Did you like this article? Let us know what you think about it! Click here
 |
|
|||||||||||
|
|
|||||||||||
|
|
|||||||||||
|
||||||||||||
|
|
||||||||||||
|
|
|
|
|
||||||
|
September 2008
Pricing and Competition in the New Zealand Air Travel Market
The domestic passenger air services market
in New Zealand can be split up into three groups: The database on hand consists of 1076 flights (e.g. flying from Auckland-Wellington on 22/08/07 departing at 7.00am is considered as one flight). The departure dates range from 22/08/07 to 10/10/07, on a weekly basis. All departure dates are on a Wednesday, this day being chosen as we believe that Wednesday represent a 'typical' weekday which is likely to capture both business travellers and holiday makers. The range of departure dates also include the school holiday period which lies from 22/09/07 to 07/10/07 - school holiday between Term 3 and Term 4 - and we hope to capture some interesting behaviour in our database.
The revolutionary internet booking system provides a means of collecting data on airfares. The airfares were collected from the respective airlines websites. The fares recorded in this study were the cheapest available adult one-way economy fares for each flight. The price information revealed a typical inter-temporal price pattern, with lower fares farther out from departure date and fares rising as departure date approaches.
A thorough paper in regards to airline cost functions was that of Swan and Adler (2006). This paper splits the aircraft operating costs into various components which include pilot, cabin crew, fuel, landing fees and two separate measures of maintenance. An investigation on these factors was done by modelling the long run average cost of a flight based on the size of the aircraft and flight distance. It is found that "most airplane costs are proportional to the hours flown, and hours flown are linear in distance" (Swan and Adler, 2006, p. 107). It is from this fact, and its relative ease of availability, that it was decided to make use of the non-stop flight distance (DIST) as a proxy for flight costs in the econometric models. Using fuel price as an additional cost measure was considered, however, this was refrained from keeping in mind the short time frame of the dataset. We should keep in mind that airlines buy fuel at hedged prices and not current market prices. Thus, it is assumed that in this short period of time, the cost of fuel for the airlines remain constant.
3.3 Measure of Market Concentration The Herfindahl-Hirschman Index (HHI) has commonly been used as a measure of market concentration. The index is calculated by squaring the proportional capacity of each firm in the industry and summing the resulting figures:
There are two measures of supply, the number of seats, and the number of flights available on a route. I was lucky enough to be able to find data on both these measures. The number of flights available on a route was information which was readily available from the dataset. As for the number of seats on offer (on a given flight) - upon further investigation it was found that clicking on the flight number reveals information on the type of aircraft which was going to be used for the flight. The airlines' websites also contain fleet statistics which reveal how many seats are available (and the possible configurations of these) on the different types of aircrafts. Hence, making use of these two pieces of information revealed the number of seats which would be on offer on a flight. To reiterate, we have two measures of the Herfindahl-Hirschman Index. HHI based on the number of seats on offer on a route, HHISEATS, and HHI based on the number of flights on a route, HHIFLIGHTS. Each of these measures does not vary over the sample time period.
In addition to the information mentioned above, the following dummy variables were also used as part of the econometric analysis:
This section builds on the existing literature on airline pricing. It reports econometric regression estimates of factors affecting average fares and aims to analyse the price discrimination in airline markets using the 1076 observations on average fares offered for flights within New Zealand. Furthermore, it tests the truth in the comments made by Air New Zealand, who in January 2007 announced that "fares will fall by up to 26%". It will also be investigated if the airfare structure of Qantas has indeed been lowered. This paper will present two models which will allow us to test the following hypotheses for the New Zealand domestic air travel market:
This sub-section will outline the econometric models which will be used in our analysis. The first model uses the same specification as in Hazledine (2007).
where PWAVK is the weighted average price per flight per kilometre. The weighting is assigned taking into account the proportion of leisure to business travellers on a route. It is well known in airline markets that leisure travellers book flights well in advance while business travellers purchase tickets close to departure date given that they require added flexibility. Hence for routes which have a high proportion of leisure travellers, relatively greater weights are assigned for fares collected farther out from departure while predominantly business routes have relatively bigger weights allocated for fares collected close to departure. A further specification for this model allows us to check the fall in average fare per kilometre on domestic routes which in turn allows us test our third hypothesis. It should be noted that this specification makes use of two datasets. The first dataset is the one described of above containing 1076 flights with departure dates from 22/08/07 to 10/10/07. While the second panel dataset contains 977 flights with the same variables and for the same route pairings as the first dataset but for departure dates ranging from 08/11/06 to 20/12/06.
where DATA2007=1 for the new dataset i.e. for flights from 2007, while DATA2007=0 for flights from 2006. Hence this specification should give us a fair indication as to whether the average fare per kilometre has decreased. The passenger air services industry is also unique in the sense that airlines are able to adopt yield management (price discrimination) techniques that distinguish travellers with different willingness to pay. The yield management system deployed by airlines is aimed at filling aircrafts as much as possible and at the same time obtaining the maximum fare each passenger is willing to pay. The second model is a measure of a standard form of price discrimination i.e. charging different prices to different groups of people for a homogenous product. Hazledine (2005) uses the following variables to model the dynamic price discrimination for a small-number oligopoly case.
where PDIFF is the ratio of highest low price to lowest low price. The main variable of interest here is HHI, which is the measure of structural competition. It is used to see what effect the level of competition has on price discrimination. HHI for a monopoly route is equal to 1 while for a symmetric duopoly will be equal to 0.5. Hence a greater value for HHI implies less competition.
This sub-section presents the findings of the models outlined in the setup. The Average Price Model The coefficient on distance implies that a
10% increase in distance results in approximately 7% decrease in
average price per kilometre. It is known that for domestic New Zealand
services, most of the flight costs are on-the-ground costs which
are not directly proportional to the distance travelled. The on
ground costs include landing fees, baggage handling, boarding costs,
maintenance costs etc. and as a proportion of total costs these
fall as the flight distance increase. The signs on the coefficients of the dummy variables are all fairly intuitive. It is found that average fares during peak travel periods are just over 20% higher than fares for flights at other times. It is believed that these fares are higher to capture the lucrative business travellers who purchase close to departure. The coefficient on SOLDOUT reveals that average fares for sold-out flights are approximately 16% higher than fares for flights which are not sold-out. This would imply that airlines are able to forecast which flights are likely to be sold-out and thus add a premium to the fare. It is also found that flights which have a stopover have average fare per kilometre close to 40% greater then fares for non-stop flights. The coefficient on the next variable, HOLIDAY, provides evidence of some seasonal effect. The dataset contains flights with departure dates ranging from 22/08/07 to 10/10/07 with the school holidays falling between 22/09/07 to 07/10/07. During this period average fares per kilometre are around 7% higher than fares for flights in the non-holiday period. This might specifically be aimed at taking advantage of the increased leisure travel which is expected to occur during the school holiday season. The estimates also imply that fares on Qantas flights are up to 22% lower than fares on Air New Zealand flights.
Next the aim is to test the second hypothesis, which is whether greater market concentration (reduced competition) leads to increased price discrimination. The coefficient on the measure of structural competition, HHI, suggests that the ratio of highest low price to lowest low price (PDIFF) is 8% greater on a monopoly route compared to a symmetric duopoly. This would mean that if PDIFF is equal to 1.5 on a duopoly route, then on a symmetric monopoly it would be equal to 1.62, i.e. the highest fare is approximately 62% more than the lowest fare in the eight weeks prior to departure. This signifies that more competitive routes have a lower dispersion of fares offered and this is in line with our second hypothesis. The coefficient on PEAK suggests that peak time flights have 10% greater dispersion in fares than off-peak periods. This is not unexpected considering that these flights would mostly be occupied by business travellers who purchase close to departure date. Hence airlines would try and create greater market segmentation so as to take advantage of the travellers varying sensitivity to prices. Leisure travellers are highly price sensitive and purchase well in advance when airfares are lower while business travellers who have higher willingness to pay end up paying the premium end of the fares. The finding of this section is consistent with the standard textbook perception that as competition increases, firms price closer to their marginal costs and hence price dispersion decreases.
The paper now presents the findings on whether Air New Zealand and Qantas have lowered their domestic New Zealand airfare structure in 2007 compared to 2006. A certain degree of caution should be exercised when interpreting these results due to the inability to control for time varying factors. However, the critical market concentration variable, HHI, was adjusted for between the two time periods. The focus is on the two key variables, HHI and DATA2007. The coefficient on HHI implies that average fare on monopoly routes for the combined time period is approximately 14% greater than the average fare on symmetric duopoly routes, ceteris paribus. Previously, when regression was performed just for the August to October 2007 sample, this figure was discovered to be as high as 20%. This would suggest that the gap between monopoly and duopoly pricing has actually increased. Hence this should provide greater incentive for regulation authorities to allow and encourage a greater level of competition in the New Zealand air travel market. As for the coefficient on DATA2007, our estimates suggest that average fares per kilometre for the period August to October 2007 are around 9% lower than they were for November to December 2006, keeping everything else constant. However, a closer analysis of the fare decrease reveals much more interesting results. Adding a new dummy interaction term, DATA2007*QANTAS, enables us to differentiate the fare decrease across the two airlines. The approximations imply that Air New Zealand fares have decreased in 2007 on average by approximately 7% compared to 2006. In contrast, Qantas fares seem to have been lowered on average by as much as 17%. Given the fact that Qantas fares were already lower than those of Air New Zealand, these new results would suggest that the gap between Air New Zealand and Qantas pricing has actually increased in 2007. This is further evidenced by the graphs below. They are illustrated for routes on which Air New Zealand faces competition from Qantas.
Air New Zealand Qantas Figure 4 shows the mean fares on the Auckland to Wellington route for both Air New Zealand and Qantas from eight weeks prior to departure to a day before the flight for years 2006 and 2007. It reveals that the price level seems to have decreased on the Auckland to Wellington route in 2007. But the gap between the two airlines pricing structure appears to have widened. The figure also shows the varying degrees of price discrimination. In 2007, Qantas keeps fares fairly stable up to about two weeks before the flight but then fares start increasing. As for Air New Zealand, in 2007, they still maintain an active yield management scheme with fares increasing smoothly up to around a week before the flight and then increases sharply in the last week.
Air New Zealand Qantas Figure 5 illustrates prices for the Auckland to Christchurch route. Once again here we have evidence of lower average fares in 2007, however, the two airlines pricing appear to be closer together.
Air New Zealand Qantas Figure 6 shows prices for the predominantly tourist route Auckland to Queenstown. There appears to be a slight indication of higher average fares for Air New Zealand and lower average fares for Qantas in 2007 compared to 2006. This is highlighted by the expanded gap between the two airlines pricing structure.
The final illustration shows prices for the Christchurch to Queenstown route. Remarkably, Air New Zealand fares appear to be considerably greater in 2007 in comparison with 2006. However, given that Queenstown is a popular destination for ski enthusiasts, this is likely to be due to the ski season which lasts from June until October. Hence the 2007 dataset includes this period while the 2006 dataset does not. Qantas's fares on the other hand appear to be lower with hardly any price discrimination. Figures 4 to 7 demonstrate that our result, that Qantas fare cuts are greater than those of Air New Zealand, appear to be valid. In fact, the graphs also raise the possibility that in response to decreasing prices on some routes, Air New Zealand may have gone on to raise prices on some other routes to perhaps make up for lost revenue. If this is the case, Qantas would appear to be more disadvantaged to these fare cuts. We should keep in mind that these results are comparing fares of flights at different times of the year. Fares for 2007 are collected for flights in the August to October period while fares for 2006 were collected for flights in the November to December period. It is expected that fares for this latter period would generally be higher. If this was to be the case, our results would be overestimating the fare decreases and underestimating the fare increases (i.e. compared to the case if the fares for 2007 were also collected for flights in November and December).
This paper has analysed new price data for domestic New Zealand flights in August and October 2007. There is a significant variation in average price per kilometre offered on different routes with longer distance routes having average fares per kilometre much lower compared to shorter distance routes. The average price analysis revealed that fares on monopoly routes are on average up to 20% higher than on a symmetric duopoly, a finding which implies that greater competition does lead to lower average fares. It is also found that routes on which Qantas operate have fares up to 22% lower than routes on which Air New Zealand has a monopoly. Hence if the proposed alliance of Air New Zealand and Qantas was given approval by the authorities, there would have been substantial lessening of competition on domestic New Zealand routes. The results in this paper also exemplify that the extent of price dispersion diminishes with lower market concentration. This finding is consistent with the standard textbook perception that as competition increases, firms price closer to marginal costs and hence price dispersion decreases. However, given the low R2 of 0.01 for the price discrimination model, this result is not conclusive. A possible expansion could be to look at how competition affects the high-end and low-end fares separately. All these results are fairly consistent with other airline pricing and competition literature in the sense that allowing for a greater level of competition would lead to lower average fares and reduced price dispersion. Further, this paper makes use of additional price data for domestic flights in November and December 2006 to investigate whether there has been a shift in the price regime. It finds that average fares per kilometre in the New Zealand domestic air travel market are approximately 9% lower for the period August to October 2007 than they were for the period November to December 2006. Moreover, Qantas price cuts appear to be significantly greater than those by Air New Zealand. It is further established that in response to lowering fares on some routes, Air New Zealand might have reacted by increasing fares on some other routes. However, the price cuts outweigh the price increases and hence this is reflected by a fall in their overall average fare structure. There were two particular limiting factors
to this study. Firstly, the number of tickets sold at each price
point was information not available. Secondly, using distance as
a cost proxy does not give accurate information on the costs associated
in operating a route for each airline. One should look to obtain
or construct a cost measure which can be directly attributable to
a particular airline. An interesting focus for future research on
New Zealand airline pricing could involve the recent entrant Pacific
Blue. In particular, one could collect fares post and pre entry
and perhaps study the impact the entrant has had on the pricing
of the incumbents, Air New Zealand and Qantas.
| |
 |
|
|
 |
Copyright 1998 -
Celway
Group. All Rights Reserved. |
 |
|