighton., Marsh, D., 2000. Learning from abroad: the role of policy transfer inrary policy-making. Governance 13, 524.

Elster, J.,453

Fudenbemetr

GoodchiLond

Hardin, RScien

Hardin, RHargrea

LondHavel, M

urbaHealey,

PropJones,A.

ChicKalai, E.,

61, 1Keogh, G

UrbaKorthals

Dutc33 (1

Kreps, DOxfo

Luce, R.DDove

Mu, L., Mtiona

MyersonCam

NeedhamRout

Needhamthe N

Needhamis im

Oliver, P19, 2

Olson,MOstrom,

sachOstrom,

EconSalskov-

menTaylor, Nvan Dee

RadbVanderK

of thVan der

chooVan Din

ploitVerhage

Ashg, Medearis, D., 2009. Considerations of the obstacles and opportuni-malizing cross-national policy transfer to the United States: a casee transfer of urban environmental and planning policies from Ger-ironment and Planning C: Government and Policy, advance onlinen.

VonNeumann100, 2953

Von NeumannPrinceton3 0.9 2.70 0 0.6

Marxism, functionalism and game theory. Theory and Society 11 (4),

Levine,D.K., 1993. Steady state learningandNashequilibrium.Econo-, 547573.Munton, R., 1985. Development and the Landowner. Allen andUnwin,

1. Collective action as an agreeable n-prisoners dilemma. Behavioral(5), 472.2. Collective Action. Resources for the Future, Inc., Washington, DC.ap, S.P., Varoufakis, Y., 2004. Game Theory: A Critical Text. Routledge,

iitanen, K., 2007. Urban game as a computer assisted simulation ofelopment. Maanmittaustieteiden paivat 44, 121127.92. An institutional model of the development process. Journal ofesearch 9, 3344..GameTheory:MathematicalModelsofConict.HorwoodPublishing,.r, E., 1993. Rational learning leads to Nash equilibrium. Econometrica045.cy, E., 1999. Property market efciency: an institutional perspective.dies 36, 24012414., W., 2006. Stagnation in housing production: another success in thenners Paradise? Environment and Planning B: Planning and Design114.0. Game Theory and Economic Modeling. Oxford University Press,

ffa, H., 1957. Games and Decisions: Introduction and Critical Survey.lications, Inc., New York.007. Game theory of price decision in real estate industry. Interna-nal of Nonlinear Science 3 (2), 155160.1991. Game Theory: Analysis of Conict. Harvard University Press,, MA.006. Planning, law and economics: the rules wemake for using land., London.007. Dutch Land Use Planning: Planning and Managing Land use inlands, the Principles and the Practice. Sdu Uitgevers b.v, The Hague.erhage, R., 1998. The effects of land policy: quantity aswell as qualitynt. Urban Studies 35 (1), 2544.93. Formal models of collective action. Annual Review of Sociology0.. The Logic of Collective Action. HarvardUniversity Press, Cambridge.90. Governing the Commons. Cambridge University Press, Mas-.0. Collective action and the evolution of social norms. The Journal ofPerspectives 14 (3), 137158.n, D., 2006. Learning across borders: the case of Danish local govern-rnational Journal of Public Sector Management 19, 673686.8. Urban Planning Theory since 1945. Sage Publications Ltd., London.A., 2006. Speltheorie en strategische besluitvorming in organisatie.niversity Nijmegen, Inaugurale Rede, Nijmegen.n, E., Lambooy, J.G., 1993.A theoretical framework for the functioning

ch property market. Urban Studies 30, 13811397.en, E., Needham, B., 2008. Land readjustment for value capturing:he right policy instruments. Town Planning Review 79 (6), 486505.J., De Kruif, J., Van der Krabben, E., 2008. Ex-ante evaluatie Grondex-et. Ministry of Spatial Planning, The Hague.

002. Local Policy for Housing Development: European Experiences.ldershot., J., 1928. Zur Theorie derGesellschaftsspiele.MathematischeAnnalen20., J., Morgenstern, O., 1944. Theory of Games and Economic Behaviour.University Press, Princeton, NJ.

A game theory approach to the analysis of land and property development processesIntroductionWhy game theory?How does game theory work?GamesPlayersStrategyOutcome and payoffSolution concepts

Case study: residential land development in the NetherlandsMunicipal strategies for greenfield residential land development in the NetherlandsThe New Land Development Act and its impact on land and property development

Game constructionFramework for the gamePlayersStrategiesPayoffs

The game modelsFindings

ConclusionAppendix AReferences

a game theory approach to the analysis of land and property development processes

Documents

stakeholders decisions

development of land

stakeholders strategieshealey

case study

land andevelopment

decisions result

game adevelod

game theoretical modelling