How can we know if government spending is money-financed?

I have a proposal that would improve aggregate demand management. I call my proposal "money-financed government expenditure on sacrificing goats".

I propose that every time the government sacrifices one goat, the Bank of Canada is required by law to increase the base money supply permanently by $1 billion, relative to what it would otherwise have been. To say the same thing another way, the Bank of Canada should be required to pay the government for all goat sacrifices, at a price of $1 billion each.

If animal rights people object to my proposal, I am quite willing to modify it so that the "goat sacrifice" is purely symbolic. No (real) goats are involved at all. The government just announces that one "goat" has been "sacrificed". But the price is still $1 billion per "goat sacrificed". The government can spend that extra $1 billion seigniorage whenever it likes, on whatever it likes.

This is how an increase in G (money-financed goat sacrifices) would increase aggregate demand:

The money base is mostly currency, and the demand for currency depends on Nominal GDP. If people hold (say) 5% of annual nominal income in currency, a permanent increase in the stock of currency of $1 billion would require a permanent increase of NGDP by $20 billion. Even if Canada were temporarily at the Zero Lower Bound on nominal interest rates, the commitment by the Bank of Canada to increase future NGDP by $20 billion would increase expected future NGDP, which would increase current aggregate demand, which would increase current NGDP.

If my proposal were implemented, the Zero Lower Bound would cease to be a problem. The government could sacrifice enough goats to lift the economy off the ZLB.

But there is one big problem with my proposal: how could my proposed law be enforced? The lawyers would have a lovely time arguing about that counterfactual conditional clause in the law. We know, roughly, what the monetary base is now. But how could we ever know what the monetary base would have been if the government had not sacrificed one goat in the past? If we can't answer that counterfactual conditional question, we can't figure out if the Bank of Canada did in fact pay $1 billion to finance the goat sacrifice. How can we ensure that goat sacrifices are in fact money-financed?

We would need an agreed-on economic model. We could require the Bank of Canada to use that model to issue a forecast for all future levels of the money base, conditional on number of goats sacrificed. And we would then check to see that those forecasts were reasonable, and were all $1 billion apart for every additional goat sacrificed, at all future times. Then we could check to see if the actual money base matched the Bank of Canada's previous forecasts, given the actual number of goats sacrificed in the past. But if the model was wrong, or if something else unexpected happened, so the Bank of Canada's forecasts turned out to be wrong, we could never tell whether the Bank had made a bad forecast, or was violating the new law. Hmmm.

Maybe it would be much simpler just to require the Bank of Canada to raise its NGDP target path by $20 billion every time the government sacrificed a goat. And then check to see if actual NGDP followed the new target path. The effect on the monetary base would be much the same, and the law would be much easier to enforce. And laws that are easier to enforce are much more credible.

I see that Jordi Gali has a proposal (HT Mark Thoma) that is very similar to mine. He too wants goat sacrifices to be money-financed. The only difference is that Gali wants the government to sacrifice real goats, and to pay the going market price for goats.

This sounds a bit wasteful to me. Why sacrifice real goats, when sacrificing symbolic goats does the job just as well? And even if sacrificing $1 billion worth of real goats has (say) twice the effect on aggregate demand of sacrificing $1 billion worth of symbolic goats, we could easily sacrifice $2 billion worth of symbolic goats instead.

But the biggest problem with Gali's proposal is that it has exactly the same problem as mine. How would we enforce his proposed law that goat sacrifices be money-financed?

How would we ever know whether an increase in government spending was or was not money-financed? Only if the central bank announces a permanently higher target path for NGDP, or a permanently higher target path for the price level (or a temporarily higher inflation target), and if it hits that new target path, would we know that the increase in G was in fact money-financed. But if the central bank made that announcement, and if it were a credible announcement, we wouldn't need the increase in G.

Money-financed sacrifices of real goats will not really be money-financed if we cannot enforce the law making them money-financed. And if we can enforce the law making them money-financed, we don't need a real goat.

Comments

[Mustsign topost]

this line of thinking is what made enron successful

[Frances Coppola]

Isn't sacrificing a virtual goat the same as issuing a bond? I mean, if the government actually hasn't got a goat to sacrifice, it is really issuing a sacrifice IOU.

[Nick Rowe]

Frances C: It's issuing a $1 billion bond, and the Bank of Canada printing $1 billion in currency to buy that bond, and the stock of currency staying permanently $1 billion higher than it otherwise would have been, so the government never has to pay interest on that bond or redeem it. (Or, equivalently, it does pay interest to the Bank of Canada, but the Bank of Canada immediately returns that interest payment to the government, which owns the Bank of Canada, and gets all the Bank of Canada's profits.)

[Frances Coppola]

Exactly, that's what I thought. So why bother with the goat at all, then?

[Nick Rowe]

Frances: yep. The goat isn't needed. It was just a literary device to help me make my point. A money-financed increase in Government spending is equivalent to an increase in government spending accompanied by an announced permanent increase in the NGDP level-path target (or price-level path target).

I'm not sure if my literary device (is that the right word?) worked or not. But it was hard to explain the isomorphism (and is that the right word?) between money-financed increases in G and increases in G accompanied by an announced new NGDP target. And getting across the idea that it was the announced new NGDP (or price level) target that was doing most of the work, and that without that announcement the increase in G would not in fact be money-financed.

[Frances Coppola]

So what you seem to be saying is it doesn't matter a jot whether government finances itself with money or bonds, what matters is the nominal anchor. I think I agree with that. And I think it is going to be an increasingly important insight.

I read Gali's piece earlier, by the way. He still seems to be thinking of money financing as an emergency crisis response, which to me entirely misses the point (and does carry a risk of hyperinflation, since the nominal anchor can be lost in a crisis). Noah Smith on monetizing Japan's debt is far more interesting - he seems to be in agreement with Adair Turner that monetization is a) inevitable b) unlikely to be a major disaster in an ageing society where maintaining positive NGDP at all is the challenge.

[Nick Rowe]

Frances: "So what you seem to be saying is it doesn't matter a jot whether government finances itself with money or bonds, what matters is the nominal anchor."

Yes, but I would say you can't even tell whether the government finances a change in government spending with money or bonds, unless you look at what happens to the nominal anchor when it changes government spending.

Suppose, for example, that the central bank followed a Friedmanite k% rule, only for the money base MB. (Not a good policy, of course). So the central bank makes MB grow at (say) 5% per year. If an increase in G was matched by an increase in k from 5% to (say) 6%, then we would be able to say that the increase in G was money financed.

And if it was a 2% inflation target instead, an increase in G would only be money-financed if it raised the inflation target to 3% at the same time. Etc.

We always finance *some* government spending by printing money. (It's roughly 0.25% of GDP, with 5% NGDP growth and 5% MB/NGDP ratio.) But if you want changes in G to be money-financed, the nominal anchor must change at the same time.

BTW, did you see my post on fractional reserves? It was partly inspired by your ultra-liquidity post.

[Lord]

If you want this to be credible, you need to announce the penalty for not attaining the result and make it high enough for them to care.

[Frances Coppola]

No, I missed that! will read

[Christiaan Hofman]

But who says the government will actually sacrifice goats (or whatever) when they need to? My experience is that they will actually refrain from sacrificing goats exactly when they should. I base this expectation on Europe. There they don't sacrifice goats, they do the exact opposite: they bring back goats to life, and force the central bank, by law, to supply the corresponding negative billions. (As a manner of speaking.) So why would you think otherwise? They're politicians, not economists!

[Frances Coppola]

Christiaan, why on earth do goats need to be sacrificed at all?

[Nick Rowe]

Lord: New Zealand did that, IIRC. The Governor's salary got cut if inflation deviated from target. In Canada we didn't bother, and they still kept to the target.

Christiaan: Oh dear. Yes, you are right. At least, for the Eurozone.

[Philippe]

" Even if Canada were temporarily at the Zero Lower Bound on nominal interest rates, the Bank of Canada to increase future NGDP by $20 billion would increase expected future NGDP, which would increase current aggregate demand, which would increase current NGDP"

so at the ZLB your theory seems to depend entirely on expectations. The CB says "NGDP will be higher in the future!". Why should people believe this? Does the CB have any means by which it can directly increase NGDP in the present, which doesn't depend entirely on people believing its promises about the future?

[Frances Coppola]

Nick - read your fractional reserves post now. Welcome to my world! My only slight quibble is that I think ultra-liquidity and "excess capital" is forcing R down to Rm, rather than the Bank forcing Rm up to R. Positive interest rate on bank reserves prevents R falling below zero. But the effect is the same - indifference between money and other assets because R=Rm for all values of Rm. A permanent liquidity trap, if you like.

[Nick Rowe]

Frances C: I thought it might be up your street!

You are right. R falling towards Rm is more in line with what's been happening recently than my thought-experiment of Rm rising towards R. But the results should be about the same.

Philippe: that's not really the main point of the post. You are caught up on those concrete steppes again!

If the central bank increases Ms, then Ms > Md, so people want to get rid of the excess money, by spending it or lending it. And that increases the demand for goods, which increases either Y and/or P, and so NGDP increases. But the concrete steppes effect this very instant is peanuts compared to expectations. How much income I earn this second is almost irrelevant to my actions. The number of seconds in my future life is much bigger than one. My expected income tomorrow, and next week, and next month, and next year, and next decade, matter far more. The present is a very very short period of time.

[Min]

"Only if the central bank announces a permanently higher target path for NGDP, or a permanently higher target path for the price level (or a temporarily higher inflation target), and if it hits that new target path, would we know that the increase in G was in fact money-financed."

Sorry, no we wouldn't. Too many assumptions.

[Nick Rowe]

Min: picky!

[Philippe]

"My expected income tomorrow, and next week, and next month, and next year, and next decade, matter far more."

but CB promises about the future have to be credible to affect expectations and behaviour. The CB could promise that NGDP will equal X in a year's time, but if it has no way to directly bring that about ('concrete steppes'), then its promise will not be credible.

If the only direct mechanism at the ZLB is the 'excess money'/ 'hot potato effect', then it is the effectiveness of this mechanism which will determine expectations.

[Nick Rowe]

Philippe; do you expect, and do you think 100% of the population expects, with 100% certainty, the ZLB to be permanent/forever? I don't.

[Frances Coppola]

Nick,

Yes, either way yield curves flatten. This is death to commercial banks, of course. This is why we are seeing creeping nationalization of maturity transformation, mostly via off-balance sheet devices such as sovereign guarantees and GSE-type structures. Governments becoming banks. That's your scenario 2, isn't it?

[Nick Rowe]

Frances: the death of commercial banks comes from increasing competition by the central bank reducing the gap between R and its Rm. Which is like a flattening of the yield curve. (Though perhaps we should put "liquidity" rather than "term to maturity" on the horizontal axis of that yield curve).

Whether that results in my scenario 1 or scenario 2 depends on what the central bank/government chooses to do. If it buys lots of assets (QE++++) it's scenario 1. If it helicopters money it's scenario 2.

[Nick Rowe]

In other words Frances: yes! (We are saying the same thing in different ways, and I always have to translate you into macro-speak!)

[Philippe]

forever is a pretty long time. I don't think the economy will remain in the doldrums forever, but the question is what the CB can actually do to change the situation.

[Roger Sparks]

It seems to me that the counterfactual to a "money-financed" government spending increase is a "debt-financed" government spending increase.

If this is correct, then "money-financed" should mean money supplied to government from the Central Bank where the CB holds the government debt. On the other hand, "debt-financed" would mean that the increased money spent by government would first be borrowed from the private sector.

Given these two choices of money for government spending, it would be easy follow the money to source to see if the goat-sacrifice law has been followed.

[Nick Rowe]

Philippe: ??? I have already answered that question!

One last time. If you think the ZLB ends in 20XX and the central bank can raise NGDP in 20XX, and NGDP in 2014 depends on expected NGDP in 20XX, the central bank in 2014 can make a commitment to a higher level of NGDP in 20XX, and that will increase NGDP in 2014.

What determines the price of a strip bond that matures in 20XX? The government's commitment to pay the owner $1,000 in 20XX. Change that $1,000 to a different number, and that affects the price of the bond today, even though the government can do "nothing" today.

[Nick Rowe]

Roger: Suppose the central bank increases the money base by $1 billion today, to pay for the goat. Then reduces the money base by $1 billion tomorrow. The only thing the central bank has paid for is one day's interest on $1 billion.

[Philippe]

"the central bank in 2014 can make a commitment to a higher level of NGDP in 20XX"

The central bank can only make a credible, believable commitment to a particular level of NGDP in 20XX if it actually has some means to directly bring that about (concrete steppes). So the effectiveness, or perceived effectiveness, of those means (such as the 'hot potato effect') is crucial.

Saying that "NGDP will probably be higher at some point in the future" is not enough in itself to change behaviour today. There's too much uncertainty involved.

[Nick Rowe]

Philippe: "So the effectiveness, or perceived effectiveness, of those means (such as the 'hot potato effect') is crucial."

Of course!

If apple producers increase the supply of apples, the price of apples goes down.

If money producers increase the supply of money, the price of money goes down.

The price of money is the reciprocal of the price of apples (and bananas and carrots).

[Frances Coppola]

Hehe. And I always have to translate you into finance-speak!

[Too Much Fed]

Frances Coppola, I have some questions about the post:

http://worthwhile.typepad.com/worthwhile_canadian_initi/2014/09/fractional-reserves-and-the-optimum-quantity-of-money.html

Would you be willing to answer them? Thanks!

[Nick Rowe]

Philippe: "Saying that "NGDP will probably be higher at some point in the future" is not enough in itself to change behaviour today. There's too much uncertainty involved."

I don't think I will buy a fire extinguisher or fire insurance. I'm not 100% certain that my house will catch fire.

[Too Much Fed]

Let's say RGDP = 1%, price inflation = 2%, and productivity = 1.5%. Now change to RGDP = 1%, price inflation = 4%, and productivity = 1.5%.

What happens to hours worked in both cases?

[Frances Coppola]

TMF, yes of course, fire away.

[Frances Coppola]

TMF, price inflation by itself makes no difference to hours worked.

[Roger Sparks]

Nick,

I don't understand this comment:

"Roger: Suppose the central bank increases the money base by $1 billion today, to pay for the goat. Then reduces the money base by $1 billion tomorrow. The only thing the central bank has paid for is one day's interest on $1 billion."

I guess the CB could call the goat an asset and make that entry on their books. The CB would call the money they paid a liability. The public sector would have $1 billion and one less goat.

I guess the CB could tax the money back tomorrow. After paying the tax, the public would have no money and no goat.

You can see that I don't understand the comment.

[Philippe]

"If apple producers increase the supply of apples, the price of apples goes down"

That's only true if all else remains equal.

"I don't think I will buy a fire extinguisher or fire insurance. I'm not 100% certain that my house will catch fire."

Buying a fire extinguisher or fire insurance is about insuring yourself against risk, i.e. trying to reduce risk in the face of uncertainty. Borrowing or spending savings to increase your level of investment, or to increase your level of consumption, simply in the hope that your income will be higher in the future, involves taking on risk in the face of uncertainty.

[Nick Rowe]

Roger: if I lend you $100 forever, at 0% interest, you are $100 richer. If I lend you $100 for one year, at 0% interest, when other interest rates are 10%, you are $10 richer.

Philippe: Announcing an NGDP target reduces the uncertainty over future NGDP.

[Roger Sparks]

Thanks Nick, I think I see now.

"Roger: if I lend you $100 forever, at 0% interest, you are $100 richer. If I lend you $100 for one year, at 0% interest, when other interest rates are 10%, you are $10 richer."

When the CB is the lender, the $100 loan causes internal movement within the private sector. When the loan is paid back at 0% interest, the $10 in interest the handler receives is a pure transfer between agents within the private sector.

Thanks, that helps.

[Philippe]

"Announcing an NGDP target reduces the uncertainty over future NGDP"

only if the CB actually has the means to directly bring it about (concrete steppes). So the 'excess money' or 'hot potato effect' is crucial at the ZLB - it is a means by which the CB could potentially directly bring about higher NGDP, and it is thus a potential basis for making credible promises about future NGDP.

However, increasing the supply of base money only makes the price of money go down if all else remains equal. So the 'hot potato effect' might not work, if all else does not remain equal. This in turn could undermine the credibility of CB promises about future NGDP.

[Nick Rowe]

Philippe: the driver of a car can keep close to the speed limit despite other things (hills, headwinds) not staying equal. He adjusts the gas pedal accordingly. And can certainly do better than a driver who doesn't even try to keep close to the speed limit.

(And while cars have finite horsepower, a central bank rarely runs out of paper and ink.)

[Miami Vice]

If people made investment or consumption spending decisions based on expected future ngdp, wouldn't there be a need for a ngdp futures market?

There isn't so, i don't think they do.

It seems to me this expectations thing suffers fallacy of division and free rider problems. or not?

[Philippe]

you're just saying that the CB increasing the monetary base when at the ZLB is like pushing on the gas pedal, when it's possible it could be more like pushing on a string.

[Nick Rowe]

Philippe: now you are starting again from the beginning. You are back at the ZLB. We've been here twice already.

Miami: suppose nominal interest rates are 5%. And you have just decided how much to invest and consume. Then your crystal ball tells you that NGDP will drop by 50% next year. You know there will either be 50% deflation, or a massive recession with 50% (approx) unemployment. Or some mix of the two. Would you revise your plans? I certainly would. I might be one of the 50% unemployed. I won't be able to sell the extra output from my investment. Or I will pay 55% real interest rates. And then you hear that everyone else's crystal ball is telling them the same thing, and they are revising their plans too.

[Philippe]

"You are back at the ZLB"

I'm assuming it is possible to be at the ZLB, given that is supposedly where we are at.

I'm not sure in our discussion where we were supposed to have left it.

Do you mean that if the CB originally had a NGDP target, we would never have ended up at the ZLB in the first place?

The concrete steppe the CB can take when not at the ZLB is to reduce the short term interest rate. When this no longer works, it's at the ZLB. Then the concrete steppe it can take is to increase the monetary base further. If the CB's promises about future NGDP are to be credible, these concrete steppes have to be effective. At the ZLB, they may not be.

[Nick Rowe]

Philippe: no. I meant our argument is going round in circles. Every time I explain something, you circle back around and start again at the beginning. It has happened twice now. Forget it.

[Philippe]

Oh, I thought I was disagreeing with your explanations.

[Miami Vice]

If I'm not mistaken you've assumed risk averse agents in your model (zero lower bound). If national nominal income grows 5% my income doesn't necessarily grow by 5%. Why, if you can free ride and capture some growth without taking any undue risk, would risk averse agents take risks in excess of internal projections of demand growth? Wouldn't you need to assume capacity is fully utilized or that inventories (of outputs and inputs) are insufficient to buffer against a positive demand shock combined with easy substitution away from your product or service. Or, if real(or just specific to your company) output growth is stagnant, its not possible to hedge against inflation (or benefit from inflation) or benefit from the rise in rates caused by the the expectation of 5% ngdp growth. How much can real interest savings be? How large is the supply of marginal investment spending that will be coaxed into existence?

Sure crystal balls, self fulfilling prophecies and ngdp changes that are an order of magnitude different would have wildly different reactions and outcomes. Assuming away the possibility that ngdp misses the declared target reduces the discussion to being absurd IMO. Don't you have to assume rational agents assume that 5% national income growth means personal income growth of 5%.
How's this for absurd? How much more (less) food would you eat today if I told you food would be more(less) expensive in the future. Food cannot be stored.

[Nick Rowe]

Miami: "If national nominal income grows 5% my income doesn't necessarily grow by 5%."

And if NGDP grows by 0%, my nominal income doesn't necessarily grow by 0%.

I face risk either way. But the risk is higher if the central bank announces no target for NGDP at all, and just lets everyone guess what it will do in future.

[Miami Vice]

"If wishes were horses, I'd be eating wish-meat every night."
-Moe Szyslak

[Nick Rowe]

The central bank can print money. That's not an idle wish. It can make that money as scarce, or as plentiful, as it likes. NGDP is the value of output, measured in terms of that money.

[Miami Vice]

I know, I was the one who brought up the fallacy of division. Thanks for repeating.

You're the one who thinks he can make a horse drink. I'm saying the Horse will drink when it's thirsty.

Would Ben Bernanke have been able to refinance his house if Janet Yellen anounced a 5% ngdp target? Because that's what I hear you saying.

[Miami Vice]

The Q1 2008 survey of professional forecasters was estimating ngdp would grow by 4.8% in Q4 2008 and 5.6% in Q1 2009 (5% in 2009). Q4 2007 estimate for Q4 2008 ngdp was 5.1%. The 2008 Q2 survey estimate for Q4 2008 ngdp was for 4.5%.
(annualized %)

How many quarters lag does it take for monetary policy to gain traction? The base began expanding in the summer of 2008. The same time as ngdp forecasts for Q4 2008 were revised lower to 3.1%. Actual 2008 ngdp was -0.9%. What actions would you have expected the fed to have taken to prevent ngdp from falling in Q4 2008? They took action once they realized growth was going to come in lower than had been previously forecast.

Yes I know they weren't targeting ngdp. I doubt that means they aren't/weren't concerned with its growth path. You're probably going to write that if they had been targeting ngdp then this all would be a non issue since they would have hit their target then because forecasts would be more accurate/credible. crystal ball

http://www.philadelphiafed.org/research-and-data/real-time-center/survey-of-professional-forecasters/

[Miami Vice]

How Do Forecasts respond to Changes in Monetary Policy? By Laurence Ball and Dean Croushore

http://www.philadelphiafed.org/research-and-data/publications/business-review/2001/q4/brq401dc.pdf

[Miami Vice]

Forecast Disagrement in the Survey of Professional Forecasters by Keith Sill

http://www.philadelphiafed.org/research-and-data/real-time-center/SPF-forecast-dispersion_Q3-2010.jpg

[Miami Vice]

65% of the time the Bank of Canada undershoots their 2% inflation target since 1993. Except 2009 the Bank of Canada would have been within the range of respectability, analogous to the inflation target range of 1-3%. Had the target been ngdp they would have undershot the target only 45% of the time 1999-2011. Why are you so confident ngdp targeting would have so wildly transformed monetary policy to the point of guaranteeing 5% ngdp growth?

[Miami Vice]

analogous to the inflation target range of 1-3%, had they been targeting ngdp.

[Nick Rowe]

Miami: On average, the Bank of Canada has hit its 2% inflation target almost exactly. See my old posts here and here for the data, and for why I switched to supporting NGDP level path targeting.

You need to understand the distinction between growth rate and level-path targeting. Under level-path targeting, the central bank makes up for past mistakes. So if it targets 5%, but misses and actually hits 4%, it targets 6% the following year (or 5.5% for 2 years). This ensures that drops in NGDP will be temporary, not permanent.

[tim]

Nick when you come up with a stable money demand function we will buy all this stuff. This is the very old monetarism. With its pros and its vast fallacies. Try harder

[Nick Rowe]

tim: you are missing the point. If the central bank announces a permanently higher NGDP level-path target when the government sacrifices a goat, does it really matter whether or not MB rises in exact proportion?

[tim]

Can you please tell me why on earth all central banks are targeting inflation by controlling interest rates and NOT targeting NGDP by controlling money supply? This is a "positive" question not a normative one. Tell me your opinion. In and out the ZLB

[Nick Rowe]

tim: OK. But I'm going to break your question into two/three parts:

1. Why IT vs NGDPLPT?

In the late 1980's the Bank of Canada was all set up to target NGDP (I can't remember if it was NGDP growth rate or NGDP level path). They had the model all ready. The new Governor came in and said that people would understand inflation but not NGDP. And the (forgotten the name, but small business owners group) said they wanted to know what inflation rate the Bank was aiming for, to help them negotiate wages with the unions. So the Bank adopted Inflation Targeting.

Historical accident, in other words. And the rest of the world followed. (Sorry Kiwis, for leaving the RBNZ out of this story, but your story was a little different, and you weren't really doing IT right originally anyway ;-) .)

2. Why use i as an instrument rather than MB?

I don't know. I'm not sure it really matters much. My guess is that it's because central bankers were historically bankers, and commercial banks set i (plus terms for loans). So they always thought in terms of i. Plus the influence of Wicksell maybe?

3. At the ZLB they couldn't cut i, so they had to increase MB.

[Too Much Fed]

Frances Coppola said: "TMF, price inflation by itself makes no difference to hours worked."

Exactly. And in my example for both cases, hours worked fall. So targeting NGDP may not help and could even harm the labor market if prices rise and RGDP falls.

[Too Much Fed]

http://worthwhile.typepad.com/worthwhile_canadian_initi/2014/09/fractional-reserves-and-the-optimum-quantity-of-money.html

Frances, I don't see a capital requirement there. I believe that would make a difference.

Sorry about the delay.