It gets asked in every recession. Recovery requires an increase in demand. "But where will the demand come from?"
When I was young and foolish I would answer "housing". Which was actually a fairly good guess in the past. But when I got older and more devious I would refuse to answer. "If an armchair economist like me really knew the answer to questions like that, we wouldn't need a market economy; we would just make me central planner". Or "If I really knew the answer to questions like that, I would be very rich and sitting on a beach somewhere".
But "housing", even if it were the right answer, isn't really answering the question. It's not really a question about which sector of the economy will be the leading edge of the revival. It's not really a question about which particular group of people or firms will be the first to increase spending. It's not a micro question at all. That's not what they are asking. It's a macro question.
What people who ask the question are really looking for is some source of demand that comes from outside the system. Exogenous demand. From exports, perhaps. If I could persuade people that the Chinese would develop a sudden taste for Canadian-produced goods, they would find that a perfectly satisfactory answer to their question. Where will the demand come from? China. OK.
But the logic of that answer always ends up with everyone wanting to increase exports to Mars.
Macroeconomics is ultimately about closed systems. (Sorry guys, but open-economy macroeconomics, and especially small open economy macroeconomics, isn't really macroeconomics.) Demand cannot come from outside the system. There is no outside.
Demand comes (mostly) from itself. That's the answer that makes no sense whatsoever to most people. It's the logic of the Old Keynesian multiplier. The Hawtrey-Kahn-Keynes-Clower multiplier contains an important truth that is missing from nearly all modern macroeconomics.
The short side of the market determines quantity traded. Quantity sold is whichever is less: quantity demanded; or quantity supplied. If there is excess supply of goods in aggregate, then realised sales of goods, and income from those realised sales, is demand-determined. And if people are unable to realise their plans to sell as many goods as they wish (if they face Clowerian quantity constraints) then their demand for goods will depend on their realised sales, which is demand-determined.
Demand creates income. And income creates demand. So demand creates demand. That's the fundamental insight of the Old Keynesian multiplier that was lost in the New Keynesian Euler equation.
Now I'm going to bring together two very unlikely bedfellows: Keynes and Say.
It's income from the realised sale of newly-produced goods that provides the wherewithal to purchase those same newly-produced goods. And if some people save part of their income and lend it to others to spend (on investment or consumption) more than their income it makes no difference in aggregate. Whether spent or lent, (nearly) all income is spent.
That version of Say's Law (there are many versions, most having nothing to do with Say), which says that people in aggregate plan to spend all their income, is very nearly right. The marginal (and average) propensity to spend (on consumption plus investment) is (very nearly) equal to one.
If the marginal propensity to spend is one (if the slope of the Keynesian Cross AE curve is one), then the Old Keynesian multiplier is infinite. An infinitesimally small exogenous increase in desired expenditure is sufficient to bring the economy to "full employment", where the supply constraint bites and stops further expansion.
That, actually, is very close to my view of macroeconomics. It's an ungodly mix of Keynes and Say, seen through a Clowerian lens.
But it's only "very close". It's not exact. And that version of Say's Law, in which aggregate demand is determined by and equal to aggregate income, is only "very nearly" correct. There's something missing.
What's missing is money. Add monetary disequilibrium to the mix of Keynes and Say.
None of the above makes any sense in a barter economy. The very distinction between aggregate supply and aggregate demand only makes sense in a monetary exchange economy, where we sell goods for money and buy goods with money. Money is the medium of exchange. As Yeager noted, there are always two ways to get more money: sell more goods; and buy less goods. If you face Clowerian quantity constraints on selling more goods, because there's an excess supply of goods, you can still buy fewer goods if you want to get more money.
It is money, and only money, that makes Say's Law false. If, in aggregate, we wish to hold more money than we currently hold, we will plan to spend less than our income. If, in aggregate, we wish to hold less money than we currently hold, we will plan to spend more than our income. An excess demand for bonds won't falsify Say. If there's an excess demand for bonds we can't buy more bonds, because the quantity of bonds is supply-determined. And if we can't buy more bonds, we have to spend our income ouselves. Or hold more money.
If the desired stock of money were identically equal to the actual stock of money, at all levels of income, then the Old Keynesian Cross model would have an indeterminate equilibrium. Any level of income between zero and "full employment" would be an equilibrium. At any level of income, demand would equal income, and income would equal demand. And so an infinitesimally small increase in the supply of money, or decrease in the demand for money, would be enough to create a self-perpetuating increase in demand, increase in income, further increase in demand, to get the economy to expand to where the supply constraint stops any further expansion. Or where inflation reduces the real value of the money supply, or leads the central bank to take away the punchbowl.
So don't ask where the demand will come from. It comes from itself. And from an excess supply of money.
Worthwhile Canadian Initiative 
Kevin: exactly! You pass the test. You can predict exactly what I would say.
“Okay, being a creative person, I offer to manufacture some asset backed securities–using the goods that aren’t being sold as the collateral behind the security. In exchange, I get base money now; I promise to pay you more base money later, but then you only have recourse to the pile of goods I just “sold” you if I don’t.”
No one would make this loan. First, the value of collateral is its resale value, so that means only short term loans can be secured by inventories of consumption goods, and the value of the loan will be less than the expected resale value of the collateral.
Longer term loans would beed to be collateralized by capital goods, but again the re-sale value of the capital goods needs to be more than the loan amount. The residual is your equity in the real asset that you have purchased with the loan, and you generally need skin in the game, and cannot acquire capital goods for free without supplying any capital yourself.
RSJ writes: No one would make this loan.
Precisely. The excess demand for money cannot be met by producing more securities except in the narrow case where those securities are money substitutes and can be used in place of base money (because they are secure claims on money).
An asset backed security will have this precise property when the goods markets would clear. Ergo, there is no excess demand for asset backed securities generally.
And if you cannot make it work with asset backed securities, I find it difficult to believe you can make it work without collateral.
So by exclusion, we’re left not with savings generally but savings particularly as the unmet demand to accumulate money balances as the problem.
Jon,
My most simple model would be that, at the margin, an excess demand for money corresponds to a deficient demand for consumer goods, resulting in an unwanted accumulation of inventory of consumer goods, which is classified as investment.
Making the further simplifying assumption that inventory is financed in the first place by bank loans, and that bank loans are offset by deposits, this means that an excess demand for money shows up as deposits held by households. These deposits are banking liabilities, offset by the inventory financing that is the banking asset. Excess demand for this money means that velocity of this money has hit zero, at the margin under discussion. That zero velocity is operationalized as a holdback of money balances by households due to their excess demand for money.
So far, this story should be consistent with Nick’s story, even though I’m framing it in an ex post sort of way, as the sort of behaviour that must be observed logically in connection with an actual excess demand for money.
The next step is to recognize that this excess demand for money has resulted in an increase in household equity relative to the counterfactual where the market for goods and services would have cleared and households purchase goods that they consume immediately. (This counterfactual position accords with the assumption that consumer goods once purchased are not classified as investments on the balance sheets of households. And that is consistent with the fact that there is an effective equity drain from household balance sheets associated with a counterfactual purchase of consumer goods.) This comparative increase in household equity due to the excess demand for money is deployed as an asset in the deposits that correspond to the collapse in monetary velocity due to the increased demand for money.
This comparative increase in household equity (again compared to the counterfactual of a cleared market) corresponds to an increase in saving. That’s what saving is – an increase in equity. It’s not the money or the bonds or the house or the investment or any of the assets that offset the equity – it’s the equity itself, which is the right hand entry on the balance sheet.
So the increase in the demand for money has been associated with a comparative increase in equity and saving, and in this case a comparative increase in money holdings of households.
Given such a base case position for this example, it is easy enough to consider all sorts of micro permutations whereby those households who initially hold the excess money choose to swap their money for other assets – assets that are not part of the flow of new goods and services – these include financial assets such as bonds, stocks, mutual funds, or “old” real assets such as existing houses. None of these swaps affects the macro configuration of the assumed excess demand for money or the corresponding excess saving that has accumulated. I see nothing special about bonds or the supply of bonds in this context.
Bottom line is that of course the forces in play must involve an excess demand for money, because money is the medium of exchange. But they must also involve an excess demand for saving, because that is inherent in the required behaviour of the accounting identities that are consistent with the granular detail of the particular asset in play – which is money as the medium of exchange. So my conclusion is that it is nonsensical to attempt to divorce the excess demand for money from the excess demand for saving. The former is driven by the fact that money is the medium of exchange and the latter is driven by the fact that accounting identities must behave in such a way to accommodate the assumption of an excess demand for money. The two characteristics are inextricably linked in the description of this type of recession dynamic. It is a matter of recognizing that money is the medium of exchange, and that saving is the accounting requisite that accommodates the assumed money behaviour. If money is the medium of exchange, then one implies the other and vice versa – excess demand for money and excess demand for saving.
Jon,
You’ve lost me. The walrassian framework of excess demands assumes that endowments are fixed.
It is a partial equilibrium assumption.
But endowments are not fixed.
Endowments change in real time, because the economy is producing, consuming, and investing in real time.
In that case, it is not true that the sum of demands must equal zero. That is only true in the special case of instantaneous production, or in an economy without production.
Why can’t we just agree to switch to a barter economy without production and all get along?
LOL, yeah we need to shift out of time and into meta-time.
I really think that the relevant distinction between monetary economies and barter economies is one of time. To buy something for money is not to exchange a real good for another real good, but to exchange a real good for a paper claim, with the assumption that later you can exchange your paper claim for another good. And equivalent to that is the ability to buy something without an endowment (e.g. borrow) and then later repay (when your endowment increases).
But as soon as production requires time, then it will be funded by selling some form of paper claim. Whether those paper claims pay interest or not doesn’t matter. Now you have individuals transacting across time, but the economy as a whole cannot, and this can lead to the savings/investment miscoordination.
The only true barter economy is one in which no one attempts to transact across time (i.e. sell now and buy later, or buy now and repay later), which means no production, no storage technology, and Say is right in that those wanting to sell must also want to buy at the same time.
But as soon as they have the option to sell now and buy just a few moments later, or buy now and then sell a few moments later, then you are in a monetary economy again, in which, at any point in time, the demand to sell does not need to equate to a demand to buy.
It’s a real shame that people get their intuition from statics, and then view dynamics as an advanced concept.
Sandwichman
“The only way principal + interest can be paid back is through the creation of more money through credit.”
I’m surprised at you here. This is not true. Not when you consider time and the money is relent as it is paid back. Steve Keen addressed this.
P.S. There are implications about the velocity of circulation here and the rate of growth – but in general what you say is not true. It is confusing a stock and a flow.
Nick
Why do you ctiticize MMT for adhering to the state theory of money. Its obvious that the state theory is a fact. The US$ was a creation of “the state” as was the canadian dollar, the euro, the yen, the deutsche mark. In fact I’ll venture to say that in virtually every stable modern economy it is the states money that is being used. The economies with multiple competing currencies are not places that most of us want to live and do business. Knowing this to be true it follows then that the state is not constrained by tax revenue for its spending decisions, only by its politics. A stable state with a stable state currency is the best place to live and do business.
A social construction of reality……….. that sounds an awful lot like a state.
No, you’re quite right there I made an unnecessarily absolutist statement. I would suspect, though, that you’ll find rare episodes historically in which a contraction of credit hasn’t been followed by a rash of defaults.
But, as I’ve tried to point out before, my overstatement was not the main point I was trying to make. And it’s disconcerting to see folks gloating about an infelicity but not addressing the issue I raised (or the quote from JMK).
That point I was trying to make is that debt service and final demand for goods and services are not inherently proportional. If they happened to be proportional or if demand was expanding faster than debt service costs for a period of time, there would be no need — during that period, anyway — for credit to expand. There comes a time, though…
Gizzard: most states have nationalised money. But monetary exchange can and did evolve without the state. And a particular choice of which good to use as money can evolve without the state.
The state is a social construction of reality. We make it, by believing in it. It does not exist apart from our belief in it. The state is made out of the same stuff as money. The state is not ontologically prior to money.
In defence of Sandwichman: the demand for money will presumably depend not just on GDP, but also on transactions in intermediate goods, and also on the flows of loans and repayments of loans, and interest payments on loans. All (monetary) transactions will create a demand for the medium of exchange. But a debt that is just accumulating at the rate of interest, without a flow of interest actually being paid, will presumably not create a demand for money.
Nick, I don’t get it. If gold is the MOA, then wages are set in gold terms. If wages are sticky, that means they are sticky in gold terms. Otherwise gold is not the medium of account. If the real value of gold doubles, and gold wages are sticky, then real wages double and you get mass unemployment. What am I missing?
If you argue wages are paid in beer, and the wage in beer is sticky, then you have violated the assumption that gold in the MOA. You can’t have it both ways.
Scott: Assume W the price of labour in terms of gold is sticky, and P the price of beer in terms of gold is equally sticky. If the real value of gold doubles, then W needs to halve but won’t, and P needs to halve but won’t, but W/P doesn’t need to change at all. So if wages are paid in beer the labour market can function normally. MRS=W/P=MPL. (except in the market for gold miners).
Nick, If gold is the MOA, then it’s nominal price is fixed. If its nominal price is fixed, then its real price can only double if the price level falls in half. If you are claiming that goods prices are fixed in terms of gold, then you are claiming that the real price of gold cannot change.
So it seems to me that your actual assertion here is that the real price of gold cannot change if it is the medium of account, not that it could change, but it wouldn’t matter.
This is exactly my problem with the (Keynesian) model. We know that some prices are fixed and some are flexible. That means that in the real world the real value of gold can rise (even if it is medium of account), and cause unemployment.
Yes, one could envision a model where all prices were fixed, but it would have no real world applicability.
It also seem to me that your model assumes that monetary shocks cannot cause changes in real wages, because wages and prices are equally sticky. But we know that’s false. Monetary shocks do affect real wages.
Scott: a thought-experiment:
Start in equilibrium. Barter economy. Hold all prices fixed in terms of gold. Then the demand for gold increases (or the supply of gold falls). All prices need to fall, but none do. But W/P doesn’t need to change, and the market for beer/labour still clears, MRS=W/P=MPL, even if the markets for beer/gold and labour/gold don’t clear.
Now the same experiment with a monetary exchange economy. There is no beer/labour market. There’s a beer/MOE market, a labour/MOE market, and a gold/MOE market. If all 3 prices (W, P, and Pm) are fixed in terms of gold, then only the gold/MOE market fails to clear. But if Pm falls to clear the gold market, it will cause the beer and labour markets to fail to clear, output and employment to fall, even though W/P is still at the correct level. If P falls to clear the beer market, then W/P must rise to equal MPL at the suboptimal level of employment. So it looks like the real wage is too high.
Nick,
I tried creating a picture-book version of your argument. I don’t know if it helps, but it was fun fooling around with Paint.
Kevin: that’s beautiful!
Nick
Just because monetary exchange evolved in many places without the state is NOT a point against MMT. The fact is virtually ALL modern economies use state money, whether historically this has always been true is completely irrelevant. MMT describes the realities of a state money system, the ONE WE ACTUALLY HAVE. Why spend time talking about money systems which we do not use. The problems we are having are not arising from the fact we have a state money system, they are arising from the fact that we have a state money system and too many economists want to act like we have something else.
We have a system which can never run out of “money”, so why do we create a financial system that continually acts like we are? Its beyond ignorant and bordering on criminal.
Nick, Hope I’m not beating a dead horse, but I’m not sure what you are assuming. If you say gold is the MOA, and ipso facto we assume that the nominal price of gold is fixed, and we assume that the real value of gold doubles, and we assume all nominal goods prices (in terms of gold) are fixed, then we have a logical contradiction. I must be missing something obvious, because it seems like the way you set up the problem, gold is called the medium of account, but (de facto) it’s not the medium of account.
If we are going to discuss whether changes in the real value of the medium of account are important, then we MUST talk about that situation in terms of a model with flexible prices. That’s because if prices don’t change, it is impossible for the real value medium of account to change.
The only possible exception would be if you wanted to make a queuing cost argument. The nominal price of goods doesn’t change, so people try to get the more valuable gold by standing in long lines trying to sell goods for gold. It seems to me that would lead to a recession, even under barter.
Nick, Thanks for pointing out (February 05, 2011 at 03:19 PM) that a lottery win is simply a transfer between different people and has (at least arguably) no macroeconomic implications. I was well aware of that. I thought that point was so obvious that it did not need spelling out.
Thus my basic point still stands. Which is that when someone wins a lottery, their spending rises. Thus if the bank balance of everyone in the country is boosted, then spending in the aggregate will rise (assuming there are no capacity constraints, or put another way, assuming unemployment is above NAIRU).
And I suppose I shall have to point out that I am well aware of the dangers of extrapolating from micro to macro, but in this case as far as I can see it works. Indeed, the idea that boosting everyone’s bank balance boosts employment is central to Modern Monetary Theory.
Which is why if economists and others simply learn a bit about differential equations and statistical mechanics, none of this would be anything but crystal clear because the entire discourse of explaining things in terms of Say or Clower is popped out as simply multiple behaviors of a single solution family.
Basically you are doing what is simple in math by describing the math in many paragraphs of words. It’s no different from explaining multiplication by exhaustively reciting all the individual numbers in a multiplication table. Strictly correct, sometimes gets the right answers, is error-prone and just a bit bizarre when you know there’s an easier way.
Do I actually imagine math will ever become part of the standard discourse. Not any time soon. The old guard always has to die off before any disruptive improvement can ever be adopted. Probably not in my lifetime.
Yes, I know various financial “rocket scientist” know all this, but consider that they went into finance rather then their own profession – they probably were NOT all that technically.
JG: nearly all economists do learn an awful lot about differential equations and statistical mechanics. (Far more than I have learned, let alone remembered). And they normally do talk about stuff like this in math, rather than words. Math is very much the standard discourse in economics. That’s what we teach the students. This post is the exception.
But doing it in math is absolutely no guarantee of getting it right. There are loads of mathematical macro models that totally miss these points. If anything, doing it in math seems to make it easier to miss these points. They can’t see whether they are modelling a Walrasian or a monetary exchange economy.