Scott Sumner  

Nick Rowe on John Cochrane's Neo-Fisherian model

PRINT
The Spoils of War... National Origin as Nurture Eff...

Nick Rowe and I have fairly similar views on what's right and what's wrong with Neo-Fisherism. Recall that Neo-Fisherism is the view that shifting to a policy of higher nominal interest rates will lead to higher inflation, especially if the policy persists for an extended period of time.

There's a grain of truth in this claim, but at the same time the policy implications are trickier than the Neo-Fisherians seem to assume. This reflects the age-old distinction between one time level shifts in the money supply, and permanent shifts in the growth rate of the money supply. A one-time increase in M reduces interest rates, whereas an increase in the money supply growth rate increases inflation, and hence nominal interest rates. But here's the problem, every permanent increase in the growth rate of money begins with what looks like a one-time increase. And the fact that prices are sticky makes this all even harder to sort out. So how do you identify easy and tight money?

Nick Rowe has a very clever post that beautifully lays out what's going on here. He uses the money supply, so I thought it would be interesting to translate his claims into exchange rate language, which is the approach I've used in analyzing Neo-Fisherism. Here's Nick:

They key question to ask is this: when the . . . central bank announces an increase in the . . . . interest rate that it pays on . . . money (call it Rm), what does it announce about the growth rate of the stock of . . . money?
[You need to read Nick's entire post to understand why I added the ellipses.] Here's my translation into forex language:
The key question is when the Bank of Japan announces an increase in the foreign exchange value of the yen, what does it announce about changes in the expected future appreciation of the yen in forex markets?
Now let's go back to Nick's post:
If the central bank announces that Rm increases by 1%, and at the same time announces that money growth increases by 1%, then we get Neo-Fisherian results. The inflation rate increase by 1%, but the opportunity cost of holding money is unchanged (the increased Rm and increased inflation cancel out), so there is no initial jump up or down in the price level.

But if the central bank announces that Rm increases by 1%, and at the same time announces that money growth will not change, then we get an initial drop in the price level, because the opportunity cost of holding money has fallen so the demand for money has increased, but there is no subsequent change in the inflation rate.


And my translation into forex language:

If the BOJ announces that the spot exchange rate is unchanged, and at the same time announces that the expected depreciation of the yen increases by 1%, then we get Neo-Fisherian results. The inflation rate increase by 1%, but the opportunity cost of holding money is unchanged (the increased Rm and increased inflation cancel out), so there is no initial jump up or down in the price level.

But if the BOJ announces that the yen suddenly appreciates by 1%, and at the same time announces that the future appreciation of the yen does not change, then we get an initial drop in the price level, but there is no subsequent change in the inflation rate.


You might object that prices are sticky and hence PPP does not hold in the short run. But Nick's example faces the exact same complication:
If we assumed prices are sticky rather than perfectly flexible, that initial drop in the price level would take a few years of deflation to work itself out.

Nick claims that the key mistake made by both New Keynesians and Neo-Fisherians is that they look at interest rates and ignore the money supply:

It's not enough to ask what happens if the central bank changes the deposit rate of interest. We must also ask what the central bank does with the money supply. And the New Keynesians (Neo-Wicksellians) are to blame by deleting that second question, by deleting money from their model.
It's very useful to also look at the money supply, but not essential. What is essential is that you look beyond interest rates, to a monetary measure that gives an unambiguous reading on the distinction between level shifts and growth rates shifts. The money supply does this, but so do exchange rates, or the price of gold (in the 1930s).

In one sense Nick's money supply approach is superior to my forex approach. It's more general in that it applies to both closed and open economies. Obviously the exchange rate approach only applies to open economies. But there are an almost infinite number of similar ways of measuring the "price of money", which do apply to closed economy models. Thus you could replace the spot exchange rate with the one-year forward NGDP futures price, and the expected appreciation in the exchange rate with the expected depreciation over time in the one-year forward NGDP futures price. That applies equally well to a closed economy model. And I would argue that it is superior to the money supply, as it incorporates the effects of both money supply and money demand shocks.

Here's Nick on the Great Recession:

What actually happened in the Great Recession? Why wasn't there a deflationary spiral when central banks were constrained by the ZLB? The answer is simple: the price level (and real output too, if prices are sticky) did not spiral down to zero because central banks did not let the money supply spiral down to zero. In fact, they did the opposite. They did "QE" (aka Open Market Operations). Empirical puzzle solved.
I agree, but I'm not sure this will satisfy Cochrane. He might argue that changes in QE didn't seem to impact the price level, and hence that QE was ineffective. So price stability remains a mystery.

I'd reply that there was little correlation between the amount of QE and the inflation rate for standard "thermostat" reasons. If policy were effective you'd expect to see no correlation.

Cochrane might reply that this excuse is too clever, and that I'd need evidence.

I'd reply that the evidence is in the asset markets, which responded to news of QE as if it were effective.

And that final point is my principle objection to modern macro (New Keynesian and Neo-Fisherian.) There is too little interest in the response of asset prices to policy shocks, and what that tells us about the structure of the economy. These real time prices tell us a lot, if you only bother to look.




COMMENTS (9 to date)
Ted Sanders writes:

Would you mind adding a link, please? (I'm guessing you meant to but forgot.) I briefly googled from my phone but could not locate Nick's post.

Ted Sanders writes:

Here it is: http://worthwhile.typepad.com/worthwhile_canadian_initi/2016/12/john-cochrane-on-neo-fisherianism-again.html

Apologies for my poor Google-fu and commenting. Being on my phone is my excuse for both.

Scott Sumner writes:

Ted, Thanks for the reminder.

Andrew_FL writes:
the key mistake made by both New Keynesians and Neo-Fisherians is that they look at interest rates and ignore the money supply

Irving Fisher must be turning in his grave.

Will LS writes:

It's still curious how most empirical papers investigating this show inflation picking up after an increase in the fed funds rate. Part 9 from Cochrane's recent paper : https://faculty.chicagobooth.edu/john.cochrane/research/papers/fisher.pdf

Jeff writes:

@Will,

Not really curious. It's the standard thermostat effect Scott mentioned. If you have your thermostat set at 70 degrees Fahrenheit, and the outside temperature rises from 70 to 90, the indoor temperature will start to rise, the air conditioner kicks on, and the indoor temp returns to 70. A really sophisticated thermostat might even measure the outdoor temperature and calculate that it should start the AC even before the indoor temp starts rising. If you are an observer who sees only the indoor temp and whether or not the AC was running recorded once every five minutes, it may appear to you that the AC being on is associated with higher indoor temps. A statistical causality test may even tell you that the AC being on "causes" the indoor temp to be higher.

The economy is not plumbing or some other inanimate physical system. There are purposeful actors involved who react in real time to what is going on.

Will LS writes:

@Jeff,

Yes, there all types of reactions which may blur the underlying relationship between interest rates and inflation. I have yet to see anyone systematically test what Prof. Sumner is describing here by using asset prices.

The thought experiment that you have described and conventional MM wisdom may still point us to the right direction, but the empirical implications are still important for theories like NK, NeoFisherism, FTPL...

Jeff writes:

@Will,

Asset prices do not react (much) to events that were anticipated by almost everyone. The Fed's action this week was nothing if not anticipated, as most of their actions are. The few times they have surprised people, asset prices have not behaved in the way you would expect if the Neo-Fisherians were right. Scott has pointed this out a bunch of times, although I guess most of those were on his other blog, TheMoneyIllusion.

Will LS writes:

@Jeff,

I agree, hence even more reasons to be skeptical of systematic tests like VARs. You need to take into market anticipations and other developments — a bit like what Prof. Sumner did in Midas.

Comments for this entry have been closed
Return to top