Tuesday, September 1, 2015

Intuitive Neo-Fisherism

John Cochrane's blog post, Whither Inflation, is excellent reading. Basically, John takes a mainstream New Keynesian model and shows how it has Neo-Fisherian characteristics - if the central bank raises the nominal interest rate, inflation increases. Noah Smith seems to find this interesting, but he's got some problems with it. I'm going to attempt to answer his questions.

First, Noah says:
What about the Volcker disinflation, when Fed interest rate hikes were followed by disinflation?
Volcker was Fed Chair from August 1979 to August 1987. During that period the time series for the fed funds rate and the pce (year-over-year) headline inflation rate looks like this:
The broad story in that picture is that Volcker began his 8-year term with high nominal interest rates and high inflation, and ended it with much lower nominal interest rates and much lower inflation. The scatter plot, with points joined in temporal sequence, looks like this:
In that chart, inflation and the fed funds rate more often than not are moving in the same direction. Volcker didn't do Cochrane's experiment in reverse - the fed funds rate comes down gradually during the disinflationary period - but I don't see an inconsistency between what we see in the above 2 charts and what is coming out of Cochrane's experiments with the NK model.

Second, Noah says:
...are we sure we want to think about interest rate policy as a series of interest rate pegs, each of which people believe will last forever?
Typically, in analyzing monetary policy, we want to study the operating characteristics of particular policy rules. That is, we specify the feasible set of actions a central bank can take under particular contingencies, specify a policy rule as a mapping from states of the world to actions by the central bank, and then ask how the economy functions under that rule. Cochrane's clearly interested in that, but he wants to show you a simple policy experiment so you can understand what's going on: Suppose the central bank increases the policy rate at a given point in time, permanently. What happens? This can help build your intuition so that you can handle the more difficult exercise of working out the optimal policy rule.

Finally, Noah says:
But the last reason we should be a little wary of the Neo-Fisherian idea is that it goes against our basic partial-equilibrium Marshallian idea of supply and demand.
For Noah, what's going on in the NK model is not intuitive. What does he mean? We can start by consulting an online dictionary:
Intuition: the ability to understand something immediately, without the need for conscious reasoning.
That fits closely with how an economist thinks about intuition. Over time, we accumulate knowledge of economic theory - models, basically - by working with them. For a lay person, partial equilibrium Marshallian supply and demand is not intuition. He or she has no clue about supply and demand. Sometimes Marshallian intuition works - for economists. I see a piece of economics that is unfamiliar, but I can recognize elements of supply and demand in it, and then I get the idea. But Cochrane is dealing with an issue that is dynamic, it's general equilibrium, it involves how the economic agents in the model think about the future. Why would I expect that static Marshallian intution would work? And if it doesn't seem to work, why should that make me "wary" of the idea? Looks like I have the wrong intuition, and I have some learning to do.

Here's some intuition. See if this works for you. Roughly, we can separate the short-term nominal interest rate into three components at any point in time:

R = F + L + r,

where R is the observed nominal interest rate, F is the Fisher effect (or inflation premium), L is the liquidity effect - the effect of monetary policy on the real interest rate - and r is the long-run real interest rate, which is determined by non-monetary factors (we'll neglect Tobin effects and such). Next, consider Cochrane's experiment. There is a one-time, permanent increase in the nominal interest rate. In some models we would have to worry about what actual monetary policy actions (asset swaps or settings for administered interest rates) would be required to support this market interest rate, but in the NK model Cochrane works with, that's not an issue. The central bank can set the nominal interest rate, and that will induce a dynamic path for F and L. I think most economists would have intuition for the long-run effect. In the long run, the increase in L induced by the increase in R is zero, and F increases one-for-one with R. In the long run, the Fisher effect dominates.

In the short run, L moves in the same direction as R. That's the non-neutrality of money in the sticky price NK model. Consumption declines on impact, and then grows over time, at a declining rate. If you have taken a PhD macro course (or even some undergrad macro courses ) you should have some modern macro intuition. That intuition tells you how consumption smoothing relates to asset pricing. In this instance, with the representative consumer facing growing consumption, the real interest rate will be higher than it would otherwise be (the liquidity effect) because the representative consumer would like to smooth consumption over time (by borrowing against the future), but cannot. But consumption growth is falling over time, so the liquidity effect (the increase in L) declines over time. This implies that, with R constant, the Fisher effect (the increase in F) must rise over time.

So, we have the liquidity effect - the increase in L - initially positive and converging over time to zero, and the Fisher effect - the increase in F - rising over time and converging to the increase in R. But what is the Fisher effect on impact? I think the intuition of some people would tell them this impact effect should be negative. That intuition might come from thinking about Phillips curves. Maybe this Phillips curve intuition exists because people are ignoring the empirical evidence. Maybe people have some familiarity with reduced form NK models. Typically, in those formulations - essentially what Cochrane has written down - Cochrane's equation (2) is a "Phillips curve." So, you could forgive people for using their Phillips curve intuition in an attempt to understand what is going on the standard NK model. But, as it turns out, their intuition doesn't help them in this case.

It's not clear why anyone would expect the impact effect of the increase in R on F to be negative, if he or she understands what is going on in the model. And, indeed, Cochrane shows us an example in which the impact effect on F is positive, though there are other equilibria in which the impact on F is negative - in those other equilibria, the liquidity effect has to initially be that much stronger.

What would happen in models where we actually include money? An example in which there are short-run nonneutralities - liquidity effects - is the class of segmented markets models. In these models, the first round effects of monetary policy affect only a segment of the population - those closely-connected to financial markets. Then, an open market operation affects people asymmetrically. One very simple model of this type is Alvarez, Lucas, and Weber (AER, 2000). In that model, there are two types of people, traders and non-traders. When an open market purchase by the central bank occurs, the traders receive a cash injection, which is a temporary increase in their wealth, so traders then consume more. This induces a liquidity effect, as it is the consumption of traders that determines the price of nominal bonds in this economy. Consumption is temporarily high for traders, so L is temporarily low, and this reduces R.

If we do Cochrane's policy experiment in Alvarez/Lucas/Weber (ALW), we can find an equilibrium in which the impulse responses look like Cochrane's. In response to a permanent increase in the nominal interest rate, the inflation rate increases over time, and ultimately the increase in F is equal to the increase in R. But in the ALW model, there is a path for money growth that is needed to support the permanent increase in the nominal interest rate. In ALW, the quantity theory of money holds in a very simply way - total output is constant, and the price level is proportional to the money stock. So, ALW tells us that, to support a permanently higher nominal interest rate, what is required is higher and rising money growth.

So, that's reassuring. Cochrane's results hold in a widely-used macro model, and they hold in other monetary models with liquidity effects. So, for people who know how those models work, the results should be intuitive.

16 comments:

  1. Interesting stuff. I posted a couple responses here: http://noahpinionblog.blogspot.com/2015/08/non-intuitive-neo-fisherism.html

    I hadn't thought about things in R = F + L + r terms before, so that's interesting. Of course I agree that Marshallian partial-equilibrium intuition is not the only intuition or always good intuition.

    ReplyDelete
  2. I learned something here.

    I learned that you can connect the dots in a Fred dot-plot with lines.

    Thank you.

    ReplyDelete
  3. Volcker entered his term during an oil shock that demanded large resets in relative prices and wages. If the reset, is upward (which is desirable to limit recessions and high unemployment) then higher than average inflation is necessary. If not allowed to spiral out of control, inflation will subside naturally once the reset is complete. Volcker left his term after Carter energy policy (fiscal and regulatory) was fully implemented and business and the economy had largely adjusted to energy by conservation and energy switching.

    Why did inflation not return when unemployment rates returned to low levels in the 90s? No major economic shocks demanded large relative resetting of wages and prices. The oil shock inflation had been addressed by Carter's energy policy. Many people cannot admit that Carter's policies were correct and they fixed the majority of the problem. Some economists believe that only monetary policy is important or only monetary policy should be used to guide an economy so dismiss out of hand very important regulatory and fiscal policy and very important interactions with monetary policy. So Volcker as the monetary avenger is credited with the fix and Carter who addressed the underlying problem, dismissed. This rewriting of history prevents people from learning the lessons of past successes that may be politically inconvenient today.

    By raising interest rates, Volcker made financing of energy conservation and energy switching much more expensive and possibly slowed the implementation that addressed the underlying cause of inflation: the oil shock. Volcker caused much misery and unnecessary unemployment with his misguided over the top policy. Of course he is praised by the wealthy elite special interests and their minions for diminishing the power of labor and setting the table for corporate execs to grab more of the economic pie at the expense of labor that has seen stagnant wages.
    -jonny bake

    ReplyDelete
  4. Every time the Fed started QE (more money printing), interest rates rose and vice versa. Is not that evidence that we are in the Neo Fisherian world?

    QE-1 was formally adopted in March 2009, when the U.S. T-Bond yield was 2.53%, but by the end of QE-1, in March 2010, the yield had moved up to 3.83%, for a rise of 1.3% points. When the Fed launched QE-2 in November 2010 the T-Bond yield was 2.62%, but 3.16% when QE-2 was ended in June 2011 – a rise of 0.5% points. QE-3 began at the end of 2012, when the T-Bond yield was 1.76%, by the end of QE, it was 2.4%.

    This same dynamic happens every time that the Fed cuts rate in a deflationary environment. It is not conclusive but it is evidence.

    ReplyDelete
    Replies
    1. QE is something else altogether. If it works at all (in terms of ultimate effects on output and inflation), I don't think it works like conventional monetary policy - but that's another story.

      Delete
    2. I am just talking about QE's effect on interest rates (not the real economy). It seems to effect expected inflation the same way. Actual inflation is another matter altogether.

      Delete
  5. You have a reply here, in case you didn't know:
    https://thefaintofheart.wordpress.com/2015/09/01/steve-williamson-should-change-the-name-of-his-blog-from-new-monetarist-to-new-fisherite/

    ReplyDelete
    Replies
    1. "New monetarism" is just a name, like Tom. There's no specific message in the name.

      Delete
  6. "It's not clear why anyone would expect the impact effect of the increase in R on F to be negative, if he or she understands what is going on in the model."

    Certainly, what the effect is in Cochrane's two equation model should be pretty clear. But I think the intuitive resistance to the idea that the impact effect on F is positive comes when we think about translating this effect to real economies.

    For my own part, this has nothing to do with the Philips curve, but rather things like the widespread existence of term debt. Most notable is that an expected permanent increase in rates would cause an immediate reduction in the real value of holdings of public debt. It's not hard to add something like this into Cochrane's model and get a negative impact effect on F.

    ReplyDelete
    Replies
    1. Sure, if you want to talk about a different model, we could explore that. But I disagree with this:

      " It's not hard to add something like this into Cochrane's model and get a negative impact effect on F. "

      The baseline NK model has complete markets, so there's no role for debt, and we can't talk about the prices of long-maturity bonds going down when short rates go up, and that having any significance. Potentially what you're talking about is interesting, but the NK model is not the place to start in thinking about the issue.

      Delete
    2. I had in mind something like the valuation formula for government debt Cochrane has used himself in his "Monetary Policy with Interest on Reserves". So we add in an equation saying that B / p is the present value of future real surpluses, with B as the nominal value of government debt and p as the price level and real surpluses as the exogenous policy instrument. This pegs the price level, so it eliminates the multiple equilibria problem in Cochrane's two equation model. Which path we get then depends on how B responds to a change in expected future interest rates (which depends on the term of the debt). If it responds enough, we get one of the paths that involves an initial deflation.

      This seems to me to be a natural extension of Cochrane's model, particularly as a way of dealing with the multiple equilibria issue.

      Delete
    3. I had thought, though this is not entirely clear to me from the NK literature, that something like that is always in the model. That is, the policy rule can give you equilibria in terms of paths for the inflation rate, but you need something to pin down the price level,and B/P = present value of surpluses could give you that.

      Delete
  7. Hi Stephen,
    I'm trying to get to grips with the Neo-Fisher stuff as something of a layperson. Hoping you can help. So far I can see the evidence for it in some of the other pieces showing rates up inflation up and more recently and importantly rates down and inflation down. I don't understand the mechanism or flows in the economy that would make it happen though. With the standard Phillips curve approach, I get it - low rates increase demand, reduce project hurdle rates, hiring cycle, lower mortgages, falling unemployment and then tighter labour market, higher wages, higher prices etc. I can understand the view that we have not seen the inflation models predicted. To my (simple) mind, is this just a function of time - we are starting to see wages and inflation pick up now, the models I don't think predict a relationship that is instant or expressly direct in nature, more a loose framework. When looked at that way they are starting to to work? Also, can any inflation model survive the huge commodity price falls,and USD rise weve seen over the last couple of years when one woul'd have predicted via the Phillips curve inflation to have picked up. Absent those, and in the domestic sectors like rents and services, inflation doesn't look to be too far away from the more usual approaches. It seems like a lot of the case is made looking at other countries - Sweden, Switzerland as evidence the current standard approach isn't working, but these to me are evidence that the world is interconnected, and the exogenous factors keeping inflation down are impacting them as well - China over capacity and oil prices. The evidence presented looks compelling on the face of it, but I don't understand the transmission mechanism for higher rates to lead to higher inflation now? The only way I can see it happening is through a reverse of the idea that over time the return on capital drops to the cost of capital, and that hiking rates would lead to higher interest costs getting passed on. But that seems like a bit of a stretch overall? Apologies if I'm missing something obvious here?

    ReplyDelete
    Replies
    1. "With the standard Phillips curve approach, I get it..."

      The problem is that that story actually doesn't make sense, though your reaction is a standard one - it seems to be intuitive. A tight labor market should make wages higher in units of goods (the real wage is higher), but it doesn't mean that prices or wages should be increasing at a higher rate in terms of money. And it doesn't work empirically, for example during the last few years inflation was falling and unemployment was falling.

      "I don't understand the transmission mechanism for higher rates to lead to higher inflation now..."

      Higher nominal interest rate makes interest-bearing assets more attractive relative to cash. So people start to dump cash and its price goes down, i.e. the price of goods in terms of cash goes up - inflation.

      Delete
    2. Thanks very much for your reply. I still dont follow the logic. Couple of clarifying points if I may:
      "And it doesn't work empirically, for example during the last few years inflation was falling and unemployment was falling."

      This was the point i was getting at - arguably it hasnt worked the past few years for exogenous reasons, Europe, China, oil, strong USD and the fact that we haven't been at NAIRU that long - at least in the US when we got there or thereabouts wages have started to pick up. At least for me its too early to say its not working. I looked at the original Phillips curve scatterplot, its not a super tight relationship that would lead you to think of a mechanical in level and time reaction function, particularly given the factors above.

      "Higher nominal interest rate makes interest-bearing assets more attractive relative to cash. So people start to dump cash and its price goes down, i.e. the price of goods in terms of cash goes up - inflation."

      Apologies - this part i just don't follow. If you have higher nominal rates, i can see interest bearing assets are more attractive - wouldnt this lead to higher savings rates? if people "dump cash" but are still buy interest bearing assets, i dont see how that leads to inflation - to me it sounds like it take money out of the economy, less demand for goods and services. In this world - how are the prices of things getting bid up, or company pricing power moving up? It seems like an identity rather than a practical model.

      Delete