DEVELOPING A NEW SPATIAL COMPUTABLE GENERAL EQUILIBRIUM MODEL FOR NORWAY Wiljar Hansen Institute of Transport Economics, Oslo, Norway

1. INTRODUCTION Larger infrastructure investments often generate considerable debate between those seeking to justify the investment and those trying to refute the need for that particular infrastructure extension. The favorable argument among those justifying the proposed investment is the presents of wider economic benefits not captured by the traditional cost-benefit analysis conducted as part the investment appraisal. With a reasonable degree of perfect competition in most markets, most transport researchers claim that the cost-benefit framework will capture all benefits associated with the investment since the users will want to pay exactly their value of the traffic improvement. However, as also noted by most transport researchers, most markets are not perfectly competitive. Market imperfections lead to economic benefits that are not adequately reflected in the transport appraisal. There is a large literature on the wider economic benefits of transport infrastructure investments (Mohring Jr and Williamson 1969; SACTRA 1999; Oosterhaven and Elhorst 2003; Vickerman 2007; Lakshmanan 2010), using a variety of scientific methods where the most common are (Tavasszy, Thissen et al. 2002):      

Micro surveys with firms Estimation of quasi production functions Partial equilibrium potential models Macro and regional economic models Land Use / Transportation Interaction models Spatial Computable General Equilibrium models

The aim of this paper is to propose a new Spatial Computable General Equilibrium (SCGE) model for Norway. The proposed model is in the new economic geography (NEG) tradition and an extension of the existing Norwegian SCGE model, PINGO (Ivanova, Vold et al. 2002; Vold and JeanHansen 2007). SCGE models are spatial and operational extensions of General Equilibrium models. Modern classics in theoretical general equilibrium analysis are Debreu (1959) and Arrow and Hahn (1971). A review of CGE modeling is found in Shoven and Whalley (1992). This paper is structured into 6 sections. The following section gives the rationale on why the traditional cost-benefit framework is unable to capture the wider economic benefits of infrastructure investments. The section discusses the market imperfections in the transport sector, in the product markets and the labour market. Section 3 presents SCGE modeling as a tool to quantify the wider economic benefits associated with larger infrastructure investments. © Association for European Transport and contributors 2010

1

Section 4 presents the current Norwegian SCGE model PINGO and its shortcomings. Section 5 proposes further development to the PINGO model in order to capture wider economic effects of traffic improvements, and section 6 concludes the paper. 2. WIDER ECONOMIC BENEFITS OF TRANSPORT INFRASTRUCTURE INVESTMENTS Vickerman (2007) define wider economic benefits to be all economic benefits not captured in the direct user benefits of the type which are normally analyzed in a well constructed transport cost-benefit analysis after allowing for environmental and other directly imposed external costs. In transport appraisal it is traditionally assumed that a well specified costbenefit analysis (CBA) will capture all the relevant impacts on the economy. Simply put; In a CBA one adds up all the benefits associated with a policy alternative, subtract all the costs, and choose the alternative that maximizes the net present value. As long as all markets are perfectly competitive the user benefit will equal the total benefit of the investment (Kanemoto and Mera 1985; Jara-Diaz 1986). Adding spillover effects in a perfect competitive environment will only result in double counting (Mohring 1993). Neither the transport sector nor the transport using sectors are perfectly competitive. If the markets are imperfect, i.e. the price on important market goods exceeds marginal cost, such a deviation from the first-best solution implies that the traffic improvement produce impacts in other sectors of the economy not evening out (Jara-Diaz 1986). Thus, market imperfections may lead to an under estimation of the user benefit of the project (Venables and Gasiorek 1998; SACTRA 1999). The benefits not captured by the direct change in consumer surplus are labeled wider economic benefits. There are many reasons for market imperfections, the most common reasons being taxes and subsidies and market power, where i.e. economies of scale may lead to unregulated market power in product markets. The pervasive departure from perfect competition in the NEG literature is the Dixit-Stiglitz model of monopolistic competition (Dixit and Stiglitz 1977). Here, product differentiation allows the producers to charge above marginal cost, but competition drives the monopoly profits to zero and prices equal to average cost. The Dixit-Stiglitz model has become the workhorse for most NEG modeling. The NEG tradition acknowledges that in a perfectly competitive environment price mechanisms alone are unable to endogenously generate economic agglomeration. In order to build a model with the purpose of analyzing the formation of economic agglomeration, one has to depart from the notion of a perfect competitive economy (Fujita and Thisse 1996). Lower transport costs due to new or improved transport infrastructure may lead firms to exploit their economies of scale and hence influence the location of economic activity. In the NEG literature there seems to be consensus on the two main forces behind agglomeration and regional dispersion: 1. Increasing returns to scale in production, and 2. market power through product differentiation

© Association for European Transport and contributors 2010

2

Following the core-periphery model of Krugman (1991), NEG demonstrates that the interplay between these two main forces and factor mobility and transport costs give rise to agglomeration in general equilibrium models (Fujita, Krugman et al. 1999; Fujita and Thisse 2009). In the NEG literature high transport costs yield equilibrium with even dispersion of economic activity, while low transport costs give rise to agglomeration. In this setting transport costs work as a centrifugal force while increasing returns to scale work as a centripetal force. A synthesis of the pioneering works of NEG is found in Fujita, Krugman et al.(1999). In addition to the product market being imperfect, the labour market is imperfect as well, both at the national and at the regional level. The gap between the gross wage for the employer and the net wage for the employee, as well as the regional immobility of workers and inflexibility of wages, indicate a market with strong imperfection characteristics.

3. SCGE MODELING IN TRANSPORT ECONOMICS SCGE models are spatial extensions of Computable General Equilibrium (CGE) models. The MSG model of the Norwegian economy (Johansen 1960) is widely credited for being the originator of the CGE modeling tradition. A CGE model builds on a benchmark equilibrium dataset that accounts for all the economic transactions in a base period. A Social Accounting Matrix (SAM) is used to represent the equilibrium situation where all the economic agents and goods are represented. In the SAM matrix the columns typically represent the economic agents’ accounts while the rows represent markets for goods and factors of production. A CGE model is then a solvable system of equations that reproduce the equilibrium dataset. This reproduction is made by making assumptions on the market structures and functional forms of the preferences and technologies, and by setting parameter values on the substitution elasticities. The system of equations describes the behavior of economic agents (households, firms) and institutions, the structure of the markets (goods, assets, production factors), and the interaction between these. The CGE models are then used to estimate the reactions on the economy of exogenous changes in policy, technology or other external factors. The spatial extension of CGE models is achieved by an explicit representation of the transport of each commodity within each region and between all pairs of regions in the model. Multiregional SCGE models typically aim at quantifying regional effects of transport infrastructure investments or changes in transport policy. An infrastructure investment or policy change lead to changes in the trade costs which produce repercussions in the transport using sectors and other related markets. SCGE models embrace the entire economy, making these models specially suited for analyzing wider economic benefits of transport investments through the link between the transport sector and the transport using sectors, acknowledging that an exogenous change in one sector may produce repercussions throughout the economy.

© Association for European Transport and contributors 2010

3

The main advantages of SCGE modeling for transport appraisal lies in the ability to compare outcomes of different equilibrium states (Tavasszy, Thissen et al. 2002), such as (Oosterhaven and Knaap 2003): 

 

Benefits of generalized transport cost reductions due to changing prices, production, consumption and trade, while holding the number of firms and workers per region constant; showing what could be labeled as short-run effects; Benefits of transport cost reductions when the number of firms per region is allowed to change; showing medium term effects; Benefits when the number of workers is allowed to change too; showing the long run effects of new transport infrastructure.

The model in Bröcker (1998) was the first European example of an SCGE model and has been an inspiration for many following European SCGE models. In the Bröcker model regional welfare effects of transport related policies are quantified. An example of a sophisticated European SCGE model is the Dutch RAEM model (Ivanova, Heyndrickx et al. 2007). Bröcker and Mercenier (2009) provide a tutorial on the theory behind SCGE modeling in transport economics. 4. THE PINGO MODEL AND ITS IDENTIFIED SHORTCOMINGS PINGO is a SCGE model for prediction of regional and interregional freight transport in Norway. The model has been developed in two stages at the Institute of Transport Economics in Norway by Ivanova, Vold et al.(2002) and Vold and Jean-Hansen (2007). PINGO is a static SCGE model where space is explicitly represented in the form of freight transport costs between the regions in the model. The original PINGO structure is part of a neo-classical general equilibrium modeling tradition assuming constant return to scale and perfect competition in all markets. This implies that all prices are equal to marginal cost and that technical development or other changes that leads to lower unit costs automatically leads to lower prices, hence, all cost reductions are passed on to the consumers. In a perfect competitive economy with constant returns to scale there are no wider economic benefits of investments meaning that price mechanisms alone are unable to endogenously generate economic agglomeration (Starrett 1978). The initial development of PINGO was based on the models proposed by Hussain (1996) and Bröcker. However, PINGO differs from these models in the representation of the transport sector and transport costs. The transport costs in Bröcker (1998) is based on Samuelson’s (1954) iceberg model where transport costs take the form of shrinkage en route. In PINGO the goods specific freight transport costs between pair of regions are calculated in the Logistics model, based on (deJong, Grønland et al. 2005; deJong, Baak et al. 2007; deJong, Ben-Akiva et al. 2008). PINGO and the Logistics model together yield a national freight model system for Norway with elastic demand, where the Logistics model is used as a sub model to PINGO and vice-versa. The Pingo model is predicting future demand both in the base and alternative scenarios between different regions, given exogenous growth from a national applied general equilibrium model, MSG6 (Heide, Holmøy et al. 2004). The © Association for European Transport and contributors 2010

4

supply part of the national freight modeling system consists of the Logistics model. In the version 2 of PINGO (Vold and Jean-Hansen 2007) the number of regions are extended. In this second version of the model the Norwegian economy is represented by 21 regions covering the 19 Norwegian counties, one off-shore zone and one zone representing foreign trade. Each region shelters nine different production sectors producing 32 commodity groups, as well as six service groups and six investment types. In addition there are sectors for private and public consumption, as well as a specific government sector at national level for taxes/subsidies and sectors for export/import. Interregional trade in PINGO is modeled via the so-called pooling concept. The Chenery – Moses model, (Chenery 1953) and (Moses 1955), introduced the pooling approach in interregional trade. This approach is often chosen in general equilibrium modeling in order to keep the data requirements low. According to the pooling concept no direct link exists between the producer and the consumer since all commodities produced by a sector in a region is pooled by a transport agent prior to their delivery for final or intermediate use. Each region in the model harbors such transport agents who are responsible for pooling commodities. Figure 1 illustrates the structure of the PINGO model.

Figure 1: The structure of the PINGO model PINGO is implemented in the GAMS/MPSGE programming package (Rutherford 1999). The PATH solver (Ferris and Munson 2000) is employed to provide solutions to the Mixed Complementary Problems (MCP). Ivanova, Vold et al (2002) identify a range of shortcomings to the first version of the PINGO model. The authors provide the following list of further possible developments in order to improve the reliability of the model.     

Estimation of elasticities Improve the description of import Mobility of physical capital and labour Segmentation of household groups Include economies of scale effects

© Association for European Transport and contributors 2010

5



Better forecasts.

The second version of PINGO (Vold and Jean-Hansen 2007) does little to mend these shortcomings. Version 2 of PINGO mainly focuses on further disaggregation of sectors and commodity groups, and updating the SAM matrix to 2003 as the base year. 5. NEW DATA AND PROPOSED CHANGES TO THE MODEL In this section of the article the proposed changes to the PINGO model are presented. However, one has to bear in mind that an expansion of the model increases the data requirements, the variables involved in the calibration process, and the equations. There is a trade-off between wanting to expand the model into yet another dimension and the possibility of not reaching equilibrium at all. New data: Commodity flow survey The new commodity flow survey provided by Statistics Norway enables us to update the benchmark dataset used in the PINGO model to 2008 as base year. The new data will be used to establish new origin-destination matrices in the Logistics model and hence provide equilibrium transportation costs between pairs of regions for all the commodity groups in the model. In order to maintain the consistency in the input data in the model all the data will be updated to 2008 as base year. In the commodity flow survey conducted by Statistics Norway each responding company is asked to provide the following information for every delivery point in 2008:      

The postal code Aggregated weight or volume Unit of measurement (tons or m3) Number of deliveries Turnover or value of goods Who paid for the transport (buyer or seller)

This new data will provide new and valuable information on the geographical origin of the intermediate products in the different sectors. Market imperfections and increasing returns The perfect competition structure of the PINGO model makes it unsuitable to study wider economic benefits of transport infrastructure investments. In order to build a model with the aim of analyzing such indirect economic effects one has to depart from the assumptions of constant returns to scale and perfect competition. The Dixit-Stiglitz model of monopolistic competition is the pervasive departure from perfect competition in the NEG literature. This model offers an alternative to the Arrow-Debreu framework by integrating imperfect competition and increasing returns. Monopolistic competition was first established by Chamberlin (1933) who wanted to formalize a industry configuration where

© Association for European Transport and contributors 2010

6

  

Each firm faces downward sloping demand There is no profit A price change made by one firm has negligible effect on the demand faced by the other firms.

Monopolistic competition assumes that each sector is characterized by a number of homogenous firms operating under the same technology producing slightly differentiated products. Every firm produce a single variety under increasing returns to scale. And there is free entry and exit to the markets, hence, zero profits. One of the characteristics of the monopolistic competition model that makes it especially appealing is the large group assumption that abstracts the model from strategic interaction between firms. Fujita (1988) was the first to apply Chamberlin’s monopolistic competition and Dixit-Stiglitz’ product differentiation in a spatial setting In the Dixit –Stiglitz version of monopolistic competition a representative consumer who embodies the aggregated preferences of the population is used as a simplifying approach. This representative consumer has symmetric preferences, i.e. no product can be ranked over another product based on the price. This form of product differentiation is in contrast to the Armington approach since there is no preference of one product over another meaning that no product can be preferred based on its origin. The substitution elasticity of a sector measures the degree of monopolistic competition in that sector. If the substitution elasticity of a sector moves towards infinity the degree of competition tends towards being perfect. In Sundberg (2009) the specification of the transport agents’ technology allows parameterization between perfect and monopolistic competition. Hence, in the market clearing, as the elasticity between varieties goes to infinity, the demand system moves toward a standard Armington approach under perfect competition. The level of competition in each sector is an issue of discussion when modeling the demand structure in the model. In the monopolistic competition approach commodities produced in different regions, and abroad as well, are perfect substitutes in the consumption bundle. Most domestic SCGE models have a foreign region for imports and exports and the substitution elasticities between domestic and foreign products are treated differently in different models. The RAEM model assumes Armington elasticities of substitution in international trade meaning that imported and domestic products are imperfect substitutes. On one hand the Armington assumption in international trade is a realistic assumption since there is a tendency towards predominance of domestic products in a consumers consumption bundle. On the other hand one can argue that the consumers often are unaware of the origin of the products consumed and even if the consumer has a preference for domestic products the domestically branded product is often assembled from imported parts. In this paper it is proposed to depart from the assumptions of constant returns to scale and perfect competition found in the two first versions of PINGO. The strength of the “ love of variety” approach is often overrated in the standard © Association for European Transport and contributors 2010

7

monopolistic competition approach (Ardelean 2006) in light of this result a parameterization of the substitution elasticities in the model will be investigated. Integrating passenger transport in the model In order to analyze the short run effects on the labour market of an infrastructure investment it is necessary to include passenger transport and commuting in particular in the model. There are several possible ways to include passenger traffic in the model. The Dutch RAEM model (Ivanova, Heyndrickx et al. 2007) separate between business, commuting, shopping, education and travel purposes of the passenger trips. The amounts of trips associated with each of these purposes are generated according to a set of trip generating functions. These trip generating functions take into account time and money costs of the travel as well as a set of attraction factors. Another option for passenger transport integration that could be investigated is to include the national passenger modeling system with the national freight modeling system. That is to expand the dimensions of PINGO by including the generalized passenger transport costs calculated in the national passenger transport model in PINGO in the same manner as the freight transport costs from the Logistics model are included. The national passenger transport system consists of two models, NTM5 for long distance passenger travel and RTM for short distance passenger travel. As noted earlier, changes in the transportation costs as a result of an infrastructure investment may lead to changes in the location of economic activity. These agglomeration effects are results of market power and increasing returns to scale. Changes in the location of economic activity will lead to changes in the commuting patterns, in addition to the direct changes in economic benefits resulting from decreased generalized passenger transportation costs. Integrating changes in the generalized passenger transport costs in the SCGE model enables us to quantify the wider economic benefits from an infrastructure investment found in changes in the commuting patterns. Segmentation of households The development of the model will investigate segmentation of the households according to their income deciles. The total income per household is typically calculated as the sum of its labor income and capital income, taking account for the regional labor supplied in other regions. The households’ consumption budget consists of their net income plus social transfers and unemployment benefits, minus savings and money spent on transport. The current version of PINGO follows the traditional assumption of one representative consumer in each region with aggregated preferences. Segmenting the households after income level will enable a better description of the variance in consumption preferences and enable a better modeling of the labour market effects of an improvement in infrastructure.

© Association for European Transport and contributors 2010

8

Migration Migration is a long run effect of changes in the localization patterns of economic activity. Agglomeration or regional dispersion as a result of infrastructure improvements may in the long run lead to migration. In addition to this, lower generalized transportation costs may lead to an enlargement of the catchment area by migration to regions with lower housing costs. The trade off between commuting and migration is affected by the cost characteristics in the different regions, i.e. the housing costs etc., as well as the generalized transportation costs. Studies also show that people with higher education are willing to commute longer distances than people with lower education (Harsman and Quigley 1998; Trendle and Siu 2005). The proposed modeling of migration will at large follow Ivanova, Heyndrickx et al. (2007). A separate migration model that clears the labour market will be investigated. Migration could be modeled as a nested logit model where the upper level of the discrete choice model decides on migration or not, while the destination for the migration is decided on level two. A similar proposition was made in Ivanova and Eriksen (2004) in a one region model, where households’ choices on resident-job pairs of location are represented using a logit model. In their model, households’ location choices are influenced by prices of goods, availability of housing, location specific wage levels as well as commuting times between all pairs of locations. 6. CONCLUDING REMARKS Market imperfections lead to economic benefits not adequately reflected in the cost benefit analysis traditionally used in transport appraisal. The current structure of the Norwegian SCGE model PINGO unable us to study the wider economic benefits from transport infrastructure investments or transport policy changes. This is due to the perfect competition and constant returns to scale assumptions in the model. The current version of PINGO is also unable to show the indirect repercussions of an infrastructure investment on the labour market and the location of economic activity. The following table summarizes the identified shortcomings in the PINGO model and the proposed changes presented in this paper.

© Association for European Transport and contributors 2010

9

Table 1: Identified shortcomings in PINGO and proposed changes to the model. Current PINGO model Perfect competition and constant returns to scale

One representative consumer with aggregated preferences. Immobility of labour, medium term effects. Immobilty of labour, long-run effects.

Proposed changes Monopolistic competition and increasing returns to scale. Parameterization between monopolistic and perfect competition. Segmentation of households after income deciles. Integrating passenger transport Migration sub model to PINGO

The use of SCGE models for transport appraisal has gained popularity as the computational power provided to researchers has grown and with the development of robust and efficient algorithms for solving the mathematical problems associated with SCGE modeling. Plans for larger infrastructure investments often result in considerable public debate on the methodology used in the transport appraisal. Those in favor of the investment often refute the traditional CBA used in investment evaluation. Development of a modeling framework with the ability to capture possible wider economic benefits of infrastructure investments may contribute to enlightening this debate.

REFERENCES Ardelean, A. (2006). How Strong is the Love of Variety?, Working Paper, Purdue University. Arrow, K. and F. Hahn (1971). General competitive analysis, Holden-Day San Francisco. Bröcker, J. (1998). "Operational spatial computable general equilibrium modeling." The Annals of Regional Science 32(3): 367-387. Bröcker, J. and J. Mercenier (2009). "General equilibrium models for transportation economics (*)." Handbook in Transport Economics. Chamberlin, E. (1933). The Theory of Monopolistic Competition, Cambridge: Harvard University Press. Chenery, H. (1953). Regional analysis. The structure and growth of the Italian economy. H. Chenery, P. Clark and V. Cao-Pinna, United States of America, Mutual Security Agency: 98-139. Debreu, G. (1959). Theory of value, Yale Univ. Pr.

© Association for European Transport and contributors 2010

10

deJong, G., J. Baak, et al. (2007). Calibration and validation of the logistics module in the Norwegian and Swedish National Freight Model Systems. Deliverable 5a, Significance (Leiden).

deJong, G., M. Ben-Akiva, et al. (2008). Method Report - Logistics Model in the Norwegian Freight Model System. Deliverable 6A. , Significance, Den Haag. deJong, G., S. E. Grønland, et al. (2005). The Development of a Logistics

Module in the Norwegian and Swedish National Freight Model Systems. Deliverable 4: Final Progress Report on Model Development., Rand Europe (Leiden) and SITMA (Oslo).

Dixit, A. and J. Stiglitz (1977). "Monopolistic competition and optimum product diversity." The American Economic Review: 297-308. Ferris, M. and T. Munson (2000). "Complementarity problems in GAMS and the PATH solver1." Journal of Economic Dynamics and Control 24(2): 165188. Fujita, M. (1988). "A monopolistic competition model of spatial agglomeration* 1:: Differentiated product approach." Regional Science and Urban Economics 18(1): 87-124. Fujita, M., P. Krugman, et al. (1999). The spatial economy: cities, regions and international trade, The MIT press. Fujita, M. and J. Thisse (1996). "Economics of agglomeration." Journal of the Japanese and international economies 10(4): 339-378. Fujita, M. and J. Thisse (2009). "New economic geography: an appraisal on the occasion of Paul Krugman's 2008 Nobel Prize in Economic Sciences." Regional Science and Urban Economics 39(2): 109-119. Harsman, B. and J. Quigley (1998). "Education, Job Requirements, and Commuting: An Analysis of Network Flows." Knowledge and networks in a dynamic economy: festschrift in honor of Åke E. Andersson: 261. Heide, K., E. Holmøy, et al. (2004). "Macroeconomic Properties of the Norwegian Applied General Equilibrium Model MSG6, Reports 2004/18." Statistics Norway. Hussain, I. (1996). Benefits of transport infrastructure investments: a spatial computable general equilibrium approach. Umeå, Umeå universitet. Ivanova, O. and K. S. Eriksen (2004). The influence of market imperfections on the evaluation of investments - a SCGE model approach. Nectar cluster meeting 20. march 2004. Amsterdam. Ivanova, O., C. Heyndrickx, et al. (2007). "RAEM: version 3.0."

© Association for European Transport and contributors 2010

11

Ivanova, O., A. Vold, et al. (2002). PINGO A model for prediction of regional and interregional freight transport. Version 1. Oslo, Institute of Transport Economics. Jara-Diaz, S. (1986). "On the relation between users' benefits and the economic effects of transportation activities." Journal of Regional Science 26: 379-391. Johansen, L. (1960). A multi-sectoral study of economic growth, NorthHolland Amsterdam. Kanemoto, Y. and K. Mera (1985). "General equilibrium analysis of the benefits of large transportation improvements* 1." Regional Science and Urban Economics 15(3): 343-363. Krugman, P. (1991). Geography and trade, the MIT Press. Lakshmanan, T. (2010). "The broader economic consequences of transport infrastructure investments." Journal of Transport Geography. Mohring, H. (1993). "Maximizing, measuring, and not double counting transportation-improvement benefits: A primer on closed-and open-economy cost-benefit analysis." Transportation Research Part B: Methodological 27(6): 413-424. Mohring Jr, H. and H. Williamson (1969). "Scale and" Industrial Reorganisation" Economies of Transport Improvements." Journal of Transport Economics and Policy 3(3): 251-271. Moses, L. (1955). "The stability of interregional trading patterns and inputoutput analysis." The American Economic Review 45(5): 803-826. Oosterhaven, J. and J. Elhorst (2003). "Indirect economic benefits of transport infrastructure investments." Across the Border. Building upon a Quarter Century of Transport Research in the Benelux: 143–62. Oosterhaven, J. and T. Knaap (2003). Spatial Economic impacts of Transport Infrastructure Investments. Transport projects, programmes, and policies: evaluation needs and capabilities. A. Pearman, P. Mackie and J. Nellthorp, Ashgate Pub Ltd. Rutherford, T. (1999). "Applied general equilibrium modeling with MPSGE as a GAMS subsystem: An overview of the modeling framework and syntax." Computational Economics 14(1): 1-46. SACTRA (1999). Transport and the Economy. London, Standing Advisory Committee on Trunk Road Appraisal. Samuelson, P. (1954). "The transfer problem and transport costs, II: Analysis of effects of trade impediments." The Economic Journal 64(254): 264-289. © Association for European Transport and contributors 2010

12

Shoven, J. and J. Whalley (1992). Applying general equilibrium, Cambridge Univ Pr. Starrett, D. (1978). "Market allocations of location choice in a model with free mobility." Journal of Economic Theory 17(1): 21-37. Sundberg, M. (2009). "Essays on spatial economies and organization." Tavasszy, L., M. Thissen, et al. (2002). Pitfalls and solutions in the application of spatial computable general equilibrium models for transport appraisal. Trendle, B. and J. Siu (2005). "Commuting Patterns of Sunshine Coast Residents and the Impact of Education." Australasian Journal of Regional Studies, The 13(2): 221. Venables, A. and M. Gasiorek (1998). "The welfare implications of transport improvements in the presence of market failure." Report to SACTRA. Vickerman, R. (2007). Recent evolution of research into the wider economic benefits of transport infrastructure investments. Vold, A. and V. Jean-Hansen (2007). Pingo - a model for prediction of regional and interregional freight transport in Norway. Oslo, Institute of Transport Economics.

© Association for European Transport and contributors 2010

13