by
Professor of Economics, Stanford University
Economist, The Bank of Italy
Economist, Cleveland, Ohio and
Professor of Economics, Boston University President, Economic Security Planning, Inc.
July 2000
Effective retirement planning begins with the establishment of appropriate goals for saving. When these goals are too high, households excessively sacrifice present well-being to sustain high living standards in the future. When these goals are too low, households excessively indulge immediate desires at the expense of future living standards. Thus, with inappropriate saving goals, households necessarily experience undesired changes in living standard, including (but not limited to) abrupt adjustments. In contrast, with appropriate saving goals, households “smooth” consumption, thereby avoiding these undesired changes.
Consumption smoothing is a fundamental prediction and prescription of modern economic theory. This theory is predicated on the well-known life cycle model of household behavior, which proceeds from the premise that each household is motivated by a sense of wellbeing that depends both on current satisfaction and on expectations of future satisfaction. The principle of consumption smoothing follows directly from the law of diminishing returns: individuals are well advised to reallocate dollars from time periods in which they are consuming a great deal (and in which incremental dollars therefore add relatively little to well-being), to periods in which they are consuming relatively little (and in which incremental dollars are therefore particularly valuable). To economists, consumption smoothing is the central purpose of saving.
The methods of traditional financial planning are inconsistent with, and in some instances antithetical to standard economic doctrine. The hallmark of these methods is the establishment of an asset target, derived either from income or spending objectives. Unfortunately, the latter objectives are not in turn derived from the principles of consumption smoothing. As a result, traditional financial plans frequently guarantee dramatic swings in spending as the household ages.
Despite this fundamental shortcoming, virtually every financial planning software package available today embodies the traditional targeted-saving approach. In some instances, users are asked to specify future spending or income targets with reference to current spending or income, but this is a far cry from consumption smoothing. To illustrate, imagine a household that is attempting to set a spending target. If it selects the target with reference to current spending, and if its current spending is not sustainable over its planning horizon, then it will be told to reduce its current consumption and save more than is required for consumption smoothing. Were the household to follow this prescription, it would experience an undesired surge in spending at retirement. To put it differently, the household would sacrifice excessively to obtain the benefits of a higher living standard after retirement. Conversely, if the household’s spending is currently less than it can sustain over its planning horizon, it will be told to increase its current consumption and save less than is required for consumption smoothing. Were the household to follow this prescription, it would experience a sudden and undesired drop in living standard at retirement. In other words, the household would deprive itself of a more satisfactory retirement by consuming excessively in earlier years. Setting future income targets with reference to current income is even less likely to generate a sensible path for consumption. A household’s current income may fluctuate because of one-time bonuses, temporary unemployment, enrollment in higher education, childcare, and/or a variety of other factors.
By proceeding from spending or income targets, traditional financial planning techniques implicitly require households to perform the most complex and important planning tasks by themselves. To establish goals that are consistent with consumption smoothing, households would need to consider a wide range of factors, including current and future household composition, the age and likely lifespan of each spouse, current and future labor earnings, special expenditures and receipts, social security benefits, current net worth, income from taxable and non-taxable assets, current and future contributions to retirement accounts, current and future federal and state taxes, asset returns, current housing and future housing plans, and borrowing constraints. Each of these factors interacts with others, and none can be evaluated appropriately in isolation. Consider, for example, future housing plans. Downsizing or upsizing a home would alter the future path of housing expenses, mortgage and property tax deductions, saving, capital income, and federal and state taxes. To determine the impact of housing choices on sustainable living standards, one must solve a complex dynamic programming problem. One cannot expect most individuals to understand intuitively the nature of the solution to this problem when selecting targets for future spending or income.
Mindful of these overwhelming complexities, traditional financial planners often advise households to set their targets using simple rules of thumb, such as 70 percent income replacement. Unfortunately, these seemingly straightforward recommendations are often highly inappropriate, and many households, deferring to the planner’s expertise, adopt them uncritically. In following such advice, a household smoothes its saving, rather than its consumption, and its living standard potentially fluctuates wildly from year to year.
The shortcomings of existing financial planning techniques led three of this paper’s four co-authors to develop a new financial planning software package known as Economic Security Planner, or ESPlanner.1 The software’s underlying algorithm determines each household’s maximum sustainable living standard, as well as the rates of saving and levels of life insurance holdings required to preserve that living standard through time. This paper begins with a description of ESPlanner. It then compares the economic approach to financial planning (as embodied in ESPlanner) with the conventional approach (as embodied in Quicken Financial Planner), drawing heavily on Gokhale, Kotlikoff, and Warshawsky (1999). Finally, it expands on Bernheim, Gokhale, Forni, and Kotlikoff (2000) by using ESPlanner to determine appropriate saving goals for typical Americans approaching retirement. In the course of accomplishing this final task, the paper compares our results with the findings of Engen, Gale, and Uccello (1999), Moore and Mitchell (2000), Warshawsky and Ameriks (2000), and other studies.
It is important to realize that the principle of consumption smoothing applies to the individual, rather than to the household. When individuals smooth consumption, household expenditure shifts with the arrival of new family members, and with the departure of old ones. Since larger households benefit from economies of scale with respect to shared expenses, consumption smoothing on the part of each individual does not requires household expenditures to increase proportionately with the size of the household; a smaller increase typically suffices. Accordingly, ESPlanner actually smoothes a measure of the household’s living standard, which depends on consumption per adult-equivalent (based on children’s ages), accounting for the economies of scale that are associated with family size.
It is also important to realize that the principle of consumption smoothing does not apply to all household expenditures. Exceptions occur when particular expenditures are either nonrecurring or difficult to modify. Examples include college tuition and housing expenses (downpayments, mortgages, and property taxes). ESPlanner deducts these special expenditures directly from income “off the top,” and smoothes the living standard derived from all remaining expenditures.
Application of the consumption-smoothing dictum is also sometimes limited in practice by institutional constraints. For example, lenders are often reluctant to extend unsecured credit. To smooth consumption, households with rapidly growing income must borrow against future receipts. If they cannot, then their consumption may rise (and even fluctuate) with income. Accordingly, ESPlanner smoothes consumption to the greatest extent possible, subject to the limitations on each household’s ability to borrow.
The principle of consumption smoothing applies to decisions about life insurance, as well as to decisions about saving. Households use life insurance to moderate the impact of a family member’s death on the survivors’ living standards. Moreover, decisions about life insurance and saving are inextricably linked. Current budgeted expenses must include an adequate allotment for life insurance premiums, and saving must be sufficient to cover future premiums. Accordingly, in deriving a financial plan, ESPlanner solves simultaneously for the ideal levels of saving and life insurance. It thereby ensures that survivors can sustain the same living standard as the intact family, irrespective of which family member dies, or when they die.
Naturally, many things can change following the death of a spouse. The survivor may move to a new house, change jobs, or return to work. He or she may incur additional child care expenses, or revisit plans to send to child to an expensive private university. To accommodate these important possibilities, ESPlanner permits households to undertake contingent planning. In particular, each spouse may specify different levels of earnings, special expenditures, and tax-favored retirement contributions in the event that he or she is widowed. Changes in contingent plans often have substantial effects on appropriate life insurance holdings.
Required Information. To apply the principle of consumption smoothing while accounting for the various considerations mentioned above, ESPlanner requires the following inputs.
Demographics. ESPlanner solicits the birth dates of the household head and spouse as well as the birth years of children under age 19. Children are assumed ordinarily to remain in the household through age 18. Each spouse must also specify a maximum length of life, which refers to the limit of the individual’s planning horizon (note that this differs from life expectancy). The program smoothes consumption over this horizon, thereby protecting household members from the possibility that they might outlive their resources. Users also identify their state of residence, which is used to determine applicable tax rates.
Standard of Living Index. Economic theory allows for the possibility that any given household might prefer either a rising or falling standard of living to one that is constant through time. A household might also prefer to change its level of consumption upon retirement because it anticipates increased spending on activities that are complementary with leisure, and/or reduced spending on activities that are substitutes for leisure. ESPlanner accommodates these possibilities by permitting users to specify how they would like their living standard to change through time. By adjusting a living standard index from its default value of 100 in any year or collection of years, a user can customize the shape of its living standard profile (for example, it can specify that its living standard is to grow at the rate of 1 percent per year, or is to decline by 10 percent at retirement). The program then determines the highest current living standard, as well as the associated financial plan, that is consistent with the characteristics specified by the user.
Labor Earnings. For each spouse, ESPlanner solicits current labor income, as well as the amount that he or she would expect to earn if widowed (contingent earnings). Separate information is collected on employee wages and self-employment income. It is necessary to distinguish between these forms of income because they are treated differently under the payroll tax. Each spouse also specifies a retirement date and a growth path for labor earnings through retirement, as well as a (potentially different) retirement date and earnings growth path that would apply in the event he or she is widowed. To speed data entry, users may elect a default that sets the contingent earnings path equal to the joint-survivor path. Users may specify future income in either present-year (real) dollars, or in future-year (nominal) dollars.
Special Expenditures and Receipts. ESPlanner also provides users with the ability to specify non-recurring (or briefly recurring) expenditures and receipts. Each special expenditure must be designated as either deductible or non-deductible, and each special receipt as either taxable or non-taxable. For each briefly recurring item (such as college tuition), users provide a start date and an end date. Each spouse specifies special expenditures and receipts that would apply in the event he or she is widowed. To speed data entry, users may elect a default that sets the contingent expenditures and receipts equal to the joint-survivor values. Users may specify future expenditures and receipts in either present-year (real) dollars, or in future-year (nominal) dollars.
Estate Plans. In many instances, an individual may wish to leave a bequest in excess of the amount required to sustain the surviving spouse’s living standard through his or her maximum lifespan, and to sustain the living standard of each child through age 18. Accordingly, ESPlanner permits users to specify special (incremental) bequests, including resources to defray death-related expenses (such as funerals).
Net Worth. Information on net worth is essential for accurate financial planning. ESPlanner separately solicits data on non tax-favored and tax-favored assets. In the case of tax-favored accounts, each spouse’s holdings are detailed, as well as a) the last year he or she will contribute to the account, b) the first year he or she will start withdrawing from the account, and c) the year he or she will stop withdrawing from the account. Users may select one of two options for withdrawing tax-favored balances: uniform withdrawals, or the smallest legally permissable withdrawals.
Saving. An individual who attaches a high value to liquidity may be reluctant to tie up too large a fraction of his or her net worth in tax-favored accounts. Conversely, someone less concerned about liquidity may wish to maximize tax-favored holdings. Accordingly, ESPlanner permits the user to determine the composition of his or her saving. Users indicate current non-tax-favored saving, as well as current and intended future employee and employer contributions to tax-favored accounts (for both joint-survivor and widowed contingencies).
Housing. For most Americans, housing represents both a major expense and an important store of wealth. Accordingly, ESPlanner solicits information on both primary and secondary (vacation) homes. Homeowners estimate current market value, provide information on loans, and detail current expenses. Renters also list housing-related expenses. Users describe future plans concerning refinancing and moves (including upsizing, downsizing, liquidation of second homes, shifts between homeowner and renter status, and so forth).
Pensions. Accurate financial planning requires detailed information on work-related retirement benefits. ESPlanner treats defined contribution (DC) accounts as tax-favored assets. Each spouse separately supplies information on defined benefit (DB) pensions, including the year or years in which benefits will be received, projected amounts (either lump-sum, annual, or both), whether the benefits are indexed to inflation, and the level of benefits received by a survivor.
Social Security. Social security remains an important source of retirement and disability income for many Americans. ESPlanner uses past and future earnings in covered employment to estimate benefits for those who are not yet collecting benefits. Its benefit calculator considers eligibility rules, early retirement reductions, delayed retirement credits, benefit recomputations, the phased increase in the normal retirement age, the earnings test between ages 62 and 65, family benefit maximums, the wage indexation of Average Indexed Monthly Earnings, and the price indexation of benefits once they are received. All these elements feed into the determination of retirement, spousal, mother, father, child, and widow(er) benefits.
Economic Assumptions. To undertake meaningful financial planning, one must make a variety of assumptions about the economic environment. ESPlanner supplies default values for all critical economic parameters, including the inflation rate, nominal rates of return on tax-favored and non tax-favored assets, the degree of economies in shared living, child-adult equivalency factors, the maximum amount the household can borrow (apart from mortgages), future rules governing payroll taxes and social security benefits, and the share of total non tax-favored capital income accruing in the form of long-term capital gains. Users are permitted to substitute alternative values for these parameters.
Taxes. Meaningful financial planning requires proper recognition of tax liabilities. Accordingly, ESPlanner calculates federal and state income and payroll taxes. For each future year and for each survival state (both spouses alive, wife alive and husband deceased, husband deceased and wife alive), estimates of federal income taxes reflect deductions and exemptions, the partial taxation of social security benefits, the earned income tax credit, the child tax credit, the phaseout of deductions and exemptions at higher income levels, the indexation of tax brackets to the consumer price index, and the preferential taxation of long-term capital gains. In computing deductions, the household is assumed to itemize if eligible expenses (principally mortgage payments, property taxes, state income taxes, spousal support payments, charitable contributions, and other designated special expenses) exceed the standard deduction. Estimated state income tax liabilities (for each year and for each survival state) reflect state of residence, as well as the specific exemptions, deductions, and rate structure appropriate for that state. In computing both federal and state taxable income, the program deducts, as appropriate, contributions to tax-favored accounts and includes, as appropriate, withdrawals from these accounts. Finally, the determination of FICA taxes accounts for the ceiling on covered earnings, which applies for to the portions of the tax that finance retirement and disability benefits, but not to the portion that finances Medicare.
ESPlanner’s Recommendations. ESPlanner’s principal outputs are recommended time-paths for consumption expenditure, non tax-favored saving, and term-life insurance holdings (for each spouse individually in the case of married couples). All outputs are displayed in current-year (i.e., real) dollars. Recommendations for saving and life insurance are compared with current choices.
In this context, “consumption” refers to all spending over and above “off-the-top” items, including housing expenses, special expenditures, life insurance premiums, taxes, and net contributions to tax-favored accounts. Recommended consumption expenditures vary from year to year when the household’s composition changes, and when the household moves into or out of a liquidity-constrained period. Naturally, recommended household consumption may also change over time when the user has expressed a preference for a rising or declining living standard (as indicated by the values of the standard of living index, discussed above).
Recommended taxable saving in any year equals the household’s total income (non-asset plus asset income) minus the sum of (a) recommended spending on consumption and insurance premiums, (b) specified spending on housing and special expenditures, (c) taxes, and (d) net contributions to tax-favored accounts (contributions less withdrawals).
Recommended levels for term life insurance are either positive or zero.2 If recommended term insurance is positive for a particular potential decedent (the household head or, if married, the spouse) in a particular year, and if the decedent dies at the end of that year, the surviving household will have precisely the same living standard as the household would have had absent the decedent’s premature death. If the potential decedent’s recommended insurance in a particular year is zero, the surviving household will have the same or higher living standard if the decedent dies in that year. These statements are, of course, conditional on complete execution of the recommended financial plan and on the correctness of underlying economic assumptions and information concerning future income, current asset holdings, special expenditures, and so forth.
Although the derivation of the recommended financial plan involves a complex dynamic programming algorithm, reports and recommendations are easily interpreted. Moreover, from an inspection of the reports, it is readily evident that the program achieves the objective of consumption smoothing, thereby identifying the highest sustainable living standard for the household.
Illustrating ESPlanner. Al and Peg, a married couple, decide to formulate a detailed financial plan. Currently (as of 2000), Al is 50 years old and Peg is 45. They reside in the state of New York. They have two children, Kelly, age 15, and Bud, age 13. Al and Peg each plan to work through age 65 earning, respectively, $25,000 and $100,000 each year. These figures (and all others mentioned in this illustration) refer to year-2000 dollars. If Al were to die, Peg would still earn $100,000, but if Peg were to die, Al would switch jobs and expect to earn $40,000. The couple plans to send each child to college for 4 years and to spend $30,000 per child per year on tuition. Al and Peg wish to allocate $5,000 each for their funerals. Anticipating a desire to pursue costly leisure activities during retirement, they decide to specify a 10 percent increase in living standard upon Al’s retirement. They currently own and live in their home. The house has a market value of $300,000. Annual property taxes are $5,000, annual homeowners insurance is $750, and annual maintenance averages $1,500. Al and Peg have 25 years remaining in a 30 year mortgage; their current mortgage balance stands at $200,000, and they pay $2,200 each month. When Al is 70 years old, they plan to sell their home and rent an apartment for $2,000 per month (in today’s dollars). Each spouse works in social security-covered employment and the past covered earnings of each spouse grew smoothly to their current values. The couple wants to set aside $100,000 by 2020 (when Al is age 70 and Peg is age 65) as an emergency fund for medical expenses. If only one spouse is alive in 2020, they plan to put only $50,000 aside.
Table 1 shows ESPlanner’s annual non tax-favored saving, consumption, and life insurance recommendations. Table 2 details the couple’s future spending, including its consumption, housing expenses, special expenditures, life insurance premiums, and funeral expenses. Table 3 is a balance sheet --it tracks the household’s non tax-favored assets. Table 4 displays Al and Peg’s income. Non asset income refers to labor income, pension income, and social security benefits. All of these first four tables assume each spouse lives to his or her maximum age of life.3
Consider first the consumption recommendations in Table 1. (Table 1 here) Recommended discretionary expenditures equal $58,018 through 2004, the year Kelly goes to college. At that point, consumption falls to $49,622. It drops again to $40,486 in 2006 when Bud leaves the household and goes to college. Consumption remains at this level until 2015 when Al reaches age 65. At this point, consumption rises by 10 percent to $44,534 in accordance with Al’s and Peg’s desire to have a 10 percent higher living standard in retirement. Finally, in 2046, when Al is deceased, consumption falls to $27,834, since there is only one mouth left – Peg’s – to feed. Note that the ratio of consumption when Al and Peg are both alive ($44,534) to the value when only Peg ($27,834) is alive is 1.6. This reflects our assumption that, with the addition of a second adult, spending must increase by a factor of 1.6 (i.e. by 60%) to preserve the same living standard.
In contrast to the household’s living standard, recommended non-tax-favored saving fluctuates widely. It is positive until the children go to college, negative when they are in college, positive after they leave college, and negative once Al and Peg are retired. Note that Al’s and Peg’s non-tax-favored saving is largest immediately prior to Peg’s retirement. This is what one would expect. In their younger years, Al and Peg have to pay their mortgage and college tuition for their children. After these obligations have been met, Al and Peg can concentrate on saving for retirement. The largest increment to their liquid assets occurs when they sell their home. Their highest rate of dissaving occurs when they make special expenditures. Likewise, as indicated in Table 2, total spending also fluctuates more than discretionary spending, due to changes in special expenditures, housing costs, life insurance premiums, and funeral expenses. (Table 2 here)
ESPlanner recommends that the couple initially obtain $468,868 in insurance on Peg’s life. Over time, Peg’s recommended life insurance declines, and reaches zero at age 64. For Al, recommended life insurance is zero. Even without life insurance, Peg and the children would enjoy a higher material living standard were Al to die than were he to live.
By inspecting the balance sheet in Table 3, one can readily see that ESPlanner’s consumption recommendations are affordable. (Table 3 here) This balance sheet tracks the evolution of the couple’s non tax-favored net worth. The change in the household’s non tax-favored net worth from one year to the next equals recommended non tax-favored saving. This flow, in turn, equals the difference between the household’s income, which is detailed in Table 4, and the sum of a) its net contribution to retirement (non tax-favored) accounts, b) its total spending, and c) its taxes. (Table 4 here) Note that household net worth is never negative. This implies that the plan is feasible. Note also that net worth is zero when Peg reaches her maximum lifespan. This implies that that there are no unused resources. It is therefore infeasible to increase consumption in any year without reducing it in another year. Thus, the program has identified the highest consumption profile with the characteristics that the couple desires (an unchanging living standard, except for a 10 percent rise at retirement).
By referring to ESPlanner’s survivor reports, one can verify that surviving spouses are also able to maintain the couple’s accustomed living standard (assuming, of course, that all information and assumptions are correct and that the couple fully executes the complete plan). Table 5 details Al’s recommended spending assuming that Peg dies at age 46, one year after adopting the plan. (Table 5 here) Recommended consumption for Al declines when the children leave the household and then rises by 10 percent when Al reaches age 65. Note that when Al is living by himself, the ratio of his consumption in any year to the corresponding value in Table 2 is 1 divided by 1.6. Given our assumption concerning the magnitude of household scale economies, this implies that Al is enjoying the same living standard as a survivor that he would have enjoyed had Peg not died.
By inspecting the balance sheet in Table 6, one can readily see that ESPlanner’s consumption recommendations are affordable for Al if Peg dies. (Table 6 here) This follows from the fact that Al never goes into debt. Note also that net worth is zero when Al reaches his maximum lifespan. This implies that that there are no unused resources. It is therefore infeasible to increase consumption in any year without reducing it in another year. Upon Peg’s death, Al’s non tax-favored wealth is $605,992. This amount equals the couple’s $158,647 in non tax-favored assets at the end of 2001 plus the $452,345 in term insurance recommended for Peg in 2001, less the $5000 payment for Peg’s funeral. Were the couple to purchase less insurance on Peg’s life, Al would not be able to finance the same living standard as a survivor.
Limitations. Although ESPlanner considers many key factors that enter into saving and insurance decisions, it is important to acknowledge that some relevant factors are omitted. Two specific omissions merit discussion. First, the software does not take into account the uncertainty of future income or expenditures on necessities, such as non-insured health care costs. To understand the implications of uncertainty, users must perform sensitivity analysis, examining a variety of alternative scenarios to assess their exposures and vulnerabilities. Second, the software does not account for possible changes in marital status, such as remarriage after a spouse’s death. To some extent, the remarriage option may mitigate financial vulnerabilities associated with the risk of a spouse’s death. There are, nevertheless, legitimate reasons to ignore this possibility. Arguably, the choice of whether to remarry should not be dictated by financial necessity. In addition, the economic well-being of a remarried individual may be determined by his or her financial status prior to remarriage, insofar as this affects bargaining power within the new marriage (see e.g. Lundberg, 1999). Finally, remarriage after a spouse’s death is less common among older individuals.
Table 7 summarizes some key findings from Gokhale, Kotlikoff, and Warshawsky (1999), who compared the economic approach to financial planning, as embodied in ESPlanner, with the traditional approach, as embodied in Quicken Financial Planner. (Table 7 here) We present consumption, saving, and life insurance recommendations for three actual households – a low-income, young married couple with no children, a upper-income, middle-aged married couple with two children, and a high-income, older married couple with adult children. Although these three cases are hypothetical, they are based on actual households.
In deriving financial plans with Quicken Financial Planner (QFP), Gokale, Kotlikoff, and Warshawsky attempted to emulate the manner in which a somewhat sophisticated household might use the program. After soliciting current spending levels, QFP asks the user whether he or she wishes to spend the same amount in the future. We assume that most households would, at least initially, answer this question in the affirmative. Using information on income and net worth, the program then determines whether the desired expenditures are feasible. If planned spending is not feasible, the user must adjust planned expenditures downward. If planned spending is feasible, the user can choose to adjust planned expenditures upward. We assume that a sophisticated household would follow this procedure iteratively until it determined the highest feasible level of consumption. This manual process of “trial and error” is time consuming. Consequently, we doubt that even sophisticated households would further fine-tune their expenditure plans to accommodate changes in household composition (such as the arrival and departure of children from the household), borrowing constraints, or other factors that ESPlanner handles automatically.
Gokhale, et. al. (1999) compared the two software programs’ recommendations for 24 different households. Saving and life insurance recommendations differed dramatically in each case, with the discrepancies generally increasing with the complexity of the case. In some instances, ESPlanner recommended substantially more saving in early years and substantially less saving in later years than did QFP. In other cases, the opposite was true. The differences in life insurance recommendations were more systematic, with ESPlanner generally recommending significantly less life insurance than QFP.
Differences in saving recommendations primarily reflect ESPlanner’s adjustments for household demographics, economies in shared living, and borrowing constraints, as well as its different, and more detailed, approach to the calculation of federal and state income taxes and social security retirement benefits. Differences in life insurance recommendations reflect the same set of factors, plus two additional considerations: ESPlanner allows for contingent planning
(i.e. changes in plans concerning employment, expenditures, housing, and investment upon the death of a spouse), and it fully accounts for social security survivor benefits when determining insurance needs. The following three cases illustrate some of these considerations.
The Young, Low-Income Couple. For this case, we assume that each spouse is 35 years old in 1999, and each retires at age 65. They plan to have two children, one in 2001 and one in 2003. The husband earns $43,000 initially, declining by 2001 to $35,000 and staying constant in real terms thereafter. The wife earns $37,000 in 1999, zero in 2000, $35,000 in 2001, $36,000 in 2002, $37,000 in 2003, and $38,000 thereafter. The husband receives a gift from his father of $10,000 in 1999 and 2000. Special expenditures include nominal truck loan payments of $4,500 in 1999 and 2000. The couple also plans to spend $20,000 on college tuition for each child between the ages 19 and 22. The couple allocates $5,000 for each spouse’s funeral, but does not wish to leave incremental bequests over and above the level necessary to assure survivors of an undiminished living standard. The couple’s current assets include $14,000 in taxable accounts, as well as $3,000 in an IRA under the wife’s name. The wife intends to contribute $1,200 to her IRA annually until she retires. She will begin withdrawing funds from her IRA at age 65 in equal annual installments. The couple purchases a house in 1999 for $150,000. They make a downpayment of $15,000 and take out a $135,000, 30-year mortgage. Monthly mortgage payments, including principal and interest, total $990. Annual housing expenses include $2,500 in property taxes, $400 in homeowner’s insurance payments, and $2,000 in maintenance. Both spouses will begin collecting social security retirement benefits at age 65. The calculations presented in Table 7 reflect ESPlanner’s default assumptions concerning economic parameters, including a 6 percent nominal interest rates on taxable and non-taxable assets, a 3 percent inflation rate, and a non-negativity constraint on non-housing wealth (equivalently, no unsecured borrowing).
If the couple follows ESPlanner’s recommendations, it will never encounter liquidity constraints; nor will it ever accumulate a significant stock of taxable assets. According to the table, the couple is advised to consume $26,866 initially and $38,500 when both children are present in the household. QFP, on the other hand, recommends constant consumption of $26,920 as long as both spouses are alive, regardless of whether children are present. ESPlanner’s recommendation is more reasonable, in that decisions concerning spending and saving should account for the costs of child rearing. Both programs indicate that husband and wife should have similar life insurance holdings, reflecting their similar economic contributions to the household. However, ESPlanner recommends significantly less life insurance than QFP.
The Middle-Aged, Upper-Income Couple. For this case, we assume that the wife is 40 years old, and the husband 39, in 1999. The couple resides in Massachusetts. They have two children, one born in 1991, the other born in 1993. The wife does not work. The husband earns $200,000 in 1999 and 2000. Starting in 2001 and continuing until his retirement at age 55, the husband expects to earn $100,000. They plan to send each of their children to college for four years at a cost of $30,000 per child per year. They allocate $5,000 for each spouse’s funeral, but do not wish to leave incremental bequests over and above the level necessary to assure survivors of an undiminished living standard. The couple’s taxable assets are $225,500. The wife has an IRA with a 1999 balance of $84,700, and the husband has a 401k with a 1999 balance of $148,000. Both plan to withdraw their non-taxable assets (thereby making them subsequently taxable) at age 59. The couple currently saves $11,765 per year in taxable forms. The husband plans to contribute $9,500 to his 401k plan each year and expects his employer to contribute $6,000. The wife does not intend to make additional IRA contributions. The couple owns a $475,000 house with annual property taxes of $5,200, annual maintenance of $1,500, annual homeowners insurance of $500, and a 29-year $170,000 mortgage with monthly payments of $1,131. Each spouse intends to begin receiving his/her social security retirement benefits at age 62. The calculations presented in Table 7 reflect ESPlanner’s default assumptions concerning economic parameters.
QFP’s and ESPlanner’s recommendations for consumption, taxable saving, and life insurance differ dramatically. For example, compared to ESPlanner, QFP recommends more than twice as much insurance on the husband’s life. These findings are traceable to several factors. First, ESPlanner recommends that spending should decline sharply when the children leave the household. Initial consumption (with children present) therefore exceeds the level recommended by QFP. Second, ESPlanner’s estimate of the couple’s short-term tax liabilities is significantly higher than QFP’s. This is primarily attributable to ESPlanner’s treatment of Massachusetts income taxes, which imposes high rates on capital income. Third, unlike ESPlanner, QFP does not allow the employer’s matching 401(k) contribution to rise with inflation-induced increases in pay. Fourth, QFP’s social security benefit estimates are lower than ESPlanner’s.
An Older, High Income Couple. For this case, we assume that the husband is 64 years old and the wife is 57. The husband intends to work for two more years, earning close to $400,000 over this period. The couple has a variety of large special expenses in the short run, including an expensive home renovation. The husband has two pensions providing close to $200,000 (nominal) annually, and he expects to begin receiving this income as soon as he retires. The couple allocates $5,000 in funeral expenses for each spouse. In addition, the couple wants to provide gifts or bequests for its children totaling $2 million as of 2025. The couple’s taxable net worth is close to $3 million. The wife has a small IRA account, and the husband has a 401k account worth close to three-quarters of a million dollars. Each spouse elects to withdraw the smallest amount of funds from these tax-favored accounts permitted by law. The couple owns a house with a market value of $1,200,000. Annual property taxes, maintenance, and homeowners insurance total $6,000, $13,000, and $1,000, respectively. There is a 25-year mortgage on the property with a balance of $525,000 and a monthly payment of $3,318. The couple plans to sell its home in 2025 and thereafter rent a home for $4,000 per month. The calculations presented in Table 7 reflect ESPlanner’s default assumptions concerning economic parameters.
Neither QFP nor ESPlanner prescribe life insurance for either spouse in this case. However, recommendations for consumption and saving diverge considerably. According to ESPlanner, the household can spend $204,510 in 1999 (on items other than housing, taxes, life insurance premiums, and special expenditures), whereas QFP indicates that it should spend no more than $186,880. This discrepancy is, in large part, attributable to the treatment of taxes. ESPlanner’s estimate of the couple’s 1999 tax liabilities is $182,449, whereas QFP’s estimate is $237,681. This large (30 percent) difference is apparently attributable to the deductibility of certain special expenditures, which ESPlanner recognizes, while QFP does not. ESPlanner’s estimate of the couple’s tax liabilities actually exceeds QFP’s by the time the husband reaches age 75, but then falls below QFP’s. Due in part to the presence of very large, short-term special expenditures on home remodeling, which are not captured properly by QFP, ESPlanner recommends that the couple dissave $317,615 in 1999. In contrast, QFP recommends that the couple dissave only $138,380.
Implications. As these results demonstrate, it is extremely difficult to achieve consumption smoothing with traditional financial planning tools, even when one uses these tools in a relatively sophisticated way. Households that rely on these tools will most likely experience significant, predictable (though unintended), and avoidable changes in living standard over the courses of their lives.
The importance of accurate financial planning deserves emphasis. Research indicates that individuals do change their financial decisions in response to information and guidance, particularly when provided through employers (see Bernheim and Garrett, 1999, Bayer, Bernheim, and Scholz, 1996, Bernheim, 1998, and Clark and Schieber, 1998). Faulty information can therefore lead to poor decisions. In a related, ongoing project, the authors of this paper are conducting a study among users of ESPlanner at Boston University, in an effort to assess the impact of financial planning on behavior. Preliminary results are encouraging. Despite the program’s internal complexity, it has proven accessible, user-friendly, and understandable. Many users indicate their intentions to follow its recommendations. We plan to assess behavioral changes through follow-up surveys and by tracking pension plan activity.
Over the next two decades, a significant fraction of Americans belonging to the 75 million-member baby boom generation will reach retirement age. Impending retirement magnifies the importance of saving, particularly for those who are currently over fifty. Moreover, in planning for retirement, boomers must recognize the possibility that fiscal pressures may eventually force lawmakers to cut social security benefits. Short-term surpluses notwithstanding, the social security system may be as much as 40 percent underfunded: in terms of present value, projected tax contributions represent only 60 percent of projected benefit payments. While the Trustee’s of the social security system acknowledge some degree of underfunding, their annual Trustees Report understates the problem, for two reasons. First, the Trustees truncate their long-term projections at 75 years; second, they make excessively optimistic assumptions concerning future changes in longevity.
To understand the baby boomers’ needs for retirement saving, Bernheim, et. al. (2000) used ESPlanner to derive financial plans for a sample of individuals drawn from the Health and Retirement Survey. We contrasted recommended levels of saving under two alternative policy scenarios. In the first “base case” scenario, Congress avoids reductions in social security benefits. In the second “fiscal distress” scenario, Congress is forced to reduce benefits by 30 percent in 2015. In the remainder of this section, we summarize the key findings of that study. We also extend our earlier analysis by exploring the implications of some alternative assumptions.
The HRS. The 1992 wave of the Health and Retirement Survey (HRS) collected information on the families of 12,652 individuals. Though the survey contains a great deal of economic and demographic data, it does not include all of the information required by ESPlanner. We imputed the values of all missing variables; see Bernheim, Forni, Gokhale, and Kotlikoff (1999) for details. We restricted our analysis to households satisfying the following criteria: (1) the head’s age was between 51 and 61, (2) information on social security earnings in past covered employment is available for both the head and the spouse (if any), and (3) the respondent answered all critical survey questions. We excluded an additional 141 because their economic resources were insufficient to cover their housing costs and other off-the-top expenditures. Our final sample consisted of 1714 married couples and 1145 single individuals.
Recommended Saving Rates. Table 8 reports the median recommended saving rates for HRS households. (Table 8 here) The saving rate is defined as non tax-favored saving divided by income. Note that this measure of saving excludes contributions to or withdrawals from retirement accounts. Our measure of income also excludes net contributions to tax-favored accounts. We sort heads into two age groups – 50 to 55 and 56 to 61 --and then further decompose them into categories based on household income, marital status, race, and education. We present results for both social security policy scenarios. Our calculations assume a 6 percent nominal interest rate and a 3 percent inflation rate.
Consider first our base case policy scenario, which involves no social security benefit cuts. Focus for the moment on the 50 to 55 year-old age group. The median recommended saving rate those with incomes below $15,000 is very small --only 1 percent. At the other extreme, for those with incomes over $100,000, the median recommended saving rate is fairly high --17 percent. For households with incomes between $15,000 to $45,000, and for those with incomes between $45,000 to $100,000, the median recommended saving rates are 13 percent and 14 percent, respectively. The strong positive relation between recommended saving rates and income is, in large part, attributable to the progressive structure of the social security benefit formula, which provides lower income individuals with significantly higher rates of earnings replacement.
For the older subsample (those between the ages of 56 and 61), we find again that median recommended saving rates increase sharply with income. With the exception of the lowest income group, these rates also rise steeply with age; they range from 17 percent for older households with incomes between $15,000 and $45,000, to 23 percent for older households with incomes over $100,000.
To some extent, households can achieve these saving rate targets by reinvesting income earned from previously accumulated assets. It is therefore natural to wonder whether reinvested capital income is sufficient to reach the targets, or whether households must also put away significant fractions of take-home pay. To examine this issue, we calculated recommended rates of saving for non-asset income. Specifically, we adjusted the recommended saving rates by subtracting non-tax-favored capital income from both the numerator and the denominator of the ratio. For the lowest and second-lowest income segments of both age groups, median recommended saving rates are essentially unchanged. For households with incomes between $45,000 and $100,000, the adjustment reduces the median recommended saving rate among younger households from 13 percent to 12 percent, and leaves the median recommended saving rate among older households unchanged at 17 percent. Hence, most households do need to save significant fractions of non-asset income.
Recommended saving rates tend to be higher for single individuals, non-whites, and those without college education. For example, among non-white households between the ages of 56 and 61 with incomes between $15,000 and $45,000, the median recommended saving rate is 23 percent. This is 6 percentage points higher than the corresponding rate for whites and nonwhites combined. Likewise, in the same age and income group, the median recommended saving rate is 23 for single households, compared with 14 percent for married couples. Though these systematic differences are important, it is essential to bear in mind that they mask considerable variation within groups. In formulating useful financial plans, the particular circumstances of a household are often much more important than its general demographic characteristics.
The Impact of Potential Social Security Benefit Cuts. Table 8 also contains recommended saving rates for the second “fiscal distress” policy scenario, in which lawmakers cut social security benefits by 30 percent as of 2015. The results differ dramatically from those obtained for the base case scenario. Consider, for example, married households in the lowest income category. The median recommended rate of saving rises by 10 percentage points for the younger age group, and by 12 points for the older group. In the second lowest income group, recommended saving rates rise by 8 and 7 percentage points, respectively, for the younger and older age groups. Among high-income households, the increases are smaller, but still significant. Recommended rates of saving out of non-asset income also rise sharply; for the middle-income groups, these rates range from 16 to 22 percent.
Qualitatively similar conclusions follow for plausible alternative values of the key economic parameters. For example, with an 8 percent nominal (5 percent real) rate of return, the median recommended saving rates for 50-55 year olds are, respectively, 1, 11, 11, and 10 percent for the first through fourth income categories. The corresponding figures from Table 8 are 1, 13, 14, and 17 percent. For 56-61 year-olds, median recommended saving rates with the higher rate of return are, respectively, 1, 16, 17, and 20 percent for the first through fourth income categories. The comparable figures from Table 8 are 0, 17, 20, and 23 percent. Thus, recommended saving rates are lower with the higher interest rate. However, recommendations are still highly sensitive to assumptions about social security benefits. For our second policy scenario, median recommended saving rates among 56-61 year olds are, respectively, 4, 17, 16, and 12 percent for the first through fourth income categories, assuming a nominal return of 8 percent. Among 50 to 55 year olds, the comparable figures are 3, 21, 21, and 22 percent. Each of these rates is significantly higher than the corresponding figure for the base case policy scenario.
Alternative Assumptions: Lifespan, Market Performance, Retirement, & Nursing Home Care. Table 9 explores the sensitivity of our results with respect to various alternative assumptions. (Table 9 here) First, we recalculate recommended saving rates assuming a maximum lifespan (for both the respondent and the spouse) of 100, rather than 95. Second, we assume that there is an immediate 30 percent decline in the market value of stocks and other financial assets, after which these assets earn the same return as in our base case. Third, for each household, we accelerate retirement by two years. Fourth, we assume that respondents and spouses must each accumulate a reserve fund sufficient to defray the costs of nursing home care at $15,000 per year for five years (in present dollars). We recognize that the cost of nursing home care may exceed $15,000 per year; however, other spending presumably declines when an individual is institutionalized. The $15,000 figure is intended to represent the net increment to total expenditures. Finally, we consider the combined effects of all of these assumptions, along with the fiscal distress scenario examined previously.5 For purposes of comparison, Table 9 also summarizes our findings for the base case and fiscal distress scenarios.
A quick glance at the table reveals that the fourth assumption (saving for nursing home care) has the largest impact on recommended saving rates. It is particularly important for those with the lowest levels of income, raising recommended median saving rates by 18 to 30 percentage points. These figures may be somewhat exaggerated, in that low income families are more inclined to rely on Medicaid, even though this tends to reduce the quality of care received. Nevertheless, our results suggest more generally that low income families may need to save at high rates if they wish to establish non-trivial emergency funds.
Increasing the maximum lifespan from 95 to 100 years has a more modest effect on median recommended saving rates, which generally rise by 1 to 2 percentage points, with the exception of older, low income, single individuals, for whom the increase is 5 percentage points. A 30 percent decline in asset values also has a relatively small effect on recommended saving rates, except among high-income households. Finally, accelerating retirement by 2 years has a sizable impact on recommendations for particular subgroups. For example, the median recommended saving rate rises by 5 percentage points for married couples between the ages of 50 and 55, with incomes between $15,000 and $45,000. In some groups, recommended saving stays constant or declines. This occurs because the acceleration of retirement renders some households unable to cover housing expenses and other off-the-top commitments.
When we consider the combined effects of all four assumptions along with the fiscal distress scenario, recommended median saving rates rise dramatically for all subgroups; generally, they are in the neighborhood of 30 to 40 percent. For example, among the lowest-income married couples, the median recommended saving rate increases from zero to 39 percent.
Moreover, the number of households with infeasible planning problems (that is, those who can no longer cover off-the-top expenditures) rises from 141 to 346. This statistic sheds additional light on the degree of undersaving and financial vulnerability among HRS households. As Table 10 indicates, lower-income households are more likely to present infeasible planning problems. (Table 10 here)
The above analysis suggests rates at which households should be saving, but doesn’t compare these recommended saving rates with actual ones for the simple reason that the HRS data does not provide a reliable measure of actual saving. A number of other studies have, however, attempted to assess saving adequacy either indirectly or directly.
Kotlikoff, Spivak, and Summers (1983) is an example of an indirect approach. They examined the issue by comparing household’s entire lifetime sustainable consumption, calculated based on entire lifetime resources, with remaining lifetime sustainable consumption calculated based on remaining lifetime resources. They concluded that, absent Social Security’s forced saving, a significant fraction of the elderly would suffer a decline in living standard in old age. Bernheim (1994) studies saving directly by comparing actual saving behavior of baby boomers with recommended saving generated by a stylized life-cycle model. He reports that typical baby boomers are saving only one third of what they need to maintain their current living standards. Bernheim and Scholz (1983,1986) use the 1983 and 1986 Surveys of Consumer Finances to compare the median change in wealth by education group across the two survey years. They find inadequate growth in wealth among non-college graduates compared to that implied by a optimizing life-cycle saving model.
Warshawsky and Ameriks (2000) use Quicken Financial Planner to assess saving adequacy among participants in the 1992 Survey of Consumer Finances. They find that over half of the sample will run out of money at some time in the future if they try to maintain their living standard. Their analysis is predicted on a number of strong assumptions associated with their inference of household’s current spending levels, but their findings are highly suggestive.
The assessment of these and other papers that, left to their own devices, households will undersave is questioned by Manchester (1994), Hubbard, Skinner, and Zeldes (1994), and Engen, Gale, and Uccello (1999). Manchester argues that today’s baby boomers have saved more than did their parents. Hubbard, Skinner, and Zeldes argue that undersaving may be optimal for many low income households who expect to rely on the future largess of the government’s Medicaid and income-support programs. And Engen, Gale, and Uccello argue that apparent low saving rates may reflect temporarily low income. They also claim that median wealth holdings by age are as high, if not higher, than what one would predict based on optimal life-cycle choice.
The study that comes closest to ours is Moore and Mitchell (2000). They too use the HRS data and consider how much households need to save to maintain their pre-retirement living standard. Their methodology differs in many respects from our own. For example, they assume that housing is fungible, whereas we assume that homeowners retain their homes. Our assumption seems to accord more close with the findings of Venti and Wise (this volume) and Caplin (this volume). Venti and Wise state (page 3) that “In general, we see very little reduction in home equity that can be construed as converting home equity to liquid assets for purposes of supporting non-housing consumption. And Caplin finds very little evidence of the use of reverse mortgages to convert housing equity into income streams that can support consumption expenditure. This and other differences in methodology aside, Moore and Mitchell’s principal finding – that the median HRS household needs to save 16 percent of its income to preserve its living standard – is consistent with the tables discussed above.
Traditional financial planning is based on targeted saving. The approach requires households to choose future spending or income levels, and to save to meet associated targets. Since setting an appropriate target is a highly complex problem, households are often encouraged to rely on rough rules of thumb. A priori, there is no reason to believe that this method of planning will help the household achieve a smooth and sustainable living standard. An alternative method of financial planning, rooted in economic theory (and embodied in a new software package, ESPlanner), does not require households to undertake complex aspects of planning by themselves. Instead, it derives a saving target for each household by determining its highest sustainable living standard, as well as the levels of saving and life insurance needed to preserve that living standard.
The economic approach to financial planning (as embodied in ESPlanner) generates dramatically different recommendations from traditional approach (as typified by Quicken Financial Planner). Although the differences in saving recommendations are large, they are not systematically high or low. For some households, Quicken Financial Planner recommends far too little saving compared with ESPlanner; for others, it recommends far too much. Differences in life insurance recommendations are also typically large, but they tend to be more systematic, with the traditional approach generally overstated life insurance requirements.
Based on an examination of several thousand households, we conclude that the vast majority of Americans approaching retirement need to save at quite high rates – rates that are much higher than those commonly observed. This conclusion is strengthened once one considers the potential for cuts in social security benefits, gains in longevity, a sudden and significant decline in the stock market, and increases in the costs of nursing home care.
We are grateful to the National Institute of Aging for research support and to Economic Security Planning, Inc. for permitting the use of Economic Security Planner (ESPlanner) in this study. The opinions expressed in this paper are those of the authors and not necessarily those of Boston University, Stanford University, or The Bank of Italy. Sections of this paper draw heavily on Gokhale, Kotlikoff, and Warshawsky (1999) and Bernheim, Forni, Gokhale, and Kotlikoff (2000).
References
Bayer, Patrick J., Bernheim, B. Douglas, Scholz, John Karl. “The Effects of Financial Education in the Workplace: Evidence from a Survey of Employers.” Mimeo, Stanford University, 1996.
Bernheim, B. Douglas. “Financial Illiteracy, Education, and Retirement Saving.” In Living with Defined Contribution Plans. Eds. O.S. Mitchell, and S.J. Schieber. Philadelphia, PA: University of Pennsylvania Press, 1998: pp. 38-68.
Bernheim, B. Douglas, and Daniel Garrett. “The Determinants and Consequences of Financial Education in the Workplace: Evidence from a Survey of Households.” Mimeo, Stanford University, 1999.
Bernheim, B. Douglas, “The Adequacy of Saving for Retirement and the Role of Economic Literacy,” in Retirement in the 21st Century … Ready Or Not…, Washington, D.C.: Employee Benefit Research Institute, 1994, pp. 73-81.
Bernheim, B. Douglas and John Karl Scholz, “Private Saving and Public Policy,” in Tax Policy and the Economy, James M. Poterba, ed., vol. 7, 1993, pp. 73-110.
Caplin, Andrew, “The Reverse Mortgage Market: Problems and Prospects,” this volume.
B. Douglas Bernheim, Lorenzo Forni, Jagadeesh Gokhale, and Laurence J. Kotlikoff. “How Much Should Americans Be Saving forRetirement?” American Economic Review, 90 (2) May 2000: 288-292.
B. Douglas Bernheim, Lorenzo Forni, Jagadeesh Gokhale, and Laurence J. Kotlikoff. “The Adequacy of Life Insurance: Evidencefrom the Health and Retirement Survey.” NBER Working Paper No. 7372, October 1999.
Clark, Robert, and Sylvester S. Scheiber. “Factors Affecting Participation Rates and Contribution Levels in 401(k) Plans.” In Living with Defined Contribution Plans. Eds. O.S. Mitchell, and S.J. Schieber. Philadelphia, PA: University of Pennsylvania Press, 1998: pp.69-97.
Engen, Eric M., William G. Gale, and Cori E. Uccello, “The Adequacy of Household Saving,” Brookings Papers on Economic Activity, vol. 2, 1999, pp. 65-165.
Bernheim, Forni, Gokhale, and Kotlikoff, 7/5/00
Gokhale, Jagadeesh, Laurence J. Kotlikoff, and Mark Warshawsky. “Comparing the Conventional and Economic Approaches to Financial Planning.” NBER working paper, no. 7321, August 1999.
Hubbard, Glen R., Jonathan Skinner, and Stephen P. Zeldes, “The Importance of Precautionary Motives in Explaining Individual and Aggegate Saving,” Carnegie-Rochester Conference Series on Public Policy, 40, June 1994, 59-125.
Kotlikoff, Laurence J., Avia Spivak, and Lawrence H. Summers, “The Adequacy of Savings,” The American Economic Review, 72 (5), December 1982, 1056-69.
Manchester, Joyce, “Baby Boomers in Retirement: An Early Perspective,” in Retirement in the 21st Century … Ready Or Not…, Washington, D.C.: Employee Benefit Research Institute, 1994, pp. 63-67
Mitchell, Olivia S., and James Moore. “Retirement Wealth Accumulation and Decumulation: New Developments and Outstanding Opportunities.” Journal of Risk and Insurance, 65 (3), December 1998: 371-400.
Moore, James, and Olivia S. Mitchell. “Projected Retirement Wealth and Saving Adequacy.” In Forecasting Retirement Needs and Retirement Wealth. Eds. O.S. Mitchell, B. Hammond, and A. Rappaport. Pension Research Council. Philadelphia, PA: University of Pennsylvania Press, 2000.
Venti, Steve and David Wise, “Patterns of Housing Equity Use Among the Elderly,” this volume.
Warshawsky, Mark J., and John Ameriks. “How Prepared are Americans for Retirement?,” in Forecasting Retirement Needs and Retirement Wealth. Eds. O.S. Mitchell, B. Hammond, and A. Rappaport. Pension Research Council. Philadelphia, PA: University of Pennsylvania Press, 2000.
Bernheim, Forni, Gokhale, and Kotlikoff, 7/5/00 Bernheim, Forni, Gokhale, and Kotlikoff, 7/5/00 Bernheim, Forni, Gokhale, and Kotlikoff, 7/5/00 Bernheim, Forni, Gokhale, and Kotlikoff, 7/5/00
| Table 1: Annual Recommendations for | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Saving, Consumption, and Life Insurance | |||||||||||
| (Al and Peg) | |||||||||||
| Year | Al 's | Peg's | Non Tax- | Consumption | Al 's Life | Peg's Life | |||||
| Age | Age | Favored | Insurance | Insurance | |||||||
| Saving | |||||||||||
| 2000 | 50 | 45 | 1,751 | 58,018 | 0 | 468,868 | |||||
| 2001 | 51 | 46 | 2,395 | 58,018 | 0 | 452,345 | |||||
| 2002 | 52 | 47 | 2,591 | 58,018 | 0 | 431,413 | |||||
| 2003 | 53 | 48 | 3,208 | 58,018 | 0 | 408,272 | |||||
| 2004 | 54 | 49 | (18,153) | 49,622 | 0 | 377,605 | |||||
| 2005 | 55 | 50 | (17,775) | 49,622 | 0 | 347,278 | |||||
| 2010 | 60 | 55 | 21,981 | 40,486 | 0 | 148,952 | |||||
| 2015 | 65 | 60 | 23,138 | 44,534 | 0 | 0 | |||||
| 2020 | 70 | 65 | (107,898) | 44,534 | 0 | 0 | |||||
| 2025 | 75 | 70 | (11,297) | 44,534 | 0 | 0 | |||||
| 2030 | 80 | 75 | (11,804) | 44,534 | 0 | 0 | |||||
| 2035 | 85 | 80 | (12,334) | 44,534 | 0 | 0 | |||||
| 2040 | 90 | 85 | (12,863) | 44,534 | 0 | 0 | |||||
| 2045 | 95 | 90 | (18,441) | 44,534 | 0 | 0 | |||||
| 2050 | 95 | (20,162) | 27,834 | 0 | 0 | ||||||
| Source: Authors' calculations, based on hypothetical family characteristics. | |||||||||||
| Table 2: Annual Decomposition of Spending | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| (Al and Peg) | |||||||||
| Year | Al 's | Peg's | Consumption | Special | Housing | Al 's Life | Peg's Life | Excess | Total |
| Age | Age | Expenditures | Expenditures | Insurance | Insurance | Funerals & | Spending | ||
| Premium | Premium | Bequests | |||||||
| 2000 | 50 | 45 | 58,018 | 0 | 32,881 | 0 | 1,052 | 0 | 91,951 |
| 2001 | 51 | 46 | 58,018 | 0 | 32,135 | 0 | 1,080 | 0 | 91,233 |
| 2002 | 52 | 47 | 58,018 | 0 | 31,410 | 0 | 1,123 | 0 | 90,551 |
| 2003 | 53 | 48 | 58,018 | 0 | 30,706 | 0 | 1,130 | 0 | 89,854 |
| 2004 | 54 | 49 | 49,622 | 30,000 | 30,023 | 0 | 1,123 | 0 | 110,768 |
| 2005 | 55 | 50 | 49,622 | 30,000 | 29,360 | 0 | 1,115 | 0 | 110,097 |
| 2010 | 60 | 55 | 40,486 | 0 | 26,322 | 0 | 760 | 0 | 67,568 |
| 2015 | 65 | 60 | 44,534 | 0 | 23,702 | 0 | 0 | 0 | 68,236 |
| 2020 | 70 | 65 | 44,534 | 100,000 | 24,000 | 0 | 0 | 0 | 168,534 |
| 2025 | 75 | 70 | 44,534 | 0 | 24,000 | 0 | 0 | 0 | 68,534 |
| 2030 | 80 | 75 | 44,534 | 0 | 24,000 | 0 | 0 | 0 | 68,534 |
| 2035 | 85 | 80 | 44,534 | 0 | 24,000 | 0 | 0 | 0 | 68,534 |
| 2040 | 90 | 85 | 44,534 | 0 | |||||