Tim Peach directed me to a graph in Paul Krugman’s International Economics textbook (coauthored with Maurice Obstfeld and Marc J. Melitz.) It’s a very elegant display of the long run effect of an acceleration in the money supply growth rate (and roughly describes the US economy from 1963-73):
Here’s how to read these graphs. Graph A shows the money growth rate increasing, say from 5%/year to 8%/year (a steeper slope to the line). Graph C shows that the growth rate of the price level also increases due to the quantity theory of money. Prices take a sudden jump at time=0, which we’ll consider later.
In Graph B you see the nominal interest rate jump up at time=0 due to the Fisher effect—higher inflation leads to higher nominal interest rates. Because the nominal interest rate is the opportunity cost of holding money, this causes money demand to fall at time=0, or if you prefer it causes velocity to rise. Going back to graph C, this drop in money demand explains the sudden jump in the price level at time=0. The bottom line is that when money growth accelerates, prices rise even faster than the money supply, as there’s less demand to hold zero interest money. Both the growing money supply and falling money demand push prices higher.
And in Graph D we see the exchange rate follow the price level, due to purchasing power parity. BTW, an increase in the price of euros means the dollar is actually depreciating.
This is roughly what happened during 1963-73, and the analysis is right out of Milton Friedman’s monetarism. Money growth accelerated, inflation accelerated, nominal interest rates rose, and the dollar depreciated. The process was less smooth than you see here, because in the real world prices are sticky and hence the price level does not jump discontinuously when money growth accelerates.
This exercise helps us to understand the difference between New Keynesians like Krugman and MMTers. Both groups tend to agree on policy at zero interest rates, favoring fiscal stimulus and not worrying about crowding out. Both are skeptical of the efficacy of monetary policy at zero rates (although MMTers are even a bit more skeptical than New Keynesians.)
It is when nominal interest rates are positive that the two schools of thought sharply diverge. When I try to explain the views of MMTers I get shot down. But that’s never stopped me before, and so I’ll try again here. I believe that MMTers would start by claiming that the Fed can’t increase the money supply growth rate, as money is “endogenous”. If you insisted that they consider what would happen if the Fed persevered in trying to force more reserves into the economy, they’d argue that this would drive rates to zero. Krugman is arguing that faster money growth raises interest rates in the long run. This difference helps to explain why Krugman differs from MMTers on the existence of a “money multiplier”.
MMTers would claim that driving interest rates to zero would cause banks to hoard most of the new money, and thus it would not have a multiplier effect. They’d also suggest that it would have relatively little impact on the price level, as aggregate demand is determined by spending, not the stock of bank reserves.
Both New Keynesians and monetarists argue that the Fisher effect is very important when thinking about the long run effect of a change in the money growth rate. In contrast, MMTers seem to pretty much ignore the Fisher and income effects, and view interest rates as being set by the central bank via the liquidity effect. An exogenous increase in the money supply growth rate would lead to lower interest rates, in their view. Because central banks target interest rates, MMTers assume money is endogenous. They basically ignore the vast empirical literature on the “superneutrality of money” when the money growth rate changes.
To be a good macroeconomist, you need to hold two models in your head simultaneously. One is the long run flexible price classical model, such as Krugman illustrates in the graph above. The other is the short run sticky price model, which has special characteristics at zero interest rates. Krugman’s a brilliant macroeconomist, and thus is not attracted to MMT models that have no tools for evaluating the long run impact of changing money supply growth rates.
READER COMMENTS
Daniel
Dec 9 2020 at 3:04pm
Thanks, this is great!
It would also help the reader to hold these two models in their head with another set of graphs for the short run that demonstrate those special characteristics at zero.
Figure 14-13 (I think, I’m looking at slides) is a start.
But it would be great to have a reconciliation (for the non-economist) between these two aside from “one is short-run and the other is long-run so they aren’t completely inconsistent with each other.” In reality, we get a series of short-runs stitched to one another, while the long-run graph, though intended for a long time interval, could just as easily be seen to represent dynamics over the course of an extremely short-run (i.e., the Fed doesn’t just blip money into existence instantly, it takes, say, OMO over time, however quickly). As you note, where there is a discontinuity, there is unrealism, and it makes it harder to express clearly what the models actually expect.
Scott Sumner
Dec 9 2020 at 5:33pm
In practice, it depends on how long it takes for the change in policy to be credible, and thus how long it takes to affect nominal interest rates via the Fisher effect.
Mike Sproul
Dec 9 2020 at 3:34pm
“MMTers would claim that driving interest rates to zero would cause banks to hoard most of the new money, and thus it would not have a multiplier effect. They’d also suggest that it would have relatively little impact on the price level, as aggregate demand is determined by spending, not the stock of bank reserves.”
If the Fed pegs the dollar to gold at $1=1oz, and keeps enough assets to always be able to honor the peg, then new dollars would only be issued when they are wanted. New issues of money would not be “multiplied”, and would not affect either interest or prices. It’s the Fed’s assets and liabilities that determine the value of the dollar, not the level of the peg.
If the Fed suspended the peg for a day, a week, or 86 years, then the above result would not change.
Scott Sumner
Dec 9 2020 at 5:32pm
That isn’t even entirely true under the gold standard. In the 1920s, if the Fed did an open market purchase of bonds it reduced the Fed’s demand for gold reserves. This reduces global gold demand, and boosts the global price level. The reverse happened in the early 1930s, when central banks hoarded gold.
Under fiat money, gold is no longer the medium of account. The price level is determined by changes in the supply and demand for base money. The public has no way to return unwanted base money to the Fed. That’s why open market purchases are inflationary, a point well understood by (efficient) asset markets.
Mike Sproul
Dec 9 2020 at 9:20pm
So-called “fiat money” normally starts out pegged to metal. The peg is initially suspended nights, weekends, and holidays. Nobody is such a stickler as to claim that money is backed and pegged on Friday afternoon, but becomes fiat money on Saturday afternoon.
But what if the peg is suspended from 1797 until 1821? What if the Bank periodically makes noise about resuming the peg during that time, until it finally does so in 1821? Did the pound go back and forth between being pegged and fiat during that time? How about the Fed’s suspension from 1934 until today? Rather than claiming that pegged money transforms to fiat money, we should say is that money is backed and convertible one day, and then becomes backed and inconvertible the next. The key word is “backed”, and as long as money is backed (by the issuer’s assets) it should not be called “fiat”.
“The public has no way to return unwanted base money to the Fed.”
Yes they do–loan repayments and bond trading, to name two. Even if the Fed closes those channels of reflux, and the public finds itself with excess base money, the public can offset that by holding less privately-issued money, and that money will reflux to its issuer.
For the record, nobody (including me) denies that as monetary demand for gold falls, gold will lose value. What I do claim is that when any bank issues new money, in exchange for equal-valued assets, the value of that money will not change relative to gold.
Scott Sumner
Dec 10 2020 at 2:46pm
I define “money” as the base, so there’s no automatic reflux of excessive money issue according to the way I define money.
As for periods of temporary inconvertibility, I’ve written extensively on 1933-34. The reason why gold mattered during that brief fiat money period is that for 9 months the current market price of gold was viewed as an indicator of where gold prices would be decades out in the future. That was because it was widely (and correctly) assumed that FDR would again peg gold prices and that this new price would hold for decades. It did, staying at $35/oz. in global markets from 1934-68.
That’s no longer true of gold prices, which is why gold no longer plays the important monetary role that it played during the brief 1933-34 period of fiat money.
We’ve replaced gold with currency notes as the medium of account. Now all that matters is the supply and demand for base money.
Mike Sproul
Dec 10 2020 at 5:49pm
Suppose the Fed increases the quantity of federal reserve notes by 10%, and refuses to allow them to reflux as loan payments or as payment for bonds. The public has only to swap the unwanted 10% of FRN’s for federal funds. Bingo, 10% reflux. Furthermore, your “money=base” definition aside, the public can also reduce its holding of private checking account dollars by enough to offset that 10% boost in FRN’s. Most textbooks would say that since M1 didn’t change, the price level wouldn’t change. (I’m guessing you break with the textbooks here.)
As for expected pegs, the government can only set a peg at $35/oz it it has enough assets to defend that peg. Thus the value of the dollar is set by the assets and liabilities of its issuer (Just like any other financial security.), and not by supply and demand . (Which would make money unique among all financial securities.)
Market Fiscalist
Dec 9 2020 at 4:37pm
‘“MMTers would claim that driving interest rates to zero would cause banks to hoard most of the new money, and thus it would not have a multiplier effect. They’d also suggest that it would have relatively little impact on the price level, as aggregate demand is determined by spending, not the stock of bank reserves.”’
I am not convinced this is correct. MMTeres favor an interest rate target of zero. At this rate they believe that banks would make loans to any credit-worthy customer who wanted a loan (presumably with some kind of profit and risk markup so the actual rate would be above zero). The central bank would then adjust the quantity of base money to allow the banks to have sufficient reserves to hold against all the loans they would make at this zero rate.
I am sure that MMTers recognize that if an MMT regime overnight reduced the target rate from (say) 5% to 0% this would increase lending and spending and put upward pressure on NGDP. They would not see this as much of a problem though as fiscal policy could be accordingly tightened to prevent inflation. And any effect on unemployment would be mitigated via their Job Guarantee Program.
As far as I can see MMTers literally believe that any interest rate target combined with the right fiscal policy (and the JGP) can keep the economy at full employment with low inflation. They just happen to prefer zero as their rate of choice.
They claim there is tons of evidence that ‘natural rate theories’ are bogus (for example: http://bilbo.economicoutlook.net/blog/?p=4656) and if one accepts that (and I don’t) their theories make some sense.
Scott Sumner
Dec 9 2020 at 5:27pm
You said:
“I am sure that MMTers recognize that if an MMT regime overnight reduced the target rate from (say) 5% to 0% this would increase lending and spending and put upward pressure on NGDP.”
You may believe that, but it’s not what the MMT textbook says:
https://www.themoneyillusion.com/mmt-bleg/
You said:
“MMTeres favor an interest rate target of zero.”
Real or nominal? They don’t seem to pay much attention to the distinction between these two concepts. What if the Fed targets inflation at 10% and then sets an interest rate target of zero?
Market Fiscalist
Dec 9 2020 at 6:36pm
‘You may believe that, but it’s not what the MMT textbook says:’
I agree that much of what MMT says about interest rates is ambiguous. I see that the quotes from your earlier post refer to the effects of a government deficit causing ‘demand pull inflation’ rather than a change in the target interest rate. So one possible reading is that they saying that if the government increases the deficit but doesn’t change the target interest rate then the inflationary effects will be the same no matter what proportion of the new deficit end up as bonds v money at the prevailing interest rate.
‘Real or nominal? They don’t seem to pay much attention to the distinction between these two concepts. What if the Fed targets inflation at 10% and then sets an interest rate target of zero?’
I assume they mean nominal but they don’t generally specify. In any case I think they believe that MMT policies could deliver inflation at 10% with either real or nominal rates set to zero. They just tell the fed what interest rate to target and then the government adjusts the deficit to hit the 10% inflation target. As there is no natural rate anything is possible!
Scott Sumner
Dec 9 2020 at 8:09pm
But they claim that open market purchases are irrelevant. And since they acknowledge that OMPs do reduce interest rates, they are implicitly saying that’s also irrelevant.
You said:
“‘Real or nominal? They don’t seem to pay much attention to the distinction between these two concepts. ”
I’d say that’s a pretty big problem.
As for zero interest rates in a 10% inflation world . . . let’s just say I hope they don’t believe that’s feasible for any expected period of time.
Market Fiscalist
Dec 9 2020 at 11:26pm
‘As for zero interest rates in a 10% inflation world . . . let’s just say I hope they don’t believe that’s feasible for any expected period of time.’
For Wicksellian’s that would clearly not be feasible. But if one assumes (as MMTers appear to) that interest rates have only distributional and not equilibriating consequences and that inflation can be generated by fiscal policy alone (independent of anything the CB does) then any rate of interest and rate of inflation can be compatible with full employment.
Scott Sumner
Dec 10 2020 at 2:48pm
And that’s precisely what’s wrong with their model. A negative 10% real interest rate for an extended period of time would be extremely distorting, which all economists who are not MMTers can easily recognize. The system would quickly blow up.
Alex S.
Dec 9 2020 at 8:14pm
I think the distinction between NKE/monetarist and MMT with interest rates > 0 might hinge on whether the monetary base increase is ***expected*** to be transitory or permanent and that this relates to your response to the first comment regarding whether the central bank’s commitments to a particular policy are credible or not.
Scott Sumner
Dec 10 2020 at 2:52pm
Milton Friedman said in 1975 that in the past 200 years macro had merely gone “one derivative beyond Hume.” I’d say MMTers haven’t even addressed permanent changes in the growth rates of nominal aggregates. If they have, it’s not in the macro textbook I just read.
Rajat
Dec 9 2020 at 10:44pm
Off-topic, but I find it odd that a New Keynesian like Krugman has both prices and the exchange rate jumping simultaneously. By contrast, the Dornbusch model has ‘sticky’ prices rising gradually to the new equilibrium following a one-off boost to the money supply – that’s what generates the ‘overshooting’ of the exchange rate. Here, the real exchange rate never changes. So Krugman is not only a New Keynesian who is an MMTer at the ZLB and a monetarist when rates are positive, but also a New Classicalist. 🙂
Scott Sumner
Dec 10 2020 at 2:53pm
This is a flexible price model he’s showing; he also discusses Dornbusch overshooting, and undoubtedly views that as the more realistic case.
This case might occur in some Latin American contexts where there is an extreme loss of confidence in a currency.
Comments are closed.