A Transaction-cost Analysis of the Land Development Process

Transcription

1 Urban Studies, Vol. 41, No. 13, , December 2004 A Transaction-cost Analysis of the Land Development Process Edwin Buitelaar [Paper first received, November 2003; in final form, June 2004] Summary. The production of our built environment involves many transaction costs. We want to avoid unnecessary transaction costs. Nevertheless, we hardly take account of transaction costs when considering different ways of organising the development process. The most important reason is probably the difficulty in identifying them. This paper shows why transaction costs are important and offers a method for identifying them. With this method, it is possible to compare the efficiency of different institutional arrangements. Introduction Transaction cost economics is not new. The nature of the firm, which was written by the founding father of new institutional or transaction-cost economics Ronald Coase, appeared in The entrance of this branch of economics into planning theory, however, is of a much more recent date. 1 Many economic analyses of planning were (and still are) based on traditional welfare economics (see, for example, Harrison, 1977; Heikkila, 2000). Webster and Lai (2003, p. 219) give two reasons for the growing attention given to Coasian economics in the past decade. The first is the Nobel Prize Coase received in the early 1990s. The second is that the old paradigms have lost their power in a period when the role of public land-use planning is under pressure. Alexander introduced the consideration of transaction costs into planning theory with his article A transaction-cost theory of planning in There, planning is treated from a transaction-cost perspective as being a process of co-ordination such as is analysed in general terms in institutional economics. This line of thought can be extended to the development process. 2 The development process can be regarded as a sort of production process: at a particular location, the built environment is changed i.e. produced (see Figure 1). 3 The production process has often been approached from a neo-institutional economic perspective (see, for example, Williamson, 1975, 1985). Since Alexander s article, transaction cost economics has been applied several times to market versus government issues in planning theory (Lai, 1994, 1997; Alexander, 2001a, 2001b; Buitelaar, 2003; Webster and Lai, 2003). Although these contributions provide Edwin Buitelaar is in the Nijmegen School of Management, University of Nijmegen, Thomas van Aquinostraat 3, PO Box 9108, 6500 HK Nijmegen, The Netherlands. Fax: e.buitelaar@fw.ru. The author gratefully acknowledges helpful suggestions on drafts of this paper, especially from Barrie Needham (University of Nijmegen), but also from Ernest Alexander (APD Alexander Planning and Design), Daniel Bromley (University of Wisconsin Madison), Roelof Verhage (Institut d Urbanisme de Lyon), Margo van den Brink, George de Kam and Arnoud Lagendijk (all of the University of Nijmegen) Print/ X On-line/04/ The Editors of Urban Studies DOI: /

2 2540 EDWIN BUITELAAR Figure 1. A simple model of the production process. Source: Christopher (1998, p. 13). us with useful insights, they lack (sufficient) operationalisation for empirical application. In this paper, I provide a transaction costs framework for comparing institutional arrangements in Western capitalist countries. These institutional arrangements can be regarded as devices for co-ordinating the development process, ranging from market to network to hierarchical forms. I operationalise the framework by focusing on user rights regimes, a type of institutional arrangement that is particularly important for land use planning. In addition, I provide a method for identifying transaction costs, which can be applied empirically. Before the operationalisation, I show which theoretical purposes a transaction-cost analysis might serve. In order to do that, it will be made clear what transaction costs are and why they are important. After that, I show that transaction costs can shed a different light on the discussion about efficiency by introducing the concept of process efficiency, which differs from the often used allocative or output efficiency. This is followed by the operationalisation of the framework, an empirical application to the development process of a housing site in the Netherlands which is used as a starting-point, after which reference is made to transaction costs in other institutional frameworks of planning in other countries. The paper ends with some concluding remarks. Transaction Costs Conceptually Transaction costs refer to all costs other than the costs of physical production (Lai, 1994, p. 84). Transaction costs emerge because we do not have perfect rationality and complete information as is assumed in many neo-classical models. Instead of that, there is bounded rationality, incomplete information and opportunism. Transaction costs are the costs that are incurred to increase the information available to us and to reduce uncertainty. For instance, before we buy a car we try to get information about prices, quality and service. Therefore, we compare different brands and different dealers. Those activities cost time, money and effort, in other words: transaction costs. Although the concept transaction costs suggests that only the costs around a transaction are involved, it is often used more broadly. If the costs of transacting between supplier and demander are too high, one of those two might decide to incorporate the activity of the other by acquiring it from the other. For example, a car factory might decide to incorporate the activity of distributing cars, because the transaction costs of outsourcing are too high. This shift towards a hierarchical form of organisation is called vertical integration. The transaction costs might now be lower because there is no transaction. But instead of that, the organisation has the internal costs of co-ordinating the activity that has been incorporated. These costs are agency costs, but are often also called transaction costs. 4 Some (for example, Hazeu, 2000, p. 74) argue that, because of the confusing nature of the term transaction costs, co-ordination costs is a better term. Nevertheless, because everyone is familiar with

3 TRANSACTION COSTS AND DEVELOPMENT 2541 Figure 2. Specifying the costs of the production process. transaction costs, and to prevent confusion, I continue using this term. Transaction costs can be further specified. First of all, there are (as already said) the costs of acquiring information, like a comparison of brands. In the case of land use planning, it can be, for example, research into residential preferences. These transaction costs are called information costs. Uncertainty can also be (and often is) reduced by using institutions. Landed property, for instance, means that there are rights and consequently (see Bromley, 1991) duties with respect to a particular piece of land. These reduce uncertainty about what can and cannot be done with that land, and by whom. North (1990a, p. 3) defines institutions as: the rules of the game in a society or, more formally, the humanly devised constraints that shape human interaction. Institutions are created or changed, after which they are used. 5 The creation and use (see, for this distinction, Furubotn and Richter, 1991, p. 8) of institutions involves costs: transaction costs or, more specifically, institutional costs. In Figure 2, I have summarised the different costs of the production process in order to specify the concept of transaction costs. As will become clear, this paper focuses on the costs of creating and using institutions. It should be noted that costs are incurred to get more information about the institutions. Then, they are regarded as institutional costs, because they can be seen as part of the costs of using the institution. The costs of acquiring information independent of the institutional arrangements are what I call the information costs. The identification and measurement of transaction costs are not routine tasks. Hardly anyone who has tried this, has achieved completely satisfying results (Furubotn and Richter, 1991). Ball (1998, p. 1503) even argues that it is impossible to measure transaction costs. He concludes that, because of this, the empirical validity of transaction-cost minimising theory remains unclear. Transaction-cost theory as a positive theory is indeed questionable (see also Poulton, 1997). The doubts relate not only to the measurability of transaction costs, but also to the assumption that institutions move towards the lowest transaction costs i.e. efficiency. But many institutions persist while they are sometimes highly inefficient (Buitelaar et al., 2004). Whether measured or not, according to Furubotn and Richter (1991) transaction costs have a significant heuristic value. Heuristic in this sense means that identifying transaction costs is useful. But, why is it useful and we should we take (more) account of transaction costs? Transaction costs do not contribute directly to the output of a development process, like land, bricks, concrete and trees, in other words the production costs. From a perspective of cost efficiency, transaction costs can be seen as dead weight losses that have to be minimised. It is assumed (see also Webster, 1998) that the fewer the transaction

4 2542 EDWIN BUITELAAR costs, the more smooth and efficient the development process. Nevertheless, transaction costs are hardly incorporated in economic analyses. An example of this, given by van Keulen and Hoek (2003), is the public private partnership (PPP). People that favour PPPs claim that these partnerships add value to the project. But they often forget to mention the transaction costs. In the HSL Zuid (the Dutch part of the high-speed train connection between Amsterdam and Paris), for instance, high costs had to be incurred for consultancy, before the government chose a PPP. This paper focuses on the institutional costs, which are the costs of creating and using institutions in a development process. Reducing these costs increases the (process) efficiency of the development process in which they function. Consequently, comparing the costs of different institutional arrangements can be seen as comparing the efficiency of the development process. The general assumption behind this is that different institutional arrangements have different transaction costs (more precisely: institutional costs) borne by different people. Ideal types of institutional arrangements that I distinguish are: the market, the hierarchical and the network type. 6 Comparing the efficiency of different arrangements might contribute to a better design and use of institutional arrangements bearing in mind, hwever, that other criteria (for instance, effectiveness and legitimacy) are at least equally important. Process Efficiency Efficiency is a central concept in many economic analyses. Minimal transaction costs can be seen as maximum process efficiency (see also Webster, 1998). This efficiency concept is different from the traditional allocative efficiency. Allocative efficiency is an efficiency criterion, used in welfare economics, which indicates whether resources are allocated optimally. The optimum that is often used to refer to an efficient allocation of resources is the Pareto optimum. If an allocation is not optimal in the Pareto sense, it means that there is some reallocation of resources that would improve the welfare of at least one person without reducing anyone else s (Heikkila, 2000). Or in other words: there would be a possibility of increasing the valued output with the same (or even less) valued input. Therefore, allocative efficiency is also called output efficiency (Webster, 1998). Process efficiency does not emphasise the output. It differs from allocative efficiency in the sense that the output and the input of production costs are taken as given. Then, different ways of co-ordinating the given inputs to create the given outputs might be compared. One could see it is as a question like: how efficiently do we produce the built environment? The welfare economic question might be: how efficient is the use of the resources that produce the built environment? Allocative efficiency and process efficiency cannot be completely separated. More transaction costs lead to a greater input and hence a reduced allocative efficiency. In addition, when one wants to compare different development processes, the production costs and the output will never be completely the same. One must therefore be aware that differences in production costs i.e. input and output can have consequences for the transaction costs. Only the costs that are the result of the creation and use of institutions should be taken into account for determining the process efficiency and not the transaction costs that are caused by a difference in allocative efficiency. In my research, I focus on the concept of process efficiency, and not the welfare economic approach, for several reasons. The first reason is that I want to establish what the costs are of different institutional arrangements. Many economists use allocative efficiency to justify their statement that the market co-ordinates more efficiently than the government. The basis for this assertion goes back to Adam Smith s invisible hand. But what allocative efficiency measures is the allocation of resources. This is the result of various substantive choices of agents and agencies. Therefore we cannot, for instance, as is often done (see, for example, Ellickson,

5 TRANSACTION COSTS AND DEVELOPMENT ; Pennington, 2002), blame zoning (or other instruments of hierarchical co-ordination) for inefficient resource allocation. Zoning in itself is a neutral tool. It can, in principle, produce the same type of development with the same resources as the market would. The substantive choices made in a zoning plan can lead to the sub-optimal allocation of resources; the tool zoning cannot. The use of zoning (like any other form of co-ordination) involves transaction costs. These costs have to be regarded when we want to determine the process efficiency. So again, process efficiency judges ways of coordination, where allocative efficiency is more suitable for judging substantive choices. The second reason lies in the fact that welfare economics assumes that all collective objectives can be incorporated in the one criterion of allocative efficiency. I regard that as a simplification of the goals of collective action. Allocative efficiency is just one criterion (although an important one); another criterion could, for instance, be social stability. It should also be noted that many transaction-cost economists (for example, Williamson, 1995) see transaction costs as the overarching concept for judging different institutional arrangements. But, transaction costs are also just one criterion (besides, for instance, effectiveness, justice and legitimacy) for evaluating different forms of coordination. The third and last argument for using process efficiency is that the Pareto optimum, which is often referred to in order to operationalise allocative efficiency, is an unattainable situation. When there are still some gains from trade that have not been exhausted, the market is seen as inefficient. This means that there remain social costs (i.e. negative externalities). 7 In the welfare economic view, government intervention can in a situation of market inefficiency lead to a redistribution of resources, in order eventually to reach the Pareto optimum. 8 I suggest (see also Williamson, for example, 1995; Alexander, 2001a, 2001b) a comparison of different (existing) institutional arrangements on the basis of their transaction costs. Comparing the costs of different forms of co-ordination is a more realistic and practically manageable approach than comparing a practical to some ideal situation, like the Pareto optimum in welfare economics, or the zero transaction costs ideal in institutional economics (see North, 1990b). This is also referred to as remediability: it compares alternative feasible forms of governance, rather than referring to some hypothetical ideal (Alexander, 2001b; see also Williamson, 1995). It is important to note that we should not only take account of the total amount of transaction costs, but that we have to be aware of the distribution. Keogh and D Arcy argue that Instead of seeking a judgement on whether the property market as an entity is efficient, the institutional approach allows the possibility that the property market process may for example, be efficient for users but inefficient for developers (Keogh and D Arcy, 1999, p. 2412). Or, as Lindblom says what one person or group applauds as efficient another person or group may deplore as inefficient (Lindblom, 2001, p. 126). Operationalising the Framework Transaction costs are rarely measured in development processes. If transaction costs are at all quantified by the different parties involved, they are usually put as a nonspecified item on a balance sheet like, for example, administrative costs. Here, I put forward a method for identifying transaction costs. This is a necessary step before they can be measured. I do not provide a method for measuring transaction costs, but there is no theoretical difficulty in doing this, only the practical difficulty of the high (transaction) costs of doing it in terms of time, money and effort. To identify transaction costs, it is import-

6 2544 EDWIN BUITELAAR Figure 3. The user rights regime. ant to distinguish them from production costs. I use the model of the neo-classical market as point of reference to make the distinction. The difference, it is often argued, between neo-classical economics and new institutional economics is the lack of attention given to institutions and transaction costs in the former (see, for example, Eggertsson, 1990; North, 1990a). A neo-classical market is assumed to function frictionless and smoothly. Property rights are perfectly and costlessly specified and information is likewise costless to acquire (North, 1990a, p. 11). 9 This leads to a situation where there are only production costs (or transformation costs as North calls them) and no transaction costs. So, if we want to carry out a transaction-cost analysis of the development process, we must ask the question: would the costs that we find also be incurred in a neo-classical development process? If the answer is yes, those are production costs, if the answer is no they are transaction costs. This sounds too simple, but it will become clear that it is a very useful way to approach all the activities in the development process. I focus on the costs of creating and using an institutional arrangement, together the institutional costs. 10 The institutional arrangements that are particularly important for land use (and hence for land use planning) are what I call user rights regimes (see, for an extensive elaboration, Buitelaar, 2003). A user rights regime is a set of formal rules that defines and attenuates the right to use land and in that way co-ordinates changes in land use on a particular site. 11 This is done by means of rules from both the property rights regime and the spatial planning regime (see Figure 3). Institutions from the property rights regime are directed to facilitating exchange between the property owners, and between them and all others who do not have a right, but do have an interest, in that land. In most capitalist countries, we have the institution of private property with all its associated rights and duties (Bromley, 1991). The spatial planning regime provides tools to restrict the exercise of these private rights in the collective interest. It thereby regulates the market in land for example, by a zoning or land use plan. There are different types of user rights regimes, ranging from a market type, in which there are hardly any planning regulations, to a hierarchical, by which the market in user rights is strongly regulated. In addition, we have the network model of coordination, which is more directed to negotiation and collaboration. A user rights regime is used and (partly) created in a development process. Creating and using it involves costs. The property rights regime (the rules for the market in rights in land) is created at a national (or federal or state) level and used at a sitespecific level. The spatial planning regime is also created at a national level. The difference with the property rights regime is that the laws and rules are mainly procedural.

7 TRANSACTION COSTS AND DEVELOPMENT 2545 Table 1. User rights regimes and transaction costs Elements of the user rights regime in the development process Possible transaction-cost generating factors 1. Land exchange Yes or no Number of parties involved Conflict of interest Information about future possibilities Delineation (assignment) of rights Information re the delineation of rights: land registry or not Use of hierarchical means (such as compulsory purchase, pre-emption rights) 2. Making land use or zoning Yes or no plan/building ordinance, etc. (i.e. Stakeholder participation land use regulations) Appeal Number of parties involved Conflict of interest Structure of the plan: legally binding indicative Structure of the procedure: administrative political 3. Agreement (between, for Yes or no example, developer and Number of parties involved municipality) Conflict of interest Structure of agreement/contract: detailed flexible 4. Planning permission Yes, no or conditional Number of parties involved Conflict of interest Possibility of negotiation (planning gain) Possibility of appeal and the actual use of it Structure of the procedure: administrative political Structure of the permission: conditional unconditional The substantive filling-in occurs at a sitespecific level. The filling-in can on the one hand be seen as using the possibilities created by the spatial planning regime and on the other hand as creating the user rights regime. This creation can be in the form of a zoning or land use plan, after which the user rights regime is applied for instance, in the form of submitting and deciding upon planning permission or for a private property transaction. The level where property rights and spatial planning regimes are created, is not taken into account in the transaction-cost analysis of an individual development process, for the actors in that development process have no influence on this institutional environment (and the changes within it). It is therefore to the user rights regimes that I confine myself. What belongs in practice to these regimes and what does not, is shown in Table 1. In this table, we find four important points in the development process where the user rights regime is created (for example, land use plan) or used (for example, land exchange). To determine their affect on transaction costs there are several aspects that need to be taken into account and by which different regimes can be compared. In the second column of Table 1, some possible transaction-cost generating factors are enumerated.

8 2546 EDWIN BUITELAAR The first category in Table 1 deals with all the land transactions during the development process. These exchanges are facilitated by rules from the property rights regime. Using these institutions involves costs. The costs are further determined by how rights are delineated, the information about that, the information supply about future possibilities, the number of parties that want to acquire the land and whether or not hierarchical measures like compulsory purchase can be or are used. The second category that determines the height of the transaction costs is whether or not a land use/zoning plan is made. Related factors which might generate transaction costs are, for instance, the structure of the plan-making procedure and the plan, stakeholder participation, the number of stakeholders, conflict between them and appeal possibilities and use. The third category is agreement. In some development processes, agreements like restrictive covenants or (the British) planning agreements are signed among different parties to influence or determine the exercise of user rights. Again, the number of parties and their conflicts are important, as well as the structure of the agreement (detailed versus broad) for the size of the transaction costs. The last category is planning permission. Not every system (as we will see) has a land use plan and planning permission. If it does, it is important to see whether it is a political or administrative procedure, whether the permission is conditional or unconditional, if there is room for negotiation or appeal, how many parties are involved and whether their interests conflict. In the case study, all the (see also Table 1) categories will appear. The case study is used as a brief illustration of a transaction-cost analysis, in particular of the costs of and in a user rights regime. The case study serves as a startingpoint to discuss different designs of user rights regimes and the different costs of creating and using them. The approach is generally applicable to most Western countries. The countries must at least have the institution of private property, surrounded by rules from the property rights regime, possibly supplemented with rules that regulate land use from the spatial planning regime (see Figure 3). The Application to a Dutch Case To illustrate the user rights framework and the method for identifying transaction costs, it has been applied to a Dutch housing project. In this exploration, the sequence of activities will be briefly reconstructed, the transaction costs will be identified and the costs that can be attributed to the user rights regime will be dealt with. In addition, on several occasions reference is made to alternative regimes that have a different effect on transaction costs. The Dutch case is used as an illustration of the framework and as a starting-point for a discussion about different regimes and their transaction costs. Examples of alternative regimes are used to show that the framework does not only apply to the Dutch case, but can be used to analyse other Western capitalist regimes. The case that I have chosen is a housing project, called Lange-Veerstraat, of 71 dwellings in the municipality of Hilversum (between the cities of Amsterdam and Utrecht) in the Netherlands. Lange-Veerstraat is an urban regeneration project that (officially) started in 1994 and has not been finished yet. The motives for the municipality of Hilversum to start the project were the untidy appearance of the site, the high interest costs of the land the municipality acquired for a never-finished project in the early 1980s and the fact that a large part of the site was vacant after a school had been moved. The municipality pursued an active land policy, by acquiring most of the (remaining) land on the site. Most land was fairly easy to assemble, but one plot (of a printer) took a relatively long time to acquire. Finally, an agreement was reached in which both parties agreed upon the price and a new location. The municipality could not use compulsory purchase, because there was no legally binding land use plan (bestemmingsplan) at that time, which is a prerequisite for compulsory purchasing the land.

9 TRANSACTION COSTS AND DEVELOPMENT 2547 The project has been realised in close cooperation between the municipality and a developer. Although the municipality took the initiative and bought most of the land, the developer also acquired a plot and was involved from the start of the project. In March 2003 both parties signed an agreement in which the acquired land would be exchanged between the municipality and the developer. The municipality got all the land that is needed for public space and the developer got all the land necessary for its building plan. A clause was built into the contract that specified that the developer would not start the building process before 60 per cent of all the houses had been sold. Simultaneously with all the land development activities, a land use plan (bestemmingsplan) was established for the whole inner city of which the project is a part. Initially, the land use plan was not approved by the province on formal grounds. In 2002, it was finally approved by the province. The building plan that was made by the developer and approved by the municipality has been incorporated in the land use plan. Transaction and Production Costs The method described above for identifying transaction costs has been applied to the case study. In Table 2, a distinction has been made between production and transaction costs. For the sake of this paper, the following simplification has been made. It is impossible to reveal all the separate costs in a process. For instance, Lange-Veerstraat has already taken 10 years, in which many different costs have been incurred. Therefore, they have been identified on a more aggregated level by equating them with activities in the development process. The project start-up, for example, is an activity in the development process of Lange-Veerstraat in which many different costs can be found, like labour costs, print costs, deliberation between several departments within the municipality, deliberation between the municipality and the developer, etc. To reveal all these costs separately is practically impossible and theoretically unnecessary. The project start-up in itself is regarded as a transaction cost. The transaction costs are borne by various parties. Although there are many parties involved in a development process, the focus is on the developer and the municipality, because they are the most important parties that produce the built environment and therefore bear the majority of the transaction costs. To structure the development process, a division into three stages has been made the exploration, the plan-making and the execution stages (Koppenjan and van Ham, 2002). In the exploration stage, there is only one column, transaction costs, for there are no production costs. As said in the previous section, the neoclassical analysis of the development process forms a useful point of reference for distinguishing between production and transaction costs. The costs that contribute directly to building a house are the production costs; those costs would also have to be made in a frictionless neo-classical world. The other costs are the transaction costs, which have to be made because the acquisition of information, the creation and use of institutions are not without costs. To give an example, the acquisition of land would also be an activity that is needed in a neo-classical market : if you want to build a house you need land. 12 So, the price of land is a production cost. But negotiation and contract-making around the acquisition of land would not have existed in a frictionless market and are therefore transaction costs. Another example is the making of a building plan. A design is a necessary activity in the production process of a house or a housing block. The deliberation about the design between the municipality and the developer, in order for the developer to get planning permission, is part of the transaction costs, as are the costs of the negotiation between the developer and the architect.

10 2548 EDWIN BUITELAAR Table 2. Production and transaction costs in different stages of the development process of the Lange-Veerstraat in the municipality of Hilversum Exploration Plan making Execution Transaction costs Production costs Transaction costs Production costs Transaction costs Development plan inner Provisional Preparation for Demolition and Making city design of the land acquisition preparation of the specifications building plan from printer land Establishing and giving Definitive design Making more Tendering Application for project assignment to of the building detailed planning the project leader of the plan preconditions permission municipality Project start-up (Amicable) Administrative Preparation of the Treating acquisition of process of building site planning land approving the application provisional design (building plan and public space) Research project Move printer Deliberation with Building Putting out to boundaries architect about the tender provisional and definitive design Research design Provisional Deliberation with Maintenance after Delivery control possibilities design for public the architect of the project has space the municipality been finished about the public space Research residential Definitive design Preparation and Sale and letting Calculations of preferences and for public space starting the the land value possibilities bestemmingsplan procedure Assessment of Sales agreement Draw up sales Preparing sale ecological situation agreement and letting Historical research with More detailed soil regard to soil pollution research at the location of the printer Exploration of the Administrative buildings that had to process of move Research into commercial possibilities Research into existing and preferable planning regulations Traffic research Research on underground infrastructure Establish pre-conditions Administrative procedure of the project Preparation of the selection of the architect approving the definitive design (building plan and public space)

11 TRANSACTION COSTS AND DEVELOPMENT 2549 User Rights Regime I focus on the transaction costs that are caused by the particularities of the user rights regimes. Costs can be attributed to a particular user rights regime when they do not occur in one or more other user rights regimes. To show the regime specificity of the costs, the regime in this case will be briefly compared with alternative regimes. In the development process of Lange- Veerstraat, there are several institutions that relate to the right to use landed property that are similar to the ones shown in Table 1: for example, the sales contract in which the two parties agree upon the exchange of land (and hence the right to use it) and the future uses (see category 3 in Table 1) or the legally binding land use plan in which the right to use is attenuated (see category 2 in Table 1). The activities in this particular land development process that can be attributed to the user rights regime are represented in bold (see Table 2). I discuss them separately and show how they are different from other user rights regimes. In the exploration stage, we hardly find any costs that can be attributed to the user rights regime. Most activities are explorations of the current situation and future possibilities. The only activity is the research into existing and preferable planning regulations. Those costs would not have existed in a regime with no planning regulations at all. An example of this is El Paso County, one of the counties in Texas that border Mexico. In El Paso, there are no zoning, no sub-division, no housing or building codes and no infrastructure requirements (Kushner, 2003, pp ). In El Paso, the user rights regime is only formed by rules from the property rights regime. If in El Paso someone wants to develop a piece of land, he does not need to consult the planning regulations, which saves costs. But, on the other hand, this fails to provide any certainty about how neighbours will use their plot. Negotiation and deliberation with them are then needed to reduce one s uncertainty. Depending on the degree of conflict between the neighbours, this can raise transaction costs significantly. In other words, agents and agencies try to acquire information and reduce uncertainty, one way or another. The costs associated with the different ways can nevertheless vary significantly. The same is true of the distribution of costs. These differences are not only dependent on the regime, but also on the initiating party. In the Dutch case, the municipality took the initiative and acquired most of the land, which led to a situation in which the municipality bears most of the transaction costs (as we can see in Table 2). In England, for instance, local authorities rarely take the initiative and therefore it is often the developers that have to bear the costs of exploring the particular user rights regime. In the plan-making stage, we find most of the costs associated with the user rights regime. In the case study, the municipality acquired most of the land in an amicable way. Compulsory purchase was not possible because at that time the existing land use plan did not provide a legal basis for doing that. Therefore, the municipality was dependent on the willingness of the landowners. Most of the landowners were easily persuaded, but one plot took a long time, because an alternative location had to be found for the owner. Compulsory purchase could have brought matters to a conclusion, but it also has its costs. In the plan-making stage, the bestemmingsplan has been prepared and agreed (which took approximately two years because of an appeal). The bestemmingsplan comprises the whole inner city of Hilversum and not only the project site. Nevertheless, the costs of agreeing the plan can partly be attributed to the project. In an alternative regime, like the English, the amount and the distribution of costs might be different. The plan making of the indicative local plan in England is probably cheaper than the bestemmingsplan, because it is territory-wide, while in the Netherlands the plans are much smaller and legally binding. For instance, in

12 2550 EDWIN BUITELAAR the city of Nijmegen there are approximately 800 bestemmingsplannen, each covering only a small area. However, granting or rejecting a planning permission might involve more transaction costs in England. In England, there is room for negotiation about the content of the planning permission, possibly resulting in a planning agreement, while in the Netherlands an application for a planning permission (see the execution stage) is either granted or not, depending on whether it fits within the plan or not. In addition, the legal certainty the bestemmingsplan provides, reduces uncertainty and transaction costs for the landowners and developers. So, a user rights regime that involves fewer costs at an early point in the development process might, because of that, bring along more costs at a later stage, and vice versa. In El Paso, there are no plan-making costs for the county whatsoever, for there is no plan or anything that comes close to it. However, if there is no plan, there is hardly any certainty for the landowners. The landowners need to acquire information about each others land development intentions individually and cannot rely on a centrally developed plan or other forms of regulation. Another example where the amount and the distribution of transaction costs are different, is in the case of ballot box zoning (see, for example, Staley, 2001). This is a referendum process whereby a development initiative is placed on the ballot for public voting. Although it increases legitimacy, it increases the total transaction costs, which are partly dispersed among the citizens. A counter-reaction can be found in Oregon where the Supreme Court has transformed the zoning system from a political system into an administrative one (Kushner, 2003, p. 231). This should make the zoning process more efficient, from a transaction-cost perspective, but it breaks with the growing concern for citizen participation. Another transaction cost in the plan-making stage in the Dutch project is drawing up the sales agreement. This cost would not have been made (especially by the municipality) if the municipality had not pursued an active land policy. On the other hand, because it did pursue this policy, the municipality was able to impose conditions on the way land was used after the sale. This increased the influence of the municipality. If the municipality had chosen a passive land policy, but nevertheless wanted to keep a grip on the outcome, it would have needed to use other ways, such as persuasion and/or financial conditions, which also involve costs. In other regimes, it is often other parties that enter into agreements. In Houston Texas, for instance, there are thousands of privately established restrictive covenants, besides basic building codes, that govern land use (Siegan, 1970, p. 79). This obviously leads to a different distribution of costs. Other costs in the plan making stage that are mainly incurred by the municipality are costs associated with the administrative process of approving the provisional and definite plan for both the buildings and the public space. In England, deciding upon planning permission usually takes much more time, money and effort than in the Netherlands. In the Dutch bestemmingsplan, many of the spatial trade-offs are made. Granting or rejecting planning permission then is often an administrative procedure. In England, the local plan is indicative and the real trade-off between spatial claims is made in the planning permission stage. If planning permission as in England and the Netherlands was not needed, those costs would not have been necessary. But again, probably different costs, like the costs of signing restrictive covenants (as in Houston) are then needed. But in that case the distribution of costs might be different. The last costs that need to be mentioned are the costs associated with control. Those costs can be called ex-post costs, for they are made after using the user rights regime. After planning permission has been granted, the municipality checks the way a building is being, and has been, built. If there were no planning regulations, there probably would be other transaction costs and a different distribution. For instance, landowner A

13 TRANSACTION COSTS AND DEVELOPMENT 2551 would monitor whether his neighbour, landowner B, builds his house in the way they agreed upon in some contract. If landowner B does not, landowner A can go to court to enforce the contract. In England, most planning permissions are conditional (see, for example, Cullingworth and Nadin, 2002). Conditions might relate to the submission of further design details that have to be approved by the local planning authority, after which the condition is discharged. This can add significantly to the transaction costs. Concluding Comments Transaction costs are used in many texts in a heuristic sense. We get an understanding of what they are and how they affect the allocation of resources, but the practical relevance of the concept often remains unclear. In addition, it is hard to measure transaction costs, because they are numerous and ill-specified. It is therefore not surprising that empirical applications are rare. In this paper, I have put forward a method for identifying transaction costs in concrete development processes that can be found in most capitalist countries. A useful way to identify transaction costs is to distinguish them from the production costs. In a development process with perfect information and no institutions, all the costs that occur are production costs. Such development processes do not exist in the real world, but nevertheless form good points of reference to identify transaction costs. With every activity (in the development process), we can ask whether or not it would have been undertaken in a neo-classical development process. When we know what the transaction costs in a development process are, we can compare ways of co-ordinating the development processes. In the previous section, a comparison between user rights regimes has been made fairly briefly, but it is even more interesting to compare (significantly) different case studies in more detail. Here, I suggest some significantly different case studies that might be worth comparing. First is the case of a Dutch development process which (as we have already seen) is relatively hierarchical in its development control system and is often carried out by an active land policy of the municipality. Another interesting case would be an English case study with more private initiative and a development control system with more room for negotiation. A third case which is significantly different from the previous two, and therefore very interesting, is the case of Houston (Texas). In Houston, we find (unlike almost all other US cities) a non-zoning tradition. Although there is no zoning, it does not mean there is no planning, for the planning department is the biggest planning department of all US cities (Neuman, 2003). The result of such an investigation should be a comparison of the transaction costs of significantly different regimes and a consideration of process efficiency. It probably will not be a conclusion like: the creation and use of regime A is more efficient than that of regime B. A more nuanced analysis could lead to assertions like: in stage X, regime A involves fewer costs than regime B but, in stage Y, it is the other way around. And regime A is more efficient for agency K, while regime B is more efficient for agency L. This will shed a more qualified light than the often-heard argument that the market is more efficient than the government in co-ordinating changes in land use. Notes 1. Land use planning had been approached earlier from a new institutional economic angle by both lawyers (like Ellickson, 1977) and economists (like Fischel, 1978), but their attempts received little attention from planning theorists. 2. Healey defines land development processes as the transformation of the physical form, bundle of rights, and material and symbolic value of land and buildings from one state to another, through the effort of agents with interests and purposes in acquiring and using resources, operating rules and applying and developing ideas and values (Healey, 1992, p. 36).

14 2552 EDWIN BUITELAAR 3. Although the analogy with the production process is made, I do not intend to focus on capital flows and their determining role in the development process, and human interaction in general, as in many Marxist production-based models of the development process (see Gore and Nicholson, 1991; Healey, 1991). The analogy simply emphasises the supply side of development. In addition, production-based models hardly take account of the role of institutions; as will become clear, this paper does. 4. If the agency costs of making a product are perceived to be high or too high a company might decide to hive-off that activity. 5. Sometimes, creation and use coincide. For instance, with designing the Town and Country Planning Act in the UK in 1947, the development plan as an institution was created. If the local government makes a development plan it uses the Town and Country Planning Act and at the same time creates local rules that have to be taken into account when planning permission is decided upon. 6. Thompson et al. (1991) distinguish between three models of co-ordination: market, hierarchy and network. I have applied these three models to land use planning (Buitelaar, 2003). In general, it comes down to the assumption that in markets it is the price mechanism that co-ordinates, whereas in hierarchical forms it is command-and-control and, in network forms, collaboration that play a central role. 7. There are also people (like Bromley, 1991, p. 83) who argue that the term social costs is incorrect and that most of the negative externalities are individual costs. 8. Coase (1960, see also, for example, Eggertsson, 1990, pp ) argues that the distribution of resources i.e. the initial assignment of property rights does not matter, in a situation when the property rights are clearly delineated and there are no transaction costs associated. This thesis became famous as the Coase theorem. Coase introduced the notions of institutions, in particular property rights, and transaction costs. Although this has been a major shift in the welfare economic paradigm, Coase s goal was, like the traditional welfare economists (such as Pigou), allocative or output efficiency. Therefore, Coase can also be called a welfare economist. 9. North (1990a, p. 17) argues that, although hardly any economists believe that the behavioural assumptions reflect human behaviour, they do build many of their models upon those assumptions. 10. I must add to this that the change and use of institutions do not only give rise to the transaction costs, but can also increase the production costs. For instance, the application of some environmental act might require soil decontamination. These costs are production costs, but do not exist in a neo-classical market. So, transaction costs are the costs (other than the production costs) that would not exist in a neo-classical market, not being the costs of the physical production of the built environment. I will give no further attention to the latter costs. 11. The bundle of rights consists of the right to use (usus), the right to income (usus fructus) and the right to transfer (abusus) (see de Alessi, 1991). The right to use a piece of land (i.e. the user rights over that land), and the restrictions upon it, are of particular interest to land use planning. 12. It should be noted that, because a price has to be paid for a piece of land (like any other good), it suggests that there is an institution like private property, while a neo-classical market is supposed to be institutionless. References ALESSI,L.DE (1991) Development of the property right approach, in: E. G. FURUBOTN and R. RICHTER (Eds) The New Institutional Economics: A Collection of Articles from the Journal of Institutional and Theoretical Economics, pp Tübingen: Mohr. ALEXANDER, E. R.(1992) A transaction cost theory of planning, Journal of the American Planning Association, 58, pp ALEXANDER, E. R.(2001a) A transaction-cost theory of land use planning and development control; toward the institutional analysis of public planning, Town Planning Review, 72, pp ALEXANDER, E. R.(2001b) Why planning vs. markets is an oxymoron: asking the right question, Planning and Markets, 4 (on-line journal). BALL, M.(1998) Institutions in British property research: a review, Urban Studies, 35, pp BROMLEY, D. W.(1991) Environment and Economy: Property Rights and Public Policy. Oxford: Blackwell. BUITELAAR, E.(2003) Neither market nor government: comparing the performance of user rights regimes, Town Planning Review, 74, pp BUITELAAR, E., JACOBS, W. and LAGENDIJK, A. (2004) Institutional change in spatial governance: illustrated by Dutch city-regions and Dutch land policy. Paper presented at the GaP Congress Mind the GaP: The governance of