Volume 9 Issue 4 - October 2008
Researcher Q & A
Q & A with John Karl Scholz and Ananth Seshadri
What motivates your longstanding research interest in retirement resource adequacy?
You can’t help but get the impression reading the financial press – whether Fortune or Forbes Magazines or even reporting in the Wall Street Journal – that Americans are preparing poorly for retirement. This view is supported by anecdotal evidence (we see a lot of activity at the shopping malls) and by low observed personal saving rates, as measured by the National Income Products Accounts (NIPA) ratings. So the prevailing wisdom is that people aren’t saving enough. This is also manifest in policy discussions. There are many tax preferences granted to saving, presumably to help people prepare adequately for retirement, for example, 401(k)s, IRAs, and other pension plans. There is always some interest in trying to expand these tax preferences, generally justified by calling attention to the precarious state of Americans’ retirement planning. One of us (Karl Scholz) did some earlier work with William Gale looking at a period (1981-86) when tax-favored IRAs were available to all taxpayers reagrdless of income levels. Gale and Scholz found that these incentives did very little to increase household saving. The money that went into these accounts was largely money that would have been saved otherwise or money that was shifted from one tax-favored account to the tax-favored IRA. So one motivation for our interest in the degree to which Americans are preparing well for retirement is that it ties into tax policy debates about the wisdom of expanding tax preferences for saving.
We began working with our student, and now friend and Urban Institute colleague, Surachai Khitakatrkun using the rich data provided by the Health and Retirement Study (HRS). We saw an opportunity to take a new approach to assessing the degree to which people are preparing well for retirement by taking seriously the implications of an augmented life-cycle model. The intuition of the life-cycle model is that households want to try to try to equate the discounted marginal utility of consumption across periods, meaning, loosely, that they try to equalize the well-being they receive from the last dollar they spend each period of their lives. We augmented the simple life-cycle model to account for the possibility of uncertain earnings, uncertain health shocks in retirement, and uncertain life expectancy. We looked at the implications of that model for what households should be doing to prepare for retirement. Specifically, we used the model to evaluate what households should be saving (and consuming) under any outcome given the three types of uncertainty that we can account for in the model: lifespan, earnings, and health shocks.
Once we have the decision rules for households that tell us what they optimally should do, we can compare that to what they actually do. Through procedures established at the University of Michigan and the Social Security Administration, we were able to get actual earnings realizations that households in the HRS had. With the optimal decision rule for any realization of earnings and the earnings that households actually received, we can calculate the household’s optimal consumption (and saving) in each period, which then allows us to calculate their optimal wealth holding at the time we observed them in the HRS in 1992. Given the assumptions of the model (most importantly, the coefficient of relative risk aversion, discount rate, and the real rate of return), these optimal targets are the amount of net worth – excluding Social security and defined benefit pension wealth – that the household should have accumulated to maximize its lifetime well-being.
When we do that exercise, we find two very interesting things. The first is that the HRS cohort born between 1931 and 1941 is overwhelmingly at or above their optimal wealth target. That is a surprise given the gloom and doom scenario commonly portrayed in the financial press. The second is that the outcome of that model fits the data remarkably well. This is somewhat at odds with at least one current of thought in the literature that suggests people’s behavior is not forward-looking, as our behavioral model would suggest.
Are younger people saving less?
A question arises as to whether it is just that cohort – born and raised during the depression – who is saving well. Is this behavior somehow unique to them? There’s a strong presumption out there that younger people are behaving less responsibly that their elders. So the work we currently report on looks at a broader range of age groups in the HRS. This is where the value of the HRS is really manifest. We were able to compare across several birth cohorts to see if the results from our earlier work still hold. We looked at two generations born before the original HRS cohort, the original HRS cohort, and two generations born after, the so-called War Babies (born 1942 to 1947), and the Early Baby Boomers (born 1948 to 1953). And our benchmark year is 2004, rather than 1994.
When we examine the implications of the augmented lifecycle model for this broader set of households, we find that our earlier results still hold. Households in every birth cohort were overwhelmingly at or above their optimum wealth targets. For the youngest group, the numbers were slightly less favorable. But even so, only 10% of that cohort fell short of their optimum target. We think the evidence is quite strong that Americans born before 1954 were accumulating sufficient wealth to maintain their accustomed standard of living in retirement, at least as of 2004. It remains to be seen how the financial market disruption in 2008 will affect our views.
What impact do children have on household savings?
The kids puzzle derives from the juxtaposition of a number of observations. First, our own analysis suggests that Americans are largely on target for achieving retirement living standards that are consistent with their pre-retirement living standards. Second, a casual look at data on asset balances (whether from the HRS data or data from the Survey of Consumer Finances) suggests that people have not accumulated very much financial wealth. Third, conventional financial planning advice suggests that replacement rates - the fraction of pre-retirement income that should be replaced by retirement income - should be anywhere between 75 and 90 percent. Those three observations do not square up very well, and the question is, what’s going on? In addition, the distribution of wealth is much more dispersed than earnings. Explaining this discrepancy has been a challenge in the field.
It turns out that children have an important and underemphasized role in understanding the distribution and dispersion of net worth in our society. We can illustrate the finding by giving the example of two couples, who are otherwise identical, but one has five kids and the other has no kids. The couple with five kids is eating peanut butter, and the couple with no kids is going to restaurants. That has really important implications for accumulation of retirement resources. It takes far fewer resources to maintain the living standards of the husband and wife with the five kids after the kids have left the house than it does for the couple with no kids. Children literally eat up a large share of household resources. Once they have left the nest, the wealth needed to maintain the living standards of the adults remaining in the household is substantially less than what is needed for the childless couple.
To study carefully the effects of children, we extended the model we had developed by incorporating endogenous fertility, which means we actually model the decisions people make to have children, in addition to the life-cycle consumption problem. We do that because some of the policy experiments that we’re interested in running, like altering the transfer system, may have implications for fertility.
Where are you headed next?
There are a lot of questions raised by our research to date that we find really fascinating. Rather than setting an arbitrary retirement wealth target, our research derives model-based estimates of optimal wealth targets. However, it’s very hard, outside the model, to translate the results to something that might be useful to people. We are interested in translating these model-derived wealth targets into a sensible set of rules that people can use when thinking about their retirement planning.
Farther off in the future, we are interested in trying to evaluate the impact of rising health care costs on personal saving. You might expect to see a higher level of precautionary saving as people anticipate large out-of-pocket expenses related to health shocks late in life. There’s quite a bit more to do to look systematically at how health shocks can affect people’s consumption and saving behavior.
John Karl Scholz is a professor of economics at the University of Wisconsin - Madison, where he specializes in public economics. In 1997-98 he was the Deputy Assistant Secretary for Tax Analysis at the U.S. Treasury Department, and from 1990-91 he was a senior staff economist at the Council of Economic Advisors. He directed the Institute for Research on Poverty at UW-Madison from 2000-2004. Professor Scholz has written on the earned income tax credit and low-wage labor markets. He also writes on public policy and household saving, charitable contributions, and bankruptcy laws. In 2007, he received the TIAA-CREF Paul A. Samuleson Award for Outstanding Scholarly Writing in Lifelong Financial Security (with Ananth Seshadri and Surachai Khitakatrakun).
Ananth Seshadri is a professor of economics at the University of Wisconsin-Madison, where he specializes in Macroeconomics and public finance. He has written on the causes and consequences of demographic change and the effects of technological change in accounting for various demographic patterns. His research has appeared in leading economics journals, including The American Economic Review, Review of Economic Studies, and the Journal of Political Economy. He was awarded a research fellowship from the Alfred P Sloan Foundation in 2006.