I must be a masochist.  I feel like Humphrey Bogart, about to slide back into that leech-infested water.  But here goes:

Suppose we pick a fairly “normal” year, when NGDP growth and nominal interest rates and unemployment are all around 5%.  It might be 2005, 1995, 1985, whatever.  The exact numbers aren’t important.  Now the Fed does an OMP and doubles the monetary base by purchasing T-securities.  They announce it’s permanent.  What happens?

One MMT answer is that the Fed can’t do this.  It would cause interest rates to change, and they peg interest rates.  But the more thoughtful MMTers seem to be willing to let me do this thought experiment, as long as I acknowledge that interest rates would change and that it’s not consistent with actual central bank practices.  I’m fine with that.

So let’s say they double the base and let rates go where ever they want.  I claim this action doubles NGDP and nearly doubles the price level.  MMTers seem to disagree, as I haven’t changed the amount of net financial assets (NFA) at all.

But here’s the Achilles heel of MMT.  Neither banks nor the public particularly wants to hold twice as much base money when interest rates are 5%, as that’s a high opportunity cost.  So they claim this action would drive nominal rates to zero, at which level people and/or banks would be willing to hold the extra base money.  Fair enough.  But then what?  You’ve got an economy far outside its Wicksellian equilibrium.

The MMTers like to talk about cases where large base injections did coincide with near zero rates—The US in 1932 or 2009, Japan in the late 1990s and early 2000s.  But those were all economies that were severely depressed and/or suffering deflation.  I find it hard to believe that you could cut rates from 5% to 0% in a healthy economy without triggering an explosion of AD, especially if the economy was already experiencing normal levels of NGDP, normal growth in NGDP, and normal unemployment levels.  The closest example might be the US after WWII, but remember that people (wrongly) expected deflation after the war, and by 1951 the Fed gave up on that policy due to rapidly rising inflation.

MMTers forgot that the nominal interest rate is the price of credit, not money.  The Fed can’t determine that rate, it reflects the forces of saving supply and investment demand.  Hence an attempt to set interest rates far below their correct level in savings/investment terms (the Wicksellian natural rate), would trigger an explosion in AD, and much higher inflation.  Central banks know this, which is why after the inflationary 1965-1981 period they adopted the Taylor Principle.

That’s the flaw with MMT; it’s not net financial assets that matters, it’s currency.   And the Fed doesn’t set interest rates, markets set interest rates.  The Fed can briefly push them out of equilibrium (due to sticky prices) but this triggers big changes in AD and the price level.

The whole point of my Quantity Theory of Money post (and especially the Canadian/Australian comparison) was to smoke out their views of currency and the price level.  It was hard sifting through all the comments, which were often on side issues, but it seems they regard base money as just another financial asset.  But it’s not, which is why their view of monetary policy is wrong.  Indeed in a sense they don’t even have a theory of monetary policy, they have a fiscal theory that implies open market operations don’t matter.   But the Canadian/Australian data tells us that currency does matter, and NFA is the wrong aggregate to look at.

This is my very last MMT post . . . until the next one.

PS.  Quiz question:

1.  Sumner claims that a 5% NGDP growth rule will lead to roughly 5% NGDP growth, 5% interest rates, and 5% unemployment.  What would a 3% NGDP target lead to?

Answer:  About 3% NGDP growth, about 3% interest rates, and about 5% unemployment.

PPS.  The previous MMT post has 292 comments, and counting.  I may not have time to answer all the comments here.  In that case I’ll answer the 1% or 2% that actually comment on what I say here.

I recently did a post trying to figure out whether there are any non-quantity theoretic models of the price level.  It led to one of the most intense debates I’ve ever seen in my comment section, and even other bloggers chimed in with posts.  But no one came forth with a non-quantity theoretic model of the price level.  It is very important that any monetary theory be able to explain why prices aren’t 100 times higher, or 100 times lower.  Thus I’m more inclined than ever to think the QTM is the best starting point for monetary theory (although obviously it’s not literally true that M and NGDP grow at the same percentage rates.)

I wasn’t able to fully grasp how MMTers (“modern monetary theorists”) think about monetary economics (despite a good-faith attempt), but a few things I read shed a bit of light on the subject.  My theory is that they focus too much on the visible, the concrete, the accounting, the institutions, and not enough on the core of monetary economics, which I see as the “hot potato phenomenon.”  This is the idea that the central bank controls the total quantity of money, but each individual controls their own personal “money supply.”  So if the Fed injects more money into the economy, something has to give to equate money supply and demand.  Initially there is too much money in circulation, and people pass the excess balances to one another like a hot potato.  This process drives up NGDP, until the public is willing to hold the new quantity of money.

Importantly, it’s very hard for individual people to see how this process works, as the Fed injection of cash doesn’t make anyone richer.  They swap cash for bonds, at fair market value.  But if no one is richer, why should AD go up?

The easiest way to see the process work is to imagine an economy without banks, where the new money goes right into circulation as currency.  Most people can instinctively grasp that more currency, without any increase in real goods being produced, will lead to inflation.  But when you add a banking system it’s much harder to see the hot potato effect, because now the new money can show up as either cash or bank reserves.  It looks like individuals who didn’t want to hold excess cash, could simply put it in the bank.  But of course the bank usually doesn’t want to hold a lot of excess cash (reserves) either, and so you can still have the hot potato effect.

Now let’s look at an example, first from my perspective, then theirs:

The Fed wants to raise the price level by 10%.  So they decide to suddenly increase the monetary base by 10%, and then continue on with the same money supply growth rate as before.  This should cause a 10% one-time rise in P, and in NGDP, compared to the no-action alternative.  But if they actually did this in a modern economy, it would create a big mess.  NGDP doesn’t change immediately, so it’s be hard to generate demand for that extra cash.  Even so, the Fed can literally force base money into the economy, by selling [I meant buying] bonds.  I believe the MMTers, argue that trying to do this would drive rates to zero.  That may or may not be true; they tend to overlook that the interest rate isn’t just the price of money, it’s also the price of credit.  So a highly expansionary policy can increase interest rates under certain conditions, for certain maturities.  But let’s assume rates did go to zero.  Then AD would rise, and eventually NGDP would increase 10%.  At that point the public is willing to hold the larger cash balances, and the nominal interest rate returns to its original level.

Because this process would be messy, real world central banks would use a much more subtle process, involving signaling.  They will signal the desire for 10% higher NGDP though various mechanisms–a higher inflation target, a lower exchange rate, or most commonly, a lower fed funds target rate.  If credible, this signal will boost AD.  To some that seems like handwaving (the inflation target more so than the interest rate.)  It’s actually an implied commitment to provide 10% more base money at that future date when NGDP is 10% higher.  But in that case the cause of the higher NGDP (more cash in the long run) seems to occur after the effect (higher NGDP growth.)  To many people, that is deeply disturbing.  An observant reader will have noticed that cause doesn’t actually follow effect in this case, the true cause of everything is a sudden expectation that future levels of currency will rise by 10%.  So cause actually does precede effect.

My hunch is that the MMTers are fooled by this process.  They probably have a better understanding of modern central banking than most non-MMTers, certainly better than mine.  They see the central bank targeting rates, and when the target changes, there is often almost no immediate change in the monetary base.  Instead, things like loans may pick up.  To prevent the interest rate from deviating from the target, the central bank is virtually forced to respond to those bank actions by adding more reserves.  This makes the monetary base seem endogenous, and in the extremely short run it is, at least under modern central bank practices.  In the future, the advent of IOR may make central banking resemble the MMTers model even more closely.

Nevertheless, even though money seems endogenous, it actually isn’t.  A permanent peg of the interest rate would leave prices unanchored, or indeterminate.  Thus the central bank moves rates around in a fashion that will eventually move the monetary base around in a fashion that will tend to keep P and NGDP on the target growth path.  So the base is actually doing all the heavy lifting, even though the specific procedure used by central banks makes it seem like the tail of the dog.

That’s why it’s so important to do thought experiments with monetary regimes lacking a banking system.  This allows us to first work out the basic principles of what determines the price level, i.e. what determines the value of money.  Once we’ve done that we can ask whether adding banking actually changes anything fundamental.  I say it doesn’t, but obviously the MMTers disagree.

Still it seems to me that anyone attacking my position first needs to develop a model of the price level (not inflation, but the level of prices.)  I’m convinced that only the QTM can do this, and still explain why Australia and Canada have similar price levels but Canada has more than 5 times as large government liabilities.  My answer is that both countries have similar currency stocks (per capita.)  And it’s the currency stock that matters; not total government liabilities.

The best way to understand modern sophisticated central banking is to study the most primitive monetary system possible–a medieval king debasing his money in a country lacking banks.  The causal chain between debasement and inflation is no different from the causal chain between OMPs of T-securities and inflation, at least in the long run when nominal rates rise above zero.

This is sort of a response to some Keynesian/fiscal theory/Post Keynesian/MMT theories I’ve seen floating around on the internet.  Theories that deny open market purchases are inflationary, because you are just exchanging one form of government debt for another.  But first a few qualifiers:

1.  If the new base money is interest-bearing reserves, I fully agree that OMOs may not be inflationary.  That’s exchanging one type of debt for another.  If it does raise inflation expectations (as QE2 did) it’s probably because it changed expectations of future monetary policy.

2.  If nominal rates are near zero, the situation is complex–I’ll return to that case later.

So let’s start with an economy that has “normal” (i.e. non-zero) interest rates, and non-interest-bearing base money.  How does the price level get determined in that case?  I’m told there are some theories of fiat money that suggest it must evolve from commodity money.  I don’t agree.  I think the quantity theory of money is all we need.  Suppose you dump 300,000 Europeans on an uninhabited island—call it Iceland.  The ship also drops off some crates of Monopoly money, and they’re told to use it as currency.  Assume no growth for simplicity.  Also assume no government and no banking system.  It’s likely that NGDP will end up being roughly 15 to 50 times the value of the stock of currency.  Once you pin down NGDP, then you figure out RGDP using real growth theories, and voila, you’ve got the price level.  At this point you might be thinking; “you consider ’15 to 50 times the currency stock’ to be a precise scientific solution?”  No, but it gets us in the ball park.  It tells us why prices are not 100 times higher than they are, or 1000 times higher.   BTW, prices in Japan are 100 times higher than in the US, and Korean prices are 1000 times higher.  I don’t see how other theories can even get us into the right ball park.

I’m going to illustrate the problem of non-QTM theories of the price level with a comparison of the US  Australia and Canada.  Here are some national debt figures from The Economist:

For simplicity assume Australia’s net debt was zero in 2007.  In Australia NGDP is about 30 times the currency stock.  Canada is similar.  (The US NGDP was only about 18 times the currency stock in 2007, because lots of our currency is hoarded overseas.)  This ratio is determined by the public.  The base also includes reserves, but in normal times like 2007 we can ignore those if we aren’t paying interest on reserves.  The opportunity cost of holding reserves is simply too large for banks to want to hold very much.  So the central bank determines the nominal base, and the public determines the ratio of NGDP to the base (aka velocity.)

Because Australia and Canada are fairly similar countries, I can get a reasonable estimate of each country’s price level as follows:

1.  Notice that their RGDP per capita is similar.

2.  Find the NGDP in one country (say Canada.)

3.   Find the currency stock in each country.

4.  Assume their NGDP/currency ratios are similar (roughly 30.)

Then all I need is Australia’s currency stock to estimate the price level in Australia.   Now suppose it was true that OMOs didn’t matter.  In that case the aggregates that would be important would be the entire stock of government liabilities, currency plus debt.  But as you can see, Canada’s was many times larger than Australia’s.  (Recall that in both countries currency is only about 3% to 4% of NGDP.)  If you looked at total government liabilities you’d get nonsense, you’d estimate Canada’s price level in 2007 to be between 5 and 10 times that of Australia, as its debt was 23.4% of GDP (so debt plus base was about 27% of GDP), vs. about 3% to 4% in Australia.   The base is “high-powered money” and interest-bearing debt isn’t.  Demand for Australian cash is very limited; you just need a little bit to smooth transactions in Australia.  Double it and the value of each note falls in half.  Double the amount of Australian T-bonds, and it’s just a drop in the bucket of a huge global market for interest-bearing debt.  The value of those bonds changes hardly at all.

Now suppose that in 2007 the US monetized the entire net debt, exchanging $6 trillion in non-interest bearing base money for T-securities.  And suppose this action is permanent.  The monetary base would have increased about 8-fold, and the QTM tells us the US NGDP (and price level) would also have increased 8-fold.  In that case our situation will be much like that of Australia; we’d have a monetary base, but no interest-bearing national debt.  So our price level would be determined in the same way Australia’s price level is determined.  NGDP would be some multiple of the base, depending on the public’s preference to hold currency (including foreign holdings of US currency.)   But since our base (and currency stock) went up 8-fold, if the ratio of NGDP to currency remained around 18, then the level of NGDP would also increase 8-fold.  That shows OMOs do matter, at least if I’m right about the public’s demand for currency usually being some fairly predictable share of NGDP.

Here’s my problem with all non-QTM models.  Suppose I’m right that only the QTM can explain the current price level.  Then it stands to reason that only the QTM can explain the price level in 2021.  Then it stands to reason that only the QTM can explain the inflation rate between 2011 and 2021.  Now it is true that a change in the money supply will have certain effects on nominal interest rates, economic slack, etc, depending on whether the monetary injections were expected or not.  And you can try to model the inflation rate using those changes in interest rates, economic slack, inflation expectations, etc.  But that’s really a roundabout way of getting at the problem.  If the QTM says that the price level in 2012 will be 47% higher due to changes in the monetary base, plus changes in the public’s desire to hold currency as a ratio or NGDP, then either the non-QTM approaches also give you the 47% answer, or they are wrong.

Here’s a nautical analogy.  You can estimate how fast a cigarette boat was going by looking at the size of the engine, the throttle setting, and so on.  That’s the direct approach, the engine drives the boat.  Or you can estimate its speed by how big its side effects were (the size of the wake, how loudly seagulls screeched as they got out of the way, etc.)  The engine approach is the QTM.  That’s what drives inflation.  (God I hope at least Nick gets this, otherwise I’ve totally failed.)  The Keynesian approach is to look at epiphenomena (like interest rates and slack) that may occur because wages and prices may be sticky to some unknown extent.  It’s like looking at the wake and trying to estimate what sort of boat went by.

OK, what about at the zero bound, aren’t cash and T-securities perfect substitutes?  Maybe, but if they aren’t expected to be perfect substitutes in 2021, then  a current OMO that is expected to be permanent will have the same impact on the expected long run price level as an OMO occurring when T-bill yields are 4%.

Of course central banks don’t target the base, they adjust the base until short term interest rates are at a level expected to produce the right inflation rate.  It’d be like adjusting the throttle until the wake looks about the right size to hit the target speed.    And in the future they might go even further away from money supply control, if they pay interest on reserves.  In that case they’ll be adjusting rates and the base in a more complicated pattern, both money supply and demand will change.  But the currency stock will still be non-interest bearing for a while, so that relationship will continue to hold.

What would cause a revival of monetarism?  That’s easy.  We just need to return to widely varying trend rates of inflation, as we saw in the 1960-1990 period.  In those decades countries might have 5%, 10%, 20%, 40% or even 80% trend inflation.  As that settles in, and people expect it, the various epiphenomena of unexpected money go away (liquidity effect, slack, etc.)  And everyone goes back to explaining inflation by looking at growth in the non-interest bearing monetary stock.  It’s the only way.  The best example was in the hyperinflationary early 1920s, when even Wicksell and Keynes, the two great proponents of the interest rate approach, became quasi-monetarists.  Needless to say I have very mixed feelings about the prospect of a revival of monetarism.

So here’s my question:  Are there any non-quantity theoretic models of the price level?  Theories that could explain the difference between Australian and Canadian and Japanese and Korean price levels?