Wednesday, September 2, 2015

Matthew Yglesias and the economic theory of everything


Matthew Yglesias apparently knows the economic theory of everything. He says these are effective policy options:
  • Printing money to buy longer-term government bonds (quantitative easing)
  • Printing money to buy foreign government bonds (currency devaluation)
  • Abandoning inflation rate targeting in favor of price level (or nominal income) targeting
  • [Negative interest rates]
How do you know this Matt? What theory tells you this?

The Fed is having trouble making its inflation target and a level target would involve inflation going higher than it currently is.

The giant rounds of QE don't seem to have had a very strong effect on output in the US (Mark Sadowski says it's about a 10 to 1 ratio of QE to price level, and I agree that is how much effect it would have if it wasn't approximately zero [1]). The impact in Japan and the EU have also been muted.

Switzerland's devaluation had zero effect on increasing inflation (deflation followed). [Although there seems to be some ability to induce deflation with the traditional method of decreasing the money supply as I will talk about in a future post -- more consistent with my idea of a maximum inflation rate.]

And macroeconomists didn't seem to think negative interest rates were even possible until last year, but after reading some blogs [2] Matt knows it's a viable monetary policy option. Basing monetary policy on program buying by pension and mutual funds (a commonly discussed mechanism for negative rates, along with the inconvenience of holding billions in physical cash) seems like exactly the kind of argument Lucas was warning about with his famous critique.

Sure, these are existing policy options. Janet Yellen could resign, for example. However Matt implies that these options will actually do something even though our past experience shows they have limited impact. That statement is of course theory and counterfactual dependent. But that is my point.

I don't disagree with Matt's conclusion that the Fed shouldn't raise rates. And the initial reason is much better (don't raise rates now just so you can lower them later). But he goes on to say you shouldn't worry about policy ineffectiveness because there are effective policies (he knows they're effective because he understands economic theory of everything).

I think a bit of praise from Scott Sumner has gone to his head.

Footnotes:

[1] In the information transfer framework, none of these options will do anything in our current situation.
[2] I think it's mostly Scott Sumner's blog.

17 comments:

  1. Lol... you're on a roll today.

    I noticed today that Noah Smith doesn't erase Marcus Nunes' (João Marcus) self promoting posts, even when he leaves them two at a time!

    Also, you might get a kick out of this interchange.

    ReplyDelete
  2. Jason, I don't understand how the information transfer framework could predict that "it wouldn't do anything in our current situation" if the Fed announced a nominal income target (Yglesias' words), which I interpret to be a permanent regime change to an NGDI level target that advances 4.5% each year, where OMO is guided by an open NGDI prediction market.

    If you're right that nothing would happen, then the prediction market would presumably tell the Fed that --- that no matter how many federal bonds, agency bonds, foreign exchange, and precious metal the Fed bought, the Fed would fall short of its +4.5% NGDI target. The Fed will say, "wanna bet", set IOR to negative 0.25%, and buy everything in sight. Buy all of the bonds and precious metal in existence, and then go to work seeing if you can print enough USD to acquire the entire monetary base of yen and euros. You really think that would have no effect on NGDI? What does the Information Transfer Framework predict would happen to all those trillions (quadrillions, if needed) of dollars? They'll all sit in negative-rate reserve accounts? Drive the Dow to 30,000 or 300,000? Get stored as paper currency in a vault the size of a mountain?

    You might be assuming that nothing that radical could happen because of political constraints, but that's what "adopting a national income target" would entail. If political constraints mean the Fed can't adopt one of Yglesias' options... well, that's different from actually trying it and finding it has "no effect" as you asserted.

    I find it frustrating that so many people think the Fed is impotent. The Fed is the monopoly issuer of fiat currency. How can it be impotent? In fact, the Fed, if it chooses, is omnipotent over any single nominal aggregate, up until people abandon the US dollar as a medium of account.

    I know you are smarter than me (I say that earnestly, not sarcastically), so maybe I'm really missing something here. I'd love to be enlightened because if NGDILT (or NGDPLT) is a dead end, I can save my money and stop funding researching it.

    Thanks,
    -Ken

    ReplyDelete
    Replies
    1. Ken have you read Jason's draft paper? I'm just curious. I've made it to page 8 (of 39) so far.

      The ITM does allow for hyperinflation, but it's a different solution to the equations (if I understand correctly). It's always been somewhat mysterious to me where the dividing line is exactly.

      I always wondered if it's possible to pop back and forth between the two solutions somehow (to produce a desired overall outcome). I think I've asked Jason about that before... it was a long time ago, and my memory of his response is murky.

      Delete
    2. Hi Tom,

      I'm sorry, I've been trying to read Jason's paper, but I'm so goddamned slow to get it through my thick head. It's especially frustrating for me because I love the intuition behind information transfer economics. After all, a gas isn't a room-sized atom, and a gas is not composed of all-knowning hyper-rational atoms with rational expectations of what all the other atoms will do. Instead, each atom propagates information about conditions around it, and the effect of that information transfer is an increase in entropy until (absent external shocks) we get the heat death of the universe (full-commodity equilibrium) or all matter collapsing to a single point (monopoly). The analogy is so beautiful.

      But, I can't follow the details. For example, I think Jason models the price as a detector propagating information from demand to supply. But why that direction? Doesn't the price also propagate information from suppliers to customers?

      But I do mean to try to really get through it, because I think Jason is onto something very important. We really can do better than assuming lifetime-utility-maximizing rational-expecting agents or throwing up our hands and saying it's too complicated to analyze. Yes, we cannot model how every atom bounces off of every other atom, but we can model aggregate behavior anyway. So, hopefully I'll have a more intelligent response someday.

      -Ken



      Delete
    3. "But why that direction? Doesn't the price also propagate information from suppliers to customers?"

      That's a great question. One I've been meaning to revisit myself. Jason has a post from more than a year ago (maybe two years ago) with some friendly diagrams of little people and the products they buy and a very simple explanation of why the market transfers information. I don't know if that addresses your question here, but maybe it does (I couldn't find it just now, unfortunately).

      Given that the direction of information transfer is sound, I've been trying to understand the intuition in the first 7 or 8 pages. It's stuff I've seen before, but as with so many of Jason's posts, there's always been parts I've understood clearly and parts that are more murky that I just glossed over. I think I finally have a handle on that first bit (which dates to the 1st month of this blog BTW!). Anyway, we've been having a good thread on the subject here if you're interested:
      http://informationtransfereconomics.blogspot.com/2013/04/supply-and-demand-from-information.html

      Delete
    4. Hi Ken,

      I wasn't saying I know for a fact that the options Yglesias provides won't work, I was taking issue with Yglesias presenting them as options that he knows for a fact will work without disclosing what underlying theory he was using. If he had started his post out with "Scott Sumner says X will work ...", that would have been fine. I do present some evidence that it at least isn't obvious they'd work.

      Imagine Janet Yellen was the chair of the Federal Energy Board and told us we're running out of oil, coal and natural gas so we're going to have to cut back. Now Matt Yglesias comes along and says we don't have to cut back because:

      • We can switch to LED lightbulbs
      • We can increase solar power
      • We can build nuclear fusion plants
      • We can begin to use quantum singularities

      Now LED lightbulbs would do a little bit. And solar is rapidly becoming a viable option (but cannot currently replace fossil fuels). However nuclear fusion hasn't been demonstrated to work and the theory behind quantum singularities is really more Star Trek than reality. That bullet list wouldn't be a good argument against the FEB's policy.

      The idea that the Fed is a monopoly issuer of fiat currency and can inflate at will isn't as cut and dried in the information transfer framework because the information content of a dollar isn't constant. If it was, then things would be simple and the Fed could inflate at will. But (if the IT framework is right) the information content of a dollar is based on output measured in dollars and number of dollars. It's almost as if there is a kind of "special relativity of money" where instead of clocks slowing down when you go faster with a constant speed of light you have the value of a dollar changing based on the amount of dollars in the economy while keeping your units of information constant.

      This of course may not be correct and maybe the real reason for the apparent fall in information content is entirely based on a trend in expectations, not a trend caused by some objective mechanism.

      As an aside -- I think exchange rates may actually be a liquid NGDP futures market, albeit one with potentially too high a volatility to be useful. More on that to come ...

      Delete
    5. Ken and Tom,

      Here is that post on the direction of information flow ...

      http://informationtransfereconomics.blogspot.com/2014/09/which-way-does-information-flow.html

      Since I assume information equilibrium most of the time -- I(D) = I(S) -- it actually doesn't matter the direction. However, the linked post addresses some of the issues you mention.

      Delete
    6. Jason, that link is much more informative regarding directionality of information flow than the one I had in mind. Thanks. The one I had in mind you have a permanent link to in the right hand column (I just now realized):
      http://informationtransfereconomics.blogspot.com/2014/03/apples-bananas-and-information-transfer.html

      Delete
    7. Thanks Jason. I've got to figure out your framework, so I can make really sense of a phrases like "information content of a dollar is based on output measured in dollars and number of dollars" and "the value of a dollar changing based on the amount of dollars in the economy while keeping your units of information constant."

      What we saw was the monetary base expanding 360% (going from $875B (Oct 2008) to $4000B (June 2014)) while the value of a dollar falling about 7 or 8% over the period (CPI from 220 to 238) and the real output (of the dollar-denominated economy) expanding about 7% over the period (RGDP from $15T to $16T). It's been fun watching various groups struggle to explain this --- why doesn't all this cash do something other than sit in bank reserves.

      It sounds like the Information Transfer explanation involves a shock to the amount of information in a dollar. Which sounds odd, but I (sigh) still don't understand the IT framework well enough to really evaluate.

      For what it's worth, the MM explanation, which I think is right, is that the markets assume that the Fed's inflation commitment is hard, and therefore the "extra" money will be destroyed at the first whiff of inflation. This money loses its "money-ness", and becomes effectively equivalent to a short-term bond, hence it doesn't get spent, every increase in M reduces V, and there's no inflation. If the Fed was credibly committed to keeping as much money around as needed to hit its NGDP target, the market would respond very differently to the Fed's monetary expansion. In other words, the only reason the Fed can't seem to generate inflation is because no one believes the Fed wants inflation.

      -Ken

      Delete
    8. This comment has been removed by the author.

      Delete
    9. The Fed has shown that it can put brakes on the economy, but it has not demonstrated that it has an accelerator. Sure, monetary easing has resulted in increased economic activity, but the failure to reach inflation targets since the financial crisis is telling evidence.

      Delete
    10. Hi Ken,

      The temporary injection view is the best explanation I have in the information transfer model of why (electronic) reserves and short term (e.g. 3 month) interest rates are connected, while physical currency (harder to take back once it is in the economy) is related to long term rates (e.g. 10 year). And it is physical currency that determines inflation.

      The reason the reserves haven't been converted into physical cash and hence inflation is that the long term interest rates are not at the zero bound.

      I discuss this more here:

      http://informationtransfereconomics.blogspot.com/2014/02/the-link-between-monetary-base-and.html

      RE: information content

      That is an attempt to put the information transfer index in words. In the inflation model, k = log NGDP/log M0 and when k is large, you have a quantity theory of money economy and when k is near 1 you have a liquidity trap economy. Essentially, each dollar carries more information when k is large than when k is small (one dollar state matches up with many NGDP states when it k is large and with fewer when k is small).

      I talk about this a lot more here (with more words than math):

      http://informationtransfereconomics.blogspot.com/2014/06/money-unit-of-information-and-medium-of.html

      It basically comes down to the unit of account (measured in information content) being different from the medium of exchange (measured in quantity of money) when they come together to create the price level.

      Delete
    11. Jason, thank you for another thoughtful reply. I hope to make sense out of this, and ultimately be able to explain in Information Transfer terms why NGDPLT would lead more quickly to full employment than current monetary policy. Intuitively, by adjusting the money supply to stabilize NGDP, the markets have less work to do --- less information that needs to be transferred --- than an interest-rate-targeting monetary policy regime where insufficient monetary expansion at the ZLB leads to economy-wide spending shortfalls that individual market participants have a hard time distinguishing from localized slack demand. i.e., to any one producer, an AD shortfall is indistinguishable from a shift in preferences away from whatever I'm producing. I'm not sure how realistic my goal is here because (a) I am totally shooting in the dark, and (b) I imagine any model here will have to involve different information transfer coefficients in different circumstances, which will look "hacked" or "rigged" to get the result I want unless I'm somehow able to empirically establish those different coefficients. (Sorry if none of that actually makes any sense).

      -Ken

      Delete
  3. Tom, thanks for the pointer to the conversation on the April 2013 thread. I can see you have worked through a lot of the issues I've been stumbling on. I have a lot to go through here in my copious free time :-)

    ReplyDelete
  4. Jason, I've followed you from the beginning, and I believe that I have a decent grasp on your modeling. My question is why we don't treat bank reserves as net debt liabilities. There just may be immense preexisting leverage that immunized the effect of greater reserves and base money. Hence little NGDP growth stemming from reserve creation: the new reserves are simply collateralizing preexisting debt. A low leverage economy would show much more NGDP growth.

    ReplyDelete
    Replies
    1. Thanks for following along!

      I am not entirely sure I understand your mechanism, however -- is it like pumping air (reserves) into a vacuum (created by the liabilities)?

      Delete
    2. The point is that bank reserves are effectively equity capital. In fact, that's all they might be. If banks show themselves to be massively undercapitalized due to risk assets that go bad, then all the fed may be doing is recapitalizing bank balance sheets. Reserves may simply plug the equity hole. Why should we then call this a monetary expansion? Printing reserves into high leverage and impacted bank balance sheets could be like pumping air into a vacuum and wondering why you aren't getting positive air pressure.

      Delete