Cantillon effects and non-SUPER-neutrality = does fiscal policy matter?

Scott Sumner and Bill Woolsey have been fighting valiantly against the Austrians. The fight is about "Cantillon effects" -- non-neutralities of money that are supposed to arise not from the increase in the money supply itself but from where exactly that new money enters the economy.

Sometimes you get a clearer answer to a question if you change the question. That's what I'm going to do here.

Let's stop asking whether the effect of a change in the level of the money supply depends on where that new stock of money enters the economy. Instead, let's assume the money supply is growing at a constant rate, and ask whether it matters where that constant flow of new money enters the economy. We are simply redirecting that constant flow of new money, not changing the stock and redirecting it at the same time. It's easier to keep our heads straight that way.

Suppose there is zero real growth, base money is 10% of Nominal GDP, and base money is growing at 10% per year. So there's a steady 10% annual inflation, and a steady flow of new money equal to 1% of GDP.

1% of GDP isn't peanuts. But it's not very big either. And I've stacked my assumptions to make it bigger than it would be in most normal economies. If NGDP is growing at 5% per year, and base money is 5% of NGDP, the flow of new money would be 0.25% of NGDP.

I will assume the government owns the central bank. Any profit the central bank gets from printing money is government revenue. We can consolidate the government's and central bank's balance sheets, so any interest the government pays on bonds owned by the central bank is a wash. The government is paying that interest to itself. The central bank earns profits of 1% of GDP, and gives those profits to the government, which decides how to spend them.

Let's compare 5 different ways the same flow of new money could enter the economy.

1. Interest on money. The baseline scenario is that all new money is paid as interest on existing money. So every year people earn 10% interest on their money, which exactly offsets the 10% depreciation through inflation. It's a wash. It's really just like a stock-split.

2. Helicopter money. Scenario 2 is that all new money is paid as transfers to the population.  It's no different from scenario 1, except that the opportunity cost of holding base money is higher (because it depreciates through inflation but doesn't pay interest), so the real stock of money would be smaller. (Plus some individuals might get lucky and others unlucky depending on whether the helicopter does or does not fly over them.)

2a. Cut taxes. The government uses the new money to cut taxes. Same as scenario 2. A tax is a negative transfer payment. So a tax cut is a transfer increase.

3. Debt reduction. Scenario 3 is that all new money is handed over to the government, which uses it to retire government bonds. Scenario 3 is the same as an open market purchase of bonds by the central bank. Scenario 3 is identical to scenario 2, plus a tax of 1% of GDP used to pay down the debt. Does it matter if the government increases taxes to pay down debt?

4. Government spending. Scenario 4 is that all new money is handed over to the government, which uses it to buy goods. Scenario 4 is identical to scenario 2, plus a tax of 1% of GDP used to buy goods. Does it matter if the government increases taxes and spending? Obviously it matters whether the government buys bridges or schools or roads or whatever. One gives us more bridges and the other gives us more schools, or roads, or whatever.

5. Other financial assets. Scenario 5 is that all new money is used to buy some other financial asset, like shares in IBM. Since the government owns the central bank, it doesn't matter if it is the government or the central bank that buys the IBM shares. Scenario 5 is identical to scenario 2, plus a tax used to buy IBM shares. Scenario 5 is also identical to scenario 3, plus a bond-financed purchase of IBM shares. Does a bond-financed purchase of IBM shares matter? I expect it depends on how close substitutes the two assets are in people's portfolios.

Sorry. What was the question again? Was it: "Do Cantillon effects matter?" Or was it "Does fiscal policy matter?" I can't tell the difference. There is no difference. "Cantillon effects" are just another name for "the effects of fiscal policy".

OK. I suppose that Austrian economists believe that fiscal policy matters. I suppose it does. And monetary policy has fiscal implications, because a faster growth rate of the money supply will mean bigger seigniorage profits for the government (as long as we stay on the left side of the Laffer Curve).

But the size of those fiscal implications depends on the ratio of the monetary base to nominal GDP.

In Canada, non-interest paying currency is currently around 4% of nominal GDP. An increase in the inflation target from the current 2% to 12% would mean a 10 ppt increase in the growth rate of the money supply. That would be a very big change in monetary policy. Even if the currency/NGDP ratio stayed the same at 4% (it would fall), the fiscal implications of that very big monetary policy change would be 0.4% of GDP.

What do you think would be the bigger deal: increasing the inflation target from 2% to 12%, or changing taxes or government spending by 0.4% of GDP?

[Update: hoisted from comments. J. V. Dubois says:

"You guys really do not get the gist of what Nick is saying?

1. Seigniorage of government controlled money IS TAX on base money
2. "Where" and how much of newly printed money is injected - even if that has any effect on redistribution IS FISCAL POLICY.

If you say that it makes a great deal of difference (deforming long-term capital structure etc) if money is injected into salaries of government employees vs purchase of bonds, then you by the same argument think that it makes a great deal of difference if government decides that from now on it spends 0.25% of GDP gathered in taxes on one versus another.

If you for instance say that Bond Dealers gather undue profits from these bond operations of the size of 0.25% GDP - then by the same account you have to be outraged that those very same bond dealers gather undue profits from regular yearly government deficits an order of magnitude higher. It is a problem of interest group capturing government and preventing competition from access to the bond market, it is not a problem of money printing."]

Comments

[Alex Godofsky]

I thought of that too.

[Nick Rowe]

Greg Ransom @11.22 December 5th. I'm sort of on the same page as you there. But I would say that "money" (medium of exchange) is the "trace data" and loose joint. Money matters, and matters a lot. But why does it matter where money is injected?

[Frank Restly]

Nick,

For credit demand, liquidity preference, and productivity to be constant a lot of things have to happen that don't exist in reality.

1. For increases in debt levels to always be matched with increases in real growth (constant productivity) we need either a non-resource bound economy (think Star Trek energy to matter converter) or we need a never ending stream of value added enterprises.
2. For constant liquidity preference we would need some sort of command and control economy (think of coal mines and company stores).
3. For constant credit demand we would need credit rationing - say Friedman's constant money growth recommendation.

Assuming you can get all three then what I said is true.

[Nick Rowe]

Frank: think about worm gears. Or think about balancing a pole upright in the palm of your hand. If you want the top of the pole to go North, do you move your hand North?

[Gret Ransom]

Nick, let's redirect the flow of government bonds and Fed bond purchases by giving taxpayers the option to buy bond rather than pay their taxes. What if the Federal Reserve begins buying these bonds.

What do we have going on here?

We could add other background premises to the story, but lets start with that.

Monetary policy or fiscal policy .. at what point is which, which.

If we switched from what we have today to this alternative regime, does anything at all happen?

[Nick Rowe]

Greg: "Nick, let's redirect the flow of government bonds and Fed bond purchases by giving taxpayers the option to buy bond rather than pay their taxes."

That doesn't work. Nobody would pay taxes if they had the option to buy a bond instead (for the same amount of money).

[Edmund]

Idle ruminations: if it doesn't matter how the money goes in, does it matter how the money comes out? Is reversely the process qualitatively different?

[Edmund]

Blast - "Is reversing the process"

[Frank Restly]

Nick,

The original question posed by Alex was:

Most of us assume that market forecasts of inflation have at least some predictive value. But fine, let's throw that one out. Which do you think is more likely:

1) If the Fed cuts nominal interest rates, inflation will probably go up (relative to what it otherwise would have been).
2) If the Fed cuts nominal interest rates, inflation will probably stay the same.
3) If the Fed cuts nominal interest rates, inflation will probably go down.
4) None of the above

And my original answer to this question was 4) none of the above. Through some manipulation of the equation I presented (take derivative of both sides with respect to the nominal interest rate) you can show that the nominal interest rate is positively correlated with the inflation rate. However, that assumes that no other variables (liquidity preference, productivity, credit demand) change with respect to the nominal interest rate.

You can make the argument that credit demand and liquidity preference are negatively correlated with the nominal interest rate to which I would agree. But when Alex and I were taking the derivative we did not allow for that (assume all other terms constant).

And so my answer is still 4).

[Frank Restly]

Nick,

"That doesn't work. Nobody would pay taxes if they had the option to buy a bond instead (for the same amount of money)."

Nobody would buy bonds if they realized that there were no taxpayers to make the interest payments on the bonds.

[Nick Rowe]

Frank: read my post. That's what happens when you let engineers/mathematicians loose on economics. They mistake correlation for causation. Some causal processes don't run in reverse (worm gears, poles balanced upright).

Edmund: that one should work in reverse too. ;-)

[K]

Nick: "Oh Christ"

Yup. That's why I said this is at the core of the thinking you have to adopt if you believe in the RBC model. Since you can't deny that the CB can control the nominal rate then in order for the real rate to be determined by the free market at the equilibrium natural rate, you *have* to believe that every move in the nominal rate is accompanied *one for one* by an equal move in expected inflation. If the CB hikes by 10%, inflation will rise by 10% as required in standard RBC.

BTW, your first example *really* doesn't work :-)

Frank,

If your answer is 4 then you are saying that inflation is on average independent of the nominal rate. In that case the real rate, on average, moves up by 1% when you move the nominal rate up by 1%. To claim that the CB can't control the real rate the burden is on you to prove that expected inflation goes up on average by exactly 1% for every 1% increase in the nominal rate. That's a high hurdle. You're simply deluded if you think a 10% rate hike is going to cause an inflation spike.

[Frank Restly]

Nick,

My original answer was 4) None of the above. I did not say that the positive correlation between nominal interest rates and inflation was an accurate reflection of reality nor did I say it implied direct causality between the two.

I believe you are arguing for sticky prices and / or market failures (hence the reference to worm gears working one way - debt expansion, but failing the other way - repayment of debt). It is a valid point. The model I presented does not allow for credit default - it assumes that all debts are paid in a timely fashion.

Sorry for the confusion.

[Nick Rowe]

K: "That's why I said this is at the core of the thinking you have to adopt if you believe in the RBC model. Since you can't deny that the CB can control the nominal rate then in order for the real rate to be determined by the free market at the equilibrium natural rate, you *have* to believe that every move in the nominal rate is accompanied *one for one* by an equal move in expected inflation. If the CB hikes by 10%, inflation will rise by 10% as required in standard RBC."

That is a very good point.

"That's why I said..."

Where did you say it previously? (Sorry, my brain is fried, from too many comments.)

"BTW, your first example *really* doesn't work :-)"

You mean worm gears? Depends on the gearing, and the friction, doesn't it? My toy trains couldn't be pushed to make the motor turn.

[Frank Restly]

K,

"If your answer is 4 then you are saying that inflation is on average independent of the nominal rate. In that case the real rate, on average, moves up by 1% when you move the nominal rate up by 1%. To claim that the CB can't control the real rate the burden is on you to prove that expected inflation goes up on average by exactly 1% for every 1% increase in the nominal rate. That's a high hurdle. You're simply deluded if you think a 10% rate hike is going to cause an inflation spike."

You are reading way too much into what I said. I said my answer is 4 because answers 1,2, and 3 do not accurately reflect what I believe. Inflation is not on average TOTALLY indepedent of the nominal rate. There are other factors at work. If the fed does a 10% nominal rate hike, if productivity does not increase to accomodate that 10% rate hike, if credit demand does not fall to accomodate that 10% rate hike, and if liquidity preference does not fall to accomodate that 10% rate hike then one of two things will happen:

1. A lot of borrowers will default on their loans (something that is missing from my model)
2. You will get higher inflation

[K]

Nick,

I said it yesterday at 12:24pm in this comment thread.

As far as your "first example" goes, I was talking about the current post, not the old one. The old one is great. In the current post I was objecting to the idea that paying 10% on money would cause inflation. By arbitrage, nominal interest rates would have to rise to 10% which (as we're discussing now) would definitely result in disinflation, not inflation. That's why it's very different from the helicopter drop in the second example.

[John S]

Edmund,

"The Austrians love the market. In the market for ideas, Austrian economics hasn't done well. Should this not give one pause?"

Hayek #1 seller in "Theory of Economics" on Amazon:
http://www.amazon.com/gp/bestsellers/books/2602?ref=sr_bs_1

Alexa rankings:
- Zerohedge: 1,963
- Ron Paul: 9,176
- Mises.org: 18,072

Meanwhile, other popular economics sites...
- Marginal Revolution: 43,237
- Freakonomics.com: 43,389
- Scott Sumner: 239,824
- Steve Keen: 260,609
- Greg Mankiw: 313,373

Not too bad. Like it or not (and there are many times that I don't), look forward to seeing a lot more "Internet Austrians" in the future.

[Greg Ransom]

"Greg Ransom @11.22 December 5th. I'm sort of on the same page as you there. But I would say that "money" (medium of exchange) is the "trace data" and loose joint. Money matters, and matters a lot. But why does it matter where money is injected?"

The language of "injected" isn't particularly helpful.

Money, credit & highly liquid substitutes for money snowball in the banking and finance sector in ways tied to lengthening of production processes, with pushes from Fed Reserve policy, pushes which lower risk and perceptions of risk on some margins, change relative price relations etc. It might be added that too big to fail is part of monetary policy.

Things would be different if the Fed were targetting pushes toward individuals who prefer expensive back rubs to long production goods and other assets, for example, by distributing free cash that can only be used in its first use for back rubs, validated by licensed back rubbers.

[Nick Rowe]

Greg: compare two policies:

A: The Fed gives $1 billion new money to people who are very impatient to consume now.

B=B1+B2, where B1. The Fed buys bonds with $1 billion new money. B2. The government issues $1 billion bonds and gives the proceeds to people who are very impatient to consume now.

A is identical to B. Are the effects of A identical to the effects of B?

[Greg Ransom]

In scenario A the government has no bonds, in scenarios B the government has outstanding bonds or issues bonds.

Tomorrow the government defaults on its bonds.

How is A identical to B?

[Greg Ransom]

Let me roll with my snowball metaphor.

Snowballs can roll and snowballs can met, snowballs call roll in thick wet snow and snowballs can find themselves hitting bumps or hills in the road and slowing down. Snowballs roll even faster are steeper slopes.

Now, snowballs can roll on their own and they can melt on their own.

Now Fed actions and expectations of Fed actions and implicit too big to fail guarantees from the Fed -- and changes in the realities and expectations of all of these, can give the snowball a push. And the can give the snowball new fresh snow. And they can assure people that is will all be downhill rolling (the Greenspan put, too big to fail, Keynesian 'aggregate demand management' scientific assumptions).

Now, the 800 lb gorilla of the economy is the financial and investment sector. This is the snowball that gets the special care and feeding of the Fed and the government more generally.

I don't think it helps a lot to use the metaphor of saying the Fed or the government "injects" anything into the snowball.

I think its better to say they are rolling a snowball, pretending they can make it roll as fast and as big as they can make it -- and every time its about to hit a cliff sized hill, all they have to do is make it snow and tilt the playing field some more to keep it rolling. Cliffs don't exists if the Fed just keeps that snowball ripping.

[Frank Restly]

Nick,

Presuming a demand for goods that are consumed (purchased) with money
Scenario A: The Fed creates money to meet a demand for money equal to the demand for goods times the price level.
Scenario B: The government borrows money from the Fed equal to the demand for goods times the price level. Government then creates an additional demand for money through taxation to make the payments on the bonds.

Scenario A - Demand for money = Demand for goods x price level
Scenario B - Demand for money = Demand for goods x price level + Taxation

If demand for money and demand for goods is constant in both scenario A and scenario B.

Scenario A - Price Level = Demand for Money / Demand for Goods
Scenario B - Price Level= ( Demand for Money - Taxation ) / Demand for Goods

[K]

Greg Ransom,

"Tomorrow the government defaults on its bonds. How is A identical to B?"

The bonds are issued by the government to *itself*. It's pure accounting, and it can me modified at any time with zero impact. In scenario A, for example, the treasury could just issue a $1Bn bond and give it to the Fed (then the accounting is identical to scenario B). No difference, default or not.

[Greg Ransom]

"Proceeds" -- I thought there was a sale of the bonds to the public here.

"The government issues $1 billion bonds and gives the proceeds to people who are very impatient to consume now."

The bonds are issued by the government to *itself*.

[Greg Ransom]

Nick, if the bonds are at 150% interest, compounding daily and maturing short term, and most taxes are payed by folks who are investors, it looks to me like we have a process which shortens the production process and eventually sterilizes cash.

[Greg Ransom]

K -- We are talking about publicly traded bonds, I thought. If the government defaults on its bonds, taxpayers are off the hook, and the Fed and banking system lacks the reserves which give solvency to their balance sheets. There is a time structure component of flows of money and resources implicit in bonds. That's why they are anything at all. If there wasn't a time component to bonds, they'd just be money, right?

[Greg Ransom]

Are there special bonds that are exchanged only between the Fed and the US government and are not traded with the public?

What if we eliminated the bonds and just made the US government pay a tax whenever the Fed credits money into the US governments account. What if there was a delayed time element on the tax -- a tax payed over a period of time.

And what if the right to these taxes could be publicly traded. Do we now have a "bond" again?

Isn't the key thing the fact that government bonds compete with investments in time-taking production processes among those who refraining from consumption in the promise of some future return?

If we have government bonds circulating and being traded, we have an alternative to production goods investments trading almost as money, the way stocks can trade almost as money.

But the distributional effects are different.

[Greg Ransom]

To be explicit, we have a process that works to shorten the production process at both ends, not just one.

I wrote,

"Nick, if the bonds are at 150% interest, compounding daily and maturing short term, and most taxes are payed by folks who are investors, it looks to me like we have a process which shortens the production process and eventually sterilizes cash."

[Nathanael]

Finally, I found out what "Cantillon effects" are supposed to mean.

But I don't like your dissection. (Admittedly, I'm not an Austrian and you were arguing with them.)

As far as I'm concerned, the only relevant "Cantillon effect" is the question of whether the money enters the economy in the hands of people with a *high propensity to spend the money* (poor people, mostly), or whether it enters the economy in the hands of people with a *low propensity to spend the money* (rich people, mostly). This determines the velocity of money, the multiplier effect, etc.

It's the only relevant "how the money gets into the economy" question: shovelling the money into zombie banks which dump it in vaults to "repair their balance sheets" doesn't help the economy at all, while shovelling it to poor people who immediately spend it on food (and the supermarkets spend that on labor, and pay truck drivers to move the food, and then the money goes to the farmers, who spend it pretty quickly too) does help the economy.

"What do you think would be the bigger deal: increasing the inflation target from 2% to 12%, or changing taxes or government spending by 0.4% of GDP?"

Well, this depends very much on where the money goes. Money to the poorest has a *massive* multiplier, so 0.4% of GDP directly to homeless people would have a definite effect. Meanwhile, changing the inflation target... depends on the transmission channel through the banks, and if the transmission channel is broken, then you can raise the target to 5000%, and there still won't be any inflation, the money will just sit in the pockets of the bank executives.

Is this really in dispute? I'm using extreme examples deliberately. In the range of "cut taxes on professors" versus "building bridges" versus "cheaper mortgage lending" I'm sure the "Cantillon effects" are not macroeconomically important. But in the range of "money in the bank accounts of billionaires" versus "food stamps", they pretty clearly are macroeconomically important.

I think I agree with you -- "deciding where the money will enter the economy IS fiscal policy" is something I agree with -- but I'm just not clear on what you're saying.

[Jacques René Giguère]

Nick: "deciding where the money will enter the economy IS fiscal policy" reminds me of an earlier comment I made this winter:
"fiscal policy is for those who don't have money to play games with and whose portfolio choice is between milk and bread"...