Arnold Kling  

Back to Basics

PRINT
Why People Hate the Octomom... Murphy on the Paradox of Thrif...

Robert P. Murphy and Bryan seem to me to be confusing things.

Murphy starts with an illustration of the paradox of thrift from Keynesian economist Steve Fazzari. I don't like Fazzari's illustration, so perhaps my quarrel is with him.

To me, the paradox of thrift is simple.

Y = C + I
C = A + cY

Y is income, C is consumer spending, I is investment, A is autonomous consumer spending, c is the marginal propensity to consume out of income.

Fix I at $100, A at $100, and let c equal .9 Then Y = $200/(1-.9) = $2000. Savings, S = I = $100.

Now, suppose that consumers try to increase saving, by reducing A or c. If A goes to $50, then Y = $1500 but saving stays at $100. Instead, if A stays at $100 and c drops from .9 to .8, then Y = $1000 but saving stays at $100. The paradox is that consumers cannot increase saving. Only investors, with their animal spirits, can increase saving.

Let me see if I can put this in terms of a homely example. Our family decides not to go out to eat this week, because they want to save and buy a new TV next year. However, our interest in a new TV does not get conveyed to the TV makers. So, the restaurant cuts back employment and output, but the TV makers do not increase employment and output. Total employment and output fall. Somebody somewhere reduces their saving because they don't have a job. That person would have bought a TV next year, but now they cannot afford it. So our family gets the TV, and the TV maker does not regret failing to make more TV's.

Now let's turn to Bryan.


If the Internet boom never happened, I say that the private sector would have spent on something else, or lent it out to someone else to spend on something else.

In other words, the spectacular increase in employment would have occurred with or without the Internet boom. The Internet boom is nothing but a "just-so" story that we tell about the increase in employment. Maybe, but in my opinion there was a genuine fluctuation in employment, and I think that it was genuinely caused by the Internet boom. I'm with Brad DeLong on that one.

Going back to our homely example. Suppose that the TV makers, instead of failing to realize that our family is saving for a TV next year, over-estimate the number of TV's being saved for. Consequently, they invest in new TV-manufacturing capacity this year, causing investment, I, to go up, which in turn causes saving, S, to go up, along with employment and output. If the TV makers are lucky, the saving will go toward purchases of new TV's next year. However, if it turns out that the TV makers were wrong, next year they will have to lay off workers and we could have a general slump (assuming those workers cannot quickly be re-deployed in the economy).

I think it really is possible to have fluctuations in employment. Bryan's thinking applies in a frictionless economy that is always at full employment, but I don't think it applies in reality.


Comments and Sharing


CATEGORIES: Macroeconomics



COMMENTS (21 to date)
Blackadder writes:

Our family decides not to go out to eat this week, because they want to save and buy a new TV next year. However, our interest in a new TV does not get conveyed to the TV makers. So, the restaurant cuts back employment and output, but the TV makers do not increase employment and output. Total employment and output fall.

Suppose that, instead of saving the money, the family decides to give it to their friend Maynard, who agrees in exchange to give them the money to buy a TV at the end of a set period of time. Maynard then spends the money on whatever. Do total employment and output fall? No.

Now suppose that, instead of lending the money to Maynard, the family puts the money in a bank, and the bank lends the money to Maynard. Again, total employment and output don't fall, even though the family is saving more.

The problem with the paradox of thrift is that it assumes that when money is saved it just kind of disappears from the economy. But of course that's not what typically happens.

Unit writes:

The more I read about these "macro" arguments, the more it all sounds like hocus pocus. Sorry but my head is spinning.

KipEsquire writes:

"Fix I at $100..."

Stopped reading after that.

Increased saving results in increased investment, at least dynamically. S=I is an identity, not an equilibrium condition.

And increased investment of course results in increased output -- again, at least dynamically.

Comparative statics is the last refuge of Keynesian scoundrels.

Gavin Andresen writes:

So: you decide to save for a TV-- where do you park your money?

If you put it under your mattress, it seems to me we've got the paradox of thrift-- you'll spend it later, but in the meantime you've put somebody somewhere out of work.

If you give it to a bank, then it seems to me the outcome will depend on how wise the banker is. If the banker is good at spotting great, productive investments (maybe a new restaurant on the other side of town that is more productive than the one you were thinking of eating at, or maybe that newfangled internet TV startup) then I don't think there's a paradox-- money's gone from a less productive use to a more productive use, and overall we're better off.

If the banker is an idiot and invests in over-priced Las Vegas real estate... well, we're seeing how that turns out.

newt0311 writes:

@KipEsquire

Agreed. Here's an interesting thought experiment.

Let us fix Y at $1000 instead of fixing I. Then reducing A by $50 increases I by $50. Viola!!! we can alter investment by altering consumption.

Not really. The fact is that we have two equations and 5 unknowns. Thus, as long as we fix 2 to 3 quantities, the others are just up for grabs. Those equations are useless for predicting causality.

Nathan Smith writes:

The idea of the "paradox of thrift" depends on investment markets being somehow dysfunctional. The phrase "animal spirits" is key here. It expresses a contempt for the possibility that investors are rational agents who respond to incentives. If they are, the story falls apart.

The idea that the collapse in investment in the 1930s was a result of "animal spirits," as if it had nothing to do with the Hawley-Smoot Tariff and taxes being hiked to confiscatory rates and the anti-business slant of FDR's policy and rhetoric, is so bizarre that one has trouble believing in Keynes' ingenuousness. It's pretty obvious that when the government is engaged in arbitrary prosecutions against leading industrialists like Mellon and Sam Insoll, destroy the value of Commonwealth and Southern shares by jumping into the utilities business, encouraging runaway unionization, and raising taxes again and again to rates over 60%, 70%, 80%, you'd be a sucker to spend a lot of money now on investments that will pay off later. The paradox of thrift can indeed occur if savings cannot be translated into investment, but the dearth of investment is a function of sharply increased political risk.

In the 1930s, and today.

Niccolo writes:

Prof. Klein,


Coming from an entrepreneur, I assure you, we do not rely on absolute measures of what one family is expecting to purchase in order to expand production. This is simply not how investors and entrepreneurs work - as natural risk takers we expand, meeting the needs of the unemployed gentleman.

In this example, say that the TV maker does not know for a fact that the family will want a TV, but assumes that some family will. They thus expand production, and with the prior assumption of the family's increase in consumption, they're investment pays off and the free market for competing goods wins out again. An expanding and successful sector grows, a declining sector that needs to reconsider either budget or marketing declines.

To the point of frictionless economies, I agree, but I honestly don't think it says much to point out that economies are processes and part of adjustments in processes will mean a snapshot of time that appears to be an increase in unemployment and not merely a time lag between adjustments.

Kit writes:

As the family saves wouldn't the cost of capital decrease for the TV maker triggering them to invest for future production?

Bill Woolsey writes:

Those comments that refer to "parking" money, have the right idea.

The problem is that many of these comments fail to recognize that this is a real issue. Breezing by with the implicit assumption that no one will do that, is a mistake.

People, in fact, do hold money. The amount of money they choose to hold can change. What is the market process that brings that back to equilibrium?

Regardless of whether the money is "in the bank or not," changes in the amount of money people want to hold will impact spending in the economy.

Getting from people choosing to hold more money, so they cut their current expenditures, to businesses buy more capital goods, is problematic.

An increase in the demand for money requires an increase in the real supply of money, which could occur by a lower price level or else by an increase in the nominal quantity of money.

Or, of course, perhaps government borrowing will convince people to reduce how much money they want to hold. And perhaps inceasing the quantity of money (if it is done in certain ways) will cause people to choose to hold even more money.

These are the central questions--at least if you are asking what determines total spending in the economy.

Bill Woolsey writes:

Those comments that refer to "parking" money, have the right idea.

The problem is that many of these comments fail to recognize that this is a real issue. Breezing by with the implicit assumption that no one will do that, is a mistake.

People, in fact, do hold money. The amount of money they choose to hold can change. What is the market process that brings that back to equilibrium?

Regardless of whether the money is "in the bank or not," changes in the amount of money people want to hold will impact spending in the economy.

Getting from people choosing to hold more money, so they cut their current expenditures, to businesses buy more capital goods, is problematic.

An increase in the demand for money requires an increase in the real supply of money, which could occur by a lower price level or else by an increase in the nominal quantity of money.

Or, of course, perhaps government borrowing will convince people to reduce how much money they want to hold. And perhaps inceasing the quantity of money (if it is done in certain ways) will cause people to choose to hold even more money.

These are the central questions--at least if you are asking what determines total spending in the economy.

Bob Murphy writes:

I agree with the commenters who are saying Arnold basically assumed his conclusion, yet doesn't seem to realize how simple his argument is. We can drop most of the other stuff and just focus on:

S = I = $100.

If investment is fixed because of animal spirits etc., and savings=investment by accounting, then obviously changes in the propensity to save can't alter savings, unless they influence animal spirits. QED.

At this point, one response is to point out that all of this is in nominal terms. I.e. even if people stuff money under their mattress and nominal incomes fall, prices fall too, so real income doesn't fall as much, and real savings does increase.

I didn't get into this point in the EconLib article, but maybe I should have since I think I left Fazzari an escape hatch.

Gary Rogers writes:

The first equation makes sense: Consumption plus savings equals income. The second equation does not. Spending decisions are made based on propensity to consume times discretionary income not total income. Autonomous consumption, if I understand it correctly, is consumption that comes out of savings or from borrowing and therefore cannot be included without relating it back to the change in investment.

I think when you look at it this way, the effects of past debt on future consumption become more obvious and there is no paradox.

David W. writes:

Algebraically, I agree that those equations show that result. I don't see how those equations relate to reality, though - not only the problems with the identities pointed out by Gary, Gavin, Newt, etc, but also things like assuming the 'marginal propensity to consume' is a single value, rather than a function of income.

It's true, when model building, the whole point is to simplify things so that they're more easily understood - but you can't simplify out the important parts, or it no longer reflects reality! That's why physicists can assume a spherical cow, but biologists can't.

Arare Litus writes:

Dr. Kling;

With your equations if you have a society of imprudent spenders (c=1) you get infinite income; I take this to indicate that the equations ignore finite velocity of money (and likely other effects**). I have not sat down and studied macro (not even "101"), but it seems aggregate variables and equations are much too simplistic to sufficiently capture reality to reasonably guide policy (though many with PhD's in economics disagree with this apparently!).

**My naive view is that value is ignored in the Keynesian perspective - the internet boom (or any growth) occurred because we discovered a new source of wealth and value, and therefore invested in it. This creates value - and as the true value is a revealed quantity we cannot capture this in equations. Recessions seem to occur when we find the true value was less than hoped, or when shifts in production are delayed by stickiness (I may be misusing stickiness here - I mean human and other capital is optimized for other means of production, and have to be retooled to extract wealth from the new sources).

Speaking of back to basics:

The links on your "Learning Economics" e-version of your text do not work, and the .wma files does not seem to work (not in my browser, or the audio programs I have tried).

I have started reading your posts and find I learn well from your presentations [I was drawn to the blog via Dr. Caplan (via Russ Roberts EconTalk), and happened to hit on his book club - now I'm starting to read other posts and authors here - very nice; thank you all GMU clan]. If you could find time to update your Learning Economics links/audio files it would be greatly appreciated.

Cheers,
Arare

Niccolo writes:

Again, I just want to note that Arnold's example implies that entrepreneurs are not really risk takers, but are merely channels for data.


This is simply not correct for many entrepreneurs - I would wager almost every entrepreneur. Simply put, it assumes that entrepreneurs are not risk takers; it assumes that entrepreneurs don't exist.

Arnold Kling writes:

I do not claim that the simple equations I put up are true of the real world. They are the textbook rendition of the paradox of thrift. I think it is coherent. The key assumption is that investment demand is not sensitive to interest rates. If you don't think that there will be demand for TV's in the future, then you don't undertake investment even if interest rates are low.

Again, I do not claim that is true in the real world. But Keynes' argument for the disconnect between savers' decisions and the animal spirits of investors deserves some consideration. I didn't see that articulated in the Fazzari piece or in Murphy's commentary

Bob Murphy writes:

Arnold Kling wrote:

Again, I do not claim that is true in the real world. But Keynes' argument for the disconnect between savers' decisions and the animal spirits of investors deserves some consideration. I didn't see that articulated in the Fazzari piece or in Murphy's commentary

Did you not like my fancy model with time subscripts? :) I did try to address your point (in that part of the article). I was saying that if intertemporal plan coordination gets screwed up, the question then becomes: how best to recoordinate? And I think letting markets set prices does it best, rather than having politicians start throwing variable amounts of money at politically-connected sectors.

El Presidente writes:

Arnold,

The key assumption is that investment demand is not sensitive to interest rates.

Not exactly. I appreciate the distinction you make between reality and theory. I think the assumption is rather that investment demand is not sensitive to interest rates alone. That is, return on investment matters for capital allocation, volume of investment, and MPS, but without a market for a product the return is non-existent. This is where "animal spirits" (I really hate that term), or the prospect of economic profits and all associated social and material advantage solicits additional investment from existing income. A promise of higher return must be _believed_ in order to change total investment. Interest rates matter because they set the ROI necessary to induce this behavior, to tell us whether or not a market should exist for a given product at a given time. They tell investors what base to add to their risk premium before they will consider increasing investment. You are right that this is all wrapped up in "animal spirits".

Arare Litus writes:

Dr. Kling,

"I do not claim that the simple equations I put up are true of the real world. They are the textbook rendition of the paradox of thrift. I think it is coherent."

I don't think anyone is making a strong claim that you do state the equations capture reality (my statement, for example, of "(though many with PhD's in economics disagree with this apparently!)" was not some snide comment on you, but a disclaimer of my ignorance of the subject, as well as my observation that policy is being driven by this mindset right now). I personally was simply pointing out the equations clearly break down and have strong limitations (and therefore paradoxes are not surprising). This very approximate nature of the equations require an extremely careful presentation, to guide focus to the key mechanism.

It may be clear to you that the rendition is coherent, and that "Only investors, with their animal spirits, can increase saving.", but for those without the background (such as myself) there seems to be a lot hinging on the investors/spirits mechanism which is not explicitly explained in the post or the equations - how do these investors increase saving? [is the point simply that the equations do not capture the "animal spirits"? - and therefore only they can take up the slack? In that case the interesting next question is: how? Without the explanation of how, the assumption that animal spirits pick up the slack remains an assumption, supported by the argument, but not demonstrated.]

As the equations are a stylized presentation, and it is clear you have a sharp mental image of what is going on, I am hoping that what is implicitly there is drawn out more. If the matter is merely that the equations do not capture things, and that therefore the animal spirits are the only mechanism left for something that we do clearly observe, the it would be clarifying if this is explicitly stated - as, I for one, include the how? question in the paradox: this may simply be misunderstanding of what the paradox and its explanations are addressing.

I think a lot of the confusion regarding this paradox, and explanations of it, are "hidden priors" that inform the presenter, but not clearly visible to the audience and discussion participants. This may simply be due to basic ignorance on the audiences part, but it also seems that the confusion is widespread among knowledgeable economists [for example Russ Roberts econtalk discussion with Steve Fazzari clearly indicated a mismatch between the parties - with the key question on this paradox not being resolved in the discussion, which had to be dropped due to lack of progress in the available time; also, in the very next post on the blog, David Henderson indicated he thought you are flawed].

It simply is not clear (to me) what the equations are suppose to be capturing and indicating: the fact that they are flawed and stylized promotes confusion unless the argument is very clearly presented, piece by piece, with all priors, steps, and conclusions indicated. I believe the confusion on this paradox is widespread (hence "paradox"), even among economists.

Perhaps Bryan Caplan and Robin Hanson should debate this paradox in their up coming GMU Econ Society debate.

Dan Clementi writes:

Bob Murphy wrote:

if intertemporal plan coordination gets screwed up, the question then becomes: how best to recoordinate? And I think letting markets set prices does it best, rather than having politicians start throwing variable amounts of money at politically-connected sectors.

BINGO! I've been force-feeding myself the economics blogosphere for months now, and I conclude that Bob is right. IMO, short-term negative effects to the overall economy probably occur upon ANY sudden disruption in expectations (including the restaurant owner suddenly seeing his revenue slump). So what else is new? Aren't we interested in how to reach LONG-TERM prosperity? I can't see how we get there by pumping air (credit) into bubbles that have already burst (housing, cars, stocks, etc.) Further, I can't see how we get there by forcing capital and human resources through an intermediate step of government allocation. It can only delay their proper deployment.

Charlie Poole writes:

This statement was in the article above: "I think it really is possible to have fluctuations in employment." Wow! What a revelation. I look forward to hearing whether you think it is possible to have fluctuations in inflation.

This is the age of boundless confusion...

Comments for this entry have been closed
Return to top