Trading in the downstream European gas market: a successive oligopoly approach.

Second, we consider more traders throughout Europe. Table 5 shows O-ND equilibria when there are three (case "All-3") and nine ("All-9") traders active in each country. As the numbers of traders increase from one or two (O-ND, Table 4) to three and then nine (Table 5), and finally, in effect, infinity (PC-ND, Table 4), retail prices and trader profits decrease monotonically, while wholesale prices, social welfare, producer profit, and consumer surplus all increase. The decreases in residential and industrial retail prices (due to more competitive trading) are much greater than the increases in wholesale prices (resulting from higher production increasing the marginal cost of supply). When all nine traders are active in all eight countries, the results tend to the competitive trader outcomes, although traders still earn significant profits. Meanwhile, as might be anticipated, the less competitive case "All-3" yields price and throughput results roughly halfway between the competitive trader (PC-ND, Table 4) and monopoly/duopoly trader (O-ND, Table 4) cases.

8. INCOMPLETE MARKET OPENING

This section focuses on the effects of asymmetric market opening in Europe. We assume that selected countries (Austria, Belgium, France, Italy) will not open their gas market completely, i.e., households in those countries will stay captive. For these captive markets, prices are regulated and consumption is defined by (18) or (25b), where the 1995 consumption is taken as the constant in the latter equation (IEA, 1997). All other circumstances are the same as in Section 5, so the analysis is done for the benchmark and four alternative cases of market structure. This allows comparison with complete market opening (Table 4).

Table 6 shows that trader profits become positive with incomplete market opening, perfectly competitive traders and no price discrimination. This results from the use of equation (1) to calculate profits, as the difference between border and end use prices for captive sectors exceeds the assumed cost of within-country distribution d[c.sub.ng]. But it is not credible to assume that competitive traders would continue operating at a profit; a more reasonable scenario is that government regulators would alter regulated prices, taxes, or subsidies to avoid this outcome. For simplicity, we assume that this adjustment takes the form of some lump sum transfer (e.g., fixed customer charge or refund) that does not affect consumption.

Table 6 also shows the prices of natural gas in the captive markets. Incomplete market opening, compared to the cases with complete opening in Table 4, is advantageous for the consumers that stay captive when traders are oligopolistic. Prices for households in Austria, Belgium, France and Italy are 20-26% lower than in cases O-ND and O-D in Table 4. Other countries and industry and power generators in the four countries mentioned face slightly higher end-use prices because of the increase in marginal cost (0 to 1%). Lower prices result in 10-25% lower trader profits, while producer profits increase by 5-13%. (12) In contrast, in the case of competitive traders when no price discrimination is allowed (benchmark and PC-ND), captive customers face higher prices. Producer profit, consumer surplus, social welfare and production are somewhat lower. Results in the case of price discrimination combined with competitive traders (PC-D) are ambivalent.

9. DISCUSSION AND CONCLUSIONS

This paper describes the empirical model GASTALE and shows several illustrative analyses of the European gas market using this model. GASTALE extends and applies the successive oligopolist model of Greenhut and Ohta (1979) to a situation in which there are multiple consumer markets separated in space while upstream producers have nonlinear production costs. GASTALE makes an explicit distinction between upstream producers and downstream traders in the gas market. It is possible to simulate alternative strategies for producers and traders (oligopolistic or perfectly competitive). Liberalisation of the gas market can be examined with GASTALE in several ways: allowing consumer groups to be either eligible or captive; varying the assumed behaviour of traders between perfect competition and oligopoly; constraining price discrimination; and varying the number of traders.

A number of simplifications have been made in GASTALE that should be addressed in future work, as we discuss later. Nevertheless, the model is the first to explicitly address the sequential oligopoly nature of the European gas market. We present several sets of results that illustrate how the interactions of oligopoly in production and trade can affect market outcomes, although the model's simplifications imply that specific numerical results for particular sectors should be interpreted cautiously. Our model results show that as a result of our assumed linearity of within-country transmission tariffs (no scale economies), traders make no profits above a normal return to capital in a perfect competitive market. But if traders are oligopolistic, they make a profit and the level of this profit depends on the ability of producers to price discriminate at the border. End-use prices converge to prices corresponding with perfectly competitive trading when the number of traders increases.

Although it is often thought that vertical integration stimulates market power and puts the consumer at a disadvantage, the opposite might be true. Our results show that, given the oligopolistic structure of the upstream industry, it is important to prevent monopolistic/oligopolistic structures in the downstream gas market. As Tirole (1988) states: "What is worse than a monopoly? A chain of monopolies."

In general, the economic literature (Tirole, 1988) concludes that where there are both upstream and downstream oligopoly, vertical integration between upstream and downstream is favourable for consumers. Vertical integration prevents double marginalization, i.e., two successive mark-ups, and end-use prices would be lower. This suggests that in the case where monopolistic or oligopolistic competition between downstream gas companies cannot be prevented, vertical integration should be supported (or at least not be discouraged!). The conclusion is confirmed by the results of Section 5 in which a comparison was made between the behaviour of the competitive and oligopolistic traders. Case PC-D can be interpreted as representing the case of vertically integrated gas companies. In PC-D, producers set their border prices with the knowledge that the traders will not charge a second margin on the prices, consistent with our Stackelberg assumption. Therefore the most optimal end-use prices, from the point of view of producers, are set and their maximum profit is attained. (Alternatively, PC-D can be viewed as simulating a situation in which every producer integrates vertically by creating a trading operation in each country, and those operations displace the assumed independent traders.) In contrast, if independent traders form an oligopoly and there is no vertical integration (case O-D), the traders also set a margin on the end-use price. Consequently all end-use prices are higher, whereas consumer surplus and social welfare are lower compared to vertically integrated companies. (13) Considering these results, vertical integration indeed should not be discouraged in case oligopolies dominate the trading market. The best form of vertical integration would be to allow producers to enter national markets by forming their own trading operations. This possibility should be simulated in future work.

Our model has several limitations that should be addressed in future research. First, price and welfare effects depend on the assumed elasticities. For now, our sensitivity analyses show that the main conclusions concerning the undesirability of successive oligopoly are unaffected by variations in elasticities. However, the magnitude of the effects and their distribution among different consuming sectors are impacted. Therefore, better price elasticity estimates are needed for a more disaggregated set of consuming sectors. For instance, market models for electricity (e.g., the power module in PRIMES (Capros et al., 2000)) could be used to obtain that sector's elasticity for gas, considering how gas competes with other boiler fuels.

Second, we have incomplete information about new TPA tariffs. Most countries are still developing TPA tariff structures and they are not (yet) public. To the extent that those tariffs depend on load and distance, it may be desirable to further divide consuming sectors by customer size and location. Third, price discrimination is incorporated at the level of producers, i.e., on the border prices. The traders are still allowed to discriminate between end-consumers, which they do. However, partial arbitrage (for instance among industrial and generation customers) could mitigate that discrimination, and could be simulated in GASTALE. (14)