Using a quasi-experimental approach, the paper examines how the introduction of a social security system in Korea affects household saving. The relatively short history of social security in Korea provides an exogenous source of variations both in time-series and cross-section. For time-series variation, periods 1983 to 1987 and 1989 to 1991 are used in various combinations. For cross-sectional variation, government employees with special pension plan are used as control group. The paper finds that a social security system crowds out household saving. Estimates show that the effect of Korea’s national pension on household saving ranges from -0.9 to - 1.8 percentage points, and -1.4 percentage points on average. When the group-specific time trend is considered, the crowding-out effect is estimated as -3.4 percentage points to -9.3 percentage points, and -5.9 percentagepoints on average. The crowding-out effect might get larger as a social security system matures.