When I was in graduate school, Benjamin Friedman’s paper on targets, instruments, and indicators of monetary policy (appears to be gated) was assigned in several courses. So I think of it as a classic, but mine may be an idiosyncratic perspective.
A target is a policy goal: Unemployment. Inflation. Nominal GDP.
An instrument is something that the Fed controls. The three old-fashioned textbook examples are the amount of reserves (or reserves plus currency), the required reserve ratio, and the discount rate. More recently, the Fed funds rate is the instrument that economists focused on. Even more recently, there is the size of the Fed’s balance sheet.
An indicator is something that the Fed can watch to see whether the economy is moving toward or away from its target. There are plenty of such indicators: private forecasts of NGDP, high-frequency data, such as retail sales figures, etc.
As I see it, one of Scott Sumner’s objectives, in his blog and in his new book The Midas Paradox which I have just started reading, is to get people to pay less attention to certain indicators of monetary policy. In particular, interest rates and the quantity of money are not reliable indicators, in his view. He wishes that policy makers would forget about such indicators. They should instead turn instruments in order to hit the target.
For example, in recent years, he has said that all you need to know to say that money has been too tight is to look at the growth rate of NGDP. It dropped way below trend, which tells you that monetary policy should have been looser. If you insist on having an indicator, you should use the forecast for NGDP. But even if you respond only to NGDP after it is reported, you should have had a looser policy.
In the 1930s, we did not have a lot of the data that we have today, including NGDP. Sumner regards the Wholesale Price Index as the best target variable available. As I understand it (Scott, if you read this, please correct me), he thinks that the instrument that mattered most at the time was the ratio of gold reserves to currency. When this is high, government is hoarding gold and tightening monetary policy. When this is low, government is dis-hoarding gold and loosening monetary policy.
The private sector also can hoard gold, and this has the same effect as a monetary tightening. If I understand Scott’s thinking correctly, when the private sector does more hoarding, if the central bank wants to hit its nominal target it will have to do some offsetting dis-hoarding.
My own view is that the connection between instruments and targets is very loose. In the current environment, think of the Fed’s instrument as M0, which is currency plus reserves. Think of the money used for transactions as Mt, which is some complex (and variable) weighted average of currency, checking accounts, money market funds, credit lines, frequent-flyer miles, you-name-it. Because these two definitions of money are so different, the Fed can turn its dial a long way without any result, and then when it starts to get results they could end up all over the map.
This is also my instinct for the 1930s, but to be fair I need to read through Sumner’s book before I make up my mind.
I do not understand why your worldview does not predict hyperinflations and hyperdeflations to happen all the time.
Why?
I believe that prices are inherently predictable unless something dramatic happens to destabilize them. People use prices as information. For relative price movements to convey information, we cannot have volatility in prices in general. So people expect prices in general to be stable and predictable. Those expectations are almost always self-fulfilling. One of the few ways to escape that stability is for the government to run deficits that go beyond its capacity to fund by borrowing, so that it starts printing money at exponentially increasing rates. Otherwise, my baseline prediction is quite the opposite of frequent hyperinflations or hyperdeflations.
Once a little bit of inflationary or deflationary expectations set in at random, what prevents them from self-amplifying?
Why? There are times where they do, deflation in The Depression perhaps, and inflation expectations in the 70s, perhaps, but what implies that would be the rule rather than exceptions of self-regulation?
If I understand Arnold correctly, he believes that in normal circumstances the price level is mostly independent of monetary policy. That prices are at the level they are at, not because anything is regulating them to be at that level, but as a naked equilibrium. They are what they are because that’s what people expect them to be, and people expect them to be that way because that’s the way they’ve been. What I’m asking is why would you expect that equilibrium to be stable?
Just because something is path-dependent does not mean that it is unstable. The engineering-type economist wants to be able to compute the equilibrium price level mathematically, preferably as a function of the money supply. I am saying that the price level could be X if last month it was X, but it could be 10% higher than X if last month it was also 10% higher than X, with no difference in the money supply or any other variables. That is not unstable, but it is path-dependent
the thing I understand least about Sumner is which instruments he is proposing adjusting and how. if the discount rate is near zero, which lever (instrument) do we need to turn in order to hit these targets? what does the fed do to increase the supply? it seems like many of the things done simply increase asset prices, not NGDP.
I think loose-tight money are important, but poorly defined.
One key monetary mechanism is thru loans, so there are two key numbers:
Amount loaned out, and amount asked for.
So if $200 bln is asked for but only $100 bln is loaned (base 1/2), that is “tighter” than if $150 bln is loaned (base 3/4), but “looser” than if only $80 bln is loaned (base 2/5).
However, the total amount is also important. What if $400 bln is asked for?:
a) is this equal in tightness if $200 bln is loaned ? (2xbase 1/2) I think not, I think it’s looser.
b) what about if $160 bln is loaned? (2xbase 2/5) More loans turned down (tight), but more total loans (loose). I think this is actually a bit looser
And I think the above holds even with different interest rates, where interest rates have an indirect effect on the loans given out. Lower rates usually go with more asked & much more loaned. The ZLB talk is because there are too few total loans going out.
Do you know of any loan based definitions for loose-tight money?
Failure to be clear about what tight & loose actual practices are makes it tough to discuss other policies.