STATUS AND INTERTEMPORAL CHOICE Debraj Ray Arthur Robson New York University Simon Fraser University Bianjun Xia Simon Fraser University May 1, 2008 1 ABSTRACT The idea that utility or happiness depends on the comparison of one’s own consumption to that of others can be traced back, at the very least, to Veblen (1899). Nevertheless, neoclassical economic theory has typically assumed that an agent’s utility depends solely on the absolute level of consumption. This stance has been harder to maintain in the face of a large body of empirical evidence on an individual’s tendency to evaluate her consumption in comparison with consumption of others, and there is now a growing literature that includes this relative concern, or status, as an additional argument in the utility function. This note considers a preference for status as fundamental. It takes the extreme view that the individual does not care about the consumption of particular goods per se, but only how she ranks in the distribution of consumption. This view makes it possible to straightforwardly reconcile an evolutionary basis for preferences with the obvious fact that absolute consumption levels now vastly exceed any plausible level in hunter-gatherer societies. Despite this apparently fundamentally distinct basis for choice, this paper establishes a startling behavioral equivalence between the relative concern model and an absolute concern model of intertemporal consumption choice. That is, the equilibrium consumption time path followed by individuals with a pure preference for status is the same as that followed by individuals with a logarithm utility function in absolute consumption. This result holds for a large class of preferences for status. There may then be no observable difference at all between a model with a pure concern for status and a conventional model. 2 “Now here, you see, it takes all the running you can do to keep in the same place.” Lewis Carroll, Through the Looking-Glass 1 Introduction The idea that utility or happiness depends in part on comparing one’s own consumption to that of others can be tracked back, at least, to Veblen. Nevertheless, neoclassical economic theory has typically assumed that an agent’s utility depends solely on the absolute level of consumption. This stance has been harder to maintain in the face of a large body of empirical evidence of an individual’s tendency to compare her consumption to that of others (for example, Frank (1985), Brown, Oswald and Qian (2004), and Dynan and Ravina (2007)). There is also now a growing literature that includes this relative concern as an argument in the utility function in addition to the absolute consumption level (for example, Robson (2002), Becker, Murphy and Werning (2005), Hopkins and Kornienko (2004)). This note takes the view that people’s preference for relative status is fundamental and arises as a result of long-run evolutionary process. Further, for the sake of simplicity, it adopts the extreme view that the individual does not care about the consumption of particular goods per se, but only how this consumption ranks relative to others. Such a view has the additional merit of reconciling an evolutionary basis for preferences with the obvious fact that absolute consumption levels now vastly exceed any plausible level in hunter-gatherer societies. We construct a theoretical model of consumption over time in which people care only about status. The agent’s status is her ranking in the social distribution of consumption in each period. Despite this apparently fundamentally distinct basis 3 for choice, this paper establishes a startling behavioral equivalence between the relative concern model and an absolute concern model. Specifically, the equilibrium consumption time path followed by individuals with pure preference for status is the same as that followed by individuals with a logarithm utility function in absolute consumption. This result holds for a large class of preferences for status. 2 Model and Results Suppose there is a continuum of measure one of agents, where each agent i is endowed with intial wealth w ≥ 0. The cumulative distribution of the wealth is denoted by G(w), assumed to be continuously twice differentiable. The associated density function is denoted by g(w). The agents allocate their wealth over time. Time is discrete and the horizon is infinite, i.e., t = 1, 2, ....1 The agents can borrow or save any amount at interest rate r. The agent cares about her relative standing in the economy ,where this standing is the agent’s consumption level relative to that of others. Specifically, an agent with wealth w has status at time t given by Ft (ct (w)), where ct (w) is her consumption at time t, and Ft (·) is the cumulative density function of consumption at time t. The associated density function is denoted by ft (·). (We will only need to consider differentiable Ft (·).) The agent values her status according to some continuously twice differentiable function v(·) with v 0 > 0. Her objective is to maximize the present 1 The results remain valid for any finite horizon also. 4 discounted value of the utility of status over time: max {ct } s.t. ∞ X v(Ft (ct )) t=1 ∞ X t=1 (1) (1 + ρ)t ct =w (1 + r)t where the discount rate is ρ > 0. We first make an assumption about the preference for status and the cumulative distribution function. Assumption 1. v(G(·)) is strictly concave in w. The key result is as follows: Proposition 1. Under Assumption 1, there exists a strict Nash equilibrium in which the equilibrium status for any individual is constant over time. Therefore, Ft (ct (w)) = G(w), for all t and w, where ct (w) is the equilibrium consumption level at time t. Indeed, in this equilibrium, every agent behaves as if she were independently solving the following standard problem: max {ct } s.t. ∞ X t=1 ∞ X t=1 ln ct (1 + ρ)t (2) ct =w (1 + r)t Proof: It is straightforward that the optimal time path of consumption in (2) is 5 given by c∗t (w) = P w 1 (1+ρ)t (1 + r)t (1 + ρ)t ∀t = kt w, say, where kt = (3) (1+r)t (1+ρ)t P 1 (1+ρ)t Consider then the intertemporal choice of a particular individual facing (1) in a population where all other individuals choose c∗t (w) = kt w. It follows that Ft (kt w) = G(w) so that Ft (ct ) = G(ct /kt ). That is, this particular individual solves the following problem: max {ct } s.t. ∞ X v(G(ct /kt )) t=1 ∞ X t=1 (4) (1 + ρ)t ct =w (1 + r)t The first-order conditions for an interior solution to this are then: v0 (G(ct /kt ))g(ct /kt ) = λ(w)kt (1 + ρ)t (1 + r)t ∀t (5) for some Lagrange multiplier λ(w). However, given the definition of kt , these firstorder conditions are satisfied at ct (w) = kt w where λ(w) = v0 (G(w))g(w) X 1 . (1 + ρ)t Given Assumption 1, these first-order conditions are sufficient for the unique global maximum to the individual’s problem. Furthermore, the following uniqueness result holds: Proposition 2. Consider any pure strategy Nash equilibrium, as defined in (1), 6 where all consumption levels are positive, Ft (·)) is twice continuously differentiable, and the support of G contains 0. Suppose individuals have access to any fair bet at any date, but strictly prefer to decline all such bets. Then this Nash equilibrium can only be the equilibrium characterized above, and Assumption 1 holds. Proof: Consider any Nash equilibrium as described above. Since individuals strictly prefer to decline all fair bets at any time t, v(Ft (·)) is strictly concave for all t. It follows that (1) has a a solution for consumption ct (w) that is a strictly increasing and continuously differentiable function of w, since this solution is interior.2 Hence ct (w) = pt (w), say. Furthermore, it must be that F (pt (w)) = G(w) so that Ft (ct ) = G(p−1 t (ct )). Indeed, given that all other individuals are choosing in accordance with pt (w), a particular individual’s problem is max {ct } s.t. ∞ X v(G(p−1 t (ct ))) t=1 ∞ X t=1 (6) (1 + ρ)t ct =w (1 + r)t which has the first-order conditions for an interior solution −1 v0 (G(p−1 t (ct )))g(pt (ct )) = λ(w) dpt (1 + ρ)t dw (1 + r)t ∀t (7) for some Lagrange multiplier λ(w). For a Nash equilibrium, it must be that ct = pt (w) so v0 (G(w))g(w) = λ(w) dct (1 + ρ)t dw (1 + r)t ∀t (8) That is dct (1 + r)t = K(w) dw (1 + ρ)t 2 Strictly, we make the slightly stronger assumption that 7 ∀t d2 dc2t (9) [v(Ft (ct ))] < 0 for all ct and all t. for a suitable constant K(w). Since X dct dw (1 + Hence r)t P ct (1+r)t = w, it follows that = 1 = K(w) X 1 . (1 + ρ)t (1+r)t dct (1+ρ)t = P 1 = kt . dw (1+ρ)t Since the support of G contains 0, it must be that ct (0) = 0, so that ct = kt w ∀t as before.3 Finally, since v(Ft (ct )) = v(G(ct /kt )), Assumption 1 must hold. 3 Conclusions This paper considers the behavioral implications of a model of intertemporal choice in which individuals care only about relative consumption. It may well be that a relative concern is more fundamental than absolute concern from an evolutionary perspective. However, there is a startling equivalence of the present model of status and a conventional model with a logarithmic utility. It could then be very difficult to distinguish empirically between these two possibilities, despite their apparently radically different assumptions. 3 If the support of G does P not contain 0, there are multiple solutions for consumption of the form ct (w) = αt + kt w, where αt = 0 and ct (w) ≥ 0 remains true. 8 References [1] Becker, G., K. Murphy and I. Werning (2005), “The Equilibrium Distribution of Income and the Market for Status,” Journal of Political Economy, Vol. 113(2), pp. 282-310. [2] Brown, G. J. Gardner, A. Oswald and J. Qian (2005), “Does Wage Rank Affect Employees’ Wellbeing?” IZA Discussion Paper No. 1505. [3] Dynan, Karen and Enrichetta Ravina (2007), “Increasing Income Inequality, External Habits, and Self-Reported Happiness,” American Economic Review, Vol. 97, 226-231 [4] Frank, Robert H. (1985), Choosing the Right Pond: Human Behavior and the Quest for Status, Oxford University Press [5] Hopkins, Ed and Tatiana Kornienko (2004), “Running to Keep in the Same Place: Consumer Choice as a Game of Status,” American Economic Review, September, 1085-1107 [6] Robson, Arthur (1996), “The Evolution of Attitudes to Risk: Lottery Tickets and Relative Wealth,” Games and Economic Behavior 14 , 190-207 [7] Robson, Arthur (2002), “Status, the Distribution of Wealth, Private and Social Attitudes to Risk,” Econometrica, Vol. 60, No. 4, 837-857 9