Retirement, Pensions, and Social Security.
Gary S. Fields and Olivia S. Mitchell
This little book has a most simple and inviting title that gives promise to the student of social security and employee benefits of a few hours of interesting, informative, and possibly innovative reading. Such, unfortunately, is not the case. Rather, the book is written for those who worship, either blindly or well-informedly, at the altar of mathematical statistics. To read the book thoroughly requires a knowledge of, and a faith in, such esoteric matters as multinomial logits, ordered logit models, regression models, intertemporal utility functions, reoptimization, intertemporal budget sets, plan-specific dummy variables, discrete choice models, etc.
At this point, the devotees of this cult may well be rising in righteous wrath and saying "Who is this ignoramus to be criticizing this glorious work which has been proclaimed to be `By far the most ambitious analysis of retirement behavior undertaken to date, -- a definitive synthesis of the existing research and a solid foundation for continued empirical analysis' by an eminent reviewer." This reviewer will first attempt to defend himself by setting forth his belief that the elegant methods of analysis are basically flawed when examined in the light of reasonableness of the procedures, including the basic data and underlying specific methodologies.
This reviewer does not doubt that the authors have excellently used their analytical tools. However, the real question is whether the material to which they applied their tools was worth the time and effort. Multiple regression analysis and similar elegant procedures can readily enchant the user to believe implicitly in the resulting findings, because after all, the results came out of a computer on reams of impressive printouts." The ultimate question that may be asked is how well multiple regression analysis would have performed in late September 1987 to project the Dow Jones Industrial Average for even only a few months.
But, after all, the "little boy" in the fable of the emperor's magic clothes will raise the question as to the validity of the data input, the methodology to which the data were subjected, and whether many unknown or unknowable elements of the long-distant future were ignored. To be more specific, let me cite some examples in the book that lead this reviewer to question the underlying procedures involved in the analysis and the resulting general conclusions reached. First, however, let it be stated that these conclusions -- which will be described briefly later -- are not too surprising qualitatively, if not quantitatively.
So, let us separate the wheat from the chaff (the mathematical-economics material, which is incomprehensible to the general student of social security). The authors present the following major conclusions:
(1) The present value of future income from work, private pensions, and Old-Age, Survivors, and Disability Insurance (QASDI) benefits increases as the age at retirement increases.
(2) In some private pension plans, the present values of the pensions decrease as the age at retirement increases -- because, unlike under OASDI, there are not "actuarial" reductions for early retirement (before age 65).
(3) Workers who have high present values of retirement income retire earlier than those who do not.
(4) Workers who expect to have higher present values of retirement and work income by deferring retirement do postpone retirement.
(5) About 75 percent of retirement decisions are based on economic factors and only 25 percent on health factors.
(6) The average person values a percentage increase in the present value of retirement and work income relatively less than the same percentage increase in leisure years in retirement. However, there is great dispersion in tastes about the mean, indicating great individual differences as to the importance attached to leisure versus income.
(7) A 10-percent reduction in the general level of OASDI benefits would increase the average retirement age by only 1 month.
(8) Decrease in the reduction factor for OASDI benefits taken at age 62 from the "actuarial" rate of 80 percent in present law to 55 percent would increase the average retirement age by 3 months. This was the ill-fated proposal of the Reagan Administration in 1981 -- and illogical too, because it would not have saved, over the long run, as much money as hoped for, since persons could still really retire at age 62, but not claim benefits until age 65 (thus avoiding the "actuarial non-bargain").
(9) Increase in the Normal Retirement Age (at which unreduced benefits are first available) from 65 to 68 would increase the average retirement age by only 1 1/2 months. (Note that present law, as a result of the 1983 Amendments, makes only a 2-year increase.)
(10) Increase in the Delayed-Retirement Credit of 3 percent per year of delay (non-compounded) to 6 2/3 percent would increase the average retirement age by only 1/4 year. (Note that present law, as a result of the 1983 Amendments, provides for the current 3-percent DRC to rise to 8 percent for those attaining the Normal Retirement Age after 2008.) The authors erroneously used a 6 2/3-percent DRC as being "the same as the early retirement factor" (page 15) -- i.e., the 20-percent reduction divided by the three years from 65 to 62; they should have used 8 1/3 percent instead -- the 20-percent increase compared with the 80-percent base benefit (or a 25-percent relative rise) and then divided by three.
The authors criticize the actuaries at the Social Security Administration for "assuming a 2.3-year response to a three-year increase in the normal retirement age" (page 123). Their criticism resulted from the failure to recognize that the proposal considered by the SSA actuaries was different than the one which they analyzed. The former increased both the NRA and the minimum early-retirement age, while the latter increased only the NRA. Because the SSA actuaries considered "retirement" only as the period when benefits are paid, it is only natural that a sizable change in the average retirement age would occur under the proposal which they were considering (e.g., a person who really retired at age 62 would be considered as retiring at age 65 under the proposal, as against being considered as retiring at age 62 under present law if benefits were claimed then).
This reviewer doubts the validity of the results of the authors as to the effect of increasing the Normal Retirement Age. The mechanistic procedures used cannot take into account significant changes in viewpoints and work philosophy that can occur when people eventually realize "magic" applies to age 65 as "the" retirement age.
Further, the reviewer is skeptical about the results of extensive, elaborate computations when flaws in the methodology are apparent and when important assumptions are not described. As an example of the latter, the authors do not describe the mortality basis of their present-value computations (a 2-percent real interest rate is properly used, thus adequately taking into account the OASDI COLA provision). An assumption of steadily-decreasing mortality rates should have been used (which would have resulted in difficult computational procedures); such practice is followed in the SSA actuarial estimates.
Two flaws in the methodology may be noted. First, the flawed benefit-computation procedure in the 1972 Act was assumed to apply, and be fully understood, by persons aged 60 in 1970; this reviewer doubts whether many covered workers have any good idea as to the level of the OASDI benefits, let alone this group knowing that they would get windfalls, and then they "planned according to the 1972 rules in computing their initial Social Security benefits" (page 37). Second, the method of computing OASDI benefits as used to evaluate different methods of reforming the program is seriously flawed when it states that "the bend points will increase with the consumer price index" (page 114); the changes from year to year (not necessarily "increase") are based on changes in nationwide average wages.
Reviewer: Robert J. Myers, Professor of Actuarial Science (Emeritus), Temple University.
|Printer friendly Cite/link Email Feedback|
|Author:||Myers, Robert J.|
|Publication:||Journal of Risk and Insurance|
|Article Type:||Book Review|
|Date:||Mar 1, 1989|
|Previous Article:||Pension Plan Management Manual - Administration and Investment.|
|Next Article:||Personal Financial Management, a Financial Planning Approach, 2d ed.|