Nate Silver blogged yesterday at the New York Times about airfare, and he’s a smart guy who knows his regressions but he appears not to know very much about air travel.
His post is titled, Which Airports Have the Most Unfair Airfares? but his methodology is seriously flawed.
Almost as a throwaway he tosses out as fact that it’s best to shop for airfare on the weekends. He cites a paper that claims to show greater dispersion in airfares but “the same fare level on average.” And while it’s true that there’s less shopping for price insensitive business travel over the weekends, there are also fewer airfare sales, partly as a function of just having fewer scheduled uploads to the computer reservations. Prices change more quickly during the week, and more people are working during the week to change and adjust those prices than they are on the weekend. Still, some of the best deals are found on weekends precisely because that’s when price sensitive shoppers tend to have time to shop, and also because mistakes or fare wars last longer over the weekend, it takes more time for an airline to load new higher fares.
There’s no magical time to search for airfare, and much better tips can be found in various posts on this blog (just use Kayak.com, learn fare rules and find the lowest fare then search for flights with availability in those buckets, learn to dump fuel surcharges, search for cities where you can get cheaper fares while connecting through the city you want to travel to, etc). And there are similar suggestions for getting the best hotel deals as well.
Silver is certainly correct that on average some airports are much less expensive to fly out of than others, such as Milwaukee being a whole lot cheaper than originating in Chicago when traveling to many markets. Milwaukee is served by more low cost carriers and sicounters, Chicago is congested and simply doesn’t have the gate space or the ability to quickly turnaround an aircraft that’s necessitated by the business models of some of those carriers. And Chicago also has more price insensitive business travelers than Milwaukee does (agent problems inherent in spending other peopels’ money, also people with higher opportunity cost of time).
But when Silver assets that flying out of Newark is 25% more expensive than flying out of JFK, he misses:
- That Newark has far more non-stop destinations than JFK, as Wandering Aramean notes. (And while that may entail lower costs, it also involves greater demand and New York’s airspace, gates, etc. mean that demand is met by limited capacity.)
- That JFK is home to JetBlue, which drives much of the data for the single airport comparison.
Beyond simple comparisons Silver wants to get at “not where are the average fares highest, but where are they the most unfair.”
But he’s on pretty shaky ground when he tries to construct a moral framework for airline pricing.
I doubt anyone would dispute that it’s fair for an airline to charge you more for traveling a longer distance, so distance is something we’ll need to control for. I’d also argue that it’s fair to charge you more for flying into or out of a smaller market where there is less demand for air travel. Fewer economies of scale are available in a place like that: the airline can run fewer profitable flights each day, so the costs of ground services like check-in and baggage handling will be spread over fewer passengers, and aircraft may be idle longer.
At the same time, smaller airports tend to be served by fewer airlines, and lack of competition can lead to higher prices; that isn’t fair to the traveler. Nor is it fair if your home airport is one like Memphis, where discount airlines like Southwest have had trouble breaking in to compete with full-fare carriers, despite years of trying.
What we need is an approach that distinguishes airfares that are high because of monopoly pricing from those on routes that are legitimately expensive to fly.
Silver appears to build a “cost of production” moral theory of airline pricing, and posits monopoly as the cause of markets that he believes ‘ought’ to have more service than they do (and the lack of service is ‘unfair’ to the traveler). Except that his cost of production model even fails to understand airline costs, as he abstracts away from many of the key ones in his analysis.
Silver explains how he massaged his data. I’m not sure his ‘cleaning up of data’ to remove uber-expensive and uber-cheap fares makes sense, though I also don’t know whether his conclusions were sensitive to these adjustments. He looks at only domestic tickets flown in coach, but assuming that he means what he says — flown in coach, rather than purchased in coach — then presumably he excludes the trips of many frequent flyers regularly receiving upgrades. That has to skew the results.
Then he waives his hands at the extra distance and costs of connecting flights:
The first factor is the distance traveled — we use the distance from the origin airport to the destination as though it were a nonstop flight, whether or not there was a layover along the way (since the airline is not exactly doing the passenger a favor by routing her through, say, Baltimore on her way from Buffalo to Atlanta, increasing the number of miles flown).
I’d dispute the characterization that the airline is ‘doing something’ to the passenger by virtue of their routing, rather passengers generally select their routing. But that’s beside the point.
He wants a cost of production model of fairness, but connecting flights are inherently more expensive since takeoffs and landings are more expensive than flying distance. Somehow those costs don’t get factored into his notion of a fair price.
And connecting through a hub means costs associated not just with takeoffs and landings but also with a second airport, second gate agent, catering (even for just soft drinks), ground service. And hubs operate with significant labor costs that are often idle outside of banks of flights, in order to service passengers quickly. The hub and spoke model is about planes waiting on passengers rather than passengers waiting on planes, and that’s costly. (And is precisely what creates an opening for a carrier like Airtran in Atlanta which offers frequently longer connecting times — passengers waiting on planes instead of the other way around).
He also seems to draw conclusions from the data when his data doesn’t support causality.
Prices are higher the more the legacy airlines dominate an airport, but they also tend to be a bit higher where Southwest has a large share as opposed to other low-cost carriers like AirTran and JetBlue. (Southwest is cheap, but it isn’t quite as cheap as some of these up-and-coming airlines and now represents something of a middle ground.)
It’s true that Southwest isn’t that cheap and its costs also aren’t that low by many measures, it’s also true that Southwest is pretty darned methodical in picking the cities it serves and will often go in and lower prices relative to what came before, but is picking markets that can still support something other than rock-bottom pricing. Nothing in the claim suggests Southwest is high- or higher-price, and there’s no comparative analysis here of markets before and after Southwest entry. He simply claims too much.
Also, prices tend to be higher when any one airline dominates an airport, regardless of whether it is a legacy carrier or a low-cost one.
Fair, but several reasons for this. Sure, lots of competition drives down price but the very reason that those markets are dominated — outside of available gates and air traffic and subsidies — is often the same reason that prices are high, rather than the cause of the high prices. Otherwise additional airlines would add service . . .
More proof that Silver is out of his depths looking only at statistics to draw conclusions without knowing much about his subject matter — the case of Northwest Regional Airport in Arkansas.
Perhaps the most unfairly priced airport in America is Northwest Arkansas Regional in Fayetteville. Economy-class round trip tickets cost an average of $527 there, $158 above fair rates.
The reason, no doubt, is because the traffic there is dominated by business travelers on their way to and from the headquarters of Wal-Mart in nearby Bentonville. Although Wal-Mart is famous for its sensitivity to prices, and has saved consumers billions of dollars over the years, its pricing power evidently does not extend to the air fares its executives and clients have to cough up.
Of course, Silver has no idea what Wal-Mart pays in its corporate deals for airline tickets.
Northwest regional airport doesn’t have a mix of traveler types, it needs high frequency but with low number of seats per aircraft. Because it’s drawing business travel but not leisure traffic. It can’t support larger aircraft with lower seat costs as a result, and so the cost per enplanement is going to be much higher than at either larger metropolitan destinations or vacation destinations. And that’s precisely to cater service that matches the demands of air travel passengers!
None of which makes flights in and out of the airport “unfairly priced.”
Similarly, I’d love for Silver to stop and think about the following claim he makes.
Memphis, about 28 percent overpriced, is nearly as bad, because Delta controls about two-thirds of the passenger traffic and FedEx ties up a lot of the airport’s flight capacity with its shipping hub.
Can Silver explain, then, why Delta is de-emphasizing service through its Memphis hub? Further, he recognizes capacity constraints (oddly, in of all things the Memphis context more clearly than the New York one!) as a limiting factor on supply which influences price. And yet this is the only place where he acknowledges the role that cargo plays in airline pricing! Flights will often exist and cover fixed costs through their cargo operations, and then to the extent that they’re being serviced by passenger aircraft will see any passenger revenue as incremental. Cargo operations can generate cheap prices, but not all cities are major cargo centers.
Silver doesn’t get everything wrong, though:
Passengers in areas far from large cities — like La Crosse, Wis., or Minot, N.D. — tend to be the worst affected.
Indeed, Northwest Airlines execs used to say about this part of the country, “It’s cold, dark, and nobody wants to go there. But it’s all ours! (Cue “brouhahahaha!”).”
But even airports like the one nearest to where I grew up in Lansing, Michigan (about $117 overpriced) can go through a death spiral of sorts. Lansing is within a reasonable driving distance of Wayne County Airport outside Detroit, so relatively few people will choose to fly from Lansing unless its fares are competitive with Detroit’s. The airlines, rightly or wrongly, may take that trend to signify a lack of demand, and may cut service, with further price increases on the remaining flights. Before long, the only passengers who regularly fly out of an airport like Lansing are those who are extremely insensitive to price, creating a semi-stable equilibrium of limited but very expensive service.
Ding ding ding. You’re one of the few folks who wants to fly out of Lansing, rather than take advantage of the huge economies of scale out of nearby Detroit, you’re going to pay a premium to do so. Silver’s moral model, though, doesn’t really allow him to say that this is actually quite ‘fair’ instead of ‘unfairly priced’ based on distance and abstracting away from connections.
Indeed, despite the use of statistics, the exercise appears to reveal far more about Silver’s biases than it does about the way in which airfares are constructed.
Ultimately the important things to understand about airfare from a policy level is that even with a sense that airfares are currently “high” (and jet fuel sure is ‘high’!), they’re far lower than they were pre-deregulation… both on an inflation adjusted basis and in many cases even in nominal dollars. The beneficiaries have been consumers, not airlines, and certainly not airline shareholders as airlines have been wildly unprofitable over the past decade.
Some markets are more expensive than others, driven by the mix of passengers, barriers to entry, factors that generate delays and additional costs like congestion and weather, and the cargo environment. Few can claim airlines have been making “unfair profits” for the past decade, even if they can establish a better ‘moral’ theory of airline pricing than Silver does.
Instead of worrying about which airfares are “too high,” recognize that they’re lower now with airlines in control of pricing than they ever were in the regulated era when central planners were in charge of pricing. And recognize further that markets with significant subsidies for their air service tend to be on the higher side as well, there’s not a very strong empirical case for intervention. No matter what kinds of cute things you can do with statistics and with the imprimatur of the New York Times…