In this thesis the relation between dependency rate and saving level in Korea is analyzed. Starting in the early 1970s, the children born during the baby boom period began enter the working age population, relieving the country of high dependency burden. By the way, saving rate has showed a persistent increasing tendency during the period of 1960-1981. Prior to estimating the coefficients of saving function, the feasibility of estimation is considered. The result of the test shows that domestic saving has not been sufficient to fund the domestic investments. This result accords with the history of Korean economy. In two-gap model terminology, the result means that we may identify and estimate ex-ante saving function. Now, according to Mason model, the dependency effects --- level effect and timing effect --- on saving rate are considered. The regression result shows that there is only a timing effect but not a level effect. But in reality, the multicollinearity between independent variables makes this result useless. So reserving the Mason``s implication, another testing method is contrived. The result by this method shows that there is only a level effect but not a timing effect, which is opposite to the above. Further it is proved that the Mason model is long-run or steady sate saving function. We may think that Mason``s model is appropriate to the cross-section data rather than to the time-series data. To verify the above fact, new saving function is derived by asset accumulation approach, and we find that the new function can be interpreted as dynamic saving function. This dynamic saving function explains well the saving behavior of our country, and dependency rate is shown to have a negative effect on saving level. Finally, simultaneous equation model which considers the simultaneous determination of saving and investment is built up.