Using a CGE Model for Analyzing the Macroeconomic Impact of the Grand Paris Express Project

Using a CGE Model for analyzing the Macroeconomic impact of the Grand Paris Express project on the Ile-de-France Region

Haykel Hadj-Salem, Aboulkacem El-Mehdi, Hubert Jayet, Quentin David, Hakim Hammadou and Moez Kilani,[1]

Abstract

This paper presents the first version of a new Computable General Equilibrium (CGE) model aiming at analyzing the expected macroeconomic impacts of a new infrastructure project, the so-called “Grand Paris Express” (GPE), on the Ile-de-France region. The GPE consists in a new high-speed circular underground train linking suburbs around Paris.

We present a new Social Accounting Matrix(SAM) developed for the model. This SAM provides a detailed representation of the transportation system and includes the land markets. This first version focuses on the public transportation sector.

The simulations focus on the impact of a shock in the transportation sector on the various sector of the economy of Ile-de-France. The objective of this analysis is to identify public policies that can improve the implementation of the Grand Paris Express. These policies will be related to transportation reforms itself, but also to real estate market and to the labor market.

Keywords: CGE model, Social Accounting Matrix, Transport sector, Ile-de-France region, monopolistic competition.

Table of Contents

Introduction

1.Literature review

2.Outlines of the CGE model and objectives of the analysis

3.The structure of the CGE model

3.1The Social Accounting Matrix

3.1.1 The general structure of the SAM

3.1.2 The structure of the SAMIF

3.1.3 Balancing of the SAMIF

3.2.The current version of the CGE model for the Ile-de-France Region

3.2.1Production

3.2.2Incomes and saving

3.2.3Demand

3.2.4Prices

3.2.5International trade

3.2.6Transport

4.Interpretations of the simulations

Conclusion

Appendix 1: Data sources

Appendix 2: The balanced SAMIF

Appendix 3: The results of the simulations of the series 1 from the CGEMIF (with GAMS software)

Appendix 4: The results of the simulations of the series 2 from the CGEMIF (with GAMS software)

Appendix 5: The results of the simulations of the series 3 from the CGEMIF (with GAMS software)

Introduction[2]

This paper presents the first version of a Computable General Equilibrium model aiming at analyzing the expected macroeconomic impacts of a new infrastructure project, the so-called “Grand Paris Express” (GPE), on the Ile-de-France region. The GPE is a large infrastructure project, leading to the creation of new express subway lines all around Paris. The lines are expected to start receiving passengers in 2023 for the first line, in 2030 for the last line.

The size of this new infrastructure, its high speed, and its circular shape in a region where transportation is almost exclusively radial, will radically change the level and geographical structure of accessibilities within the Ile-de-France Region. The operator in charge of this infrastructure, the Société du Grand Paris (SGP)has engaged in a large research program for analyzing the impact of these changes.

Passenger transportation has an important influence on location decisions of households and firms. It also has strong impacts on everyday decisions related to urban trips an agent may, or may not, undertake. Improving public transportation, either through a better quality (less crowding, higher frequency, newer vehicles, etc.) or through transit network development, can contribute to reduce car usage but may also generate economic benefits that cover the development of some commercial activities, increased productivity in some economic sectors or better efficiency in the labor market. We have been commissioned by the SGP to develop a modelaiming at quantitatively evaluate these impacts.

This new transport infrastructure is expected to impact various economic aspects of the Ile-de-France Region: extra economic growth, employment, real estate market, regional attractiveness, etc. To evaluate these impacts, we rely on a CGE model of the Region Ile-de-France where the local economy is represented through a number of representative agents and goods. The current version of the model displays several innovative features.

First, weinclude a detailed representation of the transportation system. This model is built with the objective to focus on transportation activities and their relationship with agents’ behavior. Since we focus on the GPE project, only passenger transportation is considered in the transportation part and freight transport is included in the activity of firms. At this stage, road congestion and crowding in public transportation are assumed to depend on passenger transportation only. It is not difficult to extend the model if some data on freight transport were made available.The Social Accounting Matrix (SAM) takes into account the public transport sector. The model includes supply and demand functions for public transport. Moreover, the demand is directly derived from micro-geographic foundations. More precisely, the determination of the demand for public transport takes account of the spatial distribution of households across locations with various levels of accessibility and transport costs measures. These micro-geographic foundations allow us to determine the impact on demand of the changes in accessibility and transport costs induced by the GPE.

Second, we study the interaction between the transportation system and the agents’ decisions. This question is more and more addressed through the development of land use and transport infrastructure models (LUTI models). Although these models provide valuable information on how human activities will evolve when a large transport project is made, they remain silent when it comes to macroeconomic impacts that cover economic growth and job creations. These issues are crucial since transportation infrastructure projects require large financial funds. The objective of the model is to provide a tool to assess the macroeconomic impacts of such infrastructures. Location decisions are not the focus of our analysis, but we rely on the outcome of LUTI models on this question (there are three LUTI models running on the case of the GPE). The CGE model we develop relates agents’ choice (consumers and firms) with transportation opportunities and costs. For example, with easier travel possibilities, a user may consider a new leisure activity that he couldn't consider at the initial situation. Expenditures from this agent will be revenue for other agents.

Third, the model is not a standard Walrasian model with perfectly competitive markets. Most goods and services are differentiated and produced under increasing returns, leading to monopolistically competitive markets.

Fourth, the model, in its final version, will include a representation of the regional labor market and the influence of transport infrastructure on its behavior. The Ile-de-France Region currently suffers from a high level of spatial mismatch between local labor markets, particularly for low skilled workers. The new infrastructure is expected to reduce the impact of the spatial mismatch. Our model is able to measure these frictions and their macro-economic impact. Using micro-geographic foundations, we will study the impact of the project on the aggregate equilibrium on the regional labor market (obtained from the aggregation of micro markets imperfectly connected to each other). The Grand Paris Express will impact the aggregate labor market through changes in the degree of connection between micro-markets.

Fifth, the model includes land inputs and a Real Estate sector providing housing and industrial premises. Through the inclusion of land inputs, we aim at analyzing the impact of the development of a new infrastructure on land rents and the housing market; moreover, we try to provide a tool for studying the interactions between the infrastructure and the land use policies developed by local authorities. Once again, the representation of the land market is derived from micro-geographic foundations in order to take account of the heterogeneity of land with respect to accessibility and the changes in accessibility induced by the Grand Paris Express.

The paper is organized as follows. We present a literature review in the next Section. In Section 2, we present the CGE model of the Ile-de-France Region. The structure of the model is presented in Section 3. Section 4 provides the outcome of various scenarios and simulations and we conclude in Section 5.

1.Literature review

If the literature on computable regional equilibrium models is large, there are much less models taking account of the regional or multi-regional dimension and even less explicitly taking account of transportation activities and their impact on the economy, particularly passengers transport.

Typical models like Monash MMRF for Australia (Giesecke and alii, 2008; Naqvi and Peter, 1995) do not explicitly include transportation activities. The impact of transportation only appear in exogenous transport of goods. The same holds for the models elaborated by Bröcker (1998; 2001; 2004; 2013). If, in the last version of his model, Bröcker makes an explicit link between transportation costs and infrastructure, transportation only appear through transportation costs without any representation of transportation activities and passagers transport is completely neglected.

A more explicit treatment of transportation activities is provides by RAEM (ivanova and alii, 2007), which includes a sector devoted to transportation of goods. A more developed solution is proposed by Conrad and Heng (2002), who include four sectors devoted to transport activities and transport inputs used by the other sectors;however, they focus on transport of goods only. RAEM also explicitly takes account of passengers transport,and it is one of the very few models including this dimension. RAEM takes account of passengers transport through a a gravity submodel using a general transport cost.

To the best of our knowledge, the CGE model including the most explicit representation of passengers transport has been elaborated by Mayeres and Proost (2001, 2004). Their model includes a highly detailed representation of transportation activities, both for goods and passengers, with eleven transport modes. They explicitly represent the choice of transport modes and they take account of congestion and pollution.

To the best of our knowledge, there is no regional model making an explicit link between transportation, labor markets, and the real estate sector. This type of link appears in urban models (Anas and Kim, 1996; Anas and Liu, 2007), but here we are closer to the family of LUTI models than to the family of macroeconomic regional models.

2.Outlines of the CGE model and objectives of the analysis

Accounting for the global impact of an infrastructure requires a framework where we show how the relevant agents interact and how monetary flows are organized. The building block of such a CGE model, as in traditional models, is the social accounting matrix (SAM). The elaboration of a SAM generally relies to the country's national accounts. As we focus on transportation, an important part of the required information is not available from national accounts and we have to infer it from distinct sources. Another difficulty relates to the local dimension of the model. Economic accounts are available at the national level and we had to estimate local aggregates for the Paris region. The details of these processes are discussed in the subsequent sections.

The main structure of the model is summarized in Figure 1. The main actors are Households, Firms, the Government and transportation operators. Land developers are also considered in our model, but we do not include them in the illustration for the sake of clarity. Investors are included to account for the transport investment decisions, but they are part of the government in our model. Transport operators manage public transportation. Both households and firms use transportation facilities. They decide on the transport mode (private of public) to use and the trip timing (in a dynamic framework) as a function of their conditions.

Households (or consumers) maximize a utility function given their budget constraint. They have to allocate their time between work, which increasesincome, and leisure which directly increases utility. A typical household makes a first decision on how to use public and private transport. Congestion increases with the number of road users. It depends on the number ofhouseholdsusing private transport. Congestion reduces travel speed and increases travel distance. The generalized travel cost has a monetary part and a time part (opportunity cost of time spent in transportation). The number of users of public transport is assumed to determine the service frequency (Mohring Law) and the level of crowding in public transport. Note that, given the current situation of high crowding in main transit lines in Paris (RER A, RER B, etc.), it is unlikely that frequency decreases significantly. We should expect that the decrease in the number of passengers will mainly lead to a comfort improvement.

Firms, and in particular those providing services, are also concerned by passenger transportation. Time savings from better transportation supply can increase production since transport time is reduced. Our model is based on production functions that include transportation as an intermediate input. We make a distinction between economic sectors on the basis of their use of passenger transportation. For example, services rely more on the development of passenger transport than other sectors. The model we present here share some similarities with the model of Brocker and Mercenier (2012), but their paper does not proceed for an empirical analysis and remains limited to a microeconomic description of the model.

Transportationhas other indirect impacts on economic activities. First, easier market access generated by an improvement in transport benefits to some activities. For example, commercial centers, are more successful when they are well connected to public transportation, well connected to the highway network and when they offer sufficient parking space. In this case, the investment in public transportation is equivalent to an increase in demand.

Second, improved transportation can reduce frictions in the labor market. This line of research has been theoretically discussed in Zenou (2006), but very few empirical studies have been undertaken on the subject. When transportation opportunities expand, job search is more efficient since workers explore job opportunities on a wider geographical area. Matching in the labor market works better. Our model takes into account this impact and includes the impact of transport facilities on the job market.

In Figure 1, a transport operator is in charge of planning and managing activities in public transportation. Making this operator distinct from the government allows us to easily account for the financial flows and cost in public transportation. Transport activities are subject to some network effects. In particular, congestion increaseswith the number of road users and, when the loading of public transport vehicles is high, crowding appears (comfort quality decrease). Overall, the GPE project is expected to reduce both of these external costs.

Both households and firms benefit from transport services. As we have discussed above, households benefit from better accessibility and save time in commuting. Firms benefit from the reduction in the generalized transport cost(time savings for professional trips). Households sell their labor force to the firms. In return, the firms distribute revenues (wages, profits) to households. The main part of these revenues is used to buy goods to firms. The remaining part is savedand may be used for investments in the private sector. The government supervises transport policy and may impose constraints on operators.

Figure 1: a graphic representation of the model

3.The structure of the CGE model

3.1The Social Accounting Matrix

3.1.1The general structure of the SAM

As every CGE model, our model rests upon a social accounting matrix, the SAM of the Ile-de-France Région (SAMIF). The general structure of the SAMIF, presented in Table 1, is a standard one for a mono regional model. Each row of this squared matrix displays the revenues received by an account, while the corresponding column displays the expenditures for the same account. The matrix has five main types or accounts, for the factors of production, the economic agents, the activities, the products and theaccount forsavings and investment.

Table 1: Basic Structure of the SAM.

Receipts
Expenditures / Factors of production / Institutions / Activities / Products / Saving-Investment / Total
Factors of production / Value Added / Factor returns
Institutions / Revenues from factors supply / Transfers / Indirect taxes / Revenues of institutions
Activities / Production at factor cost / Production at factor cost
Products / Final consumption / Intermediate inputs / Investment / Absorption
Saving-Investment / Savings / Total savings
Total / Revenues from factors supply / Total expenditures / Production at factor cost / Production at market prices / Total Investment

The SAMIF has however several original features, linked to important aspects of the model we want to develop. First of all, it includes a description of activities, factors and outputs linked to passengers transport. This choice implies that, besides the standard sectors, we add an activity devoted to passengers transport. This activity uses private inputs (e.g. vehicles and labour) and public inputs (transportation infrastructure). Second, it includes a real estate sector, which combines land, labor and capital for producing housing and business premises.

Overall, there are 23 accounts in the SAMIF:four factors of production (labor, private capital, public capital, land), four institutions(households, firms, government, Rest of the world), seven activities (real estate, manufacturing, trade, passenger transportation, business services, non-market services, and other branches), seventypes of goods (one for each activity) and aSavings-Investment account for dynamic analysis.

To the best of our knowledge, this is the first SAM produced for the Ile-de-France Region. Therefore, the SAMIF has been built from scratch for the base year 2012, combining information from a large variety of sources[3]. When regional data were not available, we applied the Location Quotient approach (LQ) on the relevant national data using two allocators: the number of employees and the number of firms. The LQ approach is sometimes criticized, but it is still the most used method when the scarcity of regional data obliges to rest upon national data and it provides a good approximation.