Winner of the New Statesman SPERI Prize in Political Economy 2016


Thursday 14 June 2012

Helicopter money, Inflation targets and Quantitative Easing


For economists

                I’m often asked about helicopter money, and occasionally there are calls for this policy to be implemented. (Here is a recent example.) The way I think about this, at a very basic level, is that it is equivalent to a call for a temporarily higher inflation target.
                Think of an economy with just consumption and a government. Consumers are Ricardian. The government issues money and indexed debt, and it pays the interest on debt using lump sum taxes. There are two periods. The second period has flexible prices, but the first involves sticky prices, a demand deficiency and a ZLB.
                In this set up, a tax cut in the first period financed by issuing debt does nothing, because taxes rise in the second period to pay back the debt. What happens if the tax cut is financed by printing money? Prices must rise in the second period as money is the only nominal quantity, such that real balances in the second period are unchanged. What about the first period? Taxes have fallen, so it is tempting to say that lifetime income must have increased. However that tax cut needs to be kept in the form of cash, and not spent, because prices are going to rise diluting holdings of real money. So people save that tax cut as money. (I guess it’s like the Ricardian case, except it’s the inflation tax that you are saving for.)
                However there is a real effect in the first period, because with nominal rates at the ZLB, higher expected inflation reduces real interest rates, which means it’s sensible to bring forward some consumption into the first period. So helicopter money works through raising expected inflation, and not through any income effect, in this very simple set up. I think this is consistent with some old posts by Tyler Cowen and Brad DeLong. (There is a paper by Buiter (NBER 10163) which talks about this issue, but my pdf reader does not recognise the WPMathA font, so I cannot read the maths. If anyone can help me here I’d be very grateful.)
                In this very simple context, calls for helicopter money are equivalent to calls for a temporary increase in the inflation target. The policy only works because inflation increases. If the government printed money to spend on its own goods and services in the first period, then there would be a direct positive demand effect in addition to the inflation effect, but this direct demand effect could be achieved by a balanced budget fiscal expansion, without the need to print money. 
                If the government cut taxes by printing money, but then reduced the stock of money back again in the second period by raising taxes, then nothing happens in this simple model. The fact that there is more money in the first period does nothing, and consumers are no better off. We do not have any of the imperfections and margins in this simple model that QE is thought to work on in the real world.
QE involves a temporary exchange of bonds for money, if we consolidate the government and central bank. How do we know it’s temporary? First, because central bankers say it is. But, second and more importantly, because central banks appear totally wedded to their inflation targets. Data on inflation expectations suggest this is also what the private sector believes. If QE was not temporary, we would get the equivalent of helicopter money. In the QE case, and unlike helicopter money, the bond stock falls, but we can get back to helicopter money by cutting taxes by issuing bonds, which as the model is Ricardian does nothing.
                So it seems to me that calls for helicopter money are isomorphic to calls for a temporary increase in the inflation target, and none the better or worse for that. But if I’m missing something important here, do let me know.

14 comments:

  1. I think that the Ricardian hypotheses could be discarded by channeling the money to the young at the sole expense of those not expected to be living at period two.

    As a matter of fact the tax to be levied at period two could be a tax imposed on the transfer of the taxable estate after a person dies while the money could be given exclusively to the young.

    This would be equivalent to just a premature transfer of wealth from a generation to the other since the young would be receiving at period one what they would be entitled to receive anyway at period two and to do so at no discount as the discount rate would be zero.

    ReplyDelete
  2. "Temporary increase in the inflation target" could mean a couple different thing: higher inflation now, and normal inflation later; or higher inflation now, lower inflation later. I assume you mean the first?

    ReplyDelete
  3. If the problem is a balance sheet recession where many more agents than usual are liquidity constrained then a helicopter drop allows a significant chunk of an agent's debt to be repaid immediately, reducing the number of agents who are liquidity constrained. Higher inflation will eventually achieve the same outcome - but not immediately, which suggests they're not isomorphic?

    ReplyDelete
  4. "In this set up, a tax cut in the first period financed by issuing debt does nothing, because taxes rise in the second period to pay back the debt"

    So why does government spending in the first period financed by issuing debt do something? I'm assuming you don't believe that RE nullifies the effectiveness of government spending? As long as r<g, tax cuts can also be money-financed and don't need to be analysed in a debt-financed framework, even with Ricardian consumers. It's the same argument as for government spending, just with different level of government spending in period 1, which is irrelevant in the infinitely lived case. Indeed, a money-financed tax cut is indistinguishable from a money-financed fiscal transfer - both are helicopter drops and the BoE's recent musings on implementing tax cuts/ fiscal transfers through thee PAYE system reflect this equivalence.

    One can analyse a helicopter drop in the interest/inflation framework, but a drop in the real interest rate is not necessitated. You get to that conclusion only by assuming that nominal interest rates will remain paribus at ZLB, but this is a restrictive assumption. They need not be. In the Wicksellian/ New Keynesian framework, a helicopter drop can simply end up increasing the natural rate of interest to equilibrate it, rather than decreasing the actual (market ) real rate of interest. Along with a rise in inflation expectations, this can result in higher nominal interest rates, consistent with the monetarist observation that the nominal interest rate is not indicative of the stance of monetary policy.

    The big benefits of helicopter drops are - 1) it repairs household balance sheets, so to the extent private spending is liquidity/ risk aversion constrained, it is absolutely direct in its effect. 2) it increases inflation expectations directly by debasing the currency, the tried and tested method of banana republics, rather than hoping and praying that the central bank is the nominal god. 3) it channels money to households rather than to presumably rent-seeking government officials and does not differentiate between sectors 4) it is mildly progressive, as opposed to asset purchases which may be regressive.

    Pure money-financed fiscal transfers/ tax cuts can be best understood, as Steve Waldman at Interfluidity put it, as the equity financing of the private sector, without nationalization. If you worry about the inefficacy of monetary policy or of its regressive effects (http://www.macroresilience.com/2012/06/04/the-case-against-monetary-stimulus-via-asset-purchases/), helicopter drops should be your first port of call in a world that has currency and thus cannot have significantly negative nominal rates.

    Increasing government spending can be favoured over helicopter drops only if you're that the slightly higher marginal propensity to spend of the government counterbalances the standard public choice concerns as well as the private sector debt/ balance sheet concerns.

    Note that that microeconomics of public finance decisions remain intact and if infrastructure must be invested in, then it must be invested in. Irrespective of the business cycle.

    As for the specific claim in your last paragraph, the big difference is that helicopter drops do not presume the omnipotence of the central bank in the face of a variety of structural factors, which inflation targets do.

    ReplyDelete
  5. Ok, read your post again and obviously you recognise the equivalence of fiscal transfers and tax cuts. My apologies for unnecessarily spending time on that.

    Reading parts of Buiter paper, he does not even require the r<g case, and he is making his case through the more monetarist 'irredeemability of fiat money' angle. He talks about a fiscal/wealth effect on private demand, and he rejects the 'save money to pay for future tax' hypothesis of yours/Krugman's by proving that all that is needed to avoid this outcome even with forward-looking Ricardian consumers and a balanced budget constraint is that the money stock grows at a rate higher than the nominal interest rate on money. (Notice that in a framework of endogenous money stock + super-neutrality of money + zero credit risk of the government, this is exactly equivalent to r<g. But nevermind.) Buiter' making the standard monetarist/ New Keynesian case that helicopter drops and QE are equivalent, as long as the increase in base money is expected to be permanent. For the purpose of your argument, if QE was expected to permanently increase base money, it would also work precisely through raising inflation expectations. In Buiter's model QE = helicopter drop = raised inflation target, as long as we assume a perfectly credible central bank and permanence of base money rise.

    De Long's position is slightly different - he is highlighting the greater possibility of convincing permanence in a helicopter drop, but still prefers direct fiscal policy differential marginal propensity to spend of the government. This is a valid point but it is uninformed by public choice, as I mentioned earlier. Tyler Cowen explicitly recognises inflation expectations but mainly from a game-theoretic microeconomic perspective, rather than the Wicksellian real rate perspective. It is similar in vein to David Glasner's recent resuscitation of Keynes from the prism of Armen Alchian. http://uneasymoney.com/2012/06/13/alchian-on-the-meaning-of-keynesian-involuntary-unemployment/

    None of these mental/ formal models talk about household liquidity and balance sheet constraints, which may be the primary channel through which helicopter drops work.

    Ultimately, helicopter drops do work through inflation expectations, yes. They are preferable to QE/inflation targets because it seems easier to convince the private sector of their non-repetitive and permanent nature, because they are not regressive, and because they directly debase the currency. They also look preferable to fiscal policy because of reasons of public choice.

    ReplyDelete
  6. Try Foxit Reader instead of Adobe. Works for me.

    ReplyDelete
  7. I make certain alternative assumptions, and draw different conclusions:

    1. Consumers are not Ricardian.

    2. Inflation expectations play a very small role in actual consumption and investment behavior across the economy.

    3. Even if they did play a large role in actual consumption and investment behavior, inflation expectations have proven very resistant to previous policy episodes publicly perceived as "money printing" - that is QE. So why should money-financed fiscal transfers have any different effect on inflation expectations than did money-financed transfers through the banking channel?

    4. In an economy with significant underemployment of material and human resources, money-financed government purchases, public investment and pure gifts of purchasing power are unlikely to cause actual price increases. The additional money is easily absorbed by an expanding economy. M and Q go up. Employment goes up. P and V stay the same.

    5. Economies can stabilize in a prisoners' dilemma condition of high unemployment: That is, individual firms recognize that if they and all of the other firms in the economy hired more people and invested in expanded production, then they would collectively generate more worker income and higher aggregate demand sufficient to warrant the costs of expanded output. But no individual firm can affect the hiring decisions of all of the other firms. So unless individual firms know that all those other firms will hire and expand production, it is not rational for that individual firm to hire and expand production.

    6. There are a variety of valuable but currently unproduced goods and services that experience has shown private sector entrepreneurialism does not and cannot deliver, and that can only be produced by the public sector. In fact, there are always far more of these goods and services than there are resources to produce them. There is never a good reason, then, to eschew expanded public enterprise during periods when private enterprise is flagging.

    So I would say that the chief reason for money-financed fiscal expansion has little to do with generating higher inflation or expectations of inflation, and much more to do with (i) the government's ability to break the prisoners' dilemma logjam of a low-employment equilibrium, and (ii) the inherent benefits of public enterprise, especially when these can be accomplished without significantly diminishing the purchasing power of money already in circulation.

    ReplyDelete
  8. Distribution matters.

    Net debtors who received helicopter money would presumably split the windfall between paying down debt and extra consumption. The poorer the recipient, the more would likely be spent.

    This argues for a helicopter drop over England's Northern cities.

    ReplyDelete
  9. I'm not sure the toy 2 period model can capture temporary versus permanenet or persistent inflation expectations changes. You'd need at least 3 periods. The problem with calls for changing the inflation target is that in the long run you'd expect to get the same real interest rates with higher equilibrium/flexible prices nominal rates, and at least in the nk model you're using the output gap must be stationnary and eventually converge to zero. Temporary changes in inflation expectations would work, though I haven't yet seen a clear prescription for such precise control by the central bank over inflation expectations. The main reason helicopter drops don't necessarily do much in modern nk models (and in economies with modern payments systems) is that cash in advance constraints are usually satiated already at least at the frequencies relevant to macro analysis (that's the essence of cashless economy models). Therefore, central bank attempts to raise the money supply directly need not increase real money balances. This leaves the tricky inflation expectations channel.

    ReplyDelete
  10. Nice little brain twister, but please do not try to draw real world conclusions from this. We have had enough of that.

    ReplyDelete
  11. You can actually download the fonts from Corel officially
    The fonts are available at
    http://www.corel.com/6763/downloads/WordPerfect/wpwin/9/wpfonts.exe

    This link may be useful for this problem:
    http://www.warnercnr.colostate.edu/c...s/fontlist.htm.

    ReplyDelete
  12. I draw my conclusions from looking at the world and using logic and a whole different array of models to organise all the information overload out there. I draw conclusions, because in the end conclusions will have to be drawn, actions will have to be taken even though we will never actually understand the world as it is. I suspect this is quite a common practice in fact. If you're not happy with this feel free to run your VARMA(p,q) model or linear regression and believe the coefficients actually tell you something about how the world works as opposed to just being en efficient device for unconditional forecasting for a few quarters.

    ReplyDelete
  13. Simon, I am disappointed here because the model is too simple and leads to the wrong result, and thus does not inform policy.

    The place where the model runs importantly afoul of reality is in the assumption that there is a government, and everyone else, and the "everyone else"'s are all equal.

    One of the reasons we are in a demand depression now is because of increasing income distribution inequality. The more money goes to rich people, the more saving there is and the less spending there is, because the super-rich are already satiated. Thus, a key effect of any fiscal transfer to ordinary consumers is an immediate boost in spending, no matter what future taxes and inflation expectations are, becuase the 99% is going to spend regardless of those things, and the 1% weren't spending anyway. Plus, the future taxes and inflation primarily affect the 1%, because you can't tax people who don't have money and inflation affects people holding money more than those who don't.

    Please re-run your analysis with two groups of consumers: a Ricardian 1%, and a 99% who do no planning for future taxes, have no inflation expectations, have no savings, and pay no tax. I recognize this model is deficient in various ways but it's a lot closer to reality than yours, and will (I believe) lead to a much more useful (and intuitive) result.

    Kenneth Duda
    kjd@duda.org

    ReplyDelete
  14. First of all, thank you for this blog. Apart from all its scientific relevance, I do appreciate the fact that you are probably the most unbiased analyst I've come across in the blogosphere.

    Now, the question I have to ask you. I am still starting to learn Economics, at the very start I shall say, so the only thing I can actually work with "well" is the old Keynesian model and the standard, classical one, the General Equilibrium.

    I know this will be completely wrong in scientific terms, but I tried to use these models (Keynesian for the first period, Classical for the second) to try to understand what you said back here.

    Is it correct for me to say that, in the first period, with the increse in money supply, interest rate raises from the hypothetical ZLB, and private consumption also raises (eventually crowding out due to even less investment?).

    In the second period, as a result of these higher interest rates and now going "classical", consumption will be lower than with previous market conditions and therefore demand for money will decrease, and that is why prices should go up, that being the supposed "higher inflation expectations".

    Is this close enough? Thank you in advance, and sorry for bringing you back to such an "old" post... although still shockingly relevant.

    ReplyDelete

Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.