تجزیه و تحلیل تعادل عمومی قابل محاسبه مالیات صادرات در صنعت پشم استرالیا

We solve for Australia's optimal export tax on wool using a computable general equilibrium model — an aggregated version of the Monash model. A key aspect of the analysis is the way in which we model short-run and long-run comparative statics. As opposed to varying the Armington elasticity which measures the degree of substitutability between domestic and imported goods, we contrast the unrestricted movement of primary factors of production with a specific factors representation. We find that while results are virtually unchanged for the range of Armington elasticity values we employ in our sensitivity analysis, the specific factors specification has a significant impact on model results. In addition, we provide an explanation for why there are differences between our optimal export tax results and those generated by the Johnson [Johnson, H.G. (1965), Optimum Tariffs and Retaliation, in International Trade and Economic Growth, London: George Allen & Unwin Ltd., pp.31–61.] inverse elasticity formula. These results indicate that it is necessary to be cautious when interpreting optimal export tax estimates based on the Johnson inverse elasticity formula.

Despite no longer “Riding on the Sheep's Back” wool is still an important industry for Australia. In 2000–01 Australia exported $3.9 billion dollars worth of wool, almost 98% of production, second only to wheat in terms of magnitude of agricultural production. These exports accounted for 74% of world raw wool exports while the next largest was New Zealand with 15% (ABARE, 2001a and ABARE, 2001b). Given the importance of Australian wool production it is of little surprise to find that there is an extensive literature examining many aspects of the industry (e.g., Hinchy and Fisher, 1988, Bardsley, 1994 and Cashin and McDermott, 2002). However, for an industry with such a large share of world trade, and in principle the ability to exert market power, there is minimal research examining the adoption of an export tax for wool. This lack of attention is even more surprising as wool is not a homogeneous product. As Beare and Zwart (1990) explain, the physical characteristics of wool are important in determining end use. Wool from New Zealand is coarse and used in non-apparel production. Wool from Australia is much finer and appropriate for use in the production of apparel. So wool from Australia and New Zealand can be considered different products, providing even greater support for the argument that Australia could exert some market power. To date, the only papers that examine the effects of an export tax on wool are Alston and Mullen (1992) and Edwards (1997). Alston and Mullen do not directly consider the export tax. They are interested in how R&D in the wool industry should be funded. Alston and Mullen recognise that a wool tax levied on grower output to fund R&D is essentially tantamount to an export tax because almost all wool produced in Australia is exported. Edwards provides a more detailed review of the theory underpinning the adoption of an export tax for wool. Both papers infer the likely size of an optimal export tax for wool based on the approximation of Johnson (1965) and Cordon (1974). The first objective of our paper is to estimate the optimal export tax for wool in a comprehensive modelling framework. We employ a Computable General Equilibrium (CGE) model to examine an export tax on wool for Australia. Our research adds to the CGE literature that examines the economic implications of introducing an export tax for primary commodities (e.g., De Santis, 2000, Warr, 2001, Warr, 2002 and Wiig et al., 2001). However, unlike the existing literature we are considering the production of a primary commodity in a developed economy. Existing research on export taxes has typically focused on developing economies (for example, Bangladesh and the production of jute (Repetto, 1972, Hwang and Mai, 1999 and Ahammad and Fane, 2000)). Furthermore, we compare the optimal export taxes generated by our CGE model with those derived using the Johnson (1965) inverse elasticity formula. When other distortions exist (as in our CGE model), Johnson's inverse elasticity formula needs to be revised to account of the effect that the export tax on wool has on tax revenue derived from other distortions. We use the CGE model to identify those distortions in the dataset which cause our estimates of Australia's optimal export tax on wool to deviate from that predicted by Johnson's inverse elasticity formula. This adds to the literature (e.g., Rodrik, 1989 and Yilmaz, 2006) that has previously relaxed assumptions (i.e., domestic perfect competition and no retaliation by other countries) for the derivation of optimal export taxes. The second objective of this paper is to show how a specific factors modelling frame-work can be used to contrast short-run and long-run comparative static results in a CGE model. Typically, such comparisons between short and long-run results are made by varying the Armington elasticity which measures the degree of substitutability between domestic and imported goods. In our model, sensitivity analysis on the Armington elasticity shows that results are insensitive to different specifications of this parameter, which is in contrast to other research (e.g., Irwin, 2003). Instead we argue that it is more appropriate to model short and long-run behaviour by assuming that some share of primary factors in production is specific in the short-run, and then allow these specific factors to become perfectly mobile between sectors in the long-run. The use of the specific factors approach to examine differences between long and short-run is not uncommon in the literature (see Mayer, 1974, Mussa, 1974 and Schweinberger, 2002). Indeed, there is a growing body of empirical evidence to indicate the importance of specific factors (e.g., Magee, 1980, Grossman and Levinsohn, 1989 and Hiscox, 2002). However, what differentiates our paper from existing CGE research is the way in which we model specific factors. As Bhagwati and Srinivasan (1983) observe, the conventional all-factors-specific model (Haberler, 1950) and the one specific factor model developed by Jones (1971) are extreme cases “with few counterparts in reality” (p. 92). The approach we take is to model a particular share of labour and capital as being specific in the short-run, and we reduce this share until in the long-run we arrive at a specification equivalent to the Heckscher–Ohlin model. The paper is organized as follows. In Section 2 we review the various strands of literature that our paper draws upon and will add to. The CGE model and dataset, including a description of the industry and commodity aggregations and relevant model parameters, are described in Section 3. In Section 4 we describe results, including the level of the optimal tax on wool exports as a function of the elasticity of demand for wool by Australia's trading partners. Concluding comments and areas for further research are presented in Section 5.

In this paper we have modelled and examined the imposition of an optimal export tax for Australian wool. Since Australian wool exports account for almost 75% of world trade in wool, we model Australia as facing a less-than-perfectly elastic demand curve for wool. We use a CGE model and an aggregated version of the dataset in the Monash model to solve for Australia's optimal export tax, for different export demand elasticities for Australian wool. When the model is benchmarked to a rest-of-world demand elasticity consistent with estimates in the literature, an optimal tax on wool exports leads to static long-run welfare gains of over $1.4 billion 1994 Australian, slightly less than 0.34% of GDP. Even when the model is benchmarked to a much more elastic export demand elasticity, an optimal export tax of 40% would yield static long-run welfare gains of over $492 million 1994 Australian. In addition, our results contribute to the literature by explaining and identifying why we observe differences in optimal tax estimates using a CGE model and the Johnson inverse elasticity rule. As we demonstrate, the existence of various distortions in the CGE model can give rise to significant differences in the optimal export tax estimates presented. Furthermore, we have shown that these distortions can work in different directions depending on whether we are considering the short-run or long-run. Since wool is produced in three distinct zones in Australia, results of the model show how the export tax on wool affects different wool-producing regions. Wool production is most important in the Pastoral zone, which suffers a much larger decline in wool production under an optimal export tax policy than the Wheat sheep and High rainfall zones. The export tax on wool also gives a competitive advantage to domestic wool users. While these domestic value-adding industries in wool see a large increase in production in the short-run, these gains are much smaller in the long-run and as such provide little support for increased investment in wool processing. While short-run and long-run results in CGE models are typically contrasted by varying the Armington elasticity to which the model is benchmarked, our model results are insensitive to different specifications of the Armington elasticity. Instead, we assume that some share of primary factors in production are specific in the short-run, and model the long-run by allowing this share of specific factors to go to zero, so that results in the long-run are consistent with the traditional Heckscher–Ohlin model. What is apparent from the results we generate is that our specific factors specification does yield important differences in modelling results between the short and long-run. An important feature of our model is the choice of the share of primary factors assumed to be specific in the short-run. The literature gives no indication of how this share parameter should be chosen, so we set this parameter at 25% and acknowledge that there is no empirical evidence to support this choice. Since our results indicate that this specific factors approach is important in proxying short-run and long-run outcomes, further empirical research is warranted to provide guidance on appropriate choice of this important parameter.