On not quite getting Lloyd Metzler

30 years ago, IIRC, my colleague Steve Ferris said I should read Lloyd Metzler's Wealth, Saving, and the Rate of Interest, because he thought it was a great paper. Yesterday Brad DeLong said the same thing. David Glasner also thinks it's a classic paper. When three very good but very different economists recommend a paper that strongly, you figure they are probably right.

But when I re-read it now, I get the same impression I did 30 years ago. It's an OK theory paper, but nothing special. And not very important at all empirically. But the guys who think it's an important paper are at least as smart as me. Maybe I'm not quite getting something?

Lloyd Metzler sets up a model with two financial assets: money (issued by the central bank); and "shares". Those shares are real rather than nominal assets, because (unlike bonds) they do not promise a fixed dollar payment. Those shares are net wealth, because they represent claims on the real productive assets of firms. But he assumes shares held by the central bank are not included in the wealth of the public. He does not say this, but it is as if the central bank is owned by foreigners. (If the central bank is owned by the government, and if the government is owned by the people, and they count government wealth as part of their wealth, like under Ricardian Equivalence, we would get a different result). The demand for money, and desired consumption and saving, depend on wealth and the rate of interest.

Helicopter money is neutral in the model (all prices are flexible and there are no nominal bonds). An increase in M via helicopters causes an equiproportionate change in P, so M/P is unchanged, and real wealth is unchanged, and the real interest rate is unchanged.

Vaccum cleaner shares is not neutral in the model, for obvious reasons. If the foreign-owned central bank confiscates some of people's shares, their real wealth drops. That drop in real wealth will have other real effects, and also nominal effects. Those real effects include the real rate of interest.

An open market operation, where the central bank issues new money and buys shares, is equivalent to helicopter money (neutral) plus vacuum cleaner shares (non-neutral), which equals non-neutral.

The message Lloyd Metzler draws from this is that monetary policy (in the sense of open market operations) will affect the equilibrium real rate of interest. So the real rate of interest is partly a monetary phenomenon. OK. And helicopter money has different effects from an open market operation. OK.

But how big is this effect? How big is this difference? The assets of the central bank are a very small part of total wealth.

Let's ballpark it. If we capitalise future income at 10%, we get a conservative estimate of total wealth as 10 times annual GDP. And if the assets of the central bank are around 10% of annual GDP, we get that the assets of the central bank are 1% of total wealth. Even if the central bank doubled its size (and did so in secret, and bought shares before the price of shares roughly doubled), that amounts to a 1% confiscation of total private wealth. That's small, compared to the effects of other tax policies, recessions, and natural disasters. And that's for a truly massive change in monetary policy, that roughly doubles the equilibrium price level.

Now in the real world, central banks mostly buy government bonds, not shares in companies. Does it matter much if central banks do helicopter money rather than buy government bonds in an open market operation? Very simple arithmetic tells us the difference between them: helicopter money minus open market operations = helicopter bonds. If helicopter bonds has a big effect, then there is a big difference between helicopter money and an open market operation.

How big a $ helicopter drop of bonds would you need, to have the same effect on the equilibrium price level as a $1 helicopter drop of money? If there were a economically significant effect of helicopter drops of bonds, we would expect it to show up in the data. Do countries with higher government debt/GDP ratios also have lower base money/GDP ratios, other things equal? Because that is what we would expect to find in the long run and cross-country data, if both helicopter bonds and helicopter money increased the equilibrium price level. And how big could that effect possibly be, measured as dollars of base money equivalent to dollars of bonds? Debt/GDP ratios range enourmously, from near 0% to over 200% of GDP. Base money/GDP ratios don't vary anything like that amount, in absolute terms. They are normally around 5% to 15% of GDP.

If it really mattered whether the central bank issued base money via helicopter drop or by open market purchases of bonds, so that a $1 helicopter bonds had the same effect as some non-negligible fraction of a $1 helicopter money, the evidence would be staring us in the face in the long run and cross-country data. We would see a big negative slope (slope, not elasticity) if we plotted debt/GDP against base/GDP. And countries with 200% debt/GDP ratios would already have exploded into hyperinflation, because they couldn't make base money negative enough to compensate. It doesn't; they haven't. (Scott Sumner made a similar observation once.)

To sum up, the sort of wealth effects from monetary policy that are in Lloyd Metzler's model are peanuts. I can think of other wealth effects from monetary policy that are not in his model that are probably much bigger. And if prices are sticky, I could definitely get much bigger wealth effects. What sort of recession would we get, if the central bank suddenly decided to halve the price level, and how big an effect would that have on wealth? The last thing I would think about is whether the central bank halved the price level via open-market operation sales of bonds or by confiscating half of everyone's money.

My take-aways from re-reading Metzler's paper, and thinking about it when writing this post:

1. Even though we normally think of "pure" monetary policy as an open market operation, the theoretically "pure" monetary policy is a helicopter drop. (Patinkin said much the same thing, but after Metzler, so Metzler gets the credit, unless someone else said it even earlier.)

2. Does fiscal policy matter much? To the extent that fiscal policy matters much, it can't be through wealth effects.

3. We should be paying more attention to the temporary/permanent question, both for monetary and for fiscal policy, and less attention to what the new money is spent on.

But maybe I'm missing something, that other people can see?

Comments

[Nick Edmonds]

Nick,

Are you suggesting that if holdings of net financial assets are a significant determinant of private sector spending, then high debt to GDP ratios should be positively correlated with inflation (other things being equal)? Because although it might apply to a pure helicopter bond drop, I don't think that should generally be the case.

[Market Fiscalist]

Say a person has an income of $100 a year all of which they spend and net wealth of $1000 which he keeps $200 in money and the rest in interest bearing bonds.

Then they become nervous about the future and want to increase there savings to $1100 so they plan to cut spending to $90 until they achieve this goal. They plan to keep only $180 in money and the rest in bonds so demand for the unit of account has actually fallen.

The CB wants to get their spending back up to $100 to avoid a recession. They have 2 choices.

1. Just give them enough additional paper wealth to make them feel comfortable spending $100 per year. Here the increased spending is mainly driven by higher wealth relative to the current price level.

2. Buy bonds off them for newly created money, and hope they spend some of the new money. I have a bit of trouble with this mechanism because try as I might it seems like the effect of these purchases is to lower the interest rates on bonds, and is it this lower interest rate rather than the increased money holdings that drives the increased current spending.

So the things that drives the level of spending in the present (for any given level of "animal spirits") are

1. Change in wealth relative to current price level
2. Change in interest rates causing inter temporal spending to chnage

Of course these things are related (a change in interest rates caused wealth to be revalued). But the story does seem more complex than "increase the monetary base and changes in spending will follow". In my example where they induce more spending via "helicopter drops" that add to people's wealth, then the spending target may be achieved with no increase in monetary base at all (as all the new wealth will be held in bonds).

[Market Fiscalist]

I think an implicit assumption I made in my comment is that when the CB uses OMO to hit an interest rate target this is the functional equivalent of allowing individuals to swap between cash and bonds (that pay that given interest rate), so that the ratio of bonds/money in the economy is entirely demand driven for any given rate.

[Tom Brown]

Market Fiscalist, you write:

"...try as I might it seems like the effect of these purchases is to lower the interest rates on bonds"

Is that a comment specifically about your thought experiment here, or is that your view in general? If the latter, Woodford has some thoughts on that in this interview.

[Market Fiscalist]

Well, it didn't used to be my view until now , but I came to that conclusion when thinking this through and now I can't seem to shake it off.

[Market Fiscalist]

I mean obviously it lowers the price of bonds - its the bit about this being the main cause of the subsequent higher spending (rather than the corresponding increase in base money) that I changed my mind on.

[Tom Brown]

Nick, are you saying that you agree that "the theoretically "pure" monetary policy is a helicopter drop?"

[Market Fiscalist]

(Apologies for my stream of random comments.)

If all money was electronic and the CB could adjust up or down the value of everyone's balances they could exercise a very pure form of monetary policy. It is my view however that the effects would mostly be "wealth" effects (they spend more because their net wealth has increased after the CB makes an upward adjustment) rather than just because they now have more cash.

Suppose the CB had the power not only to adjust the size of people's balances but also could adjust the ratio of their bond/money holdings. If the CB changed this ratio to increase the quantity of base money , while still trying to target the same interest rate, then this change would have no effect. They would be forced (other things equal) to convert all the base back into bonds and there would be no HPE driving increased spending.

[Nick Rowe]

Nick E: I don't think I would put it that way. If real output demanded depends (positively) on both M/P, and B/P, then for two countries that have output=potential, and all other things (like inflation) equal, we should see a negative correlation between M/NGDP and B/NGDP. And not just a negative correlation, but a big enough negative slope to matter. (e.g., if money and bonds were exactly equivalent in their effects on demand, so demand depends on (M+B)/NGDP, the slope should be minus one.)

And given the massive variance we observe in B/NGDP ratios, at least some of which must be exogenous, if that slope was anywhere near minus one, the evidence would be staring us in the face, because the effect would be big enough to swamp all the other things that aren't equal.

[Nick Rowe]

MF: "If all money was electronic and the CB could adjust up or down the value of everyone's balances they could exercise a very pure form of monetary policy. It is my view however that the effects would mostly be "wealth" effects (they spend more because their net wealth has increased after the CB makes an upward adjustment) rather than just because they now have more cash."

And since base money is such a tiny percentage of total wealth, and the fluctuations in base money are swamped by other fluctuations in wealth, if your view was correct it would have been impossible for the Bank of Canada to keep inflation at the 2% target.

[Market Fiscalist]

"And since base money is such a tiny percentage of total wealth, and the fluctuations in base money are swamped by other fluctuations in wealth, if your view was correct it would have been impossible for the Bank of Canada to keep inflation at the 2% target."

Weren't they relying on an interest rate effect? They adjusted the interest rate so that people changed their inter temporal spending decisions to keep inflation on track.

Had they instead used fiscal policy as the main tool then why would the size of the base be relevant ? They could have made people richer by giving away either bonds or base and people would decide what mix to keep based on interest rates and "animal spirits".

I agree however that if wealth fluctuates randomly with no direct connection to fiscal/monetary policy , and if the % of wealth consisting of things the CB/govt control is small, then that may undermine my views: need to think about that.

[Nick Rowe]

MF: "Had they instead used fiscal policy as the main tool then why would the size of the base be relevant ?"

Because they didn't use fiscal policy to stabilise demand at all, until the ZLB hit. Instead, there were some massive changes in fiscal policy in the late 1990s (big deficits to big surpluses) that were done for totally different reasons. But the Bank of Canada kept inflation at 2% regardless.

[Market Fiscalist]

Yes, I see that based on the numbers you suggest in your post then controlling NGDP via adjusting people's wealth via helicopter drops would involve unfeasible large changes in the qty of base money and bonds issued.

However it still seems that it is interest rate policy affecting inter-temporal spending that is doing most of the heavy lifting in monetary policy, rather than just the qty of base money being adjusted, and the degree to which it will be effective will depend upon degree of elasticity of current spending to changes in the interest rate.

So: The qty of base money can be adjusted either by giving it away, or by lowering interest rates so that people will willingly hold (and spend) more base as a % of their total wealth. Which is more effective is an empirical question - but "back-of-envelope" type calculations would indicate that interests rate policies will have a bigger effect than "helicopter drop" policies for a give change in base.

[Nick Rowe]

MF: suppose you give money away, and people don't want to hold extra. What would that do to: their spending; interest rates?

[Nick Edmonds]

Nick,

That doesn't sound right to me. If either B/NGDP or M/NGDP are away from their long term equilibrium levels, I can see how that might impact on demand. But the actual equilibrium ratios will depend on all sorts of things like institutional and regulatory structure that are going to be very different from country to country and across time.

[Nick Rowe]

Nick E: take the extreme case, where output demanded = NGDP.D[(M+B)/NGDP, other stuff]

Holding other stuff constant, if output demanded = potential output, we should observe a perfect negative relationship between M/NGDP and B/NGDP, with a slope of minus one. For example, suppose the government increases B/NGDP for some reason unrelated to trying to increase demand. We should see M/NGDP fall by an equal amount to keep output at potential.

If other stuff is not constant, that will mess up the correlation and flatten the slope. But if there is enough exogenous variance in B/NGDP we should still see something of that slope.

[Market Fiscalist]

"suppose you give money away, and people don't want to hold extra. What would that do to: their spending; interest rates?"

They would spend some and buy bonds with the rest. If the CB didn't fully accommodate the increased demand for bonds by issuing more then interest rates would fall until people were happy with the bond/money split. The price level would probably rise too and this would in turn have an influence on both interest rates and the demand to hold money.

[Nick Rowe]

MF: Yep. And if M doubled and P doubled, the demand for money would rise along with the supply of money, and we would be at a new equilibrium, with the same real (inflation-adjusted) spending, and the same real lending and borrowing and interest rates as before. That's the classical quantity theory of money.

[Market Fiscalist]

Yes, if you wait for P to adjust up or down to match the new wealth/spending ratio then everything will be back in equilrium.

But I thought we were talking about a scenario where people were spending less and sticky prices were preventing a quick adjustment to the price level so we needed monetary or fiscal policy (whichever is optimal) to bring spending (and NGDP) back on target.

It still seems to me that whatever is driving the slowdown in spending its probably not a shortage of money per se, but the fact that people when uncertain about the future will want to have a higher ratio of total savings (whatever form they are kept in) to spending.

I would welcome your view on my earlier thought experiment "Suppose the CB had the power not only to adjust the size of people's balances but also could adjust the ratio of their bond/money holdings. If the CB changed this ratio to increase the quantity of base money , while still trying to target the same interest rate, then this change would have no effect. They would be forced (other things equal) to convert all the base back into bonds and there would be no HPE driving increased spending."

[HJC]

Nick: If you were to include central bank assets in the private sector's wealth then, to avoid double counting, you would have to remove money from wealth. Then there would be no model. Instead he treats the assets held by the central bank as being offset by the central bank's monetary liabilities. Also note the footnote on page 109 which discusses his treatment of the income from central bank assets. But your point about the size of lambda to have any real effect may well be right.

[CMA]

"We should be paying more attention to the temporary/permanent question, both for monetary and for fiscal policy, and less attention to what the new money is spent on."

Heli drops if conducted directly by fed without purchasing assets should be permanent because there is no claim on the feds balance sheet. If the fed has purchased a treasury to inject money then permanence is more doubtful and reliant on the fed constantly offsetting income provided by the asset it holds.

[Min]

"it is as if the central bank is owned by foreigners."

Anticipating the ECB.

[Jon]

Yes... Interestingly growth in assets has outstripped growth in money for a while. Perhaps this explains why expectations and forward guidance seem so much more important in this century.