The classical consumption models (Modigliani’s Life-Cycle hypothesis and Friedman’s Permanent Income hypothesis) tell us that consumption is dependent on life-time income. This is based on the assumptions of credit market access (so we don’t have liquidity constrained individuals) and certainty. In short this means that consumption will only change if income changes, and a temporary income change will cause consumption to rise by less (i.e. MPC is low) than a permanent income change (i.e. MPC is close to 1).
Due to the theory of consumption smoothing – whereby individuals prefer to have similar incomes over two periods (or a lifetime) than extremities in either period – we would expect change in consumption to be low over a lifetime. If incomes fall in one period – due to an exogenous change in income (perhaps due to a business cycle recession) – then we would expect individuals to borrow from future income, so that C1 and C2 didn’t differ by much. This would imply that even with income shocks, consumption shouldn’t be volatile.
According to Hall current consumption is based on expected life-time income and incorporates all known information under the assumption of rational expectations. He postulated that current consumption should be independent of any future income changes, as any such changes would already be incorporated into current consumption. Therefore this would mean that changes in consumption should be unpredictable and follow a random walk. It would be unpredictable because it would only change if income unexpectedly changed as a result of a permanent shock. The implication of this – assuming these conditions hold in the real world – is that a policy change will only affect consumption if it is unanticipated.
Hall tests this theory, that changes in consumption should be unpredictable, using post-war US data, and his findings show that there is no evidence to suggest that past consumption data has any affect in predicting future consumption levels.
The way that Hall conducted his test though ignores the Pull of Instant gratification, whereby an individual may prefer utility now than in the future (and so discounts future consumption). Hall assumed constant marginal utility over time, something which this theory would seem to disprove in practice. Moreover, Hall’s results that consumption changes are unpredictable is only valid if we take rational expectations: quite a strong assumption, as some individuals may make their decisions based on heuristics, or from following the crowd, and may not have access to perfect information, nor would they have unlimited processing capacity to make their decision.
Other assumptions that Hall made include a quadratic utility function, so that marginal utility is linear in consumption.