A Prisoner at 30,000 Feet

My beautiful wife is quite the bargain hunter—she excels at finding the lowest cost source of anything on the Internet.  So we were thrilled the other day when she found fares from DC to LA over the coming Thanksgiving holiday weekend that were about $100 per seat less than the next lowest airline.  She quickly booked the tickets and learned, only at the very end of the reservations process, that this particular airline (AirTran) charges $6 extra per person per segment in order to reserve a seat.  Suddenly, our family of five was paying $60 above the advertised price just for the privilege of sitting in seats 14A-E on each flight.

We all know how crowded Thanksgiving flights can get.  So we had a good chuckle at the thought of what would happen if everyone travelling on AirTran that day refused to pay this $6 per seat extortion.  It is in the airline’s best interest that everyone have an assigned seat so that boarding the airplane will be more orderly (Southwest Airlines is the exception that proves this rule).  If all of the passengers refused to pay extra for the privilege of choosing their own seat, AirTran would have to do it for them and at a cost. 

So, how is AirTran able to get its passengers to pay to reserve a seat while the passengers have an incentive not to pay?  Because each individual passenger faces what in economics is known as a classic Prisoners’ Dilemma.  The Prisoners’ Dilemma is an example of a problem from game theory in which two (or more) players, unable to cooperate, each individually making rational, profit or utility­–maximizing choices, realize a worse outcome than they would have achieved had cooperation been permitted.  The standard example is of two prisoners who would suffer only a small penalty if they would each keep their mouths shut.  But, because of some promised benefit of confessing if they are the only one confessing, both confess and doom themselves to a harsher penalty than if they had kept quiet.

In the AirTran example, the passengers would all be better off if they could agree to not pay the reserved seat fee.  However, each individual also has an incentive that is contrary to the best interest of the group.  The first passenger to go against the group and pay the reserved seat fee gets choice seats.  Further, no one wants to be the only passenger not to have reserved seats—that could get you stuck in a middle seat, or worse!  Consequently, everyone (or at least everyone who cares) pays the reserved seat fee.

The payoff matrix below gives a visual example of a typical two-player Prisoners’ Dilemma-style game and the payoffs earned by each player.  In the game, two travelers who do not know each other and are unable to communicate are separately asked to pay a fee for a reserved seat.  If both travelers choose “Don’t Pay” the airline assigns them seats at the airport for which the traveler derives no extra benefit but also incurs no extra cost (net value = 0).  If one traveler “Pays” while the other does not, the paying passenger is rewarded with a good seat (value = +8) but must pay $6 for it (net value = +2).  Meanwhile, the non-paying passenger is stuck with a crummy seat next to the lavatory (net value = -10).  If both choose to “Pay” then both pay $6 and are randomly assigned a seat (net value = -6). 

 

 

 

		Traveler A
		Pay	Don’t Pay
Traveler B	Pay	A = -6 B = -6	A = -10 B = 2
	Don’t Pay	A = 2 B = -10	A = 0 B = 0

In this setup, each traveler will strictly prefer to pay the reservation fee no matter what the other traveler chooses.  Consequently, both will choose to pay the fee even though each would be better off if neither paid it.  

Notice, however, that this outcome depends crucially upon the set up of the game.  If we were to change the values the travelers place on each choice, allow the travelers to communicate, or change the number of times the two travelers come across each other in the same situation, the results of the game may change dramatically.  Under certain circumstances (for example, allowing the game to be repeated but not allowing the players to know when it will end), it is possible to have an outcome where neither traveler ever pays the seat reservation fee.

For an expert witness, game theory models like the Prisoner’s Dilemma may be very useful for describing how people might behave in simplified versions of real-world settings.  However, the assumptions made about the game’s set up must closely fit the facts in evidence for a game theory model to have any value or credibility.  The game in my matrix is very similar to my actual experience with AirTran.  Yet, I’ve made an assumption about the value that travelers will place on a “good” seat.  If instead, some travelers don’t consider any seat to be “good”, then the net value of “Paying” when other travelers “Don’t Pay” may be very different, possibly even negative.  This would make the outcome of the game opposite of the outcome the resulted from my assumption.  Making assumptions that are supported by facts is the only way to avoid the perception that your assumptions drive the results.

Meanwhile, we sucked it up and paid AirTran the $60 to have reserved seats.  I’m not happy about it.  Still, I figure that it beats showing up at the airport with three young kids the day before Thanksgiving and having us all spread around the plane in various middle seats.  But wait!  That’s a different payoff matrix…

Update: Immunity Challenges

To update my post from two weeks ago, today’s Wall Street Journal is reporting that the Department of Transportation (DOT) has approved the plan for Continental Airlines to join United Airlines’ Star Alliance code-sharing venture on certain international routes.  The Department of Justice (DOJ) previously objected to this code-sharing arrangement on antitrust grounds.  DOT appears to have limited the alliance in response to some of DOJ’s objections, but did not go as far as DOJ had requested. 

Immunity Challenges

One of my favorite features of reality TV shows is the competition between contestants for “immunity” from the voting process in which someone gets kicked off the island, learns they aren’t the biggest loser, or goes home without the bachelorette.   Immunity, as far as reality TV is concerned, means that whatever questionable behavior the contestant committed during the week becomes moot as he or she becomes untouchable, at least for this week.  Back here in the real world, the concept of immunity, specifically, antitrust immunity, made headlines today.

The Wall Street Journal reports that the DOJ’s Antitrust Division is recommending that DOT deny antitrust immunity to the proposed addition of Continental Airlines to United Airlines’ Star Alliance code sharing network.  United and Continental had applied for antitrust immunity for code sharing on a number of international routes.  DOJ recommends that DOT limit the number of routes covered by the expanded Star Alliance.  The agency seems to be concerned that the Star Alliance members would be able to limit competition and raise prices on certain routes in their proposal to DOT. 

What’s the big deal?  Back in the 1990s, airlines began using code sharing agreements in lieu of mergers to achieve operational efficiencies.  In a code sharing arrangement, two or more airlines will separately sell seats on each others’ flights.  For example, Continental has a hub in Houston.  United has a hub in Chicago with non-stop flights to Munich.  Continental only flies from Houston to Munich via New York and Amsterdam.   Yet, by code sharing, Continental can book passengers on any of United’s flights, giving the passengers more options.  A passenger may purchase a ticket from Houston to Munich with a stop in Chicago from Continental, but fly on United planes. 

The airlines derive several benefits from code sharing arrangements.  First, code sharing allows the carriers to offer larger networks to their passengers without the costly expense of expanding.  As an additional benefit, by expanding volume over the existing network, airlines are able to fill more seats, eliminating excess capacity.   

Another more important benefit is that the U.S. Department of Justice (DOJ) and Department of Transportation (DOT) grant limited antitrust immunity to those code sharing arrangements they approve.  Approved alliances are allowed to share certain sensitive data about the shared routes, such as prices, without being accused of illegal price fixing.  Code sharing is not feasible without this immunity.

Back in the 1990s and through the recent Bush administration, DOJ was concerned about airline mergers reducing competition on many routes.  Consequently, airline mergers were almost never approved.  Even with both airlines undergoing bankruptcy troubles at the time, DOJ challenged a proposed merger between United and USAirways in July 2001.  Code sharing is an alternative to merger—­a way to achieve some of the efficiencies a full-blown merger might have produced but without the potential anticompetitive consequences.

Today, however, it seems that even code sharing will be frowned upon under the new administration.  It is unclear whether this is a sign of stricter antitrust enforcement or whether DOJ’s recommendation is isolated to this particular case.   Either way, DOJ’s recommendation is backed by economic analysis.

DOJ has likely looked at the number of airlines independently serving the routes in question and try to determine whether prices on the route would likely rise due to a Continental/United alliance.  Presumably, DOJ has also tried to determine whether other airlines could easily enter those routes should prices rise.  DOJ has also probably looked at whether Continental and United would create substantial cost savings through their code sharing alliance. Further, the economists at DOJ might have looked to see if similar arrangements on similar routes resulted in fare increases and, if so, they could use those results to forecast what would likely happen after Continental and United’s agreement took effect.

There’s still a chance that this code sharing agreement will be allowed.  DOJ has only made a recommendation.  DOT has the final say in approving the agreement.  How will DOT vote? Stay tuned! 

Teaching Your Kids About Safe Sexting

Last fall I was watching a football game on TV with my sons – ages 8 and 7.  During a commercial break, a public service announcement came on that repeatedly admonished me to talk to my kids about sex.  As soon as the ad was over, my sons turned to me and asked “What’s sex?”  Thankfully, I was a quick thinker that day.  I said “Sex is another way of saying ‘boy or girl’.  If someone asks what sex you are you say ‘boy’ or ‘male’.”

Thankfully, my sons don’t read the newspaper past the sports and comics sections or they would have recently seen articles about “teen sexting,” the sending of sexually explicit pictures attached to text or e-mail messages by teenagers.  A recent study found that 20% of teens sent nude or semi-nude photos using their cellphones or e-mail. 

More recently, an article by “The Numbers Guy” in the Wall Street Journal questioned that study’s findings.  One criticism was that the study defined “teen” to include 18 and 19 year-olds (legal adults), as well as younger teens.  Turns out the younger teens were half as likely to share provocative pictures of themselves.  The article also raises questions about whether the study is truly a representative sample of all teenagers.

Ideally, a survey relies on respondents from a representative or random sample of the group or population of all possible respondents.  A randomly drawn sample will not be perfectly representative of the group it purports to represent because of sampling error.  Natural variation in selecting the sample at random will occur.  For example, if the population is 10% African-American, the random sample may have a larger or smaller proportion of African-Americans.  Ideally, the sampling error will be small and unbiased (meaning that if you repeat the sampling procedure, you will not consistently over-represent or under-represent African-Americans in the sample).

In the case of the sexting study, the sample may not have been a truly random sample of all teenagers.  The sample they used may have been what is called a convenience sample (despite the authors’ best intentions).  The authors of the sexting study got their sample of respondents from an on-line panel.  This panel may over-represent teens who spend a lot of time on-line and under-represent teens who do not.  Even if the study authors selected their sample at random from their on-line panel, they may not have a representative sample because their sample would systematically leave out teens that do not use the internet or mobile phones as frequently as those that participate in the on-line panel.  Yes, it is hard to conceive of a teenage these days that does not spend a lot of time on-line or on their cellphone, but that stereotype may be what led the study authors to feel that their on-line panel was sufficient.

Now convenience samples are not necessarily bad.  They may be very informative.  Sometimes, they may be the best sample you can get.  A random sample can sometimes be expensive to obtain.  Depending upon the nature of the question you are trying to answer, and the budget you have to get that answer, a convenience sample may be sufficient. 

The trick is in knowing what kind of sample you are using and the limitations that places on what conclusions you can draw.  For example, sexting may not be as big a problem as the study would have us believe.  However, the study still raises legitimate questions about what children are doing on-line.  If you want to discuss appropriate on-line behavior with your teenage son or daughter then it probably doesn’t matter to you whether one-in-five or one-in-ten of their peers are sexting.  But, if you want to make or evaluate public policy decisions that would affect how teens interact on-line, then having a study based on a random sample is more important.

I use both random samples and convenience samples in my consulting work.  In each case, I make sure that my clients and I are aware of the benefits and limitations of the type of sample I choose to use.  Having this knowledge lets my client and me draw appropriate and defensible conclusions from our analysis of the sample.  Consultants that ignore the distinction between random and convenience samples place their clients at risk.  In litigation, ignoring the distinction may be all it takes to get an expert witness’ analysis excluded by the court.

In my role as a parent, I used a convenience sample of commercials viewed during a football game one autumn day to make an important policy decision.  I now Tivo all football games that my children and I want to watch so that I can fast-forward through all of the commercials.  We’ll have a more detailed discussion about that public service ad when my kids are a little older.

Bowling for Senators

In 1998 my Alma Mater, Tulane University, had an undefeated football season, won it’s bowl game, and finished ranked #7 in the nation.  They were not considered for a Bowl Championship Series game and precluded from competing for the mythical national college football championship.  Several other “non-BCS” schools have since similarly gone undefeated and yet been shut out of a national championship game.

 On Wednesday, the Senate Judiciary subcommittee on antitrust announced hearings on whether the Bowl Championship Series violates U.S. antitrust laws.   Here’s an interesting question that the Senate subcommittee should consider:  Who exactly has been harmed by the BCS?

Here are a list of candidates:

Non-BCS Schools— These are, in a sense, the BCS members’ competitors.  They are in practice excluded from the BCS championship game because the human and computer polls that the BCS uses to rank teams appear to be biased against these schools.  However, the BCS rules do allow non-BCS member schools to participate in BCS bowl games under certain, unlikely circumstances.  So, is it the BCS’s fault that non-BCS schools haven’t qualified?  And what are the damages?  Have the non-member schools received reduced bowl game revenue, reduced TV appearances, and fewer recruiting opportunities?  Would these schools be better off under a playoff? 

Note that the alternative to the BCS is not necessarily a “March Madness” type playoff system.  The alternative could be to revert back to the bowl system of the 1970s or 1980s.   Would non-member schools be better off under the old system?

Other Bowl Games—  These too are competitors to the BCS.  It used to be the case that bowl games would bid for unaffiliated or uncommitted teams to play in their games.  Over the years however, to reduce the risk of undesirable match-ups, more and more bowls have committed to accepting the conference champions and runners up.  In the meanwhile, there seem to be more bowl games now than there were before the BCS. 

The BCS is simply a group of bowls that were all fortunate enough to have commitments from the major conference champions.  The BCS too has reduced risk and (arguably) offered a more desire able product by matching the top teams from their stable of schools.  Have other bowls been prevented from luring more desirable match ups because of the BCS?  Would they be better off under a playoff?  How would they have fared under the old bowl system?

TV Networks and Advertisers — Would the networks pay more or less money for a playoff?  Would advertisers pay more or less for a playoff?   Would the networks pay more or less money under the old system?  What about advertisers?

College Football Fans — How have fans been hurt?  Are ticket prices at the BCS schools higher under the system than they would be under an alternative system?  And are ticket prices for fans of Indiana University, a BCS member that has never played in a BCS bowl, affected the same as ticket prices at Ohio State, which has played in 3 BCS title games and is one of the dominant programs in the sport?  Are bowl game tickets higher or lower priced than tickets would be to playoff games?  And don’t fans actually enjoy arguing about who is number 1?

I’d love to see these questions asked by the Senators running the hearing.  Obviously however, the Senate hearings are political grandstanding.  The BCS is unpopular because it is a flawed way to crown a national champion.  But if the BCS is to be deemed anti-competitive, it must be done by demonstrating a competitive alternative.  What would the world look like if the BCS had not formed and would consumers be better off?  These are the questions that would be asked (and hopefully answered) by a true antitrust inquiry.

In the meantime, whether there is a playoff system or the BCS opens up more to non-member schools is sort of moot for Tulane fans.  In the ten seasons since going 12-0, Tulane has had only two winning seasons.  They were 2-10 last year.  It’s safe to say they will not be picked to win the national title this year.  Roll Wave!

Good Morning, Here’s the News

House Speaker Nancy Pelosi has proposed that newspapers receive some form of exemption from antitrust laws.  Now Attorney General Eric Holder has come out in support of this proposal.  Granting antitrust exemptions is something that should be approached cautiously.  In this case, it may not even be necessary.

Consider a town with two newspapers, one of which is losing money and could soon go out of business.  The other paper could purchase its competitor.  This transaction would likely pass the “failing firm” defense.  Mergers, even ones that would create a monopolist, are often allowed if one of the merging parties would go out of business because you would have a monopoly in either scenario.

Suppose instead of outright merger, the two newspapers enter into a production joint-venture similar to the one in Detroit.  The papers would share certain cost centers, like printing presses, but would maintain independent editorial boards and maybe even independent sales departments.  The two papers would still compete for circulation and ad revenue.  Given that there is precedent for such an arrangement, such a joint venture would also likely survive an antitrust challenge.

Finally, in nearly every antitrust case, a central point of contention is the definition of the “relevant market.”  The relevant market is essentially the market that is in danger of being monopolized.  A narrow relevant market – for example, the market for newspapers in San Fransisco — would imply that a merger between the two newspapers creates a monopoly.  A broad relevant market — for example, including all news sources such as paper, broadcast, and internet available in San Fransisco — would imply that a merger between the two newspapers might have little impact on the price of news and information.

Policymakers should consider whether newspapers compete with a wide variety of news sources, or just with each other.  The answer to this question might be found by considering why many newspapers are in such financial trouble to begin with.

One Flu Over the Cuckoo’s Nest

Despite receiving a flu shot back in October, this week I came down with the flu.  I came home from the doctor’s office on Monday with a armful of prescription medications.  The doctor was nothing if not thorough.  Problem is, I have no idea if any of these drugs actually helped me.  The side effects of one of them are exactly the same as my flu symptoms.  I’ve been sick for five days.  Would I have really have been any worse off if I hadn’t taken any of these medications? 

I will never know the answer to this question.  There is no way for me to know if I’d have been better or worse off if I had skipped the fancy prescriptions.  In my case, there is no control group with which I can compare what my condition would be without the prescribed drugs.

“Natural experiments” are what economists and other social scientists refer to when we want to test our theories using control groups that are created naturally in the real-world instead of in a controlled laboratory or classroom environment.  There is a field called experimental economics, where economists construct physical experiments complete with live test subjects and control groups that they create, to test economic theories.  However, many, if not most of the types of markets that I study in the course of business do not lend themselves to controlled laboratory or classroom experiments.  Instead, I try to find natural experiments.

Consider, for example, the possible effect on prices of a merger between two competing chain stores.  Before the merger has been consummated, it’s difficult to really know what the market will look like after the merger.  However, suppose these chains compete in some cities but not in others.  I can compare prices in cities where both chains have locations to prices in cities where only one chain or the other has locations.  If prices are higher in the cities where only one chain operates than in cities where both chains compete with one another, then I may infer that the merger would cause prices to increase.  If there is no meaningful difference in prices in cities where both chains compete versus cities with only one chain, then I may infer that the merger will have little or no impact on prices.

Of course, it’s not as simple as it sounds.  For a natural experiment like this to work I need to account for a host of other factors.  These may include the presence of other competitors and differences in demographics (such as population or median income) of the different cities in the experiment.  There may be other types of factors that are specific to the particular industry as well.  Further, the effects of these confounding factors need to be examined individually or collectively.  A natural experiment can be ruined, and consequently laughed out of court, if some factor independent of the merger turns out to better explain price differences than the presence of either of the merging chains. 

Unfortunately for now, I still don’t know whether all of the medications I’ve been taking have done any good.  For what it’s worth, my daughter caught the flu at the same time as I did but at 3-years old is too young for any of the medications I was given.  She’s been well now for a couple of days.  Naturally!

Are You Smarter than a Third-Grader?

The other night I attended an alumni event for Indiana University’s Kelley School of Business.  As I sat there, I couldn’t help but thinking about the professors I had in college.  Sure, some were quite good classroom instructors, but many had obviously not put much thought into how they presented their subject matter (these do NOT include any of my former professors who also read my blog).

Universities have historically assumed that people who are the leading researchers in their field are also the best people to convey their subject matter knowledge to students.  Consequently, universities invest little in teaching their faculty how to teach.  This is critical because the knowledge presented in the traditional lecture is not always presented intuitively, making it harder for the students to learn.

In a traditional lecture, a professor may first present a new term, give its definition, and describe a number of examples or counter-examples.  For example, in teaching a class about monopoly, a traditional lecturer may say a monopolist is the sole producer of a good, and having no competition, is able to charge a profit-maximizing price such that the marginal revenue of the next unit it produces equals the marginal cost of production.  To understand this definition, the students must understand what is meant by ‘marginal revenue’, ‘marginal cost’, and ‘profit maximization’.  They must also understand how this definition differs from a firm that faces competition.  This is not an intuitive explanation of monopoly.

Now, imagine how you would explain monopoly to a third grader.  This is something that I will actually have to do next Monday when I speak to my son’s third-grade class.  If I throw out terms like ‘marginal revenue’ and ‘marginal cost’ I would be lucky to hold the class’ attention as well as Ben Stein’s character did in “Ferris Bueller’s Day Off”. 

But, monopoly theory can be explained in a way to which a third-grader can relate.  For example, suppose you are the only seller of Hershey’s miniatures in Nottingham Elementary.  In fact, you are the only candy salesperson in the whole school.  What price would you charge for each piece of candy?  If the candy cost a nickel a piece, could you charge 10 cents?  a quarter?  a dollar?  If you charged a really high price, would anyone buy the candy from you at all?  Now suppose someone starts selling Snickers bars at the school.  Could you charge the same price and still sell the same amount of candy?  Why or why not?  Suppose instead of Snickers bars the other person also sells Hershey’s miniatures?  What if instead of chocolate bars, the other person sells broccoli?

Notice, this line of questioning requires no prior knowledge of economics.  It asks for intuitive answers to a situation that anyone, even a third-grader, can easily imagine or has previously experienced.  And because the answers are already part of the students’ imagination or experience they will feel much more comfortable when I explain that in this example they were a monopolist who was able to charge a high price when there was no competition, would have to lower its price when a close substitute was introduced, but may not have to change its price at all if a more distant substitute or non-competing product is introduced.  This is a much more effective way to communicate the idea of a monopoly.

This style of explanation also works outside of the traditional classroom setting.   I wouldn’t advocate speaking to a judge or juror in the same manner I would speak to a third-grader.  However, judges and juries cannot be expected to have any more exposure to economic theory or statistics than a third-grader, or a college freshman taking introductory courses for that matter.  It is the expert witness’ job to teach his audience what they need to know – and make sure they stay awake and engaged in the process.  If the most brilliant expert starts explaining monopoly to a jury by using the terms ‘marginal revenue’ and ‘marginal cost’, he will not be persuasive.  However, an expert that can get his audience to relate his testimony to their own experiences will be highly effective.

We all learn better when we can relate to the subject matter. So, how will I do explaining monopolies to my son’s third grade class?  I’ll let you know Monday. Mental note:  make sure to bring lots of chocolate bars — that will keep their attention!

Update: Stephan gets results!

On Monday I wrote about how I was asked to pay $2 for a can of soda on my flight between DC and Phoenix.  Later that day, US Airways announced that it will no longer charge for soft drinks on any of its flights beginning March 1 (here’s the press release).  How’s that for influence!  (Thanks to my LECG colleague Al Bremser for the tip!)

US Airways’ change of policy is a great example of competition at work.   Of course, if you want a pillow and blanket on your flight, you will have to pay $7 for the “Power Nap Sack“.  Happy flying!

Substitutes and Complimentary Drinks

As I write this, I’m sitting on an airplane from Phoenix to Washington.  The airline on which I’m flying has a policy of nickel-and-diming.  They charge extra for checked bags.  They offer certain open economy seats for sale (only $15) when you use on-line check-in, although there is no difference in legroom between the seat you were first assigned and the $15-extra seat.  However, the airline’s policy which generates the biggest complaint from the passengers around me is that it now charges $2 per can for what had previously been complimentary soda.

As a passenger who would like to enjoy a complimentary Coke Zero, I loathe this policy.  As an economist, I hate to say, I have to admire it.  After years and years of flying, the airline finally figured out that, once the plane is in the air, there are no substitutes for its formerly complimentary beverage service.

Now, I have no idea whether this airline is actually making a profit from this change in policy (although the profit-margin on a single can of soda is probably $1.75 or more).  There are still a number of passengers who will buy a beverage from a vendor at the airport before boarding.  But once you are in the air, they’ve got you.  In the air, they’ve got market power.

Market power is the ability to charge you $2 for a can of Coke Zero that would cost you 50 cents from the hot-dog vendor on the corner of 17^th and I Streets.  The hot-dog vendor, sells me a can of Coke Zero for 50 cents because, if he charged more, I could easily walk across the street and buy the same thing from a different vendor.  All of the street-cart vendors in DC have arrived at a competitive stalemate, and therefore all charge about the same price for soda (I hear that prices are higher across from the White House but that’s only because the tourists don’t know they can walk an extra block or two to visit my vendor).  My airline, however, has no competition once the plane is in the air.  The airline knows that it will not lose many passengers by charging for soda that it used to give away. 

In antitrust litigation, assessing whether a firm has market power is not always clear cut.  It requires some investigation and analysis.  This same airline may not have (much) market power when it comes to selling tickets on its flights.  Other airlines (such as Southwest) also fly non-stop between Washington and Phoenix so that I probably paid a competitive fare for my ticket.  However, flying between Washington and Charlotte, or Washington and Atlanta may be a different story.  Those routes are only served non-stop by one airline and it’s not easy (for a host of reasons that I’ll save for another day) for other airlines to add flights in Atlanta or Charlotte.

These factors – the number of competitors or substitute products, and the easy of entry into the market by new competitors or substitute products – are important in determining whether market power exists.  I might compare prices in one market with few substitutes and difficult entry to prices in another market with many substitutes and easy entry (holding everything else constant, of course) to see to what degree these factors (as opposed to others) explain a difference in prices between those two markets (this is the benchmarking analysis I mentioned in my earlier post).  I could also look to see if, over time (again, holding everything else constant), entry and exit into the market I’m examining had any effect on prices (this is the before-and-after analysis from my earlier post).  If prices are not very responsive to the entry and exit of new products or other firms then that is a good indication that market power may exist.

Of course, knowing all of this doesn’t change the immediate dilemma facing the passengers on my flight.  Thankfully, I bought a bottle of soda before boarding!