When I listen to discussions of monetary policy, I’m often struck by the extent to which outdated terms and concepts are fundamental to the conversation. I don’t mean that the academic discourse is backward—in fact, it moves along rather quickly. But discussions among practitioners, and in the financial press, fall back on the same obsolete ideas over and over.
Obsolete idea #1: The CPI as a central indicator for monetary policy
When the conversation turns to inflation, most of the time it’s assumed that we’re talking about the Consumer Price Index, as if that were the only price index that could conceivably matter. In reality, of course, there are plenty of other price indices—the Producer Price Index, the GDP deflator, nominal wages—with an equal or better claim to relevance in monetary policy. The CPI is not even close to the correct measure for the purposes of either output stabilization or efficient nominal contracting.
Non-economists, of course, often assume that the Consumer Price Index is important because they have a deeply flawed understanding of what the Fed does. They think that inflation is bad because stuff costs more, making them poorer—but this isn’t really “inflation”, it’s the relative change in prices and wages. To the extent that the Fed controls “inflation”, it controls the overall price level, which includes both wages and consumer prices; it can’t repeal the laws of supply and demand and make workers’ labor more valuable in terms of goods and services. In fact, any attempt to do this (by, say, implementing sudden deflation and exploiting the fact that wages are stickier than prices) would be deeply contractionary.
The Consumer Price Index is a perfectly sensible way to make cost-of-living adjustments to Social Security. But there’s no reason save sheer inertia for it to be considered so fundamental to monetary policy.
Obsolete idea #2: the NAIRU
The “Non-Accelerating Inflation Rate of Unemployment” dates back to Milton Friedman and his dispute with the Old Keynesians. At the time, it was an excellent contribution: based merely on a short-term correlation, the Old Keynesians (despite having no good theory to back themselves up) thought that there was a long-term tradeoff between inflation and unemployment. Friedman argued, to the contrary, that there was a “natural rate of unemployment” pinned down by real factors in the labor market, and that at best monetary policy could manipulate this outcome in the short run. Any attempt to keep unemployment below its natural rate would force the Fed into an endlessly accelerating spiral of inflation.
This is indeed a great conceptual insight—but that doesn’t make it a useful way to implement monetary policy in practice. The problem is that it’s very difficult to know what the “NAIRU” really is. As Nancy Stokey pointed out in a 2000 interview:
I think [the NAIRU] is not really such a good one. And I think that the empirical evidence points in that direction pretty strongly. In the 1960s people would have said the NAIRU was around 4 percent, then it crept up to around 8 percent, leveled off and remained there for a while. Now it’s crept down, and I guess it’s around 4 percent again. A number that is as variable as that over a few decades is probably not a good number to build policy around. It doesn’t seem to be a stable number.
It is far simpler and more accurate to target wage inflation instead. After all, the logic of the NAIRU tells us that an attempt to bring unemployment below its natural level inevitably leads to inflation—and, in particular, that inflation in wages is a very good measure of the extent to which labor markets are being overstimulated. If we keep wage inflation at a stable level, we can be sure that we’re staying somewhere in the neighborhood of the NAIRU: there’s no need to make clumsy guesses about what the underlying quantity might be.
Obsolete idea #3: Monetary Aggregates
Anyone learning about monetary economics is sure to run into a bewildering array of “monetary aggregates”: M1 (currency plus checking deposits), M2 (M1 plus savings deposits, small time deposits and money market deposit accounts), MZM (M1 plus savings deposits and all money market funds), and so on. One might be led to believe that these are key quantities for conducting monetary policy. This is not even close to being true.
In theory, monetary aggregates might be useful because they provide ways to represent the “M” in the equation of exchange, MV=PY. The idea is that if velocity (the “V” term) is stable, keeping “M” on a steady trajectory will help to stabilize prices (“P”) and more broadly nominal GDP (“PY”). Unfortunately, velocity is not stable, and changes in monetary aggregates (especially the broader ones) are just as likely to reflect some non-monetary financial innovation as they are to reveal any deep macroeconomic trend.
More to the point, it’s not clear why monetary aggregates should be important for monetary policy. Aside from its ability to create cash, the Fed’s direct impact on monetary aggregates is supposed to come through reserve requirements: as the supply of bank reserves expands, banks can create more deposits. But this isn’t relevant at all for monetary aggregates larger than M1, since reserve requirements today only apply to checking deposits. Moreover, required reserves are only a tiny part of the monetary base, which (in normal times) consists almost entirely of cash. Since the monetary base is the quantity the Fed actually controls, any quantity-theoretic approach to conducting monetary policy must place most of its emphasis on the demand for cash, which plays a dominant role in the demand for base money in general.
Monetary aggregates like M1 do not do this. Instead, they treat checking deposits in exactly the same way as cash. If the two were obviously close substitutes, this might be forgivable, but in reality the demand for checking deposits and the demand for cash are quite different: the former is used to provide liquidity for general transactions, while the latter is used mostly to stash money for non-transactional purposes in the form of $50 or $100 bills, often outside the US.
Since the demand for base money is so difficult to predict (and monetary aggregates provide minimal guidance), the modern instrument for monetary policy is the short-term nominal interest rate. This instrument can be used in many different ways—for instance, to implement a Taylor rule, or maybe inflation targeting. But in the day-to-day conduct of monetary policy, it is essential, and monetary aggregates are meaningless.
This isn’t to say that monetary aggregates are completely uninformative: since they place such an emphasis on deposits, they do provide a useful picture of the banking system. It’s hard to imagine, however, that they provide a particularly effective way of capturing this information—after all, that’s not what they were designed to do.
I don’t claim to have any grand theory for why these ideas persist long after they have become academically and practically obsolete. It’s probably just inertia. But it is frustrating, and I hope that someday we’ll manage to make the conversation a little less outmoded.
Matt Rognlie 
I think you might be throwing baby out with the bathwater by dismissing CPI altogether and not making a distinction between core and headline CPI inflation which seems to be what you are talking about.
Also, as far as the NAIRU, is there even any good empirical evidence to support the idea of wage-price inflation at all? I know the theory behind wage targeting a little bit but I think the idea of wage targeting is creepy.
I think that core CPI is a better measure than headline CPI for the purposes of monetary policy (it certainly is better at satisfying the criteria in the post/paper I linked), but I wanted to stress the more fundamental point that there is no reason for any variant of the CPI to be a particularly good indicator. I think this brings clarity to why complaints about the core CPI are so ridiculous—it’s not as if headline CPI is some god-given perfect measure of monetary policy.
I’m not quite sure what “wage-price inflation” means in this context. Certainly any increase in “pure” inflation (an across-the-board increase in prices, not just a relative price movement) must include increases in both wages and prices, by definition. Since this phenomenon is what the Fed is really trying to control, ex ante it is conceivable that the Fed may use either wages or prices (or both) to measure the rate of pure inflation.
I don’t think that it’s really creepy: nominal wage targeting means that improvements in productivity (above and beyond the given rate of increase for nominal wages) will show up in lower nominal prices rather than higher nominal wages. Is this really such a problem? It seems just as easy to say that price level targeting is creepy.
Sorry, by wage-price inflation I meant wage-price spirals or I guess more to the point is if there isn’t really a NAIRU then there isn’t really a AIRU. I guess I’m asking if you could point me at anything that suggest that wages are the real determinant of inflation. I know its taken for granted theoretically.
The reason I think wage targeting is creepy is that it targets only one piece of the inflation puzzle. The more “blunt” instrument of targeting prices force both parties (capital and labor) to “share the pain” Since wage increases haven’t really kept up with productivity increases anyway over the last 30 – 40 years it isn’t obvious that capitalists wont capture the difference in productivity increases above the wage target.
Of course, I guess nothing stops the owners of capital from capturing productivity gains anyway, but I think institutionalizing that practice is a terrible idea.
I’m not super familiar with the literature cause inflation hasn’t interested me too much until lately, but conclusions of wage targeting come out of General Equilibrium models in which the only costs are wages and there is no real rigidity to the response of wages to demand?
Also, I’ve noticed you’ve been bringing up the point that relative prices rises aren’t inflation. I recently came across a paper by Ball and Mankiw (QJE, 1995 p161-193) that argues that supply side inflation shocks are driven by asymmetric relative prices changes. For instance, an oil shock causes inflation because other prices don’t respond downward due to menu costs.
Oh, and I do think price targeting is creepy but for different reasons than we are discussing here!
It seems you are conflating two different propositions: (i) the various monetary aggregates are a poor measure of the quantity of money, and (ii) the quantity of money doesn’t really matter much, only interest rates and credit markets do. Point (i) has some merit, but point (ii) is nothing but an assertion. The fact that many economists have constructed models in which money doesn’t matter does not make it true.
As Nick Rowe has pointed out time and time again, the medium of exchange is special, and it’s special in a way that has nothing to do with credit. I’m not going to go into how and why here because you’ve already seen Nick’s stuff, although it doesn’t seem to have sunk in yet.
I still don’t understand why the quantity of base money is supposed to matter beyond its impact on interest rates—even the effect of complementarity the utility function on real outcomes can be expressed via the nominal interest rate as an intermediate indicator.
I’ve seen Nick Rowe’s argument. Here it is from the comments section on my MV=PY post:
As I said then, I have a lot of trouble understanding this argument (which is stated in greater depth here). Part of the problem, I think, is that the terms we’re using are garbled. If, when Nick talks about money, he is talking about liquid assets, then he may be right (and I even partly agree with him), but if he literally means base money this doesn’t make much sense. It is trivial to exchange any liquid asset for base money at any time; that’s what makes them liquid. It would be really weird if someone who wanted more base money decided to accumulate it by reducing consumption, rather than just withdrawing it from an ATM.
I agree that excess demand for liquid assets can be a big problem—it forces up the liquidity premium, which (because base money is a liquid asset guaranteed to pay zero) means that interest rates on less liquid assets are forced to increase, possibly beyond their equilibrium level. Maybe we don’t really disagree, and we’re just using the word “money” to mean different things. But if not, I still fail to understand why excess base money demand should mean anything special.
Excellent post. A few comments:
1. One way to explain your first point is as follows. The public thinks inflation is bad because it reduces their real income. And the sort of inflation they notice most (supply shocks) does reduce their real income. But inflation created by the Fed shifts the AD curve to the right, raises real GDP in the short run, and hence raises aggregate real income for Americans. (Real hourly wages may fall, but hours worked increase.) So if inflation is making you feel grumpy (in the aggregate) it’s not the sort of inflation created by the Fed.
2. I don’t think Friedman came up with the NAIRU, which says higher than normal employment raises inflation. Rather he proposed the natural rate model, which says that higher than expected inflation raises employment. And he argued the natural rate was not stable, and not suitable for policy purposes. So he would agree with you.
3. I agree that the base is the most sensible definition of M, as it’s the quantity controlled by the Fed, but disagree that monetary targeting must be done with short term rates. There is no problem with using variations in the base to peg the price of a CPI futures contract, or a NGDP futures contract. The fed funds rate is an intermediate target, and hence not very useful in a modern Svenssonian “target the forecast” monetary framework.
Peg the price of the goal variable (futures contract) and let both the base and the fed funds rate move endogenously.
Thanks for the comment.
This is probably not the right place to get into this discussion, but while I can see that pegging the price of the goal variable (future NGDP) is conceivably an alternative to the standard approach of using an intermediate target (the interest rate) plus a Taylor rule, I have some worries about implementation. It’s easy for me to imagine investors’ buying-and-selling of NGDP contracts producing massive, massive swings in the fed funds rate; after all, short-term reserve demand is so inelastic that even a tiny change in the amount of bank reserves in the system can lead to a 5 point swing in the short rate.
Perhaps under an NGDP targeting regime, we simply won’t care (after all, interest rates aren’t the target anymore), but I do worry that this kind of short rate instability would be problematic for financial markets. Correct me if I’m wrong (my history isn’t very good), but even long before the advent of modern interest rate policy, part of the justification for establishing the Fed was to stabilize the short rate, right?
1. Using CPI as an indicator for monetary policy is a bad idea but not for the reasons you give. The principal reason is that it does not take the price of financial assets into consideration, and as we know from past bubbles, the price of financial assets is what causes booms and crashes.
2. Monetary aggregates matter. It is just that the existing monetary aggregates M1, M2 etc are based on total misconceptions of the nature of money. A monetary aggregate I have developed, called Corrected Money Supply , provides a much more accurate view of the economy. Right now it shows that we are headed for a crash.
A graph can be seen at http://www.philipji.com/item/2011-05-11/how-much-longer-until-a-crash
I’m glad we agree that the CPI is a bad indicator, but I don’t think it’s quite fair to criticize it for failing to consider the price of financial assets—that is a phenomenon very different from inflation, and one that the CPI was never designed to measure. It is entirely possible for financial assets to undergo dramatic increases in price for real reasons (maybe some news increased expectations of future dividends) with no inflationary pressure whatsoever.
I agree that perhaps the Fed should start taking the behavior of financial markets more into account, but this doesn’t mean that it should conflate inflation in the price of goods and services with increases in asset prices by including them in the same index, when the two are so fundamentally different—this would be needlessly confusing without offering any new information.
Do people really talk about M1 or the NAIRU in the financial press? I mean, I wouldn’t be surprised to find someone on Seeking Alpha ranting about M1, but do writers in the WSJ or FT really talk about those things??
If so, I must be doing a good job of avoiding the shadier corners of the marketplace of ideas…
What do you mean by “wage inflation,” and how would you define it? How can it be done without using some form of price inflation and productivity measures? So how can you target wage inflation without implicitly targeting some form of price inflation, given the productivity change measure?
Aren’t these ideas dead already?
Matt, Yes, I believe the Fed was set up partly to stabilize short term rates. But that’s because the seasonal fluctuations in short-term rates made the economy susceptible to crisis in the fall, when money demand was higher and interest rates tended to rise.
Oddly enough, a suspicious number of post-Fed stock market crashes have also occurred in September/October (1929, 1937, 1987, 2001, 2008.)
NGDP targeting would also tend to eliminate the seasonality of interest rates. I actually would have no problem with the Fed doing some OMOs outside the NGDP futures market to offset what they thought was excess speculation (and perhaps reduce volatility of the fed funds rate), as long as they didn’t try to move the market price for NGDP futures away from the target. In other words, they should be free to gamble like anyone else, but if they lose lots of money then they’d have a lot of explaining to do. (My own view is that fed funds fluctuations wouldn’t cause macro problems as long as NGDP forecasts showed 5% growth expectations.)
BTW, my comment was a bit off topic, as you said. As an aside, I do favor nominal wage targeting, and long ago did a paper combining it with futures targeting.
hey Matt — you know of these people at MIT working on improving inflation measurement? its fairly cool. http://bpp.mit.edu/
No real disagreement really, but I really like NAIRU. Sure it’s no good as a policy instrument but it still has some use as a measure of labor market distress.
And with respect to the history of NAIRU….
In 1975 Franco Modigliani and Lucas Papademos coined the term noninflationary rate of unemployment (NIRU) which was later revised to nonaccelerating rate of unemployment (NAIRU). While Friedman and Phelps believed that the existence of a natural rate implied that there was no useful trade-off between inflation and unemployment, Modigliani and Papademos interpreted the NAIRU as a constraint on the ability of policymakers to exploit a trade-off that remained both available and helpful in the short run. But despite the fundamental differences that still existed between the Monetarists and the Keynesians, the NAIRU was seen by many contemporary economists as helping build a consensus about the nature of the inflation-unemployment relationship.
At first there was an attempt to estimate a fixed NAIRU for the United States that was valid for all time periods (usually estimated to be between 5.5% and 6.0%). By the late 1990s, as unemployment dropped well below that rate and yet inflation continued to be moderate, this idea was questioned. Now it is fairly well accepted that NAIRU varies over time. The CBO maintains a NAIRU data series that ranges from a high of 6.2% in the mid 1970s to a low of 4.8% since 2001.
So Please don’t take my NAIRU away from me!
Hi Matt,
Have you come across by ?
He’s an economics professor at the University of Queensland (in Australia). It was published by Princeton Press and is on a similar theme to this post.
Regards,
Rohan
Hi Matt,
Have you come across ‘Zombie Economics’ by John Quiggin?
He’s an economics professor at the University of Queensland (in Australia). It was published by Princeton Press and is on a similar theme to this post.
Regards,
Rohan
Pingback: Tuesday Early Morning Reads | The Big Picture
I have long wondered about the huge stores of cash held outside the United States. As a side note, when the US switched over to the “new” Ben Franklins (with the larger portrait), did anyone notice large amounts of old BF’s being traded into banks? Are the old BFs just out there yet?
Does the fact that so many people are willing to hold BFs mean that inflationary expectations are low?
It seems the messy reality of cash must play havoc with any finely tuned economic theories.
An intersting post.
Hard to argue NAIRU is obsolete (or should be obsolete) when regional Fed banks continue to publish research on it: http://www.frbsf.org/publications/economics/letter/2011/el2011-05.pdf
Just looking at the FOMC’s latest projections (http://www.federalreserve.gov/newsevents/press/monetary/fomcprojtabl20110427.pdf), it’s clear that some “normal” level for the unemployment rate is on their minds — that is, when they call for the unemployment rate to fall back down to between 5.2 and 5.6 percent over the longer-run.
The other two points you made are fair, but people (both economists and non-economists) will always be interested in what rate of unemployment is normal, and I argue that FOMC policymakers have some rate in the back of their minds when making MP decisions.
Pingback: Articles of the Week: | Stocks on Wall Street
Pingback: Articles of the Week: The Invisible Stock Bubble | Stocks on Wall Street
Real Bullshit. You don’t know what money is. You don’t know what the base is. You don’t know what the multiplier is. You don’t know what money is. You don’t know how to control it. You don’t understand the equation of exchange. You are FUCKING LOST.
“How do hundreds of thsdoanus people leave the workforce yet jobless claims over the past month”…So just how credible was the reporting on thess supposed job losses? It is an AP story after all…BTW maybe just maybe better worker efficiency has something to do with it assuming the job losses story is factual…
I am surprised that no economist wonders why monetary aggregates were a reasonably good economic indicator before 1980 and ceased to be so after 1982, which is when the Fed abandoned using them.
The reason is fundamental to understanding why macroeconomics is in such a mess now. Keynesians, Austrians and monetarists all assume that money is used to buy real goods and services but not to buy financial assets. Before 1980 the financial sector was relatively unimportant but after that year it grew disproportionately in importance. So although the monetary aggregates were always wrongly defined, they did a reasonably good job as economic indicators before that year and ceased to do so after that year.
This is also the reason why inflation (which measures only the prices of real goods and services) has repeatedly failed as an indicator. Before the Great Depression, Japan’s lost decade, and the Great Recession inflation was low or even negative. Central banks which looked only at inflation failed to see that it was a crash in financial markets which would cause disaster.
My ebook “The General Theory of Money” http://www.amazon.com/dp/B0080WPK2I explains this fundamental misunderstanding of money common to all schools of economics and corrects it. It also indicates that we are headed for a crash of epic proportions.