In a world in which economies are linked by international trade and capital flows, abatement by one economy induces spillover effects and has welfare impacts on other economies. It matters to understand the conditions under which both abating and non-abating economies will experience positive or negative impacts from the policy adopted in other groups of countries; it matters also to understand the results of these spillover effects in terms of carbon leakage. Chapter 7 provides the basic concepts of such an analysis and here some brief comments are added to explain the strengths and weaknesses on the results found by modelling exercises.

In static terms, without international capital mobility, the welfare costs of abatement for an open economy can be decomposed into two components (Dixit and Norman, 1984):

• costs that would be incurred if the economy were closed; and
  
• changes in the terms of trade, which are the first transmission mechanism for spillover effects (see Chapter 7).

If they require to go beyond “no regrets” potentials, binding emissions constraint comes to increasing the cost of carbon-intensive products and, if emissions arise from the production of its export goods, the abating economy benefits from better terms of trade. If, indeed, the importing economy cannot produce a perfect substitute easily, it will sell the same product at a higher price and increase its purchasing power of imported goods. The non-abating economy will symmetrically suffer a welfare loss because of more expensive imports, while the net result for the abating economy depends on the size of improvement in the terms of trade relative to the production costs of abatement. The welfare impacts are more important in the economies that are very dependent on foreign trade.

In the real world, an emission constraint simultaneously affects both export and import goods, but this does change the nature of the mechanism. Increased production of emission-intensive goods in non-Annex I regions is stimulated by both increased non-Annex I consumption and increased exports to Annex I regions. The net relative balance between these parameters is influenced by the extent to which Annex I emission constraints fall on export competing industries (when the country is specialized in such industries) as opposed to import-competing industries (when it imports carbon intensive goods). If a constraint predominantly affects export industries, it encourages increased non-Annex I production for internal consumption. If the constraint predominantly affects import-competing industries, increased non-Annex I production is mainly exported to Annex I regions. Emissions leakage is beneficial to non-Annex I economies only in the second case, since it is associated with an improvement in their terms of trade, whereas their terms of trade deteriorate in the first case.

Another factor that affects the increase of emission-intensive goods in non-Annex I regions is the effect of Annex I abatement on the intermediate demand for fossil fuels. As discussed above, Annex I abatement will reduce fossil fuel prices. Lower prices for fossil fuels will encourage the production of more emission-intensive goods and the use of more emission-intensive production techniques in non-Annex I regions ¹⁵.

So far, it was assumed that changes in the production structures in both Annex I and Non-Annex I countries result only in changes in final demand and in price structures. The introduction of international capital mobility complicates the analysis since, in addition to production costs and changes in the terms of trade, carbon constraints alter the relative rates of return in the abating and non-abating countries. If capital flows from the first country to the second in response to these changes, there will be a further restriction of the production frontier (the set of possible productive combinations) for the abating economy and an outward shift for the other economy. Factor rewards in both countries are also affected. Part of the income from foreign investment accrues to the home economy and subtracts from income in the foreign economy; abating economies are affected by changes in income and factor prices that result from changes in international capital flows, with symmetric gains for non-abating economies.

No theoretical results for complex and empirically relevant cases can be obtained as to the extent that international capital mobility modifies the conclusions of the static analysis of the role of the trade effects. However, modelling results are seldom reported on the welfare impact of changes in international capital flows, although McKibbin et al. (1999) emphasize the macroeconomic repercussions. It is still, indeed, impossible to derive clear conclusions about the role of these changes, because of the methodological difficulties in interpreting the results from complex CGE models. It is usually conceded that modelling international capital flows is one of the more contentious issues; technically indeed, such a modelling relies on equalizing rates of return on capital across countries, but, because this makes capital flows too reactive, various “ad hoc” devices are used to obtain less irrealistic outcomes. Differences in the riskiness of rates of returns are clearly relevant to explain most of the real behaviours, but how this can be “best” dealt with in a deterministic model is an open question. Progress depends on the further development of techniques¹⁶. It depends also on progress in theoretical and empirical analyses to capture more effectively how the exchange rate of currencies reacts to external payment deficits. This depends on the level of confidence on the future economic expansion of each country and how monetary policies (including the determination of the public discount rate) employed to mitigate adverse impacts can change the return to capital in a country relative to other countries.

Models reviewed in this section have in common features that must be clearly borne in mind when interpreting the results:

• They assume perfect competition in all industries.
  
• Most of them use the so-called Armington specification that identical goods produced in different countries are imperfect substitutes: it is known that the results may then be sensitive to the particular commodity and chosen regional aggregation models (Lloyd, 1994).
  
• All of the models, apart from the G-cubed model of McKibbin and Wilcoxen (1995), are long-term growth models with international trade, without explicitly modelled financial markets that affect the macroeconomic adjustment¹⁷.
  
• Emission reductions involve only carbon dioxide¹⁸.
  
• The bias in technological change is unaffected by the emissions constraints and the production possibilities frontier always lies below the unconstrained frontier. Under such a hypothesis, the aggregate impact is unlikely to be positive, but some economies may benefit from favourable changes in their terms of trade and from changes in international capital flows.

Simulation studies covered in this report were conducted prior to and after the negotiation of the Kyoto Protocol. Pre-Kyoto studies consider more stringent emissions reduction targets for Annex I regions than the average 5.2% actually adopted under the Protocol. The major findings are that Annex B abatement would result in welfare losses for most non-Annex I regions under the more stringent targets. The magnitude of these losses is reduced under the less stringent Kyoto targets. Some non-Annex I regions that would experience a welfare loss under the more stringent targets experience a mild welfare gain under the less stringent targets.

Studies using a variety of more stringent pre-Kyoto targets include Coppel and Lee (1995; the GREEN model), Jacoby et al. (1997; the EPPA model), Brown et al.1997b) and Donovan et al (1997; the GTEM model), and Harrison and Rutherford (1999; the IIAM model). The last two models are based on the Global Trade Analysis Project (GTAP) database (Hertel, 1997).

In these studies, most non-Annex I countries suffer deterioration in their terms of trade and also welfare losses. Since the analysis at the region or country level depends on the type of aggregation, it is difficult to give a comprehensive list of exceptions. The reasons for these exceptions are, however, easy to explain. Brazil and South Korea are, in many models, found to enjoy welfare gains from Annex I abatement policies because, unlike other non-Annex I regions, they are net importers of fossil fuels and have a high relative dependence on exports of iron and steel and non-ferrous metal products. In addition, in Brazil these products are far less intensive in fossil energy than in many other economies¹⁹. Brazil gains from lower prices for fossil fuel imports and higher prices for exports of iron and steel and non-ferrous metal products. Conversely, non-Annex I regions with the greatest dependence on fossil fuel exports, such as the Middle East and Indonesia, suffer the greatest deterioration. Non-Annex I regions that are net importers of manufacture goods that are fossil-fuel intensive also suffer a deterioration even if they benefit from lower oil prices.

One of the most important conclusions is that a number of those among non-Annex I regions that experienced a welfare loss under the pre-Kyoto targets experience a welfare gain under the Kyoto targets. For example, in the GREEN model, India and the Dynamic Asian Economies experienced a loss in real income in the pre-Kyoto simulation (Coppel and Lee, 1995). They experience a mild gain in real income under simulations of the Kyoto Protocol that involve varying degrees of policy co-ordination among the non-Annex I regions (van der Mensbrugghe, 1998). In pre-Kyoto simulations of the GTEM model (Brown et al., 1997b; Donovan et al., 1997), Chinese Taipei, India, Brazil, and the Rest of America were all found to experience welfare losses; with Kyoto targets (Tulpulé et al., 1999) these regions experience mild welfare gains.

There is one key reason why some regions that experienced welfare losses under the more stringent targets experience mild gains in welfare under the Kyoto targets: the changing balance between substitution and output reduction with the level of abatement. GDP losses or the required level of a carbon tax for Annex I regions are, indeed, an increasing function of the level of abatement and the milder Kyoto targets are expected to be achieved with a greater reliance on substitution relative to output reduction than the more stringent targets.

A fairly similar regional pattern of non-Annex I welfare changes is found in simulations of Kyoto targets in a number of studies in which comparable pre-Kyoto target simulations are not available. These studies include Kainuma et al. (1999; the AIM model drawing on the GTAP database), McKibbin et al. (1999; the G-Cubed model), Bernstein et al. 1999; MS-MRT, drawing on the GTAP database), and Brown et al., (1999; the multigas (CO₂, CH₄, and NO_x) version of GTEM) and Böhringer and Rutherford (2001).

IPCC Homepage