On "letting the market set interest rates"

Most people who say "interest rates should be set by the market and not by government-owned central banks" are confused. But not all who say that are confused.

People who want the government to get out of the money business altogether, and de-nationalise money and central banking, are not confused. They might be right or they might be wrong, but they are not confused. But this post is not about them. Off-topic.

Modern central banks, like the Bank of Canada for example, have an inflation target. And they adjust the overnight rate of interest to try to hit that inflation target. Does this mean that the Bank of Canada, rather than the market, sets that overnight rate of interest? Well, yes and no.

The driver of a car sets the position of the steering wheel. But if the driver has decided to drive along a particular road to get to a particular destination, it is that destination, and the bends in that road, that really determine where the driver must set the position of the steering wheel. If we replaced the conscious decisions of the human driver with a self-driving car, we would see the steering wheel do roughly the same thing. Provided the car stays on the same road in both cases.

The "destination" is the 2% inflation target. The "bends in the road" are the shocks that hit the economy. The "self-driving" car is the market. Just to make the analogy clear.

If the driver and passengers had a clear view of the road ahead, and if all agreed on how cars worked, and if all agreed on the destination, and if the driver were doing a good job of staying on the road, we wouldn't see any back-seat drivers. Some passengers might dislike right turns, and others might dislike left turns, but none would try to persuade the driver to do anything different. They might whinge about the bends in the road, but they wouldn't complain about the driver when the driver turned the wheel to follow those bends.

But if the bends in the road are unclear, or if people disagree about how cars work, or are unsure whether the driver is really trying to stay on the road to get to the agreed destination, we might see a lot of back-seat driving. And it will be very hard for people to separate their personal preferences for right or left turns from their personal judgements about whether right or left turns are needed to keep the car on the road and get to the destination. And even harder if there is no agreed destination.

Paul Krugman gives us a recent example of this. But anyone who has renewed a mortgage can recognise the same feeling in themselves: "I hope the Bank of Canada cuts interest rates!"

If the government owns and ultimately controls the central bank, and if you don't want that to change, then it does not make sense to say that the market should determine everything. The central bank must choose something. It could be an inflation target, or it could be an NGDP target, or it could be a fixed exchange rate target, but there must be something it targets. If the government is in the money business, the government must, logically, set something, and cannot let the market set everything. People who don't get that point are confused.

But if the government-owned central bank sets the destination, there is no reason why we can't let a self-driving car make all the other decisions. Set up some sort of prediction market like Scott Sumner advocates, and hook it up to the steering wheel with a negative feedback mechanism, so if the market thinks that the car is turning too far to the left, the wheel turns automatically to the right.

Then we would have an immediate response to any back seat drivers: "You think the central bank is getting it wrong? OK, put your money where your mouth is, and you should make a profit in the prediction market, if you turn out to be right!"

It's not just about getting monetary policy right to reach the destination. It's about the perceived legitimacy and transparency of the process by which we use our imperfect judgements to try to reach that destination. Juries vs judges as ways to implement the rule of law rather than the rule of man. A prediction market would be like a jury, where anyone could be a member, except if you found the defendant guilty/innocent, and subsequent evidence showed he was innocent/guilty after all, you would have to pay a fine if you were wrong, and get a reward if you were right.

The steering wheel doesn't have to be interest rates. And in my opinion it shouldn't be interest rates. It could be base money, or it could be the nominal price of some real asset. But that's a separate question.

[I hope I die before they force us all to use self-driving cars. But if I had to take the bus, it might as well be a self-driving bus, if they do better than human drivers. Monetary policy is like a bus, because we are all on it together.]

[In Canada, the 2% inflation target destination was jointly agreed on by a democratically elected government and the Bank of Canada, and the governors of the Bank are appointed by the government, so the 2% inflation target destination has legitimacy. It might be the wrong target, and in my opinion it is the wrong target, but it is nevertheless a legitimate target.]

Comments

[Market Fiscalist]

If the destination is set at "achieve 2% inflation" or "Achieve 5% NGDP rate" then most of the time a self-driving central bank will be able to steer purely by interest rate adjustment and stay on track to arrive at the destination.

Sometimes (every few decades) the self-driving CB may hit road conditions so severe that even turning the wheel to the maximum (zero interest rates) will fail to keep it on track. Something else (money creation via fiscal policy) will be needed. If done right this fiscal policy will steer the economy even when the IR steering wheel is broken. In fact it will even avoid the need for low interest rates.

So: What use is a self-driving car that works perfectly except when the road conditions are really bad ? Perhaps market-orientated fiscal policy might be a better way to steer the economy even when conditions are good IR are not zero? The destination (inflation or NGDP) could be same, and the market really could determine interest rates.

[Nick Rowe]

Market Fiscalist: think of using interest rates as an instrument as like driving a car with rear-wheel steering (or driving a car in reverse). If you want to make the car go right, you have to make the rear wheels go left. And you can't do that if you are too near the left edge of the road (the ZLB). If you *must* drive the car using rear wheel steering, then you need to stay well away from the edges of the road (adopt a higher inflation target or NGDP target (the latter also has self-equilibrating properties that keeps it nearer the centre of the road automatically, so you need less steering)). Better yet, using front-wheel steering (something other than interest rates).

The people of the concrete steppes love fiscal policy. They can *see* how it works, because Y=C+I+G+NX! Little do they realise they are (explicitly or implicitly) arguing from an accounting identity, and accounting identities are the exact opposite of concrete. They are pure framing devices. But even sophisticated New Keynesians are invoking the fiscal fairy when they advocate fiscal policy. Because it is only the *expectation that G will be falling over time from now on* that expands Aggregate Demand in NK models, and even that only works because NK models *just assume*, for no reason whatsoever, that people expect that future output always converges on full employment regardless.

[Market Fiscalist]

I don't fully understand your second paragraph. The thinking behind using IR targeting to hit an inflation or NGDP target is that lower interest rates will spur increases in C+I+G+NX and so increase Y. The "concrete steppes" used is the adjustment of the balance between money and other assets (via OMO) with the aim of adjusting C and I. But if at 0% IRs C and I are still too low then the "concrete steppes" are gone. At this point it is monetary policy that is dependent upon the expectations channel. A higher rate of inflation may give some buffer space but doesn't change the underlying story.

Using fiscal policy to hit an inflation or NGDP target also aims to adjust the components of C+I+G+NX to spur an increase in Y. The "concrete steppes" is that increases in G (money creation via govt deficits) will directly increase C or I (for example by deficit funded tax-cuts, sales subsidies or investment breaks). Unlike monetary policy these "concrete steppes" will always work - no zero bound on fiscal policy.

New Keynesians seem to have complicated this story - but I believe that expectations not backed by concrete steppes are likely to fail when most needed. Fiscal policy is definitely aided by setting the right kind of expectations - but it is not dependent upon them and doesn't fail when expectations prove hard to shift.

[Minarchiste]

This approach neglects two important things:

1) The definition of inflation (with or without asset prices?). Getting toward that 2% CPI can build up huge imbalances in asset markets that are not captured by CPI...ask Mr Greenspan about that...

2) Relative prices : what if overall inflation is 2%, but the price of some goods/assets is skyrocketting (say houses), while most other prices are flat? Again, huge imbalances in the economy.

The end result of such imbalances must correct themselves at some point. That's what we call a recession.

We could say that the driver is color-blind, wearing sunglasses at night and sees only with one eye.

[Market Fiscalist]

Using terms like "monetary policy" or "fiscal policy" may not even be the best way to look at these issues. Both polices work by changing the money supply - one by varying the interest rate and adjusting existing financial wealth between money and other assets, the other by directly increasing financial wealth.

I think the attraction of the first method is that it allows Y to be adjusted without directly affecting the relative wealth of individuals and this perhaps is a reason for it be used when it works. However because it works mainly by adjusting interest rates (this is the primary reason why people change their portfolio balance) 1) It may have perverse effects on the structure of the economy and 2) It will largely stop working if interest rates hit zero.

[Vedicculture]

You can't "decentralize" money. If the nation state doesn't do anything with money(and it completely does not) then a "market based" setting of money is "centralization". A global cabal telling countries what to do.

This post represents the ill of modern day existential thought. Completely lost and degenerated.

[Vedicculture]

You see, I don't see any imbalances at all. Inflation targetting isn't going to cause imbalances. Imbalances is capital becoming a "herd" in certain investment behavior and neglecting other investments because of poor RoR.

Capitalism's problem is this facet. After the Y2K/computer powered economic boom in the late 90's, capital had sig. excess liquidity that flew into finance like crazy which went into RE products because of RoR. As industrial disbasement continued, so did the capital flight into finance/RE. The same problem happened after WWI and was similiarly mishandled. The same after the Civil War............

The fact is, the world must decide if they are going to be a market state or a nation state. I would argue there are even better and less degenerate ways of organization but I don't see that dream coming true in the next 100 years. Market economy is a degeneration, totalitarian form of governance. Give me the nation state in the short run and we can de-evolve from there together..........

[NickRowe]

Market Fiscalist: we are wandering off-topic (my fault as much as yours), but in NK models fiscal policy works *only* through the "expectations channel", regardless of whether the economy is at the ZLB or not.

Minarchiste: "This approach neglects two important things:"

This blog post does neglect those two things, yes. Because this post is not about what is the best target for monetary policy. It's about how to hit that target.

Vedicculture: you lost me.

[Ben J]

Nick,

You will be amused - I asked a top macro guy in my department (who will be supervising my thesis starting from next year) how the level of output is pinned down in the NK models if the Euler function is only a ratio of present to future output. He looked confused for a minute, said "there are other factors, you don't have the background." I prodded him a bit more and he just told me to get Gali's book.

Got any advice? I'm not really sure where to start.

[NickRowe]

Ben: I am amused. But not shocked. In many ways he gave a good answer, because at least he understood your question. Some things we have to spend time thinking about. My guess is he will be thinking about it.

Advice? I'm useless at giving advice. Do your thesis on it? Start by adding a simple Pigou effect to the Gali model. Try to meld Gali with Samuelson 1958. See how it goes. Take it from there. (That's probably bad advice; follow it at your own risk!)

[Ralph Musgrave]

On the subject of steering wheels, I object to the fact that most economies have two steering wheels or two ways of imparting stimulus: first, the central bank, and second, “government” or “politicians” who can impart fiscal stimulus. That makes no more sense than a car with two steering wheels, each controlled by a different person.

[NickRowe]

I object to that fact too Ralph. But what can we do about it? We can recognise we have more than one objective, and so the fiscal authority can target those additional objectives. But you and I have been down that road before, and this post isn't about that.

[Kevin Donoghue]

Ben J,
"Get Gali's book" isn't bad advice at all. While I have some sympathy for Nick's criticism, I don't think he takes sufficient account of the fact that the harsh things he says about the NK model (Gali's Ch 3) apply with equal force to the Classical Model (Ch 2).

[NickRowe]

Kevin: Yep. "Get Gali's book" (and Woodford's) is very good advice.

But isn't the "Classical Model" a barter economy (or a model with a Walrasian auctioneer)? Because Say's Law would work fine in that sort of non-monetary economy. Or, if you have a monetary exchange economy where money really is in the model, my criticisms wouldn't apply.

(I actually sorta like the NK model in many ways. But what really pisses me off about NK macro is this: you've got all these really high-powered hot-shot theorists teaching this model to all these really bright students, all of whom can do ridiculously fancy math in their sleep, and none of them seem to have even thought about this really basic issue. It takes a burned-out ex-minor-university-administrator, who's an *associate* prof (that one word speaks volumes), winding down towards retirement, teaching first year, who can't understand the fancy math in the textbooks, who can't even get the math right for the pension plan assets, who hasn't read anywhere near enough modern stuff, (moi) to make an issue of it, in a *blog*. Yes I do have a chip on my shoulder, why do you ask?)

[Gizzard]

"Set up some sort of prediction market like Scott Sumner advocates, and hook it up to the steering wheel with a negative feedback mechanism,"

You mean a prediction market that has zero chance of working. Sumner advocates things he doesnt understand. People who
"do" futures markets for a living have been trying to tell him his idea is unworkable. These guys would actually like to see it work too. Another way for them to make money. Problem is it cant work. It would be nice if Scott would actually listen to people who do this stuff for a living, but he cant let reality get in the way of his theoretical models.

[NickRowe]

Gizzard: IIRC Scott did listen to their criticisms and did respond. I read those criticisms too, and they did not convince me. Simple version: you have a jar of beans, and let people take bets on whether there are more or less than 500 beans in the jar (exactly 500 and all bets are off). Then you add or subtract beans mechanically until the market odds are 50-50. Then at noon you stop betting and count the beans. Even simpler: set the odds at 50-50, then adjust the number of beans until the betting market clears.

[NickRowe]

Think of a horse race, where the faster horses are handicapped by adding weights. Except you adjust the weights until all horses are at the same odds to win.

[Dan Kervick]

Isn't the central bank's control over overnight rates much more direct than this? Beliefs about the central bank's rate choice play a role, but only because people know that the central bank has a non-belief-based mechanisms to rely on that are always effective, and it is thus futile to try to overpower the bank or out-guess it.

Changing the analogy slightly: The overnight rate is not the destination, it is the lane the car is driving in. The car is driven by a lot of people, each with their own personal steering wheel. The car has a mechanism by which the direction the car drives is determined as a resultant of the turnings of the many personal steering wheels.

But one driver has a steering wheel much more powerful than the others. Its steering wheel is so powerful that it can determine the direction with one mighty turn even if all of the other drivers turned in a different direction. As a result, once people know that the person with the powerful wheel is determined to drive in a particular lane, they also know that attempting to resist that decision is futile. As a result, most people conform right away with the decision and the person with the powerful wheel doesn't have to turn it very much.

But if that special person had a steering wheel just like everyone else's, this conforming behavior generally wouldn't occur.

[NickRowe]

Dan: I'm with you on your third paragraph (but you lost me on the first two). I would add that it is the knowledge of the special driver's intended destination, and that he will do whatever it takes to get there, that matters more than where he is turning the wheel at this particular moment.

And we need to ask: what is it about this special driver that makes him more powerful than all the other drivers? And I think there are two reasons:

1. The special driver cares only about the destination, while the other drivers care about their own profits, which depend in part on whether they are turning their wheels the same way as all the other drivers. (The central bank is not trying to maximise its own profits, but all the other banks and borrowers and lenders are.)

2. Asymmetric redeemability. The Bank of Montreal promises to redeem its dollars for Bank of Canada dollars, rather than vice versa (except when the BoC acts as lender of last resort in an emergency). (I can't think how to translate this into the metaphor!)

[Philippe]

How does acting as a 'lender of last resort' mean that the BoC promises to redeem its liabilities for Bank of Montreal liabilities?

[Gizzard]

Nick

If this was such a great idea it would have already been done. You dont think people are looking for more ways to place bets already? This would have happened outside of some Fed policy metric. Guys like Sankowski do that for a living. I think when he says its dubious to do you ought to listen. He's not saying that just cuz he doesnt like Scott. He's actually looked at how to set it up AND has set other futures markets up.

"Simple version: you have a jar of beans, and let people take bets on whether there are more or less than 500 beans in the jar (exactly 500 and all bets are off). Then you add or subtract beans mechanically until the market odds are 50-50. Then at noon you stop betting and count the beans. Even simpler: set the odds at 50-50, then adjust the number of beans until the betting market clears."

There isnt enough real time info to count the beans at noon. NGDP gets adjusted weeks later. Its not like we know what NGDP is at all times. There would be disputes about the number of beans for weeks. Thats just the measurement problem. The gaming the system problem is even bigger. Let it go you guys. Its a nice theory...... but totally unworkable.

[Frank Restly]

Nick,

"You think the central bank is getting it wrong? OK, put your money where your mouth is, and you should make a profit in the prediction market, if you turn out to be right!"

The only way to make money in a prediction market is for someone to make the opposite prediction - take the opposite side of a trade. Scott recommends a GDP futures market but fails to identify who is the issuing / redemption agent for GDP futures - who takes the opposite side of a trade when no one else will?

Most central banks are limited in what they can buy and sell - usually default risk free debt. If Scott is advocating that the central bank be the issuing / redemption agent then all that needs to happen is all other parties (you, me, everyone) bet against rising GDP and excess funds under a mattress. GDP collapses and central bank pays out - is this what Scott has in mind?

Incentives matter a lot.

[Kevin Donoghue]

Nick,
Yes, Gali's "Classical Model" is Walrasian, though he doesn't actually say that. But your criticism of the NK model applies, in the sense that E[C(t+1)]=optimal C(t+1) for all t. There's no story of how that optimum is found. And just as in the NK model, a policy of targetting the interest rate leaves the price level indeterminate.

[NickRowe]

Kevin: so we can think of the "classical" model as just the limiting case of the standard NK model in the limit as price stickiness vanishes and imperfect competition becomes perfect? Then yes, my criticism applies equally. The problem is not sticky prices or imperfect competition. I'm fine with those. The problem is the central bank setting interest rates in a monetary exchange economy. Or a model of a monetary exchange economy where money doesn't exist.

Philippe: "How does acting as a 'lender of last resort' mean that the BoC promises to redeem its liabilities for Bank of Montreal liabilities?"

Well, it doesn't do this literally, but the effect is the same. Suppose I issued NR notes that I promised to redeem for BoC notes at par. And suppose the BoC lent me unlimited amounts of BoC notes to ensure I could keep my promise to redeem. I could then issue trillions of NR notes, and force the BoC to lend me trillions of BoC notes so I could keep my promise. The BoC would then lose control of the money supply.

Gizzard: think of it this way: Canada already issues bonds indexed to the CPI. The yield differential between indexed and non-indexed bonds means a market in CPI futures effectively already exists. And that differential is one of the things the BoC looks at when it sets interest rates to target 2% inflation. When StatsCan reports the CPI it bases that number on a sample, and it does not revise that number. And CPI data is issued with a lag. All we need do is issue "Trills", that are indexed to NGDP (or a particular revision of NGDP), then switch to NGDP, and take the limit of that same process. Something cannot be impossible if it's already being done, in a small way.

Frank: the central bank adjusts its instrument until equal numbers of punters are willing to take opposite sides of the bet, so the betting market clears at 50-50 odds. See my reply to Gizzard in earlier comments.

[Philippe]

"force the BoC to lend me trillions of BoC notes so I could keep my promise"

if that were the case then how could any bank ever go bust?

[Philippe]

"People who want the government to get out of the money business altogether, and de-nationalise money and central banking, are not confused. They might be right or they might be wrong, but they are not confused. But this post is not about them."

Nick, do you think you might be able to do a post about them? That would be really interesting!

[NickRowe]

Philippe: it couldn't go bust. Banks only go bust when lenders of last resort decide they are too far gone to be worth helping, or close them down.

[Frank Restly]

Nick,

"Frank: the central bank adjusts its instrument until equal numbers of punters are willing to take opposite sides of the bet, so the betting market clears at 50-50 odds. See my reply to Gizzard in earlier comments"

Why would anyone take the opposite side of the bet against punters? Why wouldn't everyone other than the central bank be punters? The punters have the ability to affect the outcome in their own decision to consume or not consume, produce or not produce. From an incentive standpoint it is easier to do nothing than it is to do something. And if the central bank must take the opposite side of all trades (bettor of last resort) when no one else will, then it pays for all private individuals to be punters.

[Sina Motamedi]

Great analogy.

Self-driving cars are coming sooner than you think.

[Kathleen]

@Phillipe: "People who want the government to get out of the money business altogether, and de-nationalise money and central banking, are not confused. They might be right or they might be wrong, but they are not confused. But this post is not about them."

This is off topic so I will just say: read Goodhart on the Evolution on Central Banks. It is only 100 pages (main body) and is really easy to read.

[Nick Rowe]

Kathleen: I vaguely remember reading Goodhart on that too, and thinking it was good.

BTW, since the discussion here is pretty well wrapped up, if you feel like going off-topic, be my guest!