HOUSEHOLD SAVING RATES IN KOREA - EVIDENCE ON LIFE-CYCLE CONSUMPTION BEHAVIOR

Abstract

We use household consumption and income data from the 1984 Family Income and Expenditure Survey of Korea to study saving during the working life. We believe the Korean data are better than U.S. data because the nature of the survey suggests more complete consumption coverage, and, indeed, aggregation to national averages reinforces this view. This is important because it means we can calculate saving to be the difference between income and consumption rather than as wealth change, which is typically subject to considerable observation error. Furthermore, in Korea almost all household saving is undertaken by the household whereas in the United States a substantial fraction is done by firms through pension plans and by society through Social Security. In Korean data we can observe from household data the variation in saving rates with age; in U.S. data many households have little observable savings, and, therefore, we cannot easily study life-cycle saving behavior. We find that the saving rate by age follows the pattern predicted by the Life-Cycle Hypothesis: the saving rate is low at young ages, rises to a maximum in middle ages, and then falls, becoming negative after about age 65. We investigate the role of household size in the consumption rate: our data show that it has very little effect, and then only among households whose heads are young. We present a theory which says that, just as in the Permanent Income Hypothesis, the effect of family size on the saving rate depends on whether the family members are permanent or temporary: increasing the number of permanent members should have no effect on the saving rate whereas increasing the number of temporary members should decrease the rate. Our data are roughly consistent with this theory. Our conclusion is that the consumption rates in our data set are consistent with the life-cycle hypothesis both in general and in detail. Because of data limitations similar evidence is hard to find in U.S. data. (C) 1995 Academic Press, Inc.