Oligopoly Behaviour in the Trans-Tasman Air Travel Market: The Case of Kiwi International.
The Trans-Tasman air route carries more than a million return passengers between Australia and New Zealand every year. For decades, it has been dominated by the duopoly of national carriers, Air New Zealand and Qantas, but for a period of just over a year in 1995 and 1996 this settled situation was disrupted by competition from a small new operator, Kiwi International, which, having started as a charter operator, began offering scheduled services in August 1995 and exited into liquidation just over one year later, in September 1996. The period of Kiwi's presence was marked by bouts of substantial price cutting by all the competitors, and the introduction by Air New Zealand of a subsidiary airline, Freedom Air, to compete directly with Kiwi.
The responses of the incumbents were widely regarded at the time by the public as an attempt to oust Kiwi from the market, with many expecting both Freedom Air and the low prices to disappear as soon as Kiwi International collapsed. Prices have indeed risen, but Freedom remains in the market, and it has become apparent that Kiwi had suffered from other difficulties which had contributed to its collapse, including poor financial and strategic management. Thus there is reasonable uncertainty about how to regard the events that unfolded through Kiwi's brief history. That is, just how should the pricing and behaviour of the main incumbents in response to Kiwi's entry be interpreted? Were Air New Zealand and Qantas merely aggressive competitors with which Kiwi just was not efficient enough to compete, or had they engaged in predatory behaviour with the aim of ejecting a troublesome new entrant? NZ's competition authority, the Commerce Commission, investigated a complaint by Mr Ewan Wilson, the former CEO of Kiwi, against Air New Zealand's actions, but dismissed this without considering the matter of predation on the grounds that the incumbent airline is not `dominant' in this market. (Under the weak NZ competition law, it is in effect assumed that no firm which was not dominant would in fact have the power to predate.)
No doubt Air New Zealand is not dominant, but the duopoly it shares with Qantas probably is, and the interesting economic issue is whether the duopoly predated Kiwi out of the market. The traditional method for considering this has relied on comparisons of price charged with appropriate measures of costs. But apart from measurement difficulties, simple price/cost comparisons do not uniquely identify the motives of the players. It could be the entrant that has forced the issue and is predating, or it could just be that the presence of an additional competitor changes the oligopoly outcome by this amount even with no change in the underlying motivations of the existing firms.
We need to dig below the surface of the pricing data and investigate how they were generated. Such is the innovation developed in this paper, which, in essence, seeks first to identify the oligopolistic behaviour that generated the pre-entry price structure, and then asks whether behaviour was markedly different in the period of fiercest competition that preceded the exit of Kiwi International.
The paper is set out as follows. The next section is a narrative of the events preceding and following Kiwi's entry. Section 3 reviews the analytical issues involved with standard tests of predatory pricing. Section 4 develops data for pricing and costs in the Trans-Tasman air market, and applies these to the standard tests. Then, in Section 5, we write down, calibrate, and simulate the oligopoly model. Section 6 discusses the results, and Section 7 concludes the paper.
2. The Trans-Tasman Air Market and the Kiwi International Saga
The Trans-Tasman market is defined as the air routes that connect cities in both islands of New Zealand with the major cities on the East Coast of Australia: Brisbane, Melbourne and Sydney. The market is dominated by Air New Zealand and Qantas, who in 1996 had an estimated combined market share of 89%, with Air New Zealand at 47%, and Qantas 42%. The remainder of the market is served by a fringe of foreign carriers, some of which are very large airlines in their own right, but none of which have more than a very small share of the Trans-Tasman market, which they serve as a relatively unimportant add-on to their long-haul Trans-Pacific flights. For calendar year 1996 the two new operators Freedom and Kiwi had about 3% of the market each, which Kiwi achieved over the nine months that it flew in that year.
So-called `fifth freedom' rights for a foreign airline to pick up passengers at a point not in that airline's home country, and to take them to a destination also not at home are not granted automatically and almost always involve a reciprocal concession from the foreign airline's government. Under various treaties and agreements, any NZ or Australian airline(1) can fly between the two countries without restrictions on destinations or capacity. Thus entry is relatively easy for local airlines, but difficult for third-country based carriers.
In August 1995 Kiwi Travel International Airlines (later Kiwi International Airlines), with capital of $778,888, became New Zealand's second international carrier, following a profitable year as a charter operator across the Tasman. Initially Kiwi flew to Sydney and Brisbane only from the small regional cities of Hamilton and Dunedin, which had not previously been served by scheduled international flights, using a leased 173 seat Boeing 727-200. But it rapidly expanded and at its height was flying from Dunedin, Christchurch, Hamilton and Auckland to Brisbane, Sydney, Melbourne and Perth in Australia, with a total of 30 flights a week and using 2 aircraft (737-300 and Airbus A320).
The impact on prices can be seen on Table 1, which shows fares in the months that they were changed. The general impression in the media at the time of Kiwi's launch as a scheduled operator was that it was a low cost, low price budget operator, offering no-frills `nuts and cola' return tickets for as low as $399. But a closer examination of its initial pricing practice reveals a slightly different picture. At entry it was charging $678 to Brisbane, with none of the $399 seats actually available on this route because it had a 90% load factor at the standard rate. The Sydney flights were more difficult to fill and therefore Kiwi offered the $399 seats in limited numbers in order to increase the load factor on these flights. The standard fare to Sydney, while $100 below Air New Zealand's lowest, was still $549, and 70% of passengers were paying this rate.
Table 1. Prices in the Market August 1995-November 1998 Date Aug-95 Nov-95 Dec-95 Mar-95 Jun-96 Event Kiwi Freedom enters enters Air NZ/Qantas Standard Fare High Season: Brisbane 849 499 Sydney 749 499 Low Season: Brisbane 749 Sydney 649 Special Fare Brisbane 559 399 Sydney 449 399 Freedom Standard Fare Brisbane 549 Sydney 449 Special Fare Brisbane 495 349 Sydney 395 299 Kiwi Standard Fare Brisbane 678 649 549 Sydney 549 549 449 Brisbane 449 349 Date Sep-96 Dec-96 Nov-98 Event Kiwi exits Air NZ/Qantas Standard Fare High Season: Brisbane 699/799 659 Sydney 699/799 659 Low Season: Brisbane 599/699 559 Sydney 599/699 559 Special Fare Brisbane 499 Sydney 499 Freedom Standard Fare Brisbane 499/599 459 Sydney 499/599 459 Special Fare Brisbane 449 Sydney 449 Kiwi Standard Fare Brisbane Sydney Brisbane
In December 1995 competition intensified with the entry of Freedom Air, a `new' operator run by Air New Zealand subsidiary Mount Cook Airlines, with standard Kiwi and Freedom fares of $549 and $449 for Brisbane and Sydney, and `specials' below that. Furthermore, Air New Zealand itself and Qantas began to offer very cheap fares with their lowest being $449.
Initially Freedom, using a wet-leased Boeing 757 and flying from Auckland, Hamilton, Wellington, Christchurch and Dunedin, with 2 flights per week from each, achieved poor load factors (around 40%). These improved when Freedom changed its schedule in March 1996 to concentrate solely on Palmerston North, Dunedin and Hamilton, the latter two cities being Kiwi's key home markets. It was also in March 1996 that Freedom began a more aggressive pricing strategy with new specials of $299 and $349 to Sydney and Brisbane respectively. By April 1996 Kiwi had been forced to drop its standard rate to Sydney to $429 from its original $549.
The catalyst for the final blow in the price war seems to have been Kiwi's decision to expand into Auckland and Christchurch, which prompted Qantas in June of 1996 to offer fares as low as $399 to the entire East Coast of Australia (Wilson, 1996). This was immediately matched by Air New Zealand. At this time $399 fares were sold very quickly, but nearly everybody who missed out on the $399 fare was able to get a return ticket for $499. In effect $499 was the Air New Zealand and Qantas standard discount economy airfare to the East Coast of Australia. This compares with $749/$649 (High Season/Low Season) to Sydney and $849/$749 to Brisbane prior to Kiwi's entry.
In addition to the price war Kiwi had equipment failures and systems problems (eg no electronic link between accounting and reservations system), and in September 1996 went into voluntary liquidation after losing up to $8 million in passenger and creditor money. Following its demise, prices rose, as they did in 1993 when the aggressive price competitor Continental Airlines left the market.
In 1998 prices fell back. The `low' season fare of $559 covered all but school holidays and about three weeks around Christmas (when the fare was $659), and advance purchase requirements were reduced from 21 days, as they used to be, to almost zero. The primary reason appears to have been aggressive capacity expansion by both major airlines, who no longer had any code sharing arrangements(2). By May 1999, the fares had edged up to $599 to Sydney and $619 to Melbourne and Brisbane. Freedom Air remains as an operator serving the regional NZ cities, very much on a `niche' basis. Their flights do not show on computer reservation systems, and the airline direct markets through an 0800 number.
3. Standard Tests for Predatory Pricing
Did Air New Zealand (including its wholly-owned subsidiaries) and/or Qantas engage in predatory behaviour to drive Kiwi International from the market? In this section we first discuss and then apply standard price/cost based tests for predatory behaviour.
3.1 The theory of predation
The basic notion of predatory pricing is that of a dominant firm cutting its price to such a low level that it drives smaller rivals out of the market and/or deters new entrants. This action will impose short-term losses on the predating firm but these will be exceeded by the gains that it anticipates it will earn from the increase in its market power following the exit of its rivals. The dominant firm may be able to outlast its rivals because of its larger capital resources, or, if it is a multi-market player whereas its rivals are not, because of its ability to cross-subsidise from one market to another.
A number of authors have suggested methods of detecting predatory pricing and regulating it. The first and probably best known of these are Areeda and Turner (1975) who concentrate on short run profit maximisation. They argue that a firm that is selling at a short run profit maximising price or loss minimising price is not a predator, and that a necessary but not sufficient condition for predatory pricing is deliberate sacrificing of short run profits. They choose marginal cost (MC) as the dividing line between predatory and non-predatory prices. If price is below marginal cost then the firm is making a loss on at least part of its output. It could eliminate this loss by decreasing output, or, if its price is below average variable cost (AVC), by shutting down the firm. In the case of excess capacity -- where MC is below AC (average total cost) -- they argue that a price below AC but above MC should be tolerated even though it could drive an equally efficient rival from the market where that rival has insufficient capital reserves. Their reasons are that a rule which required price to be not only higher than MC but also some level above MC in order to be considered not predatory would permit the survival of less efficient rivals. Furthermore, in the short run, entry even by an equally efficient rival will be undesirable because excess capacity already exists.
On the other hand, if price is below MC it should be presumed that it is predatory. At a price below MC the firm is wasting social resources and the probability of a more efficient rival being eliminated or failing to enter is much greater. The possible exception is that such pricing could be used for promotional purposes, especially where strong brand loyalty exists or in order for a firm with declining costs to move to a more efficient level of output. When costs are increasing with output, a price below MC but above AC should be tolerated even if predatory in intent, because then more efficient rivals will be making supernormal profits and therefore will remain in the industry.
Due to the practical difficulties of estimating marginal cost, Areeda and Turner suggest that average variable cost (AVC) be used as a proxy. Thus the Areeda - Turner rule which has been widely used in United States jurisprudence (Van Roy (1991)) is that a price is predatory if it is below reasonably anticipated average variable cost. The authors contend that where MC is below AVC the rule is stricter, but that this is the correct test because a firm selling below AVC is not loss minimising as it would be better off shutting down. If MC is greater than AVC then using the proxy will allow a firm to price below MC, but the authors do not see this as a problem because if MC is greater than AVC the firm is likely to be at its capacity constraint, with the result that predatory behaviour will be unlikely as it will be difficult for the predator to satisfy any new demand.
Baumol (1996) agrees with the use of the Areeda-Turner rule, and argues further that it should be the theoretical rule as well. His main argument is that if a predatory price is defined as one that will drive an equally or more efficient rival from the market, then AVC is actually a superior standard to MC.(3) This is because while a price below MC will not always drive an at least equally efficient rival from the market -- for example where MC exceeds AC -- if price is below AVC then it is always rational for the firm to shut down.
Joskow and Klevorick (1979) argue that prices below average total cost can also be predatory. They justify this on the grounds that such prices can eliminate equally or more efficient rivals, and also that a firm with market power will only price below AC and sustain losses if it expects to earn monopoly profits in the long run which make this worthwhile. The Canadian Bureau of Competition Policy (1992) adopts an intermediate position, from which prices above AVC but below AC are treated as falling in a grey range in which determination of predatory purpose depends on the circumstances. For example, such a price may be regarded as not predatory if demand is declining or there is substantial excess capacity in the market, whereas it may be considered predatory if the firm was failing to raise prices above this level despite increasing demand.
Measuring average total costs is in practice even more difficult -- and thus likely to be controversial -- than calculating AVC, and so Joskow and Klevorick propose a two-stage procedure. In the first stage, structural characteristics of the market and the market power of the predator are examined in order to determine whether there is reasonable expectation that predatory pricing could occur. If there is no serious monopoly problem then investigation of predatory pricing should cease. If there are no barriers to entry it is unlikely that predatory pricing is taking place because the opportunity to recover losses through earning monopoly rents is non-existent.
But if the market is judged conducive to predatory pricing -- that is, there is a strong possibility that the predator will be able to earn monopoly rents -- then the second stage involves examining the behaviour through price-cost comparisons to determine whether the pricing is predatory. Their aim is to make the probability of false positive and false negative errors low when the cost of such errors is high. For example, the greater the market power possessed by the predator, the greater the cost of a false negative, or the less elastic the demand, the lower the cost of a false positive error.
The Joskow/Klevorick approach can be seen as a move towards a more sophisticated analysis of oligopolistic market structure and behaviour in possibly predatory situations. In this spirit we can also place the proposals of Williamson (1977) and Baumol (1979). Williamson proposes an output restriction rule which states that in the period after entry occurs the dominant firm cannot increase output above the pre-entry level for a period of 12 to 18 months. This is designed to take into account the effect that rules have on the strategic behaviour of firms. Using an example, Williamson shows that the output restriction rule has superior welfare consequences to the Areeda-Turner cost rule if strategic responses are taken into consideration, although this has been disputed (Lefever (1981); Williamson (1981))
Baumol suggested a restriction rule directed towards price. Under his `quasipermanent price reduction rule', firms would be free to cut prices in response to new competition, but they would not be permitted to put them up again for a certain period of time if the entrant leaves the market. Thus firms are free to respond to entry but they must ensure that their price will still cover their costs.
Even theoretically quite simple rules are likely to be problematic in practical application, and for this and other reasons -- especially, the difficulty in distinguishing predatory from worthily competitive behaviour -- some economists and anti-trust specialists believe that is unwise to attempt to proscribe it. McGee (1980) contends that predatory pricing will be rare because it will usually be irrational -- the predator will not recoup its losses. He argues that it is generally rational for the victim to `stick it out' if a predatory campaign occurs because any predatory pricing strategy is characterised by only a temporary cut in prices. Furthermore (Easterbrook, 1981), the victim has the same incentive as the predator to outlast its rival and collect the eventual monopoly rents.
Both McGee and Easterbrook argue that the `deep pockets' argument of predator being able to sustain losses for longer than its victim is not valid. The predator will be sustaining larger losses than the victim because its output will be larger, and the victim should have similar access to finance. It may also be able to call on its customers for assistance, given that these will benefit from its continued participation in the market, in the form of long term agreements to buy at the tree competitive price. But the difficulty faced by Kiwi International in obtaining financial capital prior to its demise may be an example of how even a firm with strong customer loyalty and a successful track record, but in trouble due to competitive pressure, may not be able to convince anyone to provide it with the resources it needs.
3.2 Applying standard tests
We will examine the pricing and cost behaviour of Air New Zealand's `fighting brand' subsidiary, Freedom Air, to apply the standard tests of predation.(4)
3.2.1 Calculating price and cost
`Price' is not a simple concept in air travel, because seats are usually sold at a number of different prices. We will use the total revenue per passenger kilometre (RPPK) as our measure of effective price. Freedom sold approximately 30% of their aircrafts' capacity at the `special' price of $395 Auckland-Sydney return, which is 9.2 cents/kilometer for the 4316km round trip, and we assume that all these seats were sold before standard fares are purchased. When the load factor was 40%, 75% of the seats sold will be have been at the lowest price with the remaining 25%, or 10% of the aircraft, at the standard price. Thus at Freedom's December 1995 prices to Sydney, overall RPPK at a load factor of 40% is 0.75*9.2 + 0.25*10.4 = 9.5 cents.
In 1996, load factors increased but prices fell. Our estimates of RPPK for March and May are 8.3 cents and 9.1 cents. With increases in both fares and load factors after Kiwi's exit, the December 1996 RPPK is calculated at 11.8 cents.
Freedom Air's operating costs are not separated out in its parent company's annual accounts, so that we must use Air New Zealand cost data as a proxy. First, Air New Zealand cost of international operations is estimated by assuming that the proportion of total cost attributable to providing international passenger service is equal to the proportion of total revenue provided by international passengers. This figure -- $1621 million for 1997 -- divided by the number of revenue-earning passenger kilometres carried, which was 18,440 million that year, gives a figure of 8.8 cents as the cost per passenger kilometre.
Most of Air New Zealand's international network has longer stage lengths than the Trans-Tasman route, and cost/kilometre falls as per-flight fixed costs, such as passenger processing and heavier fuel use in takeoffs are spread over longer journeys. We used (1994) Bureau of Transport and Communications Economics (BTCE) data on costs differences by length of stage to arrive at a scale factor of 1.25 for the Tasman route. That is, we estimate that, holding the load factor constant, it cost 11 cents per passenger kilometre travelled to move people between Australia and New Zealand.
Load factors (ratio of seats sold to seats available on a flight), of course, are not held constant in reality, and they have a large effect on unit costs, since only about 15% of the cost of providing a flight varies with the number of passengers (mostly food, drink; some fuel expenses). We will take this percentage to hold at Air New Zealand's average 1997 load factor of 68%, and assume that the variable costs are linearly related to the number of paying passengers, in order to adjust costs for different observed load factors.
Two time periods will be studied: the 6 month short run and the long run. The shorter the time period, the less likely it is that pricing will be found to be predatory, because the proportion of costs that are variable decreases with the time period. The 6 month short run scenario was chosen in order to approximate the period over which Freedom and Kiwi were in competition with each other.
In the extreme short run -- i.e. if the airline decides not to fly today -- the only costs that can be varied are aircraft fuel, passenger services, and landing and associated charges. In the 6 month short run, the airline can vary a great deal more, including all its flight operations costs and at least some of its maintenance costs. We estimate that 72% of total costs are variable within six months, and use this to compute estimates of Variable and Total costs per passenger kilometre for the load factors achieved by Freedom Air.
3.2.2 The comparison of price and costs
Table 2 shows prices and costs at the load factors achieved by Freedom Air in each month. Price fell below short run costs for the December to April 1996 period, and below total CPPK for all the months in which Kiwi and Freedom coexisted. In the winter months before Kiwi exited in September 1996, price seems to have been above six-month average variable costs, despite the price war, and quite substantial profit margins were earned, even on total costs, in the fourth quarter of 1996.
Table 2. Price and Cost Comparison by Load Factor and Scenario 6 Month SR Long Run Date Load Factor(%) Total RPPK VCPPK CPPK Dec-95 40 9.5 12.4 17.6 Mar-96 50 8.3 10.2 14.4 May-96 70 8.9 7.7 10.8 Dec-96 73 11.8 7.5 10.4
So do we have evidence of predatory pricing? The main reason for price being below cost in the early months is Freedom's very low load factor, which in turn could be attributed to the usual difficulties faced by new businesses in building up their clientele (Freedom, unusually for an alleged predator, is a quasi-`new' firm itself). By May 1996, load factors had built up enough to probably pull variable costs below price
But it does seem that price was below average total cost, CPPK, for the entire period in which Kiwi and Freedom were competing. Although it may have been rational for Kiwi to stay in the market if it was at least as efficient as Freedom because it could cover at least some of its fixed costs, Kiwi's lack of financial capital meant that it did not have this choice open to it. Although Air New Zealand, through Freedom, may not have engaged in predatory pricing according to the generally acceptable Areeda-Turner rule, its behaviour could have driven an equally efficient rival from the market with detrimental effects on the level of competition and overall welfare in the market.
This calls into question the approach of relying entirely on the Areeda-Turner rule because it could lead to the exit of an equally or more efficient rival which is not a desirable result. One alternative would be to have a price floor at average total cost, where a price below this might be predatory but not necessarily. In order for such a price to be regarded as predatory two conditions would have to be met, firstly, the victim would have to demonstrate it was at least as efficient as the predator but unable to sustain losses over an extended period and therefore could not choose the rational action. Secondly, the predator would be unable to show that it had priced below AC because of falling market demand or for promotional purposes. This would ensure that only equally or more efficient rivals were protected despite the price floor being above AVC, and that the predator still had the freedom to pursue actions which were aligned with the furtherance of long term competition and efficiency, such as using promotional pricing to enter a market leading to greater competition in the long run. Such an approach would be more difficult to implement and is more subjective than the Areeda-Turner rule, but the alternative under Areeda-Turner involves accepting that more efficient rivals could be driven out of the market.
If this approach is used in the Kiwi case then the first requirement that price is below average cost is satisfied, as is, clearly, the requirement of not having the financial resources to sustain losses for an extended period. As for costs, it is probably reasonable to assume that Kiwi was at least as cost-efficient as Freedom, because they were both operating 1 or 2 leased aircraft of similar type, and were flying the same routes. The Commerce Commission found that Kiwi had lower operational costs than Air New Zealand (Commerce Commission, 1997). We note too that Kiwi's load factors were always higher than those achieved by Freedom Air when the two airlines were in competition.
The final requirement is that Freedom could not justify pricing below average cost on grounds such as promotional market-building, or falling market demand. Overall, market demand was strong, but what about promotion? Freedom's initial load factors were low, but from May 1996 when a more efficient load factor was achieved the promotion justification could no longer apply. Freedom had secured itself a niche in the market but continued to price below average cost.
Thus we conclude that Freedom was probably not engaging in predatory pricing according to the orthodox Areeda-Turner rule, but that on the basis of the alternative approach set out above, its behaviour may have driven an equally efficient rival from the market.
4. Oligopoly Behaviour in the Trans-Tasman Air Travel Market
At the very least, the preceding section reveals the difficulties in resolving issues of alleged predatory behaviour with the orthodox price/cost comparison methodology. The basic problem is in knowing what to expect when structure changes in the absence of a model of oligopoly behaviour. Could more `competitive' pricing post-entry just be a natural consequence of a more `competitive' industry structure? There are, for example, well documented instances of price cuts of around 50% in US city-pair routes entered by the low-cost carrier Southwest Airlines which have not resulted in charges of predatory behaviour (Dresner et al, 1996). In this section, we attempt to move the analysis forward by interpreting before- and after-entry behaviour in the context of an explicit oligopoly model.
Our approach is as follows. First we use data and our knowledge of the industry to calibrate an oligopoly model to fit pre-entry behaviour. Then we similarly analyse the pre-exit period of intense competition. If the nature of the oligopoly game being played appears to have changed significantly after entry in the direction of more aggressive behaviour, we infer predation.
We believe that this approach represents a useful step forward from price/cost rules, but it suffers itself from at least two difficulties. The first (shared by all applied oligopoly calibration exercises) is the damage done to reality by the simplifications needed for a tractable analysis. The second problem is our lack of a well established theory of the determinants of oligopoly behaviour. Suppose we established that a settled duopoly could be modeled as Cournot-Nash and that so too could the triopoly following entry.(5) Then we could reasonably infer that post-entry price reductions were not predatory in intent (even if large enough so that revenues no longer covered incumbents' total costs), but were just what would be expected from the addition of another player to the Cournot setting.
But what if, post-entry, the nature of the oligopoly game changes? Since we might reasonably expect that conjectures become more competitive(6) as the number of firms increases (as we do in the limit of very large numbers of competitors) how can we ascertain whether the change reflects a conscious effort to behave more aggressively by incumbents? We will not be able to settle this matter, but will return to it in our discussion of results.
4.1 The Model
We make the usual tractability assumptions of linearity and constant costs. For demand, we assume that the incumbents Air New Zealand and Qantas are offering an effectively identical product, which is both horizontally and vertically differentiated from the service supplied by the entrant Kiwi International. We fold the operations of Air NZ's subsidiary Freedom Air into its parent.
That is, we will be modelling first a homogeneous duopoly, and then a triopoly with a mix of homogeneity and heterogeneity of product. The incumbents have consistently responded quickly to each other's price changes, such that they offer identical fare schedules, which is consistent with the existence of a rather broad margin of customers to whom the services offered by the two airlines are perfectly interchangeable -- no price differential can be sustained. Kiwi, however, with its focus on small regional markets, offered a distinctively different product (horizontal differentiation), which overall the market perceived as inferior (vertically differentiated) to the incumbents', as demonstrated by the entrant's price and market share being both smaller.
We write the price-dependent demand curves for incumbents (I) and entrant (E):
(1) [P.sub.I] = a - bQ + [eq.sub.E]
(2) [P.sub.E] = [Alpha] - [Beta]Q + [Epsilon][Q.sub.I]
where: Q = [Q.sub.I] + [q.sub.E] = [q.sub.i] + [q.sub.j] + [q.sub.E],
using i and j to subscript the two incumbents. The e and [Epsilon] coefficients measure the extent of horizontal product differentiation. If, say, e = b, then E's product is completely independent of I's in the marketplace -- they are not at all substitutes, because changes in [q.sub.E] have no impact at all on [P.sub.I]. If, at the other extreme, e = 0, then the products are perfect substitutes.
Total cost of firm i is:
(3) [C.sub.i] = [f.sub.i] + [c.sub.i][q.sub.i],
where [f.sub.i] are firm i's fixed costs, and [c.sub.i] is its marginal cost.
Incumbent firm i's profit function is:
(4) [[Pi].sub.i] = [q.sub.i][P.sub.i] - Ci = [q.sub.i] [a - b[Q.sub.I] - (b-e)[q.sub.E]] - [f.sub.i] - [c.sub.i][q.sub.i]
Differentiating with respect to firm i's output:
(5) d[[Pi].sub.i]/d[q.sub.i] = a - b[q.sub.i] d[Q.sub.I]/d[q.sub.i] - b[Q.sub.I] - (b-e)[q.sub.i] d[q.sub.E]/d[q.sub.i] - (b-e)[q.sub.E] - [c.sub.i]
We will be assuming [c.sub.i] = [c.sub.j] = c, which will give us symmetry in incumbent outputs:
[q.sub.i] = [q.sub.j] = q
Substituting into (5), and writing d[Q.sub.I]/d[q.sub.i] = (1+[[Lambda].sub.I], and [dq.sub.E]/[dq.sub.i] = [[Lambda].sub.IE] for the response expected by an incumbent firm of the changes in output of the other incumbent and of the entrant induced by a unit change in its own output:
(6) d[[Pi].sub.i]/d[q.sub.i] = a - 3bq - bq[[Lambda].sub.I] - (b-e)[q.sub.E] - (b-e)q [[Lambda].sub.IE] - c
These response parameters have traditionally been known as `conjectural variation' parameters, though more recently the term `competitive response' has become fashionable. We will use the older terminology here.
Then similarly, for the entrant E:
(7) [[Pi].sub.E] = [q.sub.E][[Alpha] - [Beta][q.sub.E] - ([Beta] - [Epsilon]) [Q.sub.I]] - [f.sub.E] - [c.sub.E][q.sub.E]
Differentiating and substituting as for (6):
(8) d[[Pi].sub.E]/[dq.sub.E] = [Alpha] - 2[Beta][q.sub.E] - 2([Beta] - [Epsilon])q ([Beta] - [Epsilon])[q.sub.E][[Lambda].sub.E] - [c.sub.E],
where [[Lambda].sub.E] is the entrant's conjectured response of total incumbent output to a unit change in its own output.
We assume that at any time observed industry outcomes were generated by profit maximising behaviour in a market equilibrium, so that we can set (6) and (8) equal to zero and solve them together to find the conjectural variations parameters as functions of observed outputs and the demand and cost coefficients. We have one lambda too many, and will need to assume a relationship between an incumbent firm's expectations about the response of the other incumbent and the response of the entrant:
(9) [[Lambda].sub.IE] = [Theta][[Lambda].sub.I]
With this, we get expressions for [[Lambda].sub.I] and [[Lambda].sub.E]:
(10) [[Lambda].sub.I] = [a - 3bq - (b-e)[q.sub.E] - c]/ [(b + (b-e)[Theta]) q]
(11) [[Lambda].sub.E] = [[Alpha] - 2[Beta][q.sub.E] - 2([Beta] - [Epsilon]) - [c.sub.E]]/[([Beta] - [Epsilon])[q.sub.E]]
For the pre-entry period we just solve (10), setting [q.sub.E] and [Theta] equal to zero. Post-entry, we solve for both conjectural variation parameters.
We need to make a number of additional simplifications to squeeze this oligopoly model into the template formed by the realities of the Trans-Tasman air travel market. Such simplifications are always needed, and they always represent a tradeoff between empirical accuracy and analytical tractability. Our procedures are in line with those used by other applied oligopoly analysts, for example, in the airline industry context, Brander and Zhang (1990).
First, we define the product to be a Trans-Tasman return leisure trip, and measure this by numbers of travellers paying discounted fares. This assumes that differences in the origin and destination of flights are not significant to the analysis. This simplification is probably not problematic -- the airlines themselves seem to act as though all New Zealand cities are the same distance from all Eastern Australian cities, with a standard add-on to Brisbane fares. However, there is usually some difference in prices ex-Australia and ex-NZ, which travellers cannot easily arbitrage away because of the return-flight requirement imposed on purchases of cheap tickets.
We will ignore the activities of all other airlines. The non-Australasian fringe has a small market share (about 9% in 1994), is capacity constrained, and none of its members covers all the main Trans-Tasman routes, with most of them confined to a single city pair -- for example, the largest fringe member, United, only flies the Auckland-Melbourne route.
We have lumped Freedom Air in with its owner Air New Zealand, and assumed symmetry between Air New Zealand and Qantas. The first of these assumptions is motivated solely by the desire for analytical tractability. Certainly, in a proper competition policy investigation the role of Freedom Air would be subjected to close scrutiny. By limiting ourselves in effect to the pricing activities of the two large airlines, we are being conservative from the point of view of identifying predatory behaviour, since the creation of Freedom may itself have been a predatory act.
As for symmetry, this goes slightly against the fact of Air New Zealand's larger share of the Trans-Tasman market. But we suggest that this is not due to any fundamental cost or product superiority of the New Zealand airline over Qantas, but simply because (a) each airline has a natural advantage in picking up custom from its own nationals (if only because of the flow-on from its internal feeder network), and (b) rather more New Zealanders want to travel to Australia than Australians to NZ.(7)
For `price' we use the discounted economy return fare from Auckland to Sydney -- that is, the fare below the full economy fare which carries conditions such as advance purchase and limited refundability. Pre-entry in late 1995, we take the simple average of high and low season fares, which is $699, but during 1996 the airlines offered portfolios of fares, with differing availability, and so our prices are calculated as share-weighted averages of the posted fares: $483 for Air New Zealand (with Freedom Air) and Qantas; $354 for Kiwi Air. For quantity, we use the number of leisure or non-business flights, which was 82% of the total in 1995. The output of the non-Australasian carriers is subtracted from the total. Our (annualized) outputs of return Trans-Tasman flights are 445,000 and 535,000 for each incumbent pre- and post-entry, and 52,000 for Kiwi.
To solve the functions (10) and (11) we need to assign values to the parameters of the demand curves (1) and (2) and the cost functions (3). For the incumbent price function (1), we solve for parameters a and b using actual pre-entry output and price and an estimate of the market demand elasticity from within the range suggested by our literature survey (Haugh and Hazledine (1999), Appendix 2). Then we can add the post-entry data for [P.sub.I], [Q.sub.I] and [q.sub.E] to solve for the `cross-price' parameter e, assuming that the demand curve did not shift between the two periods. The parameter e must lie in the range (0,b). If it is larger than b, this implies that the entry of Kiwi expanded the market so much that the demand for the incumbents' services actually shifted out. If it is less than zero, then Kiwi's arrival has actually contracted the total Trans-Tasman market. While neither of these eventualities is totally inconceivable, they must be considered very unlikely.
The range of (pre-entry) market price elasticities of demand that keep e within this range is approximately (-0.85, -0.65). This implies market demand slightly less elastic than the literature suggests is normal for international leisure travel. Such is probably quite reasonable for the relatively short Trans-Tasman route, for which fares are below those the airlines charge for many of their domestic flights, and it covers the econometric estimate for this route by BTCE (1995). We will use an elasticity of -0.75 for our base case, and show the sensitivity of results to changes in this. We will also show results based on the assumption of some total market growth due to population and/or income increases, which we model as reducing the slope, b, of the demand curve (1), while keeping the intercept, a, constant.
Next we calibrate the entrant's price function, (2). We have three parameters and only one data-generated equation (the situation post-entry), so two more assumptions are needed. The first of these is:
d[P.sub.I]/[dq.sub.E] = d[P.sub.E]/d[Q.sub.I],
which means that small cross-price effects cancel out -- the customers lost to the other airline by an own-price increase will return if the other airline matches the price change. This gives us:
(12) [Beta] - [Epsilon] = b - d
The second piece of `information' that we feed into the system is the price at which Kiwi International would lose (virtually) all its customers, given the actual post-entry output of the two large airlines. Kiwi's disadvantages -- limited schedule, no Frequent Flier Program, frugal service, no history of carrying passengers without crashing -- are such that it is unlikely it would even be able to even match the incumbents' fare and attract significant custom, given that even its `horizontal' niche advantage of specialised location was negated by the presence alongside it of Freedom Air in the regional markets. We will try a range of zero-demand price assumptions, starting from $450.
This price and [q.sub.E] = 0, together with ([Beta] - [Epsilon]), allow us to solve (2) for [Alpha], and then we plug in the actual values of output and price to get [Beta].
Finally, we need estimates of marginal cost. For the incumbents we use our estimate of the variable cost per passenger kilometre (VCPPK) on the 6 month short run scenario, at a 69% load factor, which was the approximate average for Air New Zealand and Qantas over the 1994-1996 period, multiplied by the total return distance from Auckland-Sydney of 4316km which gives a constant marginal cost of $338 per return flight.
As for Kiwi, its bare-bones operation out of low-rent regional airports would tend to reduce its costs relative to the incumbents', against which are its inexperience and inability to exploit such scale economies as are available. The Commerce Commission (1997) determined that Kiwi had lower operating costs. We will use a figure of $300 and experiment with higher numbers.
We are interested in the nature of oligopolistic interaction before Kiwi's entry in August 1995, and then during the 1996 period of intense competition. Did the `rules of the oligopoly game' change significantly? We will show our estimates of the conjectural variations parameters before and after entry for a number of sets of the other parameters and variables.
First, we show, as Scenario A, what we believe to be the least unlikely or middle-of-the road results, with the 1995 point elasticity of market demand in the middle of its permissible range, at -0.75. Market growth is set at zero (no shift in the demand curve between 1995 and 1996). We will, throughout, set the parameter [Theta] at 0.2. We have no precise reason to do this, but it is not an important parameter and changing its value has little effect. Here we simply report the results, saving most discussion of them to the next section.
Scenario A: Middle-of-the-Road Parameter Values Kiwi zero-demand price = $450 duopoly: [[Lambda].sub.] = -0.23 Kiwi marginal cost = $300 triopoly: 1995 market demand elasticity = -0.75 [[Lambda].sub.I] = -0.67 market growth = 0% [[Lambda].sub.E] = -1.53
Our MoR scenario reveals (a) incumbents pre-entry had conjectural variations quite close to Cournot, at -0.23; (b) the incumbents' behaviour changed substantially in the post-entry period, moving to two thirds of the way towards perfect competition; (c) Kiwi's conjectures are what might be called `super-competitive', being larger in absolute value than -1. These results are quite striking and we will be anxious to find out how sensitive they are to errors in our parameters.
First, we move the market demand price elasticity up and down within its quite restricted permissible range:
Scenario B: Less elastic market demand elasticity Kiwi zero-demand price = $450 Duopoly: [[Lambda].sub.I] = -0.28 Kiwi marginal cost = $300 Triopoly: 1995 market demand elasticity = -0.70 [[Lambda].sub.I] = -0.72 market growth = 0% [[Lambda].sub.E] = -3.0 Scenario C: More elastic market demand elasticity Kiwi zero-demand price = $450 duopoly: [[Lambda].sub.I] = -0.17 Kiwi marginal cost = $300 triopoly: 1995 market demand elasticity = -0.80 [[Lambda].sub.I] = -0.63 market growth = 0% [[Lambda].sub.E] = -1.1
We think it reasonable to infer from Scenarios B and C that our results for incumbents' conjectural variations parameters are fairly stable, but not so the estimate of Kiwi International's beliefs on the response by the incumbents to its actions. This last result is repeated when we return to Scenario A but increase the price at which Kiwi would have lost all its customers, to $500:
Scenario D: Higher value for demand-eliminating Kiwi price Kiwi zero-demand price = $500 duopoly: [[Lambda].sub.I] = -0.23 Kiwi marginal cost = $300 triopoly: 1995 market demand elasticity = -0.75 [[Lambda].sub.I] = -0.67 market growth = 0% [[Lambda].sub.E] = -3.35
Changing this parameter does not, of course, affect the incumbents' CVs, but it more than doubles the Kiwi's. Something similar happens if instead of changing Kiwi's demand we change its costs:
Scenario E: Higher marginal cost for Kiwi Kiwi zero-demand price = $450 duopoly: [[Lambda].sub.I] = -0.23 Kiwi marginal cost = $350 triopoly: 1995 market demand elasticity = -0.75 [[Lambda].sub.I] = -0.67 market growth = 0% [[Lambda].sub.E] = -3.34
It is no surprise that changes in demand and costs have similar effects, as inspection of equation (11) makes clear.
Finally, we suppose that the total market increased (due, say, to income and/or population growth) from 1995 to 1996, such that the slope parameter `b' in equation (1) is 3% smaller in absolute value `flatter' demand curve). This necessarily must lower our estimate of market demand elasticity, if the entry of Kiwi is still to have an overall market-expanding effect. The permissible range of market elasticities becomes (-0.72, -0.54). The results for a scenario with market growth and elasticity in the middle of this range are:
Scenario F: Market expands by 3%; 1995-96 Kiwi zero-demand price = $450 duopoly: [[Lambda].sub.I] = -0.35 Kiwi marginal cost = $300 triopoly: 1995 market demand elasticity = -0.63 [[Lambda].sub.I] = -0.71 market growth = 3% [[Lambda].sub.E] = -0.17
Scenario F shows that our estimate of post-entry incumbents' oligopoly behaviour is not much affected, but there is a quite substantial (about 50%) increase in the absolute size of the pre-entry CV estimate, in the direction of more competitive behaviour. As for the Kiwi International CV parameter, this now becomes rather small.
5. Discussion and Welfare Analysis
We discuss, in turn, what the results of our oligopoly model calibration exercise imply for the behaviour of the two incumbents pre-entry, their behaviour post-entry, and the behaviour of the entrant, Kiwi International, during the period that preceded its demise. We then do a simple public benefit analysis of Kiwi's contribution to the Trans-Tasman market.
5.1 The duopoly
In our base-case solution, we find that Air New Zealand and Qantas, modeled as symmetric duopolists in 1995, generated an estimate for the conjectural variations parameter equal to -0.23, thus implying behaviour just on the `competitive' side of Cournot. This estimate is quite stable to sensitivity analysis, with the possible exception of assuming a non-zero increase in the total Trans-Tasman market between 1995 and 1996 (which requires a lower estimate of the 1995 market demand elasticity).
We find this result quite intuitively appealing as a description of interaction in a fairly mature but vigorous duopoly. It corresponds more to what business people, rather than economists, would describe as `competitive' behaviour. Consider output as sales. Then Cournot conjectures (no change in rival's sales in response to successful effort to increase own sales) mean that the firm expects its rival to defend its market -- to be willing to take a price cut to keep from losing any of its customers. A small negative conjecture means that the firm does expect it can steal a few customers from its rival without the latter responding, or, even, noticing (given the usual amount of `noise' and fluctuations in sales data).
Near-Cournot conjectures are consistent with the study of multiple city pair routes in the US by Brander and Zhang (1990), who find mean values of the CV parameter equal to -+0.06 and +0.12 for American Airlines and United Airlines, respectively, over 33 routes which these airlines operate as a duopoly.
5.2 The duopolists post-entry
What do we expect to happen when two duopolists are joined by a third firm in the market? Even if conjectures do not change, the equilibrium outcome should have a lower price, basically because each firm, having now a smaller market share, perceives their demand to be more elastic. But we should expect conjectures also to alter in a more `competitive' (in the economists sense) direction. Start with a symmetrical duopoly CV of -0.20, meaning that each of the firms expects it could pick up one of every five additional customers from its rival if it increased output. What would firms in a symmetrical triopoly expect?
It seems reasonable that the lower bound for each firm's CV would now be - 0.40, which would hold if the firm thought it could now pick up one of five additional customers from each of its competitors. But the likely number is probably smaller (in absolute value) than this, because loss of a customer is a relatively bigger event to a (triopoly) firm with just 33% of a given total market than to a (duopoly) firm with half the market, and so more likely to be noticed and responded to. This implies that the bilateral firm i/firm j conjectural variations will be smaller than the duopoly CV.
In the present case the triopoly is hardly symmetric, with the entrant Kiwi's limited range of regional departure points and perceived inferior product, and we impose on the model the requirement that each incumbent's conjectured variation with respect to Kiwi be just a quite small fraction (0.2) of their conjectured response from the other incumbent.
But it is this incumbent response that is germane to the predation issue, and of course our most striking result in section 5.3 was that, far from shrinking, incumbent conjectural variations just about triple (in absolute value) during the period that preceded Kiwi's exit from the market. That is, Air New Zealand and Qantas suddenly started perceiving each other as much more competitive in the economists' sense, and much less competitive according to the everyday business person's understanding of the term.
This is really very suspicious, not just because the change in conjectures is so large, but also because of the nature of that change. Bear in mind that these competitive conjectures are actually seriously wrong (mutually inconsistent) -- each firm is now acting as though the other would accommodate most (about two thirds) of its output increases, when of course the result of such conjectures is that the other firm, thinking similarly, actually increases its output. Such a breakdown in mutual understanding does not seem to match the direct evidence on pricing responses, which rather suggests a quite well coordinated response to entry by two well-established airlines, who know each other and their shared market very well indeed. It is difficult to resist the inference that the observed behaviour was not generated by normal everyday commercial considerations, but instead was part of a bigger game, designed to alter the structure of the industry.
By July 1996 Air New Zealand had set up Freedom Air, and then Qantas, followed quickly by Air New Zealand, had cut their fares, without either airline complaining about the behaviour of the other despite these actions resulting in a large fall in fares from the average of $699 prior to Kiwis entry. Indeed, Air New Zealand subsequently seemed to blame Kiwi entirely for starting the airfare discounting war and for fares being sold `at less than an economically viable level's(8) when explaining its poor profit performance for the year ending June 30 1997. No complaint was made about Qantas and no mention was made that it was Qantas, not Kiwi, that slashed fares to the $399/$499 level in June 1996.
5.3 Kiwi International's behaviour
Our model showed estimates for Kiwi International's conjectures as being `supercompetitive', and quite variable with respect to changes in scenario assumptions. What is happening is this: given the estimated demand curve and the incumbents' output, Kiwi's own output levels are too high, in that marginal revenue is below marginal cost. The only way to rationalise this is to deduce that Kiwi's management believed that, were they to cut back output, this would be countered by even larger increases in incumbent output, such that profits would not increase.
Such conjectures make little economic sense, and could even be interpreted as evidence of predatory behaviour by Kiwi itself! However, unwise though the new airline probably was to expand from its regional base into the big city markets of Auckland and Christchurch, it cannot seriously be accused of attempting to eliminate one or both of its incumbent competitors. A more reasonable interpretation of its actions is that, in extremis, it was just trying to keep flying, with no ability to develop and sustain normal commercial pricing practices.
5.4 Welfare analysis
We conclude with a back-of-the-envelope' analysis of the net welfare impact of the Kiwi saga. The numbers are so large that no great precision is needed. On the cost side, we have the $8 million owed to the liquidated Kiwi International's creditors. Just about all of this is lost the airline had very few tangible assets, and there is no suggestion that any of this money is still around. The benefits are the allocative efficiency gains from lower prices and higher output, equal to the area of the `trapezoid' made up of the sum of the consumer surplus welfare triangle, and the rectangle of additional profit when output expands with price above marginal cost.
For the six-month period of intense price competition in 1996, these efficiency gains total more than $34 million(9). For the two years to date since Kiwi International's exit, readily available fares for Trans-Tasman travel appear to be still around 20% below their 1995 levels, implying another two years of efficiency gains totalling upwards of $50 million, of which it seems reasonable to attribute a large proportion to the threat, which Kiwi's entry demonstrated to be real, of new competition. Freedom Air remains servicing the niche regional market that Kiwi demonstrated to be viable.
Consumers, of course, have benefited much more than the net efficiency gain figures: this is an approximately 1 million flights/year market, so that every $100 on or off the average price is worth about $100 million/year to consumers. Without condoning the management practices of Mr Ewan Wilson, Kiwi's founder (for which he has been sentenced by the courts), and without attributing all or even most of these benefits as the result of entry-deterring pricing spurred by the desire on the part of the incumbents not to have to put up with a Kiwi-type operator again, it is hard to resist the conclusion that, within the amoral calculus of cost-benefit analysis, Mr Wilson was a considerable benefactor to the travelling public of New Zealand and Australia.
Kiwi International Airline's participation in the Trans-Tasman market was short in term but significant in its impact on prices, quantities and product variety. Both the incumbent airlines Air New Zealand and Qantas expanded output and engaged in significant price cutting in response to the entry of Kiwi. In the case of Air New Zealand this was done not only directly but also through the creation of what was effectively another marketing arm, Freedom Air, which was directly targeted at the Kiwi International regional market. How should these responses be interpreted?
Standard predatory pricing comparisons of price and cost were not conclusive, and, in any case, a major problem with these tests is that they do not identify the behaviour that generated observed prices, and so cannot distinguish predation from the legitimate impact on oligopolistic conduct that could follow the entry into an industry of another competitor. We therefore advance the analysis by developing an explicit oligopoly model, and by calibrating this to pre- and post-entry market outcomes. The results of this exercise were that the behavioural assumptions that must be assigned to the participants in a non-cooperative setting in order to generate the actual output outcomes observed, are implausible on the basis of what is known about the main incumbents.
We infer that Air New Zealand and Qantas ceased to play their normal non-cooperative (near-Cournot-Nash) oligopoly game following the entry of Kiwi and switched to aggressively `competitive' behaviour in order to drive Kiwi from the market. However, our inferences fall short of discovering the `smoking gun' of direct evidence of collusion to predate. And it should be admitted that, in addition to its difficulties caused by the response of the incumbent airlines, Kiwi also suffered from an array of other problems that must have contributed to its collapse.
Kiwi International had the advantage of strong public support in the Waikato and Otago provinces, which meant it was quite difficult for Air New Zealand to break into this market using Freedom. The combination of less aggressive response, lower costs, as well as strong customer loyalty may have been sufficient to give Kiwi the breathing space it needed to consolidate enough to convince Air New Zealand and Qantas that they should accommodate the new airline.
After Kiwi's collapse and the demonstration effect of Air New Zealand and Qantas' aggressive behaviour, as well as their more restrained pricing in the two years since, it is unlikely that there will be a new entrant in the scheduled airline market across the Tasman in the near future. Also, any new entrant would not have the advantage of being able to enter an unexploited provincial market niche because this is now occupied at least to some extent by Freedom, which could rapidly expand into any other cities where it is not presently operating. But the fare structure seems still to show the effects of the period of intense competition, and for that New Zealand and Australian consumers should be thankful for Kiwi International's brief but spectacular foray into the tough business of international air travel services.
(1) Nationality requirements are (a) at least 50% ownership, and (b) effective board control by Australian or New Zealand nationals.
(2) Air New Zealand and Qantas are now members of different global airline alliances. Qantas actually initiated capacity expansion in April, 1996.
(3) Baumol actually uses the concept of `Average Avoidable Costs' as his preferred variable cost measure.
(4) More detail on the calculations of price and cost is given in Haugh and Hazledine (1999).
(5) A Cournot oligopolist takes rivals `output as given when choosing its output; Cournot-Nash equilibrium is the set of outputs such that, given their others', no one of the oligopolists wishes to change its own output.
(6) (Perfectly) competitive conjectures mean that each firm expects that a unit increase in its own output will not affect the market price, which implies, mathematically that it expects the other firms to reduce their output by one unit, in total.
(7) For example, in the year to March 1995, 431,365 New Zealanders travelled to Australia, and 402,580 visitors from Australia arrived in New Zealand (NZ Official Yearbook 1996).
(8) Bob Matthews, Chairman and Jim McCrea, CEO Air New Zealand, quoted in the (Wellington) Evening Post September 2, 1997.
(9) Using the data from our simulation analysis (annualised output before and after Kiwi's entry rising from about 900,000 to about 1,170,000; price falling from $699 to $483; marginal variable costs = $338), we get the allocative efficiency gain for half a year as:
0.27[ 145 + 0.5(216)]/2 = $34,000,000
Areeda, P. and Turner, D. (1975), `Predatory Pricing and Related Practices under Section 2 of the Sherman Act', Harvard Law Review, 88, 697-733.
Australian Department of Transport and Regional Development, Air Transport Statistics, International Scheduled Air Transport, 1993, 1994, 1995, 1996, Canberra
Baumol, W. (1979), `Quasi-Permanence of Price Reductions: A policy for Prevention of Predatory Pricing', The Yale Law Journal, 89, No. 1, November 1979, 1-26
Baumol, W.(1996), `Predation and the Logic of the Average Variable Cost Test', Journal of Law and Economics, 49-72
Brander, James A. and Anming Zhang (1990), `Market conduct in the airline industry: an empirical investigation', Rand Journal of Economics, 20, No. 4, 567-583.
Bureau of Competition Policy (1992), Predatory Pricing Enforcement Guidelines, Consumer and Corporate Affairs, Ottawa
Bureau of Transport and Communications Economics (1995), Demand Elasticities for Air Travel to and from Australia, Working Paper 20, Australian Publishing Service, Canberra
Bureau of Transport and Communications Economics (1994), International Aviation, BTCE Report 86, Australian Government Publishing Service, Canberra
Commerce Commission (1997), Termination Report: Kiwi International Airlines/ Air NZ, Wellington
Dresner, Martin, Jiun-Sheng Chris Lin and Robert Windle (1996), `The Impact of Low-Cost Carriers on Airport and Route Competition', Journal of Transport Economics and Policy, September 1996
Easterbrook, F. (1981), `Predatory Strategies and Counterstrategies', The University of Chicago Law Review, 48, 263-337
Haugh, David and Tim Hazledine (1999), `Oligopoly Behaviour in the TransTasman Air Travel Market: the case of Kiwi International', Economics Department, University of Auckland, Working Paper 190.
Joskow, P. and Klevorick, A. (1979), `A Framework for Analyzing Predatory Pricing Policy', The Yale Law Journal, 89, No.2, 213-270
Lefever, T. (1981), Predatory Pricing Rules: A Comment on Williamson' s Output Restriction Rule, The Yale Law Journal, 90, 1639-1645
McGee, J. (1980), `Predatory Pricing Revisited', Journal of Law and Economics, 290-330
Van Roy, Y. (1991), Guidebook to New Zealand Competition Laws, Auckland, CCH New Zealand Limited.
Williamson, O. (1977 `Predatory Pricing: A Strategic and Welfare Analysis', 87, 284-340
Williamson, O. (1981), `Reply to Lefever', The Yale Law Journal, 90, 1646-1649
David Haugh and Tim Hazledine(*)
(*) Haugh is with the New Zealand Treasury; Hazledine is in the Economics Department at the University of Auckland. Email addresses: David. Haugh@Treasury.govt.nz; firstname.lastname@example.org. The views expressed in this paper do not necessarily represent the views of the New Zealand government. Earlier versions of this paper were presented to the New Zealand Association of Economists' Annual Conference, Wellington, 2-4 September, 1998, to the Economics Department Seminar, University of Auckland, to the 1998 Industry Economics Conference, Canberra, and to seminars at the University of British Columbia and the University of Calgary. Participants at all of these occasions, and two anonymous referees are thanked for their comments and suggestions.3
|Printer friendly Cite/link Email Feedback|
|Author:||Haugh, David; Hazledine, Tim|
|Publication:||New Zealand Economic Papers|
|Article Type:||Statistical Data Included|
|Date:||Jun 1, 1999|
|Previous Article:||The Structure and Dynamics of New Zealand Industries.|
|Next Article:||Price Stability: Some Costs and Benefits in New Zealand.|