Bryan Caplan  

The Myth of Time Preference

PRINT
The Hopeful Science... Sin City and the Bizarr...

How come interest rates are always positive? Austrian economists have a stock answer: it's because of time preference. All else equal, we prefer satisfaction sooner rather than later. If we did not have time preference, we would never consume anything, because we would keep delaying consumption. As Mises puts it:

We must conceive that a man who does not prefer satisfaction within a nearer period of the future to that in a remoter period would never achieve consumption and enjoyment at all.

Unlike a lot of Austrian doctrines, this view is plausible. But it's wrong nevertheless. You don't need time preference to get people to divide their consumption between today and tomorrow; all you need is diminishing marginal utility. If you are stuck on an island with two bananas for two days, a perfectly patient person would still want to eat one banana per day. Even though he disvalues hunger today and hunger tomorrow equally, eating one banana today assuages his hunger more effectively than saving that banana for tomorrow.

But if diminishing marginal utility is a sufficient explanation, how come the price of consuming now is always greater than the price of consuming later? Don't you need time preference to explain why interest rates are always positive? Not really. Gold has (almost?) always been more expensive than silver, but we don't need to postulate "gold preference" to explain this pattern. The greater scarcity of gold is all the explanation we need.

But before we try to explain why interest rates are always positive, we should make sure that they are always positive. In barter markets, interest rates are frequently negative. Suppose we knew the price of food would double next year. Then a pound of food now trades for half a pound of food one year from now. Translation: a negative 50% interest rate!

If this seems crazy to you, suppose food were the only commodity, and you expect a famine next year. Wouldn't you happily trade 2 pounds of current food in exchange for a promissory note good for 1 pound of food next year?

When you put it this way, it isn't too hard to figure out why interest rates on money are always positive. Unlike food, money doesn't spoil, and costs almost nothing to store. So if the interest rate fell below 0%, lenders would simply hold their money instead of lending it. (This also explains why inflation-adjusted interest rates on money often are negative. Lending money at 5% when inflation is 10% is a better deal than sitting on it).

Focusing on time preference also leads Austrians to miss another important reason that pushes up interest rates: economic growth. In the modern world, the typical person gets richer in the typical year. Once again, this gives even perfectly patient people a reason to increase their demand for current consumption. Imagine you are going to inherit $1,000,000 next year. According to the law of diminishing marginal utility, you would want to increase your consumption now when the marginal utility is high, and pay for it by cutting back your consumption in the future when the marginal utility is low. No time preference story need apply.

Of course, none of this means that time preference does not exist. It does. But you don't need it to explain the existence of interest. Diminishing marginal utility does that job. And you don't need time preference to explain why interest rates are always positive. Many aren't; and the ones that are always positive involve products that don't spoil and don't cost much to store.


Comments and Sharing





TRACKBACKS (9 to date)
TrackBack URL: http://econlog.econlib.org/mt/mt-tb.cgi/213
The author at PRESTOPUNDIT in a related article titled BRYAN CAPLAN argues that "time preference" is a writes:
    myth not an explanation for interest rates.... [Tracked on March 11, 2005 4:09 AM]
The author at Mises Economics Blog in a related article titled Time Preference Confirmed writes:
    Among the critics of Austrian economics none is so close, yet so far, from being an Austrian as Bryan Caplan. In his latest effort to discredit Austrian Economics Bryan has taken on the idea of time preference. According to Bryan... [Tracked on March 11, 2005 11:35 AM]
The author at Mises Economics Blog in a related article titled Hey Rocky, Watch Me Prove Positive Interest Rates Without Time Preference! writes:
    OK, so it's not Rocky and Bullwinkle. But it seems like it. Bryan Caplan claims to have demonstrated that time preference is superfluous to the explanation of positive interest rates, since marginal utility all by itself explains positive interest rate... [Tracked on March 11, 2005 11:57 AM]
COMMENTS (28 to date)
Mike Linksvayer writes:

Interesting article, has encouraged me to go back and digest http://www.gmu.edu/departments/economics/bcaplan/whyaust.htm soon.

Isn't it a stretch in your 'food price doubles next year' example to say there's a negative interest rate because you can get half a pound of food next year for one pound now? If you translate those quantities of food delivered now and 1 year from now into money you get a zero percent interest rate: if you know the price of food doubles next year, you're trading $1 worth of food today for $1 worth of food next year.

Maestro writes:

Good points. Related to the non-spoilage factor is my point. If you could receive a new car now or in a year, you don't need time preference to choose to take it now. You take it now so that you will enjoy it for a year longer than you would with the second option. However, this isn't necessarily so if the wear and tear factor is great enough.

Steve Miller writes:

Let me think this through, using the tired beer example. Diminishing marginal utility means that the first beer is enjoyable, the second possibly a little more, but eventually subsequent beers have diminishing marginal utility. This particular example sort of assumes that we're talking about one time period of some fixed length. I suppose what you're saying is that this can be extended to a lifetime, also of a fixed length. Let's see if that's true of the beer example. Beers consumed early in life have a higher marginal utility than beers consumed later in life? That doesn't make nearly as much sense as saying that beers consumed early in the evening have a higher marginal utility than beers consumed later in the evening. The buffet meal story in the link you gave is very similar.

But beer is weird like that, it goes through your system (my dad says you never buy beer, you only rent it). So maybe cars? Do cars purchased earlier in life have a higher marginal utility than cars purchased later in life? Or is the diminishing marginal utility about how many cars I have at once? The first car is very useful, it makes it possible to go all kinds of places. The second one is slightly less useful at the margin, but still useful (it can serve as a backup to the first car when it breaks down, etc.). The third has a lower marg. utility, and so on. But what the heck does that have to do with borrowing money to have one car now? It still seems that the reason I would borrow money to buy a car is because I prefer a car now to a car in the future. That's time preference. It seems like I'm willing to pay interest because of a time preference, not because subsequent cars purchased will have lower marginal utility. It doesn't really seem like the concept of diminishing marginal utility has much to do with time at all. If you're saying that the marginal utility of a car now is higher than the marginal utility of a car in the future, then we're saying the same thing, and it's just a matter of semantics... isn't it?

Steve Miller writes:

Maestro wrote: "If you could receive a new car now or in a year, you don't need time preference to choose to take it now. You take it now so that you will enjoy it for a year longer than you would with the second option."

Taking a car "now so that you will enjoy it for a year longer" sounds exactly like time preference to me.

Maestro writes:

'Taking a car "now so that you will enjoy it for a year longer" sounds exactly like time preference to me.'
No, at least not as I understand time preference. I want a doughnut now rather than tomorrow, not so I can enjoy it for a day, but because I want it now, I prefer it now, my time preference is for now. Maybe I'm misunderstanding time preference. If so, please correct me.

Don Lloyd writes:
If this seems crazy to you, suppose food were the only commodity, and you expect a famine next year. Wouldn't you happily trade 2 pounds of current food in exchange for a promissory note good for 1 pound of food next year?

Actually, no. If you expect a famine, your promissary note may be worth no more than the calories it provides when you eat it.

Regards, Don

Steve Miller writes:

I'd say they're both basically the same thing. With the car, because it stays useful longer, you get to use it today and all the days in between. For a doughnut the time period is measured in hours at most. But you'd still rather use it sooner than later. If you desired a doughnut and didn't have the 50 cents to pay for it, you might consider my offer to give you a doughnut today in exchange for 51 cents tomorrow.

Michael Messina writes:

I think the banana example is flawed. If you just cared about marginal utility, you would never eat them because their utility would increase the hungrier you got. Also, saying that we disvalue hunger equally both days is sort of like asking "Would you like to live today or tomorrow?". You can't live tomorrow if you die today. Also, the negative 50% interest rate on food example seems flawed, because there isn't a negative interest rate in dollar terms. The famine example gets back to the live today/tomorrow problem and there isn't any money basis given.

There are two sides to a lending transaction, borrower and lender. It seems obvious that the borrower has a time preference for reasons Steve Miller gives above, while the lender presumably has excess funds with a corresponding diminished marginal utility. However, I suggest that the lender actually has a time preference as well.

Time preference is the idea that present consumption is valued over future consumption; however, this is a reversible concept: there is an increased amount of future consumption that is valued the same as present consumption. If you are offered a promise of excess value in the future over the value of current consumption, then, rationally, you would take it. This makes the statement "time preference explains positive interest" a tautology. A time preference means a positive interest rate and vice versa. Bringing inflation into the mix doesn't affect this since once the funds are lent there is no longer any time preference at work. The positive interest rate in question is the one set at the time the money is lent.

Greg Ransom writes:

What Caplan calls "the stock answer of Austrian economists" is _not_ Friedrich Hayek's answer, so it's deeply misleading to suggest that Austrians have only one answer to this question. Hayek shows that time preference and productivity are inter-related and combine to produce positive "interest" returns on _capital_. Bringing in positive returns on _money_ adds a whole different and new level of complexity to the matter.

Of course, economists have punted on all of this -- they don't have a theory/logic of disaggregated interest/capital. In other words, they haven't produced a logic of the marginal valuation of production goods through time, as they have with the valuation of consumption goods without production. And what I mean by "punted" is simply this -- they haven't tried, they haven't looked into why it might not be possible to do so, and they haven't comtemplated the significance of their failure to do so. And you are right if you're thinking that this is about as embarrassing as it gets in a scientific discipline.

For those interested in the topic of "Austrian" interest / capital theory, let me recommend _The Austrian Subjectivist Theory of Interest_ by Ingo Pellengahr. Peter Lang: 1996.

Bob Knaus writes:

I think Mike and some others misunderstand Bryan's example. In barter markets, of course the interest rate is frequently negative, because the unit of exchange is FOOD, not money.

If you move to a market where the unit of exchange for delivery of food is money, then of course $1 of food today = $1 of food next year discounted to present value at the prevailing interest rate. But then it's not a barter market any more, is it?

This is why, despite what you read in some sleazy promotional materials, that it is impossible to be certain of making money in the commodity futures markets on seasonal prices swings.

Luca writes:

In the barter example it looks like that the factor that makes the interest rate negative is not the barter mechanism but the assumption that food is perishable. If you store some salted cod today, you can barter it for twice the stuff a year from now, and therefore the interest rate is 100% (minus storage costs) instead of -50%.

Regarding the economic growth example: isn't a difference in marginal utility just influencing my time preference, rather than being a completely different interest rate framework?

Randy writes:

This reminds me of a Finance class I took years age. I was asked to explain the time value of money. My answer was something about the money amount changing over time because time has value. He told me I was wrong - what he wanted was the definiton - but I always thought I was right.

I think Bryan is right, but I still think I'm right. Time has value. Products have value. What interest rates determine is the relationship between the value of the time and the value of the product. Or I might just be wrong...

Joseph R. Stromberg writes:

[I suppose we can expect something like this in the near future?]

Forthcoming by B***n K****n: "Why Wheels Do Not Require the Concept of Roundness."

Summary: Austrians are wrong to attribute roundness to wheels. All that is needed is a bounded constantly receding surface set upon a larger constantly flat surface. Throw in Bayesian probability (or some such thing), and we see that Austrians have been mistaken about roundness for almost 200 years. Roundness is just a wheel in the head.

Joseph R. Stromberg,
Historian in Residence,
Ludwig von Mises Institute

Mike Linksvayer writes:

Bob Knaus: The example stipulates that the price of food will double next year. However, disregarding that and taking food as the unit of exchange, it sounds to me like the next year is deflationary and you have a zero real interest rate.

Bryan Caplan: Upon re-reading, I missed the argument that diminishing marginal utility explains the existence of interest, i.e. the argument does not exist above. All you've explained above with diminishing marginal utility is why people might want to delay consumption. That seems completely orthogonal to time preference, and a non-explanation for interest.

Finally (me not knowing anything about theories of interest) isn't it obvious that interest is a price, determined by a meeting of supply and demand, and is never negative for the same reasons blowout sales might have really low prices, but never zero nor negative prices?

Nominal interest rates aren't always positive. They were negative in Japan a few years ago. The reason being http://research.stlouisfed.org/publications/mt/19990101/cover.pdf

Pete Canning writes:

"Gold has (almost?) always been more expensive than silver, but we don't need to postulate "gold preference" to explain this pattern. The greater scarcity of gold is all the explanation we need."

Actually, if no one had a preference for gold over silver gold would be cheaper than silver. Even though there is less of it.

jsale515 writes:

Bryan Caplan completely misconceives the relationship between the law of marginal utility, time preference, the intertemporal allocation of resources and the interest rate. Time preference is a "category" of human action, meaning that any act undertaken brings the actor's goal closer in time and demonstrates a preference for satisfaction sooner rather than later. Another way of putting it is that no one has an infinite time horizon. Everyone without exception has a finite "period of provision," beyond which the achievement of any goal is valueless to the actor. The term "positive time preference" is therefore redundant and the concept of "negative time preference" is logically contradictory. Zero time preference would occur only in a world in which everyone was completely satisfied at every moment of time and need not ever act--think of a world in which consumer goods rained like manna from the heavens at every spot on earth and each individual could instantaneously clone himself an infinite number of times so that his capacity for enjoying this superabundance was also unlimited and you come close to imagining a zero time preference world.

One implication of time preference is that an individual stranded on an island with an absolutely unchanging or, more accurately, evenly recurring scale of values and endowed with a limited stock of durable consumer goods that he has no prospect of ever increasing would not consume the goods all at once but at a progressively diminishing rate that roughly equalized the discounted marginal utilities of the goods over the period of provision determined by his time preference. So he would save most of his goods in the early stages of his period of provision because not to do so would mean satisfying a present want for a good (the 100th ranked today) whose marginal utility is exceeded by the discounted m.u. of some future want for the same good (the 10th ranked 5 years from today). Nonetheless his standard of living would fall progressively through time until eventually he will have consumed his last morsel and passed from the scene. So people in the real world money economy are indeed always striving to smooth out their consumption patterns through time, given their changing income prospects and anticipated fluctuations in their value scales. Nonetheless time preference is always driving each actor to an intertemporal allocation of his money income that results in the (expected) discounted m.u. of a satisfaction from a given increment of income in any prospective period of time exceeding the discounted m.u. of the sacrificed alternative satisfaction from the use of that increment in any other prospective period of time. The fact that the cumulative discount rate rises to infinity and the discounted value of a unit of income falls to zero after a finite period of time for everyone explains why no one plans for the indefinite future and illustrates how everyone demonstrates categorial time preference.

In a money economy--but not in a Crusoe or barter economy--the rate of discount on future want satisfactions is indeed reflected in a positive and unitary pure interest rate on money loans and monetary investments in the structure of production. As Mises pointed out in the early 1930's in a barter economy there is no possibility of a single interest rate because each good has a unique "own" rate of interest and there are as many rates of interest as there are goods. The reason for the multiplicity of interest rates in a money-less world is that the intertemporal exchange ratio for every good reflects considerations of both time preference and the anticipation of its unique future price movements.

Actually Bryan Caplan's post raises an interesting sociological question: why is it that so many people who have done little if any academic research in Austrian economics are continually driven to pontificate on Austrian economics? Why hasn't the same phenomenon occurred with, say, Chicago economics or New Keynesian economics? Maybe Bryan can Blog an enlightening answer on this question.

Lancelot Finn writes:

Well, I'm with you. Time preference always seemed fishy to me, because it is arbitrary. It's one of those things economics students have trouble with. "So you're saying people are irrational, they want everything now and don't think about the future enough?" "No, they're rational. The future is worth less than the present to them."

Of course, time preference is a convenient assumption for lots of economic models. But when you just assume time preference, you induce an uncritical attitude towards it. I suspect there's a lot to be learn from dropping the assumption of time preference, substituting, let's say, an assumption that consumption today and consumption tomorrow are valued equally, and then trying to explain the phenomenon of time preference with reference to the life-cycle, uncertainty, general economic growth, and personal accumulation of human capital. This could lead to studies of why the rate of time preference seems to differ across countries, and what policies or socio-cultural variables influence the rate of time preference, and whether a lower rate of time preference is socially beneficial.

In short, I think you're dead on, and I hope your insights conquer and transform the discipline.

Lawrance George Lux writes:

Bryan,
You are right, but your argument lacks some impact. The Time necessity remains investment, when you lack the capacity to Consume all of what you make. Time Preference does not explain the positive Interest rates, but the fact of alternative Investment options does. lgl

El Presidente writes:

Greg Ransom: "...economists have punted on all of this..."

That's the truth. The marginal utility of an item is distorted by the scarcity of time. Sooner or later you will die (nothing personal). The myth of onward and upward is meaningless to someone who is seized with their own mortality. Pardon the jingoism but fear and consumption is the order of the day. Hayek pointed out that monopolies (or simply anti-competitive producers) are the catalyst for unrest and socialism in Road to Serfdom . When a person realizes they are woefully outmatched by the provider of their sustenance they grow uneasy. The time preference is to consume everything one can before death. Remove immediacy from need and there is no time preference (corporations are the only economic participants that have this advantage because they are, in theory, immortal). Then marginal utility plays a role when one realizes that they cannot consume everything. So we scheme about how we can consume in a way that provides maximal pleasure and thus there are trade-offs. You cannot deal with the time preference factor that generates willingness to pay interest and the marginal utility factor that causes us to decide which items are most worth that interest as mutually exclusive. Synthesis is lame, right?

Consider this. An elderly person decides that a life of frugality as a good "worker bee" should finally be rewarded. They purchase a home at an outrageous interest rate they cannot afford. They could care less. They are about to die. They never intend to repay the interest. So, instead of making their payment, they go to Vegas and buy a sports car. Their acceptance of their own mortality has freed them from both marginal utility and time preference considerations to a significant degree.

Phil Birnbaum writes:

There are some goods that do not appreciably depreciate and decay – say, land, or a nice set of silverware.

Now, suppose I own some land. Clearly, I want to use the land, because otherwise, I would sell it and buy something else. Therefore, it’s going to take a payment to me – rent – to for me to let you use my land for a year.

How much rent? Depends, but suppose I expect to live 40 years. Even with no time preference, you’d expect me to want at least 1/40 of the cost of the land to give it up for a year.

Now, suppose interest rates were zero. I’d just borrow money at 0%, buy the land, lend it out at 1/40 (or 2.5%), and live off the income. The demand for money to do this would push interest rates up to at least 2.5%. This would happen even if there was generally a *negative* time preference for other goods.

It seems to me that it’s not only the preference for goods now instead of later, but the preference to have those same goods *both* now AND later that is the basis for a positive interest rate.

Phil

Tom writes:

But before we try to explain why interest rates are always positive, we should make sure that they are always positive. In barter markets, interest rates are frequently negative. Suppose we knew the price of food would double next year. Then a pound of food now trades for half a pound of food one year from now. Translation: a negative 50% interest rate!

If this seems crazy to you, suppose food were the only commodity, and you expect a famine next year. Wouldn't you happily trade 2 pounds of current food in exchange for a promissory note good for 1 pound of food next year?

The first example is ridiculous. Barter markets involve complex trades. They're not just about food-for-food. They're about food-for-clothing, etc. Moreover, how do buyers "know" that the price of food will double next year? And if buyers do somehow "know" that it will double, they'll bid up the price of food this year, in order to store it for use next year (not all food being perishable). Result (ignoring time preference): the price of food (in terms of other goods) will double this year, ceteris paribus. Interest rate: zero.

Similarly, if a famine is expected next year, the price of food will rise this year. And so on, as in the preceding paragraph.

Imagine you are going to inherit $1,000,000 next year. According to the law of diminishing marginal utility, you would want to increase your consumption now when the marginal utility is high, and pay for it by cutting back your consumption in the future when the marginal utility is low. No time preference story need apply.

Why should my marginal utility be higher this year than next year (unless I have a time preference for consumption this year)? This year is this year and next year is next year. I want more consumption this year -- and next year -- because additional consumption of something always has positive marginal utility. I'm willing to subdue my taste for more consumption this year only if I can be assured of even more consumption next year. That's called time preference, and that's why I require a positive rate of interest as the price of forgoing some consumption this year in favor of more consumption next year.

Try again.

Bill Woolsey writes:

The "Myth of Time Preference" is a poor choice of headline.

There was a paper making a similar point to Caplan's many years ago. Perhaps someone else can remember the author, title, ect.

That paper pointed out that "time preference" has a different meaning in neo-classical and in misian ecnomics.

In neo-classical economics, it means that there is a particular bias towards present or future consumption. A simple way to look at it is that if one had constant income and no interest, then one would consume one's income each period if one had zero or no time preference. Positive time preference (or just "time preference") means that in such a situation one would consume more now and less in the future. Consumption over time would be allocated so that it gradually decreases. One's preferences are such that future consumption is consistently discounted. Negative time preference would be the opposite. Given constant income and no interest, consumption would gradually increase.

Negative time preference wouldn't mean that people never consume, however. That is because of diminishing marginal utility. You just have gradually increasing consumption. The discounted (or premiumed) marginal utility of consumption is equated. The discount or premium shows this consistent bias in one's preferences.

Figuring out what time preference means in Austrian economics is more difficult. To the degree it is a category of human action as mentioned by sale (quoting Mises,) then it doesn't
seem to have much to do with interest. To say that by the definition of action people prefer to get satisfaction sooner than later just means that by definition, I would like to get that negative interest rate contract signed quick to be sure I have get those consumer goods in the future when I really want them. The notion that this sort of time preference implies positive interest is hard to credit.

The article I mentioned above claimed that time preference in Austrian economics really means that people care when they consume. If no one cared when they consumed, there would be no phenomenon of interest.

Of course, that doesn't require that the interest rate be positive at all times.

I think such a claim isn't controversial, but that most economists wouldn't consider this very important. Then again, most economists use the term "time preference" in a way that is aimed and understanding what the level of the interest rate would be. People's preferences regarding future and present consumption would seem to be a factor. That factor without which the phenomenon of interest couldn't exist isn't of much interest to most economists.

And, of course, the notion that the interest rate must always be positive is most likely just wrong. It just requires that at some point, more round about methods of production become less productive. Or, in neo-classical terms, the marginal value product of capital become negative.

In my opinion, this is possible. And it is even likely to happen from time to time for short periods.

One final note--Caplan's argument about money interest suggests that the interest rate can be no more negative than the storage cost of money. And what happens if there is a pure credit money? Who will issue money at zero interest if earning assets all provide negative returns?

jaimito writes:

In times of war and uncertainty, people pays for somebody storing value for them. Id est, negative interest rate.

Positive interest rate implies a mood of optimism that believes things will be better in the future. That productive enterprises will prosper, that no catastrophy is in view, peace and stability is here for ever.

Pessimists, as El Presidente pointed out, know that in the end optimists will be shown wrong and therfore will pay for security. It may be a question of temperament.

Joshua Allen writes:

The economist arguments on both sides seem to rely on assumptions of scarcity which are simply not credible:

A) People are motivated to consume their whole pile (or as much as possible) before dying. This is clearly false. The fact that the earth supports 5 billion more people today than 2000 years ago proves that, on average, we leave behind more than we started with.
B) People have time preference for finite goods like food on a desert island. Again, at least in the west, people are swimming in excess. Nobody makes decisions based on dollars per calorie. The idea is an irrelevant anachronism. The bulk of consumption is for goods which are optional, preferences for which are arbitrary (and can change in a split second), and often created from thin air (e.g. entertainment).

When the only scarce resource is time, that would seem to change the economists equation of self interest. One can gorge only so much.

Barkley Rosser writes:

I agree with those who say the food barter
example is all wet. Another problem is that
the law of diminishing marginal utility does
not hold over time. People learn to like things
as they consume them more over time. Dim MU is
not a good candidate here at all.
Paul Samuelson argued that positive real
subjective time preference arises from the fact
that we all perceive time to "pass more quickly"
as we get older. Makes sense to me.

Noumenon writes:

You have an extremely impressive set of commenters at this site.

Barkley Rosser writes:

Another reason for positive subjective time
preference is death. I am alive now. The further into the future, the more likely I'll be
dead. Hence, I am less willing to pay today for
receiving something identical further in the future, unless I have a 100% Barro-type bequest
motive, which most of us do not possess.

Comments for this entry have been closed
Return to top