School of Economics and Centre for Competition Policy University of East Anglia Norwich NR4 7TJ, United Kingdom Consumers' surplus for consumers who lack coherent preferences Robert Sugden Paper prepared for conference on Behavioural Industrial Organisation and Consumer Protection, UCL, 17–18 October 2014 1 This paper is work in progress, describing a problem rather than a solution. It’s a mix of economic theory and history of economic thought. Theory: it’s part of my attempt to reconcile behavioural and normative economics. The general problem is that conventional welfare economics assumes that individuals have coherent preferences (i.e. stable, context-independent, ‘consistent’), and uses preference-satisfaction as the criterion; but there is now lots of evidence that individuals don’t act on coherent preferences. In this paper I focus on consumers’ surplus: can we reconcile this concept with the findings of behavioural economics? Consumers’ surplus is a central concept in incremental welfare economics. In IO, it is used in evaluations of changes in industry concentration, anti-trust policy, etc. It is standardly defined in terms of individuals’ compensating variations, i.e. in terms of (assumed) coherent preferences. If in fact, coherent preferences don’t exist, what should we do? 2 A possible answer: in practice, consumers’ surplus is measured by areas under demand curves, not by discovering individuals’ preference orderings. In practical applications, demand curves report aggregations over many consumers, many decision contexts, and extended periods of time. E.g. an analysis of the alcoholic drinks market, using price of beer and total consumption per week. Standard neoclassical theory tells us that the area under the aggregated demand curve is the sum of the compensating variations for all the individuals. But.. the aggregate data don’t tell us whether individual drinkers act on dynamically consistent preferences on a micro scale (e.g. if they buy a first drink in evening with intention of buying no more, do they stick to the plan?). Even if individuals’ preferences were inconsistent, demand curve for beer might be quite stable; so standard measurement tools might not affected. But what is the normative status of the measure? 3 This is where the history comes in. In the light of behavioural economics, it’s interesting to look at microeconomics before the ‘marginal revolution’. What kind of normative analysis were economists able to do without assuming utility-maximising consumers (only rational arbitrage/ profit-seeking behaviour by merchants)? Particularly interesting: Jules Dupuit, ‘On the measurement of the utility of public works’ (1844). This is mathematically sophisticated for its time; it is the founding text of CBA. But Dupuit explicitly denies that he is assuming rational consumers. Does this give us any clues about how to use consumers’ surplus without assuming coherent preferences? 4 Dupuit’s analysis Dupuit’s topic: ‘the conditions which [public] works must fulfil in order to be really useful’. ‘Public works’ = roads, railways, canals, bridges, water supply systems… His analysis is addressed to ‘legislators’, i.e. public officials who regulate the provision and pricing of public works. In many respects, Dupuit is far ahead of his time. As far as I know, he gives: -- the first definition of consumer’s surplus; -- the first analysis of price discrimination; -- the first recognition of the distinction between pecuniary and technological externalities; -- the first recognition of the distinction between transfer payments and real resource costs. But: he does not assume that consumers act on consistent preferences – and is explicit about this… 5 Dupuit on (what is now called) the demand function: [The demand function] ‘is not known for any commodity, and it can even be said that it will never be known since it depends on the volatile will of human beings; it is today no longer what it was yesterday. It is thus of no avail to try to determine this relationship exactly by experience or groping experiment, but there do exist certain general laws to which the relationship, in its very volatility, remains constantly subject.’ Significantly, these general laws are observed statistical regularities about demand curves that sum over many consumers – they are not grounded on principles of individual choice: -- a demand curves slopes downwards; -- the slope becomes more negative as price increases (this is explained as the result of the ‘pyramid’ structure of social classes). 6 Dupuit’s account of consumer choice seems more behavioural than neoclassical …. ‘The variable, yea mobile, nature of the value of utility is indeed well known to business men and has long been exploited by them’ [i.e. by price discrimination]. The same commodity in various guises is often sold in different shops at quite different prices to the rich, the moderately well-off, and the poor. The fine, the very fine, the superfine, and the extra fine, although drawn from the same barrel and although alike in all real respects other than the superlative on the label, sell at widely different prices. But price discrimination serves a useful purpose: although the practices of price discrimination used in business work by setting ‘traps for the buyer’s vanity and his credulity’, they are often ‘more equitable and fairer than one might expect at first sight’, and might be good examples to be followed in setting tariffs for public works. 7 What does Dupuit mean by ‘utility’? ‘Hence the saying which we shall often repeat because it is often forgotten: the only real utility is that which people are willing to pay for.’ [This concept of utility] ‘is not, in the last analysis, a rigorous measure of the quality which things have of being able to satisfy men’s needs’… But political economy, being concerned only with wealth, can take account of the intensity of a wish only through its monetary expression. Political economy only bakes bread for those who can buy it...’ Implication: Dupuit thinks that utility, defined as ‘what people are wiling to pay for’, links with wealth (the subject matter of political economy). Why? 8 I suggest the link is the thought: the production and sale of a good is wealth-creating if and only if consumers’ willingness to pay for it exceeds the cost of supply. Dupuit’s criterion (for the purposes of political economy) is wealth-creation in this sense. Hence Dupuit’s interest in price discrimination. If the cost of supply can be recovered from consumers by discriminatory prices, we can be sure that wealth is being created. Dupuit says that (in later chapters of a never-written book) he will show how tariffs for public works can be fixed ‘according to rational principles, in order to produce the greatest possible utility and at the same time a revenue sufficient to cover the cost of upkeep and interest on capital’. ‘Rational principles’ seems to mean setting a discriminatory tariff such that: -- total revenue = total cost; -- no one who is willing to pay marginal cost is deterred from consuming. 9 Dupuit uses the area under the demand curve as a measure of the total ‘utility’ of a good to consumers. Implicitly, he also treats it as a measure of the potential yield of perfectly discriminating prices. And he claims to be treating the demand curve merely as a statistical aggregation of observations of consumer behaviour. Is this approach theoretically valid? As a first attempt to answer this question, I ask whether (in the absence of income effects) the area under the demand curve is a measure of total willingness to pay (= potential yield of perfectly discriminating prices), even if consumers lack coherent preferences. The answer is no, but for interesting reasons… 10 The corner-shop model Consider a consumer at three ‘periods’ in a given day (1 = morning, 2 = afternoon, 3 = evening). At each period, he buys 0 or 1 unit of food, and consumes 0 or 1 unit. (Consumption is possible only from current or past purchases.) ‘Experienced utility’ in each period is measured relative to benchmark of eating and not paying. Paying always has utility of –p, where p = price (note: this implies zero income effect, and allows us to use money as a utility unit). Utility of not eating depends on how many periods have elapsed since previous eating (or start of day). If no such periods, utility = –1; if one such period, utility = –2; if two such periods, utility = –3. Consider demand function for consumption per day, in terms of a nondiscriminatory price that is constant over the day. (Remember that practical consumers’ surplus measurements aggregate over decision episodes.) [I assume that ties are broken in favour of eating] 11 The rational consumer (forms optimal plan for day, maximising sum of experienced utility) price consumption exp utility >4 4 3 2 1 0 (0, 0, 0) (0, 1, 0) (0, 1, 0) (0, 1, 0) (1, 1, 1) (1, 1, 1) –6 –6 –5 –4 –3 0 price 4 daily demand curve for non-discriminatory prices 3 2 1 1 2 3 experienced utility gain = U[p = 0] – U[p > 4] = 6 area under demand curve = 6 maximum yield of discriminatory pricing = 6 (e.g. charge price = 4 for first unit bought in day, price = 1 for each subsequent unit; note that this works only because consumer makes integrated plan for day). quantity 12 So, for the rational consumer: area under demand curve = experienced utility gain = maximum yield of discriminatory prices (as Dupuit claimed). But what about a non-rational consumer? 13 The myopic consumer (maximises experienced utility in current period) price consumption exp utility >3 3 2 1 0 (0, 0, 0) (0, 0, 1) (0, 1, 0) (1, 1, 1) (1, 1, 1) –6 –6 –4 –3 0 price daily demand curve for non-discriminatory prices 3 2 1 experienced utility gain = U[p = 0] – U[p > 3] = 6 area under demand curve = 5 maximum yield of discriminatory pricing = 3 (consumer is willing to pay up to either 0 in period 1 + 0 in period 2 + 3 in period 3; or 0 in period 1 + 2 in period 2 + 1 in period 3; or 1 in period 1 + 0 in period 2 + 2 in period 3; or 1 in period 1 + 1 in period 2 + 1 in period 3). 1 2 3 quantity 14 If the only real utility is that which people are willing to pay for, I think the correct measure of utility is the maximum yield of discriminatory pricing. Because the myopic consumer is never willing to pay for experiences that will occur in the future, there is no tariff that will induce him to pay more than 3 to buy food over the day. So his ‘real utility’ from food cannot be more than 3. Implications for further work I need to create a definition of ‘perfect’ price discrimination which doesn’t depend on assumptions about coherent preferences, and which allows the yield of perfect price discrimination to be measured from observable data. What I know so far: in general, this measurement is not the area under the demand curve. 15 Thank you for listening. 16