The (near) inevitability, and who and when, of Helicopter Money

Helicopter Money is (almost) inevitable. The only questions are: who does it; and when do they do it. And we can't (easily) tell when it gets spent, and what it gets spent on, because money is fungible and we don't observe counterfactual conditionals.

Let's make some ballpark-correct assumptions. Assume currency pays 0% interest, and the (stock) demand for currency is 5% of Nominal GDP on average. Assume NGDP grows on average at 4% (2% long run real GDP growth plus 2% inflation target). So the central bank prints currency worth 0.2% of GDP each year on average (=5%x4%). Ignore the costs of paper and ink and administration. (It doesn't matter whether the central bank's "currency" is printed on paper, or plastic, or on computer discs. But if that currency pays interest, we have to subtract the interest rate paid on currency from the growth rate of the stock of currency, to calculate the central bank's profits.)

That means the central bank makes (average) annual profits of 0.2% of GDP from printing money. Those profits must get spent, by someone, at some time. The only questions are: who spends them; and when do they spend them?

Here is how those two questions are currently answered (in Canada, and in most countries).

The central bank buys interest-earning assets (mostly government bonds) with the currency it prints. (You can call that "QE", if you like silly new names for old boring things.) The central bank declares its profits as the interest it earns on those assets. (So the profits are declared later than if it declared the profits as soon as it printed the currency; but the present value of those two streams of profits would be the same, if discounted at the interest rate earned on the central bank's assets.)

The central bank then gives those profits to the government, which owns the central bank.

The government then decides when to spend those profits, and what to spend then on. It could spend them on real goods and services; on tax cuts; or on increased transfer payments. It could spend them years later, if it wishes. It could spend them years in advance, if it wishes (and can borrow). But sooner or later, the government will spend the profits from printing money.

0.2% of GDP isn't a really big deal, but it's not trivial either.

It would only go to 0% of GDP if: the inflation target were low enough, or real GDP growth were low enough, that NGDP growth were 0% on average; or the demand for currency as a ratio of NGDP were falling over time faster than NGDP is growing. Those things could happen, which is why I snuck in the weasel-word "near" before "inevitability".

I can imagine a world where the Bank of Canada has more power than it currently has. We could change the Bank of Canada Act to convert the Bank from a Crown Corporation (nationalised firm) to a registered charity(?) which decides how to spend the profits it earns, and when to spend them. (I am not in favour of doing this.)

I can (sort of) imagine a world where the government decides how the profits get spent, but is required by law to spend those profits immediately the money gets printed. But that law would be extremely hard to enforce, simply because money is fungible, and we can't observe counterfactual conditionals. The government could get around the law simply by cutting spending financed by taxes or borrowing by the same amount. Who can tell which dollars got spent where and when, or what the government would have spent otherwise? Only a very smart econometrician, not a lawyer or accountant.

The problem of fungibility and non-obervability of counterfactual conditionals means we also can't tell whether the profits from printing money get spent on increased transfer payments ("helicopter money" in the strict sense), or on tax cuts, or on goods and services.

[The Brits are currently debating all this. Read Simon Wren-Lewis for a good overview. Their debate looks a bit muddled to me. My recent post, which was supposed to be about some obscure Canadian monetary cranks, seems on reflection to be equally applicable to "Corbyn's QE".]

Comments

[Mike sproul]

Nick

I'm skeptical of the whole idea that the central bank earns profits from printing money. If the Bank could pay 0% on its paper money while getting 5% on its bonds, then we would have a violation of the no free lunch principle, and rival moneys would keep getting introduced until the central banks profits were driven to zero. We might end up in an equilibrium where paper money costs 3% per year to print and handle, the issuer pays 2% per year on the paper money, and the central banks bonds pay 5%--just enough to cover the costs of printing, handling, and interest. If printing and handling costs were 5%, then the Bank would actually earn zero profits, but a misguided observer would look at the 5% bonds and the 0% notes and mistakenly think the Bank is earning 5%.

The kind or profits you talk about would have to be a result of monopoly power, but in a world where rival moneys can be issued by other governments, by private banks, credit card companies, retailers, etc., that monopoly power looks pretty weak.

[Nick Rowe]

Mike: well the Bank of Canada reported $1.2 billion of "net income" in 2014 ($1 billion in 2013) on equity of only $450 million. (If I'm reading the 2014 annual report right. Wish I could find an easy source for multiple years, because I thought it was normally a bit higher than that.) Which looks like an extremely high rate of return on equity.

You can call it "monopoly power" if you like, and there are legal restrictions on notes (Canadian Tire money is barely legal, I have heard). But mostly it's a de facto monopoly due to first mover advantage and network externalities, in my opinion.

[Market Fiscalist]

A CB print up new money worth 0.2% of NGDP. This money gets dispersed into the economy and drives 4% NGDP growth (assuming a fixed 5% cash->NGDP ratio).

It seems it then makes absolutely no difference how the money is dispersed.

- The CB can save it all and buy interest bearing bonds
- the CB can give it all to the govt who spend it how they like (reducing borrowing and taxation or increasing spending)
- The CB can spend it all themselves on luxury goods and wild parties.

As long as monetary policy is entirely a matter of printing and dispersing new money it is all helicopter money !

The way the money is dispersed is entirely a matter of taste and fairness (and perhaps economic efficiency)

(But if monetary policy ever becomes a matter of destroying old money - the CB may wish they had some assets to sell)

[Nick Rowe]

MF: I basically agree. But New Keynesians would say fiscal policy matters for AD at the ZLB.

"The way the money is dispersed is entirely a matter of taste and fairness (and perhaps economic efficiency)" But that point is important (I would maybe add "legitimacy/accountability").

[Market Fiscalist]

"But New Keynesians would say fiscal policy matters for AD at the ZLB."

At the risk of going beyond my level of competence: I think New Keynesians say that because they underestimate the role of money.

They see NGDP as being best controlled by

1) interest rates being used a as a lever to influence current v future spending
2) when this stops working (because rates are zero), they see govt deficits as being an alternative way to directly boost NGDP.

In both cases they see change in the money supply as an effect not a cause of the NGDP changes.

I think that is why they see "helicopter drops" as a separate instrument - as this policy is explicitly to boost NGDP via new money creation (which is of course what monetarists see as driving monetary, fiscal, and so-called unconventional policies like helicopter money anyway).

[Market Fiscalist]

I suppose to be fair fiscal policy could sometimes work to boost NGDP without any changes in the money supply.

[Nick Rowe]

MF: Yep.

[Mike sproul]

Nick
1.2 billion net income from 450 mil equity is impressive, but 1.2 billion earned on total assets of 90 billion, not so much. It certainly doesn't look like the BOC gets a free lunch from printing currency, any more than a private bank would appear to earn profits by borrowing short term at 2% while lending long term at 5%.

[Nick Rowe]

Mike: wandering slightly off-topic: Scottish commercial banks have a limited right of note issue (they were grandfathered). I wonder what determines that limit, and if any of the Scottish banks do anything except issue up to that limit?

[Mike Sproul]

Nick:

Interesting question. I might actually do something useful and try to look that up!

[JKH]

Nick,

I follow this post, but I’m not sure what you are seeking to conclude from it.

I think the post is about the idea that if you can assign current central bank profits (calculated in the usual accrual accounting sense, based on current net interest margin effect) to an equal amount of current transfer payments, you have in effect simulated a helicopter drop as an embedded component of the overall government budget and its consolidated financing source.

And your phrase “money is fungible and we don't observe counterfactual conditionals” observes that such an assignment or connection is indeed quite arbitrary.

But what is your conclusion? I’m guessing it’s something to do with the artificial nature of such an assignment exercise, and that the effective value of a helicopter drop lies not in such an assignment or connection, but the simpler fact that it is the nature of the consolidated financing impact (money financing replacing debt financing) that is the important marginal contribution. Or, that the helicopter drop is an idea that involves an arbitrary connection between a type of expenditure (transfer payment) and a type of financing (money), and that there is no practical way of connecting one rigorously to the other. They will always be two separate decisions in effect, each with its own separate marginal contribution to the nature of government expenditure and government finance respectively.

The transfer payment is something that is a choice unto its own, financed by either bonds attracting commercial bank money or directly with central bank money. But that is a choice between two arbitrary connections.

Similarly, the money loaded into the helicopter could just as easily take the form of cheques on deposits created by government bond issue instead of central bank money.

In a super-idealized case, instead of the central bank purchasing bonds, that central bank money could instead be trucked off to build a bridge, while other bonds are used to finance transfer payments.

The connections between expenditure and finance are arbitrary in this way.

By the way, Fed profits since the financial crisis are in the ballpark of $ 500 billion. Given the size of the accumulated stock of excess reserves, and just how close 25 basis points is to zero, most of that accumulated profit has been the result of the yield curve carry between bond interest revenue and interest expense on reserves. I.e. most of it is not due to the usual contribution from currency seigniorage. So the exercise of associating central bank profits directly with helicopter drop finance is not quite right in this case.

[Ralph Musgrave]

The title of Nick’s article declares the article is all about helicopter money (i.e. simply printing money and handing it out to whoever). But the content of the article is all about printing money and buying assets.

[nickj]

re Scottish banks: is this what you're looking for?
http://www.bankofengland.co.uk/banknotes/Pages/about/scottish_northernireland.aspx

[djb]

I imagine they spent money when they issued the bond

When they buy back the bond they retire the debt

[Nick Rowe]

JKH: "I follow this post, but I’m not sure what you are seeking to conclude from it."

Mostly I was trying to clarify our thinking, rather than seeking to make specific policy conclusions.

I would draw these two policy conclusions:

1. Consider a policy of money-financed government spending/transfer payments/tax cuts. Both people who say "this would always be terrible!", and people who say "this would always be great!" are conceptually confused. They are both wrong. Because, given the NGDP growth rate target (or inflation target) it is inevitable. That which is inevitable can't be either good or bad policy. It just is. If you say it is terrible/great, what you are really saying is you think the NGDP growth rate target is too high/low. Given the NGDP growth target, the only thing that can vary is the timing.

2. Consider a policy "All newly-printed money should immediately be spent on X". This is a policy about timing, so gets around my objection above. But since money is fungible, and we don't observe the counterfactual conditional, this policy is vacuous/unenforceable. (Take the MMT job guarantee policy, where the jobs are money-financed, for just one example; it's vacuous unless you specify the rest of fiscal policy.)

Your own guesses about the conclusions to be drawn sound pretty good to me.

[Nick Rowe]

nickj: interesting. Thanks. The next page on "backing assets" gives the important details.

http://www.bankofengland.co.uk/banknotes/Pages/about/s_ni_roleofbackingassets.aspx

"To back their note issue, authorised banks may use a combination of Bank of England notes, UK coin and funds in an interest bearing bank account at the Bank of England. Bank of England notes held as backing assets may be held at an authorised location or at the Bank of England. Notes held at the Bank may include £1 million notes (Giants) and £100 million notes (Titans), which in physical terms are permanently held at the Bank."

Sounds like 100% reserves.

(Giants, and Titans. Wow!!)

[Biagio Bossone]

Nick:
you may find the following comment of interest
http://www.economonitor.com/blog/2015/09/helicopter-money-central-bank-independence-and-the-unlearned-lesson-from-the-crisis/

[Nick Rowe]

Biago: thanks.

But I don't buy your "Bang per buck" metric of "efficiency". If paper and ink were scarce, "bang per buck" would make sense, but they aren't.

I also don't buy your "direct" vs "indirect" distinction, in a simultaneous system. Consider two accounting identities:

Y=C+I+G+NX makes fiscal policy look direct, and monetary policy indirect.

MV=PY makes monetary policy look direct, and fiscal policy look indirect.

(And Y=C+S+T makes it look like increases in T will increase Y !)

[Left Coast Bernard]

Prof. Rowe,

Why doesn't the Bank of Canada stamp the bonds issued by the Canadian government that it buys "Paid in Full" in red ink and ship them to the Treasury to be shredded?

Aren't the bank's profits from the assets on its balance sheet merely interest payments from the Treasury, which it sends back to the Treasury at the end of the year?

Thanks in advance for your informed and thoughtful reply.

[Jamie]

Nick: “It’s vacuous unless you specify the rest of … policy”

Non-economist from the UK here. I agree with this very strongly. Indeed, my number one observation on blog discussions on macro policy is that no-one starts by unambiguously defining the policy being discussed.

Here are two very different policies:

Central bank decides for itself to create lots of money. Central bank gives this money to the government on the condition that the government spends the money immediately on building a new bridge.

Government decides to build a new bridge. It orders the central bank to create lots of money which it then spends immediately on building a new bridge.

These two policies would have exactly the same macro-economic effect as they both involve creating lots of money and then spending it immediately on building a bridge. However, they are very different policies.

That raises the question of what detailed definition is required for ANY macroeconomic policy in order that discussion is based on a common understanding of the policy. Here is my starter list:

Who instigates the policy and do they need the agreement of anyone else to instigate the policy?
Under what circumstances is the policy instigated?
What is the size of the policy in monetary terms?
What is the full sequence of specific events / actions required by all parties to implement the policy?
What discretion do other parties have in their actions (including compliance / refusal, timing etc)?
Which actions are currently legal and which would require a change in the law?
How would each action work and how long would it take? For example, economists sometimes say that the central bank might print money and write a cheque to each citizen. For that policy, how would the central bank identify each citizen? Who would be included and excluded? Would the bank create its own national identity register? How long would that take? How much would it cost? How would it be kept up to date?
What is the expected outcome of the policy and what uncertainty is involved? For example, if the central bank wrote a cheque to each citizen then what are the assumptions on how many people would spend the money, how many would save it, and how many would use it to pay off debt?
Under what circumstances would the policy be terminated (including successful and unsuccessful termination conditions)?
What are the major risks with the policy and how might they be managed? In particular, how might policy makers and persons in the street view the policy and why might they object? What is the worst case scenario for the policy and how would this be managed e.g. Greece in the Euro?

I am sure I have missed other obvious questions. However, my estimate is that approximately 0% of current economics blog discussions meet even these simple conditions. None of the current UK discussions on Corbynomics in blogs or the wider media meet anything approaching these conditions. That includes Corbyn’s own definition of his policy (not his fault as he’s not been given time to think); the definitions of all of the economists critiquing Corbynomics; and the definitions of all the economists critiquing the other economists critiquing Corbynomics.

What would you include in a mandatory policy definition list? Why do you think that there is no current list of this type?

[Nick Rowe]

Bernard: good question. I answered it in an old post 6 years ago!

The most important reason, in my view, is that central banks sometimes need to print money (to stop inflation falling), and sometimes need to burn money (to stop inflation rising). And if they didn't have any assets, they wouldn't be able to buy back money and burn it. It would be a one-way street, which would only work OK in an economy where everything is expanding smoothly.

Mike Sproul (who commented above) would disagree with me.

Jamie: you missed the most important thing about policy. Those are all policy *actions*, and we need to talk about policy *rules*. Because the effect of any policy action depends on whether or not it was expected, which depends on the whole pattern or rule of policy, and how people interpret that policy action. See, for example, this old post.

[Jussi]

"And if they didn't have any assets, they wouldn't be able to buy back money and burn it."

How would they get money into economy in the first place if not against assets? I know Nick has said they can just give it away - fair point but is that an efficient way? If given away through the government why not do it against government paper just to keep the books balanced?

[Nick Rowe]

Jussi: I think that's usually correct. Here is an old post, (including links to other old posts) where I argue that that there might be a case for a central bank trashing its own balance sheet, sort of like burning its boats. Here's Citi making a similar argument.

[Jamie]

Nick: “Those are all policy *actions*, and we need to talk about policy *rules*. Because the effect of any policy action depends on whether or not it was expected”

The expectations set by any policy proposal depend on the credibility of the actions required to implement that policy. My earlier comment was triggered by your use of the word ‘vacuous’ to describe a policy proposal and my observation that the current discussions in the UK about Corbynomics are also vacuous. Corbynomics has not been defined. That means that the expectations of different economists of the impact of Corbynomics range from that it represents business as usual though to that it will be a threat to civilisation as we know it. The entire debate is ridiculous, and it is ridiculous precisely because the lack of definition results in vastly different expectations on the effect of the policy.

When market monetarists say that central banks can always accelerate the economy as they can, in extremis, buy all the assets in the economy, they are saying that the existence of such an ability can set expectations. However, my assumption, and that of any other non-economists that I have ever raised the subject with, is that such actions by the central bank are probably not legal, so no expectations are set. You would have to show that these actions were legal before expectations would change. Even then, my expectation is that any announcement that such actions were legal would result in calls from pretty much everyone outside central banks for such actions to be made illegal.

When economists call for the abolition of cash so that they can introduce negative interest rates, my expectation is that negative rates on savings would cause riots in the streets. Most people blame banks, at least in part, for the financial crisis. The idea that the ‘solution’ to the crisis might involve confiscation of nominal savings by these same banks is not credible. If banks made loans on negative interest rates then my expectation is that everyone would seek to take out as many loans as possible to get the income streams from the negative interest charges, and that no-one would bother to use loans for productive investments. What would be the point?

When economists say that central banks might print money and write a cheque to everyone in the population they are trying to set expectations that such a policy would boost the economy. However, if they can’t answer the basic operational question of how the central bank would know the identity of each person who would receive a cheque then no such expectations will be set.

You used the word ‘vacuous’ earlier to describe a badly define policy. I agree. That means there must be a basic minimum definition required to avoid such vacuous policies and the uncontrolled expectations that might arise from misinterpreting such policies. What do you think is the minimum standard for the non-vacuous definition of a policy which will meet my aim of unambiguous actions and your aim of unambiguous expectations?

[Jussi]

Yep, I think I agree. Yet one can get exactly the same effect using a fiscal transfer, issue bonds to fund it and let the central bank buy the bonds. Consolidated this is the same as the central bank buying junk (sold by the guys getting the transfer in the first scenario) in the first place.

So I'm not sure trashing the balance sheet is needed (but can be done without much harm).

[Matt]

Let's do a decomposition.

Suppose that a central bank invests in short-term nominal bills, and borrows (partly) in the form of zero-interest paper currency, earning profits on the spread. When it raises the interest rate, profits are higher (assuming the interest elasticity of money demand is less than 1, which is certainly true in the low-rate region). These profits are distributed somewhere -- though it's hard to say where they go and when that happens, because of all the issues you've raised about fungibility.

Now let's move to an alternate world, in which it is technologically possible to pay interest and levy taxes on paper currency. Suppose that in Alternate World, the standard procedure of the central bank is to adjust interest rates by changing the interest rate paid on reserves and paper currency (which it equalizes between them).

Alternate World could replicate the effects of our more conventional monetary policy by:

(1) Adjusting interest rates its usual way, by moving the rate paid on currency and reserves in tandem, plus

(2) Levying a tax on currency, which it varies in line with the interest rate to replicate the spread that would exist if it wasn't paying interest on currency; then spending the proceeds of this tax however they would have been spent in the usual world.

Now, to me it seems obvious that of this decomposition, (1) is vastly the bigger deal. (2) is just a fairly tiny tax on a tiny market, which raises a tiny amount of funds and then spends them; in the course of usual monetary policy, it will vary by only 0.1% or 0.2% of GDP.

There is a very small time-varying second-order distortion induced here, and then some somewhat larger but still mild first-order transfer effects depending on who pays the cash tax relative to who receives the proceeds. We regularly ignore or downplay changes of this magnitude in tax policy all the time. Why should we pay much attention to it here? I think the answer is that we shouldn't, and that only the conventional bundling of (1) and (2) together into a single policy lever fools us into caring about (2).

[Nick Rowe]

Matt: suppose the central bank pays (say) 1% interest on currency, and also levies a tax of 1% on currency. I say it's a wash; exactly like a world with 0% interest and no tax on currency.

[Matt]

"Matt: suppose the central bank pays (say) 1% interest on currency, and also levies a tax of 1% on currency. I say it's a wash; exactly like a world with 0% interest and no tax on currency."

Sorry, I should have been clearer. I'm assuming that there are excess reserves, so that the interest rate paid on reserves determines the interbank interest rate, and that this rate determines the short rate throughout the economy.

Now, in our current world paper currency doesn't pay interest, so bundled together with adjustments in the short rate (modeled here as being driven by IOR), there's a change in the spread charged on paper currency, and consequently in the profits earned by the central bank. This is the sense in which "monetary policy" (understood as some kind of state-contingent policy rule for the short rate) currently affects seigniorage revenue.

But there's no conceptual reason why the two should be bundled together. If it was possible to make paper currency earn interest, and standard procedure was to pay it the same interest paid to reserves, then the two would be completely separate. The central bank would by default earn zero seigniorage profits, regardless of how it decided to set the short rate. Then maybe some other completely separate Cash Tax Agency would tax holdings of paper currency and spend the proceeds on something.

My view is that in this unbundled world, we wouldn't really care what the Cash Tax Agency did. It's just too small an actor, and we would see no reason to elevate it over other obscure agencies of the government.

By extension, in our present-day world, where the Cash Tax Agency is effectively part of the central bank, we probably care too much about this dimension of policy, because we can't mentally disentangle it from the part that really matters (namely setting the short rate according to some rule, which determines the short rate throughout the economy and its reaction to events).

[Interesting side note: in Woodford's 1995 paper of the fiscal theory of the price level, he demonstrated that in an FTPL world, the fiscal consequences of seigniorage revenue would be the only way in which the decomposition of government liabilities between bonds and money affected the price level. This is pretty much diametrically opposite my position here, which is that seigniorage revenue is a nearly irrelevant little speck in the ocean of the integrated government budget, and that the true power of monetary policy comes from other sources.]

[Nick Rowe]

Matt: "This is pretty much diametrically opposite my position here, which is that seigniorage revenue is a nearly irrelevant little speck in the ocean of the integrated government budget, and that the true power of monetary policy comes from other sources."

I agree, unless we are talking about Zimbabwean rates of printing money. My ballpark 0.2% of GDP is peanuts. It gets swamped by minor tax changes.

Yep, conceptually we can imagine a zero-profit (competitive?) central bank, with a quite separate government tax on currency, just like it taxes smokes, drinks, and gas. But varying the tax on currency would have monetary consequences, just like in Gessel. (In fact, the rate of interest in Woodfordian New Keynesian modesl should be interpreted as a negative tax on central bank issued money.)

[Matt]

Nick: "But varying the tax on currency would have monetary consequences, just like in Gessel."

Not necessarily - I'm thinking about a tax on paper currency that isn't applied to reserves.

In general, if paper currency and reserves are freely convertible into each other, then the following equilibrium condition holds:

(the liquidity premium on paper currency plus interest on paper currency net of taxes)
= (the liquidity premium on reserves plus interest on reserves net of taxes)
= (market short rate)

Locally, does adjusting the net interest rate on paper currency make any difference to the market short rate (holding the total quantity of currency + reserves constant, and also holding the net interest rate on reserves constant)? Yes, if and only if there is locally a liquidity premium on reserves. Paying higher net interest on paper currency encourages people to shift their holdings from reserves to paper currency until some combination of the former liquidity premium rising and the latter liquidity premium falling restores the equilibrium condition. The market short rate only changes if the liquidity premium on reserves changes, which (locally) will only happen if the market isn't saturated with base money already.

So pre-2008, if the Fed held base money constant while suddenly introducing interest on paper currency, that would push up the market short rate, due to rising demand for paper currency. Post-2008, on the other hand, essentially nothing will happen except a change in the composition of base money: the higher demand for paper currency will just come out of excess reserves, but the short rate will stay at the interest rate on reserves.

Right now, Switzerland is implementing a negative short market rate by paying negative net interest on reserves. If everyone in Switzerland collectively decided that they would rather earn 0% interest on currency, then they would withdraw the reserves in exchange for currency up until reserves became so scarce that a liquidity premium reappeared on them. At that point, as per the logic above, it would be very nice to be able to impose a tax on currency, a la Gessel. (If things never reached that point, because the public didn't want to hold that much currency, the government ideally might still decide to impose a tax on currency to equate its pecuniary return with the return on reserves, and avoid the negative seigniorage that comes from paying an above-market rate of return on currency.)

In the previous comments, I was assuming that we were in some form of the post-2008 regime, with excess reserves such that the central bank can easily manipulate the short rate regardless of the net interest paid on money. In that kind of world, you can imagine handing over the decision about the tax on money to some kind of agency separate from the central bank. (Though maybe there's not much point in thinking about this - as a hardcore Woodford-ite I'm just personally fond of exercises that separate various "fiscal" policies from what I view as the central role of monetary policy, which is setting the short rate.) In the Swiss world, on the other hand, your point is very well taken - namely that there is a crucial interaction between the pecuniary return paid on currency and what the central bank can achieve.