Positive Time Preference as Basis for Discounting

A consumer is said to have positive time preferencee if he is unwilling to exchange (or postpone) an extra quantity of consumption now for an extra quantity later unless the amount of consumption later were larger. Negative time preference entails the willingness to sacrifice a unit of consumption now for less than a unit later. Zero time preference can be defined in a similar way.

If positive time preference turns out to be a systematic human tendency (as, say, risk aversion), it means that people naturally discount future consumption, and this should be sufficient justification for discounting future benefits and costs of investment.

It should be noted that the assumption of positive time preference has been used, since Bohm-Bawerk, as one explanation of, and justification for, the existence of a positive rate of interest. We are not directly concerned with this aspect of time preference. Note, in passing, that: firstly, the existence of a positive rate of interest in a society in no way implies the validity of the assumption of positive time preference (Patinkin, p. 477, Samuelson, p. 613 n). Positive net productivity of investment is sufficient to open the possibility of positive interest rate even if time preference is negative or zero. Secondly, even if we grant that people have such a preference, it does not follow that a positive interest rate should be permitted in society. Islam and other revealed religions prohibited such things as usury, fornication and liquor for being unjust, immoral, or harmful, not for being undesired. I would, therefore, propose to view the assumption of positive time preference as a “positive proposition” to be accepted or rejected on empirical rather that on moral grounds.

a. The Case for Positive Time Preference

One trend of thought among Western economists, made prominent by, and since, Bohm-Bawerk, assumes positive time preference to be the systematic or normal pattern of behaviour. Negative (and zero) time preference is viewed as an aberration or at least a rare exception. Some economists (e.g. Alchian and Allen, pp. 199 ff) hav.e even elevated positive time preference to the status of a principle of rationality.

The basic justification for assuming positive time preference can be traced back to Bohm-Bawerk who thought that “If the marginal utility of future goods is lower because of their increased provision, present goods must be preferred” (Ekelund and Hebert, p. 300). This notion is still a popular justification among economists who assume positive time preference (e.g., authors of UNIDO Guidelines, pp. 154 and 164-65; and Harrod, p. 74). I paraphrase Harrod (slightly out of context to facilitate the discussion).

If income and consumption per head in an economy are growing, the representative person usually expects the marginal utility of consumption at some future point of time to be less than it is at present. Accordingly, if he is asked to sacrifice present consumption in return for extra consumption in the future, he would require future consumption to be larger, i.e. he would discount it. As for those individuals who expect their income to fall through time, it would be advantageous for them to save even without the prospect of larger future consumption. But on a weighted average, such individuals are a minority in a progressive economy.

b. The Case Against

Quite a few prominent economists have expressed skepticism about the very existence, or at least the positivity, of time preference (e.g., Graaf, p. 40; Baumol, pp. 411-13)1. Others have explicitly stated that a rational consumer may have either positive or negative time preference (e.g., Henderson and Quandt, p. 304; Robinson, p. 54). Still others, while not disclaiming positive time preference among individual consumers have deemed it improper for collective choices ( Cf. Malen- vaud, p. 234 and Naqvi, 115-17).

It is fair to say that the assumption of positive time preference is far from being universally accepted among economists. In the remain der of this subsection we propose to show that negative (and zero) time preference is a normal and quite a rational pattern of behaviour to be expected under a wide variety of common situations such as the following:

  1. Some counter examples: To assume time preference to be always positive would lead to the bizarre conclusion that a man who cannot invest most of his salary (to guarantee higher consumption later) should consume it all on the first day of the month, nay, the first hour of that day! If we reflect on this extreme example it should provide many insights.
  2. Another instance of widespread negative time preference in capitalist countries has been uncovered by high rates of inflation. Many consumers kept on saving in interest-bearing accounts even when the real expected rate of interest was clearly negative for certain years.
  3. Changes in income and in needs: It is commonly observed that a consumer tries to at least maintain the past level of consumption to which he is accustomed. His present sense of satisfaction or level of utility, furthermore, is a function not only of his present level of consumption but also of his expected future consumption compared with expected future needs. These two hypotheses taken together would lead us to expect that the following groups of consumers have negative rates of time preference:
    1. Those expecting a future decrease in income because of approaching retirement, or because current income is temporarily high, etc.
    2. Those expecting a future increase in their needs (sending a son to college, or getting a daughter married, etc.), with income constant.
    3. More generally, all those who expect fluctuations in income to be out of step with anticipated needs may well have negative rates of time preference among some time periods.

One glaring weakness of the assumption of positive time preference is that it disregards any changes in needs (or desired level of consumption), i.e. it implicitly assumes needs to be constant. On the other hand, it disregards the possibility of constancy or decline in expected income, and assumes it to be always rising.

Implications of the relative income hypothesis: If steadily increasing income is consistent with positive time preference, while declining income gives rise to negative time preference, then an expectation of constant income should lead to zero time preference.

Consider now Duesenberry’s relative income hypothesis which states that “the propensity to save of an individual . .. (is) ... a rising function of his percentile position in the income distribution”, i.e. it is a function not of his absolute, but of his relative income (Duesenberry, p. 45). A moment’s reflection should convince us that when the absolute incomes of consumers are rising in step, so that expected relative positions are maintained, this should lead to zero time preference. This challenges the positivity hypothesis even in its limited home ground of the rising income case.

  1. Voluntary intergenerational transfers: It is common observation that the present level of satisfaction or utility a consumer experiences is also a function of his expectation about the future level of consumption of his family. He frequently attaches greater significance to their level of consumption after his death than to his own consumption now. Prophet Muhammad (peace be upon him) is reported in authentic tradition to have said: “It is better to leave your heirs well off than to leave them poor asking people for help” (Related in Bukhari and Muslim). It should cause no surprise if some consumers derive more utility from transferring to a beloved heir an extra future pound than from consuming it themselves now.
  2. The misers: An interesting and hopefully rare group — the misers — should not be overlooked. They may be defined economically as making the accumulation of personal wealth an end in itself, not a means to increase their own or others’ future consumption. A miser may welcome reducing present consumption by one unit if it leads to increasing his wealth by even less than a unit. To him, accumulated wealth is a good entirely distinct from consumption goods, and there is no reason why the marginal rate of substitution, in his twisted utility function, may not be: several pounds of consumption for one pound-worth of wealth.
  3. The pervasive effect of uncertainty: The crucial effect on time preference of expectations about income and needs is evident in the above paragraphs. Since the future is never certain, we must consider the impact of uncertainty as such on time preference. The most likely effect of uncertainty about future income and needs, given risk aversion, is that the consumer would try to play it safe. This he can do by preparing himself, partially or fully, for the worst eventuality—vis., a sudden reduction in income or increase in needs. Such precautionary behaviour (or, if you like sophistication, such minimum regfet strategy) should reduce the rate of time preference, making it less positive or more negative. For suppose a consumer is certain that his income will increase. His rate of time preference should, therefore, be positive. Now introduce uncertainty by letting him expect his income to increase with, say, 70 per cent probability. This leaves 30 per cent probability for a decrease or no change in income, and should reduce the rate of time preference, or may even make it negative if the consumer is excessively averse to risk. We thus conclude that the introduction of uncertainty about the future or the increase in the level of uncertainty should, ceteris paribus, increase the proportion of people whose time preference is negative. It is in fact observed that people increase savings and reduce consumption during crises when uncertainty is keenly felt.

Conclusion on Time Preference

It seems fair to conclude that positive time preference is neither a principle of rationality nor an empirically established predominant tendency among consumers. It is simply one of three patterns of intertemporal choice (the other two being zero and negative time preference), each of which is rational and observable under its own conditions. The over-emphasis on positive time preference among some Western economists can, perhaps, be ideologically explained by its presumed suitability as a defence of the institution of interest against Marxist attacks.


Since time preference need not be predominantly positive, it can not provide a rationale for always discounting future costs and benefits in project evaluation.

Copy URL