Lies, damn lies and statistics

Comparing load factors on a “like for like” basis

During a twitter exchange this morning with Simon Calder about airline overbooking, I made the point that Ryanair do not report passenger numbers and seat factors in the same way as most other airlines. Given that Ryanair likes to boast about its “industry-leading” load factors, I think it is important to note this difference as it can be quite large. I thought I’d have a go at working out how big the gap is.

How to count passengers?

Most carriers report passenger numbers based on the number of seats that were actually occupied. IATA’s “glossary of terms” defines a passenger as “Any person, except members of the crew, carried or to be carried in an aircraft with the consent of the carrier.” So stowaways don’t count, apparently. There is also the notion of a revenue passenger, which excludes passengers who travelled for free. That used to make more difference than it does these days, as frequent flyer redemptions and staff travellers generally pay some sort of fee related to their travel and are therefore included.

Ryanair instead reports the “number of fare-paying passengers booked” and also calculates a “booked load factor” by dividing this by the number of seats operated. So Ryanair counts a passenger regardless of whether they they actually turned up or not. How big a difference does this make? Let’s have a look at an alternative source of passenger numbers and load factor data, the UK’s CAA.

CAA data

The CAA requires all UK registered airlines to report figures on a standard basis. For those who are interested, you can find the airline data here. The CAA defines passenger numbers as “All revenue and non-revenue passengers on air transport movement flights”.

I’ve used data from their “Table 08 2 Aircraft Type and Utilisation Individual Airlines” to calculate the passenger load factor. I used the breakdown by aircraft type so that I could split out just the short-haul flights for British Airways, to get a more direct comparison to the low-cost operators.

To start with, I’ve looked back at 2019 and compared the figures to the reported numbers for that year, shown in the chart below.

You can see that BA’s numbers line up pretty well. For some reason, the numbers reported in their Annual Report are marginally higher than the CAA figures for all aircraft types. But very close. The load factor for BA’s short-haul operations is slightly lower than the overall average, at 82.7%.

The CAA figures for easyJet cover only “easyJet UK”, the subsidiary of the group with a UK Air Operators Certificate (AOC). Back in 2019, that covered over half of their operation, so you would expect the load factors to compare closely to the overall reported figure. And they do - it doesn’t look there are any material definitional differences for easyJet.

But the same is not the case for Ryanair and Wizz. Their CAA figures for load factor are 8 and 6 points lower, respectively.

Source: Company reports, CAA, GridPoint analysis

Could the difference be explained by UK flights being much worse than the average?

Let’s start with Wizz, since we have more information to go on. The UK subsidiary accounted for about 35% of Wizz’s capacity in 2019, so there is some scope for divergence in performance. Luckily, Wizz UK files accounts and provides a figure for load factor, which must be on the same basis as the group numbers. For the 12 month period ending March 2020, reported load factor was 92.6%, slightly lower than the 93.5% reported for the whole group. The CAA data is monthly, so I can also calculate a figure for the same period - which was 86.7%. That gives us a direct estimate of the difference due purely to reporting definitions of 5.9 points. Unlike Ryanair, Wizz don’t say anywhere in their accounts that they are using a non-standard definition of passengers and load factors - but the figures suggest that as well as copying Ryanair’s business model, they have also copied its “innovative” approach to reporting.

Unfortunately we can’t perform the same trick for Ryanair, since Ryanair UK’s accounts don’t include any operating statistics. There is also a higher risk of the UK subsidiary not being representative of the group, as it only represented 0.2% of capacity in 2019. Even in 2022, the UK subsidiary only operated circa 1% of group seats. So we are going to need another source of data and luckily, the Spanish airport operator AENA collects and publishes an enormous amount of data.

Evidence from Spain

Back in 2019, Spain was Ryanair’s second biggest market, with 18.6% of flight departures (just behind Italy at 19.2%). That actually understates its significance, as many of the flights from other countries landed there. AENA reports statistics for both departing and arriving flights, so in 2019 it showed 52.9m Ryanair passengers passing though its airports on 312,000 flights. That’s 59.0m seats, almost 40% of the group total. So it ought to be a pretty representative “sample”.

52.9m passengers divided by 59.0m seats works out at a seat factor of 89.7%, which is 6 points lower than the group’s reported 96% figure, very consistent with the difference we found at Wizz UK.

How are things going in 2022?

Since we have the CAA data monthly up to May (except for Wizz for some reason), we can also see how seat factors have compared in the recovery phase. It is clear that easyJet have been consistently winning the load factor competition, despite what you might conclude looking at reported numbers from Ryanair and Wizz.

BA’s narrow-body fleet has been “in the pack”, lagging easyJet UK but consistently better than Wizz UK and similar to Ryanair UK.

Source: Company reports, CAA, GridPoint analysis

Mind the gap

The gap between Ryanair’s “booked load factor” and the standard measure appears to have grown a little in 2022, if the evidence from Spain is to be believed.

I’ve shown here the figures for the year so far. It’s worth bearing in mind that I’ve used a seat count of 189 to calculate the capacity in Spain (AENA only gives the number of operations). The company has been introducing its 737-MAX aircraft, which have 8 more seats, so if anything I may be underestimating the capacity and overstating the seat factor on Spanish flights. That would widen the gap between group reported seat factors and the evidence from Spain.

The last few months have seen a gap of more like 7 points, compared to the 6 points we calculated based on 2019 data.

Source: Company reports, AENA, GridPoint analysis

Does it matter anyway?

I’m sure the airline geeks will have been interested in this exploration of airline metrics, and the heads of revenue management will no doubt be pleased to have some ammunition to use when challenged about why their seat factors aren’t as good as Ryanair’s or Wizz’s. But otherwise, does it matter?

The way Ryanair reports its passenger numbers does make it look a bit bigger than if it used a standard approach. It also flatters its preferred metric of “cost per passenger”. Both are useful when marketing itself to investors and airports.

But I think the main reason why “normal people” and policy makers should care about this issue is that both Ryanair and WIzz argue that their superior load factor performance means that they are more environmentally friendly than other carriers. I think this evidence shows that when measured on a like for like basis, that really isn’t the case.

Overbooking

Ryanair apparently doesn’t overbook its flights, something I think that every other airline does, at least to some extent. Whilst overbooking is often criticised as being “customer unfriendly”, the reality is that it is the flip-side of offering flexibility to customers to change their plans. Without overbooking, allowing customers to book or change their plans at short notice would mean lower achieved load factors. Ryanair manages to achieve high load factors without overbooking only because it offers very little flexibility.

Could Ryanair change its policy on overbooking? It is probably leaving money on the table at the moment. However, if it does change its policy without changing the definition of how it reports load factors, it could end up with reported load factors of over 100%, which might raise a few eyebrows.