David R. Henderson  

Greg Mankiw Makes a Case (Kind Of) For a Competitor's Textbook

PRINT
The Fed shouldn't reason from ... McArdle on Foreign Interventio...

HINT: The textbook is The Economic Way of Thinking by Heyne, Boettke, and Prychitko.

UPDATE BELOW

In an excellent post today, Greg Mankiw writes about an unnamed competitor's textbook:

I happened to be flipping through another introductory economics textbook. (Yes, some people have the temerity to try to compete with my favorite textbook.) I noticed an error that is, unfortunately, all too common in how introductory economics is taught. I won't mention which book it is, because I am quite fond of the authors, and because my goal here is not to pick on one particular book but rather to draw attention to a more pervasive problem.

The issue is how one applies welfare economics to understand price controls, such as rent control and minimum-wage laws.

The sin that this book makes is to look at consumer surplus, producer surplus, and deadweight loss as if we were studying the welfare cost of a tax. The cost of a price control, the reader is taught, is the small Harberger triangle between the supply and demand curves.

This reasoning is problematic because it assumes perfect rationing. But rationing under price controls is never perfect. Under rent control, for example, apartments do not automatically go to those who value the apartments the most. The misallocation due to imperfect rationing makes the actual welfare cost of price controls much higher than the standard deadweight loss triangle.


Mankiw's criticism is on point. But it also understates the problem. The welfare cost (efficiency loss) due to price controls includes not only the Harberger triangle, and not only the deadweight loss from misallocation among buyers that Mankiw points out. It also includes the loss from wasting time in lines.

During the 1979 shortages due to gasoline price controls, I computed that people were spending an average of about 40 cents per gallon in line. At the time, the price ceiling was about 80 cents per gallon. Economists at the Department of Energy had computed that the market-clearing price would have been about $1.00 per gallon. In other words, people were spending for gasoline, in time and money, more than they would have had their been no price controls.

Greg doesn't mention this last factor, even though it's potentially huge.

I looked at Greg's textbook and it doesn't mention this factor either, although he does a nice job of pointing out the misallocation across buyers.

But there is one textbook that has a nice diagram making my point. The authors don't mention time per se; their term is "non-money costs." But they show that non-money costs can combine with money costs to make the price-controlled good even more expensive than if there were no price controls. Unfortunately, in their diagram, they do assume that the goods end up in the hands of those who value them most. So it doesn't solve the problem Greg raises. But it does show a huge cost of price controls that the unnamed textbook he criticizes does not.

That book is The Economy Way of Thinking by Heyne, Boettke, and Prychitko. In the latest edition I have (the 12th), the graph is on page 124.

UPDATE: As Alex Tabarrok points out in a comment below, his and Tyler Cowen's text does a nice job of laying out graphically the efficiency loss from misallocation. Also Greg Mankiw points out in an e-mail, "As the commentator 'Justin' on your blog points out, I am bit more thorough than you give me credit for."


Comments and Sharing






COMMENTS (15 to date)
ThomasH writes:

I agree that the analysis of price controls needs to include the costs of imperfect rationing.

I'd argue however, that an even more useful analysis would need to consider the (smaller) costs of alternative policies to achieve the same objective. Few price controls are totally accidental (as gasoline price controls were close to being in 1974) the controls being a remnant of the Nixon/Connolly attempt to control inflation with out using monetary or fiscal policy. More often, they are being used as a way to redistribute income to a particular group. I'm sure that any first year economics student could think up lower cost ways to transfer income to doctors or taxi drivers, or low income workers or poor farmers than restrictions on prices or quantities.

Justin writes:

In my copy of Mankiw's micro text, he actualy does seem to makes these points. Here are some quotes:

"When a shortage of ice cream develops because of this price ceiling, some mechanism for rationing ice cream will naturally develop. The mechanism could be long lines: Buyers who are willing to arrive early and wait in line get a cone, while those unwilling to wait do not."

and

"This example in the market for ice cream shows a general result: When the government imposes a binding price ceiling on a competitive market, a shortage of the good arises, and sellers must ration the scarce goods among the large number of potential buyers. The rationing mechanisms that develop under price ceilings are rarely desirable. Long lines are inefficient, because they waste buyers’ time."

David R. Henderson writes:

@Justin,
Thanks.
I should add that when I teach the Heyne, Boettke, Prychitko graph on p. 124, I lay out how it well could be an understatement of the efficiency loss and show how the area of deadweight loss could be even be bigger than they show.

Peter Boettke writes:

Thanks for the plug David. Of course, this is a point that Tullock really stressed in his analysis of the welfare costs of government controls, and which Walter Williams stressed in his emphasis on multiple margins of adjustment. In a system of political capitalism like our own the applications of the analysis are unfortunately abundant rather than exceptions to the norm.

Justin writes:

This mistake of miscalculating deadweight loss seems pervasive. It seems every webpage I check and every Youtube lesson on the subject gets it wrong. For example https://www.youtube.com/watch?v=RxuXVuXhaTU and Khan Academy https://www.youtube.com/watch?t=639&v=8U8egQjcPaI

Taips writes:

Lines are a hugely wasteful way of allocating rationed goods, no one denies that.

But here's the thing: it is not the price control per se that causes waste of queuers' time.
If you take Ec101 graphs at face value (or you are a British or French civil servant during WWII), a solution hits you: rationing tickets ! On paper, if allocation of the rationed good is made by tickets and not queues, the DWL goes back to misallocation + foregone trading triangle.
Rationing tickets occasion administrative costs, but people can to some extent trade their allotments, alleviating the misallocation issue (not the foregone trades triangle, unless they can somehow compensate sellers too, but then the price control is not enforced).

Now in practice it is a huge mess, bartering allotments is harder than using currency, etc., but these issues don't fit neatly into the Ec101 graph.

Alex Tabarrok writes:

There is in fact only one principles of economics book which makes all of these points and shows the student how to calculate the loss from misallocation. Modern Principles! :) You can find the analysis here

https://books.google.com/books?id=G2QdBQAAQBAJ&lpg=PA139&ots=uU95Vw0BbB&dq=random%20misallocation%20modern%20principles&pg=PA138#v=onepage&q=random%20misallocation%20modern%20principles&f=false

Tom West writes:

I computed that people were spending an average of about 40 cents per gallon in line. ... In other words, people were spending for gasoline, in time and money, more than they would have had their been no price controls.

This brings to the forefront what people like or dislike about price controls - redistribution. For the wealthy, whose time is worth a lot, the "price" of the wait is huge, while for the poor, not so much.

So, *some* people were spending more for gasoline than without price controls.

Alex Tabarrok writes:
GregS writes:

Steven Landsburg's Price Theory text gives a good treatment of this problem. Unless I'm reading it wrong (maybe all the other texts present this problem differently from Landsburg's presentation), he states that the deadweight loss from waiting in line is included in this area between the supply and demand curves. I'm looking at his fourth edition right now, so unless a subsequent edition changed the presentation or I'm misreading your post, I'm saying he called this one right.

Another point. The people who value the rationed good most highly will spend more time waiting in line, so be careful you don't double-count by listing two offsetting items. If Greg Mankiw says "An uncounted cost is that the price controlled good is distributed randomly and not to the highest valued user," and then David Henderson says "An uncounted cost is the time spend waiting in line for the price-controlled good," it sounds like there are two uncounted costs when really there are two effects that (perhaps slightly) offset each other. I could be mistaken but it looks like Landsburg's presentation avoids this double-counting.

david condon writes:

I don't like that Mankiw described this as a mistake. Leaving something out isn't necessarily a mistake. They might have just wanted to simplify the discussion and not present too many concepts at once.

David R. Henderson writes:

@david condon,
I don't like that Mankiw described this as a mistake. Leaving something out isn't necessarily a mistake. They might have just wanted to simplify the discussion and not present too many concepts at once.
If Greg has characterized this other text correctly--and I have no reason to think he hasn’t--then it is, indeed, a mistake. You’re right that leaving something out isn’t necessarily a mistake. But that’s not what these textbook writers did. They put something in that’s wrong. Their graph makes sense only if the goods end up in the hands of those who value them most and only if the queuing costs are zero. Those assumptions are wrong.

yarbel writes:

Wait, isn't there some tension between these two effects? A buyer whose reservation price is lower than actual price (=control + wait) won't wait in line. So the waiting solves some of the problem of rationing.

David R. Henderson writes:

@GregS,
Another point. The people who value the rationed good most highly will spend more time waiting in line, so be careful you don't double-count by listing two offsetting items. If Greg Mankiw says "An uncounted cost is that the price controlled good is distributed randomly and not to the highest valued user," and then David Henderson says "An uncounted cost is the time spend waiting in line for the price-controlled good," it sounds like there are two uncounted costs when really there are two effects that (perhaps slightly) offset each other. I could be mistaken but it looks like Landsburg's presentation avoids this double-counting.
Good point.
Wait, isn't there some tension between these two effects? A buyer whose reservation price is lower than actual price (=control + wait) won't wait in line. So the waiting solves some of the problem of rationing.
Yes. Good point, which, I believe is the same point GregS is making.

David Friedman writes:

The rent seeking cost of price control was included in my 1986 Price Theory.

Comments for this entry have been closed
Return to top