Scott Sumner  

The Emperor repeatedly forgets his clothes

PRINT
Hammock on the Chicken Tax and... How Reforms Have Made Donald T...

Today I'm going to criticize economists for not knowing what they mean by the term 'inflation'. But before doing so, I'd like to point to an analogy; economists have no idea what the terms 'easy money' and 'tight money' mean, even though they use these terms all the time, and they play an important role in discussing key policy issues.

Some economists think easy money is low nominal interest rates, some think it is low real interest rates, some think it is an interest rate that is below the Wicksellian equilibrium rate. Others point to rapid growth in the monetary base. Still others look at the broader monetary aggregates. Or total credit. Mundellians look at the value of the dollar. Supply-siders look at gold prices.

And we haven't even yet gotten to respectable economists! Milton Friedman say low rates are a sign that money has been tight. Mishkin said you want to look at "other asset markets". Bernanke has the best definition; easy money is rapid growth in NGDP or the price level.

Even worse, like the emperor with no clothes, economists don't even understand that the term 'easy money' has no well-accepted definition, or that most of them are using silly definitions. They talk about easy and tight money, with the serene confidence that they are not talking gibberish.

The same is true of 'inflation'. What does that term actually mean, in a world where goods are always changing? The quality of existing goods changes, and new goods are constantly being invented. The prices of many old goods, such as dictionaries, have essentially fallen to zero.

Here's Brad DeLong, discussing a Brookings paper by David M. Byrne, John G. Fernald and Marshall B. Reinsdorf

I also confess to being annoyed by:
Second, many of the tremendous consumer benefits from smartphones, Google searches, and Facebook are, conceptually, non-market: Consumers are more productive in using their nonmarket time to produce services they value. These benefits do not mean that market-sector production functions are shifting out more rapidly than measured, even if consumer welfare is rising...
Isn't "measuring consumer welfare" the point? We (a) arrange atoms (b) in forms we find pleasing and convenient, and then use them in combination with (c) information and (d) communication to accomplish our purposes. That our measures of economic growth are overwhelmingly "market" measures that capture the value of (a), much of the value of (b), and little of the value of (c) and (d) is an indictment of those measures, and not an excuse for laziness by shrugging them off as "non-market" and claiming that measuring the shifting-out of market-sector production functions is our proper business.
I'm with DeLong. I had always assumed that inflation was the increase in income (actually consumption) required to maintain a constant level of utility. When I ask millennials if they'd rather live today on $100,000/year or back in 1965 on $100,000/year, many say they'd prefer to live today, because back then there was no internet, smart phones, HDTV, and virtually no good Thai/Vietnamese/Sushi restaurants in America. I can recall when affluent people went out to eat at "Super Clubs" and dined on steak and potatoes. It almost makes me sick just to think about it. De gustibus . . .

In the past I've said that inflation is a meaningless concept. That's a bit too strong; obviously fuzzy concepts can be useful. It's not meaningless to say Zimbabwe had a lot of inflation in 2008. Rather I'd say inflation is an almost useless concept. NGDP growth (perhaps per capita) is generally a better metric of nominal shocks. So we don't need inflation.

So DeLong's debate with the Brookings economists can never be resolved, because economists don't even know what inflation is.

Even worse, as with easy and tight money, economists don't even know that they don't know what inflation is. They talk as if it's some sort of objective fact, like the height of Mt. Everest, which we ascertain with ever more accurate measurements. It's more like the size of Trump's ego---how can you agree how to measure something, when you can't even agree on what it is?


Comments and Sharing


CATEGORIES: Monetary Policy , Money




COMMENTS (17 to date)
Lawrence D'anna writes:

Shouldn't "what is easy money" always be asked in relative sense, not the absolute sense? like "is policy X more easy than policy Y?", instead of "is policy X easy?".

And if you always ask about "easiness" only in relation to an alternate policy, doesn't it all get a lot less ambiguous? Easier means nominal prices are higher. You've still got some ambiguity because some prices could be higher while others are lower. But in that case maybe it's not reasonable to expect to be able to say which policy is "easier". Like asking if a virus is "alive".

Scott Sumner writes:

Lawrence, I sympathize with that. I've come to believe that it should be easy or tight relative to the central bank's target.

Mark writes:

If several observers have access to disparate measurements of the precise position of some car (or can't agree on which measure is the 'right' one), that doesn't necessarily mean they can't agree on some useful information inferred from the disparate measurements: say, the derivative of the car's position, or how hast the car is going and in which direction.

Perhaps one could optimistically hope that different economists' definitions/measures of inflation may at least agree on when inflation is increasing or decreasing, and that that's what's most important? Of course with the car analogy, position itself may get important if, even if all the observers agree the car is headed in the direction of a cliff, one may say it must turn immediately to avoid the cliff while another may say there's plenty of time before it needs to turn.

Ilya writes:

It's simple, actually.

'Easy money' is when your uncle dies and leaves you a couple of million.

'Tight money' is when you have to choose between buying new shoes and getting the beloved family dog to the vet.

Why economists started using those terms in the macroeconomics is beyond my comprehension.

Niklas Blanchard writes:

My line is that inflation is always and everywhere a monetary phenomenon...and when it is not, it's not a phenomenon at all.

This gets at the idea that under a stable monetary regime, nominal debt (and wage) contracts easily price inflation such that it doesn't even reach the level of consciousness. I believe you have a blog titled after that idea.

J.V. Dubois writes:

Niklas Blanchard: "My line is that inflation is always and everywhere a monetary phenomenon...and when it is not, it's not a phenomenon at all"

This is absolutely not a useful way of thinking about it. For instance in Warsaw ghetto during WWII prices of food increased in thousands of percent as a result of Nazi deliberate policy of starving out its inhabitants. Needless to say it is hard to see how this episode was a monetary phenomenon in any standard sense and one can definitely argue that it was "not a phenomenon at all".

I actually agree with Scott. Inflation is almost useless concept now. It can mean "increased cost of living" - sometimes even for specific people like adjusting pensions to account for increased prices for medical services. For Austrians inflation equals to "printing money", period. I saw multitudes of articles - sometimes even from professional economists - talking about "deflation in IT" or "inflation caused by increases in the price of oil". This means that they think about inflation as partial equilibrium phenomenon. We see economists warning about "inflation" just solely based on the fact that unemployment is low.

It is a mess.

ThomasH writes:

I agree and so I think it is counterproductive to argue over whether it is mistake to say that money was "easy" in 2008. It seems more productive to say that it was not "easy enough," that the Fed should have reduced ST interest rates earlier in the year, to zero in September, and started a much larger and non-time and amount delimited program of buying long term assets (or whatever policy mix you favor).

As for inflation, I still think the main problem with inflation targeting is not the definition of "inflation" but the practical conflation of inflation targeting with having an inflation rate ceiling.

Effem writes:

My two largest expenses (by a wide margin) are rent and education. I would take both from several decades ago as quality was basically the same and prices are now several multiples higher.

And please don't say we can all learn online now. The credential is what's needed to be accepted as a productive member of society.

Rantly McTirade writes:

[Comment removed pending confirmation of email address. Email the webmaster@econlib.org to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Niklas Blanchard writes:

J.V. Dubois:

Well, if the line gives the impression that I think the concept of inflation is at all useful, then it probably isn't the best...but it's still a pithy line, so I'm going to keep it around.

That I wouldn't classify a spike in prices due to a supply shock as "inflation" in the same way as, say, printing money out of proportionality with the demand to hold money while someone else might (especially people who like to distinguish between "demand-pull" and "supply-push" inflation) is really kind of making the point.

Scott Sumner writes:

Niklas, You said:

"My line is that inflation is always and everywhere a monetary phenomenon...and when it is not, it's not a phenomenon at all."

Agreed, but that doesn't really tell us what it is, just that inflation is a reduction in the purchasing power of money. But what do we mean by the purchasing power of money?

EB writes:

If we believe that in our economy the number and quality of goods are always changing, to explain how our economy has been evolving over time we need to explain why and how those changes have been taking place. None of our macro theories can explain that view of our economy, and more important none of our measures of "aggregate" variables are good enough to provide useful information about our economy because we don´t have yet good methods to aggregate quantities and prices of these varieties over time, or to aggregate "same" services to different individuals, in particular non-market services. One measure that is useless is NGDP because it is an aggregate of varieties of apples and varieties of medical market and non-market services and none can say that he or she understands what a change in NGDP means (or what a change in the GDP implicit price deflector means). Yes, it´s time to stop the nonsense of arguing about how "the state of the macroeconomy" has been changing based on theories of a non-changing economy and on terrible measures of "aggregate" variables.

Also, the distinction between real and nominal shocks adds nothing to our understanding of their consequences. All shocks are the result of either natural forces (an earthquake) or unexpected, extraordinary human actions. If suppliers of payments services are affected by an earthquake or an unexpected terrorist attack, the economy will suffer but comparisons with the effects of similar shocks to suppliers of electricity are “empirical” questions. If suppliers of payments services make the huge, unexpected mistake of printing a fake currency and suppliers of electricity make the huge, unexpected mistake of cutting service by pressing the wrong button, we know the economy will suffer but comparing their effects is an “empirical” question.

Ricardo writes:

Your definition of inflation as "the increase in income (actually consumption) required to maintain a constant level of utility" is a good one.

The obvious next question is: since we don't have utilometers, how can we be sure that utility is constant? We can't, of course, but perhaps the next best thing is to measure actual consumer expenditures, on the theory that consumers will attempt to maximize utility.

In fact, this is what the (US) CPI does. The "market basket" is constructed using consumption data from the BLS Consumer Expenditures program, so price indexes are just expenditure-weighted aggregations of price changes. The rate at which the market basket evolves is every two years for the CPI and every month for the chained CPI.

You might consider whether it is reasonable to define a concept by its measurement. One definition of "second" is the number of oscillations in a particular crystal. Another definition of "second" is the time it takes the earth to rotate 1/86400th of its circumference. The first definition is akin to the CPI, whereas the second definition is akin to what you describe in your post. Different circumstances may call for different definitions.

JDI writes:

Some years ago I worked in a national treasury doing analysis and forecasting of inflation. It was in a small, open economy.

I spent time interrogating the statisticians a lot about how they did it. They claimed to adhere to international standards and from what I could tell they did. Still, it left room for massive discretion.

Two examples to illustrate the point:

1) New and used cars were both about 3% of the inflation basket. For simplicity the statisticians assumed the price of used cars moved perfectly in line with new ones. The problem was that one day the economy was hit by a massive AD shock. The stock of second-hand cars was fixed (in short run) and anecdotally prices fell by 20%-30% in 18 months. This was completely missed by the methodology, which meant an overestimation of the price level by a non-trivial 1/2 a percentage point.

2) Grocery-type goods were about 15% of the basket and the bulk of resources were spent on surveying them through staff going once a month to retail outlets to take down prices. Despite a relatively concentrated retail market and a small country (5 main grocery players) the statisticians measured the price of milk in about 80 outlets nationwide!


Officially, inflation estimates are never revised (as this can mess with index-linked payments), so discretionary and methodological errors compound over time.

I know this is somewhat trivial compared to the larger conceptual point you are making Scott. It's just to show that even if you take measures of inflation very seriously (like central bankers do) you are inevitably working with a problematic estimate.

Scott Sumner writes:

Thanks JDI, Housing has similar problems.

Greg Jaxon writes:

Here are two other (fringish) {in,de}flation definitions to consider:

  • Inflation is an imbalance of price speculation into commodities (vs bonds)
    Deflation is the opposite imbalance.
In this view, bond and commodity speculation (long bets) offer relatively different risk/reward ratios and speculative money flows accordingly, affecting interest rates and consumer prices in a way that keeps them coupled (albeit with one or the other leading/lagging).
  • Inflation is the monetizing of counterfeit credit.
    DisInflation is the corrective defaulting of said credits.
    Deflation is the demonetization of good credits (especially of circulation credits that arise through ordinary commerce).

Jose Romeu Robazzi writes:

Great Post, I totally agree.
One caveat, though: at high levels of repeat purchase price percentage increases(i.e. high inflation), the rise in prices become a real problem. It is hard not to be annoyed by yearly rent adjustments of 5% when you salary remains flat, or to see you favorite restaurant adjust prices in 10% and your salary remains flat ... No wonder people hate "inflation", because they associate the perceived phenomenon with being "left behind in the game", or something like that...

Comments for this entry have been closed
Return to top