Scott Sumner  

Beware of macro survey questions

PRINT
How People Get Good at Their J... A Collectivist Case for Trade,...

I've often expressed skepticism about survey questions regarding macroeconomics. Thus why ask the public whether they like inflation, if less than 5% of the public even knows what inflation is? (I ask my class if the cost of living has risen when both wages and prices rise by 10%, and 95% say no.)

I recently came across a graph in a Financial Times article that almost perfectly encapsulates my problem with macro surveys:

Screen Shot 2018-04-07 at 5.04.19 PM.png
I see three major problems with the FT interpretation of this survey of Chinese opinion:

1. The FT headline suggests that the appetite to consume has fallen sharply, but the downward blip in March looks like random noise. Overall, all three categories look pretty stable in recent years.

2. The three categories are also quite peculiar. In any principles of economics textbook, you'll see the following identities:

GDP = C + I = C + S = GDI

This implies that S = I.

Saving is the portion of income that is used to finance investment, and investment is the portion of output that is not consumed.

But if saving and investment are two sides of the same coin, then what does it mean to talk about saving and investment as two alternatives?

3. It is true that the public may conceive of saving and investment in a different way from how economists define the terms. But this leads to a third objection. What makes you think the public is going to provide any sort of coherent answer to the following question?

Considering current prices, interest rates and income levels, are you more inclined to consume, invest or save?
What do people mean when they say they are not "inclined to consume"? People who don't consume will die. Obviously people have something else in mind, perhaps "not consume at above normal levels." But the question is so vague that it's not at all clear what the answers might indicate. After all, it's always been fashionable to say that it's wise to save, wise to avoid consuming too much. In many cultures, people are taught that saving is virtuous, or at least prudent. So perhaps survey respondents are merely giving a politically correct answer.

Notice that the share of the public that believes it's a good time to consume has stayed around 12% or so, even as Chinese consumption levels have risen dramatically. My hunch is that if very few people say it's a good time to consume, it is because it sounds better to say it's a good time to save or invest.


Comments and Sharing






COMMENTS (28 to date)
Angelo writes:

As a non economist, could anyone explain why the answer is not no to the question to the class?

Antischiff writes:

Dr. Sumner,

Are you suggesting that both wages and prices rise by the same amount that there's been real cost of living increase, or merely a nominal one? I'd say there was a nominal increase, but not a real one.

gwern writes:

I assume his point is something like, you only consume a fraction of your wages with the rest being taxed or saved or spent on things not measured in inflation; so if your wages go up 10% and the cost of your basket of consumed goods goes up 10% too, you wind up ahead in real terms. (Imagine I owe you $1 and you owe me $2; I propose we double by 100% each amount; that's fair, right? Because each amount is increased the same percentage, right?) And your consumption patterns will change as your real income increases, too - a 10% increase in the cost of food hurts the rich less than the poor.

Kenneth Duda writes:

> could anyone explain why the answer is not
> no to the question to the class?

The "cost of living" is simply the amount of money required to meet basic living expenses. If the nominal price of these necessary goods and services goes up 10%, then the cost of living has risen 10%, period. It makes no difference what has happened to your wages.

See https://www.investopedia.com/terms/c/cost-of-living.asp for example.

Kenneth Duda

S D writes:

I used to work in a finance ministry in a small open economy.

We were hiring and I had to interview about 50 graduates.

All had some level of economics in their undergraduate degrees, some up to master's level


I chose a question to test their understanding of economics. Something like: "What would happen to inflation if there was a sudden and sharp depreciation of the exchange rate?".

After a while it became clear that many just didn't understand the question.

So I made it easier - simply asking "Give me a definition of inflation". And if they got that I would probe further.


At least a third of the interviewees (all recent economics graduates) could not give me something like: "a generalised increase in the price of goods and services".

So good luck asking a random sample of the population!

Scott Sumner writes:

Ken is correct.

George Selgin writes:

Antischiff has a point. There is such a thing as the real (as opposed to nominal) cost of living, which can be understood as the amount of real factor input required to produce the goods and services we consume. A wise-guy student in Scott's class might therefore reply to the question that more information was needed to answer it.

(I was that that sort of student. My teachers couldn't stand me.)

robc writes:

SD,

How would Mises do:

In theoretical investigation there is only one meaning that can rationally be attached to the expression Inflation: an increase in the quantity of money (in the broader sense of the term, so as to include fiduciary media as well), that is not offset by a corresponding increase in the need for money (again in the broader sense of the term), so that a fall in the objective exchange-value of money must occur.
Bob Murphy writes:

Scott,

I agree 100% with your critique of the survey / graph. But regarding your exasperation with your students, two things:

(1) Your question itself is influencing the answer. Coming from an economics professor, the question seems like it is testing whether the students know real vs. nominal. So it's like you're doing a double fake out.

(2) If Nancy Pelosi wanted to investigate the auto and computer industries because "cars and computers cost more today than they did in 1985," you would write a blog post praising her excellent command of economics?

David R Henderson writes:

Sorry, Ken Duda, but I'm with George Selgin and Robert Murphy here. I guess I was that "wise guy" George refers to, even though I didn't intend to be.

Mark Bahner writes:

As previously commented, prices going up 10 percent means the cost of living has gone up, no matter what happens to wages.

Another aspect is that even if (pre-tax) wages go up by 10 percent, that generally means people are paying more than 10 percent more in income taxes. So the after-tax income goes up by less than 10 percent, even if the (pre-tax) wages go up by 10 percent.

P.S. I second Bob Murphy's excellent comments. Regarding his first comment, my recollection of my student days is that I sometimes tried to guess what answer the teacher wanted, rather than what answer was correct. (Or I would answer what was correct, and then have to battle if the teacher marked it as wrong.)

David R Henderson writes:

An addendum:
I'm used to thinking of costs in real terms. So when I see that both wages and prices rose by 10%, I assume an increase in money supply preceded it. So the real cost did not rise. If I were asking my class that question, I wouldn't want just their bottom line, but also their reasoning that got them there.

Mark Bahner writes:
Another aspect is that even if (pre-tax) wages go up by 10 percent, that generally means people are paying more than 10 percent more in income taxes. So the after-tax income goes up by less than 10 percent, even if the (pre-tax) wages go up by 10 percent.

That's actually something I learned in one of my economics classes. (Take that, Bryan Caplan! ;-))

That was way back in the high-inflation late 1970s, so people were actually experiencing the tax bracket creep effect of rising wages and prices.

Floccina writes:

Would it be called inflation by economists if prices of goods and services rose say because resources became more scarce, therefore requiring more labor to produce the same output?

E. Harding writes:

There's a similar problem with issue polls; they tend to have a much higher percentage of people supporting the left-leaning option than the percentage of people who would ever actually vote for the left-leaning option when presented with a referendum at the voting booth.

Scott Sumner writes:

Everyone, The only definition of the "cost of living" that I've ever seen is the price level. If there's another definition in a textbook, can someone point me to it?

There's a serious point here. To an economist, supply and demand side inflation are both inflation. To the general public, inflation refers to prices rising while their income is stagnant. (Supply-side inflation).

Suppose I had asked the question using "price level" rather than cost of living. Would the answers have been different?

George Selgin writes:

On the real cost of living:

http://www.dallasfed.org/assets/documents/fed/annual/1999/ar97.pdf

Bob Murphy writes:

Scott,

Just to be clear, I don't have a problem with the way you're using "cost of living," I'm just saying I don't think your students are doing anything unreasonable. In fact, I would guess that the 5% who got your question "right" are worse at economics than the 95% who got it "wrong" (except if you spent time in class on this exact point).

I totally agree with your take, if we ask, "Is the cost of living higher in NYC or in Boise?" We clearly do NOT take into account the fact that nominal wages are higher in NYC for "the same job."

On the other hand, if groceries cost 1000 yen in Tokyo but 10 dollars in NYC, we wouldn't say, "The cost of living in Tokyo is 100x what it is in NYC." No, we would take the exchange rate into account.

So by the same token, if groceries cost $1000 in NYC-in-the-year-2018 but only cost $10 in NYC-in-the-year-1918 then I don't think we should naively say, "The cost of living went up 100x over the century." We should take the exchange rate into account, between 1918 dollars and 2018 dollars, right?

Alec Fahrin writes:

I personally would not understand how to answer that question intelligently, and I took multiple college-level classes in economics.

Thanks for pointing this out.

It represents the disconnect between most economic polling and what people actually do economically.

Scott Sumner writes:

George, Interesting, I had not seen that before. However I said "cost of living" not the "real cost of living". I'd add that he's talking about (the inverse of) the real wages, or the standard of living, so why confuse things by introducing the term "real cost of living"?

Consider this example: Suppose I told students that the BLS just reported that the cost of living rose by 1.9% in 2017. How would students process that information?

1.They might correctly understand that the price level rose by 1.9%. But in that case they would be wrong in their answer to my question.

2. They might assume that prices rose by 1.9% more than wages. That would be consistent with their answer to my question, but of course it would be wrong about what the BLS is actually reporting. They are reporting the rise in the price level when they say the cost of living rose by 1.9%, with no reference to nominal or real wages.

So I still believe that students are confused on this topic.

Bob, The reason I keep making this point is that the public's confusion causes real harm. Thus the public was opposed to Bernanke's attempt to raise inflation back in 2010 precisely because they thought he was trying to reduce their living standards. Actually he was trying to raise their living standards.

I understand the sense in which it is a "tricky" question, but that's partly because people actually do not understand inflation.

I'm not sure what you mean by the exchange rate between 1918 and 2018 dollars. What is the exchange rate between these years?

David R Henderson writes:

@Scott Sumner,
What got me to the conclusion that the students were right is thinking through the extreme cases. If, in one year, the price level in Germany in the early 1920s rose by 10 million percent, I don't think it would make sense to say that the cost of living rose by 10 million percent. But, given your reasoning, you would need to say that it did.

George Selgin writes:

Roger Garrison--certainly a good economist--who lives in Auburn, AL, used to like to say that the cost of living there was just as high as that of living in a city like New York, the only difference being in how those costs were paid--e.g., in rent or in tedium.

Evidently Roger was thinking in real rather than nominal terms! (He also much preferred Auburn's tedium to New York's excitement.)

Mark Z writes:

One might measure 'real' cost of living in terms of the number of hours you have to work to purchase a given amount of goods and services. What perhaps confuses things is that some of the difference between New York and Auburn (sticking with George Selgin's example) is 'nominal' (getting paid more to pay higher rent) while some of the difference is real: the higher salary in NYC may not be enough to fully cancel out the higher rent. So, Price_NYC/Price_Auburn > 1, but (Price_NYC/Wages_NYC)/(Prices_Auburn/Wages_Auburn) > 1 as well. These two inequalities say two different things. "I'm taller because I grew two inches" is not the same as "I'm taller because the definition of what an inch is has decreased." Only the former sense gets me any closer to reaching the top shelf at the store, so I expect the former matters more to people instinctively, whatever the formal definition is.

George Selgin writes:

Roger Garrison--certainly a good economist--who lives in Auburn, AL, used to like to say that the cost of living there was just as high as that of living in a city like New York, the only difference being in how those costs were paid--e.g., in rent or in tedium.

Evidently Roger was thinking in real rather than nominal terms! (He also much preferred Auburn's tedium to New York's excitement.)

Scott Sumner writes:

David, You said:

"But, given your reasoning"

Wouldn't it make more sense to say "given the standard definition of the 'cost of living'"? I'm just going with the definition that I assumed was accepted by everyone.

David R Henderson writes:

@Scott Sumner,
Wouldn't it make more sense to say "given the standard definition of the 'cost of living'"? I'm just going with the definition that I assumed was accepted by everyone.
Good point. I think this whole discussion will cause me to always distinguish between the nominal cost of living and the real cost of living.

Roy Kerns writes:

When I think about understanding inflation and COLA, I always find Mark Twain's Yankee fun to point out. It seems to me that story does not so much focus on explaining inflation as it does on how likely folks will not understand it. In fact, they will deny that understanding and continue to endorse the government policies that most facilitate inflation.


http://etc.usf.edu/lit2go/174/a-connecticut-yankee-in-king-arthurs-court/3127/chapter-33-sixth-century-political-economy/

The comments on David's originating question show the value (heh) of recognizing equivocation and need for disambiguation.

Roy Kerns writes:

oops, sorry, that should be Scott's originating question

Comments for this entry have been closed
Return to top