A Dynamic Analysis of Unemployment and Household Economic Well-Being in Korea

David C. Ribar, George Washington University
Sung Un Kim, George Washington University
Maria Laura Di Tommaso, University of Turin

As a result of the Asian financial crisis in late 1997, Korea suffered a sudden and devastating economic downturn. Official unemployment soared from 2.6 percent at the end of 1997 to 8.4 percent by the beginning of 1999; unofficial figures suggest that the unemployment rate was much higher, probably above 10 percent. While these job losses undoubtedly led to economic hardship for many Koreans, relatively little is known about the impact on individual households. The size of the employment shock suggests that the hardships were severe. Bank failures and other elements of the financial crisis likely further restricted the ability of households to reduce the effects of the shocks through borrowing and dis-saving. This study examines the effect that unemployment had on household incomes and other material measures of economic well-being using newly-available 1998-9 data from the Korea Labor Panel Study (KLPS). The KLPS is a new longitudinal survey, which began with roughly 5,000 households in June-October 1998. In each year, the KLPS asks household heads standard questions about economic circumstances of their households during the previous year including the labor force status of household members, the level of income, level of expenditures, and changes in debts and savings. It also gathers standard demographic information on the composition of the household and the ages and education levels of its members. In addition to these standard measures, the KLPS asks household heads questions regarding material well-being. These questions include whether the household experienced problems paying for food, tuition, debts, medical expenses, housing, or other expenses. The study uses these data to estimate multivariate, longitudinal models of the effects of unemployment, household composition and other characteristics on the various measures of economic well-being. Longitudinal, fixed-effect regression models are used to examine household incomes, expenditures, and savings changes. The fixed-effect specifications account for the confounding influence of unobserved time-invariant characteristics that are correlated with both unemployment and the well-being measures. A longitudinal Multiple Indicator, Multiple Cause (MIMIC) model is used to examine the measures of material well-being. The study uses this specification to examine whether the consumption-based measures of well-being can be represented by a common latent variable. An innovation in the study's MIMIC specifications is that they also incorporate controls for time-invariant effects following the methodology proposed by Chamberlain. Besides the use of a new data set and the introduction of a new empirical model, the study makes several other contributions. First, it carefully examines the relationship between income-based measures of economic well-being and material-based measures. The study examines these outcomes in the context of a life-cycle model of household labor supply and consumption. The model shows the ways in which these are related, yet distinct, outcomes. Results from the study's theoretical and empirical analyses have implications for the measurement of household well-being in future surveys. Second, the study is one of only a few that have examined dynamics in indicators of hardship. There have been numerous longitudinal studies of incomes, poverty, and expenditures as well as numerous cross-section (static) studies of other self-reported hardship measures. However, few studies have examined dynamics in hardship measures, and none has done so in the context of Korea.