David R. Henderson  

Paradox of Thrift

The Future of Macroeconomics... Thinking Outside the Beltway...

Both Arnold and Bryan have weighed in on the paradox of thrift. Arnold's response was to my mention of Bob Murphy's refutation of Steve Fazzari. Jeff Hummel read Arnold, Bryan, and Bob Murphy and wrote me the following:

The essence of Keynesian business cycle theory can be distilled down to two simple features. The first feature is that aggregate demand (AD) fluctuates with people's hoarding of money (called a fall in autonomous expenditures by Keynesians and a fall in money velocity by monetarists, both of which are equivalent to rise in the portfolio demand for money). People hoard more, and aggregate demand (i.e., nominal spending) falls. That is the essence of a Keynesian recession.
The second feature is some inflexibility in prices or wages that transforms the fall in AD into a rise in unemployment and a fall in output. It is just that simple. All the rest of the Keynesian apparatus attempts to bolster one or both of these two features.
The Keynesian paradox of thrift does not even deny that saving in the form of hoarding leads to more investment. It just claims that that the resulting investment takes the form of unplanned inventory increases. And because of price or wage inflexibility, that in turn leads to a fall in employment which in turn generates a fall in output. In other words, saving in the form of hoarding leads to a decline in real income in the short run, rather than a rise in real income in the long run. Hence, the paradox.
Fiscal policy works under these conditions, according to Keynesians, because people are indifferent between hoarding money and holding government securities. So the increased government deficit, rather than crowding out private investment, sucks money out of people's hoards and re-injects it into AD. Because of the price or wage inflexibility, as AD rises, output and employment go up.
In other words, to really refute the Keynesian Paradox of Thrift, you must either argue that (a) fluctuations in hoarding (or autonomous expenditures or money velocity) are not that severe or that (b) that prices and wages are flexible enough that it doesn't matter.

Comments and Sharing

CATEGORIES: Macroeconomics

COMMENTS (22 to date)
tom writes:

Can't you also just say that we are not in a 'Keynesian' recession if the recession is not due to hoarding of wealth but the actual loss of wealth?

Lord writes:

No. Our wealth is our productive capacity which is substantially our labor, so our loss of wealth is primarily our loss of jobs and incomes. It may not be as productive as we thought, but it is still productive at the right price. Now the right price for capital may be negative which would be an inflexibility.

Jay writes:


You can also refute (c) "that the resulting investment takes the form of unplanned inventory increases".

That is convient, because (c) is the fundamental flaw. There is a soft limit to how much inventory a company will pre-produce, and because of the price constraint, there is a hard limit to how much inventory a company -can- pre-produce.

After that point, further savings will be pushed into real investment, by which we mean "that which increases the marginal productivity of labor".

Government debt crowds out this sort of investment, which is ultimately why it prolongs the problem.

Carl Jakobsson writes:

Hummel seems to think that the Keynesian "Paradox of thrift" is true, or not, if the economy is in equilibrium or not. But from my understanding of it, as Fazzari presented it, it doesn't matter whether or not entrepreneurs have planned succesfully for the change in people's consumption pattern (from consumption to saving). That is, Fazzari seems to say that even if entreprenuers are ready for more saving and less consumption, overall wealth in the economy will stay the same.

To refute this argument one only needs to show that investment is more productive than consumption, that producing robots (or parts for robots) will create more wealth than making pizzas, and that our wealth does not depend on overall spending in the economy but on production.

So, as I see it, there's a difference between the paradox of thrift and the Keynesian theory of recession, and the recession-story could well come true even if the paradox is incorrect.

El Presidente writes:

That sounds like Keynes in a nutshell to me. Good summary.

Mohammad Osman Gani presents it a little differently. I think he provides a more useful explanation for purposes of theoretical discussion. He points to a shift in the use of money. When it is used more as a store of value and less as a medium of exchange its utility is reduced in both respects and we require more of it to do each as well as before. He presents it as a misuse of fiat money, which he says is in its highest and best use as a medium of exchange to facilitate indirect exchange and satisfy multiple coincidence of wants. He identifies the presence of money as a vital element in market clearing. I think Keynes caught on to this, but he explains it slightly differently and seems to draw unnecessary criticism for what is essentially a simple and self-evident proposition that we could theoretically measure if we chose to.

Bill Woolsey writes:

Hummel is exactly right.

What is a recession like if it is caused by a "real loss of wealth" rather than a "hoarding of wealth?" If it involves a drop in spending, difficulty in selling, and then a reduction in either production or prices, then it is a Keynesian recession. And it is caused by a hoarding of money, not wealth.

Wealth is gone, we are poorer, people won't buy as much, so we can't sell as much... that is just incomplete analysis. Reduced productive capacity creates shortages, not surpluses. Reduced financial wealth that causes people to "save" results in increased production of capital goods to make up for the lost ones.

Unless there hoarding of money.

While Hummel is correct about the unplanned inventory investment, he isn't saying that this somehow "solves" the problem. As Hummel points out, firms cut production in that circumstance. (Or maybe prices.)

What Fazzari "seems" to say isn't really at issue. If you think that Keynesians believe that consumption is more productive than investment, you missed the point.

Everyone understands that if a reduction in consumption matched by an increase in capital spending by firms the result is an increase saving in the aggregate and increase production in the future. And that this is a good thing.

Keyenesians argue that the market process by which an increase in saving would result in an increase in investment can break down.

Hummel points out (correctly) that those breakdowns always come down to "hoarding," an increase in the demand for money, which means less aggregate demand.

I think he dowplays the degree to which Keynesians try to make arguments that increases in the real quanity of money can't solve the problem of hoarding. Somehow, the hoarding always expands to stay ahead of the real quantity of money. (The liquidity trap.)

Graeme Bird writes:

"Fiscal policy works under these conditions, according to Keynesians, because people are indifferent between hoarding money and holding government securities."

Whats to refute? Where is the evidence for this JIve-ass nonsense? No real person treats a government bond as if it were part of his money-balances. Not particularly when he has so many alternatives. Perhaps this was true with World War I bonds, but if we are talking a money substitute we are talking monetary-policy proper. The War Bonds were traded pretty much as money I believe.

Suppose you steal from me and give the money to a drunk in the park. Are you saying this is important? Any extra spending resultant? That the extra spending on consumer goods rather than on, for example, small-cap shares.... Is this important?

You see there is no extra spending. None. The drunk will spend the money in wealth-consuming ways. Under capitalism properly considered the victim would likely have spent that money more productively. But the money gets spent no matter what. Since we spend anything above what our normal cash balances tend to be.

And why on earth would you take on more debt if this nonsense were true? The people of Zimbabwe have shown us that if its more spending you want thats the easiest thing in the world to acheive.

New cash injection via debt retirement, in combination with the appropriate reserve asset ratio, will allow you to hit any spending target you aim at. Is there a Keynesian out there that is willing to claim that a bond which is an increase in debt, creates more spending then cash-money created to pay down debt? Of course not.

This is important. We've just got to stop this silliness that is ruining your country and mine. But mostly yours. We've just got to stop lending credence to idiotic ideas our of a lawyer-like tendency to professional courtesy.

There is no fiscal mutliplier. Its a utopian-crock and it always was. And you've just got to stop being spineless about this.

ThomasL writes:

The paradox to me is how debt is viewed as wealth. I'd need to think on it more, but I would be inclined to agree with the argument as presented by Mr Hummel and Mr Woolsey, if we were actually talking about the saving of current income brought about by productive work.

However, if people were spending outside their means--that consumption financing business expansion beyond what was sustainable had the people been spending within their means--a contraction of both spending and production is natural. It is also a recession.

Those persons refraining from taking on more debt should not be viewed as a "paradox of thrift." There is quite simply a difference between saving money one has now (thrift), and not spending money now, which one expects--or hopes--to have in the future (debt).

I do understand that many people with income exceeding their obligations are still not spending as much as they once did. However, the economy is not reeling because of that smallish group of people. If that were the case, the best possible solution would be to let prices fall, to encourage them spending was a more valuable proposition than saving.

I mention it because it surprises me that these discussions can take place without private debt ever entering into the conversation.

Bill Woolsey writes:


For every debtor there is a creditor.

When people use their current income to pay down their debts, they are paying off their creditors. What do the creditors do with the money?

If people are already in debt and so will not borrow more to spend, what do those who would have lent to them do with their money?

You are ignoring, or implictly assume that they hoard money.

To the degree peopel use checkable deposits to pay off debts to banks, the quantity of money will shrink. But that still eaves banks with excess reseves that they must be hoarding.

I don't think that an increase in the demand for reserves or else an increase in the demand for money are impossible or unlikely. That people might hold money or banks hold reseves rather than lend seems plausible enough to me.

But that must be the source of the problem.

The individual can have too much debt and respond by spending less. Not everyone can be in that situation. For every debtor their is a creditor.

Graeme Bird writes:

Well it doesn't need to be that way. Money can be about commodities in storage and not invisible pyramids of make-believe cash brought about by incontinent debt-creation.

That most of our money is debt and some of it worthless cash is not a law of economics but rather the outcome of a vicious cycle.

Alex J. writes:

This may seem churlish, but what is the optimal amount of hoarding? Might it not change in response to changing economic conditions? If, overnight, flesh eating zombies stalk the street, the amount of economic activity is going to decrease. We wouldn't need a stimulus to aggregate demand, though. Hoarding would be optimal. Likewise, if the banking system is revealed to be run by crooks or the government is throwing around trillions of dollars who-knows-where, demand for liquidity is going to go up, because you don't know how or when you're going to need to spend it.

kurt writes:
Fiscal policy works under these conditions, according to Keynesians, because people are indifferent between hoarding money and holding government securities.
If people are indifferent to holding cash balances at a bank over holding cash balances under their mattress, it would seem to solve the problem as well, without any government help.
Grant writes:

I'm with Jay, Hummel's (a) and (b) are incomplete. You can also argue that an increase in savings does not decrease AD, which is what Russ hinted at in the podcast. An increase in savings decreases the demand for some goods (consumer goods, typically) while increasing the demand for others (higher-order goods as the Austrians say). There is no net drop demand, just a shift in demand for present consumption to future consumption.

Of course when you have a non-functional financial system, an increase in savings may result in hording because there is a breakdown in financial intermediation connecting savers to borrowers. But hording isn't the "normal" scenario that the paradox of thrift deals with.

ThomasL writes:

"If people are already in debt and so will not borrow more to spend, what do those who would have lent to them do with their money?"

You're assuming that this money exists in someone's hands, be it lender or borrower. My argument was that it doesn't. When the lender runs out of borrowers, you assume they still have money to lend, but no customers, essentially they have excess production. Is this the problem you are really observing? Lenders are awash in money, but can't find people anyone to take it?

My view is that you can only create so much debt. There is a limit somewhere, though the exact value may be impossible to state with certainty. I think we are at or near that limit on the whole, which is just made worse by the likelihood that much of that debt isn't any good, since it corrodes the lenders' assets.

Since the creditors' "future money" (new lending) is backed not in a small part by the expected return on their earlier loans -- still-in-the-future-but-not-quite-as-far money, and the repayment on many of those repayments will not be made. Where do they get the money to lend?

There is a lot of new bond investment, financing lending, but I read that as primarily a flight from stocks, which is to say a flight from investment to investment. I don't count transferring between investments in the paradox of thrift.

That flight does work to depress stock prices, worsening the capital position of many lenders even further and impacting their ability to lend. I will admit, however, that it is useful for the initial funding of government stimulus.

This fits quite well as a subset of Mr Hummel's explanation:

A) When people get really scared about the future they sell what else they have to buy government bonds. (Yep, Keynesian...)

B) When lots of people buy bonds, it is a tremendous shot of cash into the government. (Yep, Keynesian...)

As far as it goes I'd agree with it without reservation.

For the paradox itself, if the evidence suggested that:

A) People (meant generally, lenders or borrows) have the money, they just won't spend it.

I might agree with that.

However, what I think has really happened is this:

A) People (again, generally) lent money to other people.

B) The lenders used that loan asset as backing to lend more money to more people.

C) Repeat A,B about thirty times.

D) Turns out half the people that borrowed have no ability to repay.

E) ?

In my scenario, at step E, no one has money.

English Professor writes:

Isn't part of the problem related to the proposition (which can be traced all the way back to Adam Smith) that savings equals investment? And isn't Keynes's claim (implicitly if not explicitly) that in certain kinds of crises, this equivalency breaks down? That is, investments are made by people who perceive a chance for real returns, but when consumers are frightened (because many have lost their jobs and others fear the same for themselves), demand for goods declines. People "hoard" money--i.e., save it--but in such circumstances, that money will not necessarily be turned into new investment. Isn't this where "animal spirits" come in? Entrepreneurs are unlikely to launch new ventures, always a risky proposition, in this sort of economic climate. Even if interest rates drop to near zero, there will be fewer people than usual who want to borrow money and invest it. Hence, you're pushing on a string. If I understand Bryan correctly, he doesn't accept this position: he suggests that any money saved will be invested; that is, the matter is governed more by economic laws than by the pecularities of human behavior. (Or perhaps I should say, he believes that human behavior will nevertheless fall into the patterns described by economic laws.) I'm no fan of Keynes (not that it matters in someone who teaches English literature), but it seems to me that an increase in government expenditure (especially if it is "timely and targeted," which the recent bill is NOT) would do some good in sustaining demand, slowing the amount of unemployment, and alleviating fear. That is, it affects the psychology of the crisis and allows the economy to return to its proper prodcutive nature where savings will once again be turned into productive investment.

But look, I'm only an amateur, so I'm happy to be shown where I'm wrong.

ws1835 writes:

As I read these comments/critiques on Keynesian theory, etc., it seems to me that a central element of the theory is the assumption of wage/price inflexibility. i.e, the problems and financial disconnects tend to be the result of wages/prices failing to respond in a timely fashion to depressed economic conditions. This response lag is what creates the pockets of unproductive resources or unemployed labor that the 'spender' of last resort (government) is boosting via stimulus. Stimulus measures are designed to boost AD until economic conditions recover enough to once again mesh with the inflexible wage/price levels.

With that in mind, it also occurs to me that under free market conditions the opposite mechanism is in play. Wages/prices are very flexible and their response to depressed economic conditions (deflation) is very quick. Hence, in lightly regulated markets the response to depressed economic conditions tends to be very tramatic, but also very quickly resolved. I.e., recessions under such a system are sharp but short. (I am thinking of specific crashes in the 1800's and 1920-21).

Based on the above observation, it appears that the effectiveness of Keynesian theory is directly proportional to the level of economic regulation in effect. It doesn't work and isn't necessary in a free market because without wage/price restrictions the market reshapes itself very quickly. In a highly regulated market, the reallocation of labor and production is inhibited by the regulations and takes much longer to be completed. Hence, there is a significant time period during which pockets of unused resources emerge and Keynesian stimulus can be helpful. Although in that situation, properly targetting the stimulus is the key and it is difficult to do in practice.

Is my line of reasoning a good interpretation for a layman?

Lee Kelly writes:

Suppose that you are stranded on a tropical island, and must produce for yourself enough food, shelter, clothing, and tools to survive. At first you produces at the same rate that you consume (i.e. you work each day produces enough food, shelter, clothing, and tools to work again tomorrow). After living on the island for some time, however, your skills improve, and you are able to produce at the same rate as before but with less work. but you decide to forego more leisure time and keep working, and invest some of these additional resources to improve your life. So you plant crops, construct better shelter, weave new clothing, and develop more sophisticated tools. To each of these activities resources must allocated, and until they are completed no benefit will be had. In other words, to increase your quality of life, you must forego consumption in the present to increase production in the future. If you did not forego consumption, then there would be no resources with which to invest in these endevours.

The principle that prosperity is created by underconsumption applies no differently to the economy as a whole than it does for you. It is saving--production without consumption--that makes available the resources upon which future prosperity can be built. Resources are today allocated for investment via banks. Interest rates, the price of borrowing, should be calibrated to the rate of saving i.e. interest rates should rise and fall relative to the stock of resources being made available to invest with. And this can occur despite the so-called paradox of thrift.

Scrooge would have been a public benefactor because he hoarded money and lived frugally. By producing without consuming, he would have pushed down prices for everyone else. By giving more to the world than he was taking from it, he would have increased the purchasing power of everyone elses money. While wages would also fall, so long as people are underconsuming, the relative supply of labour to other goods and services would decrease, and so the purchasing power of wages would rise. But here is the real interesting thing: even if Scrooge locked up all his money in a private vault, thereby depriving banks of some reserves from which to lend, the price system would still reallocate more resources toward investment. The so-called paradox of thrift rightly notes that the banking system's total reserves do not change when someone saves. But it neglects the fact that prices fall, and thus, the purchasing power of those reserves must increase. We have seen how this might work in recent months: a $200k loan in February of this year would have bought more house, hire more workers, and command more goods than it would have two years ago. The so-called paradox of thrift is a result of thinking like an accountant rather than an economist i.e. thinking about the numbers rather than actual resources and their uses.

Unfortunately, interest rates, which ought to coordinate production and consumption across time, are centrally planned, or at least centrally manipulated. All the recent saving should be reducing interest rates from some previous high, freeing resources for investment, but no 'previous high' was permitted. There is little trust, little confidence, and what is even worse, no knowledge of what to do next. The economy has walked so far along an unsustainable growth path that few can retrace their steps to discover where we got lost. Nobody knows what the correction will look like, and discovering it from such a position that we are currently in will be long and difficult.

Bill Woolsey writes:


No one is arguing that an increase saving never results in more saving. Everyone recognizes that an increase in saving (reduced consumption out of current income) matched by an increase in investment (purchases of capital goods by firms) results in more aggregate saving and increased producing in the future.

The paradox of thrift is actually a good insight, _if_ you go one to explain how an increase in saving by the individual leads to increased capital goods spending by the firms and shows how in aggreate saving is about shifting production from current consumer goods to capital goods and so future consumer goods.

Of course, if you assume investment is constant, then aggregate saving will not increase. It is possible for one household to save and another to dissave as well. But that transfer of current consumption between households now in exchange for a reverse shift in the future doesnt' increase saving in the aggregate.

What is wrong with the Keyneisan use of the paradox of thrift is it fails to point to the monetary disquilibrium that is at the heart of a failure of invidivual saving to translate into increase aggergate saving (through increased investment.)

No one claims that individual saving _never_ results in increased aggregate saving.

Scott Sumner writes:

Keynes is right that if savings rises, interest rates fall, and velocity falls, and (holding money constant) nominal GDP will fall. And every good economist going back to Hume understood that falling NGDP is partly prices and partly real output in the short run and all prices in the long run. So what did Keynes discover? According to Hicks and Friedman, only one thing. Keynes argued that the central bank might not be able to offset the fall in velocity due to a liquidity trap. Unfortunately, the one thing Keynes discovered in the GT was wrong, as under a fiat regime the Fed can always make M go up by more than V falls. So I don't see anything useful in the GT that wasn't already known to pre-Keynesian economists. Maybe I missed something.

Graeme Bird writes:

"Keynes is right that if savings rises, interest rates fall, and velocity falls...."

Its only velocity that counts here. And you are using GDP in error. Nominal GDP would fall directly is savings increases and this would happen even if the velocity of circulation picked up.

You see you ought to be thinking in terms of Gross Domestic REVENUE. Not Gross Domestic Product.

"Unfortunately, the one thing Keynes discovered in the GT was wrong, as under a fiat regime the Fed can always make M go up by more than V falls."

This is true. And obvious. And we must assume malign ill-will by people who refuse to accept it. Its a bit of an acid-test as to who is a tribal goon and who isn't. So given that this is true is there anyone here still so famously idiotic that they think that fiscal stimulus is a reasonable policy.

Economists have to grow up, put childish ideas away, and stop misleading the public.

Bob Layson writes:

Wage earners are as inflexible as government pays them to be through the offer of bailouts to companies and benefits to workers made redundent.

Conceptually, however, redundent workers in receipt of the dole may be regarded as providing a service for which governments would happily pay them a little of other people's money: not undercutting union wage-rates, not circumventing minimum wage laws, voting for those who take care of them when ''no jobs are to be had'' and showing the inefficiency of the market and the need for interventionist government by intelligent and farsighted politicians - or, if need be, by actually existing politicians.

English Professor writes:

ws1835 wrote: "As I read these comments/critiques on Keynesian theory, etc., it seems to me that a central element of the theory is the assumption of wage/price inflexibility."

As one amateur to another, let me just say that when I first read the GT, this struck me as one of Keynes's central tenets, and it infuriated me. I inferred from his text that he was not talking about "sticky" wages but about political realities--that is, unions would resist any reduction in the wage level, so any modern economic theory had to accept wages that don't adjust downward in response to market forces. It seemed to me that he had incorporated this political situation into his economic theory, and that struck me as BAD economic theory. I haven't gone back to the GT to see if my reading was justified, but I would not be surprised to find that it is.

Comments for this entry have been closed
Return to top