Textbook of Computable General Equilibrium Modelling

Efficiency of resource allocation has become increasingly important in recent years. In the 1960s and 1970s, the Japanese economy grew so rapidly that efficiency of resource allocation was a secondary issue; however, this has changed. Growth in Japan slowed from the 1980s onward to the more moderate speeds experienced in other developed economies. Most developing economies cannot expect large aid inflows and commercial financing from developed economies any more. The focus of economists has returned to an important issue in economics; namely, efficiency of scarce resource allocation.

In this chapter, we develop the simplest possible CGE model, hereafter called the ‘simple CGE model’, with one household, two firms, two goods (bread and milk) and two factors (capital and labour).^1,2 As all the CGE models used for empirical policy analysis are extensions of this simple CGE model, it is vitally important to understand how the model is built and what implications can be derived from the equilibrium of the model.

Chapter 2 provided a mathematical presentation of the ‘simple CGE model’ with its theoretical underpinnings based in standard microeconomics. However, the general equilibrium of an economy cannot be ‘computed’ only with this mathematical model. The CGE modelling needs two more steps. The first step is to prepare a computer program of the CGE model, using GAMS syntax. This step is covered in this chapter. The second step is to construct a social accounting matrix (SAM) on the basis of empirical data and then to estimate coefficients and exogenous variables such as the share coefficient of the utility function ± _i and the factor endowments FF _h . This second step is discussed in Chapters 4 and 5.

In the ‘household utility maximization model’ presented in the last chapter, we used values of coefficients for the utility function and for the factor endowments, assumed on an a priori basis. However, to develop an empirical CGE model, various coefficients and exogenous variables of the model based on real data must be estimated. The estimation process consists of two steps. First, we collect data of the base year and construct a comprehensive and consistent macroeconomic database, called a social accounting matrix (SAM). Next, by using the SAM, we estimate the coefficients and exogenous variables of the model. The estimation method of the latter step is called ‘calibration’. This chapter explains first the structure of a SAM, then how to construct a SAM. The calibration procedure is explained in Chapter 5.

Chapter 4 detailed the construction of a social accounting matrix (SAM). This chapter is devoted to explaining the method of estimation of coefficients and exogenous variables in the CGE model, based on the SAM. In estimating these, we cannot apply standard econometric methods. This is because we usually have a limited number of observations but a large number of coefficients and exogenous variables to estimate in the CGE model. This leads to insufficient degrees of freedom if econometric methods are applied.¹ To overcome this problem, we employ a unique estimation method called ‘calibration’.

The ‘simple CGE model’ presented in Chapter 2 is equipped with only the essential features of a very basic macroeconomic model, and thus it cannot be used for empirical analyses. Here, we extend the model by incorporating the following four features. First, we introduce intermediate inputs into the production process. Second, we introduce a government into the model, where its consumption, and direct and indirect taxes revenues including import tariffs, are considered. Third, we also introduce investment and savings. Finally, we extend the model to an open economy model, where international trade is considered. We call this model the ‘standard CGE model’, whose social accounting matrix (SAM) was previously shown in Table 4.2 in Chapter 4.¹

Any empirical model depicts only part of the real economy; thus, it has a borderline distinguishing between the ‘inside’ of the model, which is represented by endogenous variables, and the ‘outside’, which is assumed to be given. In the ‘simple CGE model’ developed in Chapter 2, the supply of primary factors is assumed to be exogenously constrained by their endowments FF _h . Instead, their supply can be endogenized with a factor supply function. For example, we can incorporate consumption of time as leisure in the household utility function to generate voluntary unemployment or an inflow of foreign labour. In principle, we can freely set a borderline delineating the endogenous part from the exogenous part of the model, considering the purpose of the analysis with insights into the actual economic phenomena.

The purpose of developing CGE models is primarily for policy analysis and evaluation. We simulate various policy options to quantify their impact, and therefore we have to simulate various equilibria by solving the same model under several scenarios involving different assumptions regarding the coefficients and exogenous variables; then, the respective equilibria should be compared to evaluate the assumed policies. At the same time, the robustness of the simulation results must be examined to see whether the results are unreasonably affected by the calibrated parameters and those with the a priori determined values. This chapter consists of two parts. The first part presents an efficient way to conduct comparative statics for policy analyses, and the second part discusses sensitivity analysis in CGE modelling.

We have explained how to develop a CGE model, prepare simulation scenarios and run a simulation to analyse the impact of policy or other exogenous shocks on an economy. The simulation results in the output files show changes of various economic variables, such as sectoral output, consumption, prices and welfare, induced by such shocks. However, a mere listing of these changes does not constitute an ‘analysis’. Interpretation of the simulation results, although usually the final step, is the most important task in CGE analysis. A discussion of the simulation results can only be made persuasive by explaining them using intuitive reasoning. To do so, we have to understand how the model structure combined with the simulation scenarios leads to the numerical solutions.

Almost all the objectives of this book were accomplished by the end of Chapter 9. That is, we have developed a ‘standard CGE model’ and explained how to compute it with GAMS and interpret the model solutions. The model is equipped with minimum standard features for empirical analysis. By understanding the model structure and computer program, readers are expected to be able to develop and solve their own models. In this chapter, to model a more complicated economy we extend the standard CGE model by replacing some of its basic assumptions. This chapter is aimed mainly at advanced CGE modellers. In Section 10.1, we present a multihousehold model, in contrast to the simple and the standard CGE models, which have only one household. In Section 10.2, we relax the small-country assumption and develop a large-country model to describe the international economy better. In Section 10.3, we extend the one-country model to a two-country model with some suggestions for multicountry models. In Section 10.4, the assumption of perfect competition is replaced by that of monopoly and oligopoly. In Section 10.5, rather than import tariffs, we introduce a quantitative restriction on imports into the standard CGE model. The chapter concludes with Section 10.6, where economies of scale are considered.

Many innovations by pioneering CGE modellers and economists have made CGE models one of the indispensable empirical tools for macro-economic analysis. CGE models have been developed to analyse a wide range of policy issues. In addition to the minimal data requirements for model parameter estimation in view of the model size, the significant flexibility in incorporating various features useful for empirical analysis facilitates the widespread use of CGE models. In response to the increasing need to deal with impending policy issues, CGE models have been extended from basic versions to those better suited to modelling the complexities of a modern economy. Such extensions have been made in two directions. One is within the CGE model framework, and the other is outside of it.