We use household-level monthly panel data from Korea to study the monthly change in consumption. We believe the Korean data will produce better estimates than most consumption data: the survey methods apparently produced good measures of total household expenditures as judged by a comparison with aggregate totals; furthermore, because almost all saving in Korea takes place at the household level, rather than through pensions or Social Security, the saving rate can be accurately calculated. We fit directly, both parametrically and nonparametrically, the first-order condition on consumption derived from intertemporal utility maximization. We find that consumption increases with age until middle age, and then falls as would be required by the life-cycle hypothesis, and that, conditional on an outside measure of risk aversion, the subjective time rate of discount is within reasonable bounds. (C) 1997 Academic Press.