Why we don’t produce full-service carrier average fare data – part II

Why we don’t produce full-service carrier average fare data – part II 1 April 2019
LCC fare data

The motive for not doing it is because its riskier than playing poker with an octopus! Andrew Myers continues his explanation for our method.

In our last blog we detailed our process for estimating LCC average fares, which we believe gives the most accurate third-party picture of the European budget carrier scene in the industry. The accompanying graphic details why we don’t adopt this process on a network-wide basis for full service airlines, especially long-haul. In this blog we explain the graphic in more detail:Lufthansa Fare ErrorFirstly, there can be up to four cabins on long-haul equipment, with various load factors in each and differently configured aircraft on the same route. Next, connecting traffic has to be factored in, which will be carried at a far lower sector yield than point-to-point traffic. Not only do we have to try and estimate how much connecting traffic there is, which might vary considerably by route, but also how much the respective sectors are discounted by and allocated within the overall fare. For example, a London Heathrow (LHR) to Bangkok Suvarnabhumi (BKK) fare might be £500, while a LHR to Sydney (via BKK) fare might be £750, meaning that the LHR-BKK component might only be £375 (assuming an equal split).

Another assumption which has to be decided upon is that we need to consider corporate deal traffic, which will be carried at different proportions in each cabin and at different discounts to the fare. Then, there is likely to be some non-revenue traffic (staff passengers and crew) which also needs to be factored in, and finally an allowance needs to made for cargo.

Small errors add up

Because of the wide range of aircraft types and configurations it is impossible to accurately calibrate the model to airline accounts, especially when the airline also flies short- and medium-haul routes. This makes the accuracy of the individual study all the more important. With almost all the factors above it is very difficult to make accurate assumptions, but how much impact would a small error in each have on the overall result?

Aviation Analytics decided to have a look using an example of an seven-hour (or so) sector from LHR to New York JFK on a network carrier’s 787-9 fleet. Our model was set up with a base-case scenario, with some sample costs (around £50,000 per sector) and average fares by cabin, while assumptions were adjusted to give a profit margin of 8% on each seat sold.

The base assumptions were:

Fares of £1,000, £350 and £150 for business, premium and economy seats respectively;

Load factor of 85% in each cabin;

30% connecting traffic carried at 65% of the point-to-point fare;

20% corporate traffic at a discount of 30%;

5% non-revenue traffic (staff, FFP, upgrades etc) carried at a discount of 75%;

10% cargo revenue.

This gave an average profit per seat sold of just under £20 based on 787-9 operating costs of £50,000 per sector. Then we set up a scenario where each assumption was five percentage points out in an unfavourable direction. We subtracted 5% from all fares, increased the economy load factor to 87% while adjusting the other cabins down to maintain an 85% average. Connecting traffic was increased by five percentage points, whilst yield was reduced by 5%. The same five percentage point changes were applied to the other factors. The resultant profit per seat sold reduced from £20 to a LOSS of £25 and a margin of -13%.

Invalid modelling

While it is possible that not all errors will align in the same direction, meaning that, for example an underestimate on connecting traffic could be cancelled out by an overestimate on corporate traffic, an overall error of 5-6% is almost certainly possible (individual errors could easily be up to 15-20%). On routes such as the one above, the overall average fare can easily be 20% out, invalidating the entire forecasting model.

Aviation Analytics is able to provide good value fare data on any route (direct and indirect), and the trends and relativity in these fares can certainly be used as a good indicator of overall market conditions. On an individual route, it is possible (with a lot of research), to get an idea of average fare and profitability, but to try and accurately work out individual route profitability on a network scale with so many assumptions is simply not possible. Any company without a GDS-based system that is claiming otherwise, is simply not being honest with its customers in our opinion. That’s why we stick to looking at LCC fare data…as the results are far more robust!!

Share this page