The Effect Of Social Security On Savings

The controversies about the effect of social security on savings have run through the whole process of the global disputations about social security reform since about thirty years ago. The different answers to this question have directly resulted in different opinions about social security reform. So it is very necessary to enhance the studies on this question when the social security reform in China has been advanced continually today. This paper systematically probes into this question through analyzing the relative theories and previous experiential researches and investigating international experiences. I hope it can play the role that throws out a brick to attract jades, trigger off extensive discussions about this question in China.In the first chapter, this paper analyzes and estimates the various theories and viewpoints about the effects of social security on savings, finds that the different models result in the different conclusions. Within the framework of the traditional life cycle model, the introduction of social security may result in reducing private savings by one-to-one. While Feldstein's extended life cycle model emphasizes the induced retirement effect of social security, so draws the conclusion that the net impact of social security on savings depends on the balance between the wealth replacement effect and the induced retirement effect and can therefore only be determined empirically. The behavioral life cycle model denies the assumptions such as rational economic agents and fungible wealth in previous life cycle models by using some psychological viewpoints, so rejects the previous argument that social security depresses private savings greatly. Barro and other people's intergenerational transfer theories mainly animadvert that life cycle model rips the relations between generations, put forward the impacts of the existing altruism. They believe, if the intergenerational transfer issues such as bequest are taken into account, the argument that social security depresses private savings greatly would be weakened ulteriorly. And the theories such as imperfect market, intergenerational redistribution, substituting for private pension, myopia, recognition effect, a goal gradient will also weaken that argument farther from various angles. In a word, I consider that any model or theory can't cover all the effects of social security till today. I believe, the last conclusion should be that the net effect may be little even if the effect of social security on private savings exists because various impacts of social security on private savings in different directions counteract one another.In the second chapter, this paper analyzes some time series studies about the effect of social security on savings. The conclusions of Feldstein tend toward supporting the hypothesis that social security depresses private savings greatly, but all experiential evidences in other studies don't support it. Though there are many Feldstein's time series studies, they are all within the framework of Feldstein's extended life cycle model. And the analyses of regressive results by Feldstein are inclined to justify his extended life cycle theory. So it isn't difficult to think his empirical conclusions are to be consistent with his theory. Munnell's studies oppugn the representativeness of Feldstein's social security wealth for the first time, think it still only measures the wealth replacement effect. Moreover, Munnell notices the difference between retirement savings and total private savings. Thus her conclusion agrees that social security depresses retirement savings greatly but does not affect aggregate personal savings. Barro adds other three variables to Feldstein's model: the interaction term of unemployment rate-personal income (RU·YD), the government surplus (SUR) and durable goods (DUR), so concludes that the effect of social security on savings is not significant. Darby's study is within the framework of the permanent income hypothesis, his empirical conclusion is similar to Barro's conclusion and doesn't support the hypothesis that social security depresses private savings greatly. Leimer and Lesnoy analyze Feldstein's studies very detailedly, find that there are some problems in Feldstein's studies such as calculational error, some inappropriate assumptions and so on. If these errors are corrected, Feldstein's conclusions may be pulled down. Through comparing these time series studies, I am inclined to support the opinion that the effect of social security on savings is not significant.In the third chapter, this paper analyzes Feldstein's cross-sectional studies and his investigation into international comparison, compares his investigation into international comparison with the relative research of Barro and Macdonald. These studies of Feldstein confirm ulteriorly the conclusion of his time series that social security depresses private savings greatly. This is an easy thing because all of his studies are within the framework of his extended life cycle model whatever methods are used. in other words, he only proves the explainability of his extended life cycle model to the effect of social security on savings. Just because of this, he is easy to specify his equations with blinkers. For example, he is more likely to choose the ideas that can testify his theory in choosing and measuring variables. He is more likely to explain the estimates from econometric analyses according to his tastes. for example, he is inclined to deny the practical rationality of the regression results that are inconsistent with his hypothesis but doesn't provide a convictive explanation to the presence of these regression results. The Study of Barro and Macdonald draws a conclusion that the relation between social security and savings is dubious, which is very different from Feldstein's. They argue that the net effect of social security on savings is very little in a whole because the increase in social security wealth offsets the decrease in other wealth. Through comparing these two studies, it can be found that the specifications of equation is the most important reasons though there are many factors which play roles in. If considering many other cross-sectional studies that aren't analyzed in this paper, it can be found that the estimates in all studies are very different from that social security depresses savings, that social security has no effect on savings, to that social security increases savings. Moreover, these cross-sectional studies also face various econometric problems though they have some potential advantages compared with the time series studies that are discussed above. For example, it is difficult to generate the process of variations in social security benefit, which seems to be exogenous to observed or unobserved factors that determine households'savings; it is also difficult to obtain the information about households'expectation for social security benefit though it is necessary. This chapter also analyzes the simulation researches of Auerbach and Kotlikoff and some studies about the effect of private pension on savings. Auerbach and Kotlikoff explain to a great extent the reasons why there are so different conclusions in previous studies (i.e. errors in specification, specially unclean data ), and raise a new idea to analyze this question--simulation research. But their simulated data is not closer to the reality though it is cleaner. And their studies also don't provide a convictive explanation to whether and how much social security depress savings. The studies about the effect of private pension on savings are also far from an end. But the focus here is whether private pension increase private savings because there is a rough agreement that private pension doesn't decrease private savings.In the fourth chapter, this paper analyzes experientially the effect of social security on savings by using the relative data about 12 countries. The correlation analyses show that social security is negative correlative to savings, but isn't statistically significant. The time series analyses about individual countries show that the effect of social security on savings is very different in different countries, and is also not statistically significant. The regression analysis of international comparison shows the positive effect of social security on savings, but it is not too great. So the experiential evidences in this paper support the argument that the effect of social security on savings is not significant rather than the hypothesis that social security depresses savings greatly.All in all, there are two main reasons for the inevitable divergences in the studies about the effect of social security on savings: firstly, any theory only can cover a part of the behavior of people's consumption or savings generally because this behavior is very complex; secondly, because there are intense personal tastes in the specifications of equation, the choices of data, investigative methods and so on, these experiential studies can not form an authoritative conclusion. However, considering that different models or theories may be applicable to different teams of people, this paper can study the net effect of social security on private savings through synthesizing various theories. Just based on this, this paper draws a conclusion that the net effect of social security on private savings is not significant. The research on international data in this paper also supports this opinion. However, the writer doesn't want to oppose the reforms on pay-as-you-go systems for this, and only want to say that pay-as-you-go social security systems should not be criticized only because of this. The core of all the problems of pay-as-you-go systems should be the problems such as the financial imbalance emerges under the circumstances of the aging population, so the reforms on social security systems ought to mainly solve these problems.