In a recent podcast, David Beckworth interviews David Andolfatto of the St. Louis Fed. At one point they were discussing an Andolfatto post that suggested inflation can be modeled by looking at the supply and demand for money. (A post criticized by Paul Krugman.) David Beckworth asked Andolfatto about comment left by someone called “Traditionalist”. Here is the comment:
I see that Krugman has linked to this post as an example of “immaculate inflation”. I frequently disagree with Krugman, but I do agree with him on this point, and it’s a useful way of summarizing my arguments above.
I also think it’s a more general point – a close analogue is discussion about the effects of monetary policy that ignores the role of the short-term nominal interest rate. A lot of monetarist-ish bloggers like to criticize mainstream macro’s focus on monetary policy via interest rates, arguing that interest rates are really just an epiphenomenon, a distraction from the main mechanism. But if you try to identify the mechanisms, the actual decisionmaking at the micro level, you realize that interest rates are absolutely crucial, and if anything *money* is the distraction.
The Fed has traditionally implemented monetary policy by intervening in an obscure market that virtually no households or firms participate in; this only influences real behavior because it affects much more important interest rates via expectations and arbitrage, and these interest rates matter for decisions. Meanwhile, the frictions associated with base money are trivial for most decisionmaking because it is almost costless to convert between base money and other assets (whether we’re using an ATM for paper currency or Fedwire with electronic reserves). Even if we cared about these frictions, the only way that the Fed’s policy can affect a household’s portfolio allocation to paper currency is indirectly, through interest rates (since again, the Fed does not engage directly with households); maybe you’ll withdraw fewer dollars from the ATM if the opportunity cost of foregoing interest is a bit higher.
You can’t escape the central role of interest rates as a summary of what matters for household decisions; toy models can obfuscate it, but it’s no accident that rates are central to most larger-scale models.
I’m one of those “monetarist-ish bloggers” who regards interest rates as just an epiphenomenon. And I strongly believe that not only can the analysis of monetary policy and inflation be done without the interest rate transmission mechanism, interest rates add nothing of value to the analysis.
I would take issue with how Traditionalist describes the way that monetary policy has traditionally been implemented:
The Fed has traditionally implemented monetary policy by intervening in an obscure market that virtually no households or firms participate in
This is inaccurate. Traditionally (i.e. before 2008) monetary policy was 98% the exchange of currency notes for bonds, and 2% the exchange of electronic reserves (aka “fed funds”) for bonds. Now it’s true that at the moment of the transaction, the bonds were always exchanged for fed funds. But that’s completely unimportant, as almost all of the extra fed funds were converted into currency within a few days. It would have made no substantive difference if the Fed had directly exchanged currency for bonds, as that’s the form taken by the new money within a few days.
Traditional monetary policy best thought of as the exchange of cash for bonds. Period, end of story.
Because the public is heavily involved in the currency market, they respond to currency injections by getting rid of any “excess currency balances”. An individual can do this by bringing the money to a bank. But for society as a whole, the only way to get rid of excess currency balances is by spending the money, until NGDP is driven up to a high enough level where those currency balances are once again desired for transactions and hoarding purposes.
Thus traditional monetary policy was all about injecting enough currency to create enough excess currency balances to make NGDP grow at about 5%/year (from 1990-2007).
Yes, the Fed had a short term fed funds rate target, but interest rates played no more role in the transmission process than does the price of raspberries.
PS. If you want to use interest rates in monetary analysis, you’d be better off using them as a factor helping to explain currency velocity. The currency stock rose faster than NGDP during 1984-2008 because currency velocity slowed. And currency velocity slowed because nominal interest rates were trending downwards, and nominal interest rates represent the opportunity cost of holding currency. As that opportunity cost declined, currency demand rose as a share of GDP. And currency demand as a share of GDP is the inverse of velocity.
But of course in this sort of model, low interest rates are contractionary, ceteris paribus. They boost the demand for currency.
PPS. The currency spike in 2000 represents precautionary injections before “Y2K”. The smaller spikes each year were to accommodate the Christmas shopping season.
PPPS. Some people insist that the currency stock is “demand determined”, as if in 2008 the people in Zimbabwe suddenly had a desire for trillions of dollars in currency. Yes, the currency stock is demand determined at a moment in time (say Christmas) if you are targeting something other than currency (such as interest rates, exchange rates, or M2), but that’s missing the point. Those targets get adjusted over time to hit various macroeconomic goals, by moving the currency stock away from where it would be if the targets were not being adjusted.
PS. I have a blog post at a new AEI symposium on how the US can best compete with China.
READER COMMENTS
Iskander
May 24 2018 at 1:03pm
It surprises me how much very intelligent people want to make monetary economics not involve money.
Philo
May 24 2018 at 6:58pm
“And currency velocity slowed because nominal interest rates were trending downwards, and nominal interest rates represent the opportunity cost of holding currency.” Couldn’t you equally well say that currency velocity slowed because inflation expectations were trending downwards, and people were less and less averse to holding currency, since it was expected to lose value more slowly? This formulation doesn’t mention interest rates, at least not explicitly.
Benoit Essiambre
May 24 2018 at 7:14pm
This money focused way of seeing things keeps badly fitting the way I reason about this. I think it’s mostly because it ignores investment too much.
Holding liquid cash isn’t only in competition with holding government bonds supplied by central banks. The third and more important alternative is owning real investment such as businesses, land, durables etc.
Central banks cannot, at least in the short term, affect real returns on these investments. There will always be this third, more uncontrollable alternative to holding cash or government bonds.
When central banks lower interest rates, sure there might be a certain amount of money staying more liquid instead of going into government bonds, but people don’t tend to keep large amounts of cash as savings anyways. They invest their savings. The shift between cash and government bonds should result in a small effect on the whole economy.
Shouldn’t central bank actions result in much larger shifts from government bonds to investment.
Now if there were no net new investment created, yes things might be dominated by a “hot potato” effect. When you buy investment from someone selling it, this person ends up with a bunch of cash looking for somewhere to go.
But new investment can also be created. Low interest rates can bring people in the economy, get them working, get more businesses built. This will create more wealthy people, who then have a higher need for liquidity. These newly rich will want to sit on more cash negating the hot potato effect.
The amount of net new investment created is more function of interest rates than function of the amount of floating liquidity. It’s the cost of capital calculation. The shift in the propensity of people wanting to hold private investment versus government paper seems, to me, to be the most important effect of central bank action. In my mind, the shift in propensity to hold two kinds of government paper seems much less important to the real economy. Isn’t this the reason that low central bank interest rates are not contractionary?
Scott Sumner
May 25 2018 at 11:21am
Philo, Yes.
Benoit, You are making several important errors. First, you are reasoning from a price change. Don’t talk about the effect of interest rates on investment, without first specifying what causes the interest rate to change. Are the low rates due to easy money or tight money? It makes all the difference in the world.
Second, don’t talk about people shifting out of government bonds. When one person sells, another person buys the bond. There is no net shift in or out of government bonds.
mike davis
May 25 2018 at 11:54am
I thought about submitting this anonymously since I suspect I’m about to betray shameful ignorance but I’ll just press send and hope that in exchange for suffering public embarrassment I can learn something.
Why the fascination with currency? Here’s how my reptilian macro 101 brain thinks about all of this: Suppose I need to fix my roof and so I sell a $10,000 bond to the Fed. (Maybe I decided that now is a good time to fix my roof since interest rates are really low but that’s not essential to my story.) They give me a check that I deposit in my bank. My bank sees its reserves at the fed go up by $10K (the asset) and their demand deposit liabilities go up by $10K. Now contrast two scenarios: (1) I draw $10k in cash from the bank and use it pay a guy to replace my roof. (2) I write $10k check to the guy who replaced my roof.
In both cases I am simply transferring a debt instrument that I own to the roofer. In the first scenario the instrument says “In God We Trustâ€. In the second scenario the instrument says “In Mike We Trustâ€. It is also true that the first scenario results in the Fed converting reserves into cash. In other words, they’ve converted one form of liability, “reserves held by Mike’s Bankâ€, into another form of liability, “Federal Reserve Notesâ€. In the second scenario the Fed is also converting one type of liability into another—“reserves held by Mike’s bank†have now become “reserves held by roofer’s bankâ€. In either case it’s the transfer of the liability (aka, the spending) that causes NGDP to go up. Again, I don’t know why we worry about whether it is a transfer of currency.
I understand that (unless there is some exogenous change in the demand for currency), the open market operation will ultimately increase the stock of currency in circulation. And maybe Scott’s point is just that currency is a much better target for monetary policy than interest rates. I’m pretty sure he’s right about that (did I hear somebody say, “never reason from a price changeâ€) but why currency instead of some broader monetary aggregate?
Benoit Essiambre
May 25 2018 at 10:40pm
>Are the low rates due to easy money or tight money?
Isn’t it implicit in the context of money competing as a store of value, with uncontrolled real returns on investment. Low interest rates due to past tight money doesn’t makes sense here.
>When one person sells, another person buys the bond. There is no net shift in or out of government bonds.
I guess this depends on how you define things but is this still true when it is the central bank doing most of the buying?
Sure if the economy is at full capacity, once the central bank takes bonds and provides cash, that cash will have no choice but to hot potato the economy to higher prices, but what about when the economy isn’t at full capacity and the cash unjams the investment market, unlocks the creation of new capital, makes people work and become more wealthy and raises demand for liquidity?
And this is not the only case where hot potato won’t happen very much. Every time money is expected to retain real value better than marginal new capital, won’t savers tend to want to hold cash instead of investment? Isn’t the differential between real interest rates and real marginal investment returns an important determinant in changes in velocity?
Plato’s Revenge
May 26 2018 at 3:45am
You have just described the hot potato effect 🙂
NB the theory says NGDP will grow, not necessarily inflation. Very crudely put, if supply is constrained, hot potato -> NGDP growth will flow into inflation, if supply is NOT constrained, it will go into real growth
Benoit Essiambre
May 26 2018 at 3:26pm
I’m not saying that the hot potato effect is wrong. I am wondering if it is the best way to reason about monetary policy. I guess if you are talking narrowly about a more or less direct effect on inflation from monetary injections, the hot potato effect may do the trick, especially if you add an asterisk about velocity being affected by slack and other things like the zlb. But the more concerning aspects of central bank action are the real effects. Things like unlocking new capital to allow the economy to get near full capacity.
In a way, inflation is uninteresting. In an ideal world, money would be neutral and we wouldn’t have to care at all. Bad monetary policy blocks capital from existing and usually does this just at the wrong time. One way this blockage manifests itself is as a mismatch in real interest rates and expected marginal returns on new capital.
Scott Sumner
May 26 2018 at 8:53pm
Mike, No, I am not advocating the targeting of currency. Just making the simple point that when you produce more currency than people want, its value falls. And that means NGDP rises.
Benoit, Monetary policy has real effects only to the extent it has a nominal effect.
Bruno Duarte
May 27 2018 at 8:18am
Thank you Scott. That is really clear view of the role of debt increasing NGDP. The relationship is not straight forward and still depends a bit on the market:
Dealers regularly take into inventory a large share of Treasury issuance so that
dealer positions increase during auction weeks. These inventory increases are only
partially offset in adjacent weeks and are not significantly hedged with futures. Dealers seem to be compensated for the risk associated with these inventory changes by means of price appreciation in the subsequent week.
(https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr299.pdf)
Also, what happens to NGDP if the increase in monetary supply through debt issuance is spent on imports?
Comments are closed.