-
-
Email Event
-
-
Many argue that Americans are not saving enough for retirement. Indeed, some point to increasing household debt loads and voice concern that Americans may have to work longer or lower their standard of living. But how much is "enough" saving? This AEI event will feature John Karl Scholz of the University of Wisconsin who, with his coauthors Ananth Seshadr of the University of Wisconsin and Surachai Khitatrakun of the ERS group, has developed an economic model to be able to predict the saving that is optimal for each individual. The data on actual saving behavior that is included in the study allows the authors to explore—much more exhaustively than has ever been done before—the extent to which Americans save too little. Following the presentation of the study's findings, other experts will assess its implications.
| 9:15 a.m. |
Registration |
|
| 9:30 | Speaker: | John Karl Scholz, University of Wisconsin |
| Discussants: | Jagadeesh Gokhale, Cato Institute | |
| Jane Gravelle, Congressional Research Service | ||
| Moderator: | Eric M. Engen, AEI | |
| 11:30 |
Adjournment |
|
December 2004
Are Americans Saving Enough?
Many argue that Americans are not saving enough for retirement. Indeed, some point to increasing household debt and voice concern that Americans may have to work longer or lower their standard of living. But how much is "enough" saving? At a December 17 AEI event, John Karl Scholz of the University of Wisconsin presented an economic model able to predict the optimal saving for each individual--a model that he developed with coauthors Ananth Seshadr of the University of Wisconsin and Surachai Khitatrakun of the ERS group. The data on actual saving behavior that is included in the study allows the authors to explore--much more exhaustively than has ever been done before--the extent to which Americans save too little. Following the presentation of the study's findings, other experts assessed its implications.
John Karl Scholz
University of Wisconsin
The principle aim of this paper was to assess the optimality of household saving in America. We developed an augmented life-cycle model and used it to determine optimal household wealth observed in the Health and Retirement Study (HRS) dataset, which is a national panel study of 7,702 households (12,652 persons) collected in 1992. After making selected sampling exclusions, the dataset included 6,322 households. We then compared the results from the model to the observed wealth in the HRS.
After comparing the model to the HRS study, we concluded that the model was able to explain 83 percent of cross-sectional variation in wealth in households observed in the study. From the model, we were also able to conclude that 81 percent of households meet or exceed their optimal wealth targets. While this model does not account for all saving behavior, such as calculations for bequests, these findings do afford the following conclusions: the results strongly suggest the life-cycle model is able to predict life-cycle wealth accumulation, and changes in consumption are not determined by inadequate household savings. It is important to note that this data does not include the major stock market gains of the 1990s and that the results indicate that the model has more explanatory power than the seven other models examined in the paper. Given the results, it appears as though we are closer to being able to answer affirmatively when asked if Americans are saving optimally.
Jane Gravelle
Congressional Research Service
The life-cycle model is suspect insofar as it is a "super rational" model, raising the question of whether people really make the life-cycle calculations on saving behavior that the model predicts.
Indeed the model indicates that people are over-saving. This requires some exploration of certain factors that the paper does not address, such as bequests or the inclination of households to live in owner-occupied housing. Furthermore, the family size adjustment used in the paper should be reexamined and buttressed with more detail, such as providing tables that vary across different types of families.
Perhaps one of the largest concerns with this model is how it addresses the poor. The savings behavior of the poor is not driven by life-cycle considerations; rather, the poor are simply unable to save. This is not a calculation, as the model would assume, but a circumstance, one with considerable policy considerations for entitlement programs.
Several other issues need to be addressed: Social Security, an unfunded program, figures prominently in the model's calculations, yet does not generate wealth; and the role of defined benefit pension plans in the model's results should also be scrutinized.
Jagadeesh Gokhale
Cato Institute
It is important to note at the outset that this study is unrelated to the observed macroeconomic decline in national savings; this is a purely microeconomic study. This, however, does not necessarily diminish the policy relevance of the paper, such as to Social Security.
I am a bit ambivalent toward this study, because my own similar work differs, as do some other studies, with its final results. Moreover, there is the practical question of how households make these calculations.
This study has notable strengths. As a model it pushes the envelope in terms of sophistication. Moreover, it is rigorously defended and well tested against other models.
There are some other issues within the study that are worth examining more closely. For example, households were dropped from the dataset, some because a spouse would not answer the HRS survey. Is it possible that this could reflect a household dynamic, such as some form of discord? Are there other considerations of the marriage component that should be addressed or that could alter the results?
There are also biases of over-reporting and under-reporting of worth within the model. Whether these biases cancel each other out or if there is a net bias driving the results is difficult to say, but the issue should be examined.
Lastly there is an important policy implication to this study. Would the results of this paper hold up without programs such as Social Security? While this is a counterfactual that cannot be tested, what would happen if these programs were removed from the calculation?
Given the results of this paper, the only possible conclusion is that it "tends to show" that one of the rationales considered important for providing Social Security and saving subsidies might be suspect.
AEI staff assistant Gordon Gray prepared this summary.
