It seemed like a simple enough question. What do economists mean by ‘utility’? But after scouring the literature I’m more confused than when I began.
The Penguin Dictionary of Economics defines it as: "The pleasure or satisfaction derived by an individual from being in a particular situation or from consuming goods or services." Apparently it’s "the ultimate goal of all economic activity". The dictionary then goes on to quote Jeremy Bentham. And it’s here I begin to get confused.
It turns out that Bentham used the term in an entirely different way. When he wanted to refer to pleasure or satisfaction, he used words like ‘pleasure’ and ‘satisfaction’. For him, utility was something else:
By utility is meant that property in any object, whereby it tends to produce benefit, advantage, pleasure, good, or happiness, (all this in the present case comes to the same thing) or (what comes again to the same thing) to prevent the happening of mischief, pain, evil, or unhappiness to the party whose interest is considered: if that party be the community in general, then the happiness of the community: if a particular individual, then the happiness of that individual.
Clearly, utility is a property of the goods and services people consume, not a mental state. Utility simply means usefulness. As Frédéric Bastiat writes:
Let others lose themselves in definitions. For my part, I understand by utility what everyone understands by this word, whose etymology indicates its meaning very exactly. All that is serviceable, whether it be by nature, by labor, or by both, is useful.
This use of the term isn’t confined to the 19th century. Frank Knight writes:
The power of things to satisfy conscious wants, or quality of being wanted, is utility in the economic sense, which is equivalent to "power over conduct."
Irving Fisher took a similar line. He suggested the term ‘wantability’ as a replacement for utility. This was because he thought the term utility was misleading. The naive public, he wrote, "find it hard to call an overcoat no more truly useful than a necklace, or a grindstone than a roulette wheel."
Fisher also objected to some of the Benthamite connotations of ‘utility’. Some economists wanted to reject Bentham’s assumption that people always pursued happiness, but keep the concept of utility. They preferred to be agnostic about why people wanted what they did. For an object to have utility, it didn’t need to produce pleasure, it just had to be wanted.
Regardless of what stand you take on this issue, there are two complications involved in ascribing utility to objects. First, as William Jevons pointed out in the late 19th century:
… utility, though a quality of things, is no inherent quality. It is better described as a circumstance of things arising out of their relation to man’s requirements. As Senior most accurately says, "Utility denotes no intrinsic quality in the things which we call useful; it merely expresses their relations to the pains and pleasures of mankind." We can never, therefore, say absolutely that some objects have utility and others have not.
The utility of shoe depends on whether it fits. But not only that, it typically has no utility unless it is offered as part of matching pair. For many goods, utility depends on the availability of complimentary goods. A car is not much use without petrol, a computer is not much use without matching software and so on. And some goods are only useful to a person with skills. For example, a book is not much use to a person who can’t read.
So i t turns out that utility is a property of things when they are in certain complex relationships with people and other things. So it doesn’t make much sense to ask whether something has utility, without either spelling out a lot of background information or making assumptions.
The other complication is that people make mistakes about how useful things are. Recognising this, Ludwig von Mises distinguishes between two kinds of utility — subjective and objective:
Utility means in this context simply: causal relevance for the removal of felt uneasiness. Acting man believes that the services a thing can render are apt to improve his own well-being, and calls this the utility of the thing concerned. For praxeology the term utility is tantamount to importance attached to a thing on account of the belief that it can remove uneasiness. The praxeological notion of utility (subjective use-value in the terminology of the earlier Austrian economists) must be sharply distinguished from the technological notion of utility (objective use-value in the terminology of the same economists). Use-value in the objective sense is the relation between a thing and the effect it has the capacity to bring about. It is to objective use-value that people refer in employing such terms as the "heating value" or "heating power" of coal. Subjective use-value is not always based on true objective use-value. There are things to which subjective use-value is attached because people erroneously believe that they have the power to bring about a desired effect. On the other hand there are things able to produce a desired effect to which no use-value is attached because people are ignorant of this fact.
All this seems clear enough. But somehow the term ‘utility’ has shifted in meaning. Instead of referring to the ability to produce pleasure or satisfaction, it became a synonym for pleasure or satisfaction. This is how Daniel Kahneman and his co-authors use the term ‘experienced utility’ in their 1997 paper ‘Back to Bentham? Explorations of Experienced Utility‘.
Kahneman also uses the term ‘decision utility’. Apparently ‘decision utility’ is revealed by choices or revealed preferences. Decision utility may (or may not) refer to the pleasure of satisfaction an individual expects to receive when they consume a good.
Can anyone offer me a definition that isn’t vacuous or circular?
As I recall, there have been three major definitions of utility (U). Summarily:
(1) Bentham: the property of an object that brings about the psychological state of happiness in a (human) user of that object.
(2) Jevons, Marshall and Edgeworth: the psychological state of happiness in a (human) user of an object, which is due to the properties of that object.
(3) Hicks & Allen: the value of a function which represents the well-behaved preferences of an agent over all options available to that agent.
Later, we have iterations of (3) that involve tinkering with the state of knowledge of the agent – viz. preferences over all known probable options, or all believed probable options, and questions about whether to treat probability objectively or subjectively. Arguably, with Samuelson there is also a revision of the notion of preference itself – from an internal psychological state to sheer observed behaviour.
In (1), U is a property of an object which causes a relatively specific psychological phenomenon in the subject. In (2), U is shifted from the object to the subject. By our modern version, (3), although U represents a basic psychological phenomenon, preferences, it is now drained of content in that the preferences can be held for any reason at all (happiness, unhappiness, pleasure, pain, angst, boredom, any bizarre combination of these, or whatever). Other notable changes by the time of (3): the causal relations implicit in (1) and (2) have disappeared; strictly speaking, the subject no longer need be human (coal ticks, amoeba and perhaps computers can have utility); and moral connotations are (allegedly) expunged.
Although by the time we get to (3) there is almost no meat left on the bones of (1), and so we may like to call (3) vacuous – or more politely, thin – it is not clear that it is circular per se. The charges of vacuity and circularity are stronger when (3) is pressed into the service of explaining behaviour: Why did A choose ice cream over a vomit at t0? Because at t0, the U of ice cream was greater that the U of vomit for A, which is just to say that at t0 A strongly preferred ice cream to vomit. Why then did A choose vomit over ice cream at t1? Because at t1, the U of vomit was greater that the U of ice cream for A, which is just to say that at t1 A strongly preferred vomit to ice cream.
Hi Don
Why couldn’t we consider ‘utility’ to be the acting agent’s belief in the ability of thing A to bring about thing B generally?
Utility as the actor’s belief in the power or “utility” of thing A to produce or bring about thing B, regardless of what B is (happiness, physical effects, etc.)
A thing A has “utility” to me, to the extent I believe it is or can be “instrumental” in bringing about state B.
Just a suggestion.
Likewise, I find the concept that is almost universal in the current economics in number three. Utility as a product of given options and preferences. The problem (as I see it) is less the vagueness of the term “utility” but that fact we accept preferences as exogenous and a priori unknowable. I can see the use in this in limited circumstances.
Which also brings me to a point of confusion about one of the most common criticisms of neo classical micro, that it “can’t explain” altruism and a mass of similar phenomenon. The problem as I see it is it can explain altruism and self damaging greediness and hitting yourself over the head with a giant sausage. Any behaiviour can fit into the model under different preferences. Whilst we rely on revealed preferences, any behaiviour can be explained ex post facto.
Which is not useful at all.
On topic though, I think it would be sufficient to rely on a function definition of utility for many purposes (if not alot of verbal reasoning) if we could make this a function of more appropriate preference theory. Which would be very very difficult indeed, especially if the preferences prove (as I suspect) to be endogenous.
Edward – I’m not sure that I understand what ‘utility’ refers to in the third definition.
If, in (1) utility refers to the property of an object (a good or service); and (2) it refers to the mental state of the consumer; then would it be right to say that in (3) utility refers to something entirely abstract — a value in an equation, rather than to any entity in the world?
If so then there’s nothing more to say about utility than what’s there in the equations. It’s a bit like a maths student asking “But what is the number 1 REALLY? Where is this thing that ‘one’ and ‘1’ refer to? What properties does it have?”
I’m told that there was a shift to operationalism in the 30s (psychology also shifted but the behaviourist program became exhausted after a few decades). I assume this means that it’s no longer reasonable to expect that any of the variables in economic theory correspond to things in the world. All we can expect is that there is some agreed on way to connect the abstractions back to observations.
I interpret Kahneman’s work as a return to scientific realism and a move away from operationalism or instrumentalism.
Am I on the right track here? Or am I confused?
Richard
I love your examples. I’m imagining news reports of people across the world hitting themselves over the head with giant sausages. We cross to an economic expert …
Reporter: Dr X, why are people suddenly hitting themselves over the head with giant sausages? What’s causing this?
Expert: Well Bob, OUCH!, it seems as if OOF!, we’ve all developed a preference for hitting ourselves over the head with THUMP! giant sausages.
Reporter: Thanks Dr URGH! X, I’m glad we’ve cleared that WHACK! up. And now to our next story, why people’s preferences for real estate in Los Vegas have shifted so suddenly …
Don,
your questions are very reminiscent of those posed by Sen in his book on poverty (the Tanner lectures) where he also draws distinctions between things people want, things that give pleasure, inherent properties of things, etc. Economists are usually quite sloppy about all of this and lump them all together: economists usually presume what people want is what gives them pleasure which is then also taken to be an innate property of certain goods.
Amongst micro-theorists there is much less ambiguity though: utility is merely a construct, derivative of preference maps and only informative as it relates to other utilities. it is the mere representation of preferences (whatever those are: taking a psychological view, utility makes far more sense than preference as a first datum). Utilities merely describe a set of preference relations, but not necessarily vice versa (you cant describe lexicographic preferences in utilities).
When talking about a utility function, i.e. a mapping from input to utility numbers, then we usually think of ‘the thing whose maximisation leads to choice behaviour’, which needs no normative interpretation. Utility in that sense is mostly a tool, i.e. a handy device to quickly incorporate the notion of tradeoffs into a model.
Then there are those who do wish to implicitly or explicitly make normative statements (like the happiness researchers) who thus take utility to be an attribute of an individual’s state of mind (not an attribute of goods!).
U is now effectively just some (real) number or other, but it represents (or is supposed to represent) a preference for a bundle of entities (things and/or activities, or better(?) ‘services’ of things and/or activities) – and preferences are, I think we can fairly safely say, are psychological in their nature. So in Hicks & Allen’s (1934) version, which is the one used in standard formal theory (despite the Marshallian idea being more often referred to in elementary textbooks), there is still ‘a ghost in the utility function’, so to speak.
Yes, Bridgman’s (1927) operationalism was certainly on the make by 1934, but I don’t think this was informing Hicks & Allen. I think they were basically being motivated by the same slightly embarrassing problem that concerned Edgeworth: how the hell are we going to theorise choices in the absence of any means of actually measuring this notionally cardinal U? That said, operationalism did seem to significantly influence Samuelson’s (1938) thoughts on utility, thus giving us revealed preference theory, where, roughly, he sought to eliminate the last vestiges of ‘the inner life’ remaining in Hicks & Allen’s work. Samuelsons theory is still in the textbooks (for pedagogical purposes as far as I can tell), but the attempt get U to correspond to nothing but behaviour as failed, as demonstrated in detail by Wongs (1978) vivisection of said theory. So we are left with Hicks & Allens definition as at least our point of departure for standard theorising about choice. Slightly weirdly, in recent years utility as happiness – a la version (2) – seems to have made something of a comeback (e.g. Frey, 2008).
But we should not speak too ill of the dead. Samuelson died yesterday (13th Dec.).
* Bridgman, P.W. 1927. The Logic of Modern Physics. New York, Macmillan.
* Frey, B. 2008. Happiness. A revolution in economics. MIT Press, Cambridge, Mass.
* Hicks, J.R. & Allen, R.G.D. 1934. “A Reconsideration of the Theory of Value”, Economica, 1: 52-76, 196-219.
* Samuelson, P. 1938. “A Note on the Pure Theory of Consumers’ Behaviour”, Economica, 5: 61-71.
* Wong, S. 1978. The Foundations of Paul Samueison’s Revealed Preference Theory. Boston, Routledge & Kegan Paul.
Exactly. So Kahnemann is apparently straying from convention. But as long he defines his terms, I wouldn’t lose any sleep over it.
But what’s the ending of your post about? Are you implying that the conventional definition is circular and vacuous? If so, nothing in your post supports such a conclusion. Perhaps that’s the subject of your next post. If so, I’m looking forward to it.
Unless, of course, you’re just going to rehearse the tedious canard that if utility is defined as the thing that makes us want to acquire goods, and economists tell us we acquire goods because they afford us utility, this makes the defnition circular.
Just in case this is what you have in mind, let me head you off at the pass.
The concept of utility was initially helpful in clearing up confusions about value. In the end most theorists agreed that it made sense to distinguish between the exchange value of a commodity — the price you actually pay — from the use value, utility, or benefit, that the ultimate buyer obtains from it. Our desires can only be meaningfully quantified in terms of choices between different bundles of goods: how much benefit we get from Bundle A is measured by Bundle B that we would sacrifice to obtain it. Indeed we could get by pretty well without referring to anything called utility. But that doesn’t make the concept vacuous or circular — you try explaining the price of a fish without making the distinction, implicity or explicitly, between exchange value and use value.
The early neoclassicals assumed that if there was a systemtic relationship between price and quantity, this implied that utility must be a cardinally measurable quantity.This approach survives in introductory economics, and might explain why many people think economics rests on a concept of measurable utility. But Hicks showed this wasn’t the case — that preferences held entirely in the form of complete rankings between alternative bundles were enough to generate a systematic demand function.
Where it does becomes necessary to measure utility, however, is in welfare economics, which requires making interpersonal comparisons. In this context we sometimes use utility to denote the quantifiable benefit or ‘welfare gain’ arising from something or some circumstance, as measured by how much people would pay for it; in this sense it’s a more general measure of benefit than consumer surplus. The whole approach is value-ridden, and even if it was value-free, many types welfare gains would still be tricky to measure. But this doesn’t in any way imply that the concept is circular or vacuous. The alternatives, as far as welfare economics is concerned, are the Pareto criterion (which almost everyone agrees is inadequate) and nihilism.
James – I’m having trouble understanding your heading-me-off-at-the-pass argument. You might be right, but I’m not sure how the argument works.
In para 6 it looks like you’re saying that ‘use value’, ‘utility’, and ‘benefit’ can be treated a synonyms.
Then you say “we could get by pretty well without referring to anything called utility.” So I assume you’re saying that we could get by pretty well without referring to use value or benefit (since they mean the same thing).
Then you say “you try explaining the price of a fish without making the distinction, implicitly or explicitly, between exchange value and use value.” Which sounds like we can’t get by very well at all.
There’s no contradiction, Don. We need to be able to distinguish between the individual’s subjective valuation of something and its market value, but in most contexts we can do that without invoking an unobservable, unmeasurable entity called utility if we don’t want to. If there’s a problem with the concept it’s that it’s unnecessary, not that it’s circular or vacuous.
James – The distinction between use value and exchange value is clear enough.
When you say “we could get by pretty well without referring to anything called utility” I thought you might be referring to Samuelson’s theory of revealed preferences.
I noticed that Edward wrote: “the attempt get U to correspond to nothing but behaviour as failed, as demonstrated in detail by Wongs (1978) vivisection of said theory.”
Is the idea that we can dispense with utility controversial among economists?
I think most economists would agree that utility is a handy but dispensible way of talking about prefernces, reservation prices, welfare changes and so on.
Samuelson v. Wong, and the limits of revealed preference, is a pretty esoteric topic that your average applied economist wouldn’t have an opinion on at all. Edward might have a different view.
James – I’m not going to pretend I understand the Samuelson v. Wong debate.
Edward – I really appreciated your last comment. I especially appreciated the reading list.
You should derive some utility from the New Palgrave entry if you haven’t consuted it yet. Collison Black is an expert on Jevons.
Don:
If ‘utility’ is taken to mean something more substantive than ‘a numerical representation of a preferred bundle’ (where a set of preferences is well behaved), then I don’t think there would be much (any?) controversy at all among true-believers of the Paretian-inspired wing of the neoclassical school. [I say ‘Paretian’ because I suspect one may be able to draw a line from Pareto through Edgeworth to Hicks.] For some others it may be regarded as slightly controversial, if only because it is difficult to escape the idea that utility has at least something to do with a recognisable sensation in the mind, whether it be called pleasure or happiness or satisfaction. [I suspect this idea is hard to shake because it is, after all, the rhetorical mother’s milk on which first year economics students are raised vis-
Quotable quotes referring to ‘utility’ No. 241:
From one of the masters of derision, Herr Dr Marx:
[…] so we need not retread them here (though I think for all their flaws, they’re no worse than utility). The writers make interesting comparisons within the confines of this measure. Based on regression […]