When Bob Murphy sees this title he’ll probably think it is a dream come true. The evil inflationist finally sees reason. Well I’m afraid it’s more like his worst nightmare. I don’t propose to abolish the phenomenon of inflation, but rather the concept of inflation. And to be more precise, price inflation, which is what almost everyone means by the term. I want it stripped from our macroeconomic theories, removed from our textbooks, banished into the dustbin of discarded mental constructs.
Even worse, I propose doing so for “postmodern” reasons. I will start by denying the reality of inflation, and then argue for some substitute concepts that are far more useful. First a bit of philosophy. There is a lively debate about whether there is a meaningful distinction between our perception of reality, and actual reality. I had a long debate with a philosopher about whether Newton’s laws of motion were a part of reality, or merely a human construct. I took the latter view, arguing that if humans had never existed then Newton’s laws would have never existed. He argued they are objectively true. I responded that Einstein showed that were false. He responded that they were objectively true in the limiting case. I argued that even that might be changed by future developments in our understanding of reality at the quantum level. He argued that they’d still be objectively approximately true, etc, etc.
Here’s one thing that bugs me about discussions of inflation. People talk about whether various price indices are “accurate” as if there is some sort of objective entity that we were trying to measure. If the only good in the economy were 91 octane gasoline, then it might be reasonable to talk that way. But what is the true inflation rate for PCs? The current consensus seems to be that we need to do some sort of “hedonic” estimates, that is figure out how much better the new computers are than the old ones. But better in what sense? And where are we going with this project, what is the purpose? I can only think of one quasi-objective definition of inflation. The rise in aggregate nominal income required to keep aggregate (per capita) utility constant. Do you have a better definition? If so, I’d love to hear it.
Of course that is not what the government does, they estimate how much better some new features of a PC are compared to the old features, and call it is a quality improvement. But that is not objective at all. Who’s to say what is “better?” For me, PCs maxed out around the 386 or 486 chips, and as they get “better” I just get more hopelessly confused about how to use them. I can’t even figure out how to use our new copier, and it’s driving me crazy because I prefer to do my own copying sometimes. But I’m sure the machine is regarded as “better” by the BLS. And how about those “productivity improvements” like computers that answer your phone calls? I know, the answer is that the hedonic statistical techniques tell us the value of the improvements by estimating how much we are willing to pay for them. But that ignores all the social aspects of utility. A lot of what makes a new product more valuable is novelty. Remember when Ford just made black cars? What if they suddenly came out with a red one? It would sell for more. Now suppose they started making nothing but red ones for 20 years. At that point a black one would suddenly be an expensive novelty, and it would sell for more. Sometimes I think about the fuzzy B&W TV we had in 1965. If I replaced my current 52 inch 1080P Samsung LCD with that set, and had to hold the antenna to stabilize the picture, do you think I would enjoy it as much as I did back in 1965? Especially knowing the better sets that are available today?
I’ve noticed that many “serious” economists approved of the BLS using more and more hedonic techniques as a backdoor way of shaving a few bucks off some old lady’s Social Security check. All in the name of being responsible about entitlements. Suppose a scientific study showed that the CPI actually understated inflation. Suppose that a study showed that people liked to “keep up with the Jones,” and in order to maintain a stable utility people actually needed their incomes to rise as much as the average income in America. It’s not that far-fetched, surveys in the US show little change in average happiness levels in recent decades. (I’m not saying those surveys are right, just that it might be true.) Do you think those same economists (often small government conservatives) would eagerly adopt the view that the CPI needed to be raised based on the latest scientific evidence?
[Now I’m sounding like Paul Krugman. Shamelessly pandering to my new NYT readers.]
If I’m not mistaken the Brits tried to index their basic initial pension for new retirees to their CPI, and after a few decades of falling behind wage increases, the old folks eventually rebelled. They saw through all the hedonics. (By the way, I know this is confusing, but initial benefit levels in the US are indexed to wages, only subsequent increases are indexed to prices. So in America old folks as a class don’t fall ever further behind the young, as they were doing in Britain.)
Don’t get me wrong, I agree that “living standards” have improved a lot in the ordinary materialistic sense of the term; I’m not one of those people who says real incomes haven’t changed since the 1960s. What I’m saying is that there is no obvious way to determine how much better off we are. More stuff? Better stuff? Happier? It all seems arbitrary to me. And if inflation isn’t useful for estimating how much better off we are, then what is it measuring?
Sorry for the long intro, but I’m trying to establish that there is no reasonable, objective way of measuring inflation that even comes close to matching the concept we have in our minds, which is roughly “how much more expensive stuff is.” Or “how much more it costs to live these days.” Since I’m a pragmatist, I fall back on the Rortian question:
What’s the use of it?
We can define inflation any way we wish, in whatever way we find most useful. Before offering my definition(s), I’d like to remind people of the “circular flow” of the economy—income/expenditure. On the top we have the expenditure on final goods and services. On the bottom we have the inputs, the factors of production. It’s symmetrical. Therefore we also have two logical definitions of inflation, the average price of inputs, or the average price of output. There is no obvious reason why one is to be preferred over the other. It is all a question of which is more useful, and for what purpose. I’m going to argue that we should focus on input prices, either the average wage rate, or the average nominal income per person (and I’ll just use NGDP for simplicity.) Price inflation doesn’t seem necessary.
So here are 6 reasons to get rid of inflation:
Reason 1: NGDP growth is a better indicator of monetary shocks than the CPI. The CPI is flawed by mismeasurement of housing. But even if it were accurate, it would not reflect nominal shocks as clearly as NGDP. Starting last year, NGDP growth went from its normal 5% to almost negative 5%. In the long run this sort of reduction will eventually impact only prices. But because prices are sticky in the short run, at first output falls very sharply. The initial fall in prices may be very small, and may hide the severity of the monetary shock.
Reason 2: NGDP is a better target for monetary policy than prices. If you target prices, and there is a severe supply shock such as an oil embargo, then you will need a monetary policy that is contractionary enough to depress non-oil prices. This will also require lower real wages. Since nominal wages are sticky, if you are targeting the price level then unemployment will rise sharply during a supply shock. As Earl Thompson pointed out, in theory a wage target would do best; but because of practical problems I prefer a NGDP target, which is more accommodative during supply shocks. Of course both price level and NGDP targets respond identically to velocity shocks. Another advantage of NGDP targeting is that it leans against the wind during asset bubbles such as 1999-2000 and 2004-2006. Those bubbles almost always occur when real GDP is strong, or above the target path. Thus a NGDP target will show the need to tighten policy in a much more timely way than a price level target.
Reason 3: The AS/AD model can be taught with NGDP (or NGDP minus expected NGDP) on the vertical axis and aggregate hours worked on the horizontal axis. There is no need for the price level. If you prefer a Phillips curve model, you’ll find that nominal wage inflation works better than price inflation (especially during supply shocks.)
Reason 4: Income inflation is a better variable to include in the Fisher equation than price inflation. If the rate of price and wage inflation were identical, then obviously either would work equally well. But what if they differ? Let’s think about this example: We start with zero wage and price inflation. Nominal and real interest rates are 2%. Now suppose price inflation stays at zero, but wage inflation is expected to be 10% a year over the next 30 years. That would cause nominal (and real) wage rates to rise roughly 20-fold over that period of time. Presumably this would be due to rapid productivity gains. How would this affect nominal interest rates? Well, if you knew that 20 years from now you could earn in three minutes what it now takes an hour to earn, wouldn’t that make you much less likely to save (and defer consumption) at a given interest rate? Wouldn’t you demand a much higher interest rate to take account of the fact that an hour’s labor in the future would produce far more goods than an hour’s labor today? And doesn’t this sound pretty much like the logic used to justify the original Fisher effect? But notice that I reached this conclusion in an example with zero inflation by assumption. Similarly, the demand for loanable funds would increase if wage growth were that high, because you could take out a mortgage to buy a house at a very high interest rate, knowing that you would get rapid wage increases to pay off that loan. So in a horse race between input prices (wage inflation) and output prices (CPI inflation), it looks like wage inflation belongs in the Fisher equation, not price inflation.
Reason 5: Many economists argue that deflation can put us into a liquidity trap. I initially had the same view. But Bob Murphy points to examples of deflation that did not result in liquidity traps: America in the late 1800s, or China just this year. With respect to America I think that one factor was that the gold standard pretty much eliminated inflation expectations, so the expected rate of inflation was probably near zero. But the key point is that as long as real growth is strong (and it was in the period from 1870-1900) then you should be able to avoid a liquidity trap as long as deflation is mild. The recent Chinese case is instructive. They have had some deflation in the past year, but because they have experienced fairly high productivity growth, there is still strong demand for credit. The key point is that although the price level has fallen over the past 12 months, the Chinese NGDP is still up about 4%. That’s why I have never thought that China would slip into a Japanese-style liquidity trap. Each country occasionally experiences deflation, but Chinese NGDP growth is far higher.
I have frequently criticized Krugman’s argument that we can’t expect much from monetary policy now because the Fed is too conservative to commit to the high inflation required to get us out of a liquidity trap. But inflation isn’t what matters, it’s NGDP growth. If the Fed committed to 6% NGDP growth over the next 12 months, we’d probably get only around 2% inflation. What’s wrong with that sort of target? So once again, when thinking about liquidity traps it turns out that CPI inflation just muddles the issues, whereas NGDP growth shows what’s really going on.
Reason 6: People often talk about workers’ wages catching up to changes in the price level. But what price level do they want to catch up to? Output inflation or income inflation? Obviously the latter. Essentially there are two components to any worker’s pay increase. One is the average pay increase for all workers (i.e., wage inflation.) And the other component is the change in that worker’s wage relative to the average (which depends on industry-specific issues.) And what role does price inflation play in this labor model? No role at all.
What’s the use of inflation? It’s worthless. Let’s get rid of it. Take it out of our models. Take it out of our policy rules. If indexing is required then use wage inflation. If you can’t afford to do that for Social Security, tell old folks they’ll get wage inflation minus 1%. Remind them that although at age 70 they’ll be playing golf and taking cruises, by 90 they’ll mostly be sitting around watching TV. And by that time even 60 inch 1080P Pioneer Kuros will be considered so crappy that Walmart will be almost giving them away.
PS. Of course I’m being deliberately provocative here. The one example I am aware of where inflation affects behavior is the labor/leisure trade-off. We need inflation to explain why people work fewer hours as society get richer. NGDP isn’t enough. And of course inflation may be useful to growth theorists. What else have I missed?
[To the new readers: This is a wacky, offbeat post. If you want a serious look at my policy views please check out my FAQs. In addition, this link discusses what went wrong last year. And finally, I want to thank Tyler Cowen for his very kind review of my blog.]