Observations on the Pure Time Preference Theory of Interest

First, Tyler Cowen has a brief comment here:

“... Austrian economists, such as Mises, Rothbard, and Kirzner, have advocated a ‘pure time preference’ theory. Most advocates of the pure time preference theory, however, do not dispute that real world interest rates are determined by many of the features identified by the loanable funds theory (Kirzner 1993; see also Pellengahr 1986). The advocates of the pure time preference theory wish to isolate one component of the interest rate — the time preference component — as ‘essential’ to the nature of interest. In this regard the Misesian theory is consistent with the loanable funds approach, even if nonessentialists, such as myself, find the special emphasis on time preference to be tautologous or superfluous.” (Cowen 1997: 65, n. 11).

So Austrian supporters of the pure time preference theory of interest apparently do not dispute the neoclassical loanable funds theory as an alleged explanation of real world money interest rates. They just add an additional theory along with it.

The sense in which the expression “time preference” itself is used by Austrians is described by Robert Murphy. Murphy notes that there are at least “two distinct meanings of the term, yet many Austrians conflate them” (Murphy 2003: 65):

“In sense (i), time preference refers to the greater valuation, on the margin, of a present good over a future good, or of present ‘consumption’ over future ‘consumption.’ It is the “nub and kernel” of Böhm-Bawerk’s theory of interest. If an individual possesses time preference in sense (i), then he values a present apple at time t₁ more than he values a risk-free claim (given to him at t₁) for an apple that will be delivered in t₂. In a Fisher diagram, the slope of an indifference curve (with present and future “real income” on the two axes) at a particular point corresponds to the degree of time preference (or “impatience”) in sense (i) for the respective levels of income. In a neoclassical model, an agent possessing time preference in sense (i) has a marginal rate of substitution of present for future goods that is less than one; i.e. he would have to be promised a greater number of future units in order to sacrifice one present unit of the good. Time preference in sense (i) is an endogenous concept, since it can be affected by factors such as the relative provision of goods in different time periods. In particular, under certain (perhaps unusual) circumstances, time preference in sense (i) can be negative—for example, it is possible that two present units of a good exchange for one future unit of the same good.

In sense (ii), time preference refers to the discounting of future utility per se, according to its remoteness from the present, solely because it is in the future. Time preference in sense (ii) corresponds to Böhm-Bawerk’s “second cause” for the agio on present goods. If an individual possesses time preference in sense (ii), then at t₁ he values a claim for an apple to be delivered in t₂ less than he will value that apple when it is delivered. In a Fisher diagram, the closest thing to time preference in sense (ii) would be the slope(s) of the indifference curves along the 45-degree line, that is, when all ‘other things’ (in the diagram) are held equal. In a neoclassical model, an agent possessing time preference in sense (ii) will discount (often by a constant factor denoted by β) future utils in order to compute the present marginal utility of expected future consumption; time preference in sense (ii) thus corresponds to the marginal rate of substitution between present and future units of utility (not goods). Time preference in sense (ii) is often treated as an exogenous, ‘given’ feature of individuals’ preferences. In particular, it is usually considered to be completely independent of the supply of goods in each period, and is almost always treated as universally positive, regardless of other circumstances.” (Murphy 2003: 65–67).

So in these senses we must distinguish between:

(1) the valuing of a good now more than it would be valued at a point in the future, and

(2) the present subjective valuation of a good below that which would be experienced at the point in the future when it would be consumed.

In Mises’s Human Action, time preference is defined in abstract terms like the law of demand in neoclassical economics:

“Satisfaction of a want in the nearer future is, other things being equal, preferred to that in the farther distant future. Present goods are more valuable than future goods. Time preference is a categorial requisite of human action. No mode of action can be thought of in which satisfaction within a nearer period of the future is not — other things being equal — preferred to that in a later period. The very act of gratifying a desire implies that gratification at the present instant is preferred to that at a later instant. He who consumes a nonperishable good instead of postponing consumption for an indefinite later moment thereby reveals a higher valuation of present satisfaction as compared with later satisfaction. If he were not to prefer satisfaction in a nearer period of the future to that in a remoter period, he would never consume and so satisfy wants. He would always accumulate, he would never consume and enjoy. He would not consume today, but he would not consume tomorrow either, as the morrow would confront him with the same alternative.” (Mises 2008: 480–481).

The sign of extreme abstraction from reality is the ceteris paribus (“other things being equal”) qualification.

Here Mises is saying that positive time preference is a universal, a priori state of human action. Some critics doubt this.

Of all people, the Austrian economist Robert P. Murphy does. Murphy argues:

“... Mises ... claims, “The very act of gratifying a desire implies that gratification at the present instant is preferred to that at a later instant” ... This observation is true, but it overlooks the fact that there could have been a preceding interval of delayed gratification, throughout which the (in that interval) future satisfaction had been preferred.

In a similar vein, Mises states that, “He who consumes a nonperishable good instead of postponing consumption for an indefinite later moment thereby reveals a higher valuation of present satisfaction as compared with later satisfaction” (1966, p. 484). Again, this is perfectly true, but Mises conveniently neglects to mention that he who saves a nonperishable good instead of consuming thereby reveals a higher valuation of later satisfaction. Both possibilities are equally plausible; there are real world cases of people who consume nonperishable goods, as well as real world cases of people who save them. If the former individuals demonstrate the existence of [time preference], why don’t the latter demonstrate the existence of negative [time preference]? (Murphy 2003: 65–67).”

Let us look again at this argument of Mises:

“If he were not to prefer satisfaction in a nearer period of the future to that in a remoter period, he would never consume and so satisfy wants. He would always accumulate, he would never consume and enjoy. He would not consume today, but he would not consume tomorrow either, as the morrow would confront him with the same alternative.” (Mises 2008: 481).

Murphy’s response to this is worth quoting at length:

“This argument is valid only when the “other things” that must be held equal are construed quite broadly. Mises has in mind an omniscient, immortal agent who behaves mechanistically according to external stimuli. If such an agent decides to postpone consumption in t₁ and (by construction) all relevant factors are the same in t₂, then the agent (to be consistent) must postpone consumption again. This is the sense in which Mises believes that valuing a good implies, ceteris paribus, the desire to consume it sooner rather than later.

The first thing to note about this argument is that it is quite odd from an Austrian point of view. Austrians usually deplore the mainstream approach of modeling agents as automatons who maximize given utility functions, yet this is merely a formalization of the approach Mises takes in the above quotation. Elsewhere in his book Mises says that even intransitivity of ordinal preference rankings is not evidence of irrationality, since preferences may change between acts of choice ... . Yet in the above argument, Mises must hold preferences constant for eternity.

Second, strictly speaking Mises has only proven that an agent cannot prefer to postpone consumption. For if an agent were indifferent and every period flipped a coin to decide whether or not to consume, he would eventually consume (technically, ‘with a probability of one’). Another purist quibble is that the agent could conceivably base his consumption decisions on the time index itself. Since we are dealing with an immortal being, who can say that the agent wouldn’t prefer to consume everything on February 25, 2525? In this contrived example, today’s decision to postpone consumption would not prohibit eventual consumption, even holding everything else equal.

But the most serious drawback to Mises’ argument is that it only proves the apodictic necessity of time preference for such an immortal (and in essence, timeless) being. Mises has not shown that actual human beings must exhibit time preference in his sense. On the contrary, Mises’ argument doesn’t even rule out universally negative time preference. That is, suppose that every human being prefers, other things equal, to postpone all consumption as long as possible. Does this entail an immediate cessation of consumption? Not at all! People still need to eat, or else they will starve and miss out on their planned future consumption. Moreover, people on a sinking ship will consume all they have, since other things will not be equal if they delay their gratifications. In fact, once we seriously consider the conditions of human life as they really are, there is nothing to rule out the present world as one characterized by universally negative [time preference] in the Misesian sense. Since other things are never equal, we do not need to fear the absurdity of all consumption postponed forever. The above analysis of Mises’ arguments demonstrates that, at best, he has proven the universal validity of “time preference” only when the term is couched in an entirely vacuous sense, in which, say, present hot dogs are preferred to future hot dogs because one can only eat a hot dog in the present. Needless to say, [time preference] in this sense does not bear any relation to interest theory: It does not explain the discount on goods (such as hot dogs) available in the future, since even a decision to postpone current consumption would be viewed in this sense as “positive time preference” (at the eventual moment of consumption).” (Murphy 2003: 93–95).

So, in essence, the universality of Mises’ “positive time preference” can be proven only for “an immortal (and in essence, timeless) being”!

Those skeptical of Austrian economics from the outset will be tempted to regard the concept of universal “positive time preference” defined as “a categorial requisite of human action” as just another piece of irrelevant Austrian metaphysics.

If reformulated as an empirical statement, that (for example) people tend generally to want any particular good x now rather than later, it could be tested empirically. Whether it is true or not, its truth would be a posteriori, and if true would be just another trivially true fact about human want satisfaction, with many exceptions.


BIBLIOGRAPHY
Cowen, Tyler. 1997. Risk and Business Cycles: New and Old Austrian Perspectives. Routledge, London.

Mises, L. 2008. Human Action: A Treatise on Economics. The Scholar’s Edition. Ludwig von Mises Institute, Auburn, Ala.

Murphy, Robert P. 2003. Unanticipated Intertemporal Change in Theories of Interest. PhD dissert., Department of Economics, New York University.