Backgrounder on h money

Redrafted on November 25 

Main Points

  • In the US, interest in unconventional monetary policy is fading, as the economy seems to pull away from liquidity trap conditions.  However, discussion of unconventional policy may revive when the next recession arrives, perhaps a couple years from now.
  • In this blog post I argue that helicopter money will not be a viable option to deal with the next recession. Any program large enough to matter would be far too inflationary, unless the Fed reneged, which it would. Knowing this in advance, the Fed would not commit.
  • Any program small enough not to be excessively inflationary would not be worth changing the monetary policy regime to enact.  Moreover, announcing such a program might be met with derision, quite appropriately.
  • Many analysts seem to be confused about two aspects of helicopter money. First, at inception, a conventional fiscal stimulus is converted into h money by a promise, not a funding operation.  Second, the observation that inflation is low and at little risk of rising would not be part of a case for h money.
  • Some of the best arguments against h money are actually made by advocates of the policy, unwittingly, if I may be so bold.

Temporary escape from liquidity trap

Interest in unconventional monetary policy has faded in recent months, as the US economy has pulled tentatively away from liquidity trap.  Underlying inflation has begun to renormalize and a maturing business expansion has apparently pushed up the equilibrium real interest rate slightly.

Accordingly, the Fed’s focus has shifted from finding additional means of unconventional stimulus to determining the appropriate pace of conventional tightening.  A major, debt-financed infrastructure spending program, of the sort now being considered in DC, would tend to reinforce this shift.

Indeed, fiscal stimulus in this environment might give a larger lift to interest rates and bond yields than to the pace of GDP growth, at least over the next couple years.  At near full employment, higher demand in the public sector may require lower demand in the private sector, with financial markets doing the rationing. We have arguably already seen that mechanism over the ten days since the federal election. That has nothing to do with the return of the mythical bond vigilantes. The bond market will do what it’s told by the Fed and economy, as almost always.

However, there is a good chance that exit from liquidity trap will prove temporary.  For good or ill, the Fed seems unlikely to allow underlying inflation or inflation expectations to move much above where they were at the height of the last expansion, in the mid-2000s.  And in an environment of reduced potential economic growth and still “surplus” global savings, the natural real interest rate seems likely to remain depressed by historical standards.

Come the next recession, then, the US may find itself right back in liquidity trap, i.e., at the lower bound on nominal interest rates.  If so, then the Fed will have to figure out what tools it might use to help end the recession, quicken the recovery and again exit liquidity trap again.

Helicopter money is a promise, not a funding operation

In this post, I argue that so-called helicopter money (h money) will not be on the list of possible remedies.  A major, money-financed fiscal expansion would be unworkable because it would lead either to intolerably-high inflation or, more likely, to the Fed reneging on its side of the bargain. Knowing this in advance, the Fed would not commit, which makes the idea a non-starter practically speaking.

I should probably deal immediately with the most obvious and common retort to the claim made above.  It would be odd, many say, to worry about excessive inflation in an environment where the main economic problems are insufficient demand, depressed inflation and liquidity trap.  However, that retort seems mistaken for its failure to take account of the calibration issue. A helicopter drop small enough to avoid excessive inflation would probably have little effect, leaving aside that it would be greeted with derision. And a drop large enough to “work” would be far too inflationary, unless reneged upon, which it would be.

The calibration problem arises from the awkward fact that the relevant money stock, presumably used to finance the fiscal expansion, is just too small a fraction of nominal GDP to allow money finance to be a feasible policy option. I will get to the simple arithmetic behind this problem below.

But, first, it may be helpful to get clear what a helicopter drop even is.  According to its advocates, a helicopter drop occurs (in the US) when the Fed promises to finance a fiscal expansion with an eventual increase of the money supply, where money is defined as the permanently-zero-interest-bearing liabilities of the Fed.

What converts a standard fiscal expansion into an h drop, then, is that promise itself.  The immediate evolution of the Fed’s balance sheet is entirely irrelevant. This is a point of intense confusion, arising in part because people take the metaphor too literally OR wrongly believe that h money is some sort of extension of QE.  It will be much easier to follow the argument here if you pause to internalize that the promise matters and that the immediate funding operations do not.

When fiscal expansion is ultimately financed with money, there is meant to be no rise – relative to baseline – in the longer-term path of the federal debt / GDP ratio.  There is also, therefore, no reason to expect the fiscal stimulus to be offset by Ricardian effects, in which private actors react to higher debt and future tax liabilities by raising (desired) savings. Indeed, to the extent that the overall program might lift inflation expectations, and thereby reduce real interest rates at the zero bound, desired savings might fall, as even desired investment picks up. Such is the theory, such as it is.

For the remainder of this post, I will assume that the fiscal expansion comes in the form of increased government spending, rather than in the form of tax cuts.  I choose this assumption because it is simplifying and because it actually flatters the case for helicopter money, on the grounds the multipliers on high spending exceed those on tax cuts.

screen-shot-2016-11-14-at-7-03-05-pm

With that in mind, there are three basic, qualitative aspects of helicopter money that are often misunderstood and are therefore important to get straight from the get go.

First, when advocates say that the fiscal expansion is financed with money, they mean that the long-term path of the money stock is ultimately tilted higher (in present value terms) by the same amount as the deficit associated with the fiscal stimulus.  Another, more precise, way to put this is that the seniorage on the money stock is made equal in PV terms to the cost of carrying the additional debt.  But subject to some plausible – if not guaranteed assumptions – about money demand and the gap between real interest rates and real growth, we can simplify to just saying the money stock goes up by the amount of the initial debt.

Most people weighing in on h money seem roughly to have the correct intuition on that part. What is less well understood is that it is the money stock’s future path, rather than its immediate change, that is most relevant here.  Whether the Fed chooses, in the first instance, to buy the additional debt created by the fiscal expansion is entirely incidental, and falls under the rubric of quantitative ease (QE).

Second, we need to be clear on what the “relevant” money stock is in this context.  It is currency in circulation and most certainly not what we still somewhat misleadingly all the monetary base.  As a legacy of QE, and how the asset purchases were funded, most of the monetary base now comprises reserves, which carry a market rate of interest and thus cannot be a source of “money” finance of fiscal expansion.

Third, and most fundamentally important, the Fed cannot by impose by fiat the stock of currency in circulation.  (Among the academics who are confused by horizons, this is sometimes called the “reflux” problem, but I will not get into that here.)  Rather, the Fed can influence the longer-term path of that stock by changing the path of nominal GDP.  On the standard assumption that the Fed has little control over real variables in the long-run, the choice to lift nominal GDP above baseline is basically a choice to lift inflation for a while.  The steeper path for nominal GDP, meanwhile, will raise the currency stock by raising the transactions demand for money.

Again, I can anticipate a retort to this point, mostly because I have already heard it many times. The retort is that the demand for money and its relation to nominal GDP (or velocity if you prefer) is not stable.  That is true, and the truth of it just further undermines the case for helicopter money, by introducing uncertainty into our already daunting estimates of the amount of inflation that must be tolerated to money-finance a meaningful fiscal expansion.  Please keep in mind that it is not me who is making the case for this crazy policy idea. I am simply assessing it in terms of its own logic, such as it is.

In this context, it is best not to think of velocity as a causal force converting the money stock into nominal GDP.  Rather, should be viewed velocity as a measure of the amount of additional nominal GDP that must be delivered in the future, primarily via higher inflation, in order to money-finance a given fiscal expansion. And not knowing that measure precisely, because velocity is unstable, is a flaw in the case for h money, not a feature.

The money stock is too small to be worth taxing

With those qualitative issues in mind, let’s turn now to the quantitative issues involved here. The most important of these is that is that the amount of inflation that must be accepted in order to commit to money financing a fiscal stimulus is proportional to the size of the stimulus (relative to GDP) and to the (steady-state) velocity of the currency stock.

As the chart above shows, the velocity of currency has been falling in recent years, which at the margin actually strengthens the case for h money. But at just over 12 ½, velocity is still far too high for this policy to work.  Imagine, for example, just a moderately-sized fiscal stimulus package of 4% of GDP, delivered, say, in the form of an infrastructure spending program running at $400 billion a year for two years.  (Anything much smaller than this would probably not be worth debating how to finance.)

As a first approximation, money financing a fiscal expansion of 4% of GDP, would require that the longer-term path of both the currency stock and nominal GDP be tilted higher by about 50% relative to baseline.  To get to that higher nominal trajectory within ten years, would require running inflation by about 4% above the old inflation baseline for a decade. So if the inflation baseline were 1% a year (on the premise that inflation would be stuck below target without this initiative), then we would be looking at 5% inflation for a decade, followed presumably by a return to target inflation.

This would be our first approximation. In fact, the precise amount of inflation required effectively to monetize a fiscal expansion of 4% of GDP would be a function also of potential growth, real interest rates relative to growth, and how currency demand responded to nominal interest rates during the inflation overshoot.

So there would be quite a lot of risk on either side of the central estimate of 5% a year for a decade. To repeat, this uncertainty would be a flaw, not a feature, particularly for a central bank wishing to have an inflation objective, whatever it might be.

Obviously, the Fed would not be willing to commit to tolerating and indeed delivering such an inflation environment.  And even if they were to do so, the importance of the implied tax on the currency stock would pale into insignificance beside the inflation tax on the federal debt itself.  The currency stock is currently about $1.5 trillion, compared with federal debt of $14 trillion held outside the federal government, of which $12 trillion held outside the Fed.

Using inflation to reduce the real value of the federal debt may or may not be a good idea.  But the answer would have little to do with the logic of helicopter money, which proposes imposing an inflation tax on money only.  Moreover, there would be huge consequences for wealth distribution associated with pursuing this policy.  This is something people generally understood intuitively, before the logic of helicopter money and its “manna from heaven”, as advocates misleadingly call the tax on money, came along.

These very simple arithmetic considerations seem to disqualify helicopter money as impractical, at least as conventionally understood.  To repeat, the conventional understanding is that the use of money finance allows the Treasury to deliver a meaningful fiscal stimulus, which itself pulls the economy out of liquidity trap, in part because it is not offset by Ricardian forces.

But there is a subtler form of helicopter money that we should consider before moving on.  The subtler form relies on inflation expectations per se, rather than on the brute force effects of fiscal stimulus.  For example, the Fed might commit to financing, say, a fiscal stimulus of 0.5% of GDP, not because such a tiny fiscal stimulus would matter in its own right, but because this would signal an intention to push the path of the general price level about 6% above the baseline that might otherwise hold.

There are three points to make about this weak-form helicopter money thesis. First, it is assuredly not how the program has been sold by its advocates.

Second, if the Fed wanted to be credible in signaling a higher interim inflation target, then it would be best to do so directly.  Approaching this indirectly via a commitment to financing a small fiscal stimulus would only add to confusion, mainly because the link between money finance and the inflation consequence cannot be known ahead of time with any confidence.

Third and closely related, there is no reason to believe that tying the higher inflation objective to money finance would raise the credibility of the inflation objective itself.

So let’s stick with dismantling the case for helicopter money actually made by its advocates.

Jordi Gali does the math

Thus far, I have not elaborated much on the central quantitative issue here, which is the relationship between (properly-defined) money velocity and the associated link between the size of a fiscal expansion program and the amount of inflation that would ultimately be required to finance it.  I have merely asserted the simple intuition that the inflation consequence is proportional to both the size of the fiscal expansion and velocity, properly defined.

You may relax, I don’t plan now to address that detail.  Instead, I am going to make an appeal to authority, but with what strikes me as a fun twist. That is, I appeal to the authority of a detail- and mathematically oriented academic economist who is actually an advocate of h money: Jordi Gali.

Gali’s paper from 2014, The Effects of a Money Financed Fiscal Stimulus, is to my mind the most rigorous and quantitatively-explicit exposition of the case for such a program.  This is not to say that Gali is empirical. That would be impossible, given that a helicopter money program has never actually been attempted anywhere. (We have seen helicopter drop results, in the form of hyperinflation, but they were not intended.)  Still, Gali addresses the basic arithmetic issues in a mathematically coherent framework, and he is clear about the assumptions that go into his simulation exercises.

Rather than elaborate on the mind-numbing technical detail, over which he has the superior grasp anyway, I would urge you just to go read (at least) the first eight pages of his paper.

On page 7 you will find this simple derived result, which insists the intuitively obvious, that within a helicopter money program “the growth rate of the money supply is proportional to the fiscal stimulus, inheriting the latter’s exogeneity”, and with (steady state) velocity.

On the following page, he demonstrates – with slightly tougher-to-follow notation — that this implies that the excess inflation required by a helicopter drop is similarly proportional.

This is precisely in line with what I have argued above, and so I would just refer you to Gali for the grinding algebra.  The only minor points of clarification I would offer are first, that he uses a quarterly model, which can leave an (incorrect) impression that the some of the main quantities are off by a factor of 4.  Second, expressed at an annual rate, he works with a velocity of 16, rather than my 12 ½, seemingly because he is more interested in the steady state than in the most recent observation. Recall, that a higher velocity makes money finance more inflationary.

And yet, Jordi Gali presents his analysis as offering a case for helicopter money.  As I read Gali, there are two reasons for this.  First, his central-case simulation assumes an economy with strongly New Keynesian properties, which are very friendly to the plausibility of helicopter money. In such an economy, as Gali simulates it, a fiscal expansion amounting to just 0.5% of GDP (cumulatively) has strong short-run effects on both output and employment that would, presumably, be sufficient to knock an economy out of liquidity trap.

I leave it to you to decide if such a scenario is plausible – or just an extremely counterintuitive compound result of the assumptions that go into Gali’s simulation work, any one of which may be plausible taken in isolation.  But to be generous to the idea of helicopter money, let’s assume that the simulation results are roughly correct. If a fiscal expansion of a mere 0.5% of GDP can really knock an economy out of liquidity trap, I doubt it matters much how it is financed. And if the magic here is in the inflation signal, then I would say deliver that signal explicitly.

Second, and tellingly, even within this apparently generous simulation of the short-run macro effects of helicopter money, Gali still derives that inflation overshoots baseline by 3% in the first year and by almost 2% in the second year, after which the inflationary effects begin to wear off.

This illustrates what for me is the most important quantitative issue here. When participating in a helicopter drop, the Fed gets a lot of inflation for a small fiscal stimulus, unless it reneges on its side of the deal, which it would.

Some advocates are already talking about reneging

The Fed could renege in a couple ways, the first of which is obvious, and the second of which is not really plausible, but highlights that some of the more prominent advocates of helicopter money actually know they are barking up the wrong tree.

The first way for the Fed to renege on its side of a helicopter drop would simply be to refuse to deliver the excess inflation required to allow the original fiscal expansion to have been financed by money ex post.  As mentioned at the top of this post, the Fed would be aware of this need to renege heading in, which is why it would be very unlikely actually to participate in an h money program.  But just for the sake of argument, if the Fed were to commit to an h money program and then renege, then the implications would be fairly simple. The Fed would take a large hit to its credibility, possibly with political implications, and the original fiscal expansion would ex post end up having been financed by debt, not money.

The second way for the Fed to renege is, for me, a source of some amusement because it puts the lie to the academic’s case for h money.  Recognizing, the extreme difficulty posed by the calibration issue, Adair Turner, proposes varying banks’ required reserve ratios in a way that would allow the original fiscal expansion to appear “money” financed, while evading the inflationary consequences of that.  The sleight of hand here is to pretend that required reserves are “money” on the grounds that they appear in the monetary base and would not pay interest.

In his somewhat confidently-titled paper from 2015, The Case for Monetary Finance, An Essentially Political Issue, Turner makes the following set of claims, which are worth citing at length:

it is still possible for the government/central bank together to counteract the emergence of a higher than initially intended stimulus to nominal demand, by imposing quantitative reserve requirements to limit the operation of the banking multiplier. If banks are required to hold a given minimum percentage of total assets or liabilities in monetary base reserves at the central bank, and if that percentage minimum can be increased by the central bank to offset unwanted credit and bank liability growth, then the eventual as well as the initial impact of monetary finance on aggregate demand can be constrained.

Return of quantitative reserve requirements to the central bank policy toolkit is therefore a logical complement to a policy of using money financed fiscal deficits to stimulate nominal demand.

Monetary theory is not usually funny, but this would be an exception.  In order to ensure that the original fiscal expansion is “money” financed, while containing the otherwise unavoidable inflationary implications of that, Turner suggests imposing a tax on the banking system in the form of higher reserve requirements. He justifies this tax on the grounds that there is probably too much financial intermediation occurring in the economy anyway.

This would effectively impose a sort of tax on bank credit intermediation: but since there are strong arguments for believing that credit creation can impose a negative social externality, such a tax could be appropriate.

So this is like a Pigou tax on carbon emissions, in which banks are like the smoke stacks atop coal-fired generating stations.  Maybe that characterization is fair and maybe it is not. This blog post is not the spot for that debate. But to pretend that a fiscal expansion financed ultimately with a tax on banks delivers “manna from heaven” and avoids Ricardian effects seems quite dishonest. What Turner is proposing is actually an increase in government spending financed ultimately with an indirect tax on the banking system. That is boring old conventional government finance in disguise. It has nothing to do with money finance, except as a matter of misleading semantics.

Former Fed Chairman Ben Bernanke offers a similar “solution” to the calibration problem in a recent blog post making the case for helicopter money. The main difference with Turner is that he proposes effectively a tax on the banking system that hits all liabilities equally, rather than those that are subject (in principle) to higher reserve requirements.  His justification is not so much the logic of the Pigou tax as the idea that the banks can be made to pay for having benefited from the fiscal expansion and resulting reserves creation.  There is no need to excerpt the relevant passage, which is a bit dense anyway.  His blog post is short and you can read it here.

The most striking aspect of these “solutions” to the calibration problem is not that they would work. As mentioned, they are dishonest in extremis, and would not be relied upon by any self-respecting leadership at the Federal Reserve.

Rather these feeble arguments merely highlight that some of the more prominent and illustrious advocates of helicopter money recognize (and feel a need to confront) the fatally awkward problem of calibration. It is as if they committed to helicopter money on the basis of monetarist intuition and without having worked through the mechanics. And then having been forced to do so, they get backed into double talk.

But people actually responsible for running monetary policy are unlikely to be convinced by such double talk or by a more honest case for h money. So this option is probably ruled out, barring the US economy experiencing a crisis far more serious than mere return to recession, liquidity trap, and lowflation.

I would guess there are a couple caveats here.  First, who knows how the Fed will be staffed in the coming years? The range of possibilities is actually quite wide, although I don’t know if the tilt will be towards inflation or deflation. I am simply less confident following the election.

Second, the Fed leadership have in the past told fairly harmless fibs, for example about QE, in order to lift confidence on the hope of a self-fulfilling prophecy. I would prefer a more honest approach on aesthetic grounds, but I concede it is not a big deal. Accordingly, I would not exclude, come the next recession, the Fed leadership claiming that they are acting on “roughly effectively the underlying logic” of helicopter money, even as they assuredly are not.

The day they utter such a fib, I might not want to be short equities.  But we can jump off that bridge when we come to it. For now, the current leadership is unlikely to fool itself about this unworkable policy. Or to change the metaphor, if they find themselves on the high wire, they will be aware that this safety net is not likely below.