Table of Contents
• Supplementary materials to support book chapters
• Progress in demographic analysis methods
• Updates and corrections
• Send your suggestions to improve this site
© 2013 Leroy O. Stone. All rights reserved.
Why we Should Worry about Risk Awareness Deficits in the Older Population Concerning Future Out-of-Pocket Health Care Costs
by Leroy O. Stone
Millions of baby boomers would acknowledge the importance of taking actions designed to (a) maintain or enhance their best available health status, or (b) slow down the rate of loss of basic functional capacity (ability to do basic activities of daily living), or ( c ) achieve good levels of physical and mental fitness. (In the text below I will use the word "health" to mean variables in any of these three overlapping domains -- health status, functional capacity and fitness.) And most people do take some useful actions, so as to manage health-related risks.
But it is important to go beyond taking actions. In addition, we should do systematic thinking about networks of interrelated risks and the issue of how best to harness one's inevitably limited resources toward achieving the optimal set of risk management strategies and tactics. Many who accept this perhaps controversial dictum would be troubled to read the following remark found in a recently published paper: "… only a fifth of all respondents estimated that adverse health experience could lead to a more than 50% increase in out-of-pocket costs. To oversimplify, some people know costs; few know risk." This quote is from Allison K. Hoffman and Howell E. Jackson, "Retiree Out-of-Pocket Healthcare Spending: A Study of Consumer Expectations and Policy Implications" (American Journal of Law & Medicine, 39 (2013): 3-72).
Using the Rand American Life Panel, Hoffman and Jackson obtained data from over 1700 survey respondents (henceforth called "the sample") including a high proportion who were pre-retired or already retired. The authors state that "This study is the first to examine what people understand about their likely future healthcare spending ... ." It is the first one which I have seen that queries how respondents envisage their retirement-related risks beyond the area of financial well-being.
How people conceive of their retirement-related risks is a major topic irrespective of the fact that in some way they are already doing some amount of risk management. It is a major topic for at least two reasons: (1) the consequences of ignorance in some important risk domains, and (2) the consequences of failing to allocate risk management resources in ways that maximize the affordable protections (put simply, there may be a tendency to put too much of the available resources into a particular area of risk).
Rand Corp. has generously made available to researchers their microdata files from surveys that use its American Life Panel. Using the pertinent file, I have studied the demographic structure of the sample used by Hoffman and Jackson, and it seems to be a cross-section of internet users with an unusually large percentage who have at least a high school diploma.
Particularly worrying is the authors' finding that the performance of women was not as good as that of men. They state that "women in our survey showed a dramatically less acute understanding of future out-of-pocket expenditures than the men in our survey with regard to lifetime spending estimates. Even similar estimates given by men and women would be reason for concern in light of the evidence that women spend significantly more than men on out-of-pocket healthcare over the course of retirement." In effect, women's risk management in later life should be decidedly better than that of men.
There is supporting information in "Key Demographics in Retirement Risk Management", where Chapter 8 states that "In pursuing the identification of distinctive population segments, we found that women are over-represented in the lower-performing key demographics." In Chapter 5, Stone and Rainville state that " ... one would expect a substantially greater attention by older women themselves to the tasks of addressing their greater later-life financial risk [and researchers should find] a systematic tendency towards a higher rate of utilization of professional financial assistance among women. Why does this tendency fail to appear in several studies?"
With this concern in mind, and noting the special focus in the Hoffman and Jackson survey upon respondents' risk awareness in the area of future health care costs, I used the survey questions to devise a composite indicator of how well respondents were aware of certain later-life risk issues linked to their future out-of-pocket health care costs. This composite indicator draws upon the survey questions that dealt with self assessed familiarity with insurance programs, attention to personal budgeting activity, seeking the services of a financial planner, and adequacy of anticipation of future health-care expenses including awareness of major sources of variability of those expenses. Thus, our composite indicator focuses on aspects of risk awareness, and is limited by the questions asked in the Hoffman and Jackson survey.
Here is the distribution of the survey sample over levels of the indicator. (The original indicator values have been transformed so that the lowest is 0 and the highest is 100.) Because the data are unweighted, it is not being presented as an estimate of the pattern in the whole adult population.
The blue curve shows the pattern of the frequency distribution of respondents over scale levels, with each point being a moving average that includes its two adjacent neighboring points. The red curve is the cumulative distribution, which is computed from the un-smoothed percentages.
The sample is noticeably more heavily weighted below than above the index value of 50. For example, it is much fatter near the zero level than near the 100 level (the upper tail is much longer).
However, unfortunately, there is no basis for declaring what level of the indicator represents a useful boundary between adequate and inadequate performance. This distribution serves best as a benchmark against which to compare the distributions of key population sub-groups -- key demographics. Such a comparison is used when we try to determine what combinations of factors predict very good, or very poor, performance by a key demographic. ("Key Demographics in Retirement Risk Management" makes extensive use of this kind of comparative analysis.)
For what it may be worth, the median level of the indicator is 46, which means that half the sample scored less than 46. However, as already noted we lack a basis for declaring that 46 is a useful boundary between good performance and poor performance.
The preceding remarks were designed to set the context for the measurement of the gender difference in performance on the indicator. The next chart addresses this issue.
Due to the lack of weights for the sample responses, we need to present the male-female comparison in terms of male-female ratios as shown below. These ratios establish a statistical 'control' for the over-representation of women in the sample; but they do not address other kinds of over-representation -- e.g. that involving higher levels of education. Here is the chart that does the male-female comparison.
To make this chart, the whole sample was split into five performance zones whose boundaries were approximate quintile markers for the whole-sample distribution of indicator values. The zone of worst performance is that with scores of 27 or less, while that with best performance is the one with scores of 63 or more.
When a column ends at level 1.0 of the ratio, there is gender equality. (Note the red line at that level.) Women have the distribution with the greatest weight at the two lowest levels of performance; but the margin is slight. Men have the distribution with the greatest weight at the topmost level of performance. However, just below that level is an anomalous-looking "jump" in the ratio favoring women -- this may be due to small sample size.
In case sample size is the issue, we might lump together the second and third highest performance zones, to get a result where the gender difference would again be very slight, and probably favoring women. The net result would be to reinforce the idea that the distribution for women is slightly more heavily weighted at the lower end of scores, while that of men is more heavily weighted at the very topmost end. Keep in mind the word "slight", however, as this is generally what Hoffman and Jackson are reporting from their analysis which does not use a composite indicator.
A slight gender difference in various aspects of retirement related risk management is a recurrent research finding. In both Chapters 5 and 7 of "Key Demographics in Retirement Risk Management" and in the Hoffman and Jackson article, the authors argue that this finding represents an important challenge for public policy makers. Hoffman and Jackson suggest that insurance become a key route to be used to address the poor level of risk awareness in the pre-retired population.
It is also helpful to address some key determinants of poor performance in risk management. Every one knows that lack of the economic means to take appropriate risk management actions is a high hurdle in many families. And the more this factor is present, the greater is the need for professional guidance towards finding the optimal allocation of one's scarce risk management resources.
We should also take note of a "silent and mostly invisible elephant in the room" -- lack of basic relevant knowledge about the personal aging processes and the attendant risks and opportunities for the affected individuals, families and communities. There is much talk about how participation in educational courses seems to fail to lead to changed behavior, as Hoffman and Jackson note; but it does not follow that abject ignorance is bliss. Simple unawareness of key risk dimensions and their interrelations can have catastrophic results for persons. Similarly, even when a person delegates all the "hard thinking" to a professional advisor, it is still essential to have a basic understanding of key concepts in the discussion and be able to ask the questions that really matter for the person in question. Complacency about poor levels of relevant knowledge in a population leads to a dead-end. This point is particularly relevant to key population subgroups with relatively low social status or the economic means that impede ready access to "establishment solutions and services". They need to be very innovative about finding alternate ways to get at least some of the relevant support they need to address their pressing concerns.
These thoughts come to mind when I read the following poignant text in the Hoffman and Jackson paper: " ... work in psychology and behavioral law and economics suggests that risk is a concept that is generally not understood, and decisions with regard to uncertain future events are particularly prone to cognitive error." How is cognitive error to be fixed by taking the position that knowledge, and by extension mental fitness, do not matter?
One of the useful routes towards addressing crucial lack of knowledge is the University of the Third Age, where people from all walks of life can be suitably presented with fundamental and important knowledge which they sorely lack in the presence of later-life risk management challenges. And with today's technologies, a University of the Third Age can be managed at the community level, and that largely through online classes linked to periodic face-to-face meetings. This route seems especially pertinent in the context of the gender similarity in risk awareness and risk management performance, when in fact older women's challenges are substantially greater than those of older men.
Note: More blogs are coming in future weeks.