I noticed something a bit strange about the responses to my last post.
For the last couple of years I have been writing lots posts saying (in various different ways) that recessions are always and everywhere a monetary phenomena. That Keynesian models do not make any sense except in a monetary exchange economy (which is not, I repeat, a criticism of Keynesian models, except insofar as they ought to make that assumption explicit and show exactly where money fits into the model). And that an excess of desired saving cannot create a recession unless it creates an excess demand for the medium of exchange. With loads of weird thought-experiments involving antique furniture, hairdressers, pro-usury parties, etc., trying to make my point.
And I've always felt that I was pushing against the mainstream view. And I've always gotten lots of pushback in the comments, saying I'm mistaken.
But in my last post I said the same thing, said that Paul Krugman agreed, and got almost no pushback at all (except for two, which I will come to later).
I can only conclude that suddenly my views on this are mainstream. Simply because Paul Krugman says the same thing. (Neither of us really put forward any real arguments to support that proposition.)
Which I think demonstrates that Paul Krugman has a lot of "authority". Which is not to say it isn't warranted authority, because he obviously has a much better economist's track record than me. But I'm still a bit surprised at being suddenly so mainstream on this question.
Greg Ransom is the first honourable exception. But he's an exception that proves the rule. Greg, as an Austrian, naturally won't take Paul Krugman as authoritative on anything about understanding business cycles. Greg argued, to paraphrase, that I was just defining a recession as an excess demand for money. I'm going to come back to Greg's point.
JW Mason is the second honourable exception. He(?) wrote a post questioning the assertion that recessions are all about money, and laid out a simple model of recessions in a barter economy (that did not involve the obvious bad harvest/earthquake shock to supply). [Update: just to show it was possible, not because he necessarily disagreed with the assertion as a matter of fact.]
JW's model contains what is often called a "thick market externality". The more traders there are in a market, the easier it is for each trader to find a trading partner, so the bigger the incentive for each person to become a trader in that market. There's a positive feedback process. And if that positive feedback is strong enough, over some range, you can get multiple equilibria, some locally stable and some locally unstable. A recession is an equilibrium with a lower level of trade than in some other equilibrium.
I like JW's model. And I think it's a useful model of a lot of things. But I don't think it's a useful model of recessions. It doesn't work empirically.
When I look at recessions, this is what (I think) I see: it gets harder than normal to sell stuff; and it gets easier than normal to buy stuff (if you've got the money). In a boom it's just the opposite: it gets easier than normal to sell stuff; and harder than normal to buy stuff (even if you've got the money). This is especially true for labour as the stuff you are trying to buy or sell.
There's an asymmetry in the ease of selling and buying that fluctuates over the business cycle. And that stylised fact seems to me to be equally as important as the stylised fact that output and employment fluctuate over the business cycle.
JW's model does not fit that stylised fact. In his model, because it is a barter model, buying is selling. Both buying and selling are harder in JW's recession than in JW's boom. They are equally hard. They must be equally hard, because they are the same thing.
Another way of looking at it is to say that money gets easier than normal to sell in a recession, and harder than normal to buy. And any model that does not contain monetary exchange simply cannot speak to that stylised fact of the business cycle.
So I reject JW's model on purely empirical grounds.
Which brings me back to Greg. How should we define "recession"? Dunno. How should we define "horse"? Dunno either. But I think I know one when I see one. Most of the time, because there are a few borderline cases where, even if I can see everything clearly, I'm not sure if we should really call it one or not.
If those borderline cases are rare, it doesn't matter much. Economists are like taxonomists who can't yet observe the underlying DNA of what we are categorising. So we come up with a list of observable characteristics that tend to correlate together, and use those characteristics to define categories. And because reality is lumpy, so that borderline cases are rare, this works. If reality were smooth, and characteristics were independently uniformly distributed, we wouldn't be able to talk about recessions or horses at all.
Which of the many characteristics of a horse is the defining, essential characteristic? None. A horse is a collection of empirical characteristics that tend to go together. Same with recessions. But there's something very wrong with a theory of recessions which ignores one of those semi-definitional characteristics. "Recessions are always and everywhere an excess demand for money". Is that a definition, or an empirical statement? A bit of both, really.
But what precisely do we mean by an excess demand for money? And where do we observe it? Is the excess demand for money that causes the recession the same as the excess demand for money we observe during a recession?
(You can stop reading this post now, unless you are really into that sort of stuff.)
I know what it means to talk about an excess demand for apples. It's the quantity of apples people want to buy (given current prices etc.) minus the quantity of apples people want to sell (given current prices etc.)
I know that, if prices are sticky, so that some markets do not clear, and some people are unable to buy or sell as much of some goods as they want, these quantity constraints may spill over to affect their demands and supplies of other goods. So we need to distinguish between the notional and the constrained (effective) excess demand for apples. I go to the supermarket planning to buy 3 apples and 2 pears, but they are out of pears, so I buy 5 apples instead. My notional demand for apples is 3; my constrained demand for apples is 5. I go to the labour market planning to sell my labour and then buy 3 apples. But when I get to the labour market I find I can't sell my labour, so I decide to buy only 1 apple. My notional demand for apples is 3; my constrained demand for apples is 1.
I know where to look if i want to see if there's an excess demand for apples. I look in the apple market. And I know that what I will see there is the excess constrained demand for apples, not the excess notional demand for apples. Unless by sheer fluke there is exact market clearing for all the other goods.
What about the excess demand for money (the medium of exchange)?
We are of course talking about a desired stock of money. And that desired stock of money is like a desired stock of inventory. And we desire to hold a stock of inventory because it is costly to exactly synchronise the flows in and the flows out, especially if those flows are lumpy. (Pedants might insist that a lumpy flow is not strictly a flow, but never mind.) But we adjust the actual stock to the desired stock by adjusting the flow in and/or out.
I understand what it means to talk about an excess notional demand for money. People desire to hold a larger stock of money than they actually hold, and so will want to increase their flows of money in by selling more goods and/or will want to decrease their flows of money out by buying less goods. And I understand that this notional excess demand for money, if unresolved, will create an excess supply of goods. A general glut. Because we live in a monetary exchange economy, where all goods (OK, the goods I'm talking about) are bought and sold for money.
And I understand that if goods are in excess supply, people will be unable to sell as much as they wish. The sales of goods will be demand-constrained. So individuals will be unable to implement their plans to increase their stocks of money by increasing the flow in.
And I understand that individuals will nevertheless be able to implement their plans to increase their stocks of money by buying less, and reducing the flow out. And that this plan, even though it works for each individual, cannot work in aggregate. Because one person's purchases of goods is another person's sale of goods. On person's reduced flow out is another person's reduced flow in.
But the attempt by each individual to increase his stock of money by reducing the flow out will result in a reduction in sales of goods. A recession.
OK so far.
But at this point we need to stop talking about an excess notional demand for money. When people realise they cannot sell as much as they want, when people realise they are quantity constrained, they will adjust their demands and supplies of goods in response. We switch from notional to constrained demands and supplies. And that includes the demand for money. We must stop talking about the notional demand for money and start talking instead about the constrained demand for money.
We observe an excess (constrained) demand or supply of apples in the apple market. But money doesn't have a market of its own. Every market is a market for money. We have to look at all markets to observe the excess (constrained) demand for money.
The excess constrained demand for money will be bigger, probably much bigger, than the excess notional demand for money that was the original cause of the recession. In the labour market, we observe an unemployed worker laid off from a $30,000 per year job. He wants his job back. He wants to sell $30,000 of labour per year but can't. He has an excess supply of labour of $30,000, and an excess flow demand for money of $30,000 too. But if he got his old job back, he wouldn't keep all of that $30,000 in money. He would want to spend (say) $29,000 on other goods, and keep only $1,000 in money. And in the second year, having rebuilt his stock of money to his desired level, he might spend all $30,000 on other goods.
In that example, the notional excess (stock and temporary flow) demand for money of $1,000 creates an (indefinitely long flow) excess constrained demand for money of $30,000 per year.
Worthwhile Canadian Initiative 
Too Much Fed: either a bank has to buy something or someone has to go into debt. Mostly, by far, it’s the latter. So generally the answer is yes.
fmb: Let me give my interpretation of what you are saying: there’s a spectrum of goods, from the most liquid to the least liquid. In normal times, it is very easy to swap 2 liquid goods, very hard to swap 2 illiquid goods, and moderately easy/hard to swap a liquid good for an illiquid good.
Now I’ve lost it.
What happens in a recession? Think about that intermediate case, where a liquid and illiquid good are being swapped. Is there an asymmetry? Does the seller of the liquid good have an easier time than normal, and the seller of the illiquid good have a harder time than normal?
Because that’s what I think recessions look like. And your model seems to be looking like a generalisation of the model in my head, where “money” is just the good at the most liquid end of the spectrum.
K: “Nick: I think it matters whether they sell the peanuts to the bank (QE) or just borrow with peanuts as collateral. In the first case, the banks deposits become a proxy for the value of peanuts. In the latter, the value of money can still be independent of the price of peanuts.”
I was trying to get my head around that as well. I think I disagree with you though. In principle, you could sell peanuts for money, but the money might be denominated in something else. In other words, peanuts don’t have to be the medium of account.
Brad DeLong’s thought-experiment: suppose the unemployed could pan for gold, and gold is money. In effect, they are “selling” their labour for gold. If there were plenty of gold mines, without diminishing returns to labour input (a horizontal MPL curve), this would be a very good self-correcting monetary system. The unemployed all go off panning for gold, and the extra gold cures the recession. (It’s theoretically the same as printing CB money to pay for UI/EI, or make-work). But if the MPL curve in gold mining slopes down steeply, the self-equilibrating mechanism would be very slow to work.
Actually, I’m just going to steal what Brad DeLong said. Forget my “peanut theory of recessions”. Think about this one:
“There cannot be an excess demand for money under the gold standard. If there were, people would just go and mine gold and get more money, both individually and in aggregate.”
It is true that there could never be an excess supply of labour. The unemployed would just go off and pan for gold. But if the Marginal Product of Labour curve in gold mining sloped down steeply, the flood of unemployed panning for gold would push wages down so low in gold mining many would stay at home instead.
Liquidity isn’t quite right (re-barter-ability is more what I’m after), but I think we can proceed without straightening that out. Yes, the asymmetry you propose is arising in my model.
And yes, I’m trying to generalize your model to suggest how similar-feeling recessions could happen in a minimally interesting version of a barter economy (“MIVBE”). I could certainly accept that there’s no such thing as a MIVBE, that there are only trivial barter economies and monetary economies with nothing in between. But, I think that requires that you have a more inclusive definition of “money” than I previously expected you had. “Money” in my MIVBE is a bunch of tiny non-homogeneous barely-fungible rarely-re-traded goods.
When you say you can’t have a recession in a barter economy, are you restricting that to only trivial barter economies? I wonder if there’s a MIVBE you think has no money. I think you need very few things before something will take on enough money-like properties to create the potential for Rowe-recessions.
Scott —
I think there is a big semantic component to your distinction here. In your world, the excess demand for money is quickly satisfied by an increase in real balances with respect to stock prices and with respect to newly produced goods (NGDP has finished its descent until k is satisfied). Okay, NGDP has already fallen (stage 1 is over) and so wages and other sticky prices are too high relative to money and flexible prices. This can be reasonably be described as a relative price distortion, but it can alternatively be described as a constrained equilibrium. There is no excess demand for money at current NGDP. But… there is an excess demand for money relative to notional NGDP (NGDP at full employment). By definition, a quasi-equilibrium where the demand for money is not increasing at current depressed output but is too high to allow for full output requires either that 100% of prices are equally stick or that price distortions exist. I’m not sure it matters whether you describe that moment in terms of the stickier prices or inadequate money relative to notional demand.
Nick, You have a definition of excess demand that I have trouble understanding. When I hear the term “excess demand” I think in terms of shortages. You can’t buy as much gas as you want because the price is controlled. You can’t get those Rolling Stones tickets you want because they charge less than equilibrium. You want X, and you have less than X. But that’s not true for base money, except for a very short transition period. You want X dollars of base money and you have X dollars. There is no stock shortage. NGDP adjusts to provide equilibrium. You have an excess supply of labor, but that would still be true if workers were paid in some sort of non-base money assets, and the wages were fixed in terms of a given US dollar market value of that non-base money asset (say gold or forex).
On the other hand I can’t think of a single practical implication of my criticism of your post–which makes me think I am wrong. Everything meaningful should have a practical implication–if it doesn’t it’s probably not meaningful.
As far as the question of whether recessions are monetary, the practical implication is that monetary policy can prevent recessions due to wage and price stickiness. We both think so. But I think that’s all we can ever mean by saying recessions are monetary. One can never say recessions are “intrinsically monetary.” It’s all about policy counterfactuals. One might also be able to prevent recessions with an optimal wage/price indexing scheme, with monetary policy playing no role.
Re: Scott Sumner, substitute “excess supply of non-money goods” for “excess demand for money” and the implications follow. You can’t sell goods, because they’re being priced too high in terms of money. (Or perhaps, if you prefer, money and peanuts and all other flexibly-priced stuff.) You want to exchange X goods for Y money+peanuts, but you can’t, because there is a glut of goods and a countervailing shortage of money.
“There cannot be an excess demand for money under the gold standard. If there were, people would just go and mine gold and get more money, both individually and in aggregate.”
People can go panning for gold even today. No one has trouble selling gold.
But try telling an unemployed person to pan for gold — it’s crazy. To mine for gold today is extremely capital intensive. This has nothing to do with being on the gold standard or not.
Why not hand them a bucket and ask them to dig up some crude oil instead?
With this, as with anything else, the problem is that firms are not hiring. And firms are not hiring, even though they are earning record profits, because the return expectations placed on them are too high.
If that is the case, then there is no difference between asking the unemployed man to borrow a few hundred million to start his own mining company, or asking the gold miner to hire the unemployed man.
In both cases, it doesn’t pencil out, and not because we are on or not on the gold standard. It doesn’t pencil out but because neither the unemployed man, nor the firm that is deciding to expand production expects to earn the return demanded of them given the current environment.
As David Hull explains, horses as biological entities are not natural kinds with necessary & sufficient conditions for class membership. Darwinian biology helps us to appreciate this fact — horses are particular temporal individuals taking part in an historical process, where the larger “historical individual” of the population of reproducing and evolving replicators is tied together by historical chains of causation, and not any fixed set of Platonic qualities.
The lesson from the philosophy of biology / science is that are causal explanatory mechanisms help us to understanding various things about the nature of things — but this doesn’t necessarily happen through the classic image of providing Platonic definitions of particular historical entities or exemplars.
Nick writes,
“How should we define “horse”? Dunno either. But I think I know one when I see one.”
Nick, your reply to my objection just won’t do.
But I think you are on the right track, in pointing to patterns that — for generations — have been associated with booms & busts.
So, let’s start with first things.
Most of science begins with problem raising patterns in our experience — e.g. the problem of the concomitant origin, transmutation, adaptive specialization & geographic dispersal of species.
The puzzle raising patterns grab our attention first — then aspects of those patterns are highlighted and shaded out and their significance re-colored by alternative attempts to causally explain those patterns. Imagine examples from the history of Lamarckian or Darwinian biology.
Now, something important needs to be flagged right off the bat.
There were problem raising patterns — for which trade cycle language was used — that attracted widespread attention before much of any theorizing ever took place.
Here is what is of signal importance. These patterns showed far more structure than is suggested by your own account, Nick.
The “stylized facts” identified with 19th century booms and busts showed all sorts of structure, changes in prices and demands in different sectors of the economy across the course of the boom and bust, as sketched briefly in Hayek’s Monetary Theory and the Trade Cycle, and elsewhere.
Now, here is an important set of facts. When theory comes in, two important things happen. (1) The empirical patterns are perceived differently; and (2) what is quantified and measured changes, and even what is talked about as taking place changes. This is a common theme in the philosophy of science, well documented by Thomas Kuhn in his books and essays.
So what happens when the image of a “general glut” takes hold — and an explanatory project involving mathematical tinker toys and premised in that images begins to dominate the imagination?
Well, people stop seeing or attending to all sorts of patterns taking place in front of them — their interests are elsewhere, and their quantities and measurements are of other things.
The historical entity which is the evolving concept of “the trade cycle” evolves and becomes transformed — a concept once grounded in the complex and patterned experiences of businessmen is replaced by the narrowed vision and narrowly conditioned empirical diet of paid explorers of the “general glut” paradigm.
Hundreds of miles of mothballed lumber hauling rail cars are not seen — are not measured — and are not even imaginable. Nor Iphones & Ipads selling by the tens of millions, or boom regions like the oil lands of North Dakota. Such structural non-glut phenomena — which can exhibit itself in patterns across both geography and time — simply doesn’t come into focus when one is using the “general glut / excess demand for money” lens to (1) assemble empirical patterns; (2) see empirical patterns; and final (3) to account for them.
Very tired, so I’ll wrap up.
If we start with a more theory-neutral set of problem raising patterns that typify the “stylized facts” of hyper growth and consequent recessionary phenomena, we find that there is far more structure than is suggested or implied by the “excess demand / general glut” picture.
And these facts can be accounted for by a causal mechanism whereby specialized resources are misallocated across time — only to be discovered after the passage of time during time-consuming production processes when eventually unanticipated competition for scare goods leads to excess demands, scarcities, losses, and business failures.
The misallocation comes about via distorted prices in financial markets, snowball effects, the lack of transparency, the expansion of leverage, etc. — pathologies of the sort we’ve seen with Fed interest rate policies, Fannie & Freddie & the CRA, and all of the other pathologies recounted in a series of best-selling books on Wall Street, credit default swaps, securitized mortgages, and the housing bubble.
So “money” — the existence of credit, interest, currency, debt, banking, financial instruments, etc. — is what allows for the distortion of the economy into an unsustainable misallocation of specialized resources.
But if an “excess demand for money” occurs, this is an effect of the prior miscoordination — whose unavoidable discovery causes the collapse of shadow moneys and the illiquidity of near money financial instruments.
So so you can have discoordination / recessionary consequences from “the excess demand for money”, but these are secondary effects of the ultimate explanatory cause. I.e. the “excess demand for money” explanation is a proximate cause — which may or may not have large effects — but it is not necessarily the underlying and ultimate cause.
In other words, there is a viable and empirical sound alternative causal explanatory frame, and it can’t be ruled out by any stipulative definition of the word “recession”.
This doesn’t mean that an “excess demand for money” can’t be the cause of a recession. It simply means that there is no necessity that it must be the cause of a recession, and, indeed, this causal factor may be the causal offspring of a deeper, more fundamental ultimate causal mechanism, of the type described above.
rsj: when I said
“There cannot be an excess demand for money under the gold standard. If there were, people would just go and mine gold and get more money, both individually and in aggregate.”
you do understand I was using irony, right? I was saying something I knew to be false. I was drawing a parallel between the above and the argument I commonly hear:
“There cannot be an excess demand for money today. If there were, people would just go to the bank and borrow more money, both individually and in aggregate.”
If the first is false, so is the second. If the second is true, so is the first. I believe both are (generally) false. Only under very special (imaginary) circumstances will they be true.
Greg: interesting pair of comments. I won’t (can’t) respond fully. But I will say I agree with you in seeing the excess demand for money as the proximate cause of the recession. Something else caused that excess demand for money. And that “something else” could be a number of things. It might be the central bank suddenly reduced the supply of base money. Or bank failures. Or something that increased the demand for money.
Scott: “On the other hand I can’t think of a single practical implication of my criticism of your post–which makes me think I am wrong. Everything meaningful should have a practical implication–if it doesn’t it’s probably not meaningful.”
That’s sort of the point I was trying to make in my peanut theory of recessions post, where the price of peanuts is assumed to be perfectly flexible so you can always get more money by selling peanuts. And I was trying to make the same point again above when saying that, under the gold standard, the unemployed could always get more money by panning for gold. It’s true, but (generally) it doesn’t have any practical implications.
Nick, earning money by panning for gold is income. I would agree 100% with you if you just took all of your flow market conditions, and used a flow of money — income. Divide by t! But when you divide by t, then the central bank no longer helps, because it is conducting asset swaps.
So when you say if the first is false then so must the second, then I would say this is not the case.
Earning income by panning for gold is not the same as a portfolio shift. One is a demand for flows and the other is a demand for stocks. One is a demand for income, which OMO cannot satisfy, and the other is a demand for assets, which OMO can satisfy.
rsj: assume the flow of mined gold is fast, relative to the stock of gold. Assume the CB’s flow of asset purchases is slow, because they can only process so many trades per day. We can make them the same on the stock/flow question.
Is the flow of mined gold part of income? If all the new gold is used as money? Accountants would say “yes”, but economists would say “no”, it’s not really part of national income, any more than a counterfeiter’s income is part of income. You can’t eat it. It will cause an increase in income, if it cures the recession. But it’s not itself income. For example, if you were at full employment and had perfectly flexible prices, and diversion of labour into gold mining would lower utility, since fewer useful goods are being produced.
“Accountants would say “yes”, but economists would say “no”, it’s not really part of national income, any more than a counterfeiter’s income is part of income. ”
OK, but aggregate economics, if it is to have micro-foundations, must require that behavior is sourced from individual choices.
And individuals do not care about the national accounts, they care about their own accounts.
Assume individuals want $100 in savings in case they break a leg and can’t work for a week.
But if they only have $50, then they will all want to save $50.
Obviously they cannot all save $50 unless firms decide increase investment by $50.
But in the face of declining sales, firms may not increase investment by $50, even if the overnight rate is cut.
And if the central bank offers to swap $10 of households savings (which are in the form of capital) and replace them with $10 of cash, then households still only have $50, and they are still $50 short.
So the recession continues even as the central bank is madly rushing to exchange bonds for cash.
And the recession will continue until households believe that prices are permanently 50% lower (so that if they break a leg in 2 years, they can still get by on the 50 bucks), or until households can obtain $50 of additional savings, whether by gold raining from the sky, or export income, or some other mechanism that increases their net-worth.
Alternately, the recession may end if the government promises to provide for anyone who breaks a leg, so that the demand for saving $100 is reduced to a demand to save only $50.
But in all these cases, open market operations are impotent, and promising more inflation is harmful.
Is a “excess demand for money” EXACTLY the same thing as a “shortage of AD”? I.e., you can not observe a exess demand for money in itself – but whan a lot of people want to supply stuff and no one is bying there has to be a “excess demand for money” due to Say´s law?
Furthermore – is a “excess demand for money” always a “exess demand for savings” given current prices and subjective expectations (why else would it matter whether we barter or use money)?
Fianlly – does the “money supply” have a “price”? Is the interest rate the “price” – and, if so, is the “price” of money always zero in a static framework?
dlr, I basically agree, it is a question of semantics. The question is which semantics are most useful. I could say:
1. NGDP is too low, and because wages and prices are sticky we have suboptimal employment.
or I could say:
2. There is a shortage of money, which we know because NGDP is too low, and since wages and prices are sticky we have suboptimal employment.
Or I could say:
3. There is no shortage of money, but the demand for money rose, and in the new monetary equilibrium NGDP fell until the desired Cambridge K was reached, and since wages and prices are sticky we have suboptimal employment.
I’m having trouble seeing how monetary disequilibrium allows me to better understand the problem. Example 3 doesn’t use monetary disequilibrium, but it seems to describe the problem just as effectively. Are there any policy implications from insisting on disequilibrium?
Scott —
Are there any policy implications from insisting on disequilibrium?
I think the answer is no, but I think there are probably pedagogical implications. Policy conversations often start “in the middle,” i.e. without reference to a more complete model. It’s confusing to many people who haven’t thought through a money-centric model to connect later-stage relative price distortions with the supply and demand for the medium of account. Although sticky prices is the short answer it is often intuitively inadequate. A story that starts and ends with an inadequate supply of money relative to demand often forces the telling of a more complete picture.
Most of Nick’s best posts are about forcing people who speak different languages to tell each other some rough version of their complete model so they can figure out where they disagree. So he concludes that the QM IS curve slopes up (I don’t like that one) or that MMTers implicitly or explicitly believe there is no meaningful relationship between the real interest rate and savings/investments, so the IS curve is either vertical or can’t be drawn. Here, there is nothing to tease out because it’s the identical story, but telling it from both angles is still worthwhile because not everyone will realize it.
rsj’s last comment has me a little confused. Here the demand for precautionary savings takes the form not of a demand for liquidity, but a target ratio of net worth to income. For the community as a whole, net worth is just the stock of capital goods. So in principle, households should satisfy their demand for precautionary saving by simply purchasing less consumption goods, and more investment goods. The problem seems to be the separation between households and firms. Given that only firms spend on capital goods, households can’t buy them directly but have to buy financial claims on their future output. Now if the firm could offer those claims directly, again, there wouldn’t be a problem, but in fact those claims have to take the form of money, the supply of which is fixed exogenously. And when households try collectively to buy more money than the existing stock, that doesn’t send a signal to firms to produce more capital goods. The opposite, since what firms see is that it’s harder to sell their current output.
So I’m confused when I read this:
And if the central bank offers to swap $10 of households savings (which are in the form of capital) and replace them with $10 of cash, then households still only have $50, and they are still $50 short.
If households own capital goods, then why do we need the firms to be the ones investing in order to increase net worth? On the other hand, if the central bank is buying capital from firms in return for money, then that should send the correct signal to firms to increase investment. You can’t simultaneously assume that a change in household saving implies an equal change in investment by firms, and that households own capital goods directly.
And I don’t think we want to throw out the distinction between more and less liquid assets. I think precautionary saving really has to be demand for liquidity.
It might be time to retreat a step.
I’ve been on the side that wants to deny that demand constraints are strictly equivalent to excess demand for money. But it may be better to admit defeat at the level of abstraction we’re talking at here, and say instead that while the two may be formally equivalent it’s better to distinguish them in practice. A couple reasons:
1. In the real world, there isn’t an observable substance money (belief in a discrete measurable money stock is what distinguishes old-school monetarists) but instead a property, liquidity, which is distributed in different proportions across a whole range of assets. This makes it hard to know in practice how much, or whether, a given central bank operation in the financial-asset market is actually changing the liquidity available to the private sector. It’s much more reliable to start from the least-liquid end of the spectrum (labor) than from the most-liquid end. As soon as monetary policy involves trading money for stuff that also can function as a store of value, the net effect on liquidity gets fuzzier.
2. Even though it’s true, in a formal sense, that insufficient demand for currently-produced output equals excess demand for money (where “money” means the moneyness embodied in all kinds of assets) we really don’t see in the aggregate anything that looks like a well-defined money demand function. (Outside of housing, it’s very hard to find real activity with an economically significant responsiveness to interest rates.) I think this is because binding credit constraints are ubiquitous, and most private borrowing involves large additional costs for both sides. (It may also be important here that the most important economic units are maximizing wealth rather than consumption.) The result is that even though there is excess demand for money in the aggregate, it is hard for an injection via monetary policy to flow through the financial system to the units with excess demand. Here, I think, is the genuine coordination-failure aspect to real-world recessions. (It also suggests that the effectiveness of monetary policy historically was more dependent on the specific institutional regulatory structure of the financial system than we usually acknowledge.) So again, it’s likely to be much more effective in practice to focus on the insufficient demand for current output, rather than the excess demand for money.
I don’t know if it helps you, JW Mason, but on a previous thread Nick said he was odd in insisting that all recessions are monetary. It was in the context in which I asserted that there is more than one kind of recession.
The thread eventually came to the conclusion that when economic activity falls due to supply constraints, that is a “contraction”, not a recession. It is a supply side problem.
A “Recession”, OTOH, is a demand-side problem. Money is scarce, or there is an excess demand for money, because we observe a glut of good in terms of money but lower economic activity and there is an output gap with idle capacity.
Nick said he doesn’t agree with my dual supply-side/demand-side description of periods of economic activity and that he is odd in this.
I have made it my personal mission to get the world to describe supply-side slumps as “Contractions” and demand-side slumps as “recessions” and not confuse the two.
Thanks, Determinant. But I’m coming from somewhere different. My view is that all short-term declines in output in modern economies are demand-determined. It is never the case (outside of wartime) that an economy that is not primarily agricultural faces a short-term decline in aggregate supply of several percent of GDP over a few quarters or a year. So I don’t think contractions in your sense exist.
I think this view is widely shared. Thirty years ago, it was universal. I think it still is universal outside of economcis departments, in the world of professional forecasting in government and business and the like.
The question is, is the statement that all recessions are demand determined, the same as the statement that all recessions are due to an excess demand for money? Nick says Yes, and that’s the claim — I think — that he was acknowledging is not widely shared. I’m one of those who says No, but I’m undecided if there is a difference in principle, or just in practice.
J.W.
“For the community as a whole, net worth is just the stock of capital goods.”
What matters is the position of each individual on their life-cycle savings path. Their real market net worth is the resale value of the assets that they hold, divided by the price of consumption — we are assuming that households are saving in order to sell of their assets (to someone younger, or who is not in trouble) and use the proceeds to purchase consumption at a later time.
To be strictly accurate, what matters to households is the expected future re-sale value of their assets at the time that they expect to sell those assets and purchase consumption.
Therefore asset prices and real wealth of individuals diverges from the market price of capital equipment and software, as it includes non-produced goods such as land, and even for claims on firms that sell into the output markets, the market value of the firms take into account things such as human capital, market power, brand name, the expected future state of the above, current and future inflation, as well as current and future nominal interest rates.
None of the above factors can be purchased in the output markets. Nothing forces total household net worth to be equal to the size of the capital stock — if one could even supply a convincing definition of the latter.
Therefore it is not the case that for each dollar of income one saves, one dollar of newly produced investment goods are purchased in the output market. There is no asset-price policeman forcing this occur.
Rather what we observe is that firms purchase fewer investment goods as their market value decreases, just as households invest less in residential investment when house prices decrease.
Moreover, we observe that as households decrease their consumption expenditures, firm earnings decline, and the market price of firms declines as well. As savings is, by definition, the change in net-worth, households experience less savings in aggregate as a result of trying to save more individually.
We do not see a spill-over from the consumption goods market to the equity markets so that a decrease in the demand for consumption must be matched dollar for dollar with an increase in the demand for equities, leading to a reduction in the overall cost of equity.
I think (hope) we can all agree that the above dynamics are what we observe.
In that case, how to explain it?
This may appear to be a violation of Walras’ Law. One out is to say that there is no Walrassian auctioneer, so prices are stuck at some bad level.
Nick Rowe’s favorite explanation is to point out that Walras’ law is meaningless in a monetary exchange economy. If there are N types of goods, then there are effectively N Walras’ Laws, and you learn nothing.
My favorite explanation is similar to Nick’s but is a bit simpler. I prefer a model in which there are two different types of markets: markets for stocks and markets for flows. These correspond to balance sheet and income operations.
Buying and selling in the markets for stocks — e.g. equities, bonds, deposits, corresponds to portfolio shifts. Those are balance sheet operations. That’s what central banks do.
Buying and selling in the output markets in which consumption goods, investment goods, and labor is purchased correspond to operations recorded on the income statement. Government intervention in these markets is fiscal policy.
Everyone participates in both types of markets.
Even if we were to have a Walrassian Auctioneer, there would not be one Walras’ law, but two laws, in that the sum of excess demands for stocks is zero and the sum of excess demands for flows is zero. But an excess demand for flows does not spill-over into an excess supply of stocks or vice-versa.
Therefore, if the only firm in our economy was Walmart, then nothing forces an increased demand for shares of Walmart stock when households decide to purchase fewer goods at Walmart.
More generally, a reduction in the demand for consumption goods does not spill-over into an increase in the demand for assets. It can spill-over into an overall reduction in income. There is no need to posit that prices in either market are rigidly fixed. We do not need to shoot the auctioneer just yet. It is enough to split him in two.
And in no market can an individual purchase “savings”. Rather savings is an ex-post accounting state, not a good that is bought or sold in any market.
The core of the problem is that when individuals take steps to obtain more savings — i.e. they purchase less consumption — then the net result is that firms see their earnings decline, which causes their stock values to fall, and this decreases, rather than increases, firm demand for investment goods, as well as decreasing the net-worth of the households.
In order to make the above very simple, I need to posit that there are no consumer durables. Consumer durables throw a wrench into the process, but I don’t think this assumption does a lot of violence. As households do not purchase consumer durables in order to re-sell them at a later time to fund their retirement or in case they become ill. Just assume that households rent all consumer durables from firms.
rsj: your explanation also shows why central banks and monetary policy is not enough to end all recessions: the CB has no direct influence over the flow market.
It also shows that Nick is wrong: recessions are not an artifact of money but an artifact of the market of stocks – which can occur in barter economies as well.
Suppose you had a discrete time model. Purchases of flows and stocks are then indistinguishable. We can add them together. Every week I earn an income of $100, and can either buy $100 worth of apples, or add $100 to my stock of bonds, or add $100 to my stock of money (or any mixture).
M(t)-M(t-1) +B(t) -B(t-1) = Y(t) -C(t) is the only budget constraint.
Take the limit as the length of the period goes to zero and we move to a continuous time model. (This actually gives the bizzarre result that you would never hold a stock of money if bonds gave a higher rate of return, markets were open continuously, and all individuals were always participating in each market, because you would switch between bonds and apples via money holding the money for only a millisecond, but never mind that).
Then yes, the budget constraint bifurcates, because Y and C cannot go to plus or minus infinity.
You get M+B=A for stocks, and Adot=Y-C for flows. Where A is the total stock of assets.
But:
1. You can still meaningfully talk about the flow demand for M and B. (“I want to increase my stock of money at the rate of $1 per week and increase my stock of bonds at the rate of $2 per week”.)
2. Both discrete and continuous time versions of the budget constraints leave something very important out: money buys goods (and bonds), and goods (and bonds) buy money, but goods (and bonds) do not buy goods (and bonds). (Clower, paraphrased by adding “and bonds”). You get your income in money, spend some of that money to buy apples, spend some more to buy bonds, then keep what remains in your pocket.
Whether we model in discrete time or continuous time ought to be (approximately, in the limit) irrelevant. If it isn’t, there is something seriously wrong with what the modeller is saying about the world.
Regardless of the discrete/continuous time question, the budget constraints by themselves ignore the crucial fact of monetary exchange. Why don’t the unemployed hairdresser, manicurist, and masseuse simply barter for each other’s services, and get back to full employment that way, even if each of the three women would prefer to accumulate a flow of assets from the fruit of her labour?
To say the same thing another way: where in any of those equations do we see that it is “M”, rather than “B”, that is used to buy and sell everything else? We don’t. There is something seriously missing from a model which just has budget constraints and preferences and technology. We need to say something about what markets exist, and what markets don’t exist.
Nick, the point about distinguishing between stocks and flows has to do with the auctioneer process. Assume the stock of capital goods is 100 units. The flow of production is 10 goods per year. If your time period is 1 day, then an increase in the flow of capital goods produced — i.e. investment — will have a negligible affect on the market clearing price of capital goods. The price of capital goods doesn’t clear savings and investment flows — it can’t, as the units are all wrong.
In the continuous limit, it has zero effect. In this sense the markets bifurcate.
That is why the auctioneer, when determining prices, needs to distinguish between supply and demand for stocks versus flows.
Changes to demands in the flow markets don’t cause prices in the stock markets to shift, and that is really the point.
When households decrease consumption expenditures, interest rates need not fall.
I like to think of consumption services delivered as flows, capital goods as persistent stocks, and flows of investment and saving.
“1. You can still meaningfully talk about the flow demand for M and B”
In the auction sense, “demand” corresponds to a bid in a market. We can also use “demand” more generally to mean “desire”. But if there is no market, then we shouldn’t really talk about excess demand.
While there may well be a desire to have your bond holdings grow by a certain number of units per day, there is no market in which this desire can take the form of a bid.
Rather, having this desire, you alter all your other bids. You may decide to enter a series of bids in the labor market and in the consumption market in such a way so that your bond holdings can grow.
But the auctioneer clears the actual bids, not the desires behind the bids.
“where in any of those equations do we see that it is “M”, rather than “B”, that is used to buy and sell everything else? We don’t.”
I don’t think this is necessary. I want something as close to the walrassian model that still gives me recessions similar to what I observe in the world. As long as the price of capital doesn’t adjust as much as it should when households consume less, then I have my recession. This suggests (to me) that what is used to buy and sell isn’t a fundamental cause of the recession.
Btw, it is rational to hold deposits under any time scale, if you add transaction costs and take risk aversion into account. You could, in principle, buy shares of IBM with your paycheck, and then sell shares of IBM as you spend. But if you have to pay a small fee each time you do that, and if you know that you will spend 80% of your paycheck, then you will want to keep a certain amount of deposits around to minimize your transactions in the stock market and to minimize your exposure to changing stock prices. So you can say that whenever your consumption expenditures exceed your income, then you are hit by fee unless you have enough deposits to cover the difference.
I think this is closer to describing the reality of why people hold deposits.
recessions are not an artifact of money but an artifact of the market of stocks – which can occur in barter economies as well.
This is pretty much what I was trying to say a bit earlier in these comments (about barter economies having “hard to sell” recessions, September 16, 2011 at 11:57 AM), but Nick seemed to suggest that the stock-ish goods were playing the role of money and that they weren’t actually barter economies.
Even in barter, the hairdresser and manicurist only trade by, e.g. the hairdresser getting a manicure and delivering a “haircut bond” (perhaps maturing immediately after the manicure). If everyone is a good credit, they can create bonds denominated in their products of arbitrary maturities and trade them to satisfy any immediate consumption demand. If some people become bad credits, they will want to sell their current labor to accumulate a stock of bonds denominated in the products of good credits, but sticky prices might make that hard to do.
Is this still barter, or did I just describe a monetary economy with a zillion currencies and a very incomplete cross-rate market?
where in any of those equations do we see that it is “M”, rather than “B”, that is used to buy and sell everything else? We don’t
My goal has been to offer examples of barter economies were it is B that is used to buy and sell everything else, with no M. Indeed, I think most barter would act that way, with a zillion customized bonds being created as one leg of most transactions. Perhaps once some of those start getting re-bartered they become M.
dlr, I think you make a very good argument–it is more intuitive to think in terms of a shortage of money.
My view is non-intuitive on two levels. The hot potato effect is less intuitive than say interest rates, and the sticky price explanation of why nominal shocks have real effects is also very non-intuitive. Put them together and you lose almost everyone.
I’ll probably continue to think in terms of monetary equilibrium and sticky wages, because I don’t have a problem with the intuition (having spent my whole life thinking about it.) But I think you are right that Nick’s approach is a better way to sell the importance of money to the broader public. But even there you have to take care to distinguish between people not having enough “money” as a medium of exchange, and not having enough “money” as wealth. Most people find the latter a more intuitive explanation of demand shortfalls.
Wow, what a fantastic discussion.
rsj, I’m not sure I fully understand your reply. I’d like to go through it step by step and see if I’m following you, but a couple questions first: Would you say the story your telling here is basically equivalent to Keynes in the General Theory, or does it depart in some important way? Or is there some other body of theory that you’re summarizing here, maybe even something you’ve written? Reading blog comments without context, it can be hard to tell what’s being assumed, and what choice of words is important.
rsj: “Changes to demands in the flow markets don’t cause prices in the stock markets to shift, and that is really the point.”
Let me rephrase.
Changes to flow demands do not cause discrete jumps in equilibrium prices of stocks, unless they cause discrete jumps in stock demand.
I agree.
But does it work the other way around? Do changes in stock demands cause discrete jumps in equilibrium prices of flows?
I think they do.
Suppose people want to hold more land and less capital (but no overall change in desired flow of saving)? The equilibrium price of land rises relative to capital, and the equilibrium price of flows of newly-produced investment goods will fall relative to capital and land rentals.
Suppose people want to hold more money and less capital (but no overall change in desired flow of saving). If the stock of money does not adjust, the equilibrium price of money (i.e. 1/the price of goods) will rise. Equilibrium goods prices fall. If they are sticky, and can’t fall, we get an excess demand for money and a recession.
“While there may well be a desire to have your bond holdings grow by a certain number of units per day, there is no market in which this desire can take the form of a bid.”
That just seems wrong. It’s the bond market. I can go to the bond market every millisecond, and buy a flow of bonds. I can also go to the bond market yearly, and buy a stock of bonds.
Yes, there are lump sum costs of visiting any market, from the supermarket to the bond market. And that’s why we hold strictly positive stocks of money. And that’s why we buy stocks of apples weekly. Electricity is the only pure flow market I can think of (in my case).
Scott: “Are there any policy implications from insisting on disequilibrium?”
Suppose we lived in a frictionless barter economy. But used bling as a medium of account. Let M represent the stock of bling. Let the demand function for bling be Md=kPY (because people only care about the real value of bling, M/P, they are wearing, so they can flaunt their wealth).
A halving of the supply of bling, or a doubling of the bambridge k (Kings bollege bambridge? sorry) would cause a halving of PY if prices were flexible. But if prices and wages were sticky, it would have no effect on PY in the short run. It would cause an excess demand for bling, and nothing else.
I had to think about that. Scott’s question is a good one to ask.
Nick Rowe,
Isn’t water just as much a flow market as a stock market? Sure, you can buy bottles of water and such, but you can also buy batteries. However, I might have misunderstood the flow/stock distinction.
W. Peden: OK. If you have a water meter, it’s a flow market. I have my own well, with an electric pump! You haven’t misunderstood.
Nick Rowe,
In my part of the world (Scotland) we’re not used to thinking of water as a commodity at all: (1) it’s a nationalised industry and (2) we spend enough time keeping water out of our houses!
So a market where one pays for X within time period Y is a flow market, like pay-as-you-go monthly utility bills, whereas a market where one pays for a total quantity of set units is a stock market like groceries?
I assume that an “all-you-can-eat” market (one where you can use/consume as much as you like in an allotted time period, like a contract phone, a buffet or a rented car) is also a flow market. I can’t think of an analogues in finance, however: a loan is a loan of a set stock of money, rather than money on tap within a given time period!
Water is (usually) nationalised in Canada as well, but you often (sometimes?) have a water meter on your house, and pay so much per litre.
If you measure in continuous time, and plot how much you purchase over time, if you see only infinitely high spikes of infinitely short duration, and zeros, it’s a stock. If you see finite purchases for finite periods of time, it’s a flow. It doesn’t matter if they bill you once a month. If the amount of the bill increases with the amount used, and you have a finite use per millisecond, it’s a flow.
Did my BA at Stirling. Where in Scotland are you?
Water is a flow. LOL.
Consumption goods are consumed. Therefore we model them as flow. You consume every period and do not store the consumption up or re-sell it to someone else. Therefore we can reasonable argue that supply of final consumption output intersecting with the demand for consumption determines the price of consumption goods each period.
Capital is persistent. People accumulate it in order to re-sell it at a later time. When there is a market in which there is an existing pool of durable goods, with one group of producers selling more (indistinguishable) goods into that market, you can no longer claim that the price of durables will be set by the demand for net savings intersecting the supply curve of (new) investment. Therefore the concept of a loanable funds market in which savings and investment set rates is incoherent if capital goods are long lived.
But it needs to set the price if we are to believe that interest rates adjust to clear savings demands.
Loanable funds would still be incoherent even if we used a short-run discreet model or even if we added depreciation. Loanable funds only works if the capital stock fully depreciates in each period.
So let’s quibble about clumping. Let’s keep it simple and not get into disputations about how to classify jams and jellies.
J.W. — I will make a blog and then post a reference. On my blog, I can use latex, and make diagrams, etc. Also, I have edit capabilities 🙂
A few miles north of Stirling in Callander. Stirling is the “big city”, from our perspective! I studied my undergraduate degree in Edinburgh, however, and in a few weeks I’m going to study in Cambridge
K said: “Too Much Fed: either a bank has to buy something or someone has to go into debt. Mostly, by far, it’s the latter. So generally the answer is yes.”
Even if a bank buys something, can “debt” be involved?
Nick–
I agree with you that disequilibrium is necessary. You also agree that sticky prices are necessary. The MOA function of money is thus also necessary because it is the only realistic way we can tie an MOE to stickiness (if dollar-bling prices were flexible but bling prices were sticky, we could never have disequilibrium based on changes in the supply or demand for dollars because the bling-dollar rate would clear). I think Scott’s point is that in economies that actually exist where the MOE is the MOA, there are no implications to describing recessions in terms of the necessary quality of stickiness rather than the necessary quality of disequilibrium. You seem to be arguing (if I’m getting you right) that there are implications because it is possible to have stickiness and monetary shifts without a recession if you don’t have disequilibrium in the MOE. This is only true because disequilibrium always requires stickiness but stickiness doesn’t require disequilibrium. But since the MOE and MOA are the same then stickiness DOES require disequilibrium and it doesn’t substantively matter which way you describe it — so I don’t see any meaningful implications. But I guess you’re right that if we are thinking about policies that would eliminate a medium of exchange entirely or attempt to divorce it from the MOE, then implications sneak in.
No, production goods are constantly being turned over, worn out, re-purposed, replaced by new technology, etc.
Production goods are a flow — and they flow across time, across geographic space, and across alternative uses, etc. E.g. production goods sometimes flow toward longer term production processes — like the supplying of (wood based) housing in the production of a house — and sometimes toward short term production processes, like the supplying of (wood based) disposable chop sticks for eating a meal.
Flows have changing valuational significance across time, space & technological environments.
Production goods are a flow …
It’s fallacy derived from business firm accounting and imported into economic science to call “capital” a “stock”.
“Capital is persistent. People accumulate it in order to re-sell it at a later time.”
I studied (briefly) at the Scottish Institute of Textiles in Galashiels ( now Scottish Borders Campus of Herriot-Watt University). This blog is becoming a big family…
And York and East Anglia (Norwich) (see the Blog thread) And Cardiff (Long story in Ken Kesey’s travels)
Everyone: sorry for the multiple comments. Some computer glitch..
[No worries. I unpublished the duplicates. NR]
“Production goods are a flow …”
Knowledge is a stock.