Monday, August 22, 2016

Danger!! Crazy Neo-Fisherians on the Loose!!

Not sure how I missed this Narayana post, but better late than never. I may not be Jacques Derrida, but a little deconstruction can be good fun. Here goes.

Opening paragraph:
Some economists argue that the Federal Reserve should take a highly unconventional approach to ending a long period of below-target inflation: Instead of keeping interest rates low to spur economic activity and push up prices, it should raise rates.
Clicking on "argue" takes you to my St. Louis Fed Regional Economist piece on neo-Fisherism. This is about as low-tech an elucidation of these ideas as I've been able to muster - it's got one equation, two figures, and 3,000 and some words. In any case, apparently I'm "some economists." John Cochrane is also well out of the Fisherian closet, and we have certainly received some sympathy from others (to whom the idea is obvious - as it should be), but neo-Fisherism is hardly a movement. As you can see, particularly in Narayana's post, there's plenty of hostile resistance.

The last sentence in the quote offers you a false choice. That choice is either a world of low interest rates, which obviously spurs economic activity and pushes up prices or the alternative: increases in interest rates which, the reader would naturally assume, would give us the opposite - less economic activity and lower inflation. The basic neo-Fisherian idea is that this is not the choice we're faced with. Let's put aside the issue of how monetary policy affects real economic activity, and focus on inflation. Neo-Fisherism says that conventional central banking wisdom is wrong. A lower nominal interest rate pushes inflation down, and no one should be surprised if an extended period of low nominal interest rates produces low inflation. Indeed, that's consistent with what we're seeing in the world right now.

Let's move on. Two paragraphs later, we have this claim:
Neo-Fisherites believe that modern economies are self-stabilizing.
I've been staring at that sentence for several minutes now, and I'm still not sure what it means, so I don't think I could "believe" it. But let's give this a try. In conventional Econ 101 macroeconomics, students are typically told that, in the short run, wages and prices are sticky, and there is a role for short-run "stabilization" policy, which corrects for the short-run inefficiencies caused by stickiness. The Econ 101 story is that, in the long run, prices and wages are perfectly flexible, and the inefficiencies go away. So, the standard story people are giving undergraduates is that modern economies are indeed self-stabilizing, but "in the long run we're all dead" as Keynes said. What's this have to do with neo-Fisherism? Nothing. Indeed, most conventional models have neo-Fisherian properties, whether those models have a role for short-term government intervention or not. In this post I worked through the neo-Fisherian characteristics of Narayana's favorite model, which is certainly not "self-stabilizing" in the short run. Bottom line: Basic neo-Fisherism is agnostic about the role for government intervention. It just says: Here's how to control inflation. You've been doing it wrong.

Narayana says that neo-Fisherism leads in the direction of "unusual policy recommendations."
Suppose, for example, the long-run equilibrium real rate is 2 percent. Neo-Fisherites would predict that if the Fed holds nominal rates at 0.5 percent for too long, people's inflation forecasts will eventually have to turn negative -- to minus 1.5 -- to get the real rate back to 2. Conversely, if the Fed raises its rate target to 4 percent and keeps it there, inflation expectations will rise to 2 percent. Because such expectations tend to be self-fulfilling, the result will be precisely the amount of inflation that the Fed is seeking to generate.
That doesn't describe "unusual policy recommendations" but is actually the prediction of a host of standard monetary models. For example, there is a class of representative agent monetary models (money in the utility function, cash in advance, for example) in which, if the subjective discount rate is .02, in a stationary environment, then the long run real interest rate is indeed 2%. In those models, it is certainly the case that a sustained nominal interest rate of 0.5%, supported by open market operations, transfers, whatever, must ultimately induce a deflation equal to -1.5%. Indeed, those models also tell us that deflation at -2% would be optimal - that's the Friedman rule. So, Narayana's thought experiment is not controversial in macroeconomics - that's the prediction of baseline monetary models. Things can get more interesting with fundamental models of money that build up a role for asset exchange from first priciples - e.g. overlapping generations models from back in the day, or Lagos-Wright constructs. New Keynesians seem to like taking the money out of models altogether, in "cashless" frameworks. NK models, and fundamental models of money typically have many equilibria, which presents some other problems. Multiple equilibria can also be a feature of cash-in-advance models. But, as I discuss in this post and this one, multiple equilibria need not be a serious problem for monetary policy, as we can design policies that give us unique equilibria - with Fisherian properties.

But, from Narayana's point of view, standard macroeconomics is not standard - it's crazy and dangerous. His claim:
Traditional economic models, by contrast, predict the opposite. If the central bank raises rates and credibly commits to keeping them high, people and businesses become less willing to borrow money to invest and spend. This undermines demand for goods and services, putting downward pressure on employment and prices. As a result, the economy can plunge into a deflationary spiral of the kind that afflicted the U.S. in the early 1930s.
What "traditional models" could he be talking about? This can't be some textbook IS/LM/Phillips curve construct, as he's discussing a dynamic process, and the textbook model is static. The only "tradition" I know of is a persistent narrative that you can find if you Google "deflationary spiral." Here's what Wikipedia says:
The Great Depression was regarded by some as a deflationary spiral. A deflationary spiral is the modern macroeconomic version of the general glut controversy of the 19th century. Another related idea is Irving Fisher's theory that excess debt can cause a continuing deflation. Whether deflationary spirals can actually occur is controversial, with its possibility being disputed by freshwater economists (including the Chicago school of economics) and Austrian School economists.
The closest thing to actual economic theory supporting the idea is Irving Fisher's debt-deflation paper from 1933. That's just another narrative - you won't find an equation or any data in Fisher's paper.

The best I can come up with in terms of a genuine theory of deflationary spirals, is what can happen in Narayana's NK model if inflation expectations are sufficiently sticky, and initial inflation expectations are sufficiently low - basically, people have to start off expecting a lot of deflation. Further, in order to support a "deflationary spiral," i.e. sustained deflation, conventional asset pricing tells us that there has to be sustained negative growth in output. But if there's a lower bound on output, which is natural in this type of environment, then the deflationary spiral isn't an equilibrium. Conclusion: Narayana has things turned around. Traditional macroeconomics gives us a long run Fisher effect. Deflationary spirals are not part of any "traditional" (i.e. serious) macroeconomic theory.

On the empirical front, the "deflationary spiral ... that afflicted the U.S. in the 1930s" looks like this:
There's a body of macroeconomic history that ascribes that deflationary episode to the workings of the gold standard. Indeed, the deflation stops at about the time the U.S. goes off the gold standard. Not sure why we're using a gold standard episode to think about how monetary policy works in the current context. In modern economies, I have no knowledge of an instance of anything consistent with Narayana's "traditional model" in which increases in nominal interest rates by the central bank cause a "deflationary spiral" (if you know of one, please let me know). But, when "deflation" enters the conversation, some people will mention Japan. Here's the CPI level for Japan for the last 20 years:
This is one of my favorite examples. We wouldn't really call that a "deflationary spiral" as the magnitude of the deflation isn't high at any time, and it's not sustained. Over 20 years, average inflation is about zero. Further, since mid-1995, the Bank of Japan's nominal policy interest rate has been close to zero, and recently the BOJ has thrown everything but the kitchen sink (except, of course, a higher policy rate) at this economy in an attempt to generate inflation at 2% per year - to no avail. Note in particular that the blip in inflation in 2014 can be attributed almost entirely to the direct effect of an increase in the consumption tax of 3 percentage points.

So, that's an instance in which a form of "traditional" macroeconomics doesn't work. That traditional macroeconomics is textbook IS-LM/Phillips curve with fixed inflation expectations. In that world, a low nominal interest rate makes output go up, and inflation goes up through a Phillips curve effect. A standard claim in world central banking circles is that a low nominal interest rate, sustained for a long enough time, will surely make inflation go up. I don't know about you, but if I want to catch a bus, and I go down to the bus stop and find someone who has been waiting for the bus for twenty years, my best guess is that sitting down in the bus shelter with that person has little chance of making a bus appear any time soon.

Next, in Narayana's post, he shows a time series plot of the fed funds rate and a breakeven rate. To be thorough, I'll include other breakeven rates, and focus on the post-2010 period, as the earlier information isn't relevant:
Narayana says:
The Fed held the nominal interest rate near zero from late 2008 until late 2015 -- a policy that, according to Neo-Fisherites, should have driven inflation expectations into negative territory. Yet they stayed roughly the same for most of that period. They started to slide downward only after the Fed began to tighten policy in May 2013 by signaling that it would pull back on the bond purchases known as quantitative easing. Also, the recent modest increase in the nominal rate has not led to a commensurate increase in inflation expectations.
"According to Neo-Fisherites?" No way! Any good neo-Fisherite understands something about low real interest rates, and what might cause them to be low. Here's me thinking about it in 2010. If there are forces pushing down the real interest rate, we'll tend to get more inflation with a low nominal interest rate than we might have expected if we were thinking the long run real rate was 2%, for example. So, by 2013, yours truly neo-Fisherite was certainly not surprised to be seeing the breakeven rates in the last chart.

But how should we interpret the movements in the breakeven rates in the chart? On one hand, breakeven rates have to be taken with a grain of salt as measures of inflation expectations. They can reflect changes in the relative liquidity premia on nominal Treasury bonds and TIPS; they're measuring breakeven rates for CPI inflation, not the Fed's preferred PCE inflation measure; when inflation falls below zero, the inflation compensation on TIPS is zero; there is risk to worry about. On the other hand, what else can we do? There are alternative market-based measures of inflation expectations, but it's not clear they are any better than what I've shown in the chart.

So, suppose we take the breakeven measures in the chart seriously. The 5-year and 10-year breakevens can be interpreted as predictions of average inflation over the next 5 years, and the next 10 years, respectively. The five year/five year forward rate can be interpreted as the average inflation rate anticipated over a five-year period that is 5 to 10 years from today. Given that the interest rate Narayana is focused on here is the overnight fed funds rate, what matters for these market inflation expectation measures is the course of monetary policy for up to the next 10 years - in principle, the structure of the Fed's policy rule over that whole period. There are plenty of other things that matter as well - world events, shocks to the economy, and how those events and shocks matter for the Fed's policy rule. Narayana seems to think that the Fed "tightened" in May 2013, but I remember that episode - the "taper tantrum" - as a prelude to a period in which the public perception of the future course of interest rate hikes was constantly being revised down. A downward path for long-term inflation expectations seems to me consistent with a neo-Fisherian view of the world, with the market putting increasing weight on the possibility that nominal interest rates and inflation will remain persistently low.

Narayana finishes off in true hyperbolic fashion by raising the twin specters of the Great Depression and Great Recession:
I, too, once believed that the horrific events of the early 1930s, when economic output fell by a quarter and prices by even more, could not recur in a modern capitalist economy like the U.S. Then 2008 happened, and we all learned where a religious belief in the self-correcting nature of markets can lead us. If we want stability, we have to choose the right policies. Raising rates in the face of low inflation is not one of them.
I hope you understand by now that I think: (i) economics is about science, not religion; (ii) neo-Fisherism has nothing to do with the "self-correcting nature of markets." Do I think that Narayana's policy prescriptions are crazy and dangerous? Absolutely not. If he's right, which I think he's not, then good for him. If he's wrong, and his policies get implemented, what harm gets done? Inflation stays low, and central banks may proceed to demonstrate, through experimentation, that unconventional policies don't do much. Or maybe we find some that actually work. Who knows? We would really be in danger if the people who think of high inflation as a cure-all figure out how to produce it. But I don't think that will happen.


  1. Don't most economists believe growth breeds inflation? Our Fed doesn't want growth, because I don't think it really wants inflation. It has to protect the collateral, and keep yields low. Our Fed thinks Kocherlakota is all wet, since low rates don't breed inflation, especially when the fed controls IOR.

    However, asset inflation still exists. House prices and rents are skyrocketing due to lack of supply, while most everything else stays stable. Take Nevada. House prices are skyrocketing but the electric bill is cheaper and so is gasoline. I wonder if Wall Street controls asset inflation and how much control does it have? Certainly IOR keeps banks from lending on new homes. That way the loans it made to its cronies in 2009, which were used to scoop up houses, will be repaid with high house prices. And the Fed just lets Wall Street destroy the millennials and future generations.

  2. You know what you should do? Take the CEE model, don't chance a single equation, then called the paper "A General Equilibrium Neo-Fisherian Model." Copy-paste some discussion you made for this blog, and you have a paper in a week.

    People either will be surprised that the model is Neo-Fisherian or may not notice you swiped every equation from CEE... I smell an AER!

    1. Funny you mention that. See Rupert and Sustek's working paper:

      It's in the spirit of what you're suggesting (though I'm not sure if you're serious). The idea is to take NK models apart and figure out how they work & they find Fisherian properties. The first journal they sent this to rejected it - referees claimed that NK people already know this. Not sure why they didn't tell the rest of us.

  3. They may have got the same referees as "Market for Lemons": one referee says "this is not true", the other says "we already know this." What I find fascinating is that the paper doubles as a social experiment conducted on economists. Thanks for the paper.

    1. Standard referee reports:

      1. It must be wrong.
      2. It's right, but it's obvious.
      3. I can't understand it. It's your fault.

  4. The important point about the chart is that you can need negative real interest rates to get a nominal interest rate of 0%. There are a number of complications with measuring break-even rates, inflation expectations, etc., and it's not clear what's happening to the real rate over this time period as you suggest.

    But, if we assume that real rates can't be negative, then we can try to back-solve for market expectations assuming a low real rate. For instance, it's Jan 2010, the market is pricing a 2% avg inflation over the next 5 yrs The nominal rate is 0%. If we assume a real rate of 2% and that the market thinks that the rate will remain at zero for the next 2.5 yrs, then they have to expect the nominal rate to go up to 8% after 2.5 yrs to get a 2% real rate. That scenario does not make sense.

    The question is, can you come up with reasonable scenarios in which the nominal rate is 0% the expected inflation is 1.5-2% and the real rate is 1-2%?

    The question is can yoou get reasonable values?

  5. "But, if we assume that real rates can't be negative..."

    Why would you do that? 5-year TIPS yield is currently negative. 10-year yield has been negative in the past. Ex post real rates of return on T-bills are currently negative. Jim Bullard is working off -1.35% for a short-term real rate:

  6. Sure, but then why bother talking about liquidity, difficulties of measuring expectations, etc.

    Just say the eqbm real rate has been -2% for the past 7 years, perfectly consistent with the breakeven rates and that Narayana's chart doesn't show anything, and be done.

    1. "why bother talking about liquidity, difficulties of measuring expectations, etc."

      We've got a chart, with something that purports to be a measure of something. Might be useful to discuss how good a measure it might be.

      "Just say the eqbm real rate has been -2% for the past 7 years,"

      Not sure why I would take that approach, though I agree that Narayana can't infer what he wants to from that chart.

    2. Sure, but then I think you have to be consistent with the interpretation of these charts. What made me think that the notion that low rates are responsible for low inflation may be true is the chart that shows a strong relationship between rates and inflation. I think it was you (but maybe Cochrane), who said, if you ask an alien who has not studied any economics to look at such a chart, they'd conclude that the Fed was successful fighting inflation with low rates. I don't think you were suggesting the alien would conclude that that chart is non-sense.

      I think something else is going on during the 0% rate time period after the crisis. I don't think liquidity, etc. would explain the market expectations.

    3. As you say, the Fisher effect is obvious from eyeballing the data. We can't measure real interest rates directly but, for example, real ex post rates of return on Treasury bills have varied a lot in the post-WWII data, and there are persistent departures from average. Real rates are low in the 70s, high in the 80s, and they fall on trend from 1980 until now. So, the Fisher effect is conventional - every monetary model has this, and I think we understand it well. There is not consensus about what is going on with real rates currently. We hear a lot of stories - savings gluts, secular stagnation, scarce safe assets. You can look in my recent published and unpublished papers and find models of the latter - typically there are collateral constraints, and to the extent they bind, that gives you a liquidity premium on government debt. Price is high and the real rate is low. Once you have that, you also have a vehicle whereby monetary policy moves the real rate, and can do it permanently. Basically, a low nominal rate implies a low real rate. Fiscal policy also moves the real rate - a non-Ricardian effect - so more government debt implies a higher real interest rate, as it relieves the shortage. There are plenty of reasons why safe assets are in short supply - residual overhang from the financial crisis and sovereign debt problems perhaps, but most important: new financial regulations that increase the demand for liquid, safe, assets, relative to supply. Also, perhaps turning Treasury securities into reserves (QE) is not such a great idea, as reserves are in a sense less liquid than Treasuries.

      On these expectations measures, that's complicated, as I explained above. First, these may not be measuring what you think they are. Second, these are markets where people are taking bets on the effects of monetary policy actions far in the future, so it's very difficult to draw connections between current monetary policy actions and movements in these market-based measures.

  7. You could alternately argue that the deflation of the 1930s stopped when Roosevelt came to power and implemented policies that reduced unemployment and repaired demand.

    Are you sure that when you look at the correlation between Fed funds rate and inflation that you're not just looking at central bank policy?

    And how would raising rates stimulate investment demand and reduce savings? Seems like you're advocating destroying demand.

    1. 1. There's a fairly large increase in government spending from 1933 to 1934, but it's hard to see how that stops the deflation in its tracks. You would have to make the case for some outsized Phillips curve effect. A standard story is from Eichengreen and Bordo, who ascribe this to the gold standard.
      2. "Are you sure that when you look at the correlation between Fed funds rate and inflation that you're not just looking at central bank policy?" Not sure what you mean. Of course this has something to do with central bank policy.
      3. "And how would raising rates stimulate investment demand and reduce savings? Seems like you're advocating destroying demand." I didn't say that raising rates couldn't have a negative effect on employment and output. But "destroying demand" is a pretty strong way to put it. If central banks were always worried that they would destroy demand, they would never raise interest rates. You can't go down if you don't go up sometime. That's what people are talking about when they discuss raising inflation targets. The idea is that, if real rates are low, it's hard to do stabilization policy if you're always bumping up against the zero lower bound. Which means you have higher average inflation - and higher average nominal interest rates. But you can't have the the higher nominal interest rates without increasing them at some point in time.

  8. Raising rates off of the lows could get banks to lend more as their profits would increase. But with the Fed controlling 2.4 trillion dollars in excess reserves, would they just not lend anyway? Would the Fed even allow the neo Fisherism if it controls so much potential lending by paying IOR?

    1. The reserves are there because the Fed swapped them for Treasury securities and mortgage-backed securities. So the question is whether it makes any difference at all whether there is $2.3 trillion in reserves outstanding, or $2.3 trillion of those other assets.

    2. The reserves, then, are sterilized, Prof. So, why can't they be loaned out, or at least a small portion of them? Isn't the money supply for main street, for the real economy, not the casino derivative economy, being destroyed

    3. Thinking of banks as "loaning out reserves" is an unfortunate relic which you can still find in money and banking textbooks. It doesn't make any sense.

    4. I understand that banks can lend out up to 10 times the value of the reserves, by creating loans. But is there some rule that makes these reserves, that are exchanged for collateral, off limits in lending? Aren't they like a stimulus, where collateral is exchanged for cash? And couldn't some of that stimulus go into the real economy or is it forbidden?

    5. Banks issue liabilities and acquire assets.

    6. Thank you, Prof, for clearing that up for me. But still, the banks are sitting on 2.4 trillion or so of reserves that are in excess, that are, as you say, assets. The banks could lend 10 to 1 off those assets, injecting as much as 24 trillion dollars into the economy? I am wondering why they don't lend against those assets just a little bit? Couldn't they make more money than by simply collecting their welfare IOR?

      And the economy is really slow, so raising rates is just a raise in the banks' welfare checks. Doesn't general prosperity raise inflation so that we don't have to rely on the Fed raising rates in order to kindle inflation?

      Seems like economics is being turned on its head. This I know, the money stock, M2, is high relative to GDP which is low. That means the money supply is not trickling down to the real economy.

  9. Noah SmithAugust 1, 2016 at 2:10 PM

    Another great post! Once you stopped blogging about Paul Krugman you became the best macro-blogger in the business...hehe :-)

    That comment is so true, great post again!!

    By the way you were also the best macro-blogger when you commented on Krugman...

  10. "The closest thing to actual economic theory supporting the idea is Irving Fisher's debt-deflation paper from 1933. That's just another narrative - you won't find an equation or any data in Fisher's paper."

    Not the only but the most glaring factual error. Kinda funny that self-proclaimed Fisherites have not even read the most important paper of the economist they name themselves after.

    1. Where's the factual error? Enlighten me.

    2. Maybe he means: (100%-20%)X(100%+75%)=140% on page 346. It is an equation... technically.

    3. There is actually data in it too. See the charts at the end. The paper is basically narrative though.

  11. Stephen, I am not a Neo-fisherian anymore. I have a better model than the Phillip's curve now for inflation. The model plots core inflation on y-axis. On the x-axis is corp profits/GDP (CPATAX/GDP in Fred) minus (56%Fed rate + 44%10-year treasury rate). When the difference is highly positive, inflation is low. As the difference goes negative, inflation rises.
    The model tells me that a higher Fed rate would reduce the difference implying a higher inflation rate. However, the data points tend not to rise out of range that has existed since at least 1958. The data points stay in that range due to competitive and opportunity cost pressures. So the Fed rate at the present would have to rise to over 2% to just start allowing inflation to rise within that range. You can see an image of what I am talking about here...

    The rise of core inflation would be slow for a while if it stayed at the upper boundary. But then it may also fall away from the past upper limit as the data points move left in the model.
    So I do not agree with the Fisher effect any longer. My model shows how it might seem to work, but the model really shows that it is not a Fisher effect moving inflation at all. It is an effect of profit rates over the cost of money within a business cycle context.

    I hope you do not push the Fisher effect for much longer.

  12. The point about the bus is devastating. Good one.

    The idea of a late bus is some sort of metaphor used to motivate discussion a particular probability distribution, if I recall. No matter how long you have been waiting, the SINGLE most likely time of the arrival is right now, unless the bus ain't coming. You're saying the bus ain't coming.

    But back to subject, I think it would be fun if you could make your thesis symmetrical, by explaining how Neo-Fisherianism fit into the Volcker disinflation, which was presumably a desired result. That would be like Thomas sticking his index finger through Jesus' palm or the light bending around the sun as Einstein said it would or had to.

    Can you do it, sir?!

    1. 1. The bus metaphor story you mention sounds like something you would use in a probability theory class when you're teaching Poisson distributions. This is more a case where you don't know the probability distribution, but you have a prior and update according to Bayes rule. After 20 years you have long since stopped waiting and gone home.

      2. I've thought about the Volcker period. A key point is that Volcker's FOMC was thinking in monetarist terms. People will sometimes say that Volcker increased interest rates to reduce inflation, but I think that's the wrong characterization. Volcker actually reduced the money growth rate in order to reduce inflation. There are models in which there are liquidity effects and Fisher effects, in which a once-and-for-all reduction in the money growth rate will produce a path for the nominal interest rate which is first increasing and then decreasing, with the Fisher effect dominating in the long run. But in those models, a reduction in the nominal interest rate reduces the inflation rate.

    2. Thanks, that rings a bell. Poisson says the bus is due NOW. Bayes says think again.

      My memory could be shot, but I think I recall reading in a book called something like Economic Policy in the 1980s that Volcker did in fact follow monetarism and was not just using the targets as an excuse to crank up rates. That is his own account, to your point, I guess.

      You must be super-frustrated by what must seem like a bunch of unknown unknowns in monetary policy circles. Following Jackson Hole must be unbearable. ECB and BOJ: Mark our words, we will do whatever it takes so long as it has not shot at working.

      Plus, Eddie has figured it out and has told you never mind. That must really sting.

  13. Why not -- more tax cuts / tax rebates to tax paying individuals? Done by the Treasury literally printing more money, so the national debt doesn't increase. Not quite a helicopter drop, but similar and better -- all such printed money stimulus goes to tax payers who are the producers. And they only get back what the gov't was taking in taxes, so it's not quite free money, altho similar.

    This BIG tax rebate/ tax refund strategy seemed to work for Bush. It would most probably work again better than any other policy (given the continued regulation induced housing shortages in growing cities).

    1. "Done by the Treasury literally printing more money, so the national debt doesn't increase..."

      This would be the same as what happens when the Treasury issues debt and the Fed buys it. So, actually, the national debt does increase.