On se propose ici d'explorer la sémantique de l'hypothèse d'anticipations rationnelles telle qu'on la trouve formalisée chez Muth et Lucas. On montre tout d'abord que cette hypothèse appelle une interprétation tantôt comme condition d'équilibre, tantôt comme exigence de rationalité individuelle. On s'interroge ensuite sur la compatibilité de ces deux interprétations, avant de conclure par la négative. Il n'est même pas établi qu'une fois à l'équilibre d'anticipations rationnelles, des agents rationnels ne devraient pas en dévier. L'hypothèse apparaît comme un concept formel de stationnarité en univers stochastique plutôt qu'elle ne décrit la rationalité cognitive des agents. L'argumentation repose sur une analyse des difficultés et paradoxes qui surgissent dans les modèles de révision des croyances (ou d'“apprentissage des anticipations rationnelles”).

Under the assumption that prices are proportional to labour values, it is demonstrated that Ricardo's reasoning on the relation between profit and rent in the Principles contains two errors. Both of them point to the neglect of commodity inputs. Curiously, though, they compensate one another, so that Ricardo's conclusion about the tendential fall of the rate of profits as a result of capital accumulation when there is no technical progress remains valid.

The efficiency of demand management policies is considered in a two-sector model of a small and open economy with centralized wage bargaining. Differences in labour demand between sectors are shown to have qualitative implications for the responsiveness of wages to changes in output prices. Wage increases induced by increased demand in one sector may, thus, crowd out activity in other sectors, and this has important consequences for the effects of demand management policies on internal and external goals. An often proposed and practised policy of switching demand from tradeables to non-tradeables to ensure internal and external balance is extensively discussed.

The taxation of income is usually criticized for the efficiency loss that results from distorting the intertemporal consumption decision. This argument is rejected in a world where the return to savings is random. It is shown that a distorting marginal tax on uncertain interest income serves as partial insurance and that this effect outweighs the efficiency loss, if only the first marginal dollar in tax revenue is valued less elastically than the last dollar spent on private consumption. A consumption tax — even one with intertemporally differentiated rates — would not provide insurance under these circumstances. The explicit consideration of uncertainty thus gives support to taxes that allow differentiation between the safe withdrawal of savings and the random return thereon.

Methods for nonparametric estimation and comparison of cross section Engel curves are presented and applied to U.K. expenditure data. Real Engel curves (with quantity demanded and real total expenditure on the axes) vary over time, but their shapes are generally quite stable. Mean normalized Engel curves are defined and are found not to vary greatly over time. Consequences of such invariance for the testing of microeconomic demand models are investigated.