Working Paper: NBER ID: w1492
Authors: Laurence J. Kotlikoff; Ariel Pakes
Abstract: In neoclassical models of consumption choice under earnings uncertainty changes in consumption programs from one period to the next are determined by new information received about future earnings over the period. This proposition suggests testing the neoclassical model by ascertaining whether new earnings information explains consumption choice through time. It also suggests that actual consumption choices imbed extractable information about the extent and time resolution of earnings uncertainty. This paper derives a fairly general theoretical relationship between properly defined innnovations in consumption (noise) and revisions in expectations of lifetime earnings (news). It also clarifies the relationship between testing for the theoretical determinants of consumption and standard Euler tests that focus on theoretical nondeterminants of consumption. The chief prediction of the paper's theoretical results, that noise exactly equals news, is tested using aggregate time series data on consumption and earnings. We find that new earnings information explains only a very small fraction of the variance of aggregate consumption innovations. On the other hand, the extent of suboptimal consumption choice appears to be of little economic significance.
Keywords: Consumption; Earnings Uncertainty; Optimal Consumption Choice
JEL Codes: D91; E21
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| variance of noise (C29) | variance of news (C46) |
| consumption choices (D10) | future consumption (E21) |
| new information about earnings (news) (G14) | consumption choices (D10) |
| consumption innovations (noise) (O39) | new information about lifetime earnings (news) (J31) |
| new earnings information (J31) | revisions in expectations of lifetime earnings (J17) |