Artsie non-linearity, economics, and the concrete steppes

When economists say that something is "linear", rather than "non-linear", they normally mean it is a straight line, rather than curved. Y=a+bX is linear; Y=a+bX² is non-linear.

That is NOT what I am going to mean by "linear" in this post.

Instead, I am going to use the words "linear" and "non-linear" in the way Artsies use those words. A "linear" plot is one that starts at the beginning, then continues with day causing following day, until it gets to the end. A "non-linear" plot curves back on itself, with simultaneity, flashbacks and flashforwards, and self-referential asides where the characters wonder if they are in a story and how it's going to end.

Economics is now, and has been for hundreds of years, a non-linear discipline, in the Artsie sense of the word.

[Update: anyone who doesn't get my point, or who wants to see the point explained by someone who understands math modelling better than me, or who wants to see it from the perspective of a real scientist (not an economist) who studies predator-prey dynamics, should read Jeremy Fox.]

The very simple story of demand and supply, as we teach it in ECON1000, is a fundamentally non-linear story. If you try to tell the demand and supply story as a linear story, you always get into deep trouble. "Bad weather causes a decrease in supply, which causes an increase in price, which causes a decrease in demand, which causes a decrease in price, which causes a decrease in supply, which causes an increase in price...."

Explaining demand and supply to first year students is difficult, precisely because it is a non-linear story. They want to know whether a change in quantity causes a change in price, or whether a change in price causes a change in quantity? And in which direction will price change if quantity changes, or will quantity change if price changes? And you can't answer that question, because it doesn't make sense. We can only answer "both".

And so we draw a supply and demand curve diagram, point to the intersection of the two curves, and say that price and quantity are co-determined simultaneously in equilibrium by the supply and demand curves. Where the supply curve is the whole relationship between price and quantity supplied; and the demand curve is the whole relationship between price and quantity demanded. And explain that supply, quantity supplied, and quantity sold are three different concepts. And explain how demand, quantity demanded, and quantity bought, are also three different concepts. And tell a story of the disequilibrium sub-plot in which an excess supply causes price to fall to the equilibrium, and excess demand causes price to rise to the equilibrium. And distinguish that disequilibrium sub-plot from the main equilibrium plot in which shifts in supply or demand curves cause the equilibrium price to rise or fall.

"Equilibrium" is economists' code-word for "this is a non-linear story".

The very simple standard story of demand and supply is a non-linear story. That's what makes it so hard to teach properly. Students really really want to hear a linear story. They like linear stories, because linear stories are so much easier to understand, remember, and explain. You can even write linear stories in point-form. "S up arrow P down arrow D up arrow Q up" would say it all, if the plot were linear. The "arrows" indicate "precedence", in both the causal and temporal sense. If the plot were linear, we could easily lay out the concrete steps through which an increase in supply caused a decrease in price and an increase in quantity, taking one causal step at a time. But it's not linear.

My car has a linear transmission. The engine turns the main shaft, which turns a cog, which turns another cog, which turns the auxilliary shaft,..which turns the axles, which turn the wheels, which make the car go forward. The monetary policy transmission mechanism, even in the very simplest and crudest version of that story, is non-linear. Even calling it the "monetary policy transmission mechanism" is a highly misleading metaphor, because it leads people to expect a linear story.

The simple story of monetary policy I teach in ECON1000 is non-linear like the demand and supply story, only more so. That's because there are more goods, more supply and demand curves, more endogenous variables that get determined simultaneously, and almost everything depends on almost everything else.

I have a shameful admission to make. A couple of times, after some students still didn't get it, I broke down. I told them that this wasn't exactly right, and then wrote down: "M up arrow r down arrow I up arrow Y up arrow P up". And the students loved it. Yes! This they could really understand! They all wrote it down in their notes, memorised it, and regurgitated it on the exam.

What I did was wrong. I linearised a non-linear story. But that's not the point of this post.

The point is that the students really really wanted a linear story of the monetary transmission mechanism. And when I gave them a linear story they thought they understood it. They didn't. They misunderstood it. They misunderstood it just as badly as a student who thinks "an increase in supply causes a decrease in price causes an increase in demand causes an increase in quantity" misunderstands demand and supply.

My students were like the people from the concrete steppes. The people from the concrete steppes want a linear story of the monetary policy transmission mechanism. Sorry, but you can't have it. If you think you understand monetary policy through a linear story, you don't understand it. The true story is non-linear. There is simultaneous causation, so endogenous variables are co-determined in equilibrium. Expectations matter, so the story has flashforwards and flashbacks, where the future affects the present. Those expectations are self-referential, because the characters in the story know the author who works at the central bank, and can anticipate the future plot he plans to write, which affects their actions today. That means a loosening of monetary policy could mean that interest rates rise, not fall. You cannot reason from an interest rate change (Scott Sumner), because interest rates are endogenous variables.

The people from the concrete steppes find this all very frustrating. They want a story with a linear plot, because a linear plot is what they understand. And if they think the linear plot cannot work at the Zero Lower Bound on nominal interest rates, they conclude there must be no plot in which monetary policy works. And they think that any non-linear story is just handwaving and mystification.

Here is a very simple non-linear story, which the people from the concrete steppes won't like. We know that for any equilibrium time-path there exists a second equilibrium time-path along which all nominal variables (like NGDP, and the money supply) are higher or lower. This means that a permanent loosening of monetary policy will eventually move the economy towards one of those time-paths along which all nominal variables are higher. This means that the expectation of a permanent loosening of monetary policy will mean the expectation of an eventual permanent increase in all nominal variables like NGDP. (This is how monetary policy always works, regardless of the ZLB, because interest rate changes are an endogenous symptom of monetary policy changes, not an integral step in a linear causal chain.) An increase in expected future NGDP will increase demand today, and increase current NGDP. It would really help people to expect a permanent loosening of monetary policy and a permanent increase in future NGDP if the central bank announced a commitment to a higher future NGDG target, because hitting that target would require the central bank to loosen monetary policy permanently.

Not very linear, is it? But then economics isn't a linear subject, and hasn't been for hundreds of years. So what did you expect?

Comments

[Ritwik]

This is a very very good post. It is extremely logically coherent, remorselessly so, to invoke Keynes. It would be true, if economics was purely an abstract system of propositions. It is the most Austrian (and best) non-Austrian narrative I have ever read.

But what about the empirics?

Critics (Ashwin, Dan Alpert, FT Alphaville, BIS) are arguing with data. They're looking at oil/commodity prices. At wage growth trends. At SNB reserves. Where is yours?

Plus, there is a simple enough narrative challenge to your narrative - Sorosian reflexivity. Economies behave as the constituents of the economy behave. If financial market/other influential participants think like your undergraduates (and there's no reason to suppose why they won't), monetary policy transmission becomes linear-ish. Since their beliefs are borne out by what they see happening in the markets, they see no reason to change their linear beliefs. And we'r eback to square one and concrete steppes.

[Phil Koop]

Rather than indulging in yet another determined attack on your "concrete steps" straw man, it would have been more to the point for you to have responded to what Brad DeLong says here: http://delong.typepad.com/sdj/2012/06/a-fragment-on-the-interaction-of-expansionary-monetary-and-fiscal-policy-at-the-zero-nominal-lower-bound-to-interest-rates.html.

Because if Brad is right, the implication is that it ain't the people of the concrete steps who don't understand non-linear stories and expectations; it's Nick Rowe, market monetarist.

[K]

OK, I don't disagree with a lot of that, though a lot of it sounds more like "multiple simultaneous equations" rather than "non-linear." Of course, it's both.

I don't even have a problem with the last paragraph. So long as what you are expecting the CB to do is not beyond what's in principle achievable with its *instruments*! (yeah, I know, that's the part where you cringe... keep reading) I.e. assuming they can't buy anything other than government bonds, then the limit on the quantity of demand they can achieve via expectations is equal to the effect of credibly committing to keeping the short rate at zero *forever*. If they are allowed to buy real assets, then I think there is very little limit to what can be achieved via expectations (so long as the policy respects time consistency).

[Becky Hargrove]

Non-linearity is a very good way to explain this. Thanks. For one thing it's easier to imagine than the idea of multiple simultaneous equations that may or may not correspond with other factors.

[Ritwik]

K, I think Nick prefers to refer to multiple simultaneous equations as 'artsie non-linearity'.

[Saturos]

"You cannot reason from an interest rate change (Scott Sumner), because interest rates are endogenous variables."

And isn't that exactly what invalidates Taylor-rule thinking?

p.s. Artsie? Oh you Canadians...

[Jeremy Fox]

I can't comment on this as an understanding of how monetary policy works, but as a post on how difficult it is to teach students to think about equilibria and stability, it's spot on. And it's not just economics students who love linear, "A causes B causes C causes D"-type stories. Ecology students are exactly the same. And I too am guilty of giving in sometimes and telling the linear story, and the reaction of my students is exactly the same.

Great post.

[J.V. Dubois]

Phill: The Brad is right because he did not respond to the Bernanke he quoted. He responded to Quantitative easing conducted exactly in "concrete-steps" fashion, that is that increase in M will cause increase inflation (causality) - only expectations of CB reverting this in future will negate this very mechanical step.

So it is the very opposite to what you claim - DeLong endorses MM way of thinking. Only Brad thinks that the *only* credible way to convince the markets that the stimus is permanent is to make it physically permanent (like building bridges to nowhere). Market Monetarists disagree - the problem is not with CB (and the monetary policy it conducts) not being credible - it is that it is credible in its goal not to expand demand. It is DeLong who is horribly wrong. He assumes that CB will just watch by as fiscal stimulus "mechanically" does its magic to restore the economy.

PS: Another proof of DeLong and other Keynesians being firmly anchored in concrete steppes is that they do not work with expectations. Not really. Let's assume that for whatever reason (ZLB) they think that fiscal policy is the only game in town. Why could it not work via expectations? Like Obama saying - "Dear citizens, if the GDP growth does not recover by 2015 we will start to borrow and spend a lot of money so that the economy recovers, did I tell you my boyhood dreams about Mars expedition?"

[Britmouse]

Ritwik, on empirics, are there any central banks who have failed to hit the nominal variables they target? The SNB are hitting theirs, as are the BoE, ECB, etc; even with half of Europe disintegrating around them.

[K]

The SNB would have failed if they hadn't bought $60B Euros last month. Eventually the balance sheets of arbitrageurs will grow so full of Euro's that they will stop doing the SNB's job of satisfying the demand for Francs. Then the SNB will start buying. Then their balance sheet will get terryfyingly full of Euros too. Then they'll be scared, i.e. vulnerable. Then we attack.

[Ritwik]

Britmouse

1. SNB's monetary bazookas are not as costless as first imagined. http://www.macroresilience.com/2012/06/07/monetary-policy-targets-and-the-need-for-market-intervention/

2. BoE has openly admitted that it won't target a precise rate of inflation because it can't forecast it in the first place and has warned of higher inflation. http://www.reuters.com/article/2012/05/16/column-boe-inflation-idUSL5E8GGALG20120516
http://www.myfinances.co.uk/savings/2012/05/16/bank-cuts-growth-forecast-and-says-inflation-will-stay-above

3. ECB's target is to keep inflation below 2%. It's not really a precisely defined target. But even if we believe that it is, there are good reasons to believe that in high-wage developed economies, monetary easing as currently practised (rate cuts, QE) only leads to asset price and commodity price inflation and will not lead to a sustainable increase in aggregate demand. Thus, the ECB's asymmetrically conservative bias is not even a good test case for targeting success.
http://www.economonitor.com/danalperts2cents/2011/07/06/commodity-speculation-monetary-policy-and-temporary-inflation/
http://www.economonitor.com/danalperts2cents/2011/09/21/dear-paul-krugman-letter-inflation-would-be-dandy-but-try-to-make-it-happen/

[dlr]

First of all great post. If there's anything that non-macro macro pundits (i.e. investment managers) screw up most consistently when opining and predicting it is forgetting joint determination, particularly when it comes to supposedly exogenous interest rates. But...

I think you are probably mistaking some people who understand joint determination well for people of the concrete steppes. And the reason for this is that you are not taking the difficulty of the game theory between the Fed and the market seriously enough. I don't think it's enough to merely state that Chuck Norris can threaten to buy all the assets in the room. That's only the beginning of the question. The room is full of smart (though combat inept) people, who as a whole might well determine that Chuck is ultimately bluffing. After all, Chuck has his own incentives and his own weaknesses. Those weaknesses would come on full display long before he owned every asset in the world and started having to decide whether to vote out the HP board and how to rent out all those Vespas it owns. He would surely cave, or perhaps be relieved of his Chuckness by voters, or maybe would find that he was kicking his own butt around the room, at some point long before he could actually clear the room. And the room knows this.

And that's a wild understatement of Chuck's flaws and weaknesses. In the US, the Fed has the power to buy some unusual assets like foreign sovereign bonds and muni debt in some circumstances, and can attempt to exercise other powers creatively to buy even more stuff, but is it not at all obvious how far it can go in this pursuit before other government actors or lawyers or popular sentiment make it clear that whatever they are doing is putting their tenure -- and thus their power -- in jeopardy. I don't think it's quite enough to say Scott Sumner's "every CB that has wanted to inflate has succeeded" or to simply point out that Chuck is technically strong enough to beat everyone up if you ignore his potential weaknesses. This seems like an incomplete answer because "want" is nothing but a part of "can." Do we describe the BOJ as not "wanting" to inflate -- maybe -- but "wanting" might be a lot more complex than just changing the minds of a few chosen human beings. It is the interaction between the BOJ incentives and limitations and what the market knows about the BOJ incentives and limitations.

I think many of the Concrete steppes people take this problem seriously ("commitment problem"), and so implicitly assume that the Fed is working within some undefined limited equilibrium. And that is why they focus on the supposed hydraulics and are always so pessimistic about their effectiveness. In fact, it is exactly because they know about joint determination that they so pessimistic about many of the actual concrete steppes. Don't get me wrong, the concrete steppes are still the wrong conversation. It is distracting. But I don't know that orchestrating a preferred expectations equilibrium is as simple as just saying "printing press." The printing press needs power to run, and the market knows that. Saying that Zimbabwe can inflate doesn't seem like enough. It's easy to direct the expectations equilibrium if you're not bluffing about self destruction. If there are even potential, extreme scenarios where the Fed cannot go to the mat for its NGDP target and the market knows this, are those enough to create so called expectations or commitment traps?

[foosion]

Is monetary policy endogenous or exogenous?

A central bank, especially a rule based bank or the like, is responding to the conditions it influences

[David Pearson]

The stability/fragility tradeoff is an example of a non-linearity that macro does not capture. Credibly promise no small fires and the forest eventually succumbs to a conflagration.

Out in the steppes, actors translate a credible promise of stable NGDP into, "my VAR model says I don't need to maintain much capital or liquidity as a hedge." This eventually creates enough dry tinder to spark a conflagration. Like in a forest, a promise of no small fires leads eventually to a catastrophic one. Except its worse than the ecology example, as in the economy, actors rush to hedge once the perceive the promise is broken, creating a non-linear cascade -- a run. Thus, the 2007-2008 crisis was arguably caused by the Fed making a promise they could not keep ("no threats to the integrity of collateral"). The promise itself created a condition that made the promise non-credible.

[W. Peden]

K,

"assuming they can't buy anything other than government bonds"

There's no reason to make that assumption for any modern central bank, as far as I know.

"If they are allowed to buy real assets, then I think there is very little limit to what can be achieved via expectations"

Do private sector assets like corporate bonds qualify as "real assets" in this story?

Is this the only kind of liquidity trap you give credence to, K? Is it just a matter of short-term government debt and money supposedly being perfect substitutes when interest rates approach zero? (One of these days, someone is going to have to write a dictionary on liquidity traps. It would just be one entry with about 56 different definitions.)

[K]

W. Peden,

"Do private sector assets like corporate bonds qualify as "real assets" in this story?"

Corporate bonds qualify if they are positive beta, in which case buying them would lift the whole market. When and where spreads are very tight, corporate bonds are close substitutes for treasuries, but those times rarely qualify as periods of deficient demand, so when relevant I suspect the broad corporate bond market almost always qualify.

"Is this the only kind of liquidity trap you give credence to, K? Is it just a matter of short-term government debt and money supposedly being perfect substitutes when interest rates approach zero?"

Yes. In practice, I think buying corporate bonds and stocks would result in the required AD way before we have to debate whether the CB can further stimulate the economy by threatening to turn us into a communist state. Nick and Ashwin have both discussed such general traps. Like Ashwin, I can't dismiss it out of hand, but unlike the liquidity trap you describe, I don't have good theoretical reasons to believe in it.

[W. Peden]

K,

So, since in the real world central banks can and do buy private sector debt, the notion of a liquidity trap has no application in our current monetary arrangements. Therefore, we're in agreement that fiscal stimulus is superfluous*; at this point, I think the boundaries between New Keynesianism and Monetarism become extremely murky. Notice how it makes your qualification to your agreement to Nick's post irrelevant in all but high theory and to some extent practical policy i.e. it's a good argument for moving central banks to purchasing assets on the basis of independent and non-oligopolistic ratings agencies' assesments of quality, rather than assuming that government debt should be specially privileged..

* Although it makes sense to increase government investment during recessions (demand-side or supply-side) on the grounds of good public finance: it's cheaper to invest in a downturn and the government isn't constrained in doing so, provided it is prudent when revenues are strong.

[Mike S]

Nick,

I really liked your last post with Aswin. You can guess where I came down on that one...

But I cannot agree with this.

"Any sufficiently advanced technology is indistinguishable from magic"

I think this is where we are heading here. Our system has plumbing, it has numbers, it has...more information. Going in this direction dooms the outcome to Tyler Cowen economics "We can't do anything or ever tell what's happening, so why bother?"

[Determinant]

Economics = Social Sciences = B.A. = Artsie.

Economists are not part of The Club. Sorry Nick.

Says he who has a B.Eng.

[Min]

Don't you need a linear story to show convergence? :)

[K]

W Peden: "in the real world central banks can and do buy private sector debt"

There's plenty of doubt about the legal and political limits of that, both in the US and the EU. Certainly the Fed is prohibited from buying corporate bonds. Maybe they'll be able to structure their way around it, maybe they won't. The odds that they can't are, in my opinion, high. If the market thinks like I do, then we have significant odds of still ending up in a trap.

Do I think we need fiscal policy then?

First of all, buying real assets *is* fiscal/industrial policy as targeted risk transfers can be every bit as much a bailout as lump sum transfers. If you are going to pick and choose real assets, I request that you start by buying real estate in my neighbourhood.

Second, no, I don't think we *need* it. The 10yr bond can still rally by $16 before rates are at zero. The Fed hasn't even *tried* to communicate the conditional path of rates. How about "we'll keep rates at zero until the NGDP level reaches 5% compounded from Jan 2008. Then we'll do what it takes to keep it growing at 5%." Real rates will *plummet* and *all* real assets will rally. If that doesn't work, then we will have to try something else. Meanwhile, lets leave negative beta treasury bonds in the hands of investment portfolios where they can enable the maximum amount of real risk taking. Treasury bond QE is totally bogus.

[rsj]

Equilibrium thinking is not "non-linear", it is utopian. It's a slur to the noble field of non-linear dynamics to call the assumption that prices and quantities at the national level can be solved by drawing a few supply and demand curves and assuming that all transactions occur at the intersection of these curves. It is just panglossian bullshit.

In the absence of an auctioneer, you need to look at actual mechanisms that arise from purchases and sales, and see how the economy evolves as a result of millions of these purchase decisions occurring in real time -- rather than look at some idealized equilibrium, and assume the economy is following this magical path continuously.

[W. Peden]

K,

"Certainly the Fed is prohibited from buying corporate bonds."

?!

Why?

And doesn't this cast serious doubt on Paul Krugman's professional qualifications to advise the UK on macroeconomic policy, given that he probably isn't aware that the Bank of England can (and does) buy private debt like corporate bonds?

"First of all, buying real assets *is* fiscal/industrial policy"

Then we're in the interesting position where much of what is known and conducted as "monetary policy" in much of the world isn't monetary policy in the theoretical definitions here. I suppose that, as usual, reality will be made to accomodate theory rather than vice versa.

[W. Peden]

(I only pick on Paul Krugman because he's been over here recently telling us how to run things. Milton Friedman did the same thing in the early 1980s and also showed an embarassing lack of familiarity with UK institutional arrangements.)

[rsj]

?!

Why?

There is this thing called the "Federal Reserve Act" that specifies what the CB is and is not allowed to buy. And to whom it may or may not lend.

Because purchasing risky assets bears a possibility of loss, which corresponds to fiscal policy, and the constitution specifies which branch of government is required to do fiscal policy. Said branch tends to be jealous about delegating its powers to other institutions when it writes laws. That is why the CB can only purchase assets backed by the government. In that case, there is no risk of loss to the government.

You may also be surprised to know that there is this thing called a "Debt Ceiling", which Treasury is not allowed to exceed. Once again, prior to the creation of the debt ceiling, each bond sale required an act of congress. When this got too burdensome (at the time of the Panama Canal), Treasury was allowed to borrow up to a limit set by congress.

Once again, congress controls fiscal policy, it may allow another institution to tweak it on the edges a bit, but basically it is not going to give real fiscal authority to anyone other institution.

Attempts to do back-handed fiscal policy via the CB run into this problem. In the real world.

[Determinant]

*cough*

Strictly speaking, linearity means f(x+y) = f(x)+f(y) AND f(ax) = af(x) for all a.

f(x) = mx+b fails this second test because the test test implies that f(0) must be 0, though in popular usage (popular in scientific circles even) y=mx+b is considered linear, even though f(0) != 0.

[Diego Espinosa]

W. Peden,
A suggestion: if the activity does not stem from a unique capability of the central bank, it is not "monetary policy". It is just the action of a publicly-owned bank with a rather large balance sheet. The Fed has two unique attributes: it issues dollar reserves, and its liabilities are irredeemable. When dollar reserves are not demanded for payments, or when the liquidity premium is normal, the only monetary policy tool is signaling regarding future actions (ok perhaps also a negative IOR, but it is, strictly speaking, a tax on certain bank assets that Congress could also impose, which is why its probably illegal for the Fed to implement).

For instance, the Fed swapping ER's for risk assets simply results in a transfer of risk from private actors to taxpayers. It can be exactly replicated by Treasury (swapping Tbills for risk assets). Of course there is a signalling component: signaling that the Fed will engage in fiscal policy until its monetary policy goals are achieved.

[Nick Rowe]

Determinant: "Strictly speaking, linearity means f(x+y) = f(x)+f(y) AND f(ax) = af(x) for all a."

I didn't know that. But, AFAIK, Y=a+bX is what economists call "linear".

Some economics degrees are BSc's (but not at Carleton). But yeah, we're members of the Wannabe tribe, and don't think of ourselves as artsies!

rsj: Not all equilibria are Walrasian equilibria. E.g. there are search equilibria, where there's no Walrasian auctioneer. And not all equilibria are good. E.g. there's the Prisoners' Dilemma Nash Equilibrium. And see the comment by Jeremy Fox above? Jeremy is a real scientist, and he says he teaches the same concepts in ecology. And I don't think critters use a Walrasian auctioneer to get to a utopian/Panglossian predator/prey equilibrium.

Min: "Don't you need a linear story to show convergence? :)"

Funnily, I was just wondering the same thing myself. I don't think so, is the answer I came up with. More generally, there's no rule that says economic theories *can't* be linear. But I confess the only linear theory I can think of offhand is the old Labour Theory of Value. Socially Necessary Quantity of Labour arrow value. There must be others, that aren't defunct.

Mike S: "Going in this direction dooms the outcome to Tyler Cowen economics "We can't do anything or ever tell what's happening, so why bother?" "

I don't think it's quite that dire. Sure, it's harder to understand, explain, measure, and test, but it's not impossible. And we can look for analogies in history around the world. FDR's announcing a price level target, and raising the price of gold until he hit it, is one that I found reasonably convincing, for example. The gold price probably now lacks the salience to work as an intermediate target and indicator, of course, but there are others.

David Pearson: "The stability/fragility tradeoff is an example of a non-linearity that macro does not capture. Credibly promise no small fires and the forest eventually succumbs to a conflagration."

(I'm not sure if you are maybe using "non-linearity" in both senses there.)

Ashwin's blog has some interesting stuff related to that idea, I think. Haven't fully got my head around it yet, but I reckon he's onto something.

[Jesse]

In Engineering land Y= a + bx^2 is linear (of 2nd order). Y = a + x^b is non-linear. Linearity in the parameters, not the plot on the graph...
There are lots of nice math and proofs for linear systems. Some non-linear ones work as well, but it is much harder to have general results for non-linear systems. Linear systems also generally allow superposition, where a change in two different inputs, or contributors to inputs can be considered separately, and then added together.

[rsj]

But in the sciences, while you are certainly interested in the equilibrium points (and phase diagrams, etc), you do so because your behavioral equations produce the equilibria.

In economics, you write down the equilibria based on *preferences* (e.g. "demands" or the solution to optimization problems) not behavioral relationships, and then assume that the equilibrium is achieved.

That is why I call it utopian.

If you were to derive ODEs from behavior, and then find an equilibrium, it would be more believable, and it would be comparable to what is done in the physical sciences. I would not call that type of analysis Panglossian.

[Nick Rowe]

foosion: "Is monetary policy endogenous or exogenous?

A central bank, especially a rule based bank or the like, is responding to the conditions it influences"

It's both. Or rather, it depends on what you mean by "monetary policy", and how you model it. For example, suppose you model monetary policy as a sort of Taylor Rule, where the central bank sets i=a + b.inflation + c.Ygap. Then you are saying that i is endogenous, but the parameters a, b, and c are exogenous. In that context, does "monetary policy" mean i, or the whole Taylor Rule?

If you think something is endogenous, you need to explain it in terms of things you consider exogenous.

dlr: Thanks!

"I think you are probably mistaking some people who understand joint determination well for people of the concrete steppes."

Yep. We need to keep them distinct. Someone who (for example) argues that future central banks will push future NGDP all the way back down to where it is today, regardless of what the current central bank says and does, and that people will expect this to happen, so that's why it won't work, is not someone from the concrete steppes. I just don't think that's plausible. FDR's example suggests it isn't either. But it does all depend on what people understand by "doing nothing".

Remember too, it's not a binary game. Chuck doesn't need ex ante 100% certainty of 100% success in order to hit his target. 10% probability of 10% success increases E(NGDP) 1% of the way towards the target, which will be enough to show up in asset prices, and raise current NGDP a bit, which creates a positive feedback effect (aka multiplier), and causes people to revise upwards their expectations. And we've already seen how the mere announcement of finite QE, even without a target, affects markets.

[Jeremy Fox]

rsj: Just in case you didn't know, there are actually models in ecology and evolution, and probably other fields, in which organisms have preferences (e.g., predators choosing among various prey types so as to maximize their rate of energy gain and thus fitness). You can still derive ODEs from assumptions about those preferences, and then find equilibria and other attractors, analyze their stability, etc.

I'm sure Nick will correct me if I'm wrong, but it's my impression that economists often do the same thing. That they prefer to work with the phase diagrams doesn't imply that they've never derived those diagrams from underlying equations.

Now, if you want to complain that economists generally make very unrealistic assumptions in order to derive their equations, well, there are often good reasons to complain (which doesn't mean there aren't also good reasons why economists make those assumptions in the first place; models have many useful purposes besides "duplicate reality"). FWIW, you can make the same complaint about ecology and evolution, and many have.

[rsj]

Jeremy, predators choosing among which prey to pursue is a behavioral assumption. Once you have behavioral assumptions, you can run the system and see what the behavior leads to. The fact that the behavioral assumptions are explained by preferences is beside the point.

You still have:

preferences --> behavior --> differential equations --> find equilibria, dynamics. (non-panglossian)

Rather than

preferences --> differential equations to maximize satisfaction --> find equilibria (Panglossian)

That is very different from optimizing based on preferences, and assuming that the system follows that optimal path.

In the case of predator-prey, a preference-type solution would be for the wolf to look ahead and realizes that excess hunting leads to a poor outcome, and therefore select the appropriate quantity of sheep to eat to maximize their expected lifetime satisfaction.

And then assuming that that is what we would observe in nature. When someone points out -- no the wolf will overconsume and then perish, you do not have an army of biologists accusing the modeler of ascribing irrationality to wolves.

[K]

Nick,

I'm willing to bet that you too, are "of the concrete steppes."

Imagine three different central banks:

Bank One is the Fed

Bank Two is like the Fed but has absolutely unequivocal rules sayin that their only powers are to create reserves in exchange for treasury bonds

Bank Three is like Bank Two but can only do overnight repos and buy IBM stock. No treasury bonds.

Bank Four can only do overnight repos against treasury bonds.

Bank Five can't do anything but talk.

All of these central banks are free to say whatever they want in order to affect expectations in order to achieve their target.

Which of these banks could successfully hit an NGDP level target at all times. I say none of them. I'm willing to bet that you don't believe in Bank 5 and possibly also Banks Four and Three and even Two? No?

[K]

Oops! *Five* banks!

[Nick Rowe]

K: Bank one won't talk (at least, not about anything worth saying).

You haven't said Banks two to four can talk.

Maybe Bank three could "talk", in the same way bridge players "talk", by bidding for IBM shares? The IBM share price is a much better language than Tbill prices, because the IBM share price will be positively correlated with E(NGDP).

I would pick Bank Three.

Bank Five isn't a central bank. It can't print money.

[W. Peden]

rsj,

"There is this thing called the "Federal Reserve Act" that specifies what the CB is and is not allowed to buy."

I know that. My question is: why determine purchasable assets on the basis of the type of debt issuer rather than the quality of the asset?

"Because purchasing risky assets bears a possibility of loss, which corresponds to fiscal policy, and the constitution specifies which branch of government is required to do fiscal policy."

Sure, I understand the issue of purchasing risky assets. However, government debt is not riskless and private debt is not always riskier than government debt.

(I know that the actual REASON for the rule is that it artificially boosts demand for government debt. I'm just interested in whether or not there are any covering arguments.)

"You may also be surprised to know that there is this thing called a "Debt Ceiling", which Treasury is not allowed to exceed."

It's only a slightly less ridiculous one of those quaint antique institutions that characterise the US (as opposed to the hyper-modern United Kingdom! /irony) which puzzles me only slightly less than the rules you put on the Fed. The debt ceiling has never stopped debt growing above the prescribed levels and its key function seems to be to create even more infighting between the legislature and the executive than would already occur in the US system. I'd never thought about how it might hobble effective monetary policy, though.

"Attempts to do back-handed fiscal policy via the CB run into this problem. In the real world."

You misspelt "America" as "the real world". ;)

Diego Espinosa,

"A suggestion: if the activity does not stem from a unique capability of the central bank, it is not "monetary policy". "

No, I think that definition is deficient for several reasons: (1) in one sense, the central bank has no 'unique capabilities', separate from those monopolistic arrangements created and enforced by the state; (2) it differs from established usage, which identifies monetary policy with interest rates/monetary base operations; (3) in most countries today, the important practical distinction which 'monetary policy' and 'fiscal policy' should capture is between those actions which are carried out by the central bank and those operations which are carried out by the finance ministry.

Notice how this definition leads you, a very clever person, to say something like-

"Of course there is a signalling component: signaling that the Fed will engage in fiscal policy until its monetary policy goals are achieved."

- perhaps that's right and perhaps that's wrong. I have no idea, because I don't know what "monetary policy goals" means in this sentence.

Still, my point isn't primarily about definitions, but about macroeconomic policy. I worry that local institutional arrangements in the US can make people think that, for example, the implications of restrictions on the Fed apply to the BoE and the ECB.

[Determinant]

It's somewhat simpler in Canada, where the Bank of Canada Governor and the Finance Minister have offices down the block from one another. The Bank sits on Wellington Street just opposite the West Block of Parliament Hill, Finance is headquartered in L'Esplanade Laurier. If the Governor of the Bank of Canada needs something he can walk down the block, ask the Minister for it and the Minister can then introduce a bill in the House if he needs to.

Extra points if they both walk to one of the bars on Sparks Street afterwards.

[Diego Espinosa]

W. Peden,

Perhaps its more accurate to say that a profitable risk asset purchase by the central bank is likely to be identified as "monetary policy"; while a loss-making one will ultimately be labeled "fiscal policy". This is because when a taxpayer is presented with a bill for a central bank recapitalization, it is tough for him to escape the notion that his tax liability can only rise as a consequence of his government spending money. My sense is German taxpayers are about to realize that the ECB is quite capable of non-appropriated fiscal spending.

Of course, the central bank is exposed to losses in its LOLR function. This is a special case: the central bank controls liquidity premia by virtue of its uniquely irredeemable deposit base. Specific legislation imbues the central bank with the power to appropriate (lend) funds in cases where liquidity premia are high. According to the spirit of the legislation, in no other case should a central bank create contingent tax liabilities for taxpayers.

[Jeremy Fox]

rsj:

You kind of lost me in your reply to my last comment. Afraid I don't see what you're getting at. In many evolutionary models in particular, organisms are indeed assumed to maximize, or try to maximize, the evolutionary equivalent of "lifetime utility" or whatever economists call it, namely fitness.

I think you'd be very interested in Alan Grafen's "formal Darwinism" work linking dynamics (births and deaths) and optimization, and specifying with great precision the conditions under which individual organisms can be regarded as trying to optimize something.

You seem to be questioning whether organisms other than humans have expectations about the future that shape their actions in the present. They sure do! Even plants do. Predictive germination is an example. Plant seeds can actually "learn", based on cues present in the environment in early spring, whether it's likely to be a good or a poor growing season for them, and then "decide" accordingly whether to germinate, or remain in the ground until next year (when they'll make the same decision again).

And yes, there are situations in which predators eat all their prey and then starve. It actually happens in my lab all the time (I work on protozoan predators and prey, but they're far from the only example). It's a sort of tragedy of the commons, which optimizing organisms will be selected to avoid only in rather specific circumstances. Natural selection doesn't routinely favor what are known as "prudent predators" in the evolutionary literature.

[W. Peden]

Diego Espinosa,

"Of course, the central bank is exposed to losses in its LOLR function. This is a special case: the central bank controls liquidity premia by virtue of its uniquely irredeemable deposit base. Specific legislation imbues the central bank with the power to appropriate (lend) funds in cases where liquidity premia are high. According to the spirit of the legislation, in no other case should a central bank create contingent tax liabilities for taxpayers."

So the LOLR is indeed fiscal policy, under this definition? I see...

[W. Peden]

Determinant,

Back when the UK Chancellor took explicit responsibility for monetary policy, he could almost instaneously get what he wanted from the finance minister by asking himself (politely, of course- this were Englishmen).

If he needed the backing of the govermment to pass new legislation, he could theoretically get it via morse code by banging on the walls through to the Prime Minister living next door. Hence the phrase "Westminister village".

[rsj]

W.P,

You need to read the act. Short summary is that for bills, there is a bit of latitude as to what can be purchased, although notes funding the purchase of equities are expressly forbidden.

For bonds, it needs to be guaranteed by the U.S. There are some exceptions, but these have lots of strings attached.

However, government debt is not riskless and private debt is not always riskier than government debt.

LOL. It's not risk as modelled by a finance book that we care about. It is risk of loss to the government that we care about.

The Treasury has a claim on all Fed profits, and the Treasury must pay for all Fed losses.

Therefore any debt purchased by the Fed potentially exposes Treasury to a risk of loss, with one exception -- the purchase of treasury's own debt, or debt guaranteed by the government more generally.

The CB was specifically designed to not be a risk bearing agent, and to not lose money, despite all the kooky proposals (such as the CB selling NGDP futures) that assume it is OK for the CB to do exactly that.

Making investments and risking losing money is what the government does when it engages in public works projects and the like. There is enormous opportunity for fraud as well as favoritism, and Congress wants to keep that power to itself.

Whether you personally believe that some private sector debt is less risky than Treasury debt is beside the point.

[K]

Nick: "Bank one won't talk (at least, not about anything worth saying)."

:-)))) LOL!

I did mention that they are all free to say whatever they want. How about bank 4? Can they always hit their target?

[rsj]

Jeremy, I am not saying that plant behavior cannot be predictive and does not reflect preferences in some sense. Nevertheless, it is the behavior that you are modeling. The wolf's preference is to eat a full meal each day. But the aggregate behavior of all wolves may cause them to starve. Biologists acknowledge this fact because they observe it. They do not rule it out purely on principle.

In the same way, pre-panglossian economists observed that an excess of savings results in a decline in income and did not rule it out on principle. We can observe it even today.

[Jeremy Fox]

rsj: Um, I'm not aware of economists as a group ruling out things like tragedies of the commons or other market failures "on principle". I mean, maybe there are bad economists out there who simply assume optimal outcomes at the aggregate level, but my admittedly-limited experience hasn't brought me in contact with any. Nick for one and Paul Krugman for another have plenty of posts about how aggregate behavior of collections of people can't be understood by thinking of the aggregate as analogous with a single person ("fallacy of composition" and all that). So I'm afraid my limited reading in economics hasn't equipped me to follow what you're talking about.

I'm going to refrain from pursuing this any further, as at this point I think we're just talking past one another.

[Determinant]

Remember too, it's not a binary game. Chuck doesn't need ex ante 100% certainty of 100% success in order to hit his target. 10% probability of 10% success increases E(NGDP) 1% of the way towards the target, which will be enough to show up in asset prices, and raise current NGDP a bit, which creates a positive feedback effect (aka multiplier), and causes people to revise upwards their expectations. And we've already seen how the mere announcement of finite QE, even without a target, affects markets.

You had to bring Chuck into this, didn't you Nick? I hate Chuck. Chuck Norris, meet http://en.wikipedia.org/wiki/John_Byng, whose example caused Voltaire to write "Dans ce pays-ci, il est bon de tuer de temps en temps un amiral pour encourager les autres – "In this country, it is wise to kill an admiral from time to time to encourage the others."

England's sea dogs were kept bloodthirsty by the fact that they could be charged and shot for "Failure to do their utmost". From time to time the threat must be carried out in order for it to maintain credibility.

[W. Peden]

rsj,

I'm actually currently struggling to maintain any interest in the particular arrangements of the US central banking system, except insofar as they distort the language of economics.

If exposing the Fed to risk was the real rationale, then why is it able to conduct other risky activities, including its LOLR function? Arguably, because these activities are necessary to perform its function adequately. So why is effective demand management at the zero level bound less worthwhile than lending to banks who can't borrow on the market? (If we make the assumption that treasuries and base money are indeed perfect substitutes under current circumstances.)

As I said, the real reason is the one that explains the existence of central banks in the first place and much of the history of reserve requirements: these are means to make it easier for governments to finance expenditure without (overtly) taxing their subjects. It would be interesting, however, if there were arguments other than a bogus risk argument, because then it would suggest than the purchasing of (say) corporate bonds by the BoE, ECB etc. were somewhat importantly flawed.

As for fraud: the eurozone's problems have largely been the result of government fraud and riskiness. The ECB's holdings of private debt have not been problematic, as far as I know. However, the ECB is different in many, many respects from the Fed, which is generally my point.

[beowulf]

rsj, the Fed is a govt agency and yet is not subject to the govt debt limit. What's more the Federal Reserve Act isn't the key statute, the relevant one is the Gold Reserve Act of 1934 as amended.

"Consistent with the obligations of the Government in the International Monetary Fund on orderly exchange arrangements and a stable system of exchange rates, the Secretary or an agency designated by the Secretary, with the approval of the President, may deal in gold, foreign exchange, and other instruments of credit and securities the Secretary considers necessary."

The question of whether "instruments of credit and securities" includes shares or debt of publicly trade comnpanies has a pretty clear-cut answer earlier in the statute--- S.T.F.U. ("Decisions of the Secretary are final and may not be reviewed by another officer or employee of the Government."). The Secretary's discretion is so broad as to edge into unlimited-- remember he can designate a govt agency to act instead of (or in conjunction with) Tsy itself . Imagine for a minute there were a go agency that, for whatever reason, was NOT subject to the govt debt ceiling. Yeah I can't think of one either. :o)

"Treasury and Federal Reserve foreign exchange operations are closely coordinated and typically are conducted jointly. Operations on behalf of the Treasury are made under the legal authority of the Secretary of the Treasury and those for the Federal Reserve System under the legal authority of the Federal Open Market Committee, the central bank's policy-making group. The ESF does not provide financing to the Federal Reserve System for foreign exchange operations. Rather, the Federal Reserve participates with its own funds."
http://www.newyorkfed.org/aboutthefed/fedpoint/fed14.html

[Account Deleted]

I think that even market monetarists have to admit that there is a range of feasibility of nominal outcomes that can be achieved by monetary policy. The easiest to hit would be money supply targets, followed by currency ceilings where the central bank can buy any foreign currency with its essentially unlimited printing capacity. Targeting indices would be after that.

Aggregate NGDP targets feel more difficult because it is unsure as to how the target will be achieved.

Per-capita NGDP and Wage targeting feel even more difficult. The number of "concrete steps"/decisions to be taken by actors who are out of your control, seem to be more on the way to a wage target .

However, if market monetarists believe in expectations setting overriding eveything, then the central bank should target something that has the distributional aspect also taken care of. Targetting the median wage or even better the 25th percentile wage should be what they aim for. Do you disagree?

[Determinant]

I agree with you about the Fed, W. Peden. This is supposed to be a Worthwhile Canadian Initiative, after all.

[Peter N]

@ W. Peden

"The ECB's holdings of private debt have not been problematic, as far as I know."

YET.

If you throw a trillion euros or two (or three) at a bunch of insolvent banks, there is a certain amount of repayment risk.

The ECB is entering new territory (at least in modern times) of central bank risk for a major state with its own currency. This could blow up in their faces. The Fed may have unlimited ammunition (or not), but the ECB probably doesn't anymore (if it ever did).

[Peter N]

It's confusing, unnecessary and imprecise to create your own terminology and repurpose existing terminology. Possible terms to choose among:

Non-monotonic
Multivalued
feedback and feedforward with and without delays and nonlinear feedback functions
differential equations including nonlinear and delay and matrices thereof
sensitivity
distinctions of equilibria including multiple, unstable, and nonexistent
stability and criteria for same
distinctions between steady state, finite difference and differential models and their possible nonequivalence.
The difference between uncertainty and risk and its economic effects
The effects of different risk distributions (acual and assumed) on the behavior of the financial system
continuity
phase space
bifurcation, attractors, chaos, randomness, and the concept of chaotic control
domains of validity for models
and from economics stocks, flows and integrated flows and the problems from confusing them, with reference to the use and abuse of GDP

There is an abundance of useful terminology and correct representation available from mathematics, physics and engineering, if you choose to use it. If your representations don't serve to communicate, maybe they're the wrong representations.

"... as simple as possible, but not simpler"

[Nick Rowe]

Prakash: "However, if market monetarists believe in expectations setting overriding eveything, then the central bank should target something that has the distributional aspect also taken care of. Targetting the median wage or even better the 25th percentile wage should be what they aim for. Do you disagree?"

Some MMs do sometimes wonder about targeting (the time-path of) average nominal wages, instead of NGDP. It probably wouldn't make a lot of difference if we targeted the median wage instead of the average (unless the skewness of the wage distribution changed a lot over time for other reasons, which I expect it might).

I think (and I'm pretty sure other MMs would agree on this) that targeting the median wage rather than NGDP would probably have no effect on the long run distribution of income. It would probably have some effect on short run cyclical fluctuations in the distribution of income, but I can't figure out whether it would make inequality more procyclical or countercyclical (maybe I could if I spent a lot of time thinking about it). But it would probably not reduce long run inequality relative to an NGDP target.

[Peter N]

News flash from zerohedge:

"Fortress Paper Ltd. announces that its wholly-owned subsidiary, Landqart AG, a leading manufacturer of banknote and security papers, has had a material banknote order reinstated. This order was unexpectedly suspended in the fourth quarter of 2011 which negatively impacted the financial results of Landqart's operations in the first half of 2012."

Step by step to a crackup.

[W. Peden]

Determinant,

That raises a further question: are there any parallel restrictions on the Bank of Canada? Does it hold private debts? (I confess a lack of knowledge about Canadian monetary institutions- I only did Canadian Studies for one year and while there was a bit of economics we didn't get all the way to monetary policy!)

Peter N,

Even then, with the exception perhaps of the Spanish banks, the problems for most European banks are their holdings of sovereign debts, because in an important sense (monetary sovereignty) the eurozone states are not sovereign.

I agree about new territory. Of course, the ECB has always been a pioneer institution: never before have we had a central bank without a state. It was created to facilitate the formation of a European state, rather than to help the politics of an existing state (like a normal central bank). Cart before the horse, as it turns out. However, it's only been in tough times that the ECB's peculiarities have become important.

Anyway, apart from the fact that a government can't suffer losses from its holdings of its own debts (I am no poorer if I write a $1 trillion IOU to myself and then burn it, except from one piece of paper and one match) the eurozone crisis shows us that government debts can be just as dangerous as private debts and that this isn't just a problem of third-world defaulter nations like Argentina and Zimbabwe. If we're going to have central banks, then they should generally choose the assets they buy on the basis of independent ratings. That would give an incentive to end some of the monopolistic elements in the ratings agency business, which would benefit parties quite apart from the central bank.

[Nick Rowe]

Peter N: "It's confusing, unnecessary and imprecise to create your own terminology and repurpose existing terminology."

Agreed. And I did wonder about using at least one of those alternative terms, "non-monotonic", but I thought that wasn't exactly right either, and it might be even more confusing. In this case, since a large part of the population are non-economists, and non-scientists, and do use "non-linear" in roughly the Artsie sense, and we do hear complaints from those people that "economics is linear", I thought it best to go with their existing usage.

My guess is that a greater part of the whole population uses "linear" in something like the Artsie sense than in the Y=bX sense. Even if I'm wrong on that guess, it's not a trivial proportion of the population.

[Peter N]

The public may be a lost cause at the moment, but economy begins at home. Students deserve better.

For instance, how does supply and demand work in the automobile market? Here you have a rich but well understood pattern of supply and demand. Students would see economics as it actually works. For instance, past a certain point automobile inventories are a wasting asset. Why do manufacturers overproduce, fill dealers' lots with 120 days of inventory and end up having to give rebates? What exactly does retail price mean? What effect does lobbying have on the market (the dealer market is not free).

A lot of the confusion is of economists' own making. The imprecision of undergraduate education pollutes the whole system. After all, today's students are tomorrow's economists, businesspeople, politicians...

Especially to be avoided is appeal to obviously true general principles that are not obvious, general, provable or sometimes even true. Rational expectations would be a case in point. Its only defense is that it makes modeling easier (though not more correct). Is it an emergent phenomenon of large numbers of irrational agents? How would you prove (or even plausibly demonstrate) this?

[K]

W Peden:

"If we're going to have central banks, then they should generally choose the assets they buy on the basis of independent ratings."

Why are you talking about buying risky assets? None of the central banks under consideration (Fed, ECB, BoC) can buy private assets. Only government guaranteed ones. I'm not sure what you are talking about. The ECB can't even do QE. The only literal buying they are allowed to do is limited to the demand for paper money. Anything else is banned as debt monetization.

There are good reasons why this is done, as rsj has explained. There are also good reasons why we should clearly delineate the boundaries of the CB for the purposes of our discussion. Otherwise it's pointless. The question you seem to be discussing is "what are the limits of the power of expectations management for a central bank that has the ability to do whatever it wants". Or perhaps, that has the ability to purchase assets of a quality only known to W Peden. Or how about, allowed to do just enough to make the claim that they are omnipotent true, whatever that happens to be. The rest of us should be excused for thinking that that is not a terribly interesting conversation.

The trouble with permitting risky asset purchases is that you can securitize just about any risk you want, including almost any imaginable fiscal action. How about the CB buys loans to people whose unemployment benefits have expired? That should work.

[Peter N]

@K
"The only literal buying they are allowed to do is limited to the demand for paper money. Anything else is banned as debt monetization. "

What was that last 900 billion euros of LTR, then?

[W. Peden]

K,

I was under the impression that the ECB had been buying covered bonds and the Fed had been buying mortgage-backed securities. Perhaps I'm mistaken.

[JustBob81]

I often run into this problem teaching American Politics to undergraduates. Explaining to students that the relationship between interest groups, congress, and the bureaucracy is a complex interplay of interests, information, and power is something they find deeply unsatisfying. If you reduce it down to some kind of vulgar exchange they perk right up.

[K]

W Peden,

No, actually you're right on both counts. They both have limited abilities to buy some private assets. But it doesn't change the rest of my points. It's fiscal powers that have the potential to rescue the economy from the liquidity trap. We actually do have to understand the institutional constraints and the extent to which the CBs have fiscal powers in order to understand the limits of central bank powers. And it's not "monetary policy" because the quantity of money has nothing to do with it. It's only about the asset side.

[W. Peden]

K,

Money is an asset, so changing the mix of assets between base money and private sector assets has everything to do with the quantity of money.

Whether central banks' capacity to get out of liquidity traps is 'fiscal policy' or 'monetary policy' is entirely a matter of the words one chooses to use. Yours is about the third definition of 'monetary policy' offered so far and there are probably a dozen we haven't used yet. The important point is that classical fiscal stimulus of the type that Keynesians dream about is unnecessary, even if one makes the assumption that treasuries and base money can be perfect substitutes.

I wouldn't mind if monetary policy was entirely conducted through postive/negative IOR, if possible. It's just, as it happens, central banks can always increase demand and the Keynesian liquidity trap has no application to real world economies. That's distinct from Keynes's liquidity trap; whether Keynes was right depends on your theory of liquidity preference. (Like 'monetary policy', there are about a dozen meanings of 'liquidity trap'.)

[Dan Kervick]

Feedback is compatible with causal precedence and temporal precedence.

Nonlinearity is compatible with causal precedence and temporal precedence.

It is a mistake to regard systems as acausal or regard the causal relations as temporally indifferent simply because the distances in the causal-temporal sequencing of events in these systems might in some cases lie below the threshold that can be practically measured, or that the inquirer has chosen to measure.

It can be useful to approximate systems with a level of description in which some macro-parameters are in a steady state. But one can almost always ascend to a more detailed description of the system in which that parameter is refined into one that is no longer steady. Just think about the predator-prey example Fox used.

Crude global measures and descriptions can be useful - until they aren't. When the debates concern the causal details, different accounts of which generate different predictions about results, then it is not a sufficient response to say that you are attached to temporally indifferent and acausal equilibrium models that only approximate.

There is nothing in the world of human affairs which is analogous to the atemporal causal dependence of systems such as those studied in classical Newtonian dynamics.

This looks to me like a lot of pop scientific hand-waving fluff for not wanting to think about challenging causal questions.

[K]

W Peden: "Money is an asset, so changing the mix of assets between base money and private sector assets has everything to do with the quantity of money."

No, nothing. Treasury can get the identical effect by swapping t-bills for the same real assets. If the quantity of the medium of exchange has *nothing* to do with it, then it's odd to call it monetary policy.

"classical fiscal stimulus of the type that Keynesians dream about is unnecessary, even if one makes the assumption that treasuries and base money can be perfect substitutes."

Of course, to the extent that the central bank digs out of the liquidity trap via fiscal policy, it's exactly as distortionary as when treasury does it. Just without the democracy bit.

[Nick Rowe]

Dan: "It is a mistake to regard systems as acausal..."

I am certainly not saying it's acausal. The supply and demand curves cause price and quantity. The mistake is made by those who confuse the absence of a uni-directional uni-temporal causal sequence with the absence of causality. Not every type of causation is like billiard balls.

[W. Peden]

K,

They can, but they don't, so it isn't. If someone kills someone else by clubbing them over a head with a cricket bat, it's puerile to say that cricket bats have nothing to do with it because they could have used a baseball bat instead.

"Of course, to the extent that the central bank digs out of the liquidity trap via fiscal policy, it's exactly as distortionary as when treasury does it."

No, unless the treasury conducts the fiscal stimulus simply by buying highly rated assets. That would actually avoid many of the distortionary problems of fiscal stimulus, but if you're going to have a fiscal stimulus then the least distortionary way to do it is simply to underfund existing expenditure (since changes in M = PSBR - sales of gilts + bank lending to the non-bank private sector - non-deposit liabilities - the external balance) to the extent that is needed, which will probably be enough to do the job all by itself.

[rsj]

If exposing the Fed to risk was the real rationale, then why is it able to conduct other risky activities, including its LOLR function? Arguably, because these activities are necessary to perform its function adequately. So why is effective demand management at the zero level bound less worthwhile than lending to banks who can't borrow on the market? (If we make the assumption that treasuries and base money are indeed perfect substitutes under current circumstances.)

Because the central bank was not created to be a economic agent, nor a demand management agent.

It was created at the urging of the banks to ensure the stability of the banking system. They needed a reserve bank (a bank that could settle payments between other banks and lend to other banks) that could always deliver in times of crisis. The previous reserve banks, such as the Bank of Suffolk, weren't always big enough to guarantee an end to bank runs.

There was no larger economic role at all, other than the role of preventing bank runs. The residual, and difficult to detect "demand management" role of monetary policy came much later.

The government, then as well as know, primarily manages demand via fiscal policy. That is why, for example, we have unemployment insurance, food stamps, and other fiscal stabilizers, without which we would be in a depression.

Nick's brand of monetarism died long ago when it was tried and failed. There are no plausible mechanisms by which the CB can target (or even control) the higher monetary aggregates, and no evidence of these aggregates playing any structural role in the economy, we are left with fairies, tinkerbells, and pure obstinance. Everyone else has moved to viewing the central bank as setting the OIR, rather than quantities.

Here is an example of billiard balls -- e.g. actually looking at what people are doing and how the institutions work in order to try to understand the economy -- from the Fed!
http://www.ny.frb.org/aboutthefed/fedpoint/fed49.html:

"Following the introduction of NOW accounts nationally in 1981, however, the relationship between M1 growth and measures of economic activity, such as Gross Domestic Product, broke down. Depositors moved funds from savings accounts—which are included in M2 but not in M1—into NOW accounts, which are part of M1. As a result, M1 growth exceeded the Fed's target range in 1982, even though the economy experienced its worst recession in decades. The Fed de-emphasized M1 as a guide for monetary policy in late 1982, and it stopped announcing growth ranges for M1 in 1987.

By the early 1990s, the relationship between M2 growth and the performance of the economy also had weakened. Interest rates were at the lowest levels in more than three decades, prompting some savers to move funds out of the savings and time deposits that are part of M2 into stock and bond mutual funds, which are not included in any of the money supply measures. Thus, in July 1993, when the economy had been growing for more than two years, Fed Chairman Alan Greenspan remarked in Congressional testimony that "if the historical relationships between M2 and nominal income had remained intact, the behavior of M2 in recent years would have been consistent with an economy in severe contraction." Chairman Greenspan added, "The historical relationships between money and income, and between money and the price level have largely broken down, depriving the aggregates of much of their usefulness as guides to policy. At least for the time being, M2 has been downgraded as a reliable indicator of financial conditions in the economy, and no single variable has yet been identified to take its place."

A variety of factors continue to complicate the relationship between money supply growth and U.S. macroeconomic performance. For example, the amount of currency in circulation rose rapidly in late 1999, as fears of Y2K-related problems led people to build up their holdings of the most liquid form of money, and then it showed no increase (even on a seasonally adjusted basis) in the first half of 2000. Also, the size of the M1 aggregate has been held down in recent years by "sweeps"—the practice that banks have adopted of shifting funds out of checking accounts that are subject to reserve requirements into savings accounts that are not subject to reserve requirements.

In 2000, when the Humphrey-Hawkins legislation requiring the Fed to set target ranges for money supply growth expired, the Fed announced that it was no longer setting such targets, because money supply growth does not provide a useful benchmark for the conduct of monetary policy. However, the Fed said, too, that "…the FOMC believes that the behavior of money and credit will continue to have value for gauging economic and financial conditions." Moreover, M2, adjusted for changes in the price level, remains a component of the Index of Leading Economic Indicators, which some market analysts use to forecast economic recessions and recoveries.

In March 2006, the Federal Reserve Board of Governors ceased publication of the M3 monetary aggregate. M3 did not appear to convey any additional information about economic activity that was not already embodied in M2. Consequently, the Board judged that the costs of collecting the data and publishing M3 outweigh the benefits."

[K]

W Peden,

90% treasury bond + 10% junk bond = 100% highly rated asset. Assuming purchases of treasury bonds is irrelevant, then the purchase of $1Tn highly rated assets is equivalent to the purchase of $100Bn of junk bonds. The only thing that matters to the economy is the beta of the asset, ie the equivalent quantity of whole market exposure. And of course, targeting any specific sector/asset class is a direct subsidy to that sector. Marking an asset as AAA is a great way to fool people into believing that you aren't taking any risk. Then you buy 10 times as much of it.

Wall Street could repackage the entire junk bond market as AAA if the Fed incentivized them to do it.

[rsj]

K,

90% treasury bond + 10% junk bond = 100% highly rated asset.

Assuming this logic, the optimal policy is for Treasury to sell 1 Trillion of treasury bonds, allowing the transformation of $100 B of junk bonds into AAA assets, no?

In other words, fiscal policy.

[Nick Rowe]

rsj: you can't have the central bank do monetary policy to manage AD when it's on the gold standard. The gold standard is the monetary policy. So of course the Fed wasn't created to use monetary policy to control AD. That all changed (for the Fed) when FDR decided to raise the price of gold, to target a higher price level. Oh, and FDR didn't say "lets cut interest rates to hit our target of a higher price level". He raised the price of gold instead. And it worked.

And that dismissal of monetary aggregates might have been a little premature. Have a look, just as an example, at David Beckworth's charts. If you had no other measure of monetary policy to work with, and no other information, which would tell you that there's been a recession happening recently? David's charts, which suggest tight monetary policy and recession? Or the ultra low interest rates, suggesting loose monetary policy and an inflationary boom?

"Neither" of course is the right answer, because everything is determined simultaneously. But if I were someone from the concrete steppes, who wanted a simple billiard ball causal theory, I would look at David's charts and go "Hmmm. I bet that reduction in the money supply is what caused the recession!"

[rsj]

Nick,

the charts are M4. But the Fed does not control M4. It has never been able to hit a monetary aggregate. Therefore M* is not a measure of monetary policy at all according to the Fed. That was the quote I just posted.

All it controls is the quantity of bank reserves, and even then, it only controls this quantity when the OIR is zero. It is a far cry to go from the quantity of reserves to M4 -- or even M2! This is where your billiard ball assumption comes in -- whether it be fixed ratios or other mechanisms. You need these mechanisms to go from what the CB does control to what you want it to control.

When we point out that your proposed mechanisms don't exist, you respond with "I don't need any mechanisms. It just works". Why not have the CB target youth and sunshine? It has as much control over these as it does over M2, M3, or M4.

[Lorenzo from Oz]

The future does not affect the present. There is no information from the future. So, all action is based on expectations, so expectations about the future profoundly affect the present. But the future doesn't, except in its unknowability, its complete information opaqueness which requires us to develop expectations so we can act.

[Nick Rowe]

rsj: "When we point out that your proposed mechanisms don't exist, you respond with "I don't need any mechanisms. It just works"."

Stop making up quotes.

Stop pretending that I have not explained why I believe that monetary policy can work. You know damned well I have done this. You may disagree with my explanations, or not understand them, but you cannot repeatedly insist I have not explained my belief. Like at the end of the above post.

And stop attributing beliefs to me I do not hold. Of course central banks do not have perfect control over all nominal variables.

Or, just stop altogether.

Lorenzo: agreed. But I think promises of future actions count for something. And even "cheaptalk", which isn't even a promise to do anything, can have effects in a coordination game, like the cox in a racing eight.

[Kaleberg]

This is rather sad. It seems that failing freshman calculus is a prerequisite for ECON 100. Do you actually check the transcripts? The whole supply / demand interdependence thing is right out of freshman year calculus and physics - a depends on b, b depends on a. Hey, it's the 18th century now. We know how to deal with this what with exp(a+bi)=exp(a)*(cos(b)+i*sin(b)) and all that. Expect oscillations, expect damping, expect explosive growth. Equilibrium is just one behavior among many possible and you can't assert it, you have to prove it.

I usually blame the failings of our economists on a Soviet era like ideological blinding, but it might just be basic innumeracy. Surely it is time to move the field, or at least its mathematical underpinnings, into the 18th century.

[rsj]

And stop attributing beliefs to me I do not hold. Of course central banks do not have perfect control over all nominal variables.

OK, Nick, that is fair.

You did not say "I don't need any mechanism". I caricatured your arguments. God forbid that someone from the steppes would do that :P

But your approach suggests this. In your blog posts, you at least suggest that the central bank controls the quantity of "money", which would correspond to the control of the higher aggregates. Maybe not *perfect* control, but at least a strong influence.

The whole notion of "targeting", at least in the crude form presented here, is one without mechanisms. You see a box in which there is pressure. You have two buttons, red and blue. You press one and the pressure rises. You press another and the pressure drops. You don't need to know or care how it works, it just works. You then run some regressions, etc.

But at some point, further presses of the button stop working -- something has changed inside the box. Here, the mechanisms are important. It makes a big difference whether there really is a structural relationship that is being exploited. The quote I cited suggests that there isn't.

You and I have (I believe) a fundamental disagreement about how the channels operate. And therefore what the CB can successfully target.

When the CB tries to target something that it does not control -- what happens? It wants more NGDP, so it cuts rates. That doesn't work, so it cuts rates again. That doesn't work, so it finally hits the zero bound. Is the CB ineffective because it has hit the zero bound, or because it was trying to target something that it cannot successfully target?

It gets more interesting if the CB can only conditionally target a nominal, dependent on other aspects (e.g. the belief that house prices will rise) that it does not control. Or, if it can only asymetrically target -- e.g. it needs to cut by more than it raises rates, to have the same effect. Then you can only target, but for a limited period of time, and must always hit the zero bound at some point.

And at the same time, what about the ability of fiscal policy to affect NGDP? Do you think if we eliminated food stamps, unemployment insurance, aid to states, and started running budget surpluses "to pay down debt", that this may effect NGDP?

[Peter N]

@Lorenzo from Oz

The word expectations has too much baggage. Some expectations are of such high probability that they are knowledge of the future and others are much less so. The nature of the economic system determines the difference. Since the mathematics is unfavorable to future knowledge (high sensitivity, high dispersion, feedback, delays, bifurcations, chaos and poor information concerning the nature and values of boundary conditions), only our isolated immediate local future is sufficiently likely to be considered knowledge.

OTOH my knowledge of the present and past is extremely limited and inexact. We see and hear with our imaginations, not with our eyes and ears, which are themselves rather limited. It's endlessly amazing what we can accomplish with such poor quality inputs.

[Determinant]

I usually blame the failings of our economists on a Soviet era like ideological blinding, but it might just be basic innumeracy. Surely it is time to move the field, or at least its mathematical underpinnings, into the 18th century.

I agree. I've talked about damping, causality, and forward-looking systems before. Didn't get very far.

[W. Peden]

rsj,

"It was created at the urging of the banks to ensure the stability of the banking system. They needed a reserve bank (a bank that could settle payments between other banks and lend to other banks) that could always deliver in times of crisis. The previous reserve banks, such as the Bank of Suffolk, weren't always big enough to guarantee an end to bank runs.

There was no larger economic role at all, other than the role of preventing bank runs."

*Ahem* banker to the government.

[Nick Rowe]

A thought has just crossed my mind. A very puzzling thought.

I had always understood that uniqueness of equilibrium could only be proven under very stringent assumptions, and then only in Walrasian General Equilibrium theory. Once we relax those very stringent assumptions, or move away from Walrasian General Equilibrium theory, we cannot be assured of uniqueness. Many examples exist of models where equilibrium is not unique. And these models are not at all new. Philosophers like Hume and Rousseau can be interpreted as putting forward (implicit) models of multiple equilibria, or what might nowadays be called sunspot equilibria.

To my mind, only the most dogmatic or ignorant of orthodox economists would insist on uniqueness of equilibrium.

Why then is it that those economists who are currently in this debate most dogmatically insisting on uniqueness of equilibrium also those who think of themselves as heterodox?

Not that I think the equilibrium might not be unique. But we can't be sure it's unique.

[K]

I'm convinced it's not unique. We don't know the model of the economy. We know nothing about the distribution of possible events in the future. It's questionably even meaningful to say that there exists an objective set of probabilities. Where are they? So the equilibrium is a result of our collective subjective expectations, which could be pretty well anything and which are subject to shifting drastically and unpredictably. The idea of a unique equilibrium is a bias that economists (myself included) exhibit as a result of spending too much time in their simplistic model worlds.

[Greg Ransom]

'"Equilibrium" is economists' code-word for "this is a non-linear story".'

Except when you are talking about general equilibrium, which is the only true equilibrium state.

Partial equilibrium is, well, partial.

[Greg Ransom]

All the 'non-linear' story telling has trained macroeconomists to falsely understand the world in the sense that they mistake as the actual world a fake world lacking many of the actual core causal mechanisms at the heart of the economic organism, eg ever aspect of the coordination, pricing, lengthening and shortening of heterogeneous production processes inter-related with heterogeneous labor and heterogeneous financial instruments and monies and near monies.

[Greg Ransom]

Most heterodox economists I know of don't believe the economy is ever in any equilibrium of the sort mapped out in Arrow-Debreu, etc.

[bdbd]

An old John Hicks footnote seems apt here

Some of the most serious fallacies of traditional economics have been
due to confusion between optimum and equilibrium conditions; the
apparent influence of Dr. Pangloss upon the development of economic
thought is for the most part nothing but pure intellectual error.

—J. R. Hicks, "The rehabilitation of consumers' surplus," Review of
Economic Studies 8(2), February 1941, pp. 108-116, footnote 1, p. 112.

[rsj]

Keynes certainly believed in multiple unemployment equilibria. But to him, government policy could select among them. If you believe government cannot do that, then yes, it is sunspots.

To the degree that economic outcomes are the results of our collective choices, then of course the future is indeterminate. The belief that there is some epistemological end-point that the free market will carry us to is just, at its heart, an attempt to discourage public participation and deny our freedom to shape our future with collective action.

Which is not to say that the economists who believe that are motivated by the same factors, they are motivated by the traditions of discourse as shaped by the professional institutions. But don't forget the root of that discourse.

[rsj]

It's questionably even meaningful to say that there exists an objective set of probabilities.

The probability distribution must be endogenous, right?

I mean, the probability that you will get laid off is not independent of your consumption demand, which is supposedly determined by the probability that you will get laid off, etc. By choosing a priori that the probability distribution is a random walk around some mean and then solving for the mean, you are already closing the door to a lot of interesting dynamics.

[K]

rsj:

"The belief that there is some epistemological end-point that the free market will carry us to is just, at its heart, an attempt to discourage public participation and deny our freedom to shape our future with collective action."

Beautiful.

"The probability distribution must be endogenous, right?"

The probabilities, and even the space of the dynamics (the support of the probability space) are certainly dependent on the intrinsically unpredictable decisions of the agents in the world. I suppose that the decisions of most of the agents in the economy are largely noise, but ideas do spread and some agents' decisions are pivotal, so you cannot dismiss the systemic importance of reasoning and conscious processing. In a world as complex and unknown as the one we live in, those decisions can no longer be viewed as a necessary consequence of rational expectations of the (totally unknown) system dynamics, and instead need to be viewed as exogenous (i.e. uncomputable, unpredictable) risk factors in their own right. Human decisions are every bit as exogenous as new technology in your favourite RBC model, which is to say not predictable by the laws of economics, even in aggregate. We are free to shape our world.

[W. Peden]

rsj,

"The belief that there is some epistemological end-point that the free market will carry us to is just, at its heart, an attempt to discourage public participation and deny our freedom to shape our future with collective action."

Curious. Do you think that economics is so simple that people have to have suspicious intents if they disagree with you?

[K]

W Peden,

No, when people are wrong there are other possible explanations. For example, they could be stupid.

:-)

[W. Peden]

K,

Oddly enough, those are exactly the two explanations of disagreement that Thomas Sowell argues that left-wing people will tend to return to again and again, because they fall naturally out of their foundational epistemological views. Similarly, that term used by right-wing people like Ronald Reagan and Milton Friedman, "well-meaning" and its synonyms (a term I used to hate when I was on the left) makes a lot of sense if you take your epistemological views from people like Hayek, Smith, Popper and Hume.

[reason]

J.V. Dubois
"Market Monetarists disagree - the problem is not with CB (and the monetary policy it conducts) not being credible - it is that it is credible in its goal not to expand demand. It is DeLong who is horribly wrong. He assumes that CB will just watch by as fiscal stimulus "mechanically" does its magic to restore the economy."

I must say I found this comment very disappointing - I generally find J.V. Dubois comments incisive. But this comment only makes sense if we we ignore the effect of the zero lower bound, and the effect of policy uncertainty. If there is no danger of crowding out, there is no danger of deflationary central bank intervention. And the effect of monetary policy in those circumstances remains uncertain, the effect of direct government expenditure is much more certain and direct. That relative difference in certainty is important.

[reason]

rjs seems to me (but I'm biased) to be winning the argument here. But he is missing something important (and so is everybody else). Path dependence! Surely when we start with an analogy from ecology it must immediately come to mind! The starting point might matter!

[Peter N]

"The belief that there is some epistemological end-point that the free market will carry us to"

is a perfectly reasonable comment upon the assumption that "free markets" (as if we really had any) will always produce results that are not only more efficient but morally superior and do so regardless about the assumed initial facts.

"is just, at its heart, an attempt to discourage public participation"

however is an abridgment that has to produce more heat than light.

The dynamic seems to be that

1) some participants who claim to be on the free market side hold that as a logical and necessary consequence derivable (in their view) from unassailable moral and economic principles, any opposing arguments must be some combination of incorrect, misinformed, politically radical, dangerous and evil.

This is a particularly pernicious development of an unfortunate tendency among some economists to run the scientific method backwards.

2) As an unsurprising consequence of the human tendency to think well of ourselves and or motives and justify our actions, the rich attribute their success to hard work and intelligent decisions, rather than occupation of a position that allows the extraction of rents, or starting with an advantage.

3) People naturally want to pass their favorable circumstances on to their children

4) Free speech (in practice and in definition) has always been in tension with the old saw "Freedom of the press is limited to those who own one." The current controversy being Citizens United.

5) Legislatures have usually been subject to corruption for some part of their histories. State legislatures in the 19th century US far surpassed the worst abuses of today. A society must deal with the problem or suffer the consequences.

6) With more regulation, you have more opportunities for regulatory capture.

7) The tendency of bureaucracy to expand and entrench itself applies not just to government and industry, but also to political organizations.

This dynamic is certainly opposed to the (effective) participation of that part of the public that opposes these forces (which needn't be the same part for every issue)

"deny our freedom to shape our future with collective action."

This rhetorical flourish just confuses the issue since it has a politically slanted tone, while being too vague to be arguable.

So economics may not be particularly simple, but people espousing certain views often have suspect or self-serving intents. It should be noted that there is a distinction to be made between the motives of the particular proponent making the argument, and the political baggage attached to the argument.

Neither side has an answer to the question: "Are people whose actions or their consequences are evil, but who are acting for reasons they believe are moral or just, good or evil?" And we can ask how useful it is to dwell on the question.

[reason]

Peter N
Excellent comments all. I particularly like the fact that you identify "Bureaucracy" as being not confined to government.

[reason]

Nick Rowe
"We know that for any equilibrium time-path there exists a second equilibrium time-path along which all nominal variables (like NGDP, and the money supply) are higher or lower. This means that a permanent loosening of monetary policy will eventually move the economy towards one of those time-paths along which all nominal variables are higher. This means that the expectation of a permanent loosening of monetary policy will mean the expectation of an eventual permanent increase in all nominal variables like NGDP."

Nick, even if I accept the conditional logic, I don't see how it is possible for the Fed today to make any binding and credible commitments as to the behaviour of the Fed in the future. Just like I don't see how a democratic government (e.g. Greece) can make binding and credible committments as to the behaviour of their successors (who just might be their political opponents).

[Frances Woolley]

reason - Nick's off-line for a few days, Frances