However, initially utility was defined as a property of an object — its usefulness and capacity to procure such pleasure. For Bentham, utility was “that property of any object, whereby it tends to produce benefit, advantage, pleasure, good, or happiness”. Subsequently utility came to be understood as a feature of persons rather than objects: “the sum of pleasure and pain prevented” (Jevons, 1871). Georgescu-Roegen’s entry on “Utility” in the Macmillan International Encyclopedia of the Social Sciences, upon which I am drawing in this blog posting, provides more detail on the historical development of the concept of utility (Georgescu-Roegen, 1968).
Inherent in the concept of utility as expressed by Bentham and his inheritors was the characterization of pain and pleasure as being of the same essence, with pain viewed as a reduction of pleasure and vice-versa. This approach required a detailed calculus of pleasures and pains, leading to utility as a quantifiable measure of personal satisfaction. This endeavor was, however, eventually abandoned in light of the practical difficulties of measuring (or even indeed conceiving of) a meaningful interpersonally comparable unit of pleasure or pain (let alone determining its absolute level). The marginalist revolution, led by Jevons and Marshall, placed the emphasis on intra-personal comparisons of the marginal utility derived from distinct sources rather than on total utility experienced, in response to this problem. Later, due to the assaults on the utilitarian framework by antagonists such as Pareto and Robbins, who were motivated variously by the desire to resist the redistributivist implications of utilitarianism, and by the philosophical demands of logical positivism to separate science and ethics, the concept of utility was hollowed out further.
In the ‘modern’ microeconomics which has followed upon their criticisms, a utility function is simply a mathematical representation of an individual’s preferences (taken to have regularity properties such as completeness, transitivity and continuity) as potentially inferred from observed choices (applying the framework of revealed preference). Although this characterization avoids directly invoking a concept of subjective satisfaction, it nevertheless insists on the ability to relate pains and pleasures of various types, and thus permits a subjective satisfaction interpretation of a preference ordering. This leads Georgescu-Roegen (1968: 249) to conclude that “modern economic theory goes along with Bentham, taking as its first article the idea that pleasure is the positive, and pain the negative.” In short, pain, insofar as it appears in the framework at all, is characterized as the lessening of pleasure (as expressed by the realization of outcomes which are lower down on a preference ordering). We could say, loosely, in the words of Emerson: “For every minute you are angry you lose sixty seconds of happiness.” The modern microeconomists, as much as Bentham, happily commensurate, under the rubrics of preference and choice. Commensurate, commensurate! That is Moses and the prophets! [To paraphrase Marx’s cry in Das Kapital].
This, of course, is not a realistic characterization of the psychological experience underlying the choice process. Research in psychology does cast doubt on this oversimplification (equating pleasure with the absence of pain) or indeed contradicts the standard utility maximization framework altogether (e.g. by identifying violations of WARP). However, the issue here is not merely one of the realism of assumptions, important as that consideration is.[1] The utility framework is after all specifically designed to avoid engaging in the level of internal psychological scrutiny required to explain behavior (probing the complexities of the human mind is indeed a rather ambitious endeavor that has kept writers and psychologists busy for hundreds of years). It is intended to provide a representation of observed behavior, which can be employed as long as that behavior satisfies required restrictions.
A more interesting question is, therefore, whether the oversimplified view of how individuals derive personal satisfaction from the consumption of commodities leads to shortcomings in the utility framework such that reasonable choice patterns cannot be properly accounted for in some situations. Disregarding choice behavior that cannot adequately be modeled with a utility function might lead economists to develop too narrow a view of human decision-making processes, thereby precluding a better understanding of many choices.
In his now classic, “The Joyless Economy: The Psychology of Human Satisfaction”, drawing from research in psychology, Scitovsky draws a distinction between comfort, which is produced from the satisfaction of a need (and the elimination of discomfort), and desire, which accompanies the transition from discomfort to comfort and procures pleasure. The satisfaction of a need gives rise to both comfort (with permanent effects) and a temporary sense of pleasure. We all know from experience that the newness of a thing procures satisfactions that cannot be replicated once the newness is lost: “The advantage of a bad memory is that one enjoys several times the same good things for the first time.” (F. Nietzsche).
This distinction raises the issue of whether the representation of preferences via utility functions can adequately capture psychological distinctions which are relevant if we are to understand economic behavior, e.g. aspects of satisfaction such as ‘comfort’ and ‘pleasure’. The fleeting nature of pleasure might suggest, for instance, that utility lies in the rate of change of consumption and qualitative differences that separate new consumption from old rather than in the level of consumption as such. The standard description of individuals as deriving utility from the consumption of goods in a ‘state dependent’ way adequately characterizes the comfort associated with the satisfaction of needs, but altogether misses the transient - and ‘path dependent’ - nature of pleasure. A framework in which utility is defined over changes in consumption can potentially accommodate the Scitovskyian critique and account for the diminution of pleasure experienced from the consumption of a commodity (e.g. the baseline level of satisfaction can be recalibrated once a person gets used to things). There are examples of such conceptual frameworks in behavioral finance and economics. For instance, prospect theory introduces a situation-dependent perspective in which the value placed by individuals on financial outcomes is defined in relation to changes in wealth rather than the level of wealth (Kahneman and Tversky, 1979).
A framework seeking to comprehensively capture distinct aspects of satisfaction should also be dynamic in nature and attempt to give an account of how people adjust to new situations – i.e. how preferences evolve over time. There are some references to the time-factor in the first few chapters of “Microeconomic Theory” by Mas-Colell, Whinston and Green. We are told, for instance, that technically speaking commodities consumed at different points in time should be thought of as different commodities (Ch.2, p.18) but that for convenience many models engage in time-aggregation whereby a commodity might be defined as an amount consumed over some reference time period (a day, a month, a year…). We are also told in Chapter 1 that a set of observed choices is to be generated through repeated observations (which would imply some time dimension, and the possibility of various commodities being chosen over time). We are also introduced to WARP which, in application, assumes constancy of preferences. Finally, we learn that Walras’ law is to be understood broadly (p.23), with the consumer expending resources over a lifetime (hence the budget allowing for some deferral of consumption). These comments (taken together) seem to indicate that the framework is static rather than dynamic, with the Walrasian demand function representing the summary result of an intertemporal choice process. With consumers instantaneously choosing consumption to maximize their lifetime satisfaction, there is no need for a view of preference as potentially evolving through time.
Other research on happiness, satisfaction and well-being sheds new understanding on what makes experiences actually enjoyable and suggests that there are still deeper concerns about the utility framework. A psychoanalytic approach suggests a more complex picture of what constitutes a fulfilled life, which includes a place for both pleasant and unpleasant experiences, and indeed for unhappiness as well as happiness. Hence, the presence of the unpleasant does not necessarily mean that fulfillment over the life course is diminished (see for instance Nandy, 2012). Mihaly Csikszentmihalyi’s research on experiences that greatly enhance one’s quality of life provides an understanding of enjoyment based on the concept of “flow” — characterized by the presence of goal-directed challenges, discovery, high concentration, a sense of complete immersion in the task, and a potential to increase a sense of meaning and purpose. These ideas are reminiscent of Aristotle’s ideas concerning eudaimonia. The key insight here seems to be that happiness (and true enjoyment) does not merely consist in experiencing pleasure (which is evanescent and does not add complexity to the self) or satisfying needs, and certainly not derive from consumption as such. Rather, a person’s ability to acquire skills, perfect those skills over time, and establish a sense of purpose leads, in the long run, to significant differences in the quality of inner experience and expressed fulfillment. It is unclear that a mere snapshot of an individual’s preferences defined over a commodity space at a specific point in time can begin to capture the aspects of life relevant to fulfillment. At best, consumption choices made through time would seem to reveal or express underlying directions of self-development being shaped concurrently with consumption itself, as a person grows in complexity and life experiences accumulate. The narrow representation of preferences via a state-dependent utility function (and the framework of utility maximization) misses the greater picture (the personal trajectory underlying observed choice).
These types of considerations lead to the conclusion that a more textured understanding of the psychological underpinnings of choice is after all needed, going beyond a purely ordinal ranking. There are different shortcomings associated with an approach limited to ranking various alternatives on the basis of preferences revealed through observed choice. As we just discussed, the framework misses the long-term trajectory of personal transformation underlying choice. Further, as is extensively discussed by Amartya Sen, the standard consistency conditions (such as WARP or GARP) are overly narrow in that they ignore considerations external to the intrinsic properties of the elements of a choice set, such as the circumstances in which choice was made (for instance the identity of the person choosing, or the menu over which choice is being made) that are relevant to understanding a choice decision (Sen, 1997). The chooser dependence of preferences can be illustrated with the example of a person that would be content if someone else were to select a particular outcome (Sen’s example of sitting on a comfortable chair at a party, or picking the last mango from a fruit basket), and yet might be hesitant to make the choice himself (and instead go along with a less preferred alternative). There could be many reasons underlying this choice, such as concern for reputation (wanting to come across as a considerate person), a perceived moral imperative, the following of a conventional rule associated with ‘proper behavior’, etc. In such situations[2], the representation of preferences via a utility function has real inherent shortcomings for the representation of actual behavior. To the extent a ranking of alternatives does not adequately capture reasons for the choice (such as social norms), seemingly irrational behavior (such as WARP violations) might emerge precisely because there are good reasons to act in such a way.
Ad hoc adjustments to the framework can be made (e.g. for ‘preferences’ to be permitted to evolve over time, or preferences to be defined so as to be ‘chooser dependent’ or ‘menu dependent’). Yet this is achieved by extending the formalism at the cost of its already questionable contribution to insight becoming even more so; the more additional distinctions between different kinds of commodities are required to capture the diversity of experience, the emptier the framework becomes, with attendant loss in explanatory power and operational relevance (in our prior blog entry “Some Considerations on Rationality”, we draw an analogy with the addition of epicycles to the Ptolemaic astronomical system).
Ultimately, even when choices are consistent and can be described with an ordinal ranking, such ordering does not capture the substantive content of the relationship between a person and the actions he or she takes. For instance, choice could be the result of conflicting considerations. Sen’s distinction between the concepts of sympathy and commitment is on point (Sen, 1977). Sympathy describes situations where one’s own welfare is directly affected by the welfare of others. An action based on commitment, on the other hand, is not motivated by the concern for one’s own welfare, but rather is closely connected with one’s ethical views (Sen argues that the question of commitment is important in a number of economic contexts, including work motivation). Arguably sympathy can be folded into the utility function (with the utility of others affecting one’s own utility), but the concept of commitment is harder to reconcile with the pursuit of personal welfare and the maximization of one’s utility. Indeed, commitment may lead to some kind of personal sacrifice, for example choosing to take care of one’s ailing parent instead of pursuing one’s goal of moving to a different city and taking on a new exciting job opportunity. Much of this is missed in the tautological concept of a consumer taken to further his or her own interests no matter what he chooses as long as choices are observed to be ‘consistent’. Here, a preference ordering based on observed choices (taking care of one’s parent is preferred to the exciting job opportunity) says nothing of the potentially conflicting considerations ultimately leading to the choice structure. The preference ordering says little of real interest – it misses the true understanding of why the person has chosen to take care of the ailing parent. These types of considerations lead Sen to the conclusion that “preferences as rankings have to be replaced by a richer structure involving meta-rankings and related concepts.” (Sen, 1977: 344). By attempting to ‘put it all in the utility function’ (which one cannot in any case always do), one engages in a level of abstraction from the real motivations involved which does violence to both descriptive richness and explanatory understanding.
The trouble with utility is not only its lack of psychological realism, but also the rigidity with which it is defined over a commodity space – the lack of a process-centered view of what generates the satisfactions ultimately leading to choice. The representation of preferences by employing a utility function appears adequate only to describe the ordinal relation of preference, at the cost of capturing only very partially the textured variety of psychological phenomena giving rise to pleasure, happiness and indeed fulfillment[3].
References:
Csikszentmihalyi, M. (1990). Flow: The Psychology of Optimal Experience. Harper and Row, New York.
Georgescu-Roegen, N. (1968). “Utility”, International Encyclopedia of the Social Sciences Macmillan: New York
Jevons. W.S. (1871). The Theory of Political Economy. 5th ed. New York: Kelley.
Kahneman, D. and Tversky A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica, XLVII (1979), 263-291.
Nandy, A. (2012). The Idea of Happiness. Economic and Political Weekly, 47(2): 45-48.
Scitovsky, T. (1976). The Joyless Economy. Oxford University Press, 1976.
Sen, A. (1997). Maximization and the Act of Choice. Econometrica, 65(4):745-780.
Sen, A. (1977). Rational Fools: A Critique of the Behavioral Foundations of Economic Theory. Philosophy and Public Affairs. 6(4):317-344
[1] Our prior blog entry “Some Considerations on Rationality” explores this issue in greater detail.
[2] For a discussion of menu dependence, please refer to our other post “Some Considerations on Rationality”.
[3] We have said nothing about concepts such as need (or allied ideas such as basic requirements or elementary capabilities) which are of great worldly relevance and arguably can even less be captured by the ordinalist framework.
