May lack of information be the real invisible hand? (Work in progress.)

Pareto improving intervention may be impossible not because it does not exist, but because it is impossible for a planner to figure out one. In this paper, we show that finite sets of market data may not suffice for a planner to be able to design Pareto improving policies. In an incomplete markets economy, after asset markets have closed, there typically exist Pareto-improving asset reallocations. The question that naturally arises is: how much does a planner need to know if he is to figure out one such asset reallocation? Existing identification results (Kubler et al., 2002, or Carvajal and Riascos, 2004) require knowledge of the equilibrium manifold that may be unrealistic. If the informational requirement for finding Pareto-improving policies is implausible, market performance is not questionable on efficiency grounds. We show here that market information in a nonstationary economy does not provide enough information for the design of Pareto-improving policies, even when these policies are likely to exist. When the information available is a set of prices and endowments, we prove that there may exist multiple rationalizations in which the welfare effects of economic policy go in opposite directions. 1 Not everything will do Consider a two-period economy with uncertainty. There are I ∈ N individuals, S ∈ N states of nature and L ∈ N commodities, with commodity 1 as numeraire. There are A < S linearly independent numeraire assets. Asset payoffs in state s are rs ∈ RA. Let U be the class of all strongly concave, strictly monotone, C2 functions u: RL+ − → R. For each individual i, state-contingent endowments are ei = eis ∈ RL++ ¢S s=1 and state-contingent preferences are uis ∈ U ¢S s=1. There is no date-zero consumption, and ex-ante preferences are U i