Demand

Once the theory of demand had been rigorously formulated, economists were still prepared to allow exceptions for addictive commodities. If individuals were addicted, in the total loss of control sense, then the good would exhibit a very low price elasticity of demand. This might seem to be supported by the legacy of many centuries of failed attempts to control addictive goods by taxation and punishments [Corti (1931)]. From the public finance point of view then addictive goods would become very attractive items for sumptuary taxation which would, in essence, become a means of redistributing rents from producers and/or addicts to the rest of society. Many economists, working in the crime area, have shown a particular interest in heroin demand [for references and summary see Maynard & Wagstaff (1988, ch. 5)], coming to the conclusion that there are vertical segments on the demand curve over a fairly wide range of prices, but when prices are relatively 'high' or 'low' we switch to a conventional demand curve. At the low-price region this happens because the addict is tempted by income and substitution effects to allow their maintenance dose to drift up to the new customary level. In the high-price region the demand falls mainly because of income effects: that is, the addict is now expending so much of their resources on procuring the drug that they need to switch towards other goods. The proposal of an 'extreme seeking' approach [Barthold & Hochman (1988), Averett & Hochman (1994)], premised on indifference curves with concave segments, tries to provide a foundation for the vertical segment in terms of individuals having 'preferences that are structurally inconsistent, over some range, with the conventional axioms of demand theory'. These extreme seekers have indifference curves which are negative, as in the conventional case, but with concave and convex segments so that marginal rates of substitution are increasing in some ranges. This generates corner solutions; i.e., in a two-good model only the addictive good is consumed and it is extremely unresponsive to relative full price changes. The segmented demand curve described above, where two conventional demand curves are joined together by a vertical section, can be linked to the general dynamic model of consumer demand proposed in the pioneering 1960s econometric work by Houthakker & Taylor (1970) who applied it to data for alcohol and tobacco. Here consumption depends on a 'stock of habits', which is a function of past consumption and the depreciation rate in the stock of habits.

In the last ten years or so the myopic approach has almost completely dropped out of studies of addiction demand as the 'rational addiction' approach has become the dominant research paradigm. Rational addiction is the term used by Becker [see Becker & Murphy (1988), Becker (1992)]. It is important to return, at this point, to the issue of definitions of rationality. The work reviewed in the previous paragraph was an attempt to explain seemingly irrational behaviour (people being less well off than they could be) in terms of them being trapped in some kind of inertia in the manner of the X-inefficient agents of Chapter 4. The decision-makers are still assumed to be operating in a rational manner, so it would not be strictly correct to call these 'irrational' addiction models. The key difference in the Becker models is that the addicts have a much higher degree of foresight (indeed in the formal models it is perfect foresight) in the shape of a willingness to act on the basis of the knowledge that their habits are not a 'given' but are rather endogenous to the level of consumption. For what it is worth, this seems to be exactly what Thomas de Quincey is saying in the quote with which we started this chapter. Accounts of the withdrawal problems of his contemporary Samuel Taylor Coleridge are not so sanguine, but it could be claimed that the latter's torture was due, in large part, to a lack of knowledge on the nature and operations of opium.

Becker & Murphy (1988, pp. 675-6) make no distinction between goods in terms of their pharmacological properties as 'people get addicted not only to alcohol, cocaine and cigarettes but also to work, eating, music, television, their standard of living, other people, religion, and many other activities'.

This is a very similar list to that on the Christian website given above and it chimes with recent trends in psychiatric practice. The crucial feature of addiction, in this sense, is that the marginal utility at a point in time depends on consumption in other periods; it does not matter whether this interdepen-dency has a chemical component or is simply learned habitual behaviour. A fully rational individual plans for the sequential dynamics brought about by the intertemporal non-separability of consumption arising from addiction. The distinctive feature of the model is that, under certain assumptions, it gives rise to a linear demand equation in which price, income-led consumption and lag consumption enter. This is a general formulation which nests the myopic Houthakker-Taylor model as a special case in which the coefficient on the lead term is zero. It is also easy to work out the estimated discount rate from such a model by applying the relevant formula to the coefficients on the lag and lead terms. These types of equation are now commonplace in health economics and general economics journals, and we even have the spectacle [see Chaloupka et al. (1999)] of behavioural psychologists using baboons, rats, etc. to estimate rational addiction equations from laboratory studies where they provide addictive drugs to these animals. Many studies in the Chaloupka volume review the same evidence on the range of magnitudes of the own-price elasticity of drug use and also present their own. Broadly speaking the overall impression is of a figure in the -0.5 to -1.0 range.

An interesting, but neglected, question is the differential addictiveness of women. As smoking in conjunction with the contraceptive pill greatly increases the risk of heart attack or stroke [Jacobson (1986)], and babies of smokers, in developed countries, weigh on average 200 grams less at birth

[World Bank (1993)] than those of non-smokers, one would anticipate that women have greater incentives not to become smokers. This generally applies to other addictions also. In the UK, female smoking (measured in cigarettes per day per adult) rose steadily from 1956 to 1973 against a background of fairly stable male smoking rates. This took place in the context of a general anti-smoking environment. Reports of 1953 and 1962, from the Royal College of Physicians, drew considerable attention to the health-smoking link. In 1965, broadcasting of cigarette advertisements, on electronic media, was banned. These measures were not followed by any decline in female tobacco consumption but rather the reverse. Whilst per capita female cigarette consumption fell after 1973, the fall was less dramatic than that for men. From 1976 to 1994 the percentage of adult males smoking fell from 48 per cent to 28 per cent whilst the corresponding fall for females was from 38 per cent to 26 per cent [ASH (1995)]. US experience of trends in male-female smoking has been similar to the UK; in 1955-1966, against a background of anti-smoking information, male smoking fell whilst that of females rose. In contrast, in Taiwan only 3 per cent of women smoke, despite a heavy growth of advertising from multinational firms and a strong male smoking habit, whilst in South Africa, smoking amongst women has been low because of 'cultural traditions' [Mandela (1993, p. 99)]. There does not appear to be any country in the world where adult female tobacco consumption outstrips that of males to a significant degree.

Evidence on differential, by gender, price elasticities in cigarette smoking is somewhat mixed. The evidence on male-female differential responsiveness, from work on other addictive goods, is inconclusive. Some studies of the demand for alcohol exhibit a larger price elasticity for women [Kendell, de Romanie & Ritson (1983), Kenkel (1993)] and a study of 1834 injecting heroin users in Oslo [Bretteville-Jensen (1999)] finds that the female sample is more price-responsive than the male and also has a higher drug intake level. In contrast, a study using the discount rate criterion in a rational addiction approach to illegal drugs [Pacula (1997)] finds women to be more addictive. To further confuse the picture, a recent study by Saffer & Chaloupka (1999) using the National Household Survey of Drug Abuse finds few significant male-female differences in price responsiveness for cocaine, alcohol, marijuana and heroin.

Some additional results are available, in the econometric work on non-price factors, which seem to imply that women are more addictive than men in terms of responding to health scares [Atkinson & Skegg (1973)]. Leigh (1995) finds a deterrent effect of anti-smoking ordinances (bans on smoking in public places) on the probability of smoking, although Chaloupka (1992) using a similar analysis finds no effect of these ordinances on the amount that women smoked. One econometric study [Viscusi (1992)] using telephone survey data for 1985 deals with the issue of sex differences in the perceived risk of smoking. He finds that the perception of health risk from smoking is higher for women (p. 73), which implies they should have higher price elasticities and be more likely to give up, in the face of other deterrents, if there is no sex differential in addiction.

Clinical studies of individual smokers indicate that women have much greater difficulty quitting than men, suggesting that they may be more addicted. Jacobson (1986) reports that the Royal College of Physicians in the UK, in 1983, found that it was twice as difficult for women to remain committed non-smokers. Gender differences in quitting are also well documented in the USA [Pierce et al. (1989)]. A study in California in 1990 [Kaplan et al. (1993)] found that, amongst former smokers, women were more vulnerable than men to relapse in the second year after quitting, although there were no differences in quit difficulty ceteris paribus.

The main message of rational addiction models is that the old-fashioned view that addict consumption will not be very responsive to prices is misguided. This has the corollary that revenues from sumptuary taxes will be very sensitive to tax increases. We should also expect that high-level rationality should mean addicted consumers will be very responsive to anti-use ordinances and (valid) health information which is provided to them at low cost. Although Becker appeared to soften the Chicagoan insistence on treating tastes as given, in his well-known Nancy Schwartz lecture [see Becker (1992)], he still seems to treat the non-addicted as fundamentally similar to the addicted, except that some random variable has caused them to hold off falling into the 'trap' of addiction. He remarks that 'demand for addictive goods tends to be bifurcated: people either consume a lot, or they abstain because they anticipate that they will become "hooked" if they begin to consume. Smoking is a good example of bifurcation, for 70 per cent of adults in the United States do not smoke, while persons who do smoke generally consume at least half a pack a day' (ibid., p. 329). One problem with such observations is that the only reliable statistics available are from legalized goods and it may be that cannabis consumption, for example, in many countries is not bifurcated as extremely as that for cigarettes.

In recent presentations Becker, and those who follow his line, argue that addicts may lower their lifetime utility because the rational addiction model is predicated only on a degree of 'foresight' rather than omniscience. Supposedly the failure to optimize lifetime discounted utility comes about through an endogenous discount rate whereby substance abusers acquire an unfortunately inappropriate discount rate because of their substance consumption. This must surely lead to the bifurcation remarked on above. There is a worry about the distinction between cause and effect when we deal with the issue of bifurcation and addiction. Can a product be addictive and not lead to a bifurcated demand? Presumably it can if there is a 'high' level of consumption for everyone (a rectangular distribution) or a spread with a peak such as a normal distribution. Addiction is a somewhat pejorative value-laden label (despite Becker's claims on beneficial addictions) and it seems that a social code of regarding a good as 'addictive' is unlikely if it is widespread in usage. Nevertheless, much consumption has a thick tail of low/zero users and an opposing tail of heavy users. No doubt we can all agree that fine-cut marmalade is not an addictive good, even though it probably has a bifurcated user profile in many countries. But how do we arrive at this conclusion? Is it on intuitive grounds that marmalade does not lead to endogenous changes in the discount rate? If so is this not merely derived from the knowledge that no addictive chemicals are present and that no serious health damage is likely from 'high' usage? If addictive chemicals present in goods are forcing a person into sub-optimal choices by altering the discount rate then the word rational is surely being somewhat strained.

Not surprisingly, Becker's approach has met with criticism from well outside the 'hard core' of American economics [e.g. Tomer (1996, 1998), Cameron (2000a)] and from slightly nearer home [Akerlof (1991), Winston (1980), Gruber (2001)]. Most critiques have to stop short of claiming that addicts are totally irrational lest they cease to be regarded as 'economics' in the first place. The theory gap is filled with some notion of meta-preferential choice or a modification of Festinger's cognitive dissonance [Festinger (1957) as treated in Gilad et al. (1987)]. A time-honoured problem arises with the alternative approaches, viz. however much more plausible they may seem, it is difficult to operational-ize them. A number of papers have sought to refine and extend the rational addiction model to enhance its resistance to criticism [Orphanides & Zevros (1995), Coutoyannis & Jones (1999), Jones (1999), Suranovic, Goldfarb & Leonard (1999)]. In general such refinements of rational addiction models tend to take the route of incorporating adjustment costs into the process in the pursuit of greater realism. For example, Orphanides & Zevros model the demand-side process entry to addiction markets as a stochastic process so that people can become accidentally 'hooked'. Ultimately the rational addiction model is just an extension of the principles of discounting utility to a situation where some goods exhibit temporal complementarity whilst others do not. It has certainly encountered considerable problems in estimation. Becker, Grossman & Murphy (1994) gave highly statistically significant results, albeit with an implausibly high discount rate in terms of 'rational' behaviour given that the implicit discount rate in daily decision-making is usually under 15 per cent. Some studies have even found negative discount rates.

An alternative approach is to focus on a purely static conceptualization drawing on the more behavioural approaches to choice discussed in Chapter 2. This takes into account the risk and social context factors that pure rational choice models neglect. Initially I shall assume that the good has biological addiction properties, i.e. it contains chemicals that are directly delivered in the process of using it. Behavioural addictions with no chemical content will be discussed afterwards.

Let there be two goods: one (X) which is 'normal' in the sense that it does not deliver any chemical alterations to the brain state nor is it imbued with any learned components; and another (Z) which delivers chemical 'hits' to the brain, through the bloodstream, and also provides utility from being a constituent of 'scripts' (M) [Van Raaij (1990) pp. 168-9)]. A script is a scenario of interlocking consumption elements which have a reinforcing quality. We can distinguish between private and social scripts. In the private script one might be drinking or smoking alone as a reward for completing a solo work task, or simply because one finds an alcoholic drink relieving on a hot day, or to alleviate feelings of loneliness or tension. In the social script, the tension and excitement of the presence of other people are factors in generating the desired level of alcohol consumption. The social script element is enhanced by the impact of alcohol on lowering inhibition or cigarettes in easing the process of conversation for a stressed or nervous individual. The basic idea of the script is that the image of the consumer becomes intertwined with the experiences which are temporally joint with the consequences of the bodily absorption of the relevant chemicals. Over a process of time, these mental associations (or imaging) may be a non-separable element of the utility from the psychopharmacological processes. The mental process of association may engender brand loyalty as the style of bottle, name of the product, etc. can potentially become part of the overall pleasure.

The individual utility function is:

where H = 'hits', S is the volume of 'scripts' in which Z features, W is the perceived stock of health. There is assumed to be an exogenously given menu of scripts which the individual can work his/her way through. The delivery of scripts, hits and health is through the following production functions:

W = W(Z, X, k) = &(k) W(Z, X) + [1 - &(k)] W(X)

where

0 0

Post a comment