Printer Friendly

Tax reform and retirement income replacement ratios.

Tax Reform and Retirement Income Replacement Ratios


Wage, or income, replacement ratios are of great importance to employers in the private sector, the Social Security Administrator, other federal and state government agencies, and legislative and regulatory leaders as they formulate plans and policies regarding the implementation or modification of retirement income programs. Employees and other members of the consuming public also have a keen interest in knowing the percentage of preretirement salary that is needed to maintain, or continue, their preretirement standard of living into their retirement years. Updated information on income replacement rates is specially critical to that segment of the population that, because of personal circumstances, currently is (or should be) contemplating savings and investment decisions relative to their own future retirement. Financial planners, accountants, lawyers, insurance professionals, and actuarial/benefit consulting firms also have a vested interest in research studies focused on the wage replacement issue since these advisors are frequently called upon to provide retirement planning and personal financial planning advice to their respective clientele.

With the enactment of the Tax Reform Act (TRA) of 1986, even greater significance is attached to whether derived wage replacement ratios are still appropriate. While the income tax provisions of TRA '86 will not be fully pased-in until 1990, the majority of the key changes as they relate to reductions in both number of marginal tax brackets and permissible itemized deductions were implemented for the 1988 tax year. Since TRA '86 involves such a radical departure from the old income tax laws, e.g., a reduction in both (1) individual tax rates and (2) the number of tax brackets from 15 (ranging from 11 percent to 50 percent) to three (15 percent, 28 percent, and 33 percent), it is imperative that there be a new examination of the income replacement issue where the new income tax provisions are incorporated into the determination ratios. It is probable that the wage replacement issue will need to be reexamined every time there is a change in the tax law as major as that emobodied in TRA '86.

In examining the wage replacement issue, this study defines the wage replacement level as the amount of retirement income that will maintain the individual's preretirement standard of living. This definition of wage replacement has received considerable support in recent years as the most appropriate measure for use in pension and retirement planning. The President's Commission (1981, p. 41) stated " . . . that individuals should be able to maintain their preretirement standard of living during retirement years. Retirees should not have to experience a sudden drop in their standard of living."

This study makes no attempt to analyze the relative merits of alternative income replacement definitions, in contrast to the standard-of-living maintenance approach selected for use here. In other words, no effort is made here to ascertain whether retirees could live comfortably on a level of income that is below the amount necessary to maintain preretirement standard of living (i.e., a "needs" approach). Absolute income adequacy standards in the form of needs-based wage replacement objectives are usually established at levels meeting minimum (basic) standards of income adequacy. Two such approaches are the Bureau of Labor Statistics retired couples' budgets and the federal government's official poverty level. n1 Finally, this study examines wage replacement rates through the analysis of savings data and consumer expenditure patterns which are factored into the replacement ratio calculations along with the new tax provisions.

Prior Research

Pension and employee benefit planning texts routinely discuss the concept of wage replacement ratios and retirement income objectives that employers should consider in their retirement plan decision-making processes. McGill (1989, p. 101) states:

If a pension plan is to fulfill its basic function, it must provide an income which, supplemented by OASDI benefits and other resources, will be adequate to maintain the retired worker and his dependents on a standard of living reasonably consistent with what he enjoyed during the years immediately preceding retirement. Otherwise, some employees may want to continue on the active payroll beyond the point at which they are still functioning efficiently, and the pension plan will fail of its purpose.

McGill (1989, p. 102) also indicates that because of (1) reductions in income taxes; (2) the elimination of Social Security taxes, retirement plan contributions (if any), and work-connected expenses; and (3) the probable completion of the children's education, the fulfillment of mortgage obligations, and any savings program, ". . . it can be demonstrated that a total retirement income of 60 to 75 percent of an individual's gross earnings at time of retirement will enable him to enjoy a standard of living that is reasonably commensurate with what he enjoyed during the latter stages of his employment." Consistent with commonly-held view of other experts in the pensions field, McGill notes that the highest replacement percentage (e.e., 75 percent) is applicable to employees at the lowest salary level, with this percentage declining as employees' salaries increase. Allen, et al (1988, p. 36) offer similar advice with respect to employees who have 30 years of service with the plan sponsor.

As explained below, there are no definitive studies on the subject of wage replacement rates where TRA '86 tax provisions, together with actual consumer savings and expenditure information, have been incorporated into the ratio determinations. Thus, employers are likely to experience great difficulty in determining whether their current pension benefit formulas are inadequate, overly generous, or proper.

Prior studies indicate levels of wage replacement ratios that might be used. Most use a standard-of-living maintenance approach in defining wage replacement ratios. One major study (President's Commission, 1981), undertaken by the President's commission on Pension Policy (PCPP), reported on a number of issues and current problems facing the nation's retirement systems. One part of this comprehensive review focused on the development of retirement income goals. The PCPP wage replacement ratios, derived separately for single persons and married couples (retiring in 1980), are presented below (pp. 42-43).

Because these replacement ratios were calculated using pre-TRA '86 tax rates, they may be totally inappropriate when used as target percentages for individuals retiring in 1988 and later years. In addition to taxes, savings rates varying from 3 percent beginning with a disposable personal income of $10,000 and increasing uniformly to 15 percent for a $ 50,000 income, and work-related expenses equal to a constant 6 percent of disposable income were factored into the PCPP wage replacement calculations for both single persons and married couples.

It appears that both the savings and work-related expense assumptions may be rather arbitrary. The PCPP ratios suffer from two other potential defects since the ratio calculations do not consider property taxes or possible changes in certain age-related expenditures that may occur during the retirement years.

Dexter (1984) incorporates federal, state and local income taxes, FICA taxeS, property taxes, savings rates, work-related expenditures, together with estimates of changes in age-related consumption items in the calculation of replacement ratios. Consequently, his study probably could be viewed as the most comprehensive treatment of the wage replacement issue to date. Unfortunately, when trying to apply the study results to persons approaching retirement today, the Dexter study suffers from several shortcomings. Most notably, since the study was completed in 1984, it was not possible for Dexter to consider the potential effects of TRA '86. Possibly equally as important, Dexter's study uses dated 1973 consumer expenditure data on age-related and work-related varibles. Because the 1973 data are now 15 years old, the replacement ratios computed in the Dexter study are not likely to be appropriate today at the same levels of preretireement salaries examined in his study. For example, a $25,000 salary (one of the preretirement income levels used by Dexter) in 1973 is approximately equivalent to a $60,000 salary in 1988, assuming a 6 percent annual salary increase.

A third limitation of Dexter's study is that the changes in work-related expenditures were derived by comparing the expenditure patterns of a group of consumers below age 65 (specifically, individuals between the ages of 45 and 54) with a group of consumers age 65 and over, without any control for a worker vs. retired variable. In other words, Dexter did not screen the age 45 through 54 group to cnofirm that all of the individuals in this group were, in fact, working. Similarly, there is no assurance that only persons who were actually retired comprised Dexter's 65 and over group. (2) It is highly probably that both a significant number of consumers in the 65 and over group were still employed, and several consumers in the age 45 through 54 group were retired (or unemployed for some other reason). This drawback limits the usefulness of Dexter's estimates on work-related expenditures. Despite these and several other limitations, the Dexter study constitutes a significant addition to the literature on wage replacement ratios. Further, its methodological approach and the data sources identified in the study were of substantial value in the development of the current research project.

Wage replacement ratios were derived in Dexter's study for married couples at four preretirement salary levels: $10,000, $25,000, $50,000, and $75,000. The gross wage replacement rates (p. 41) start at nearly 82 percent at the $10,000 level and decrease to approximately 40 percent at the $75,000 salary figure. Savings rates range from 4 to 17 percent and net changes in the combined age- and work-related variable range from under 4 percent to approximately 36 percent at the highest salary level. (3)

The age- and work-related variable is a major factor explaining the relatively low replacement ratios at the highest salary levels. Further examination shows that substantial increases in gifts and contributions are largely accountable for the combined age and work effects. It is not clear why such large amounts of giving were occurring for individuals at high income levels. Whether this behavior was related to estate tax or other reasons simply is not known. In any event, it is clear that the gifts variable has a great impact on Dexter's results. (4)

A recently published report on the retirement income issue incorporates TRA '86 provisions into the replacement ratio calculations (England, 1988). (5) This Booke & Company study provides an update to the earlier PCPP study released in 1981. A major conclusion of this recent study is that wage replacement ratios significantly larger than those previously assumed are needed at higher salary levels to maintain preretirement living standards.

The Booke & Company study and the PCPP report use essentially the same methodology (and assumptions). Thus, unfortunately, the updated study suffers from the same or similar limitations as those identified earlier for the PCPP report. Specifically, actual consumer expenditure data on savings, work-related, and age-sensitive variables were not incorporated into the replacement ratio calculations. Rather, somewhat arbitrary assumptions were made for the savings rates and work-connected expenses, with the complete absence of age-sensitive expenditures in the formula.

The Booke & Company study derived replacement ratios for preretirement income levels ranging from $10,000 to $200,000. These results are displayed in a bar graph in the published report. However, actual percentages are provided in the body of the report only at three salary levels. The England report (1988, p. 2) concluded that "... as income increases from $10,000 to $30,000, the replacement ratio needed to maintain the same standard of living drops from 85% to 68%. It then rises to 73% at $50,000 and remains roughly constant thereafter as income increases."

In summary, income replacement objectives are always expressed as amounts substantially below full replacement of the employee's final salary. This, of course, is explained partly by the presumption that taxes are lower in retirement. In addition, certain work-connected expenses are eliminated. Also, the common expectation (as shown above) is that wage replacement ratios for higher-income employees should be lower than those for low-income workers. However, within a given income level, single individuals, married couples with no children, and other family unit configurations may have somewhat different income replacement needs. Thus, it cannot be assumed that a range of wage replacement ratios, where such range varies only according to preretirement income level, will be appropriate for all types of retiree (or consumer) units.

New, updated wage replacement ratios, derived from consideration of both the new federal income tax legislation and recent consumer expenditure data, should provide better standards as employers continue to evaluate their target benefit objectives. This study's research methods are described next. Findings and conclusions follow.

Research Methods

A major aspect of wage replacement determination concerns the formula, or calculation method, that should be used. Several methods are possible, with some approaches more complex than others. One approach to the calculation of replacement ratios is to express the ratio as a function of two factors: (a) the difference between pre- and post-retirement taxes including federal, state, and local income taxes, property taxes, and FICA taxes, and (b) preretirement savings (i.e., preretirement savings amounts would not be replaced and, therefore, not be available to retirees for expenditure purposes). A formula representing this basic type of replacement ratio calculation could be expressed as follows:

Replacement Ratio = PrRGP - PrRT - PrRS + PoRT/PrRGP

where: PrRGP = Preretirement Gross Pay PrRT = Preretirement Taxes PrRS = Preretirement Savings PoRT = Post-retirement Taxes

This formula could be expanded by subtracting work-related expenses in the numerator of (1) above. A more complete specification, however, would include both work-related and age-related variables in the numerator of (1). Age-related variables, with expenditure patterns that change between the pre- and post-retirement periods, could include such items as health care, housing, food, gifts, and possibly others. Collectively, the individual age-related variables could be combined into a single variable that measures the net change in age-sensitive expenditures. When both work-related and age-related variables are included, the replacement ratio formula could appear as follows: (6)

Replacement/Ratio = PrRGP - PrRT - PrRS - WRENCASE + PoRT/PrRGP

where: WRE = Work-Related Expenses NCASE = Net Change in Age-Sensitive Expenditures PrRGP, PrRT, PrRS, and PoRT are defined above

Replacement Ratio Defined

Wage replacement rates are calculated in this study using both formulas (1) and (2) above. This approach provides users with the flexibility to choose the formula that best coincides with their own philosophy. Formula (2) should be chosen when it is desired that age- and work-related expenses be incorporated into the replacement ratio calculations. In contrast, formula (1) should be used by plan sponsors whose philosophy is that differences in expenditure patterns should not be taken into account in establishing retirement income objectives. It must be remembered that both formulas implicitly assume that the replacement ratio is defined as the percentage of preretirement gross salary that will enable retirees to maintain the same standard of living which they enjoyed prior to retirement.

Since the replacement ratio formulas are partially a function of federal income taxes, the results of previous studies (i.e., those studies based on prior income tax laws) are of limited value. This study incorporates the tax law changes contained in TRA '86, and, therefore, the results should be of greater significance in today's retirement planning process. In calculating ratios, this study assumes that retirement occurs in 1988. Thus, the federal income tax rates, together with the FICA tax rates and Social Security taxable wage base, in effect in 1988 are incorporated into the calculations.

Gross Salary Levels

Wage replacement rates are calculated for eight levels of preretirement gross pay (PrRGP): $15,000, $20,000, $25,000, $31,250, $40,000, $50,000, $60,000 and $80,000. These dollar amounts represent the median values of their respective salary ranges and are shown in Table 1.

Consumer Savings and Expenditure Data

Since replacement ratios calculated under both formulas (1) and (2) are a function of pre- and post-retirement taxes and preretirement savings, estimates of these individuals variables must be determined. In addition, since taxes include federal, state and local income taxes, property taxes, and FICA taxes, separate values for these parameters also must be determined. Further, for those individuals who itemize deductions on their income tax returns, estimates of these deductions must be obtained prior to the calculation of the tax itself. Formula (2) also requires estimates of the various components of the work-related and age-related expenditure items.

Federal income taxes and FICA taxes for 1988 are calculated directly for the eight salary levels, with one exception relating to itemized deductions. Estimates for itemized deduction amounts used in calculating federal income taxes for itemizers, together with estimates for state and local income taxes, property taxes, savings amounts, work-related expenditure items, and age-related expenditure variables, were all derived from data contained in the Bureau of Labor Statistics (1986) public use tapes of the 1984 Consumer Expenditure Survey (Interview Survey). (7) The 1984 survey was the latest survey that had been released to the public when ratios were calculated for this study.

Matched Datad: For each of the eight salary levels, the expenditure patterns of two groups of consumer units are compared and the differences calculated for each expenditure variable. (8) These two groups of consumer units were derived according to certain characteristics of the "references person." The reference person is defined as:

The first member mentioned by the respondent when asked to "Start with the name of the person or one of the persons who owns or rents the home." It is with respect to this person that the relationship of other consumer unit members is determined (p. 47).

Thus, the reference person might be referred to as the head-of-household.

The two groups of consumer units are classified as follows: (1) the worker group consists of consumer units whose reference person was working (not unemployed, not retired) and was between the ages of 50 and 64 at the time the survey was conducted; and (2) the retiree group consists of consumer units whose reference person was retired and between the ages of 62 and 74 at the time the interview was conducted. These two groups are used for the purpose of contrasting the expenditure behavior of working consumer units whose reference person is nearing retirement with expenditure behavior of consumer units whose reference person is already retired. The specifications used in defining the two groups resulted in income and expenditure information on 2,544 working units and 1,217 retired units.

Table 2 shows the breakdown of the various consumer units according to preretirement salary range. Although some of the cells contain small numbers (particularly in the retireee groups at higher incomes), this may not present a significant problem because of the matching process used in this study. The data are such tthat a specific group of workers, at a given salary level, is likely matched with a retiree group that has a lower level of preretirement income.

Because working groups are matched with retiree groups, this study can be characterized as a cross-sectional analysis of pre- and post-retirement consumption patterns. The working households were not followed for a number of years (i.e., longitudinally), with subsequent examination and comparison of their post-retirement expenditure patterns with the same household's preretirement consumption and expenditure patterns. Rather, two clearly distinct groups of workers and retirees are examined. Differences in their expenditure patterns during the same period (i.e., 1984) are used in deriving the wage replacement ratios in this study. In performing the necessary analysis, subsets of the retirees were matched with subsets of the workers to derive estimates of the changes in expenditure patterns that occur in moving from a working status to a retired status. For each working group, the matching process was accomplished by comparing that group's spendable income (i.e., salary less preretirement taxes and savings) with a specific subset of retired households that had the same amount of spendable income.

In attempting to measure changes in taxes and expenditures as individuals move along their life cycles from a working to a retired status, a longitudinal study is the preferred methodological approach. However, this type of undertaking would require that numerous households and their expenditure patterns be followed over a lengthy period, probably at great cost to the surveyor. Unfortunately, the necessary time and sufficient resources were not available here to undertake a study of this type. Further, the nature of the sampling process employed by the Bureau of Labor Statistics (1986) does not permit the use of a longitudinal approach. Consequently, a cross-sectional approach is the only approach possible in this study since it relies on the BLS Consumer Expenditure Survey data.

Federal, State and Local Income Taxes: As previously indicated, federal income taxes and FICA taxes are determined according to the tax provisions in effct for 1988. For example, preretirement Social Security taxes are determined by multiplying the 1988 FICA tax rate of 7.51 percent by earnings up to the 1988 taxable wage base of $45,000. Further, the new federal income tax rates of 15 percent and 28 percent are used. The preretirement salary levels, together with the itemized deduction amounts derived in this study, are such that the 5 percent surcharge is not a factor in the federal income tax calculations. (9) This surcharge would have played a role in the calculations if higher levels of preretirement salary had been assumed. Finally, the itemized deductions computed from the Bureau of Labor Statistics (1986) data were derived by adding together only those gift and expenditure items whose deductibility is retained under the new law.

Consumer interest and certain other expenditure items, although deductible under prior federal income tax law, are not included as itemized deductions here. It is recognized that a portion (i.e., 40 percent) of consumer interest payments can still be taken as an itemized deduction in 1988. However, since the deductibility of consumer interest will be fully phased-out by the end of 1990 (according to TRA '86), these payments were treated as nondeductible items here. Although studies of the type performed here should be updated periodically in any event, this study's findings may retain their value longer due to the total exclusion of consumer interest payments in the determination of itemized deductions. (10) The drawback (or limitation) of this approach is that the total exclusion of consumer interest deductions may cause the wage replacement ratios to be slightly flawed. It is expected, however, that any changes in the ratios created by the absence of consumer interest deductions in the federal income tax calculations are minimal because (1) under the law only 40 percent of the payments are deductible; (2) the interest payments are deductible only for itemizers (i.e., individuals at the higher salary levels)--thus, fewer than the full eight salary levels are affected; and (3) only the differences in pre- and post-retirement taxes are of importance in determining the replacement ratios, and the treatment of consumer interest is the same in both sets of tax computations.

It was impossible to account for all the changes in the individual state and local income tax laws that occurred in response to the enactment of TRA '86. Thus, it became necessary to make certain assumptions in the calculation of both pre- and post-retirement state and local income taxes. In the determination of preretirement taxes, state and local income taxes were assumed to constitute the same percentage of the 1988 federal income taxes that the 1984 state and local taxes (for the worker group) comprised as a percentage of the 1984 federal income taxes, as determined from the Bureau of Labor Statistics (1986) data. A similar assumption was made in the calculation of post-retirement state and local taxes; that is, the ratio of post-retirement state and local income taxes to 1988 post-retirement federal income taxes was set equal to the 1984 ratio of state-to-federal income taxes (for the retiree group). (11) The calculation of post-retirement income taxes was further complicated at the higher salary classes because a portion of the Social Security benefits becomes taxable. (12) Since the available Social Security benefits are used to satisfy part of the needed post-retirement wage replacement, the taxation of these benefits must be factored into the income tax calculations. (13)

Property Taxes: Although estimates of property taxes can be derived from the Bureau of Labor Statistics (1986) data, these estimates are not incorporated into either formula. Upon matching the working and retiree groups according to levels of income and consumption, it was discovered that for each level of preretirement salary the amount of property taxes was actually higher for the retiree group. This result is totally inconsistent with state and local property tax laws that provide additional exemptions (and, therefore, lower property taxes) for the elderly. (14) Thus, it became necessary to eliminate the property tax variable from the replacement ratio calculations in both formulas. (15) A possible explanation for the unexpected tax results is that retirees had much higher levels of preretirement income than the workers with whom they were matched. If this is true, the retirees likely would have had more expensive homes. As such, even with reduced property taxes at retirement, the retirees' property taxes might still be higher than those of their matched worker counterparts. The property tax results obtained here are probably a direct outcome of using a cross-sectional approach. A longitudinal study likely would not suffer from this drawback.

Theoretically, the inclusion of a property tax variable should lead to a reduction in property taxes during the post-retirement period. The amount of such reduction would not have to be replaced in determining the appropriate level of income needed during retirement to maintain a person's preretirement standard of living. Consequently, the inclusion of a property tax variable should result in a somewhat smaller wage replacement ratio. Conversely, the exclusion of the property tax variable in this study causes the wage replacement ratios to be somewhat overstated, possibly by as much as 1 or 2 percent at the lower salary levels, with probably smaller overstatements at higher levels of preretirement salary.

Savings Rates: Because of its inclusion in the wage replacement ratio formulas, a savings variable must be defined so that estimates can be derived. Unfortunately, savings is probably one of the most difficult economic variables to quantify because there does not appear to be a consensus on how savings should be defined. The BLS does not compute a savings variable as such. Rather, a variable called "net change in assets and liabilities" is calculated. This variable is broader in scope than most savings variables, including the savings variable chosen for use in this study. If the net change in assets and liabilities variable had been used here, much larger estimates of savings would have resulted.

In estimating savings, this study computes a basic savings and retirement contributions variable. This variable includes the following items contained in the Bureau of Labor Statistics (1986) public use tapes: (a) net acquisition of stocks, bonds, and mutual funds; (b) net investment in a farm or business; (c) net changes in savings and checking accounts; (d) net changes in U.S. savings bond holdings; (e) net changes in money owed the consumer unit; (f) amounts received on surrender of insurance policies; and (g) contributions made by the consumer unit to retirement plans. (16)

Certainly, the choice of savings variable can greatly affect the replacement ratio calculations since none of the preretirement savings amounts are replaced in the formulas. In other words, the amount of preretirement savings creates a dollar-for-dollar reduction in the amount of post-retirement aftertax income needed to maintain an individual's preretirement standard of living. Specifically, the higher the preretirement savings rate, the lower the needed wage replacement ratio. The actual savings rates derived in this study are described later.

Work- and Age-Related Variables: A primary difference between this study and many other studies examining the wage replacement issue is that, in addition to tax and savings variables, it also examines the extent to which work- and age-related expenditures actually differ between the pre- and post-retirement periods. Other studies frequently made simplifying, arbitrary assumptions about these variables. As described above, these diffeences are estimated in this study by comparing expenditures of the age 50 through 64 working group with expenditures of the age 62 through 74 retired group. The specific set of age- and work-related expenditure variables examined are listed in Appendix A.

Reductions in work-related expenditures may arise because the individual is no longer working in the job he or she occupied prior to the time of retirement. It might be presumed that total transportation and food costs would be reduced due to the absence of commuting costs and a decrease in lunch costs. Also, certain other job-related expenses such as special articles of clothing and professional or union dues would be reduced or eliminated. Certain age-related expenditures (e.g., health care expenditures) may increase due to advancement in age or reduction in health insurance coverage. (17) Other age-related expenditures may decrease (e.g., education) because the couple's children may have left the household.

In summary, while work-related expenditures should decrease in the retirement period causing a reduction in the needed wage replacement ratio, total age-related expenditures are harder to predict a priori. Certain age-related variables may show an increase from the preretirement to the post-retirement period; other age-related expenditures may show a decrease. In some cases, it was difficult to properly classify a particular expenditure variable as either work-related or age-related because it possessed elements pertaining to both categories. Consequently, in this study the changes in work- and age-related expenditures are aggregated for purposes of presentation in the tables.

The inclusion of a gifts variable in the analysis may seem somewhat unusual. However, as described previously, the Dexter (1984) study concluded that there was a significant giving away of wealth in later years by individuals in the higher income categories examined in his study. This finding led to substantial reductions in wage replacement ratios at the higher salary levels since no portion of preretirement giving was replaced in Dexter's ratio calculations. A gifts variable was included for analysis in this study (as part of the combined age and work variable) to determine whether Dexter's conclusion still applied to consumer expenditure behavior in 1984.

Weighted Least Squares: FICA taxes are calculated simply by multiplying the preretirement salary (up to the taxable wage base) by the FICA tax rate. Federal income taxes (excepting itemized deductions) are also calculated directly, according to the tax provisions in effect for 1988. However, calculations for other variables are derived from the Bureau of Labor Statistics (1986) data. These variables include savings, state and local income taxes, itemized deductions used in the calculation of federal income taxes, and expenditure items comprising both work-related and age-related variables. As is typically found when computing estimates from sample data, the raw datapoints calculated for each of the variables examined exhibit a certain amount of unevenness when plotted as a continuous line. To achieve smoothness in the data, in moving from lower salary classes to higher income categories, a weighted least squares line was fitted to the raw values for each variable. (18) The weights used in fitting the line to the raw datapoints are the numbers in the appropriate samples in Table 2. Further, when relatively few sample observations were present (e.g., at the $60,000 and $80,000 salary levels in some instances), the weighted least squares line was fitted only to the datapoints for the remaining salary classes. Values for the omitted pay categories were then extrapolated from the weighted least squares line.

Retirement Assumptions

This study makes the following assumptions pertaining to retirement: the consumer unit for which retirement is presumed to occur is a married couple; only one of the spouses has been previously employed for any significant length of time; the spouse is age 62 at the time of the breadwinner's initial retirement at age 65; and there are no other persons, including children, who reside with, or are financially dependent upon, the married couple. These assumptions affect the wage replacement calculations in three ways. First, the amount of federal income taxes is a function of the respective ages of the breadwinner and spouse, together with the number of other dependents in the family (consumer) unit. Specifically, an additional $600 increase in the standard deduction is available for each married person who is age 65 or over, and there is a personal exemption of $1,950 available for each individual (including dependents). Second, for those preretirement salary levels in excess of $45,000 (1988 taxable wage base), the amount of preretirement FICA taxes would be higher if a two-wage-earner assumption had been made. Third, the one-wage-earner assumption, together with the assumptions about dependent children and the ages of the breadwinner and spouse, all are factors which determine the amount of retirement benefits payable under Social Security.

The decision to limit the study to a one-wage-earner family unit was made for the following reasons. First, the study assumes that retirement occurs in 1988. It is believed that a large segment of today's retirees is still comprised of the traditional one-wage-earner family configuration. While many females (whose husbands are now nearing retirement) may have been employed outside the home at some point in their lives, these women may have a sporadic employment pattern with a history of relatively low earnings. Any retirement benefits for which they qualify based on their own earnings history probably are minimal. It is anticipated that this situation will change dramatically in the future as greater numbers of women continue in the labor force, and as Congress reduces the maximum length of time for vesting under qualified retirement plans. A second reason is that this study attempts to develop wage replacement information that will be useful to employers as they contemplate possible changes in their retirement programs. It is believed that most employers are interested only in the replacement of their own workers' preretirement salaries. For practical reasons, a firm may have no interest whatsoever in establishing wage replacement objectives based on the combined income of its employees and their spouses.


The stated purpose of this study is to determine new wage replacement ratios at various levels of preretirement salary. The findings of this study are presented below separately for the two wage replacement models.

Tax and Savings Model

Table 3 displays the wage replacement ratios derived under formula (1). Lines 11, 12, and 13, respectively, show the gross replacement ratios, replacement rates (and dollar amounts) accounted for by Social Security benefits, (19) and the resulting wage replacement differences [line 10 (or line 11) minus line 12] that must come from either individual employee savings or from the employer's retirement plan. It is observed that the gross wage replacement ratio (line 11) hits a peak of 82 percent at the lowest preretirement salary of $15,000. The ratio then declines to 72 percent at the $31,250 salary level, increases only slightly to 73 percent for salaries between $40,000 and $60,000, and finally increases to 76 percent at the highest preretirement salary level of $80,000. (20)

Visually, the variations in the wage replacement percentages can be better observed by examining Figure 1 which plots all three wage replacement ratios: (1) the gross wage replacement ratio; (2) the Social Security replacement ratio; and (3) the remaining portion that must come from either the employer-sponsored plan or from individual savings. Figure 1 shows (as expected) that the portion of total wage replacement that comes from Social Security steadily declines as the preretirement salary level increases. Consequently, the remaining portion of wage replacement that must come from combined employer-sponsored plan benefits and individual savings increases significantly as the level of preretirement salary increases.

An important observation gleaned from examination of Figure 1 is the flattened U-shaped pattern of the gross wage replacement ratio curve. As seen below, a similar pattern exists for the comprehensive tax, savings, and expenditures model.

Tax, Savings, and Expenditures Model

Formula (2), presented earlier, is used to derive wage replacement ratios that take into consideration certain age- nd work-related expenditure variables, in addition to tax and savings data. Wage replacement ratios for the tax, savings, and expenditures model are displayed in Table 4. The gross replacement rates appear on line 12. The Social Security portion and the portion of wage replacement that must come from the employer's retirement plan and individual savings are presented on lines 13 and 14, respectively. The gross wage replacement ratio starts out at 82 percent at the $15,000 level, progressively decreasing (but almost leveling out) to 66 percent at the $50,000 and $60,000 salary levels, before climbing slightly to 68 percent at the $80,000 preretirement income level.

Figure 2 plots the various gross wage replacement ratios (line 12, Table 4). figure 3 shows the breakdown of the gross replacement rates between the Social Security portion and the portion from the employer plan and individual savings (lines 13 and 14 in Table 4, respectively).

Examination of line 7 in Table 4 discloses that the decreases in age- and work-related expenditures at retirement, although basically nonexistent at the lowest salary level, amount to approximately 6 to 8 percent of preretirement aftertax income at the higher salary levels. These expenditure reductions eventually lead to similar decreases in the gross wage replacement ratios, ultimately causing the gross replacement ratios under the comprehensive wage replacement model to be significantly below those ratios calculated under the tax and savings model, particularly at the higher salary levels. It is also observed that the gross wage replacement rates (as shown in Figure 2) exhibit a flattened U-shaped pattern-a pattern seen previously for the tax and savings model.

Table 4 (line 7) shows only the combined effects of the various age- and work-related expenditure variables analyzed in this study. In examining the individual component categories, it is discovered that the largest decreases in expenditures, in moving from a working to a retired status, occurred in housing, entertainment, apparel, education, and food (at home). Decreases in these expenditure categories were observed at the majority of the eight preretirement salary levels, although small increases for some of these expenditure variables were found at the lowest preretirement salary levels. It may be that many of the observed decreases in apparel, food (at home), and education are simply a direct result of children maturing, reaching adulthood, and leaving the consumer unit. It may also be that housing costs decrease due to an elimination of mortgage payments. However, other explanations for these phenomena are also certainly plausible. The transportation variable showed the largest increase (at most salary levels) in expenditures after retirement. At higher preretirement salary levels, it appears that many retirees may be spending substantial sums on vacations and travel. In contrast, transportation expenses actually decreased for retirees at the $15,000 and $20,000 salary levels, possibly reflecting both the elimination of commuting costs and insufficient income to enjoy lengthy and expensive travel. At the lowest and middle salary levels, total health care costs (including insurance) also

increased significantly in retirement.

Regarding the gifts variable, this study's findings do not support an earlier conclusion by Dexter (1984) that higher-income retirees give away a substantial amount of their wealth in later years. In fact, the expenditure data analyzed here show that gifts actually decreased after retirement. There may exist a rational explanation for this seemingly inconsistent set of results. As indicated earlier, Dexter found substantial post-retirement giving occurring at the $50,000 and $75,000 preretirement salary levels which are the two highest salaries examined in his study. However, his analysis was based on consumer expenditure data collected in 1973. Consequently, it can be argued that Dexter's results are not directly comparable to this study's findings since Dexter's $50,000 and $75,000 salary levels are equivalent, in 1988, to salaries far in excess of the highest preretirement salaries examined here. If the necessary expenditure data had been available at substantially higher income levels, this study also might have found high-income retirees giving away significant portions of their wealth.

Conclusions and Implications

Probably the most important finding of this study is that the calculated gross wage replacement ratios are significantly higher than generally has been assumed under prior federal income tax legislation. The gross replacement ratios for the tax and savings model, together with the ratios derived under the more comprehensive tax, savings, and expenditures model, are presented in Table 5. These findings are presented with the replacement ratios derived in the Dexter study (1984), the President's Commission on Pension Policy (PCPP) report (1981), and the recent Booke & Company study (England, 1988; Gibby and Dorer, 1988).

Table 5 shows that even when age- and work-related expenditures are considered, the wage replacement ratios in this study are generally still higher than those derived in the Dexter and PCPP studies. This is particularly true at the higher salary levels. Undoubtedly, this phenomenon can be at least partly explained by the increase in general salary levels over time. Other underlying reasons likely relate to changes in consumer expenditure behavior and the new federal income tax provisions contained in the Tax Reform Act of 1986. Table 5 also clearly shows the distinct differences in the wage replacement patterns between the two earlier studies and the current research project. Both the Dexter and PCPP findings support the long-held belief that gross replacement ratios decrease as a function of increasing preretirement salary levels. In contrast, the gross wage replacement ratio pattern observed in this study indicates a leveling out of the replacement ratios at higher salary levels, with a modest upturn at the highest salary level of $80,000. Earlier in this study, this pattern was described as being in the shape of a "flattened-U." The Booke & Company replacement ratios, when plotted, also exhibit a pattern of similar shape. It is likely that the similarity in the wage replacement patterns between this study and the Booke & Company study is explained by the income tax changes brought about by TRA '86. Both studies incorporated the changes in tax law. In contrast, both the Dexter and PCPP studies were conducted using prior income tax rules.

With the higher wage replacement ratios observed both here and in the Booke & Company study, it is imperative that employers, employees, benefit consultants, financial planners, and others who have an interest in retirement plan design and related financial (and retirement savings) issues reevaluate their wage replacement objectives. It is believed that the Tax Reform Act of 1986 has had a significant impact on both the magnitude of income replacement ratios and the shape (or pattern) of needed wage replacement as a function of increasing levels of preretirement income. It is no longer appropriate to assume that wage replacement ratios decrease universally as a function of increasing income. Rather, it has been shown that wage replacement ratios level out (or even increase) at higher salaries.

Appendix A

The following specific expense categories were included in the analysis of changes in expenditure behavior between the working and retired groups.

1. Age-Related

A. Health Care

(1) Health Insurance

(2) Medical Services

(3) Prescription Drugs and Medical Supplies

B. Shelter

(1) Owned Dwellings

a. Mortgage Interest and Principal

b. Maintenance

c. Repairs

d. Insurance

(2) Rented Dwellings

(3) Other Lodging

(4) Utilities

a. Natural Gas

b. Electricity

c. Fuel Oil and Other Fuels

d. Telephone

e. Water and Other Public Services

(5) Household Operations

a. Domestic Services

b. Other Household Expenses

C. Entertainment

(1) Fees and Admissions

(2) Television, Radio, and Sound Equipment

(3) Other Equipment and Services

D. Reading and Education

2. Work-Related

A. Food

(1) Food at Home

(2) Food Away from Home

B. Apparel

C. Transportation

(1) Net Outlay for New Cars and Trucks

(2) Net Outlay for Used Cars and Trucks

(3) Other Vehicles

(4) Vehicle Finance Charges

(5) Gasoline and Motor Oil

(6) Maintenance and Repairs

(7) Vehicle Insurance

(8) Public Transportation

(9) Vehicle Rental, Licenses, Other Charges

D. Occupational Expenses (e.g., Union and Professional Dues, Tools, Uniforms, etc.)

3. Gifts

(1) See the President's Commission (1980, pp. 13-20), Clark (1977), and Employee Benefit Research Institute (1982).

(2) Dexter may have been prevented from controlling for a worker vs. retired variable by the manner in which the data were provided by the Bureau of Labor Statistics.

(3) Dexter's age- and work-related variable consists of changes (between pre- and post-retirement) in the following expenditure categories: medical care, household utilities and operations, education, gifts and contributions, food, transportation, and clothing.

(4) It is debatable as to whether post-retirement increases in giving should be subtracted fully, partially, or not at all in deriving wage replacement rates designed to maintain preretirement standard of living levels. Further, because of the sheer magnitude of the giving rates in the Dexter study, it is expected that any discussion as to the appropriate treatment of post-retirement gifts in replacement ratio calculations would be highly controversial. Without trying to reach a final conclusion on the merits of including a gifts variable in wage replacement calculations, let it be noted that the real concern centers on whether retirees who give away a lot of their wealth in later years should be penalized by an employer (or through implementation of social policy) through a reduction in their targeted wage replacement ratios simply because they choose to give away their wealth rather than spending it on themselves.

(5) Gibby and Dorer (1988) also authored a write-up of the Booke & Company study.

(6) Fundamentally, this formula is the same as the one used in the Dexter (1984) study.

(7) The Bureau of Labor Statistics (BLS) has disclosed a specific set of consumer expenditure information in its published report which shares the same title with the public use tapes. This publication is frequently referred to as Bulletin 2267. Bulletin 2267 contains comprehensive statistics gathered from the 1984 survey of consumer expenditures. This survey, currently performed annually, is conducted by the Bureau of Census under contract for the Bureau of Labor Statistics. While the BLS report is comprehensive, the desired breakdowns needed for this study pertaining to preretirement salary level, worker groups vs. retiree groups (see infra), etc., are not available in Bulletin 2267. Consequently, the data required for analysis in this study, of necessity, were obtained from the public use tapes.

(8) The definition of a consumer unit as it appears on p. 46 of Bulletin 2267 is: "A consumer unit comprises either: (1) all members of a particular household who are related by blood, marriage, adoption, or other legal arrangements; (2) a person living alone or sharing a household with others or living as a roomer in a private home or lodging house or in permanent living quarters in a hotel or motel, but who is financially independent; (3) two or more persons living together who pool their income to make joint expenditure decisions."

(9) There is a 5 percent surcharge (i.e., the tax rate is increased from 28 to 33 percent) that applies to taxable income amounts between $71,900 and $149,250 for joint returns, $61,650 and $123,790 for heads of household, and $43,150 and $89,560 for singles.

(10) A contrary opinion is that a portion of consumer interest payments should be taken as a tax deduction on the assumption that many consumers will establish a home equity line of credit and transfer some consumer debt to this equity line to retain tax deductibility of the interest payments.

(11) It is impossible to ascertain the magnitude of any error that is introduced because of this particular treatment of state and local income taxes. However, since many states base their income taxes directly on the federal tax return with (at most) minor modifications, it may be that any error in the resultant wage replacement rates is small.

(12) A portion of Social Security benefits will be subject to federal income taxation if the taxpayer's adjusted gross income, when combined with 50 percent of his or her Social Security benefits (plus any tax-exempt interest), exceeds $32,000 for a married couple filing a joint return ($25,000 for an individual). When this threshold is exceeded, the amount of Social Security benefits includable in taxable income is the lesser of (a) one-half of the Social Security benefits or (b) one-half of the excess of the modified adjusted gross income, plus one-half of the benefits, over the threshold amount.

(13) Post-retirement federal income taxes are calculated only on the amount of needed wage replacement in excess of the portion of Social Security benefits that is nontaxable. Consequently, retirees who are not entitled to Social Security benefits will need replacement ratios slightly higher than the ones derived in this study.

(14) See pp. 25-26 and Appendix I of the Dexter (1984) study for a discussion of the various kinds of property tax and other state laws which provide reduced taxes to the elderly.

(15) The PCPP (1981) and Booke & Company (England, 1988; Gibby and Dorer, 1988) studies also do not include property taxes in their wage replacement ratio calculations.

(16) These specific factors were chosen because it is believed that they represent monies spent by consumers on the acquisition of assets (or wealth) for investment purposes and, as such, were not needed for meeting current consumption needs. Since there does not appear to be a set of standard factors to be included in savings, other researchers might choose to specify their savings variable differently.

(17) It is possible that out-of-pocket health care expenditures may decrease after age 65 if the combined coverage under Medicare and any employer-provided retiree medical expense insurance is more comprehensive than the medical expense plan under which the employee was covered prior to retirement. This assumes that the broader coverage is not achieved through greater out-of-pocket premium costs to the retiree.

(18) Most statistics texts contain a discussion of the weighted least squares methodology.

(19) The amounts on line 12 are based on the Social Security law in effect in 1988. In the determination of the primary insurance amount (PIA), it was assumed that individuals' salaries had progressed to the 1988 levels at a rate of 6 percent per year. Also, the total dollar amounts are comprised of the worker's benefit of 100 percent of the PIA, plus the spouse's benefit (payable at age 62) of 37.5 percent of the PIA.

(20) The explanation for this modest increase in the wage replacement ratio at the $80,000 salary level may be related to significantly higher post-retirement income taxes because of an increase in the portion of Social Security benefits that is taxed.


[1] Allen, Everett T., Joseph J. Melone, and Jerry S. Rosenbloom, 1988, Pension Planning (Homewood, IL: Richard D. Irwin, Inc.).

[2] Bureau of Labor Statistics, Department of Labor, 1986, Consumer Expenditure Survey: Interview Survey, 1984 (Washington, DC: U.S. Government Printing Office).

[3] Clark, Robert, 1977, The Role of Private Pensions in Maintaining Living Standards in Retirement (Washington, DC: National Planning Association).

[4] Dexter, Michael K., 1984, Replacement Ratios: A Major Issue in Employee Pension Systems (Washington, DC: Public Employee Pension Systems).

[5] Employee Benefit Research Institute, 1982, Retirement Income Opportunities in an Aging America: Income Levels and Adequacy (Washington, DC: Employee Benefit Research Institute).

[6] England, Colin B., 1988, How Much Retirement Income Do Employees Need? Benefits Quarterly, 4: No. 1, 1-5.

[7] Gibby, Douglas A., and Carolanne Dorer, 1988, How Much Retirement Income Do Employees Need? Compensation & Benefits Management, 5: 25-34.

[8] McGill, Dan M., 1989, Fundamentals of Private Pensions (Homewood, IL: Richard D. Irwin, Inc.).

[9] President's Commission on Pension Policy, 1981, Coming of Age: Toward a National Retirement Income Policy (Washington, DC: U.S. Government Printing Office).

[10] President's Commission on Pension Policy, 1980, Working Papers -- Retirement Income Goals (Washington, DC: U.S. Government Printing Office).

Bruce A. Palmer is Professor of Risk Management and Insurance at Georgia State University. This research project was made possible by a grant from the Alexander & Alexander Consulting Group. Invaluable technical assistance was provided the author by Fred W. Munzemaier, F.S.A., Vice President of Alexander & Alexander.
COPYRIGHT 1989 American Risk and Insurance Association, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Palmer, Bruce A.
Publication:Journal of Risk and Insurance
Date:Dec 1, 1989
Previous Article:The distribution of claims for professional malpractice: some statistical and public policy aspects.
Next Article:The transactions cost theory of insurance: contracting impediments and costs.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters