More From the “Economic Naturalist” Robert Frank

We recently posted a series of excerpts from The Economic Naturalist, a new book by the Cornell economist Robert Frank (who has another new book out this week, Falling Behind, a brief treatise on income inequality). Because the Economic Naturalist excerpts were well received and vigorously debated, we asked Frank if he would reply to some of the feedback. Kindly, he has obliged:

Guest Blog: Robert H. Frank

When I describe my “economic naturalist” writing assignment to students, I stress that it is not important that the answers to the questions they pose be correct beyond doubt. Far more important is that the questions themselves be interesting and the proposed answers economically plausible. The learning stimulated by this assignment stems less, I think, from the writing of the papers themselves than from the animated discussions provoked by the questions.

I was therefore extremely encouraged by the lively reader responses to the examples from The Economic Naturalist on this blog recently. I was encouraged, too, that a Google search a few days after the post ran turned up over 100 other web sites that had linked to it. To my eye, that’s the real beauty of the writing assignment: Once students manage to pose an interesting question, they immediately want to discuss it with others. And in the process, they find endless opportunities to refine their thinking about what a sensible answer might look like. In short, they learn a lot from these conversations, just as I did from reading your comments.

Such exchanges also provide valuable opportunities to push back, to probe the power of opposing views. So I am pleased to take advantage of an invitation to respond to some of the criticisms of my students’ explanations.

McMansions for Retirees: Several respondents objected that the phenomenon to be explained – that retirees are increasingly buying large houses close to home rather than smaller condominiums in the Sun Belt – was a statistical artifact. In the book, I cited studies purporting to confirm the trend in question, but I’m quick to concede that the practice may not be widespread in many areas. It does seem clear, however, that supply and demand in the retirees’ housing market have shifted in precisely the way my student Tobin Schilke described. Because the number of births per adult American woman has remained roughly the same for several decades, the number of children is no greater now than in the past. Yet because of the secular rise in divorce and remarriage, each child now has more grandparents than in the past (on the plausible assumption that we count the parents of step-parents as grandparents).

The upshot is that the demand for visits by grandchildren has increased relative to the supply of such visits. If we grant Mr. Schilke’s plausible assumption that having a large, conveniently located house makes visits more likely, it follows that retirees are more likely to demand such houses.

Of course, there may have been other offsetting changes in the retirees’ housing market. Rising energy costs, for example, may have reduced the demand for large houses. But that wouldn’t challenge Mr. Shilke’s interesting observations about how demographic changes appear to have altered the demand for grandchild visits.

Square Milk Containers: Regarding the proposed explanation that milk containers have square cross-sections in order to minimize the amount of costly shelf space they occupy in refrigerated storage units (in contrast to the cylindrical containers of soft drinks, which are typically stored on unrefrigerated shelves), several respondents pointed out that containers with square cross-sections could not contain the pressurized contents of carbonated soft drinks unless their walls were so thick as to make them prohibitively costly. It’s a fair point.

But a milk container of given volume could also be produced at lower cost if it were cylindrical in cross-section rather than square. Relative to a container with square cross-section, however, a cylindrical design would definitely increase the cost of storing milk on refrigerated shelves. So it seems fair to conclude that the cross-section of milk containers is dictated at least in part by a desire to minimize the cost of refrigerated storage.

Premium Prices for Black MacBooks: When its newly introduced black iPods quickly sold out in 2005, Apple discovered that customers would be willing to pay premium prices for a machine in a previously unavailable color. So when it brought out its new MacBook models the next year, it posted a higher price for the black version and had no difficulty selling them.

Many respondents apparently mistook me to be saying that Apple was somehow exploiting its customers by charging the premium. But the central point of the example was exactly the contrary. Whenever a seller produces under economies of scale, it is always possible to create additional economic surplus for all parties -buyers and seller alike – by using what I call the “hurdle” method of price discrimination.

The basic idea is that the seller offers a discount only to buyers who are willing to jump some sort of hurdle, such as mailing in a rebate coupon or settling for a machine in a less desired color. These discounts increase the number of units sold, in the process reducing the average production cost per unit. The resulting cost savings often make it possible for even buyers who pay full list price to end up paying less than they would have if the product were sold to the same price to everyone.

Although some complain that it is unfair to charge some buyers more than others for essentially the same product, in The Economic Naturalist I argue that Apple’s pricing scheme actually appears to mete out a certain rough economic justice. This will be true if, as appears plausible, the buyers who are willing to pay extra for the black machines are also the ones who value the company’s innovative design features most highly. After all, somebody has to pay for Apple’s prodigious research and development costs. Why shouldn’t these costs fall more heavily on those consumers who care most about cutting edge design?

Gas Caps on the Right and Left Side of Cars: In response to her question about why fuel filler doors are sometimes on the left, sometimes on the right (causing confusion for rental car drivers), my student Patty Yu argued that if filler doors were all on the same side (say, the driver’s side), lines at the gas pumps would be much longer during peak periods. Numerous respondents suggested other possible reasons for filler door placement. One pointed out, for example, that manufacturers tend to put the filler door on the side opposite the muffler and tail pipe, perhaps to minimize the odds of gasoline spilling onto a hot pipe during an accident. Click and Clack discussed this hypothesis on Car Talk recently, noting that, although the correlation exists, it is far from perfect.

They also mentioned a variant of another respondent’s observation that European manufacturers tend to put the filler door on the passenger’s side, perhaps to minimize the danger to a driver who runs out of gas and must add fuel to his tank while stopped at the side of a highway. Their variant was that manufacturers in countries that drive on the right tend to put filler doors on the passenger’s side, thereby to keep them farther away from shearing forces in head-on collisions. And indeed, Japanese cars do tend to have their filler doors on the left (drivers in Japan, like those in the U.K. and Australia, drive on the left side of the road). Here again, though, there are many exceptions. (My Miata’s filler door is on the left, but my son’s Subaru’s is on the right.)

By far the most common objection to Ms. Yu’s explanation was that it seemed to presume a conscious attempt on the part of manufacturers to coordinate their fuel-filler door placements – something for which there is no evidence. It is this objection that I find most interesting from a methodological perspective. Suppose manufacturers had not, in fact, coordinated their efforts in an explicit attempt to minimize the queues at gas pumps. Would that make Ms. Yu’s explanation any less plausible?

If one views product design features in an evolutionary perspective, the answer is clearly no. Darwinians argue that useful features evolve from random mutations. The eye, for example, developed from a sequence of random mutations because light-sensitive organisms were better able to locate valued objects and avoid harmful ones. The whole point of the theory is to explain how eyes came to exist without anyone having consciously planned them.

A similar point applies to evolutionary explanations in economics. If all manufacturers had happened to place fuel filler doors on, say, the left side of the car, one consequence would have been long gas lines during peak hours, because drivers in most countries would pull up on the right side of the pump. And in that case, manufacturers would have had a problem worth addressing. Ms. Yu’s explanation thus helps explain why the observed distribution of filler door placements is evolutionarily stable. Evolution, as Richard Dawkins once observed, is less aptly described as “the survival of the fittest” than as “the survival of the stable.”

In Summary: In telling my students that their answers don’t have to be the final word, I’m not saying that it’s not a good thing to be right. Rather, my point is that students are more likely to engage with our subject if we demonstrate that it can stimulate them to think about the world in interesting new ways.

I’ll mention another piece of evidence that the questions they pose meet that test with flying colors. Several hours after I had discussed a couple of examples from The Economic Naturalist in a brief interview on NPR earlier this week (“Econo-reasoning behind everyday things,” Marketplace Morning Report) a listener copied me on this e-mail in which he posed a long list of economic naturalist questions of his own. Some examples:

1. Why do phones and calculators/computers have different number pads? To wit:



2. Why do hockey games have 3 periods rather than 2 halves or 4 quarters? And why are points used to determine standings, rather than straight won-loss percentages?

3. How are railroads able to use freight cars that belong to other railroads? United doesn’t fly jets belonging to Southwest — so why should Burlington Northern let Norfolk Southern use its freight cars?

4. Why is whiskey sold in fifths?

5. There’s a metric scale for measuring just about everything — weight, distance, volume, even temperature (Celsius is derived from the metric system) — except for one thing — time. How come there’s never been a metric calendar/time system, with, say, 10 metric months of 10 metric days each, each metric day composed of 10 metric hours, each metric hour composed of 100 metric minutes, and each metric minute composed of 100 metric seconds (which would be different from the seconds currently used)? (I’m surprised that countries that use the metric system have no problem with the “non-metric” way we measure time).

6. Why is Newfoundland a half hour different from other time zones?

7. Why don’t doctors dispense medicine or employ pharmacists in their offices, so we can have one-stop health care and save a trip to the drugstore?

8. Why don’t cell phones have dial tones?

Pat McGee

On #1, I started studying user interfaces in the early '80s. We asked the same question. At an early CHI conference, someone from Bell said that it was to keep people from keying in the numbers too fast for their circuits to recognize them. Of course, in the years since, touch tone recognition technology got better. But, we're still stuck with that backwards keyboard.

(On a related note, after some research, I found that the similar story for the Sholes (qwerty) keyboard arrangement is a canard. The arrangement survived a brutal Darwinian contest against other keyboard arrangements and won because it was superior. The Dvorak keyboard came much later, and lost because it was too little too late.)



Re: evolutionary stability as an answer to the gas cap question. I think it's a poor answer, because for evolution to work, the different outcomes have to affect the ability of the organism to survive and reproduce. The equivalent for a car would be the ability of that car's design to attract buyers (thus enabling the design to be continued). I see no reason to think that car buyers take the equalization of gas cap location into consideration when buying a car (i.e. "Well, it seems about 55% of cars have their gas caps on the right, but THIS car is one of the 45% that has its gas cap on the left, so I will save a few seconds at the gas station once in a while. Therefore, I will buy this Ford instead of that Toyota").

A much simpler answer is that manufacturers put the gas cap on whichever side of the car is cheapest/most convenient for them.

I could certainly be wrong, but I think my answer is more plausible (and simpler) than the evolutionary stability answer.

As for the new questions...

I think for several of them, the answer is most likely "random chance". As systems evolve independently, and in the absence of any sort of precedent, one design choice can often be very nearly the equal of another. Designer A chooses choice A, designer B chooses choice B, and as the systems grow, those choices get locked in.

When railroads were being developed, there was no magic answer for the correct guage (width of rail separation). Rail networks were independent and not interconnected, so each designer chose a gauge based on idiosyncratic logic or guesswork (more or less). Eventually, networks grew to meet each other, and it became important to have a common standard, but switching to a common standard was expensive for those with many miles built to a different standard. For the most part, railroads swallowed the cost and standardized. But in Europe, Russia and Finland remained on a 1520 mm gauge (slightly under 5 feet), rather than switching to the standard 4' 8.5" gauge, creating inter-connection problems ever since.



Anyways, answering the actual questions... My best guesses:

1) Random chance (see my post above).
2) Random chance (see my post above).
3) First, freight cars aren't nearly as valuable (or maintenance intensive) as jets. Second, the cost of transferring cargo from freight car to freight car is relatively high. Third, our national rail network is relatively fragmented (and was even more so until relatively recently), and shipping coast to coast often requires using the rails of several different companies. Freight loading/unloading costs would become very high if cars weren't allowed to cross networks.

IIUC, locomotives (the expensive part of the train) generally stay on their own networks - it's only the relatively low value cars that change around.

4) Random chance (see my post above).
5) I've read that there were attempts at metric time, but they failed.

I suspect a metric calendar would always be a failure. Unlike units of distance, weight, and so on, where the units we choose are relatively arbitrary, for the calendar, our units are linked to things that make sense - a year is a very obvious dividing point, as is a day, and to a lesser extent, a month. A week is based on a 7 day cycle with a day for worship, that is heavily embedded in each of the major Western religions (though I'm curious if there's anything like that in the major Eastern religions). So time periods from a day to a year are based on systems unlikely to change for various reasons. Now, a day COULD be more logically broken down into a decimal system, but I suspect that inertia beat out attempts at metric 'reform'. Perhaps, given that the higher end units of time are clearly resistant to change reduced the incentive to change the lower end units (A half-decimilized system isn't much better than a non-decimilized system).

6) No idea. I wasn't even aware of this.
7) Hospitals do this already, more or less. As for regular doctors with offices - there's economies of scale. The inventory cost of a full pharmacy is high. IIUC, there are many more doctors with offices out there than there are pharmacies. Plus, consumers would likely prefer pharmacies with more convenient hours for refills and whatnot to doctors that are generally only open 9-5 or so.
8) I'd like this myself (dial tones on cell phones). My guess is that it is a cost issue, relating to when cell phones were newer and there were fewer available connections (to the cell tower, within any given area). I believe a dial tone indicates an open connection to the phone company. Keeping that connection alive for the ~10 seconds it takes to dial a phone number increases the average outbound call length by ~10 seconds.



Calculator & phone number pads are different because of research done at Bell Labs 50 or more years ago.
I don't remember the reason, but it was determined from testing that making them the opposite was important. I don't believe it was to slow you down, though.

Other research from Bell Labs discovered that the phone companies do the numbering wrong for us to be able to memorize the phone number.
Instead of the current way which is the three digit exchange first & the four digit number last, it should have been the opposite. I believe it's because we tend to use only a couple of dozen exchanges for regular calling, but the four digit section is random.


Why are freight cars interchangeable?
The same reason every different phone company can call another.
Just think what it would be like if no railroad could switch cars around.
Factories would have tracks from a dozen railroads would have tracks on top of each other.
That's pretty much what happened with the Chicago Stockyards.
Every RR went into there.

The reason the US uses 4'8.5" as the standard is that the US Government dictated that for the transcontinental railway.
Most of the southern states used a five foot gauge until after the Civil War.


Q. "Why don't doctors dispense medicine"?

A. There are too many drugs for every doctor to carry. That answer is simple!


Q. "Why don't cell phones have dial tones?"

A. Because the dial tone is created at the central office & I think it would kill the battery if there was one. As it is, cellphones are always checking which cell they are in, otherwise you couldn't receive a call.
The phone MUST KNOW where you are to deliver the call.



Newfoundland wasn't part of Canada until the late 1940's. Until then it had been a separate British dependency. I would presume that the time-zone situation is a remanant of that status in some way.


As for metric time, the French Revolutionaries tried it. They had a ten month calendar with new names for the months, I believe there were 100 minute hours & ten hours to the day, etc.


First of all, what we call a gasoline pump is actually a gasoline dispenser. The pump, like most pumps, is underground and thus not visible to the consumer. Second, most modern gasoline dispensers are two-sided, and each side dispenses the same varieties of gas at the same price. If everyone's gas cap were on the left side of the car, half the cars would approach the two-sided dispenser in, say, a northerly direction, keeping the dispenser on the left side of their car as they approached. The other half would approach this same dispenser in a southerly direction, also keeping the dispenser on the left side of their car. It would actually be a lot less chaotic than the current system. The station I usually go to offers a roughly 10-cent/gallon discount on Wednesdays, and they're only open from 9 to 6. If you guessed they're packed to the gills with cars all over the place from 5 to 6, you'd be right. The fact that different cars have different-sided gas caps makes it even worse.
One solution might be to offer longer hoses and make everyone approach the dispenser with the dispenser to the left (or right - no matter as long as you're consistent) of their car, then use the longer hose to reach around the car where necessary. Or have the hoses anchored to an overhead canopy so that both left- and right-sided cars require some stretching.



Cell phones? Because they are not connected 100% of the time like land lines.
They connect to the phone system when you hit 'send'. There's a graphic indicator telling you that you have signal, so you don't need a tone.


Time is measured intuitively on circular dials. A 1/10th wedge of a circle is not intuitively identifiable. Identifiable wedges are 1/4 and 1/6 (right angle, and angle on an equilateral triangle). Unfortunately for metric time enthusiasts, 4 and 6 are not factors of 10. The pros of measuring time with visually intuitive angles outweigh the cons of using a non-decimal system.

This issue doesn't come up for distance or weight etc. because such quantities are not cyclic. See comment above about astronomical cycles of years and days.


Venezuela is also on a half hour different time zone--so that eliminates the ties to the British answer. In both cases that makes the reported time closer to the solar time than it would be if they stuck to the full hour. Also, both are relatively isolated with a low level of economic and other direct cross border activity so there is no strong pressure for the time to match the time in adjoining countries-providences.


Milk cartons are square because they're cartons, made from covered paper. The machine Tetra Pack introduced to automate packing could take a flat preprinted and stamped sheet, fold it into the square shape, have it filled, then close the top in a quick, smooth movement. And that depends in part on the ease of folding sheets with straight folds, and on creating a container with few "open" edges that need to be glued, especially on the bottom end. A cylindrical carton is harder to make, harder to automate and more likely to leak since it needs a glued rim along the bottom.


Re: calculators

Calculators are evolved from adding machines, which traditionally had digits in columns, starting with 0 at the bottom and working up to 9 at the top. When the electronic calculator was developed, it was considered important to keep 0 next to 1, and have the numbers increase as you went upwards.

Telephone keypads are evolved from rotary phone dials, which had 1 at the top, and 0 at the bottom, after 9. The 0 was thus effectively more like a "10": When you dialed 1, the phone would click once, and when you dialed 0, the phone would click 10 times. When touch-tone dialing was developed, it was considered important to keep 0 next to 9, and have the numbers increase as you went down, like they did on the phone dial.

Residents of New York City might be more aware of this than folks from elsewhere, since the phone company charged extra for the touch-tone "service" until the mid-'90s, so most people still used pulse dialing even if they had a touch-tone keypad. Dialing a number with a lot of 9's and 0's in it could take 20 seconds or more. (Incidentally, this is why New York has an area code of 212 - the fastest code to dial on a rotary phone, for the city with the greatest population.)



Re: Hockey

In between hockey periods, the ice must be resurfaced (using the greatest of all machines, the Zamboni). Re-surfacing the ice takes a long time. There's little difference between ice that has been used for 15 minutes (standard football quarters) and 20 minute (current hockey quarters). To save money & time, they moved from four periods to three.

No clue about the win-loss vs. points reason :)


1. Why do phones and calculators/computers have different number pads?

Back when all (US) telephones were supplied by Bell, I had a sideline re-arranging touch tone keypads in 'calculator' order, mostly for accountants and others who used their calculators more than their phones.

It was a great gimmick and much appreciated by the users. It was a bit tricky, and took about half an hour per phone.

I also would occasionally do it to random phones in the executive offices without being asked. Lots of puzzled folks the next day. (-:


Regarding the previous responses on cell phones: take a look at cordless phones. They are not connected 100% of the time, either. With a cordless phone, you get a dial tone and then dial. With a cell phone, you dial first. There isn't really a reason why the two need to be different.

(On many cordless phones, you can actually dial first, then hit talk, mimicking the behavior you'd use on your cell phone)


Hockey: An overtime or shootout loss is worth more than a regular loss. Before there were shootouts, an overtime loss was worth more than a tie, meaning three points were available to the two teams instead of two. (Typically the winner gets 2 points, the loser 0; in overtime, the winner gets 2 points and the loser 1.) An interesting Freakonomics-like investigation I propose: Before the adoption of the shootout, did teams collude to get to overtime in order for the eventual losing team to be better off in the standings?


I don't know why liquor is sold in fifths. And a "fifth" is usually 750 mL these days (a fifth of a gallon is about 756 mL, so this is the most natural equivalent). But it's possible to buy liquor in 375 mL containers (I've seen them at my not-so-friendly neighborhood liquor store) and these are not called tenths. Why not?

As for gas pumps, if you believe the hypothesis that putting the hole where the gas goes in on opposite sides of the car shortens lines, you don't need any sort of coordination. If you're the designer of a car, just flip a coin to decide which side the gas hole should go on; these should even out in the long run.


I believe the fifth of whiskey system had its origins as a ration system- a fifth was the amount to be consumed (and hence sold in those units) per day!?


Venezuela is also on a half hour time zone. For both Venezuela and Newfoundland having a half hour time zone makes "clock time" closer to "solar time". So when the sun is directly overhead the clock says 12:00 noon. Moreover, Venezuela and Newfoundland are both relatively isolated and in particular their borders see very little cross-border traffic and/or commerce. Consequently neither face significant pressure to aline their time zone with their neighbors. Standard time zones were established because it facilitated railroad scheduling. But neither Venezuela nor Newfoundland have rail connections with their neighbors.


Why is gasoline always sold with a .9 cents at the end of the price, but other bulk items (for instance, flour and sugar) are not?

I've heard that putting .99 at the end of prices at retail has always been more about keeping clerks honest (they have to make change from the register) rather than enticing purchases from those unable to figure out that $1.99 is essentially the same as $2.00. So why is gas priced with the .9? What is the history? When and where did it start?

One could make the point that gas is sold many gallons at a time, so that the difference between $3.39 and $3.39.9, while not very percepible to the customer, actually does amount to something over 20 gallons, unlike the difference between $1.99 and $2 for cigarette lighters, which are purchased one at a time. But then again, why not 29.9 cents per pound for flour or $2.99.9 per yard for cloth?

And why hasn't some gas station chain ever priced their gas at .8 instead? It seems it would grab attention at a very minimal cost.