Tuesday, July 14, 2015

Garcia Schmidt and Woodford on neo-Fisherian economcs

Mariana Garcia Schmidt and Mike Woodford are lighting up the internet with a presentation on neo-Fisherian economics -- the proposition that, when we are satiated in money as at the zero bound or with interest on reserves, raising interest rates raises inflation. Noah Smith, Marginal Revolution, Brad DeLong, and indirectly at Mark Thoma's econbrowser.

This is a particularly important voice, as it seemed to me that standard New-Keynesian models produce the new-Fisherian result. i = r + Epi is a steady state in all models. In old-Keynesian models, it was an unstable steady state, so an interest rate peg leads to explosive inflation or deflation. But in new-Keynesian models, an interest rate peg is the stable/indeterminate case. There are too many equilibria, but if you raise interest rates, inflation always ends up rising to meet the higher interest rate.

What I can glean from the slides is that Garcia Schmidt and Woodford agree: Yes, this is what happens in rational expectations or perfect foresight versions of the new-Keynesian model. But if you add learning mechanisms, it goes away.

My first reaction is relief -- if Woodford says it is a prediction of the standard perfect foresight / rational expectations version, that means I didn't screw up somewhere. And if one has to resort to learning and non-rational expectations to get rid of a result, the battle is half won.

But that's only preliminary relief. Schmidt and Woodford promise a paper soon, which will undoubtedly be well crafted and challenging.

For more on the issue, here is a a previous blog post. Section 3.1 ff of "Monetary policy with interest on reserves" has a full new Keynesian model with the Fisherian result. And a wry prediction: the Fed will raise rates to head off inflation, that will cause the inflation, then the Fed will congratulate itself on having headed off the inflation.  I also suspect that models with restricted liquidity (no interest on reserves) do give a temporary decline in inflation, but without that liquidity we now will get full Fisherian results. But that's just a conjecture so far.  My last foray into learning in new-Keynesian models, which didn't end well.

Why post now? Garcia Schmidt and Woodford clearly will have a thoughtful and sophisticated paper, on what I think is a deep and important point. I hope to encourage others to read and help to digest the paper.


28 comments:

  1. Nostradamus,

    "And a wry prediction: the Fed will raise rates to head off inflation, that will cause the inflation, then the Fed will congratulate itself on having headed off the inflation."

    Before patting yourself on the back,

    First look at the fiscal response to higher interest rates - does government spending (including interest payments) increase, decrease, or stay the same? Does government curtail other spending in response to increased interest costs? Does government convert it's short term coupon paying debt to long term zero coupon debt?

    Second, look at the credit market response to higher nominal interest rates? Does total credit (public + private) expand or contract in response to higher nominal rates?

    Third, look at the currency response to higher nominal rates. Does the dollar gain ground based upon interest rate differentials between the U. S. and other nations?

    Finally, does the central bank sell off it's holdings of government debt before trying to raise the discount window rate?

    ReplyDelete
  2. John,

    "And a wry prediction: the Fed will raise rates to head off inflation, that will cause the inflation, then the Fed will congratulate itself on having headed off the inflation."

    Speaking of Irving Fisher:

    https://en.wikipedia.org/wiki/Wall_Street_Crash_of_1929

    Shortly before the crash, economist Irving Fisher famously proclaimed, "Stock prices have reached what looks like a permanently high plateau."

    Beware of economists making predictions.

    ReplyDelete
  3. You once wrote in Fiscal Fallacies:

    "However, our Federal Reserve can create as much more money as anyone might desire and more. There is about $10 trillion of Treasury debt still outstanding. The Fed can buy it."

    This must be a naïve question, but in a Neo-Fisherian world can the Fed just buy up all the assets in the world and not create inflation?

    I've always been confused about the centrality of the interest rate in the NK model. The Fed prints money and buys bonds. QE is the Fed printing money and buying different bonds. So, I have trouble thinking of changing the Fed Funds rate as being so different than QE. But presumably, buying up all the assets in the world would still being inflationary, so where did I go wrong?

    ReplyDelete
    Replies
    1. "So, I have trouble thinking of changing the Fed Funds rate as being so different than QE."

      The central bank sets two interest rates (usually close to each other). One is the fed funds rate. They do this by buying and selling short term treasuries. The other is the discount window rate. This rate is set administratively (no open market operations needed).

      A private bank with excess overnight funds can do one of three things with those funds:
      1. Buy short term U. S. Treasuries
      2. Lend those funds to another private bank, usually at a rate close to the short term U. S. Treasury rate
      3. Lend those excess funds back to the central bank

      A private bank with a dearth of funds can also do one of three things
      1. Sell any short term U. S. Treasuries that it is holding
      2. Borrow from another private bank, usually at a rate close to the short term U. S. Treasury interest rate
      3. Borrow funds from the central bank at the discount window rate

      QE isn't really a rate tool, it is a quantity tool.

      "But presumably, buying up all the assets in the world would still being inflationary, so where did I go wrong?"

      You need to read the fine print - assuming fiscal stability (meaning government is not borrowing to fund expenditures) the central bank is legally limited in what it can buy.

      Also, you need to remember that a lot of government debt is "locked up" in deferred compensation (retirement / pension / insurance plans). And so even if the Fed bought back all that debt, the end recipient would be in no position to spend the proceeds.

      Hope this is helpful.

      Delete
    2. Frank,

      I don't think you've gotten any closer to the question, which is, in a Neo-Fisherian model (for instance the NK model), can the central bank buy assets ad nauseum without it creating inflation? If so, is the model wrong or do we think that is actually true? Distinctions between discount/fed funds is superfluous to this question, in fact the distinction doesn't exist in the NK model at all.

      I don't know how in a model with money and assets you could get away with a Neo-Fisherian world, without the Fed being able to buy all the worlds assets, but I'd be curious to get John's take, because I know comparatively very little about these models.

      "the central bank is legally limited in what it can buy"

      This seems pretty unimportant to me. There is $13 trillion outstanding public debt. That seems like plenty. If the Fed can buy all that without creating any inflation, that seems like a good deal. I'd enjoy not having to pay off the boomer debt. If $13 trillion isn't enough, the Fed could move on to gold, foreign currencies, MBS, there's no shortage of assets the Fed can legally hold, and besides, it just takes passing a bill and the Fed could buy stocks/bonds, anything and become the worlds largest sovereign wealth fund.

      Delete
    3. Charlie,

      "If the Fed can buy all that without creating any inflation, that seems like a good deal. I'd enjoy not having to pay off the boomer debt."

      You are assuming that all asset owners would be willing to sell those assets to the Fed.

      "I don't think you've gotten any closer to the question, which is, in a Neo-Fisherian model (for instance the NK model), can the central bank buy assets ad nauseum without it creating inflation?"

      Since we are dealing with "just passing a bill" then yes the central bank could (with legal enforcement in a passed bill) require you to sell your house to them for $1.00. What would happen to the price of houses if it did that to you, your friends, your neighbors, et al?

      If holders of assets must sell at the request of the central bank, then presumably they must accept whatever price the central bank offers even if that price is below what the holder paid for it. So yes, the central bank can buy assets ad nauseum and push prices down generating deflation, not inflation.

      Delete
    4. "You are assuming that all asset owners would be willing to sell those assets to the Fed."

      Sure, at some price, of course. You think the 10 year treasury is going to yield -10%?

      "What would happen to the price of houses if it did that to you, your friends, your neighbors, et al?"

      The analysis is wrong. Houses would fall in value (possibly to zero), but the same amount of cash is in the economy. That cash starts chasing other goods, driving the prices up. So house values plummet, but all the other prices rise. So no, it wouldn't cause deflation, and in fact, the experiment has been run many times.

      Delete
    5. Charlie,

      "That cash starts chasing other goods, driving the prices up."

      Unless the Fed is also buying those other goods at a discount as well. You take your dollar from the sale of your house and buy a dollar can of soda. Federal reserve forces you to accept $0.10 for your can of soda. You take your $0.10 and buy a pencil. Fed forces you to accept $0.01 for your pencil. This can go on ad infinitum.

      "Houses would fall in value (possibly to zero), but the same amount of cash is in the economy."

      "Sure, at some price, of course. You think the 10 year treasury is going to yield -10%?"

      Perhaps not at any price. If I know that the central bank wants it's liabilities (dollars) to have no value and I know that the federal government wants it's liabilities (Treasury bonds) to hold their value - why would I trade bonds for dollars?

      Delete
  4. What I can glean from the slides is that Schmidt and Woodford agree: Yes, this is what happens in rational expectations or perfect foresight versions of the new-Keynesian model.

    Actually, I'm not sure that this is what Woodford is saying. It's hard to tell from looking at his slides, but it looks like he's saying that if we restrict ourselves to stable paths, the Neo-Fisherian result holds in a rational expectations equilibrium. But the indeterminacy in rational-expectations New Keynesian models might also allow for explosive paths, of the kind that you call "old-Keynesian". In fact, I'm fairly certain this is the case, especially since the "rational bubble" literature shows that explosive paths for inflation are a pretty general result in rational expectations models.

    ReplyDelete
  5. The key assumption that Schmidt and Woodford make is a Calvo-*Yun* Phillips Curve with inflation inertia. The Old Keynesians (implicitly) assumed inflation inertia (or expected inflation inertia). But the standard New Keynesian Calvo Phillips curve does not have inflation inertia (it only has price level inertia).

    Their "learning" model isn't really learning in the normal sense. It is not a process that takes place over time, as new information comes in. It's like a Walrasian tatonnement, that happens outside of time. The auctioneer calls out an inflation rate at random, people expect that inflation rate, the auctioneer solves for what the actual inflation rate would be given that expected inflation rate, and calls out a new inflation rate, and people revise their expectations accordingly, and so on. If it converges, we have tatonnement stability.

    What S&W show (it's sort of obvious really) is that pegging the nominal interest rate is incompatible with tatonnement stability. If expected inflation is above the PFE value, actual inflation would be above expected inflation, which means expected inflation would be higher still in the next round of the tatonnement process, and so on,

    I interpret S$W as confirming the old skool Keynesian/Monetarist intuition, against the Neo-Fisherian view.

    What I like about their model is they make explicit the assumption of inflation inertia. (But what they don't do, dammit, is make explicit the importance of that assumption).

    What I am less happy about is their use of tatonnement stability. It's like solving an equation by guessing an answer, plugging in your guess to solve for a new guess, then repeating, and seeing if your guesses converge. X = 0.5X does converge to X=0. X=2X does not converge to X=0. But since we have no idea about how people actually solve for the Nash equilibrium, who am I to complain.

    (But reading S&W makes me feel better about my old post, where I said that inflation inertia was the key to the standard (non Neo-Fisherian) result.)

    ReplyDelete
    Replies
    1. On second thoughts, I think I am wrong about the inflation inertia bit, but right about the tatonnement bit. Because that tatonnement process happens infinitely quickly, in meta-time, they don't need inflation inertia.

      Delete
  6. It's García-Schmidt and Woodford, not Schmidt. García is not a middle name.

    ReplyDelete
    Replies
    1. Thanks. But no hyphen on the slides nor on her webpage. I hope this is right, and my apologies.

      Delete
  7. Both New Keynesians and Neo Fisherians are wrong. The former because there is no money in their models and the latter because they confuse a definition/identity with a behavioural relationship.
    Furthermore both approaches ignore the output markets. Monetary policy is only inflationary if demand actually catches up with supply.

    Now if we include asset price inflation it becomes more complex but "ordinary" inflation is a fairly simple thing.

    ReplyDelete
    Replies
    1. Anonymous,

      New Keynesians are not wrong because of a lack of money in their model. We can conceive of a credit barter economy. New Keynesians models are wrong because in a credit barter economy there is an inordinate number of separate interest rates instead of a single interest rate.

      Delete
  8. Cochrane's argument seems entirely based on a particular kind of model, with not one iota of information or common sense received from the real world.

    It's a simple thing to observe that markets and forecasters become more optimistic about growth and pessimistic about inflation after a loosening, and more pessimistic about growth and optimistic about inflation after a tightening, It's a common-sense deduction that markets care much more about central bank policy than they do about central bank assessments of future conditions.

    Of course, if the central bank were to increase policy rates to 4%, we should expect inflation also to increase. That's because the central bank tightens in response to inflation, not because inflation rises in response to tightening.

    This is really just very, very silly, and it's wrong to blame it on Fisher, who never believed any such thing and doesn't deserve the insult.

    ReplyDelete
    Replies
    1. Tom,

      See: https://en.wikipedia.org/wiki/Fisher_hypothesis

      "In economics, the Fisher hypothesis (sometimes called the Fisher effect) is the proposition by Irving Fisher that the real interest rate is independent of monetary measures, specifically the nominal interest rate and the expected inflation rate."

      "If the real rate r is assumed, as per the Fisher hypothesis, to be constant, the nominal rate R must change point-for-point when pi rises or falls."

      Delete
    2. In this post Tom Warner makes unjustified assertions about causality from data.

      Delete
  9. "And a wry prediction:" ... does anything get falsified if this doesn't happen?

    ReplyDelete
  10. Well, if the neo-Fisherites are correct, the Fed should move to negative interest rates which result in 0 percent inflation. I guess the Fed should use QE to pay off the entire national debt, which would be a boon for taxpayers. It cannot result in inflation as the Fed would be paying negative interest.

    ReplyDelete
  11. "And if one has to resort to learning and non-rational expectations to get rid of a result, the battle is half won."

    If rational expectations is a good model, why are there still mosques, churches, synagogs, temples, covens and "alternative" medicines?

    ReplyDelete
    Replies
    1. Wouldn't you say that disproves learning even more?

      Delete
    2. Well... people do learn a lot of irrational and silly things, including to expect things they shouldn't (and to be 100% confident in them as well)... but I'm probably missing your point because that's often the case with me.

      Delete
  12. "..... when we are satiated in money as at the zero bound or with interest on reserves, raising interest rates raises inflation......" ???

    "...... happens in rational expectations or perfect foresight ............ " ???

    Raising interest rates is actually very deflationary

    ReplyDelete
  13. With the collapse in Velocity of Circulation, I wonder if either camp can lay claim to a systematic effect of raising interest rates on anything? If money demand has gone the way of the Sony Walkman then monetary inflation is dead, right?

    Join me on twitter at #FedDoNothing

    ReplyDelete
  14. The reason that many dismiss the Neo-Fisher idea is that the mechanism by which it operates in the economy is not well understood and it's implications seem to contradict conventional thinking. The Fisher relation has a lot of empirical support, especially for the U.S., but it is just a long-run equilibrium relationship and does not provide any behavioral implications, similar to PPP.

    In a world in whch the central bank pays interest on reserves (IOR) raising the IOR increases the demand for reserves by banks but it also expands reserves in the banking system since the interest on reserves is paid in the form of more reserves. So a 5% IOR means reseves in the banking system are increasing 5% per year, cetris paribus. The question then is what happens to those increased reserves? Do banks just hold them happily collecting the IOR or do they get lent out expanding the money supply? If banks are unwilling/unable to lend at low rates of interest, why should we think they will be willing/able do so at higher rates of interest?

    What good are reserves to the bank? Yes, they can earn more reserves by holding them, but what then? We could just as easily substitute lima beans for reserves and ask the same question. Eventually some of those reserves must be distributed, even if its only through dividends to the owners of the bank. But is that going to stimulate economic activity enough to move us up the Phillips curve?

    Cochrane tells the story that as banks try to obtain reserves through deposits (to take advantage of the higher IOR) they wil start to offer higher deposit interest rates which will then lead to Treasuries being sold starting a chain reaction throughout the yield curve. But exactly how this is to stimulate inflation is unclear. There is the intertemporal substitution effect and the relationship from DSGE models that real rates are proportional to consumption growth so that higher real rates lead to higher consumption growth. But that could occur through lower current consumption just as well as higher future consumption, the former leading to lower inflation while the latter to higher inflation.

    Until the mechanism of how higher nominal interest rates cause higher inflation is fully explained, I fear the Neo-Fisher hypothesis will continue to meet resistance.

    ReplyDelete
    Replies
    1. William Crowder,

      The Fisher hypothesis:
      https://en.wikipedia.org/wiki/Fisher_hypothesis

      "is the proposition by Irving Fisher that the real interest rate is independent of monetary measures, specifically the nominal interest rate and the expected inflation rate"

      The Fisher hypothesis is a link between nominal interest rates and expected inflation, not realized inflation. It does not stipulate a causal link between nominal interest rates and realized inflation.

      You can make a case that a high realized risk free rate of return in the form of interest on reserves or other forms will raise the inflation rate (all else being equal). But that is beyond what the Fisher hypothesis proposes.

      Delete

Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.