Saturday, January 16, 2016

Central Bank Obligations

The primary "liabilities" currently on the Fed's balance sheet are currency (Federal Reserve notes and coins), reserves (roughly, checking accounts of financial institutions held with the Fed), and reverse repos (ON-RRPs discussed in this post), which play an important role in the Fed's intervention to control overnight interest rates. Currency and reserves are not liabilities in the sense that private debt instruments are liabilities. A corporate bond, for example, is a promise to make a future stream of payments. Currency represents only a promise to exchange one type of note for another - the Fed will exchange new Federal Reserve notes for ones that are worn out, and it stands ready to exchange four fives for a twenty, for example. Not everyone can have a reserve account, but for those institutions that do, the Fed makes some other promises - it stands ready to exchange one unit of currency for one unit of reserves, and vice-versa. ON-RRPs are different, in that they conform to the same rules as private secured debt. In particular, in an ON-RRP transaction, the Fed posts collateral in its portfolio (Treasury securities or agency mortgage-backed securities) to secure loans that are typically made overnight (though the Fed sometimes makes use of term ON-RRPs).

But central bank liabilities are obligations in an indirect sense. That is, the Fed makes promises about the future purchasing power of its liabilities. Central bank liabilities are unique: currency is a medium of exchange which no private financial institution issues, and Fed reserves play a special role in settling large transactions among financial institutions. The Fed has determined that Fed liabilities will perform these roles well if it commits to a 2% target for the annual rate of increase in the personal consumption expenditures deflator (the PCE deflator). That's a promise - a commitment. If the Fed keeps that promise, then this should reduce uncertainty concerning the real value of payments in long-term contracts, and thus make credit markets and the markets for goods and services work more efficiently.

So, how has that been going?
We are now well into our fourth year of sub-2% inflation. Of course, oil prices that have been falling since mid-2014 share some of the responsibility, but there is some concern (see this speech by Jim Bullard) that the additional oil price decreases we have seen will further prolong very low inflation.

Added to that is concern about the public's views on future inflation - low inflation uncertainty is part of what we gain from inflation targeting. One measure of expected inflation is the "breakeven rate." A breakeven rate is the difference between the yield to maturity on a Treasury bond, and the yield to maturity on a TIPS security (an inflation adjusted Treasury bond) of the same maturity. For example, the next chart shows 5-year and 10-year breakeven rates:
Then, the five-year breakeven rate is the average rate of inflation that would have to occur over the next five years for a nominal 5-year Treasury security and a 5-year TIPS to bear the same average return. We could then take this as a measure of the average anticipated inflation rate over the next 5 years. There are three potential problems, though:

1. TIPS securities compensate investors for CPI (consumer price index) inflation. But the Fed cares about PCE inflation. The CPI is known to generate an upward-biased measure of inflation.
2. An investor in a TIPS does not make a payment to the Treasury when CPI inflation is negative - inflation compensation goes to zero whenever CPI inflation is less than zero.
3. Breakeven rates can reflect aggregate risk, which is variable over time.
4. Breakeven rates can reflect market liquidity. For example if the TIPS market is less liquid than the market in nominal Treasuries, this will bias the breakeven rate downward as an estimate of expected inflation. And market liquidity can vary over time.

Another measure of anticipated inflation is the 5-year, 5-year forward rate, which is a measure derived from bond yields of the average 5-year inflation rate anticipated 5 years from now:
There has been a flurry of concern in the last few days that all of these market-based measures of expected inflation have come down significantly in the last two years. If we believe these inflation measures, then we might be worried that the Fed's commitment to 2% inflation has become incredible - possibly market participants aren't buying it. And that would be bad. But the Atlanta Fed says we shouldn't worry. According to them, if we correct for the biases in standard market-based inflation expectations measures, inflation expectations have been quite stable.

But suppose that the Atlanta Fed is wrong. Or, suppose that the Atlanta Fed is right but, even worse, market participants are wrong and inflation is actually going to stay low, unless the Fed does something about it. But it's not as if everyone is agreed on what "doing something about it" means.

One view of inflation is that it's driven by a Phillips curve process - when output and employment are high, and unemployment is low, inflation is high. Then, the central bank controls inflation in a passive fashion, by leaning against the wind. When the economy "heats up," that produces more inflation, and the central bank cools things down by raising interest rates. When the economy "cools down" that produces less inflation, and the central bank heats things up by lowering interest rates. If this were the way the economy actually works, that would be great. Nothing subtle about it - you could get a group of 8-year-olds to run the central bank.

Here's an example where that idea didn't work:
In the early 1990s, the Bank of Japan reduced short-term interest rates - you can see in the picture that the policy rate came down and has stayed close to zero ever since. In Japan, 20 years of ZIRP (zero interest rate policy), served only to produce an inflation rate that averaged about zero. You can see the drop in inflation expectations in the early 1990s in Japan, reflected in the falling 10-year government bond yield, which stays low, and even falls further, to where it is today. In the late 1990s, Japan looked much like the U.S. today (at least with respect to the measurements in the chart). Inflation was about zero, the 10-year bond yield was at about 2%, and the central bank's policy rate was close to zero. Some people, including Naryana, think that backtracking on liftoff would be a good idea for the Fed, given low inflation, and supposing that what we're seeing in market-based inflation expectations measures is measuring what's going on. But the Japanese did that experiment - after the late 1990s they stayed with ZIRP, and inflation just went below zero.

Some have argued that, if extended ZIRP doesn't work, then there are other ways to get inflation up. Again, the Bank of Japan has been doing a great job of doing the required experiments for us. In April 2013, they embarked on a major quantitative easing project, which is reflected in Japan's monetary base as follows:
So, Japan is on the road to tripling the size of its balance sheet since the inception of the QE program in April 2013. What's happened to the CPI?
There was a burst of inflation in 2014, which appears to have had more to do with the direct effects of the consumption tax on the CPI, but inflation has essentially been absent since April 2014, in spite of a massive QE program. You can blame oil prices, but if QE actually works as advertised, it should have overcome that.

Another suggestion is that, if all else fails, the central bank's policy interest rate can always go negative. However, theory, and experience, tells us that doesn't work either. The most egregious case is Switzerland, which currently has an overnight interest rate of -0.7%, and an inflation rate of -1.4%.

So, this seems straightforward. If you're a central banker and want to increase inflation, don't model your behavior on a central bank that can't do it. Take as an example someone who's serious about it, and can produce results. Arthur Burns (Fed Chair, 1970-1978), and G William Miller (Fed Chair, 1978-1979) seem to have known the trade. What did they do?
Over the 1970s, the fed funds rate moved pretty closely with inflation. And it doesn't appear to be responding passively - through most of the period the fed funds rate leads inflation. Burns and Miller left a legacy of high inflation for someone else to clean up, but at least they showed us how neo-Fisherian policy works. Don't think in terms of Phillips curves - worry about the Fisher effect. Higher nominal interest rates mean higher inflation.

25 comments:

  1. This is just chartism, you need to present a plausible story as to *why* higher interest rates will increase demand (or otherwise why firms will raise prices, if inflation specifically is what you care about) - graphing two variables together isn't very informative, nor for that matter is getting odd results out of reduced form DSGE models. As economists it's all about the stories we tell, fundamentally - if there isn't a good story behind the data, or model result, I'm not interested.

    ReplyDelete
    Replies
    1. You're not paying attention. This is just a property of standard models that are widely used. I haven't found a model where this doesn't happen. So, the theory is there - that's the "plausible story." What you're calling "chartism" is basically blog empiricism. What's wrong with that?

      Delete
  2. If standard models are producing a strange mathematical result, and we can't provide in plain English a 'story' to go with it, my immediate instinct is to assume this is a weakness in our models. Don't you think that's a reasonable heuristic?

    As for 'blog empiricism', the problem of course is that there's no counter-factual. In Japan's case you can't show with that data that interest rate cuts simply weren't powerful enough to stop deflation, rather than that it caused deflation. Likewise, you can't show with that data that it wasn't simply Burns & Miller having good foresight.

    ReplyDelete
    Replies
    1. "...strange mathematical result..."
      That's the idea. It's not a strange result. Apparently you don't trust the theory, and you don't trust the data. Would you be happy if I told you that the nominal interest rate is like a furnace? A higher nominal interest rate heats the economy up - inflation goes up. And a low nominal interest rate is like turning off the furnace. The economy cools and inflation goes down.

      Delete
    2. "Would you be happy if I told you that the nominal interest rate is like a furnace? A higher nominal interest rate heats the economy up - inflation goes up. And a low nominal interest rate is like turning off the furnace. The economy cools and inflation goes down."

      Still lacking a 'why', a reason for this. 'Traditional' economics holds that higher rates discourage spending, discourage borrowing, encourage investors to hold more safe assets and less risky investments - that's a good reason one can construct linking *high* rates to a cooling economy, I can't do the same with low rates.

      Also it has been a while since I did my masters, but I do seem to recall that the impulse responses of NK DSGE models were expansionary in response to rate cuts, at least in the short term - and they generally weren't used for long term growth econ.

      Delete
    3. "Still lacking a 'why', a reason for this."

      We're trying to understand what a good monetary policy would be, in a context where "traditional economics" doesn't seem to help.

      "Traditional economics..."

      Which would be what, for you?

      "...I do seem to recall that the impulse responses of NK DSGE..."

      So you like NK DSGE? What do you make of this?

      http://faculty.chicagobooth.edu/john.cochrane/research/papers/fisher.pdf
      http://newmonetarism.blogspot.com/2016/01/overly-tight-monetary-policy.html




      Delete
    4. "Which would be what, for you?"

      Continue reading that sentence and I reveal what that is. I'm asking if you have a similar verbal description of the mechanisms that cause raising interest rates to raise inflation.

      "So you like NK DSGE?"

      Not especially, my point was that you kept presenting NK DSGE as if the norm of these models is to predict higher inflation from higher interest rates, but from what I *recall* the normal result in the impulse response was the opposite, it only seems to happen under special conditions (of course I may be remembering wrong).

      "What do you make of this?"

      I don't have time to read the full papers, do they contain a verbal description of the mechanisms describing how higher rates cause higher inflation?

      Delete
    5. "Continue reading that sentence and I reveal what that is."
      No you didn't. What's the model? You can make up any "verbal description" you want. But that wouldn't be economics, would it?

      "I don't have time to read the full papers..."

      Yes, I can see you're into vague recollections and "verbal descriptions."

      Delete
    6. But try this, and see if it works:

      http://newmonetarism.blogspot.com/2015/09/intuitive-neo-fisherism.html

      Delete
    7. "No you didn't. What's the model?"

      The model is supported be IS/MP or Mundell-Flemming, AS/AD and (as far as I was aware) most New Keynesian DSGE impulse response functions (but not in every case with the latter, but again that just suggests a weakness in the models to me, not the underlying intuition).

      "You can make up any "verbal description" you want. But that wouldn't be economics, would it?
      Yes, I can see you're into vague recollections and "verbal descriptions.""

      It's more than that. You must be aware of this kind of epistomology, I'm fairly sure it's a popular notion now that economic models are simply ways to ensure that the stories we tell as economists 'add up' and are internally consistent, and to solve for multiple potentially conflicting forces. The models themselves shouldn't take primacy - since they are all flawed and simplified. People like Paul Romer have shown that doing this can lead to fantasy worlds with variables that don't map to any kind of real thing. You say it's possible make up any 'verbal description' you want, but it's equally possible to make models with any crazy result given the right assumptions. I do believe economics requires BOTH a real concrete verbal description of your theory and a model to show it's consistent (some empirics might help too). If we discount anything that isn't a mathematical model, we'd have to discount nearly everything written by, for instance, the brilliant Ronald Coase; do you consider him a real economist? What about Adam Smith, if we go back further?

      Delete
    8. 1. IS/LM, AS/AD Mundell-Fleming doesn't have this effect: "...encourage investors to hold more safe assets and less risky investments..." At least when I was taught these things. Also, you understand, I hope, that those static models aren't the same thing as NK models? In any case, we have better models now. What do we want with IS-LM?

      2. "economic models are simply ways to ensure that the stories we tell as economists 'add up' and are internally consistent" and "The models themselves shouldn't take primacy" That would seem to be a contradiction. The model is the primary thing - it ensures that you're not just making it up.

      3. "People like Paul Romer have shown that doing this can lead to fantasy worlds with variables that don't map to any kind of real thing." Romer "showed" nothing of the kind. He was just complaining that he didn't get enough attention.

      4. Sure, we need the model, and we need to explain to everyone what it means. And formal modeling is a great way to run checks on what the old guys were trying to say.

      Delete
    9. "But try this, and see if it works:"

      I'm not necessarily asking for something 'intuitive', in the sense that it would make sense to a lay person or without 'conscious reasoning'. I do however want a set of statements describing the mechanisms of your model that can be comprehended.

      "Consumption declines on impact, and then grows over time, at a declining rate."

      First of all my understanding in these models is that C only grows if r > R. Also it's easier for me if we talk in terms two effects, the substitution and income effect. The substitution effect is contractionary, the income effect expansionary (over the longer term) - but it's generally intuitively understood that substitution effects dominate over the period of time demand side economists are interested in. What is wrong with that analysis?

      " But consumption growth is falling over time, so the liquidity effect (the increase in L) declines over time."

      Unless I'm missing something this assumes monetary policy won't adjust to counteract this. In fact you've just lost me, I don't really understand this, I didn't use that variable in the models I learned. I don't actually see why L would fall without R falling - either monetary policy intervenes to keep R as it is, and therefore increases L (or stops it from falling), or it stops intervening and both L and R fall, rather than F rise.

      You didn't provide any "intuition" why F would rise on impact, you just claim you don't understand why people think this if they 'understand the model' (it's obvious why people think this - lower consumption moves the economy down the 'supply curve', it's extremely difficult to rationalize a completely flat or downward sloping supply curve - entirely counter-intuitive even).

      "So, ALW tells us that, to support a permanently higher nominal interest rate, what is required is higher and rising money growth."

      Now we're finally getting somewhere. However, this doesn't work when you consider modern economies, where there is both base money and broad money. I don't regard the quantity theory of money, when considering only base money, as true any more (past a certain upper bound where banks are no longer reserve constrained) - so you keep adding more and more base money, but this has no impact on demand, while the higher rates are choking lending and *reducing* credit and thus broad money, which has a dominating and negative impact on demand.


      In actuality the best rational for neo-fisherism is basically fiscal theory of the price level, where governments are unwilling to reduce their deficits but higher interest rates are making them insolvent - so they are forced to deficit spend with new created money.

      Delete
    10. "IS/LM, AS/AD Mundell-Fleming doesn't have this effect"

      Sure, I forgot to mention portfolio theory and plenty of other models in financial economics.

      "In any case, we have better models now. "

      I'm sure I've mentioned to you previously that I dispute the idea that DSGE models are really that much better - they're both flawed in their own ways. For one thing it simply assumes away a strong mathematical possibility of total disequilibrium divergence that is totally possible when you build a DSGE model - this is an arbitrary assumption that we're imposing only because it goes against 'economic wisdom' - but if you really don't think this wisdom holds primacy over formal mathematical modelling, why aren't you spending a great deal of time assessing these divergent cases as equally plausible scenarios?

      "That would seem to be a contradiction"

      Nope, the model is a tool to make sure the logic of our stories hold, but it's the stories that actually contain the useful information and enriches our understanding of the economy.

      "He was just complaining that he didn't get enough attention."

      I disagree.

      "and we need to explain to everyone what it means."

      This is where I think Neo-Fisherism has failed.

      Delete
  3. What about fiscal policy? What woul happen if markets convinced themselves that the government will use the chance given by low rates to make 15% deficit each year? I believe that woul bring inflation. After all high rates in the 80 s brought down inflation rather than up. Probably fiscal backing is more important than interest rates, think of argentina and venezuela

    ReplyDelete
    Replies
    1. Sure, fiscal policy is important. But look at this:

      https://research.stlouisfed.org/fred2/graph/?g=2M36

      Paul Volcker brought inflation down by reducing the fed funds rate. Though actually he was a quantity theorist, and the way he conceptualized it, he brought inflation down by reducing money growth. Same thing, though.

      Delete
  4. "Currency and reserves are not liabilities in the sense that private debt instruments are liabilities. A corporate bond, for example, is a promise to make a future stream of payments."

    I don't follow. A 30-year corporate bond promises to make a future stream of payments. This includes interest and a final return of principle at the end of its 30-year term. If the corporate bond is a perpetual one, then it promises a stream of interest but no return of principle since there is no term. If that rate is floating, then the corporate bond promises an uncertain stream of perpetual interest payments.

    Given that central bank reserves promise to pay a floating interest forever and have no term, how are they any different from a perpetual corporate bond that pays a floating rate? Contra your first sentence, it would seem that reserves *are* liabilities in the sense that other private debt instruments are liabilities.

    ReplyDelete
    Replies
    1. Private financial instruments can have contingencies, of course, which are more sophisticated promises, than is the case with simple debt. In theory, we can think in terms of Arrow-Debreu contingent contracts, which is the most general construct. And in an Arrow-Debreu world, everyone keeps their promises, to deliver specific goods and services in specific states of the world. But, even if central bank reserves pay interest, those are just promises to give you more of an object in the future, where the object itself doesn't carry with it any explicit promises.

      Delete
    2. " But, even if central bank reserves pay interest, those are just promises to give you more of an object in the future, where the object itself doesn't carry with it any explicit promises."

      Ok, but perpetual corporate bonds pay out interest too. So like reserves, they are also just promises to give you more of an on object in the future, where that object itself doesn't carry with it any explicit promise. So once again, the line you're trying to draw between central bank reserves and perpetual corporate bonds is an arbitrary one.

      Delete
    3. No, that's the point. All of these private obligations are derived from the central bank's indirect promises. The central bank makes a promise that it will try to achieve 2% inflation. Then, private parties can negotiate contracts that are explicit promises in terms of the "liabilities" that the central bank issues.

      Delete
    4. Ah, ok - you're saying that central bank liabilities are the medium of account; people set prices in terms of the unit of account and denominate their own liabilities in that unit such that all private obligations derive their value from Fed notes/reserves.

      If people stopped using Federal Reserve notes and reserves as the unit of account, then those notes and reserves would become just another perpetual money market instrument. And if, in their place, the economy substituted in Google perpetual bonds, then Google's bonds would cease to be "liabilities in the sense that private debt instruments are liabilities."

      Delete
    5. "the economy substituted in Google perpetual bonds, then Google's bonds would cease to be "liabilities in the sense that private debt instruments are liabilities.""

      No. In early central banking, central bank liabilities were indeed explicit promises. For example, the Bank of England's notes were no different from the notes issued by other banks - they were redeemable in specie. Essentially callable debt. Over time, though, the Bank was granted a monopoly on note issue, and ultimately, when the gold standard lapsed, Bank of England notes became fiat money - with no promises attached.

      Your Google bond economy might look something like Canada, pre-1935, when all the Chartered Banks issued their own notes. But to get its notes to circulate, Google would have to make some promises as to redemption of their notes.

      Delete
  5. "Another suggestion is that, if all else fails, the central bank's policy interest rate can always go negative. However, theory, and experience, tells us that doesn't work either. The most egregious case is Switzerland, which currently has an overnight interest rate of -0.7%, and an inflation rate of -1.4%."

    Well, Switzerland accompanied its neg interest rate policy with the floating of its currency, which appreciated over 10% in one day. Hard to decompose the effects; deflation could simply be a function of falling import prices thanks to the much stronger Swiss franc.

    ReplyDelete
    Replies
    1. "...falling import prices thanks to the much stronger Swiss franc."

      Of course, a stronger Swiss franc is not an exogenous outcome. It's an endogenous outcome that's part of the same process we're discussing.

      Delete
  6. Problem with the Fisher effect is not that it is wrong but that it only holds in the long-run. Checking unemployment numbers or GDP data, especially of "golden-feathered" Europe, shows that we are in the short-run so this is the wrong model.

    ReplyDelete
    Replies
    1. See this:

      http://newmonetarism.blogspot.com/2015/09/intuitive-neo-fisherism.html

      Actually, the way to look at this is that there are two effects, the Fisher effect, and the liquidity effect. The liquidity effect is a short-run effect; it's temporary, and so the Fisher effect ultimately has to dominate. Further, in order for an increase in the nominal interest rate to produce a decrease in inflation, even on impact, you have to have a very large - I think implausibly large - liquidity effect. Think about it. For inflation to fall when the nominal interest rate rises, you have to have a more-than-one-for-one increase in the real rate as the result of an increase in the nominal rate.

      Delete