The seventies have seen a burgeoning of an empirical literature on labour-leisure choice grounded in the work on the systems approach to the estimation of demand systems which flowered in the sixties. Working with the Klein-Rubin utility function, Betancourt (1973) extended Lluch’s ELES system (1970, 1973) to include the labour-leisure choice. Abbott and Ashenfelter (1976) analysed United States data on hours worked using both the Klein-Rubin and the indirect addilog utility functions (although, unlike Betancourt, they did not endogenize savings). Kiefer (1977) used a Bayesian approach to the estimation of a 7 commodity plus leisure system which imposed only classical restrictions on a Rotterdam model written in logarithmic differentials. He also explored the use of a quadratic indirect utility function as a second order Taylor approximation to an arbitrary nested indirect utility function in which Box/Cox transformations comprise the inner nests (Kiefer, 1975). Phlips (1978) has used an extended Klein-Rubin utility function, in which the real stock of money appears, to integrate the demand for commodities and leisure with the demand for money. Barnett (1979) has imbedded the commodity and leisure demand problem into a version of the Rotterdam model under relatively weak assumptions about preferences. Finally, recent empirical work using Betancourt’s TELES model on Australian data has been reported by Tulpulé (1978).

A number of models of the individual’s optimal allocation of time over his life between leisure, labor supply and training have recently been devised. Their analysis is a rich source of predictions about individual behaviour — effects of taxes, wealth, tastes, ability, etc. But this information is contained in the comparative dynamics of the models, of which relatively little is known. This paper is an attempt to remedy part of the deficiency : the model of Blinder & Weiss (1976) is examined in detail by means of numerical simulations, which offer several useful insights and reveal various strengths and weaknesses of this construction.

There are three principal areas where the comparative dynamics of the model give interesting results.

(1) Usually, analysis of the individual’s labour supply takes the wage rate as exogenous and is static. A model of optimal life-cycle labour supply and human capital accumulation makes both hours of work and the wage rate endogenous, and puts them in a dynamic context. It suggests that both may change in response to taxes, i.e., taxes may be shifted partly through the work/leisure choice and partly through work/training choices, and invites comparison of the sizes of the two effects. If the effects of taxes on human capital investment are large, then analysis of the relative efficiency of alternative tax schemes which exclude these effects may be seriously misleading.

Cet article présente quelques prolongements d’une étude précédente (1), dont l’objet était de proposer une méthodologie particulière à caractère économétrique pour l’étude spatiale du phénomène de croissance sectorielle. L’approche s’inspire de l’analyse discriminante et a pour but d’expliquer les divergences régionales du point de vue de la croissance sectorielle au moyen d’un certain nombre de facteurs de localisation déterminants selectionnés à partir d’un ensemble d’indicateurs possibles. Ce modèle intègre non seulement des attractions par l’offre et la demande mais également une variété d’autres facteurs qui peuvent favoriser la localisation ou la croissance sectorielle.

De tels phénomènes de seuils ont été proposés dans un certain nombre d’articles précédents (2). Des solutions binaires (dans l’ensemble {0,1}) et des états flous (dans l’intewalle [0,1]) y ont été proposés comme cadres de références possibles. Dans la présente étude une situation intermédiaire est adoptée, le point de départ étant la théorie classique des probabilités. Une région est assignée à l’une ou l’autre de deux classes (régions à croissance élevée, régions à croissance faible) avec une certaine probabilité.

Recent contributions to the theory of the firm have recognized the impact of uncertainty on both the firm’s behavior and its equilibrium position. Models were developed by Baron (1970, 1971), Sandmo (1971) and Leland (1972) that explicitly incorporate a random output price as well as the firm’s attitude toward the risk associated with production under price uncertainty. These models assume that, in the absence of complete foresight, the firm seeks to maximize the expected utility of its profits, in contrast to the traditional profit-maximizing firm operating under certainty. It is then shown that a firm which displays aversion toward risky activities will produce a smaller output under price uncertainty than under certainty, and that, contrary to the firm operating in a certain world, the risk averse firm may alter its optimal level of production in response to a change in its fixed costs.

These and other comparative-statics results derived by Baron, Sandmo and Leland are based on the general principle of expected utility maximization, first developed by Arrow (1965) in his work on the theory of risk aversion. An alternative approach is based on the mean-standard deviation framework (M-SD hereafter) developed by Markowitz (1952, 1959) and extended by Tobin (1958, 1965).