داده های تفکیکی و تجزیه و تحلیل سیاست های تجاری: ارزش ایجاد ارتباط بین مدل تعادل جزئی و کلی

Computable General Equilibrium (CGE) models are now routinely utilized for the evaluation of trade policy reforms, yet they are typically quite highly aggregated, which limits their usefulness to trade negotiators who are often interested in impacts at the tariff line. On the other hand, Partial Equilibrium (PE) models, which are typically used for analysis at disaggregate levels, deprive the researcher of the benefits of an economy-wide analysis, which is required to examine the overall impact of broad-based trade policy reforms. Therefore, a PE–GE, nested modeling framework has the prospect of offering an ideal tool for trade policy analysis. In this paper, we develop a PE model that captures international trade, domestic consumption and output, using Constant Elasticity of Transformation (CET) and Constant Elasticity of Substitution (CES) structures, market clearing conditions and price linkages, nested within the standard GTAP model. In particular, we extend the welfare decomposition of Huff and Hertel (2001) to this PE–GE model in order to contrast the sources of welfare gain in PE and GE analyses. To illustrate the usefulness of this model, we examine the contentious issue of tariff liberalization in the Indian auto sector, using PE, GE and PE–GE models. Both the PE and PE–GE models show that the imports of motorcycles and automobiles change drastically with both unilateral and bilateral tariff liberalization by India, but the PE model does a poor job predicting the overall size and price level in the industry, post-liberalization. On the other hand, the GE model overestimates substitution between regional suppliers due to “false competition” and underestimates the welfare gain, due to the problem of tariff averaging in the aggregated model. These findings are shown to be robust to wide variation in model parameters. We conclude that the linked model is superior to both the GE and PE counterparts.

Examination of the impacts of tariff changes at a disaggregated level is important for many reasons. First, there are huge variations in tariff rates at different tariff lines for many commodities, causing serious aggregation bias in aggregate sector-based studies.1 Second, the aggregation of sectors may result in ‘false competition’: two countries that do not compete in a third market at the disaggregated level (e.g., one exports engine blocks and one auto transmissions), may appear as competitors at an aggregate (auto parts) level.2 Third, many policies are framed for specific products that are not identified among the relatively aggregated sectors. Finally, most trade policy negotiations are conducted at highly disaggregated “tariff lines”,3 which is why there has been a strong preference for Partial Equilibrium (PE) analysis (e.g. Ramos et al., 2007 and Evans et al., 2007) as negotiations begin to get seriously under way.4 As Lloyd and MacLaren (2004) note, the inability to support disaggregate analysis is a major shortcoming of CGE models. On the other hand, while comprehensive PE models may show approximate welfare measures for small exogenous changes (Kokoski and Smith, 1987), they are unable to offer a comprehensive assessment of the impact of trade policy reforms on economy-wide welfare, wages, employment and other variables of interest to policy makers. These policy indicators are typically produced by Computable General Equilibrium (CGE) models (Francois et. al., 2005; Jensen and Sandrey, 2006). This paper utilizes a combined PE–GE approach to trade policy analysis, thereby permitting us to contrast the PE, GE and combined PE–GE predictions for bilateral trade flows as well as aggregate welfare and other economy-wide variables of interest. One of the first attempts to perform a somewhat disaggregated analysis in a CGE framework was made by Basevi, 1966 and Basevi, 1968. However, there has been a surge in research efforts aimed at linking PE and CGE models only recently. This has been common in the poverty/micro-simulation literature (Herault, 2007 and Hertel et al., 2007b), in sub-regional economic modeling (Madsen and Jensen-Butler, 2004) as well as in the application of econometrics to CGE models (Han and Woodland, 2003, Hertel et al., 2007a, Bhattarai and Whalley, 1999, Arndt et al., 2002 and McKitrick, 1998). More recently, authors have begun to link CGE and PE models for disaggregated trade policy analysis. In particular, Grant et al. (2007) have proposed a partial/general equilibrium (PE/GE) framework, building on the GTAP-in-GAMS global CGE model (Rutherford and Paltsev, 2000) and focusing on the treatment of tariff rate quotas, which cannot readily be aggregated for use in a normal CGE model. Our paper draws inspiration from this work; likewise implementing a PE/GE model within the GTAP modeling framework (Hertel, 1997). We focus this paper on determining the impacts of tariff liberalization in India's automotive industry. This is an apt example for several reasons. Firstly, this is a diverse sector, not only structurally, but also in terms of the wide tariff variations across its sub-sectors.5 Secondly, India has been actively pursuing different policies for different sub-sectors of the auto industry.6 As a consequence there has been policy-driven structural change in the Indian auto industry over the years.7 Thirdly, the ongoing tariff negotiations in India are sub-sector-specific, necessitating a framework wherein tariff simulations could be done at sub-sector-level. Since the late 1990s, India has been negotiating trade agreements, covering various sectors, with East and South-East Asian countries, which are both competitors and partners in the global market for autos.8 There is a widespread concern that the domestic auto sector is very sensitive to liberalization.9 However, the government of India has been cutting auto tariffs, arguing that past tariff cuts have benefitted the industry in terms of better competitiveness, growth and employment (Ministry of Heavy Industries, 2006). So, tariff liberalization in this sector is a contentious issue. Further, the debate over auto sector reforms is relevant in a global context as well, with the potential for India to emerge as a global auto production hub as well as a consumer market. The latter is being fuelled by a rapidly growing middle-class, improved access to finance and a very low vehicle penetration ratio.10 Many studies have recently assessed the impacts of FTAs being negotiated by India within a CGE framework (Weerahawa and Meilke, 2007, Kumar and Saini, 2007 and Kawai and Wignaraja, 2007). However, none of them have utilized the kind of PE/GE framework offered by Grant et al. (2007) and developed in this paper. Using a three-region, ten-sector database derived from the GTAP 6.2 (Dimaranan, 2006), MAcMap (International Trade Center, 2006 and Bouët et al., 2004) and TASTE (Horridge and Laborde, 2008) databases, we compare the results of complete tariff liberalization in the Indian auto industry, using PE/GE, PE and GE models. We find that in addition to the differences in aggregate results, both the PE and the PE/GE model show strikingly diverse results across the sub-sectors of the auto industry, which cannot be captured by the GE models. However, our simple PE model does a poor job of predicting the changes in the size and price level of the industry. Although this could be improved by building a more complex PE model, that would still not capture economy-wide effect, which is the focus of this study. Thus, we find that the PE/GE model is superior to the GE model in terms of disaggregated impact-evaluation and to the PE model in terms of endogenous determination of aggregate supply and demand. More importantly, the PE/GE model shows lower welfare loss, higher allocative efficiency gains and lower terms of trade losses, because the GE model ignores disaggregated details of trade flows and tariffs. Apart from being among the first pieces of work developing PE/GE model to perform disaggregate analysis after Grant et al. (2007), this work uniquely contributes to the existing literature in other ways. Firstly, this is the first paper to extend welfare decomposition to the PE/GE framework. Secondly, this implementation is done in the widely-used GTAP framework, thereby aiding this large community of users in performing similar analyses, extending CGE to the tariff line. Thirdly, our auto industry example effectively shows evidence on issues such as false competition involved in the standard GE model, due to aggregation issues. Finally, the unique comparisons of the results done across the PE, PE/GE and GE models in this exercise highlight the different shortcomings of PE and GE frameworks compared to our proposed PE/GE framework. This paper is organized as follows: Section 2 outlines the modeling framework and methodology. Section 3 discusses the data sources. Section 4 summarizes the results and Section 5 concludes.

In an attempt to achieve the best of both worlds — specificity to sub-sectors and economy-wide analysis, we find the PE–GE model to be a valuable contribution. With a PE–GE model that incorporates price linkages, CET and CES nests and market clearing equations, in addition to some linking features that provide the much-needed blend with economy-wide treatment, we arrive at much more plausible results than we do using either isolated PE or GE models. (Given the lack of empirical support for the CET and CES elasticities utilized in our PE–GE model, we carried out a Systematic Sensitivity Analysis by varying them and found that all our results are clearly robust to a very broad plausible range of these elasticities.) Our PE–GE model is superior to the simple PE model, because it endogenously determines aggregate supply and demand in an economy-wide framework, which is suitable for providing inputs to the real-world trade negotiations. It is also superior to a comparable standard GE model, because it captures structural details at sub-sector-level and gives the modeler a rich set of information related to the impacts of the simulations, at disaggregated sub-sector-levels, which is again much closer to real-world situations where the policies are specific to such sub-sectors. Further, even at an aggregate level, PE–GE can yield more realistic results as it is free from the problems arising from working with the aggregate sectors. As an illustration, our results from complete tariff liberalization in Indian auto industry show two major inferences that further strengthen the evidence for the usefulness of this approach: Firstly, the PE model in isolation shows far lower changes in imports and far higher changes in prices, although it still captures, to a large extent, the disaggregate impacts in sub-sectors. Secondly, the stand-alone GE model shows lower changes in the aggregate imports and does not provide any information about the sub-sectors, which is found to be very crucial in the PE and PE–GE models, in terms of the heavy influx of imports in the automobiles and motorcycles sector from South-East and East Asian economies. Further, the substitution effect appears more pronounced in the GE model, because of ‘false competition’, as the exporters do not actually compete in the sub-sector-level as much as it appears from the aggregate level. The GE model shows just the aggregate change which is not very high and may give a misleading signal to the policy-maker that complete tariff liberalization in Indian auto sector may cause significant but not drastic changes. Thus, on both counts, PE–GE model clearly emerges as the preferred framework to address a policy issue that relies much upon the sub-sectors, which also have an economy-wide relevance. Furthermore, we extend the welfare decomposition of Huff and Hertel (2001) to the PE/GE model in order to investigate the sources of welfare gain. Comparing the welfare changes with those in the GE model, we find the overall welfare gains to be higher in the PE–GE model. There are many other notable differences, of which we highlight those in the import tax-related AE effects and the TOT effects. In both cases, all the differences could be traced back to the changes in the disaggregated model that result in different sets of changes in prices and quantities. This further illustrates the usefulness of PE/GE models for policy analysis, as welfare analysis is a very policy-relevant tool offered by CGE models.