Deaton's observation that consumption appears to be "too smooth" is easily reconciled
with the notion that consumption exhibits "excess sensitivity" to current income.
The reconciliation invokes the plausible assumption that households forecast on the
basis of a larger information set than the econometrician. The empirical case for
excess smoothness is strengthened by showing that the innovations in permanent income
can be inferred from an autoregression of income and saving, not only under the permanent
income hypothesis, but also under the excess sensitivity model. Empirically, parameter
restrictions imposed on the bivariate autoregression by the excess sensitivity model
are not rejected.