Scott Sumner  

Substitute goods and reasoning from a price change

PRINT
Mankiw's Textbook on Determina... Are Centers a Mistake?...

David Henderson has a good post on the way that textbooks teach the substitution effect. I have one other bone to pick with principles textbooks---they don't clearly explain to students how to avoid "reasoning from a price change." Start with the textbook definition of substitute goods:

If the price of good A rises, the demand for good B rises.
Now apply it to a real world example. Suppose there is a scientific study indicating that drinking coffee causes cancer. How does this impact the demand for tea? Most students will understand that tea is a substitute for coffee, and hence that the demand for tea will rise. But they won't get the right answer by applying anything they learned in economics. After all, the coffee scare would depress the price of coffee. So the demand for tea is rising despite the fall in the price of coffee.

On could argue that other things equal, a lower coffee price should lead to less demand for tea. But even that's not quite right, as "other things equal" a lower coffee price would lead to a shortage of coffee. That's because if you assume "other things equal" you are assuming that only the price of coffee has changed, nothing else. The supply and demand curves for coffee stay where they are. And in that case there is more demand for tea due to the coffee shortage.

Of course students who take advanced economics understand that a health scare in coffee causes a big increase in the demand for tea, whereas the accompanying fall in coffee prices makes the increase in tea demand a bit smaller, but that's way over the heads of principles students. I see so much reasoning from a price change that I wonder whether even 1% of economics students actually understand supply and demand. Many students believe they are taught that "in years when coffee prices soar much higher, we would normally expect to see a rise in the demand for tea," which is not at all what we are trying to teach.

In Mankiw's Principles text he says:

Suppose that the price of frozen yogurt falls. The law of demand says that you will buy more frozen yogurt.

(That's the third edition, p. 68, perhaps it's been fixed in later editions.) Of course the law of demand says no such thing. It depends why the price of yogurt fell. Defenders of Mankiw tell me that there's no problem, because he is implicitly assuming the demand curve for frozen yogurt is stable. And how many students will understand that subtlety? The vast majority will leave economics thinking that they've been taught that you can expect people to buy more when the price is lower. And then we wonder why the public is confused when economists complain about deflation.

Recently I've had lots of debates about the effect of a change in interest rates. In fact, it's nonsensical to talk about a change in interest rates having any sort of predictable effect on the economy. Rather it's the things that cause interest rates to change that will have an effect. Back in 1992, Milton Friedman complained about how the profession reasons from a price change, assuming that low interest rates always mean easy money:

Declining or low interest rates may at times correspond to easy money, but so may rising or high interest rates. To illustrate the first possibility, the short-term commercial paper rate remained around 0.75% from 1939 to 1946, while the money supply nearly tripled and the price level rose by 60%. To illustrate the second, the federal funds rate hit 20% in January 1981, and again in July, a period when both M2 and consumer prices were rising at nearly 10% a year. If the Fed can control interest rates, does anyone really believe that we would have seen the federal funds rate at 20%?

Unfortunately, I haven't seen any signs of improvement since 1992, if anything the profession is going backwards. I see people say "normally you'd expect more investment when rates are low" or "normally the government should invest more in infrastructure when rates are low" or "normally you'd expect more bank lending when rates are low," all reasoning from a price change.

Update: Speaking of Reasoning From a Price Change, there is a new podcast where Russ Roberts interviews me on that very subject.

PS. Years ago when I was teaching principles I used Mankiw's text because I thought it was the best one available, so don't take this post as trashing his book. It's actually very difficult to teach anything in economics without cutting some corners here or there.

PPS. The coffee example is not some sort of weird trick question. A priori, you'd expect roughly 50% of price changes to be due to demand shifts and 50% due to supply shifts. We'd like our principles models to be more accurate than a coin flip

HT: Benjamin Cole


Comments and Sharing


CATEGORIES: Economic Education




COMMENTS (19 to date)
Jon Murphy writes:

Let me just say thank you. I made the mistake of reasoning from a price change for years in my undergrad and it wasn't until I read your posts on it I realized what I was doing!

Thanks for pointing it out to me and for making me a better economist.

Scott Sumner writes:

Thanks Jon!

perfectlyGoodInk writes:
(That's the third edition, p. 68, perhaps it's been fixed in later editions.)

It's still there in the 6th edition, p. 70. Subsequent editions have had only superficial changes, so I doubt it's been changed.

Jameson writes:
Of course students who take advanced economics understand that a health scare in coffee causes a big increase in the demand for tea, whereas the accompanying fall in coffee prices makes the increase in tea demand a bit smaller, but that's way over the heads of principles students.

How is that way over anyone's head? Isn't that just common sense? Some days I wish I had gone into economics instead of just watching it from the sidelines as a mathematician. It just seems like this shouldn't be that hard.

vikingvista writes:

In response to this criticism, I suppose most pundits and economists would deny making a fundamental error, but rather reply "Of course, everyone knows I was assuming X." Maybe assuming government price and output controls don't exist is reasonable when they don't and usually don't exist. But all sorts of changes to supply and demand schedules are everyday market phenomena, so how can one implicitly assume those?

But maybe one can. After all, in the absence of government controls, and assuming simple rather than compound changes, if someone says that a falling price means increased consumption, what else are they saying but that the supply curve is shifting right?

If instead of telling such a person that she is reasoning from a price change, you tell her that her unambiguous assumption about supply and demand schedules is wrong, she will be immediately ready to defend an assumption that she thought was obvious to everyone in her audience.

Maybe I'm too generous, but if saying one thing (relating price and quantity changes) is effectively equivalent to saying the other thing (how supply xor demand curves are shifting), then does it really matter which one they say?

marcus nunes writes:

Scott
Did you see this example of RFPC (with great conviction):
"Seizing advantage of rock-bottom interest rates is the best way to raise them".
http://www.ft.com/intl/cms/s/0/27df678c-e511-11e4-bb4b-00144feab7de.html#axzz3XiPCue4o

jon writes:

Slightly OT, but what would the people on here recommend as good introductory texts on micro and macroecon (for someone with no background and intending to self study)?

Thanks for any suggestions.

Nick writes:

I don't see the problem ...
Sure the $ price of coffee has fallen, but people now think it may also cost their lives. Overall, the perceived purchase price of a cup of coffee has increased, and so demand for tea increases as well.
Textbook seems to work fine. Are you sure you didn't read it wrong?
;)

Steve Y writes:

I think I understand the point, but enough with the coffee, tea, and yogurt illustrations. I am trying to figure out whether the following claim is a) reasoning from a price change; b) not RFAPC; c) there's not enough info to decide (a) or (b).

"The ACA is contributing to the recent slow growth in health care prices and spending and is improving quality of care."

https://www.whitehouse.gov/sites/default/files/docs/healthcostreport_final_noembargo_v2.pdf

Thanks, btw, I always learn something from your posts.

Don Geddis writes:

@vikingvista: "if someone says that a falling price means increased consumption, what else are they saying but that the supply curve is shifting right?"

But in the real world, demand shifts are just as common as supply shifts. So you need an unstated assumption about one or the other. If you have an assumption, you should spell it out.

The problem is that falling price doesn't "mean" increased consumption. Falling price + increased consumption may imply a shifting supply curve. Or, falling price + a shifting supply curve may imply increased consumption.

But you need two assumptions, not one. You don't get anything by only assuming falling prices. There should be no conclusions, from that by itself.

Don Geddis writes:

@Nick: "the perceived purchase price"

This is not a technical term in economics. Yes, it's clear that people's utility has changed. Which suggests that the demand curve has changed. But it's not helpful to describe a shifting demand curve as a change in some "perceived purchase price". There's already a nominal purchase price. And the question being asked is, what happens to quantities sold, if the nominal price decreases?

And the answer is: it depends on whether the nominal price changed due to a supply shift, or a demand shift. Quantities might either rise or fall. You can't tell, knowing only the the nominal price changed.

"Perceived purchase price" is not something that can be measured.

Scott Sumner writes:

Jameson, Go to any university, and meet with 100 students who have already studied supply and demand. Ask them to explain why on average 300 people attend movies when the price is $10 and 100 people attend when the price is $6. You'll be stunned how few come up with the obvious answer.

I've just been arguing with a bunch of commenters who insist that lower interest rates have some sort of implication for the economy.

Lots of things in econ are easy when explained, but cause all sorts of confusion when you are simply looking at messy real world data, without any theoretical context.

VikingVista, The problem is that people are reasoning from the price change itself. A price change doesn't tell you whether it was supply that shifted or demand that shifted. But they are acting as if it does. It would be fine if they said "I'm going to assume the price change was due to a supply shift." But they are claiming that the price change shows it was a supply (or demand) shift. But it doesn't.

Thanks Marcus, I'll take a look.

Jon, Mankiw or Cowen/Tabarrok

Nick, They are referring to the dollar price.

And my criticism isn't so much with the textbooks, as the way this stuff gets misunderstood by students, and misused in the real world. The textbook writers know what I'm complaining about, and they'd say (rightly) that it's hard to get the wording exactly right and still have a readable book.

vikingvista writes:

If someone says something like, "You give me a price change, and I will be able to tell you whether the quantity increased or decreased", then I agree.

But if a person gives you both a known price *and* quantity change, even in the form of a causal statement (e.g., interest rates decreased because money supply increased, or visa versa), then how is that different from telling you how the S&D curves are shifting?

It seems like the fairer criticism is to tell that person he is wrong about the curve shifts, not that he doesn't understand fundamental principles. At least until you probe further and discover he really doesn't grasp the principles.

vikingvista writes:

Don,

"Falling price + increased consumption may imply a shifting supply curve. Or, falling price + a shifting supply curve may imply increased consumption."

Your only hangup is the causal claim? Why is that important? What matters, is that when a person says that the price falls and that consumption increases, he can only (ignoring more complex phenomena) mean that the supply curve shifted right. So there is no confusion or ambiguity. Any two automatically requires the 3rd by way of fundamental principles. Just because 2 of the 3 come in the form of a causal statement about observed facts doesn't mean that the speaker only assumed 1 of the 3.

Is it not fairer, and more likely the case, to assume a person is wrong about the facts of the supply, price, or quantity changes and NOT wrong about the principles?

Scott Sumner writes:

Vikingvista, You said:

"But if a person gives you both a known price *and* quantity change, even in the form of a causal statement (e.g., interest rates decreased because money supply increased, or visa versa), then how is that different from telling you how the S&D curves are shifting?"

That's fine, and indeed that's what I always try to do when discussing these issues.

Tommy Dorsett writes:

That Friedman quote is a GEM!

Filipe writes:

Really interesting read, thanks a lot for this.

Just a quick point of logic:

Let me take the sentence that says "If the price of good A rises, the demand for good B rises", and transform it into: if p then q.

Your real world example, "q" happened (more demand for tea). And you object noting that "demand for tea is rising despite the fall in the price of coffee" i.e. "q" despite "not-p". Yet there is absolutely no inconsistency here since the law is if p then q, and not, if q then p.

Having said that, I really enjoyed the reading. I just think this is an example you may wish to revise.

vikingvista writes:

Filipe,

Instead of 'not p or q', I think first, they are taking 'if p then q' to be 'p entails q'. Second, I think they are taking q as an unrealized prediction rather than an observed fact.

Sean writes:
Jameson, Go to any university, and meet with 100 students who have already studied supply and demand. Ask them to explain why on average 300 people attend movies when the price is $10 and 100 people attend when the price is $6. You'll be stunned how few come up with the obvious answer.

Because there was a downward shift in the demand curve?

Comments for this entry have been closed
Return to top