Just a quickie, because I'm (supposed to be) grading exams.
I want to suggest a small change in Paul Krugman's two recent posts (here and here). One that would narrow the gap between his way of thinking and (say) Scott Sumner's.
Desired saving and desired investment depend on a lot of things. One of the things they depend on is expectations of future aggregate demand and hence future (real) income. So, no matter how you define the "natural rate of interest", the natural rate of interest will depend on expected future (real) income.
The natural rate of interest plays a central role in New Keynesian/Neo-Wicksellian ways of thinking about monetary policy. If you want to loosen monetary policy, the central bank should lower the actual rate of interest relative to the natural rate of interest.
The natural rate of interest, by definition, is a real rate of interest. If the nominal interest rate is already at the Zero Lower Bound, the only way for the central bank to lower the real rate of interest is to do something (or promise to do something in future) that would increase expected inflation. That's what Paul wants the Fed to do.
OK. But there's a second thing the Fed could do that would reduce the actual rate relative to the natural rate despite the ZLB. That is to raise the natural rate itself. And raise it by doing something (or promising to do something in future) that would increase expected future real income, which would increase current desired investment and reduce current desired saving.
So, to escape the ZLB, the Fed could do something (promise to do something) that would either: increase the expected future price level (and thus raise current expected inflation and reduce the actual real rate of interest); or, increase expected future real income (and thus raise the natural rate of interest).
And Scott Sumner would say "Hell Heck, why not just add (multiply?) those two things together and get the Fed to do something (promise to do something) that would increase expected future nominal income (NGDP), especially since we don't really know much about short run Phillips Curves and so can't say much about how a loosening of expected future monetary policy will be divided into an increased price level vs an increased real income".
Another way of saying the same thing is that it might be useful to distinguish between a short-run and a long-run concept of the natural rate of interest. The short run concept takes the state of expectations as given. The long run concept assumes expectations consistent with the future economy being at the future natural rate.
This is all connected to my old post on the upward-sloping IS curve. Another way of thinking about that post is to say that the short-run natural rate may be negative (if expectations are currently depressed because the economy is in recession), but the long-run natural rate could be positive.
The natural rate of interest is not a number; it's a time-path. And the central bank doesn't observe that time-path, so when it sets the actual rate it will almost always miss the natural rate time-path. And when the economy is off that time-path, that will cause the time-path to shift. Because expectations will change. And because reality will change too, as investment changes and capital stocks (understood in the broadest sense to include human capital and the stock of employment relations) change too. So, while useful as a theoretical concept, the natural rate of interest is perhaps not so useful as a practical guide to monetary policy as the Neo-Wicksellian approach requires.
Which is perhaps why all of us, central banks especially, should stop framing monetary policy in terms of interest rates. Setting interest rates is not what central banks really really do. It's a social construction of what they do. When central banks talk about setting interest rates that is only a communications strategy, and not a very good communications strategy, especially at times like this.
[Update: Tom Hickey asks: "So monetary policy boils down to central bank communications leading to expectations?"
My response: That's 99.9% of it, yes!
Ultimately, as central bankers themselves have told me, all the central bank really controls is its own balance sheet. But a snapshot of that balance sheet at one second in time tells you almost nothing about what the central bank is doing. It's the central bank's commitment about how it will change that balance sheet over time, and conditional on what events, that really tells you what monetary policy is. A snapshot doesn't work. You need a movie camera, that can also show lots of hypothetical future paths contingent on different circumstances, if you want to see monetary policy.
For example: if you took a snapshot of the Bank of Canada's balance sheet right this instant you would see gold reserves and forex reserves on the asset side. How could you tell whether the Bank of Canada was on the gold standard, had a fixed exchange rate, or was targeting 2% CPI inflation? You couldn't.
Monetary policy is dynamic, not static.
(That last bit was slightly tongue in cheek, along the lines of "my model is dynamic, yours is static". But there's a serious point there. If you want to find out whether the Bank of Canada is targeting 2% CPI inflation, or doing something quite different, you need to read what it says it is doing (and, if you don't trust it, you might have to check to see if in the past it actually did what it said it was doing.)]
Worthwhile Canadian Initiative 
Interesting debate, especially about the question: “The problem is that households in the aggregate do not have enough income to buy all of the output that the private sector could be producing at full employment.”
And I am with Nick in this one. Imagine that the Say’s law holds perfectly. In such an environment this sentence basically means, that the households in aggregate cannot produce anything of worth in exchange for what they need/want. They are forced to sit idly (or consume their own product) forsaking any trade and we observe unemployment and recession.
If we use K’s example of 1% population fueling investment consumption boom for the rest of the 99%, this basically means that this 1% of population produces something essential (financial services?) to the rest of the population and people within that 99% cannot do anything useful within this community. There is no trade among those 99%. So if those 1% decide that they do not want that much of what 99% offer (IOU’s) people among 99% are forced to sit idly. Please realize, that there does not need to be “something” else that 1% want (yachts, monsterer SUVs). They may for example require more leisure and thus collectively restricting the amount of financial services they produce for 99%.
Hmm, now I realized if I did not find another weak spot in Say’s law. What if insufficient demand for some goods and services can be “balanced” not only by insufficient demand in another area, but also by excess demand for leisure? And the decision about the amount of leisure you decide to consume is similar to the the decision about the amount of money you want to save. This decision cannot be forced on you this potentially standing behind recessions. Maybe we may adjust the sentence “Recessions is always and everywhere a monetary phenomena, unless they are leisure phenomena” 😀
Dan: “But what I’m also interested in are plain old prisoner’s dilemma effects. Policies can only produce global macro effects by creating micro incentives that lead individual units to take actions in their limited local spheres that add up to the big collectively optimal macro action. Most businesspeople don’t think like macroeconomists. They are not basing their decisions on applied macroeconomics.”
Totally agreed. (I would just delete your “Most” businesspeople, to make it even stronger; because almost none of them think like macroeconomists.) If what you said there wasn’t true, we would have to scrap the whole of macroeconomics.
Nathan: “Here is my restatement of Dan’s sentence.
“Given firms, households etc expected income, they will not budget enough purchases in the coming year to purchase all output at full employment levels.””
Totally agreed. Though I would add that expected income is just one of the things (albeit a very important thing) that determines their planned expenditure. New Keynesians/Neo Wicksellians would add the real rate of interest as an important extra thing. Monetarists would add the stock of money as an important extra thing.
Now, start with underemployment, where planned expenditure equals expected income equals actual income.
New Keynesians would say the central bank should cut the rate of interest, (and monetarists would say the central bank should increase the money supply) so that people revise their plans, and planned expenditure now exceeds expected income. As time passes, and those revised expenditure plans are implemented, people now are surprised to find their actual sales and actual income exceeds what they had expected. So they revise upwards their expected income, and revise upwards their planned expenditure, and the recovery continues.
(Actually, New Keynesians nowadays almost never, AFAIK, spell those steps out, because they skip the whole disequilibrium process story. When monetarists talk about hot potatoes, they are spelling out that process story.)
K: “There’s no money in this story.”
So people are trying to barter yachts for haircuts, (or yacht builders services for barbers services). There would be an excess demand for yachts matched by an excess supply of haircuts.
Totally possible, but most recessions don’t look like that. There are almost no goods in excess demand.
rsj: start at the beginning. If I build a $1 million supercar, I can afford to write myself a cheque for $1 million, deposit that cheque in my bank account (the same one on which the cheque was drawn), so the cheque won’t bounce, so I can always afford to buy that supercar from myself. (Whether I would choose to buy that car from myself is another question, of course).
Now add in a second person who helps me build the car, and wants to be paid. But we can afford to buy the car together.
Now add in debt. If I have to repay my debt with the proceeds from selling the car, I can’t afford to buy it myself. But me and my creditor can afford to buy it between us.
And so on.
JV: “Hmm, now I realized if I did not find another weak spot in Say’s law. What if insufficient demand for some goods and services can be “balanced” not only by insufficient demand in another area, but also by excess demand for leisure?”
You have just rediscovered Real Business Cycle Theory!
Nathan: In case you hadn’t seen it, here is my version of the New Keynesian/monetarist (mixed together) process story.
rsj: think about it this way: for any level of output, people could afford to buy that level of output if they chose to do so. But if they chose to buy less than that level of output, then they would be unable to afford that (original) level of output.
Or, think of it this way. Suppose that, in aggregate, people always chose to spend their expected income plus a single $1 bill (per year). Then output and income would rise without limit until it hit capacity and they were unable to spend that extra $1. If I changed it to “minus a single $1 bill”, output and income would fall to zero.
Nick: Neat! I have never thought about RBC in this way and you are of course right. It seems that for me thinking about the Say’s law is the approach with most added value for me. I was able to order my thoughts about New Keynesians thanks to one online presentation by Brad DeLong on Say’s Law. And now it seems that RBC can be grasped very intuitively just by expanding on it. What is interesting for me, is that even DeLong “accuses” RBC economists (Fama, Lucas) from believing in the strong version of the Say’s law without them thinking about money, or financial instruments in a broader sense. But all the while he just discarded another equally plausible explanation of Say’s law via insufficient demand for goods and services being countered by “excess demand for leisure”.
And in the way, it is interesting to see that very similar story being played out over again. 80 years ago Keynes promoted using fiscal policy to bring us back into the realm of classical economy – back into a Say’s world. And we can see the similar story unfolding here – let’s use market monetarism to get the money out of the picture so that we can go back into the comforting RBC world where Say’s law stands firm and that makes intuitive sense for most of us.
JV: Understanding Say’s Law (different versions, when right and when wrong) is very important for understanding macro. Brad DeLong is very good on Say’s Law.
think about it this way: for any level of output, people could afford to buy that level of output if they chose to do so.
Only if it is correctly priced. 2 people — a firm spends 1 million to build a supercar, which it then tries to sell for 1.1 million. We have only 1 million in income and so cannot buy the supercar. We are wondering why the supercar has gone unsold, since by definition supply creates its own demand, etc. It hasn’t sold because it was incorrectly priced.
Now if it does sell for 1.1 million — say we take out a loan for the 100,000 and the bank creates enough money for us to buy the car. Then the owners of capital take delivery of an extra 100,000, which they turn around and invest (by purchasing our note). Now the car has sold and because it sold there is enough income to buy it. An extra 100,000 million was needed, and the banks created the additional income which then accrued as additional savings to capital.
That is ex-post. Ex-ante, there is only enough income to build it, not to buy it.
In order for there to be enough income to buy it, it has to be correctly priced.
And one can imagine, when debts are going up, that businesses succeed in selling overpriced cars for a long period of time, during a period of constantly increasing debt, and simultaneous to that, owners of capital succeed in taking delivery of excessively large amounts of capital income for a period of time.
When it stops, suddenly we don’t have enough income to buy the car.
rsj: “Only if it is correctly priced. 2 people — a firm spends 1 million to build a supercar, which it then tries to sell for 1.1 million.”
No. Even if the price is outrageous. Suppose it cost the firm nothing to build the car. It’s all their own labour, scrap materials, and a massive markup. If it prices the car at $1 trillion, the two people just write themselves cheques for half a trillion each, and buy half shares in the car each.
Nick,
“Suppose it cost the firm nothing to build the car. It’s all their own labour, scrap materials, and a massive markup. If it prices the car at $1 trillion, the two people just write themselves cheques for half a trillion each, and buy half shares in the car each.”
You are assuming some sort of coordination that happens outside of the price mechanism. For example, a legally binding commitment by the car maker to increase the income of everyone who buys cars from them. In a many-goods economy, there are no such assurances — all we have are the market prices.
In your example, the workers receive an income of 0. The firm wants to sell a car for 1 Trillion. In such an environment, it is perfectly reasonable to say that the workers do not have enough income to buy the car.
You are pointing out that there is an equilibrium in which the workers, after buying the car, receive $1 Trillion of dividend income from the sale of the car, and with this income they can, ex-post, afford to buy the car.
First, you cannot switch cause and effect like that. The car needs to be bought first if it is to be sold for more than production costs. On the other hand, if it is to be sold for less than or equal to production costs, then the workers are paid first. There is a coordination problem here. It matters who is paid first, as there is always a choice that the other side can defect.
If workers could receive guarantees that for every additional dollar of consumption spending paid, their own wages would increase by one dollar as well, then of course they would buy — it would be foolish not to. But we do not have these guarantees. Then we would be focusing on things like central bank communications strategies, since all that messy market clearing stuff would be solved by some coordinator (who, I guess, must exist but cannot be the government :P).
Or better yet, assume it takes one day to build the car, and the workers need to be paid first. So today, they are building stuff to sell tomorrow, and buying stuff built yesterday. Forget cars, just call it food.
For the workers to defect means that they work today but don’t buy all the stuff they built yesterday at the current prices. They demand lower prices today to buy the goods that they were paid to produce yesterday.
For the firms to defect, after selling all their inventory today, they decide to supply less tomorrow, meaning their labor demand curve falls today, but they still expect to sell all the inventory they produced yesterday.
rsj: “In your example, the workers receive an income of 0. The firm wants to sell a car for 1 Trillion. In such an environment, it is perfectly reasonable to say that the workers do not have enough income to buy the car……..If workers could receive guarantees that for every additional dollar of consumption spending paid, their own wages would increase by one dollar as well, then of course they would buy — it would be foolish not to.”
Capitalists (the owners of the firm) are people too. They earn income and spend it. There is always and everywhere exactly enough income to buy the goods we produce, at any price.
“There is a coordination problem here…..”
Yes! That coordination problem is exactly what I’m yammering on about when I talk about “communications strategies”!
Whether we would choose to spend enough to buy all the goods we can produce is the question. Yes, there certainly is a coordination problem; when we make those choices individually we may or may not choose to buy what we would agree to buy in some hypothetical Big Meeting. And the job of the central bank is to help us coordinate on a good equilibrium, both by printing the right amount of money so we choose to spend the right amount, and, even more importantly, by telling us it is committed to some target (like NGDP) that would act as a focal point to help us get to that good equilibrium.
It’s a cross between: the central bank acting as conductor of an orchestra coordinating the different instruments; the central bank getting the money supply right. (And the people of the concrete steppes go all funny and start talking like engineers in Yorkshire accents when I talk about the central bank as conductor of an orchestra, or coxswain on a racing eight.)
rsj @3.30. Let me translate that into my language: it takes time to build a car. Starting to build a car is an investment. Investment depends on expected future demand. Yes! That’s what this post is all about.
rsj is back! o frabjous day
Nick: “In neither of these types of models is an inability to afford to buy full employment output what stops us getting there.”
A bit late, but got back to thinking about this. There is another kind of equilibrium, technically more akin to yours and Farmer’s, but in effect, more like Keynes’. Imagine that the world has thousands or millions of stochastic factors, most of which are inactive. At any given time only a small fraction are dominant, and while they are active the agents in the world have to do their best to determine the structural parameters before new ones arise and become dominant. Though only a small fraction of the factors are dominant, the risk space is still of enormous dimensionality, especially compared to the relatively small number of periods that the agents (econometrists) have to take measurements and estimate a model. I don’t know if you are familiar with this (pdf) paper by Weitzman. Basically, even with high a high information prior (like a normal) the posterior can be hypersensitive to the exact parameters of the prior, even in the limit of an infinite number of observations. And in the real world with the comparatively tiny number of observations, our subjective posteriors may rationally be radically different from each other. The paradigm of a universe with agents who know the structural parameters of the dynamic couldn’t be more irrelevant.
If we return to our stag hunt, the reality is that I have only gone on a few hunts, and my initial vague prior was strongly conditioned by your assurance that there I would score big by hunting stag. After our epic failure, I have massively revised lower my estimate or the returns from hunting stag. And despite your vast experience, I don’t trust your estimates because our interests are not aligned. So now we suddenly have radically different posteriors, and though the structural parameters of the model are unchanged, we can easily find ourselves in a totally new equilibrium, as prices and quantities are determined by our subjective preferences and probabilities, and not by the absolute (model designer?) probabilities. In a world of such vast uncertainty, the rapidly changing and divergent nature of subjective agent probability measures (even perfectly rationally formed ones) is essentially indistinguishable from Keynes’ animal spirits. In a world of such vast and overwhelming complexity, to insist on real world conformity to a paradigm of perfect, consistent and fully processed information requires an extreme degree of pedantry.
Returning to the ability to “afford”… it’s all just a matter of expectations. “Afford” means I think I can make it. Your opinion may rationally differ. But my expectation is all it takes to move the equilibrium.
Nick,
Rereading my comment… It wasn’t my intent to accuse you, in particular, of pedantry. It was more intended as a comment on what we all do, as a profession, when we build models, and confuse the proverbial map with the territory. Anyways, sorry.
K: I didn’t take it as an accusation. No worries.
I think I see what you (might be) saying, and I think you are right. It would take me a lot of words, or math, to say it precisely. Here’s a rough rendering:
Robinson Crusoe does not know what he can afford, and does not know what he can produce, but he does know for certain that he can afford what he can produce. But this does not aggregate up. The sum (across agents) of expected abilities to afford does not necessarily equal the sum of the expected abilities to produce.
Yes! I like you succinct summary. So now we can have a pure demand (or supply) shock without any medium of exchange and without even a unit of account. Because agents in the real world know literally nothing about the expectations and plans of almost all the other agents in the economy (nor do they even know the “real” probabilities) there are essentially no constraints on their own rational expectations. And in a world of massive leverage it doesn’t take a big change in expectations to cause an asset price fall which pushes people into negative equity which in turn causes forced selling causing further large drops in asset prices at which point our aggregate income expectations and our ability and desire to take risk is so low that we are suddenly all hunting rabbit. Crusoe by himself would never. But add Friday, and all hell can break loose.
K: 2 agent model: you and me. Suppose you can only produce left shoes, and I can only produce right shoes, and we can’t communicate before we get to market and barter left for right shoes (assume we both have 2 feet). Yes, we could have a coordination failure, in which the number of pairs of shoes produced is too small.
But real world recessions don’t seem to look like that. We don’t see an excess supply of right shoes and an excess demand for left shoes. Yours would look more like a PSST problem.
And the job of the central bank is to help us coordinate on a good equilibrium, both by printing the right amount of money so we choose to spend the right amount, and, even more importantly, by telling us it is committed to some target (like NGDP) that would act as a focal point to help us get to that good equilibrium.
Here, I would argue that the central bank is just one tool, which may or may not be effective. Fiscal policy is another tool that may or may not be effective. But the track record of fiscal policy is at least as good as the track record of monetary policy, irrespective of its theoretical position. In both cases, the tools are limited, because just as households can offset changes in government debt issuance with more savings, so the financial sector can offset changes in bond purchases by the central bank with bond sales, so that the total number of bonds available to the non-financial private sector remains unchanged, and the balance sheet of the non-financial private sector is unchanged, and the income of non-financial private sector remains unchanged. In that case, the central bank is not able to supply households with more money.
When it is an income problem, then changes in the OIR may not be effective, or the effect of those changes may be too small to make a difference.
The point of my example was to argue that it can be an income problem. If firms believe that demand will be lower in period 2, then they hire less labor in period 1, and if prices are sticky, then there isn’t enough income in period 1 to purchase the goods produced in period 0 because the prices of those goods are stuck. As soon as you acknowledge that it can be an income problem, then you stop only looking solely to the central bank to orchestrate the economy, but you look to the whole toolbox: fiscal policy, financial regulations, trade policy, government investment, etc.
rsj is back! o frabjous day
Thanks! Really busy with a new career, so not much room in my head for learning econ now. But still a great forum!
On order is important/coordination failures — television dialogue between detective Adrian Monk and his assistant, Natalie Teeger:
ADRIAN MONK
What’s this?
NATALIE TEEGER
A check for $300. I’m hiring you, Mr. Monk. I need your help. I can’t do it myself.
ADRIAN MONK
Oh, this check’s no good.
NATALIE TEEGER
That’s sweet of you to say, but…
ADRIAN MONK
No, I mean it’s literally no good. You can’t cover this.
NATALIE TEEGER
Sure I can. I just deposited my paycheck on Wednesday.
ADRIAN MONK
Right. But I happen to know that that check is going to bounce. So this check is pretty much worthless.
NATALIE TEEGER
You wrote me a bad check? How could you do that?
ADRIAN MONK
I might ask you the same question.
Nick,
Your shoe swap economy is a bit too simple to make my point. What’s missing is probabilities and risk, which is why I like the stag hunt. All that’s happened in that story is that we are all hunting rabbit instead of stag despite no change in the supply of rabbit and stag. What’s changed is my tolerance for failure has declined because of my rabbit debt to you.
So now, in my subjective assessment, the balance has tipped in favour of hunting rabbit rather than investing in the stag hunt. That is very suboptimal in terms of output, but it does (barely) feed my family. If you want to get back to a more optimal equilibrium you’ll have to lower your lending rate on rabbits to shift my risk adjusted returns back in favour of the stag hunt. Where is the excess demand in this story?
The most important thing about this story is that the “objective” or “absolute” or “model designer” probabilities don’t matter. We might as well agree that there is no such thing. All that matters in determining the equilibrium is the agents’ subjective probabilities and risk preferences and also what actually happens. But given that, there is no material difference between the two following scenarios:
1) The real rate is constant at a previously optimal level, but the risk tolerance or subjective expected returns of the agents are reduced; and
2) Everything is unchanged except the real rate is increased from the previously optimal level.
The second scenario is easy to analyse and clearly results in a transition from an optimal equilibrium to a suboptimal one, with lots of deficient demand and unemployment during the disequilibrium phase while the economy transitions from stag to rabbit hunting. Or do we not agree that suddenly raising the real rate by e.g. 10% would result in a state of the economy that we would all recognize as a recession? If not, we should definitely discuss that first.