آزمون تئوری انتشار تجاری : شواهد تجربی بر مکانیزم های جایگزین برای تجارت جهانی کربن

Simulation models and theory prove that emission trading converges to market equilibrium. This paper sets out to test these results using experimental economics. Three experiments are conducted for the six largest carbon emitting industrialized regions. Two experiments use auctions, the first a single bid auction and the second a Walrasian auction. The third relies on bilateral, sequential trading. The paper finds that, in line with the standard theory, both auctions and bilateral, sequential trading capture a significant part (88% to 99%) of the potential cost savings of emission trading. As expected from trade theory, all experiments show that the market price converges (although not fully) to the market equilibrium price. In contrast to the theory, the results also suggest that not every country might gain from trading. In both the bilateral trading experiment and the Walrasian auction, one country actually is worse off with trade. In particular bilateral, sequential trading leads to a distribution of gains significantly different from the competitive market outcome. This is due to speculative behavior, imperfect foresight and market power.

There is general agreement among environmental economists that emission trading is well suited to restrict the emissions of a uniformly dispersed pollutant such as CO2 emissions (Tietenberg, 1985 and Klaassen, 1996). The idea that international emission trading between parties creates flexibility, thus allowing lower total abatement costs, has been the basis for incorporating emissions trading in the Kyoto protocol. Numerous publications have calculated the expected cost savings of international carbon trading, taking into account different market structures and restrictions on trade (see Yamin et al. (2000) for a survey). The common characteristic of these simulation studies is that they use a static framework and assume market equilibrium will be realized. Assuming well-behaved emission abatement cost functions, initial grandfathering of emission permits and price taking behavior of trading parties, the proof that market equilibrium exists is relatively straightforward (Montgomery, 1972). However, it leaves the question unanswered whether starting from disequilibrium, and in a context where bilateral trades are made sequentially at changing non-equilibrium prices, the market indeed converges to an equilibrium. Recently Ermoliev et al. (2000) tackled this problem and proved that sequential bilateral trading of emissions converges to a cost minimizing emissions market equilibrium. The trading scheme only requires that sources bilaterally state their demand price (as buyers) and supply price (as sellers) and agree on price and quantity in bilateral negotiations and contracts. This reflects the behavior in existing markets, where brokers serve the important function of bringing together the relevant information. If emission reductions are implemented immediately by sources after each round of trade, they are tied up into sunk cost. These sources may not be willing to participate in subsequent rounds of trades, which may require reversing earlier decisions on emission levels. If so, the final result will not be cost-efficient. Therefore, the dynamic market model, as specified by Ermoliev et al. (2000) assumes reversibility before irreversible real decisions are taken, by separating the price formation stage from the actual implementation stage. This problem of irreversibility can be overcome by leasing permits for a limited period rather than auctioning them in perpetuity. An alternative to a decentralised search for the cost-effective vector of emissions is a Walrasian auction of emission permits (Ermoliev et al., 2000). Such an auction also results in minimizing total emission control cost. The auction consists of two stages: a search stage, in which the auctioneer searches for the equilibrium price; and a second stage, in which transactions are made at the market equilibrium prices. The search stage is opened by the auctioneer who announces the permit price. Each participant then knows the price to pay for an extra unit of emission reduction or to receive for each additional unit of emission reduction. If the price is higher than the participant's current marginal cost, he will decide to reduce his emissions below his initial emissions level and sell permits. He will inform the auctioneer how many emission permits he wants to sell. On the other hand when the emission price is below the marginal cost of emission reduction, the source has an incentive to increase its planned emissions; it will state its demand for emission permits. The auctioneer then calculates market demand and revises the price upwards when demand exceeds supply and downwards when supply exceeds demand. It should be noted that sources have to state only the number of permits they are willing to sell or to buy at the price proposed by the auctioneer. Only the auctioneer has knowledge of demand and supply for permits. The procedure converges to an equilibrium with cost-efficient emissions. The above theoretical contributions to the dynamics of emission markets may be logically correct, that does not mean they are also true. The real world can differ from what theory predicts. The purpose of this article is to test the theory using the methods of experimental economics and apply these to international carbon trading under the Kyoto Protocol. Although our experiments deal with market adjustments over a series of iterations, we do not explicitly analyze banking or borrowing (in contrast to, e.g. Rubin (1996) and Stevens and Rose, 2002). Neither do we discuss anything other than the emission reduction commitments for the first compliance period of the Kyoto Protocol (2008–2012). Analyzing proposed policies in which critical institutional design issues confront policy makers is an obvious task for experimental methods (Smith, 1994 and Issac and Holt, 1999). Experimental methods have been used to examine policy proposals ranging from electricity rate design to scheduling cargo regimes on the space shuttle (Banks et al., 1989). Following Montgomery's (1972) seminal theoretical work on emissions trading, experiments on mechanisms for trading emissions permits began with the work of Plott (1983). With the passage of the Clean Air Act Amendments of 1990 by the U.S. Congress, experimenters had a real market to help design. Funded by the U.S. Department of Energy, teams at the University of Colorado and University of Arizona began work on auction mechanisms and permit allocation schemes which eventually led to the U.S. Environmental Protection Agency (EPA) sponsored auction for sulfur dioxide (Bjornstad et al., 1999). Contributors to Issac and Holt's (1999) edited volume on emissions permits experiments addressed potential problems for the U.S. sulfur dioxide program including speculation (Cason et al., 1999), market power (Godby, 1999) and permit banking (Cronshaw and Kruse, 1999a and Cronshaw and Kruse, 1999b). With the adoption of the Kyoto Protocol in 1997 the potential for trading greenhouse gases has now taken on an international dimension. Our carbon emissions trading experiments differ from the existing literature (Bohm, 1997, Bohm and Carlen, 1999, Baron and Cremades, 1999 and Soberg, 2000) and laboratory evidence on emission quota trade in the following ways: • Information on source pollution control costs is not regarded to be common knowledge, not even within a certain range of uncertainty; sources have only private information on their own cost function. • Both bilateral sequential trading and auction mechanisms are tested. • Our experiments are an empirical test of the theory; the other publications mainly report results of experiments. An exception to the above is recent work by Bohm and Carlen (2002). They also test the theory using carbon trading experiments. Their paper focuses on two mechanisms (additional emission quota and financial transfers) to induce developing countries to participate in carbon trading. The emphasis of our paper is, however, on the theory and experiments regarding the organization of trading via auctions and bilateral transfers. The structure of the paper is as follows. Section 2 describes the design of the experiment. Section 3 presents the results and Section 4 concludes.

This paper set out to examine the efficiency of bilateral, sequential trading and trading using auctions so as to validate the conjectures of economic theory in a laboratory context. For this purpose, three experiments were run: one using bilateral, sequential trading and two using auctions. The paper finds that, in line with the standard theory, both the auctions and the bilateral, sequential trading (with regular information on average prices) are able to capture a significant part (88 to 99%) of the potential cost savings of emission trading (see Table 10). In contrast to the usual findings, the results also suggest that not every country may gain from trading. Under the bilateral trading regime, each participant realizes substantial savings on control cost, but for some parties total expenditure turned out to be higher than when emission trading is prohibited. This is due to combination of imperfect information, speculative behavior and market power. The losses for one party are reflected in windfall profits for others. Activities such as speculation, collusion and cheating played a role especially in the bilateral trade experiment and in this respect behavior differed from the assumptions in Ermoliev et al. (2000). Yet the prediction that bilateral trade will converge to the cost-effective allocation is supported by the experimental evidence. The single bid auction appears not only to fulfil an important role in achieving the potential efficiency gains from emissions trading but also serves to distribute the gains from emissions trading so that they are closer to the expected equilibrium outcome. The Walrasian auction shows that although the market-clearing price was found quite quickly calculation errors, leading to non-cost minimizing behavior by individual participants, affect the actual total savings on abatement cost and can make parties worse off than they would have been without trading. In spite of this, the distribution of the gains in the Walrasian auction differs less from the perfect equilibrium than the distribution under bilateral, sequential trading. In the theoretical model of Ermoliev et al. (2000), it was assumed that all parties calculate correctly and do not try to affect the market price and therefore the market-clearing price is always the efficient price.Two points of discussion seem to be relevant. First, the fact that the distribution of the gains from bilateral trading differs more from the perfect equilibrium outcome than under the auctions might be relatively robust. Test runs of the bilateral trading scheme showed that the winner gained twice as much as the perfect market solution and the country with the smallest gains gained only 40% of the possible gains. In this case, only 76% of the potential gains were realized. This suggests that the auctions might not only distribute gains more evenly but may also be more efficient. Secondly, market power was also exhibited in the test runs. Russia, CEE and the Ukraine colluded and initially, agreed not to sell below $125/ton C. This strengthens our conclusion that market power is an important factor to take into account when designing the carbon market. We do point out, however, that in reality the number of sellers and buyers might be much larger than in our experiment especially if developing countries such as China and India participate, or permits are traded between companies rather than countries. Thus we may be might exaggerating the threat of market power. The absence of the USA might even increase market power on the demand since the numbers of buyers (EU, Japan, Canada, New Zealand) are fewer and could act as a block. However, in practice this does not seem to occur. We do not think that this affects our major finding that both auctions and bilateral trading are likely to be efficient in practice and that the distribution of the gains might differ significantly from the perfect equilibrium outcome, especially under bilateral trading, due to market power, speculation and imperfect foresight. Our findings indicate that two elements might be crucial for the actual implementation and design of carbon trading schemes. First, auctions not only stimulate efficiency but also equity because price transparency is created. Second, market power is potentially a significant threat; so efforts to limit it by increasing participation of more (developing) countries (cf. Bohm and Carlen, 2002), or individual firms, are relevant for a successful practical implementation.