Scott Sumner  

Never debate the impact of a price change

The Cartoon Introduction to... Quotes From a Colloquium...

There as been a lot of recent discussion about the "Neo-Fisherite" claim that higher interest rates lead to higher inflation. Noah Smith has a good summary. Unfortunately the debate has been marred by a lack of precision. What is being held constant when we talk about higher interest rates, and what is allowed to change?

Here's an analogy. What sort of impact would you expect higher oil prices to have on oil consumption? The answer depends on what causes the higher oil prices. If the increase were due to less supply (as in 1974) then you'd expect a decline in quantity demanded, and a decline in consumption. If the price increase were due to more demand (as in 2007) then you'd expect an increase in quantity supplied and more consumption. Contrary to widespread impression, the term 'consumption' is not a synonym for demand; it's a synonym for quantity.

Now let's take interest rates. If we hold the money supply constant, an increase in interest rates will usually boost velocity, which will provide a one-time boost in the price level. This fact is believed to explain the "Gibson Paradox" under the classical gold standard, when higher interest rates were associated with higher price levels.

Most of the Neo-Fisherite debate revolves around interest rate changes caused by the central bank. That helps pin things down, but still leaves a lot of ambiguity in the question. Here are some plausible claims:

1. A rise in interest rates due to open market sales (tight money) will tend to reduce inflation. This action reduces the supply of the medium of account.

2. A rise in interest rates due to higher interest on reserves will tend to reduce inflation. This action raises the demand for the medium of account.

These two examples largely explain the mainstream hostility to the Neo-Fisherites. But consider these two examples:

3. Raising the inflation target from 2% to 20% might well be associated with both a higher rate of inflation and higher nominal interest rates. I said "might" because it depends on all sorts of factors, such as the maturity of the bond yield being used as "the" interest rate.

4. Here's a more interesting and more plausible example. Suppose the Bank of Japan decides to peg the yen to the dollar. Also assume that before this (surprise) action, Japanese inflation was near zero and their nominal interest rates were about 2% below US nominal interest rates. Assume the US has a credible 2% inflation target. If the BOJ switch in policy is credible, then the interest parity theorem implies that Japanese interest rates should rise to the level of US interest rates (ignoring default risk.) And of course purchasing power parity implies that Japanese inflation should (in the long run) be roughly equal to US inflation under a pegged rate (ignoring the Balassa-Samuelson effect, which is small for fully developed countries.) Thus it's quite possible that a decision by the BOJ to permanently peg the yen to the dollar could simultaneously raise inflation and raise nominal interest rates.

Just to be clear, PPP often doesn't hold very closely, so I'm not claiming this would definitely occur, just that there are mainstream economic models that plausibly generate a seemingly "strange" neo-Fisherite result.

So why have I been somewhat hostile to the neo-Fisherite claim? Partly because the actual claim being made seems maddeningly vague to me, and partly because the way the claim is usually presented seems wrong. That is, as I read the debate they seem to be claiming that higher interest rates achieved through ordinary central bank procedures (the tight money policy in #1 above) might be inflationary. If I am right that this is their claim, then I'd say the claim is simply wrong. And we know it's wrong from the way asset markets respond to surprise changes in the fed funds target. If I misunderstood the claim of the neo-Fisherites, and their claim was closer to #3 or #4 then I'm still somewhat annoyed, because in that case I don't think they've done a good job of explaining their hypothesis.

Never debate the impact of a price change. Debate (if you must) the impact of the thing that caused the price change. This has implications for much more than the neo-Fisherites. It explains why the debate over the Fed's "low interest rate policy" of 2003-04 is almost complete nonsense. People talk about interest rates as if they are a policy, whereas they are the effect of various monetary policies (and also other economic forces.) Those underlying monetary policies must be IDENTIFIED before we can have an intelligent debate. Unfortunately we (i.e. the economics profession) have not yet reached that point. We don't (as yet) have a coherent question to debate. First we need to identify the stance of monetary policy. In 2003 Ben Bernanke suggested a sensible way of doing so, but was rejected by the profession. So he gave up. We are left with essentially nothing. We have no coherent technique for identifying the stance of monetary policy. There's nothing to debate.

PS. David Glasner has a very interesting post on this debate. In the end I'm less pessimistic than David about the indeterminacy problem of fiat money (i.e. why does it have value?) I see the 2% inflation target as being somewhat credible, and I think the public (correctly) believes that if and when paper money is no longer used, the government will redeem outstanding paper money for assets at a rate roughly consistent with the 2% inflation target. It's not a hard peg like a gold standard, but it's credible enough to overcome the indeterminacy problem.

At the same time I'm sympathetic to David's claim that this issue is more complicated than many people assume, and one purpose of this post is to explain at least some of the complications. If something good comes out of this debate it might be that the profession is forced to recognize that "the emperor has no clothes," i.e. that we have no coherent way of thinking about monetary policy. Not just what is does, but what it is.

Comments and Sharing

COMMENTS (5 to date)
Philo writes:

As you write: "Unfortunately the debate [about what would be the effect on inflation of higher interest rates] has been marred by a lack of precision. What is being held constant when we talk about higher interest rates, and what is allowed to change?" The interest rate (let's assume we have implicitly specified a time-period) is not an independent variable; other variables--independent ones--must change to produce a higher rate. So there are different scenarios in which interest rates rise, leading to different effects on the interest rate.

If we could focus just on some among the independent variables, the collective specification of which would completely define the economic situation (since all the dependent variables would thereby be determined), we could unambiguously--precisely--ask what would be the effect of a particular change in *them* (holding all the other independent variables constant). I think this is what you are getting at when you write: "Never debate the impact of a price change. Debate (if you must) the impact of the thing that caused the price change." This "thing" will be the congeries of changes in independent variables that determined the price change (holding all the other independent variables constant).

The price of any good or service--including the price of waiting/abstinence (for a certain length of time), which is the interest rate (for that time-period)--is clearly a dependent variable, determined by more basic variables. But which are the *independent* variables? That seems a very deep question.

marcus nunes writes:

Scott, In the linked post I argue that NGDP relative to target, but not inflation or interest rates, appears to give a consistent indication of the stance of monetary policy.

Kevin Erdmann writes:

Scott, I forget where I saw this recently, but someone mentioned the effect of large amounts of excess reserves, which means that, with IOR, banks are creditors at the target rate instead of debtors.

So, if the Fed left IOR at .25% and targeted a 2% FFR, depending on how many securities they would have to sell to reach the target, could that be inflationary? To the extent that it was pulling cash out of the economy, it would be coming out of excess reserves. And, if the differential between IOR and FFR was high enough, banks might be incentivized to expand credit, as they would be less indifferent about holding reserves.

Scott Sumner writes:

Philo and Marcus, Good point.

Kevin, If they did that they'd first have to remove all the ERs from circulation, to drive the market fed funds rate up to the 2% target. In that case there would be no ERs for banks to loan out. Or maybe I should say "loan out" in case any MMTers are reading this. :)

Kevin Erdmann writes:


That assumes that banks are reserve-constrained. Is that debatable? If banks are capital constrained, then it seems like rates could rise even as reserves remain high. But, I wonder, even assuming that the policy would empty the excess reserves, if the end result could lead to a net expansion of bank balance sheets.

I would have expected reserves from QE to replace treasuries on bank balance sheets, but it doesn't look like this happened. Banks basically took the deposits and sent the cash to the Fed, so that the extra deposits and excess reserves seem to be sitting on top of the bank balance sheets, kind of unrelated to their operational balance sheets. And, I don't think those deposits and excess reserves really affect their capital constraints. So, I am not so sure that if the Fed started sucking up reserves, that the banks would respond by trading reserves for treasuries. Non-bank savers would remove their deposits from the banks and exchange them for treasuries. This would be deflationary, through a kind of reverse-hot-potato effect, I suppose. But, if the liquidity effect of OMO was strong enough to raise rates as that process happened, the banks would be incentivized to try to get some of those reserves into higher return assets. Wouldn't this partly come down to the size of the flow of Fed OMO and the stock of bank excess reserves, and if the stock of reserves is high enough, this would lead to more bank asset expansion? And, if the expansion of bank assets was more powerful than the decline in QE related bank deposits, could this be expansionary? I'm not 100% sure what would happen, but it does seem like some of the standard effects could get flipped on their heads.

Comments for this entry have been closed
Return to top