再说要认真看书学习的一例:卢卡斯论效率与分配

THE ECONOMIC JOURNAL

MARCH 1992

The Economic Journal, 102 (March 2), 233-247

Printed in Great Britain

ON EFFICIENCY AND DISTRIBUTION

Robert E. Lucas, Jr.

It helps us to see the actual world

to visualize a fantastic world.

Wallace Stevens

[邹恒甫按:这是何等尖刻和幽默的引用啊!我们能做道吗?不能! 我们敢做吗? 不能!敬请大家认真地读一下此论文的导言部分,好好想想Wallace Stevens的话同卢卡斯的文章的关系:简直是苦刹研究收入分配和平等的经济学家!]

This article is concerned with the possibilities for allocating resources in an economy with the following characteristics. There is a constant, perpetual endowment flow of a single, non-storable good. There are many households, all infinitely lived, and all with the same ex ante preferences over time paths of the consumption of this single good. Each of these households is subject to shocks to its preferences, unpredictable even to itself, that give it a high urgency to consume in some periods and a low urgency to consume in other periods. These shocks are independent from household to household, so with a large number of households they will average out in any given period, with urgent consumers just balanced out by the less urgent.

Allocating resources in this setting means just distributing the given endowment, at each date, across these different households. There is no production and no contemporaneous goods-for-goods exchange. But however this allocation is done it must be consistent with the fact - the crucial assumption of the analysis - that these individual shocks are purely private information. The only way anyone can obtain any information about any particular consumer's urgency to consume is to ask him, and there is no way to audit or verify the answer this consumer chooses to give.

Of course, these data about tastes, technology, and information are not in themselves sufficient to determine how resources are allocated. This is why I referred to the 'possibilities' for allocation a moment ago. One could take a normative point of view and consider how a hypothetical, beneficent planner would distribute the consumption good across households, under various assumptions about the information available to him. Or, one could allocate property rights to the endowment stream, set up a system of markets, let households trade and see what allocation they come up with. This approach, too, will give different answers depending on the assumptions one makes about trading possibilities. I will pursue all of these directions. In each case, given an initial distribution of wealth (of entitlements to current and future consumption) each specific method for allocating resources will imply a complete description of the way this society's wealth distribution evolves over time. What is striking to me in comparing the distributional dynamics implied by different allocative mechanisms is how radically they differ. Starting from a position of ex ante equality, we will see everything from perfect equality in perpetuity to convergence to a stationary distribution to inequality that grows without bound.

All of these possibilities will be seen to arise in a society of essentially identical households, free of the issues of class and race that so complicate questions of distribution in actual societies. Perhaps it is a mistake to try to think about distributional questions at all in a context that abstracts from these distinctions.

Certainly this choice will dictate more than a little caution in drawing conclusions from our theoretical analysis about distribution in actual societies. But the idea that a society's income distribution arises, in large part, from the way it deals with individual risks is a very old and fundamental one, one that is at least implicit in all modern studies of distribution. If we cannot think clearly about this issue in the abstract context I have described, what hope is there for dealing- with more realistic situations?

I will illustrate these distributional possibilities with concrete examples drawn from research in which Andrew Atkeson and I are currently engaged (Atkeson and Lucas (1991)). Our work has many antecedents, but I find it substantively most instructive to view it as contributing to a line of inquiry initiated by Truman Bewley ( 1983) and Edward Green ( 1987) Bewley's paper was the first attempt to imagine in detail what an entire society would have to look like if the behaviour of individual households in it were to be consistent with Friedman's (1957) permanent income hypothesis. Of course, it is this step that converts Friedman's theory of individual behaviour into a general equilibrium theory of distribution. Bewley took a particular market structure, necessarily incomplete in the Arrow-Debreu sense, as a given. Green took matters a step further to take the information structure of the economy as given, and then derived the efficient allocation implied by this information structure under specific parametric assumptions about consumer preferences.

Since then many others, notably Taub, Phelan and Townsend, Marimon and Marcet , and Thomas and Worrall ( 1990) have used theoretical or numerical methods or both to work out the implications of other specific assumptions on preferences and information structures. As empirical work by Townsend, Mace, Cochrane, and others amply demonstrates, the Bewley-Green viewpoint leads to new and extremely interesting ways of interpreting data on household income and consumption expenditures. I think it has equally radical implications for the way we think about distributional dynamics.