Can faster income growth reduce well-being?

Recent evidence suggests that the phenomena we are dealing with are not confined to the rich in the U.S. The Los Angeles Times of September 27, 2000, for example, carried the U.S. Census Bureau's report on sustained income growth in the 1990's. However, the Times also carried this observation: “Middle-income households are working more hours than ever to stay ahead,” said Lawrence Mishel of the Economic Policy Institute...From 1989 to 1999, a typical married, middle-income couple with children added 279 hours—about seven weeks—of work a year...“That's a ton of work, and the growth in hours explains most of growth in income over this last decade,” he added (emphasis added).

The assumption of identical relative factor-intensities may be dispensed with if our ‘small open economy’ assumption extends, not only to the market for capital, but also to the markets for the two goods: the economy would then face exogenous terms of trade, which we take to be constant in the main text, and time-varying in the Appendix.

The July 2001 Special Issue of the Journal of Economic Behavior and Organization, on ‘Subjective Well-Being and Economic Analysis’ (Richard Easterlin, ed.), is also of interest in this regard, as is the October 1998 issue of the Journal of Public Economics, on ‘The Economics of Status’ (Giacomo Corneo and Thomas Piketty, eds.; see also Corneo and Jeanne (1997)). Dynamic models incorporating consumption externalities, especially with more than one good, are rare, and a couple are discussed in Section IV.

Frank (op. cit., p. 14) approvingly quotes Thorstein Veblen, on the status- enhancing or maintaining effects of luxury goods consumption: ‘since the consumption of these..excellent goods is an evidence of wealth, it becomes honorific; and conversely, the failure to consume in due quantity and quality becomes a mark of inferiority and demerit’. The externality-inducing effects of luxury consumption are a key element of Frank’s argument. With slight re-interpretations, our formal model is equally applicable to the situation of the well-off in the U.S.: the rise in their incomes over time could, as various studies have argued, be attributed to skill-biased technological improvements (as well as, Frank suggests (Chapter 3), to the spread of ‘winner-take-all’ markets), and C _2t would now denote the per capita status-goods consumption of well-off persons as a group.

It is readily shown that this result does not depend on the logarithmic form of the utility-of-leisure term, and holds for general iso-elastic functions. Regarding the constraint ℓ _t ≤1, we thus only have to verify that it is satisfied at t=0. Since, from Eq. (6), it is satisfied if F=0, and since, from Eq. (5), λ ₀ is a continuous (and increasing) function of F, it will continue to be satisfied for a range of positive values of F. We assume that F is such that ℓ₀ is strictly less than 1—the empirically relevant case.

Some technical points should be noted here, bearing in mind that our intention is simply to exhibit the possibility that dU*/dx can be negative: (i) we assume τ≥1, so that\(\tau ^{{1 \mathord{\left/ {\vphantom {1 \beta }} \right. \kern-\nulldelimiterspace} \beta }}\) increases or remains unchanged as β decreases; (ii) to keep λ ₀, the shadow price of wealth, unchanged in Eq. (11) as β is reduced and α is increased, we simply have to assume that these changes are made in such a manner as to keep the right-hand side of Eq. (5) unchanged at an unchanged value of λ ₀; (iii) we could, for the same purpose, introduce an additional parameter, say ξ (initially equal to 1), multiplying the \({{\left( {c^{{1 - a}}_{{1t}} } \right)}} \mathord{\left/ {\vphantom {{{\left( {c^{{1 - a}}_{{1t}} } \right)}} {1 - \alpha }}} \right. \kern-\nulldelimiterspace} {1 - \alpha }\) term in Eq. (1), and reduce it as τ is increased to keep the right-hand side of (5), again, unchanged at an unchanged value of λ ₀.

Not only would such a tax, if in place, necessarily result in an increase in x being welfare-increasing, but for a given x it would shift the time-path of c _2t logarithmically downwards, and the time-paths of c _1t and \({\left( {{\ell }_{t} - {\mathop {\ell }\limits^ - }} \right)}\) logarithmically upwards − examples of reduced conspicuous consumption in favour of increased welfare-enhancing ‘inconspicuous consumption’ (Frank, pp. 90–2). Investment subsidies and lump-sum transfers are advocated by de la Croix and Michel (1999) in an interesting alternative framework in which the externality is intertemporal, children taking their parents’ (total) consumption as a benchmark (see also de la Croix and Michel (2001)).

Not only would γ differ across goods and individuals, but so also would τ and β: an individual in a high consumption bracket may thus end up paying a high marginal tax rate, even though the goods he consumes are characterized by a low γ (but a high τ or low β). On the other hand, Ng Yew-Kwang suggests (private communication) that a useful complement to Frank’s proposal is to have additional excise taxes on those items which are quite likely to be characterized by strong status–good externalities for most people (such as the luxury watches ranging upwards in price from $17,500 that Frank mentions (pp. 16–17), as well as pleasure craft, ultrapremium wines and premium cigars (Chapter 2)). Such excise taxes would imply that the across-the-board consumption taxes that Frank advocates could be lower than they would otherwise be, reducing the distortions arising from heterogeneity of goods and agents discussed above.

Cooper et al. argue that the primary purpose of quality improvements in the status good is to enhance its prestige value in the eyes of consumers. It is highly implausible to suppose, however, that in the process no absolute consumption benefits are reaped. Is a Jaguar preferable to a Mercedes only because it has higher prestige value, or also because it is more luxuriously fitted out, offers a smoother ride, and the like?

Mention should also be made of the works of Cole et al. Concerns for relative wealth in their models arise from matching concerns, and the former in turn affect savings decisions (their 1992 multi-period analysis) or labour supply decisions (their 1995 single-period analysis): in the latter, conspicuous consumption may occur as a signaling device if wealth is unobservable. In practice, however, a few observable determinants and correlates of wealth (such as occupation and location and size of residence) are generally fairly accurate indicators, so it is unlikely that their approach can by itself account for consumption of a wide range of status goods. Stated differently, higher relative consumption of status goods may have an intrinsic desirability in the eyes of many people, quite apart from its instrumental value as an indicator of wealth. Indeed, after writing these lines I came across the following observation by James Coleman (1990, p. 130, quoted in Becker and Murphy (2000, p. 105)), ‘status, or recognition from others, has long been regarded by psychologists as a primary source of satisfaction to the self. That is, an interest in status can be regarded as being held by every person’.

In an extended analysis, γ and β may be functionally related, rather than independent parameters: certain goods are more income-elastic because they are better signals of status, the demand for which is income-elastic. This also suggests that the causal sequence in a model of the Cooper et al. type should be the reverse of that postulated by them: instead of quality improvements inducing rising expenditure shares, the rising demand for status fuels more rapid quality improvements in those goods which are more visible indicators of status.

It should be pointed out, though, that there does not exist universal agreement on the facts. Davis and Henrekson (2003, pp. 33–4) remark, ‘While more research is needed to settle the issue, the widely held view that Europeans enjoy more leisure than Americans may be a myth founded on an overly narrow conception of work activity [that excludes time spent on home production] and a lack of comprehensive, internationally comparable time-use data’.

Two further insights provided by the analysis in the Appendix are: the additional restrictions necessary to ensure that c _2t , or total spending on good 2, rise faster than c _1t ; and additional influences on dU*/dx if it is assumed that the increase in x at time 0 affects the trend rate of change of the relative price of good 2.

Other, institutional considerations may have played a role over a longer time horizon. For example, 50 years ago the prevailing social norm was for housewives in middle-income families with (young) children not to work. Work opportunities for housewives may also have been more limited then, but the role of social norms, and of societal and individual value-systems generally, should not be overlooked. An interesting recent model along these lines is Hazan and Maoz (2002).