knope2001 From United States of America, joined May 2005, 2555 posts, RR: 31 Posted (10 months 5 days 19 hours ago) and read 2546 times:
With more and more short-haul markets getting cut at Southwest/AirTran, I decided to spend some time playing with public stats to evaluate things. This gets pretty involved, but hopefully (if you care to) you can follow along with how I came up with this stuff.
Looking at Q4 2011 DoT stats (the most recent I have access to when it comes to fare data) Southwest+AirTran flew 102 nonstop markets under 350 miles. Here are those markets, with distance and average fare paid. For example, DAL-HOU is 239 miles and the average fare paid by local DAL-HOU passengers was $149.72.
Next, I went to the T100’s (stats with actual flights flown, capacity, and onboard loads) to figure out how many nonstop flights were actually operated in the quarter. Using this, and the local passengers per day from the DoT stats, I came up with the average number of local passengers on each flight. Multiplying that by the average local fare, I came up with average local revenue per flight.
[Note that this only works in markets where all or nearly all local passengers flew on the nonstop flights. For this reason there were three markets I pulled out of this analysis…HRL-AUS, HRL-SAT, and LBB-AUS. All three of those appear to have significant local traffic that doesn’t fly nonstop. ]
Why is this meaningful? Local traffic is generally where the money is. Here’s an illustration from this quarter. Southwest got an average fare of $185.08 for passengers flying from Cleveland to Tampa, and typically that meant flying CLE-BWI-TPA. However had Southwest sold those same two seats to a local CLE-BWI passenger ($127.10) and a local BWI-TPA passenger ($166.70), they’d have brought in $293.80, or 58.7% more revenue.
.Here’s an example of how this was calculated. Going back to DAL-HOU, there were 2174 passengers per day carried locally between Dallas and Houston on Southwest (2174*91 days =197834 for the whole quarter). Southwest flew 3977 flights between DAL and HOU (both ways) in the quarter, giving 49.7 local passengers on the average flight. Then multiply that times the average local DAL-HOU fare of $149.72, and you get $7447.45 of local revenue generated on each DAL-HOU flight.
Here’s the table for all 99 markets under 350 miles (102 minus the three markets referred to above)
Now that we have a number for local revenue, the next thing is to try to get information on how much revenue connecting (including thru traffic) passengers brought in. That’s a tougher task. First the easy part – figuring out how many onboard passengers were connecting . Easy enough to get the average onboard loads, and when you subtract out the local passengers the rest are connections. Going back to our DAL-HOU example, the average flight in Q4 carried 102 people. Subtract out the number of number of locals (49.7) you end up with 52.3 connecting passengers on each flight.
The tougher part is to estimate how much revenue the average connecting passenger brought in. There are two components to this. First, you have to figure out how much a ticket for the whole trip would be (for example DAL-HOU-TPA average fare was $209.82) and then figure out how much to attribute to the DAL-HOU leg and the HOU-TPA leg.
First, to figure out how much a typical ticket for a connecting passenger would be, I went into the DoT stats and averaged together Southwest and AirTran in short haul markets under 500 miles, in medium haul markets up to 1250 miles, and long haul markets over 1250. I came up with $140 for short, $183 for medium, and $228 for long haul.
Next you have to figure out how much to attribute to the short-haul feeder segment. In a short-haul market it’s easy…50/50 to each segment. In medium haul I decided to attribute 35% to the short hop feeder segment and 65% to the medium-haul segment. And for long haul I decided to split revenue 30/70.
That comes to an average of $67.48 in revenue brought in for the short-hop leg for connecting passengers. But in fact connecting passengers probably brought in more revenue in stronger Southwest markets, and less in ones where Southwest had to cut fares to fill seats. How to weight it to express this difference? Well, I decided to use the relative strength or weakness of fares in the local market as a surrogate for the relative strength or weakness of the connecting fares. The average local short-haul fare in these 99 markets was $128.15. So the relationship between that average local passenger revenue (128.15) and the average connecting passenger revenue (67.48) is 0.526. Here’s how that works. DAL-HOU local fare is an above-average $149.72, and 0.526 of that is $78.75 as the amount of revenue per connecting passenger. That makes some sense, as those are strong markets for Southwest. Toward the other end of the scale, a market like PVD-PHL which had below-average local fares probably have weaker connecting revenue too, as seats need to be filled. PVD-PHL averaged $110.05, and 0.526 of that is $57.89 as average revenue for connecting traffic.
Sooo…instead of just using a flat $67.48 revenue per connecting passenger, that number slides up or down a few notches, using the fare strength in the local market as surrogate to judge the fare strength in connecting markets. It’s about as good as can be done without proprietary information. Once we have the estimated revenue for each connecting passenger, that can be multiplied by the actual number of connecting passengers onboard to get connecting revenue. Add that to the local traffic revenue derived earlier, and now we have total revenue per trip.
Finally, we can’t compare revenue per trip without adjusting for the different sized aircraft. (Remember these include AirTran which has both 117-seat and 137-seat aircraft.) So dividing the average revenue per trip by the average number of seats gives us average revenue per seat in each market. Here are the 99 markets ranked by revenue per seat. I also included if Southwest dropped it or not (apologies if I made any mistakes on that list).
One final note on this. One might suggest that the final step would be to factor in distance and go for a revenue per available seat mile (RASM) number. That could be done, and the results are somewhat similar but not quite a stunningly clustered with nearly all the dropped routes at the top. I’m not sure that, when we’re talking about a large group of short-hop flights it is important or necessarily meaningful to adjust for distance. At 254 miles STL-SDF is 50% longer than BWI-EWR at 169 miles, but it decidedly does not cost 50% more per segment to operate
Anyway, the list of short hop markets ranked by per-seat revenue seems to match pretty closely what Southwest has actually been doing. And it suggests which short routes might be in danger, versus which are probably pulling their own weight. There can be exceptions, of course – PHF-LGA was probably cut because it was decided it didn’t make sense to stay in both PHF and ORF. CRP-HOU probably won’t go because it’s the best way to serve CRP, and perhaps connecting yields from CPR are especially high. This data doesn’t benefit from the sort of proprietary information that the airlines themselves have, but it seems to match pretty closely what’s going on.
point2point From United States of America, joined Mar 2010, 1961 posts, RR: 1 Reply 4, posted (10 months 5 days 16 hours ago) and read 2192 times:
A lot of effort in this I can see. And I'm assuming all is accurate here, this is a table probably similar to what number crunchers produce.....
And it's now easy to see why ABQ, GEG, PHL, and BOI were hit with their cuts. And all along here we thought that WN just didn't like those cities, eh?
I guess we wouldn't know at what rev/seat WN will stop the cutting, but I guess that has to do with the number of planes that they have, and what markets longer than 350 miles are in comparison. And it seems like HOU-DAL, OAK-BUR, MSY-HOU, and SNA-SJC will not go away for a long, long time.
knope2001 From United States of America, joined May 2005, 2555 posts, RR: 31 Reply 7, posted (10 months 5 days 12 hours ago) and read 1956 times:
Quoting point2point (Reply 4): A lot of effort in this I can see. And I'm assuming all is accurate here, this is a table probably similar to what number crunchers produce.....
Thanks....it did take quite a bit of time. Certainly there's some judgment call in this, including determining connecting revenue and how much to attribure to the short-hop feeder flight. There's just no way to know exactly how much revenue every connecting passenger supplied in every single individual market. And while I've heard the 35/65 rule before as how to split revenue between feeder flights and "destination" flights, there's nothing that says that's how Southwest does it. However, that my work ended up with nearly all of the WN+FL markets axed at the very bottom suggests it's reasonably in the ballpark.
Markets not dropped near the top of the list probably have danger. In many cases those surviving-but-weak short hops serve a purpose, but Southwest is probably looking at how to serve that purpose in another way. For example EWR-BWI serves to tie Newark into the the southern network , but once EWR-ATL-xxx connections are possible to fill that need, I'd expect EWR-BWI to end.
Just as interesting is seeing who ends up at the other end of the scale. The comparably strong results of core Texas and intra-California routes are noteworthy. I can't help but wonder what Southwest would look like if one of the California airlines had survived and thrived in the way Alaska has. To a great extent it's prety amazing how much intra-California flying is their unchallenged domain, and in that way it's not surprising that Southwest is so huge in the domestic segment.
Cubsrule From United States of America, joined May 2004, 21242 posts, RR: 19 Reply 9, posted (10 months 5 days 10 hours ago) and read 1772 times:
Quoting FlyPNS1 (Reply 8): This is one of WN's younger routes....launched less than 2 years ago. It will be interesting to see if it survives.
STL-MKE is too. GSP-BNA may fit in to the "network need" category. The only other similar route they could fly is GSP-STL, but it's not clear to me that that would necessarily be a stronger local market, and it's 250 miles longer.
One thing that surprises me, though, is how strong several short STL markets are (MDW, OMA, BNA). STL-BNA particularly surprises me, because that's another fairly "young" route for WN.
I can't decide whether I miss the tulip or the bowling shoe more
point2point From United States of America, joined Mar 2010, 1961 posts, RR: 1 Reply 10, posted (10 months 5 days 9 hours ago) and read 1711 times:
Quoting knope2001 (Reply 7): Certainly there's some judgment call in this, including determining connecting revenue and how much to attribure to the short-hop feeder flight. There's just no way to know exactly how much revenue every connecting passenger supplied in every single individual market. And while I've heard the 35/65 rule before as how to split revenue between feeder flights and "destination" flights, there's nothing that says that's how Southwest does it.
For all extents and purposes, if a company says publicly that there's a 35/65 split system-wide, it usually isn't that far deviant from any individual route within the system, although there of course may be some on the outer edges as anyone familiar with a Bell Curve would understand. But other than a few above that you mentioned such as HOU-CPR or BWI-EWR, and maybe a few more on each edge to balance themselves out, most of the routes there seem like they would be very few points off the norm.
And with that (just quickly figuring off the top of my head), the dollar amounts for the rev/seat, using your system, at most probably wouldn't be more than maybe a couple of dollars off any any direction in most markets. In the long run, the table does more that likely give a good picture of the short-haul scenario of WN.
Sometimes, with instinct, reasoning, and focus, and just knowing circumstances and how to apply numbers, it is doable to get somewhat of an accurate picture of what is happening inside of a company (or situation) using public data. And it seems that your efforts lead you to that place where things makes sense, and maybe you even surprised yourself with your conclusions?
Cubsrule From United States of America, joined May 2004, 21242 posts, RR: 19 Reply 11, posted (10 months 3 days 7 hours ago) and read 1264 times:
Quoting knope2001 (Reply 7): In many cases those surviving-but-weak short hops serve a purpose, but Southwest is probably looking at how to serve that purpose in another way.
I pondered this point while flying on Southwest 1430 (BHM-BNA) this afternoon, and there's something larger here. The short hops that are getting dropped are, in some sense, victims of WN's success. In many cases, network expansion has caused WN to be able to serve the traffic on those short hops in more sensible ways. Taking IND-MDW as an example, the addition of service to DEN increases the possible westbound connecting flows that could be served over MDW, and today IND-west coast passengers have 5 connecting options (PHX, LAS, DEN, MCI, HOU). I pulled up a couple of random IND-west coast city pairs in today's schedule. IND-ABQ has 10 legal connections. Only two of those are over MDW. IND-PDX has 7 legal connections, again with 2 over MDW. Connecting passengers just don't need MDW as a connecting point in most cases.
Now, that's not to say that some of these routes don't have some network value. Today, WN 1283 (DAL-BHM-BNA) carried 68 DAL-BNA passengers. Adjustments will be necessary to accommodate those passengers, but WN has never really been about operating flights with that kind of a local to non-local ratio. FL accounting may be the cause of the change, but in a way, it's just WN going back to its roots.
I can't decide whether I miss the tulip or the bowling shoe more