تعریف سیاست های کارآمد در یک مدل تعادل عمومی: یک رویکرد چند هدفه

Macroeconomic policy makers are typically concerned with several indicators of economic performance. We thus propose to tackle the design of macroeconomic policy using Multicriteria Decision Making (MCDM) techniques. More specifically, we employ Multi-objective Programming (MP) to seek so-called efficient policies. The MP approach is combined with a computable general equilibrium (CGE) model. We chose use of a CGE model since it has the dual advantage of being consistent with standard economic theory while allowing one to measure the effect(s) of a specific policy with real data. Applying the proposed methodology to Spain (via the 1995 Social Accounting Matrix) we first quantified the trade-offs between two specific policy objectives: growth and inflation, when designing fiscal policy. We then constructed a frontier of efficient policies involving real growth and inflation. In doing so, we found that policy in 1995 Spain displayed some degree of inefficiency with respect to these two policy objectives. We then offer two sets of policy recommendations that, ostensibly, could have helped Spain at the time. The first deals with efficiency independent of the importance given to both growth and inflation by policy makers (we label this set: general policy recommendations). A second set depends on which policy objective is seen as more important by policy makers: increasing growth or controlling inflation (we label this one: objective-specific recommendations).

Macroeconomic policy makers are typically concerned with several indicators such as growth, inflation and unemployment rates, and the level of public deficit. In this sense, policy making can be viewed as a problem with several objectives, some of which may conflict with one another. For example, an active anti-unemployment policy could increase inflation; a greater domestic growth rate could be harmful to the balance of trade, and so on. (See refs. [28] and [30] and [36] for selected discussions and analyses of the multi-objective nature of policy making.) The well-known area of Multi-criteria Decision Making (MCDM) offers techniques designed to deal with problems in which there are multiple conflicting goals.1 It thus appears reasonable to tackle the design of macroeconomic policies using MCDM techniques. More specifically, we explore the use of Multi-objective Programming (MP),2 which specifically seeks so-called (Pareto) efficient solutions. For current purposes, we say that a policy is “efficient” if it is not possible to find an alternative that allows improvement in the value of some objectives without harming the value of others. Importantly, knowledge about which policies are efficient or inefficient is of real practical value since, if a given policy is known to be inefficient, guidance can be provided for improvement going forward. In order to operationalize our MCDM approach, an analytic representation of the economy under study is needed. In the current case, we use a computable general equilibrium (CGE) model. Such structures have been used extensively since the 1980s in the evaluation of public policies and other simulation exercises, both in developed and developing countries. (See refs. [4], [18], [24], [26] and [31], and [32] for selected recent applications of CGE models, and [22] for a discussion of the current state-of-the-art.) CGE modelling is especially attractive for policy makers since, being consistent with standard economic theory, it allows one to measure the effects of a specific change (e.g., a given policy) on the most significant economic variables such as prices, production levels, tax revenues, and income distribution. The principal contribution of the current paper is a methodological proposal for policy making that is both operational/practical and consistent with economic theory. Moreover, it combines two analytical tools that, to the best of our knowledge, have not previously been employed together: CGE modelling and MP. The approach can thus be used to design efficient policies and/or determine if any given (real or potential) policy is efficient or not. A second contribution of the current research is its application to a real economy. In Section 2, we present an application using the Spanish national Social Accounting Matrix (from 1995) in which growth and inflation are chosen as policy objectives. In Section 3, key results are presented: The trade-off between growth and inflation is assessed, an efficient policy frontier constructed, and the observed policy compared to this frontier. Selected recommendations are offered to improve the observed policy in terms of efficiency. In Section 4, some additional extensions and applications of our methodological proposal are presented. Section 5 concludes the paper, offering some guidelines for future research.

In this article, we have presented a methodological approach for the design of public policies accounting for the fact that policy makers are usually concerned with a variety of conflicting criteria. We thus address two types of readers: first, researchers interested in economics in general, and designing macroeconomic policies in particular; and, second, policy makers interested in an operational tool for the design of rational and practical macroeconomic policies. We thus offered general guidelines to model policy with several criteria (i.e., policy objectives) and provided a definition of an efficient policy. We then presented an application to illustrate the potential usefulness of our approach. A CGE model calibrated for the (1995) Spanish economy allowed us to quantify the trade-offs between real growth and inflation when designing fiscal policy, and to construct a frontier of efficient policies in terms of these two factors. The observed policy was tested in terms of efficiency with results showing that the combination of growth and inflation was strictly above the efficient frontier. We were thus able to conclude that the existing policy displayed some degree of inefficiency with respect to the two economic objectives. A key contribution of this paper is the application of the Paretian concept of efficiency to the field of policy design. Efficiency is, to be sure, a desirable property of macroeconomic policies, since any inefficient policy could be unambiguously improved in terms of a given set of objectives. This is a relevant issue in practice since, if the applied policy is, in fact, inefficient, we can argue that it should be re-oriented to ensure that resources are managed in a (more) rational way. As we have shown, multi-objective programming, employed in conjunction with a CGE model, appears to be a useful approach to identifying efficient policies in practice, and to assessing the efficiency of a given (real or hypothetical) policy. Based on our analyses, a key conclusion/recommendation is methodological in nature, and not particularly surprising: once policy objectives have been stated and understood, decision makers should implement an appropriate procedure to ensure that relevant instruments are used in an efficient manner. Our results illustrate how not doing so may lead to inefficient (and, hence, unsatisfactory) economic results. At the same time, the proposed model provides selected and specific recommendations for the case under study. These can be grouped in two categories: the first includes so-called general efficiency recommendations. These include changes to be made in fiscal policy for the sake of efficiency (in order to drive the economy to the efficient frontier) independent of any focus of policy makers, i.e., either increasing growth or reducing inflation. In our case, these recommendations include, first, increasing public expenditures in Sectors 5 (Chemicals) and 6 (Machinery and Transport) while reducing them in Sector 9 (Services). It is also recommended that taxation be tempered in traditional sectors such as agriculture, extractives, energy and food, but increased in more dynamic sectors such as machinery and transport, and construction. The second group of specific policy recommendations (objective-specific recommendations) depends on where the policy focuses, i.e., on increasing growth, or on controlling inflation. It is interesting to note that, for each of the analyzed policy-making problems (H and V), the first group includes 21 policy recommendations while the latter includes just eight. Conventional economic wisdom suggests that fostering growth and controlling inflation are two very different objectives and thus require different, even conflicting, policy measures. Nevertheless, our results suggest that there is significant overlap between fiscal policies needed to pursue both objectives in a general efficiency-seeking framework. The current article is the first in a line of research aimed at designing public policies with multiple criteria. In this regard, we have introduced selected extensions and applications of our approach, such as a procedure to reduce the set of eligible policies and alternative sets of policy objectives. Nevertheless, there remain additional extensions that could be addressed in future research. One possibility is the use of alternative multi-criteria methods. Although we used multi-objective programming (for reasons explained earlier), other MCDM techniques could be used, and done so within different problem context(s). For example, in applications with a large number of objectives, where it is not realistic to seek a global optimum, but, rather, to satisfy “reasonable” aspiration levels, goal programming might be a more suitable tool. As another possibility, interactive MCDM methods provide a way to “fine-tune” policies by incorporating the preferences of the policy makers once initial policy suggestions have been offered. At the same time, there are some meaningful ways to improve and enlarge the CGE model itself. A natural extension would be to address the dynamic aspects of the economy.