National Academies Press: OpenBook

Assessing Knowledge of Retirement Behavior (1996)

Chapter: 2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS

« Previous: 1 INTRODUCTION
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

2
Income and Wealth of Older American Households: Modeling Issues for Public Policy Analysis

Alan L. Gustman and F. Thomas Juster

This paper is concerned with the economic behavior determining the income and wealth of older American households, and with our capacity to analyze the effects of public policies determining their income and wealth. It begins by providing a structure for relating the outcomes of interest to leading behavioral models. This is followed by descriptive statistics indicating the relative importance of the major components of income and wealth. The paper then assesses the current state of models, describing what is known about the behavior of individuals and firms that affects income and wealth determination, what is not known, and what kinds of models and data are needed to do an adequate job of understanding income and wealth outcomes and the effects of policies meant to influence these outcomes. The final part of the paper considers an array of policy changes that might be expected to influence the income and wealth of older households.

CONCEPTUAL OVERVIEW

Figure 2-1 indicates the major components of income and wealth of the older population and the elements of behavior of individuals, of markets, and in the public sector that determine these income and wealth outcomes. Box A repre

The authors are grateful to the National Institute on Aging for research support. Helpful comments by Gary Engelhardt, Anna Lusardi, Andrew Samwick, Jonathan Skinner, Tim Smeeding, and Steve Venti are very much appreciated.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

FIGURE 2-1 Framework for analysis from the perspective of an individual's decisions.

sents a matrix of income and wealth outcomes. The outcomes are delineated by type of income, but the matrix is meant to represent the full dimension of time and cohort effects, as well as the various sources of heterogeneity in outcomes. Box B lists the behavioral decisions. They include the basic decisions of labor supply and savings, as well as other behaviors that must receive attention for a full understanding of wealth and income. Below that in Box C is an array of market-determined outcomes that are taken as exogenous to the individual, such as the features of the pension plan, determined by the employer subject to market constraints. The right-hand box, D, lists some of the basic categories for the

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

policies that will be discussed. The arrows indicate that each box helps to determine elements in other boxes.

A central focus of the paper is the behavior of individuals and the efforts of firms to accommodate the behavior and preferences of individuals. Our aim is to determine whether the analyses of the various dimensions of behavior indicated in Figure 2-1 are sufficiently reliable to support policy analysis, and if not, what additional is research is required to improve the quality of the available behavioral models. As a result of our analysis, we conclude that among the key dimensions of behavior, we have greater confidence in the retirement models for use in policy analysis than in models of saving or pensions. Models of savings and pensions continue to wrestle with behavioral issues that remain unresolved at a more fundamental level than the questions that confront retirement modeling. But we also find that none of the models of behavior in any one area takes sufficient account of the behavior along other dimensions. These behaviors may interact in important ways, and most of the literature ignores these interactions.

COMPONENTS OF INCOME AND WEALTH OF OLDER HOUSEHOLDS

To measure the importance of the major components of income and wealth, we begin with descriptive data on income and wealth outcomes for recent cohorts. The relevant data include the Current Population Survey (CPS), where we use 1992 income data analyzed by Grad (1994); the Survey of Income and Program Participation (SIPP), where we use 1991 wealth data analyzed by Poterba, Venti, and Wise (1994); the Health and Retirement Survey (HRS), where we use income data from 1991 and wealth data from 1992; and the Asset and Health Dynamics Among the Oldest Old (AHEAD) survey, where we use income data from 1992 and wealth data from 1993.

These data sets are complementary. CPS covers the entire age range, as does SIPP, but the income and wealth data from these surveys, especially the wealth data, tend to be underestimates relative to either HRS or AHEAD; in both HRS and AHEAD, new survey technologies have been introduced that result in substantially smaller biases resulting from missing data components. But while HRS and AHEAD appear to have data of somewhat higher quality, they represent particular cohorts and do not include a full range of age distributions. HRS includes the birth cohorts of 1931 to 1941, while AHEAD includes the birth cohorts of 1923 and before.

The data in Table 2-1 provide a useful overview of the sources of income and wealth for households that are in a transition stage between work and retirement—those 65 to 69 years of age. For these households, earnings comprise a little under 30 percent of total income; Social Security comprises another 30 percent, while pensions and income from assets each comprise a bit under 20 percent. On the asset side, if both Social Security income flows and pension

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-1 Sources of Money Income and Wealth for 65- to 69-Year Olds

Income and Wealth Source

Share of Aggregate Income of Aged Units (%)

Average Wealth From Indicated Source as a % of Total Net Worth

Earnings

28.9

 

Social Security

29.9

31.9

Pensions

18.8

19.9

Financial and other assets

18.7

27.8

Public assistance

0.7

 

Other income

3.1

 

Housing

 

20.8

Total

100.1

100.4

 

SOURCE: Income percentages are from Grad (1994, Table VII.1). Grad's figures are based on the 1992 CPS. Wealth percentages are computed from Poterba, Venti, and Wise (1994, Table 1). Their data are from SIPP and are reported for 1991.

income flows are capitalized, the two make up a little over half of total wealth, with the other half coming from conventionally measured net worth—the sum of financial and other assets and housing equity.

From these data we can see the extent of the underestimation problem in the measurement of assets: Assets in Table 2-1 are based on data from SIPP, where conventionally defined net worth (the total of financial and other assets plus housing equity) for these 65- to 69-year-olds had a mean value of roughly $150,000 in 1991; SIPP households aged 55 to 64, a younger age group that typically has smaller net worth than the 65-to-69 group, had a mean 1991 total of about $140,000. In contrast, the HRS data on net worth for households between 51 and 61, where asset holdings would be expected to be smaller still, had a mean 1992 value of approximately $240,000—a figure that is approximately 60 percent larger than the Table 2-1 estimate. And the AHEAD net-worth data, for an age group much older than the SIPP 65- to 69-year-olds and one that would therefore be expected to have a much lower asset level, had a 1993 mean value for conventionally defined net worth of approximately $170,000—higher than the 65- to 69-year-olds in Table 2-1.

Note that the data in Table 2-1 are missing some components. For example, one would generally prefer to include imputed income from housing equity in the income definition and might well include imputed values for services provided by Medicare and Medicaid. The wealth data might include present discounted values for Supplemental Security Income, other welfare payments, and transfers. According to Hurd (1990b, Table 18), these are not negligible sources of income or wealth. It is also worth noting that although the capitalized value of future Social Security or pension benefits, which is included as part of wealth, appropriately reflects the consumption value of these assets, it represents a relatively

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

inflexible source of economic support. To some extent that is also true of housing equity, although here the inflexibility lies as much in consumer choice as in legal or institutional constraints. But it is certainly clear that Social Security wealth cannot be bought and sold, nor can pension wealth, and while housing equity can be bought or sold, most households appear to treat housing equity as an immutable fact of economic circumstances, not an asset that can be used to smooth consumption flows in future years.

An important feature of both income and wealth is their heterogeneity among households. Households are always heterogeneous in the types and amounts of income and wealth that they own, but that heterogeneity is likely to grow as households move into older age groups. And both distributions, especially the wealth distributions, are highly skewed in that medians are small relative to mean values.

The heterogeneity of the income sources received by older households is documented in Tables 2-2 and 2-3. Table 2-2 shows the detailed structure of income from HRS (for age groups in their 50s), and for AHEAD (for age groups

TABLE 2-2 Income Components, HRS and AHEAD, in Thousands of Dollars

 

Age of Household Head

Income Source

51–55a

56-61a

70–74b

75–79b

80–84b

85+b

Earnings

40.6

32.4

4.0

1.5

1.0

0.2

Pensions

1.0

2.8

16.6

15.9

12.4

10.4

Social Security

0.1

0.2

10.2

10.4

9.4

8.2

Private pension

0.9

2.6

6.4

4.5

3.0

2.2

Capital Income

5.1

5.9

2.1

2.9

2.6

1.8

Disability

0.4

0.6

1.1

0.8

0.7

0.8

Welfare

0.3

0.3

1.1

0.8

0.7

0.8

Unemployment

0.4

0.3

1.1

0.8

0.7

0.8

Other

0.2

0.2

1.1

0.8

0.7

0.8

Total

48.0

42.4

24.8c

21.9c

17.8c

15.7c

Household members other than respondent or spouse

4.9

4.7

3.0

2.8

3.7

5.0

Total

52.9

47.1

27.8

24.7

21.5

20.7

a1991 HRS data.

b1992 AHEAD data.

cDerived from an independent question, not from summing the components. The sum of components is generally lower than the above total for technical reasons (mainly the use of unfolding brackets for the total income question).

SOURCE: 1991 HRS data for ages 51 to 61; 1992 AHEAD data for ages 70 to 85+.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-3 Components of Income, AHEAD Sample

Income Component and Age Group

Mean for Sample

Mean if Greater Than Zero ($)

Percentiles ($)

10

25

50

75

90

Wage Income

 

 

 

 

 

 

 

70–74

4,059

18,989

0

0

0

0

8,000

75–79

1,463

14,934

0

0

0

0

0

80–84

1,013

15,586

0

0

0

0

0

85+

162

12,776

0

0

0

0

0

Capital Income

 

 

 

 

 

 

 

70–74

2,089

6,923

0

0

0

240

4,620

75–79

2,941

9,396

0

0

0

318

4,800

80–84

2,605

9,182

0

0

0

300

5,000

85+

1,798

6,987

0

0

0

0

3,600

Social Security Income

 

 

 

 

 

 

 

70–74

10,246

10,259

4,440

6,516

9,480

13,000

15,900

75–79

10,374

10,390

4,800

6,420

9,210

12,696

16,308

80–84

9,384

9,411

4,608

6,000

8,208

11,724

14,712

85+

8,248

8,276

4,236

5,550

7,578

9,492

12,000

Pension Income

 

 

 

 

 

 

 

70–74

6,404

10,660

0

0

2,680

7,218

16,080

75–79

4,539

8,647

0

0

1,200

4,644

11,604

80–84

2,964

6,714

0

0

0

2,964

9,540

85+

2,247

6,366

0

0

0

2,247

6,900

Other Income

 

 

 

 

 

 

 

70–74

1,109

6,126

0

0

0

0

2,400

75–79

773

4,652

0

0

0

0

1,861

80–84

740

4,076

0

0

0

0

1,861

85+

773

4,189

0

0

0

0

2,160

Other family Members Income

 

 

 

 

 

 

 

70–74

3,026

19,870

0

0

0

0

13,162

75–79

2,839

19,719

0

0

0

0

13,162

80–84

3,689

22,379

0

0

0

0

13,162

85+

4,979

21,386

0

0

0

6,571

13,162

Total Family Income

 

 

 

 

 

 

 

70–74

27,778

 

8,280

12,460

19,304

29,723

49,200

75–79

24,754

 

7,428

10,374

16,539

25,004

42,374

80–84

21,515

 

6,528

9,398

13,992

22,561

38,352

85+

20,723

 

6,564

8,400

13,000

20,476

35,064

in their 70s and 80s). Total household income is of course substantially larger for the HRS sample than for AHEAD, partly because of cohort differences in earnings but mainly because of age differences. The HRS households are about 30 years younger than the AHEAD households that are ages 80 and over, and thus there has been 30 years worth of economy-wide improvement for the HRS cohort compared with the AHEAD 80 and older cohort. In addition, the AHEAD data

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

reflect the fact that the replacement of earnings by the sum of Social Security benefits and pensions is at much lower than a one-to-one ratio for most households.

The income component data in Table 2-2 do not contain any surprises. Earnings are the dominant source of income for the HRS cohorts in their 50s, while pensions, especially Social Security, are the dominant source of income for households in the AHEAD cohorts. Earnings are relatively unimportant for households in the AHEAD age range, and the pattern of private pension income, which is about three times as high among AHEAD cohorts in the 70- to 74-year-old group as in the 85 and older group, is explainable both by cohort differences and by the growing incidence of widowhood among older households in the AHEAD cohorts. Finally, earnings of household members other than the HRS or AHEAD survey respondent and spouse are an important source of family income, especially for the oldest AHEAD cohort (those 85 years of age or older).

The heterogeneity of income components for the AHEAD cohort shows up very clearly in Table 2-3, which contains sample means, means for households with positive income in a particular category, and percentile distributions. Not surprisingly, hardly any AHEAD age groups have wage income, all the way up to the 90th percentile. Equally surprising to some, capital income is zero for the entire lower half of the AHEAD distribution, is only a few hundred dollars for AHEAD cohorts up as high as the 75th percentile, and becomes a substantial sum only when we get around the 90th percentile, where the amounts are several thousand dollars rather than several hundred dollars. The only income sources that are at all widely distributed among AHEAD households are Social Security income, which virtually everyone receives, and pension income, which is received by the upper half of the distribution in the younger age cohorts and by the upper quarter in the older age cohorts. For each age category and percentile, pension income exceeds Social Security income in the AHEAD sample for only one age cohort and one of the percentiles shown—the 90th percentile for the 70-to 74-year-olds. As we note later, it looks as if this pattern will be a bit different when the HRS cohorts get to be in the AHEAD age range, although that will depend in part on changes over time in the proportion of jobs providing pensions and in the proportion of pensions that contain survivors' rights.

The heterogeneity in economic status among older households is even more pronounced when we examine the data on net worth. Tables 2-A1 to 2-A7 in the Appendix contain estimates of total net worth, net worth in the form of housing equity, and net worth less housing equity for various HRS and AHEAD classifications of households. For HRS, we divide the sample into couples, single men, and single women (Table 2-A1), by racial/ethnic groups (blacks, Hispanics, and all others including whites, Table 2-A2); and by 1991 income (Table 2-A3). For the AHEAD sample, we show data for couples and singles in Table 2-A4 (total net worth) and Table 2-A5 (net worth excluding home equity), as well as for racial/ethnic groups in Tables 2-A6 (total net worth) and 2-A7 (net worth exclud-

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

ing home equity). For the AHEAD data, we show separate estimates for four age groups: 70 to 74, 75 to 79, 80 to 84, and 85 and up.

The principal message from this set of net-worth tables is that wealth is highly unevenly distributed among the older population: both in the HRS and the AHEAD samples, married couples have substantially higher levels of wealth than single men or single women, even after implicit correction for household size; the disparities are substantially larger for net worth less housing equity than for total net worth, since housing equity itself is somewhat more evenly distributed than most other assets; minority households have substantially fewer assets than whites by an order of magnitude of 4 or 5 to 1 in mean values, and an order of magnitude more like 10 to 1 for net worth less housing equity.

For the AHEAD data, there are of course substantial differences by age group as well as by family composition and racial/ethnic group. In general, older households have smaller net worth, other things equal, although the differences by age are surprisingly small for couples up through the age of 80 to 84. The most striking disparities in the AHEAD data are those shown by the tabulations of net worth excluding home equity, both by family composition and by racial/ethnic group, divided according to age group. For the family composition data, fully half the single women had net financial assets excluding home equity of under $ 10,000 regardless of age—the 70-to-74 group and the 85 and older group have just about the same (minimal) assets. Single men are a little better off, but fully half of this group have under $20,000 of net worth excluding home equity. For couples, in contrast, the median net worth excluding home equity is a little over $55,000 for the 70-to-74 age group and is still about $20,000 for the 85 and older age group. As would be expected, there are some very wealthy subcategories of households in the sample: AHEAD households in the 90th percentile among couples have over $400,000 of net worth excluding home equity, and almost $600,000 net worth in total, for those with heads age 70 to 74. Even for couples age 85 and up, AHEAD households have over a quarter of a million dollars in net worth excluding home equity at the 90th percentile and over $400,000 of total net worth. For minority households in the AHEAD sample, it is essentially correct that fully half of all black and Hispanic households have close to zero net worth excluding home equity, regardless of age, and even at the 75th percentile, neither black nor Hispanic households have as much as $20,000 of net worth excluding home equity in any of the AHEAD age groups. To all intents and purposes, most minority households can be thought of as having negligible financial asset holdings in old age.

Income Distribution Issues

The income and wealth data show substantial disparity among households, and the disparities appear to be a bit greater for households 70 and over than for

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

others. One way to look at issues of income distribution is to examine the economic circumstances associated with private pensions.

Two alternative models might be contrasted. In one model, the existence of a private pension, since it involves a cost to the employer, provided less current income during the working years. On that model, households with substantial pensions would have had less current income than other comparably situated households; during retirement, they would be expected to have more pension income than others, but less income from capital and from Social Security benefits.

In an alternative model, the market is such that jobs that carry pensions are also apt to carry higher current income than other jobs, perhaps because only those with high wage rates want pensions, given the tax advantages of pensions and the low Social Security replacement rate for high wage jobs. Hence households with pensions will have more favorable economic circumstances generally as they move toward retirement. On that model, households with substantial pension income would also be expected to have substantial capital income relative to other households (because of their higher current income while working) and to have higher Social Security benefits (again because of their more favorable current income while working). In short, an important issue is, do households with jobs that carry substantial pensions have offsetting differences in other sources of retirement income, or do the differences tend to cumulate—those with pensions having more of other forms of retirement income as well?

Tables 2-4 and 2-5 show comparisons for both the HRS and the AHEAD sample. For HRS (Table 2-4), we contrast households in which both respondent and spouse have jobs with pension rights, households where one has a job with pension rights and the other a nonpension job, and households where neither has

TABLE 2-4 Earnings and Capital Income by Pension Status, HRS Households Working for Pay and Not Self-Employed

 

 

Mean Values ($000)

Whether Pension Income

% of Cases

Earnings

Capital Income

Total Income

Singles

 

 

 

 

Yes

66.8

29.0

2.4

37.1

No

33.2

14.9

1.3

23.3

Couplesa

 

 

 

 

Both Have

57.1

64.4

4.0

75.0

One Has

34.6

49.7

2.9

62.1

Neither Has

8.3

36.7

5.3

49.2

NOTE: All categories weighted by the HRS household population weight.

a Both spouses in the couple households are working.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-5 Pension, Social Security, and Capital Income by Whether Pension Income and Age, AHEAD Householdsa

Whether Pension Income

% Having Pension Income

Mean Pension Income ($)

Mean Social Security Income ($)

% Having Capital Income

Mean Capital Income ($)

Mean Total Income ($)

All cases (N=5,457)

 

 

 

 

 

 

Yes

50.6

8,915

10,981

39

2,869

25,908

No

49.4

0

8,532

20

1,722

15,507

Age 70–74 (N=1,818)

 

 

 

 

 

 

Yes

58.3

10,340

10,964

37

2,290

28,237

No

41.7

0

8,815

20

1,979

18,053

Age 75–79 (N=1,415)

 

 

 

 

 

 

Yes

50.0

8,504

11,385

43

3,727

26,626

No

50.0

0

9,041

19

1,795

16,047

Age 80–84 (N=1,120)

 

 

 

 

 

 

Yes

42.3

6,803

10,404

39

3,443

21,092

No

57.7

0

8,200

20

1,866

14,011

Age 85+ (N=794)

 

 

 

 

 

 

Yes

34.1

6,099

9,011

39

2,773

19,064

No

65.9

0

7,202

19

1,131

11,148

NOTE: All categories weighted by the AHEAD household population weight.

a All income is for respondent and spouse only.

a job with pension rights although both have jobs. We also show data for singles who work, with and without pensions on their job. We tabulate current earnings for these HRS households, and also tabulate capital income. For the AHEAD households (Table 2-5), we divide the sample into households receiving some pension income versus those receiving none, and tabulate pension income, Social Security benefits, capital income, and total income for each of the AHEAD age groups.

It is clear enough from the data, especially the AHEAD data, that favorable economic circumstances cumulate rather than offset. Both for the HRS age range and for the various AHEAD age ranges, households either expecting or receiving pension income have substantially higher nonpension income (capital income and job earnings in the case of HRS households, Social Security benefits and capital income in the case of AHEAD households). For the AHEAD sample, where the differences are clearest and the analysis is least ambiguous, households receiving pension income have close to twice as much total income as other households.

Finally, we show a mapping of the relationship between health status and both income and wealth for HRS households (see Appendix, Figures 2-A1 to 2-A4).

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Although these data come from a cross section and therefore are not helpful on questions of causality, the strength of the relationship between health and wealth or health and income is quite remarkable, whether measured by relationships involving means or medians. For the net-worth measure, couples where both spouses are in either excellent or very good health have net worth in the area of $400,000, while households where both spouses are in fair or poor health tend to have net worth of less than a quarter of that. These differences are even sharper for median net worth and are almost as large for either mean or median household income. Interestingly enough, if there is any difference in the relationship between household financial variables and health for the male or female spouse, it appears that the health of the female spouse is more systematically related to the household's financial well-being than the health of the male spouse.

MODELS OF INDIVIDUAL AND FIRM BEHAVIOR EXPLAINING RETIREMENT INCOME AND WEALTH

This section briefly considers what we know and do not know about the dimensions of behavior that are central to an understanding of how policies affect retirement incomes and retirement wealth. The discussion covers labor supply decisions, savings behavior, pension plan determination, and the determination of Social Security income at the level of the individual. Also discussed are the behaviors determining family structures and transfers, and the demand for housing.

Income from Earnings Based on Labor Supply Decisions

To project earnings, it is necessary to explain patterns of labor force participation among older workers on jobs offering different wage rates.1 Thus, an important part of the approach to understanding the determination of the earnings of older individuals, and the effects of public policies on earnings, is to apply a conventional intertemporal model of labor supply and to use the labor supply outcome together with the wage to determine earnings.

The conventional model for explaining labor force behavior into retirement is dynamic. Subject to a series of constraints, including the wage offer for full-time work, the wage offer for part-time work, and the rate of pension accrual (including its option value), as well as the corresponding rate of accrual in Social Security benefits and other factors, the individual reaches a decision on whether to continue working full time, part time, or not at all. In this approach, the parameters of a utility function are estimated so as to maximize the likelihood of observing the sequence of outcomes realized for each individual, subject to the constraints created by the elements of the opportunity set (Fields and Mitchell, 1984; Burtless and Moffitt, 1984, 1985; Gustman and Steinmeier, 1986b;

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Berkovec and Stern, 1991; Rust, 1990; Stock and Wise, 1990a, 1990b; Lumsdaine, Stock, and Wise, 1990, 1992a, 1992b).

Different versions of the model have proved capable of explaining some major features of data reflecting labor supply of older individuals and their retirement. These include the spikes in the retirement hazard, most commonly at age 65, but at 55 for some and 62 for others. The models also explain the fact that two thirds of retirees proceed directly from full-time work to retirement, the relatively short duration of partial retirement, the coincidence of retirement decisions by husbands and wives, and some (about a quarter) of that portion of the trend to earlier retirement observed from 1970 through the mid-1980s (see Gustman and Steinmeier, 1986b, 1994a; Anderson, Gustman and Steinmeier, 1994).

Recent contributions have been made on a number of dimensions, adding to the richness of the dynamic specification, considering reaction to risk as well as expected values, incorporating interdependence of decisions at the level of the family, entertaining the possibility of retirement behavior that is influenced by liquidity constraints, and enriching the array of nonfinancial considerations employed in the model.2

Despite all of this work, there are many basic questions about retirement behavior that have not yet been addressed. Although we have made important progress in improving the dynamic and stochastic structures of retirement models, considerable work remains before we incorporate in a single setting the full array of behaviors in relation to uncertainty, mistakes, revisions, surprises, and random shocks. For example, Rust (1990) and Berkovec and Stern (1991) omit any consideration of the relation of pension incentives to labor market outcomes, by either focusing on the portion of the sample without a pension or simply ignoring the existence of the pension. In estimating these models, it will be important to measure pension incentives using data collected from worker descriptions as well as from firms.3

Findings based on the HRS by Brown (1993) suggest that some workers who accept an offer of a retirement window from one employer actually continue working, sometimes accepting another full-time job. This implies that some elements of mobility models (e.g., as in Allen, Clark, and McDermed, 1993; Gustman and Steinmeier, 1993) will have to be incorporated within retirement models that, to date, have analyzed the departure from a full-time job as if it always involved a substantial cessation of economic activity. Brown's findings are especially troublesome for studies that define retirement as exit from the payroll of a single firm. What is required for the models of Lumsdaine, Stock, and Wise (1992a, 1992b), for example, is to incorporate wages from work after retirement, as in the study of retirement from the Air Force by Ausink and Wise (1993), but allowing for the joint choices of work or retirement after leaving the job.

One test that is natural to impose on available studies is whether a model satisfactorily explains major social trends and policy outcomes. The weakness of

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

available labor supply and retirement models is reflected in the limited ability of such models to explain the trend to earlier retirement satisfactorily. For example, for the period from 1970 through the mid-1980s, simple decomposition analyses that would explain the trend on the basis of changes in the composition of workers' demographic characteristics or employment cannot successfully explain the trend to earlier retirement (Ruhm, 1992; Anderson, Gustman, and Steinmeier, 1994). With the use of structural retirement models, some of the trend (about a quarter of the trend in the 1970s and 1980s) can be explained by the changes in pensions and Social Security.4 We still do not have enough confidence in the implied substitution and income effects from these models to have a firm handle on the causes of these trends.5 Also disturbing in this context are the strong implications of the pension literature that pensions are designed to meet the preferences of covered workers (Gustman, Mitchell, and Steinmeier, 1994), so that even those portions of the trend to earlier retirement that are associated with changing Social Security and pension benefit formulas may reflect the effects of changing tastes, rather than the changing incentives in the opportunity set.

Cohort-specific characteristics may be important determinants of retirement and of retirement trends. Some potential candidates for cohort effects include the change associated with the increasing participation of women throughout their lifetime, the changing structure of the division of labor in the household (more male hours, fewer female hours; see Juster and Stafford, 1991), and the changes in employer attitudes induced by the different histories and expectations about labor force commitment of the members of different cohorts. The literature does not, however, do a very good job of isolating cohort effects when explaining trends in retirement. Only time will tell whether these changes are adequately represented by differences in measurable characteristics among families or whether behavior will differ among subsequent cohorts because of cohort-specific effects.

Efforts are just beginning to expand the methodology available for examining the interdependence of family retirement decisions in a structural model, for incorporating measures reflecting job conditions or difficulty of work, and more generally for understanding the relation of financial measures, and of imperfections in capital markets, to retirement outcomes (see the discussion in Hurd, 1993, and Rust and Phelan, 1993, for example).

The roles of imperfect information, complex calculations, and responses to uncertainty remain to be sorted out and satisfactorily modeled for inclusion in behavioral analyses of relevant policies. An extensive set of questions in HRS on expectations and attitudes about risky choices has been used to analyze relevant aspects of decision making in the context of the complex choices facing the potential retiree (Barsky et al., 1993), but this information has not been incorporated in a structural retirement model.

Many questions also remain about the relation of behavior to expectations. Formal models of the retirement decision assume that workers make decisions in

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

each period to maximize utility over their remaining lifetime. While some theoretical progress has been made in modeling such decisions in the presence of uncertainty, empirical work inevitably pretends that workers either know the value of future income streams associated with various choices or make rational expectations forecasts, and that they also know the ''length of the planning horizon"—that is, how long they will live. The analysis of the relation of retirement expectations to incentives is encouraging (Hurd and McGarry, 1993a), but many questions remain unanswered. Previous research has suggested that there will be some whose expectations are unreasonable (Mitchell, 1988; Gustman and Steinmeier, 1989; Bernheim, 1988, 1989). In particular, there is a tail of the distribution with individuals who expect their benefits to be larger than is called for by their firm's pension plan, given their work and earnings history. More generally, we need to know how expectations are formed, how they are revised, and what differences there are in the formation of expectations and behavior for those who correctly report their constraints and for those who do not.

When the role of health status is measured in structural retirement models, it is measured by relatively direct questions about whether the individual suffers from health problems that impede work or other activities.6 Researchers are aware that reported health status may involve an ex post rationalization, with an individual reporting he retired owing to ill health when that was not the motivating factor. This problem will be substantially remedied once panel data become available for estimating structural retirement models with HRS. The medical information is sufficiently detailed that it will be possible to isolate the effect of poor physical health from that of self-rationalization in the health measures used.

Savings and Wealth Determination

Savings and consumption analysis are major areas of economic research in both microeconomics and macroeconomics. From the perspective of aging research, analyses pertaining to life cycle, precautionary, and bequest motives are of particular interest.7 Sophisticated econometric models have been estimated with microdata on the basis of equations derived with each of these behavioral motivations in mind, and recent work has attempted to explain wealth and savings outcomes on the basis of more than one of these behavioral motivations. There has also been useful empirical research that imposes somewhat less structure but focuses on major features of the data related to cohort and age effects. In addition, there is a line of research that questions whether savers can make the sophisticated calculations called for by dynamic stochastic models of savings.

Life-Cycle Analysis

A basic prediction of the simple life-cycle model is that with a rising mortality hazard, once the sum of the time preference and the mortality hazard exceed

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

the interest rate, consumption and wealth will fall with increasing age. A good deal of relevant evidence has been collected and is presented in Hurd (1990b), who argues that if panel data are used and are corrected for the effects of mortality, profiles of bequeathable wealth do turn down at plausible ages and are consistent with the life-cycle model.8

A number of anomalies have encouraged researchers to expand the model of savings beyond a perfect capital market/life-cycle specification. Nevertheless, the forward-looking consumer continues to characterize the central agent of many models of savings behavior.

Precautionary Models of Savings

The individual faces a wide range of risks that might promote saving. If there are liquidity constraints or other factors creating incomplete markets for insurance, then a precautionary motive for saving may be important. If there are high rates of time preference, then the precautionary motive may dominate savings behavior even if the capital market is perfect. In precautionary models where agents display prudence, a precautionary motive may lead them to choose not to borrow, so they appear to be credit constrained.

A basic fact puzzling researchers of savings is why such a large number of individuals reach retirement with little or no savings. This raises the question of whether the predominance of an alternative motivation for saving beside the life-cycle model, and in particular precautionary saving, might account for the wide heterogeneity in observed outcomes. Other researchers have been motivated by evidence that suggests to them that the young don't borrow, the old don't decumulate, and consumption growth is positive even when the interest rate is low or negative in a certain period (see Zeldes, 1989a). An analysis of precautionary motives must explain both how expectations are formed and what the reaction is to risk. In microlevel studies there have been analyses of the properties of some of the major risks facing the individual, including pensions, earnings, length of life, and ill health (Skinner, 1988; Carroll, 1992; Guiso, Jappelli, and Terlizzese, 1992; Samwick, 1993b, 1994; Hubbard, Skinner, and Zeldes, 1994b). In the context of these analyses, it is also hoped that it will be possible to solve such puzzles as why consumption tracks income so closely (see Hall and Mishkin, 1982; Carroll and Summers, 1991), a phenomenon that suggests the importance of liquidity constraints (Zeldes, 1989a), and perhaps also to better understand how the saving motivation is affected by the availability of insurance (Hubbard, Skinner, and Zeldes, 1995).

Researchers are investigating the best way to categorize and measure the effects of and the reactions to risk, distinguishing risk aversion from prudence, analyzing their properties, and exploring the impact of these different features of preferences (Kimball, 1990, 1993).9

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Bequest Motive

An alternative motivation for saving is to provide a bequest for one's heirs. The bequest motivation may be simple, as in dividing a fixed sum among one's heirs, or it may be complex, as in leaving a benefit in accordance with need or specifying a strategic arrangement in exchange for certain services from one's children (see, e.g., Becker, 1991; Bernheim, 1991; and Bernheim, Shleifer, and Summers, 1985). The fact that one leaves a bequest does not mean the bequest was intended or that it was the amount that would have been delivered in a world with perfect foresight. In the absence of efficient annuity markets or insurance by families, individuals may be forced to underconsume to avoid outliving their assets (Kotlikoff and Spivak, 1981; Davies, 1981; Abel, 1985). Nor is it optimal for everyone to want to leave a bequest—for example, children may be better off than their parents. Also, it is possible that consumption declines with age and bequests follow because, for some, the capacity to consume declines with age (Borsch-Supan and Stahl, 1991).

The evidence on the operation of a bequest motive is very mixed. The question is how important the bequest motive is relative to other motives for saving, including the basic life-cycle motivation. Consider the contradictory findings in a recent paper by Smith (1994). On the one hand, he finds that HRS respondents who believe that leaving an inheritance is very important have accumulated significantly more wealth ($85,000) than those who do not think that bequests are important.10 On the other hand, using panel data from the Panel Study of Income Dynamics (PSID), he finds, consistent with Hurd (1987),11 no linkage between savings and number of children ever born. Kotlikoff and Summers (1981) suggest that the bequest motive accounts for the bulk of observed wealth, while Modigliani (1988) disagrees (see, however, Kotlikoff's, 1988, reply).

Although the life-cycle model considers the calculations of the individual or couple in isolation, research on bequest motives involves adopting intergenerational or family-based models for analyzing consumption and savings. There is interest not only in the extent to which bequests are a motivation for savings, but also in what determines the amounts of bequests and the division of bequests among children and others. (Analysis of the effects of intergenerational linkages include Becker's analyses collected in Becker, 1991; Barro, 1974; and Bernheim, 1991.)

Research Integrating These Motives

Efforts at integrating these various explanations for savings do not yet involve estimating full structural models. One approach is to create simulation models on the basis of parameters obtained from original estimation as well as from other independent sources and to use the models to simulate the paths of

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

asset accumulation. For example, Hubbard, Skinner, and Zeldes (1995) attempt to integrate a precautionary model of savings with a life-cycle model, while including the effects of asset-based means testing in social insurance programs. A related approach assumes that high rates of discounting govern the life-cycle motivation, but there also is a precautionary motive operating in the context of a model with liquidity constraints, and creating the need for buffer stocks against adverse events (see Carroll, 1992; Carroll and Samwick, 1994; Samwick, 1994). The relative weight given in these models to life-cycle retirement savings and precautionary savings varies. Nevertheless, the models do incorporate responses to risks such as those from earnings variation, health outcomes, and uncertain length of life. These models do seem capable of reproducing important features of the data, in particular, low levels of savings in early years, differences in savings among income groups, and savings and then dissaving through the life cycle.

Research that Imposes Only Limited Structure

Some empirical analyses of asset composition rely on specifications that impose as little structure to the underlying model and error terms as possible, employing considerably less structure than some formal empirical models of life cycle and precautionary savings. (See, e.g., the models of Venti and Wise, 1987, 1990, and Poterba, Venti, and Wise, 1993, for studies that attempt to separate cohort and age effects on savings.) When not much structure is imposed, that limits the ability to predict the effects of policy changes; if the structure that would otherwise be assumed is incorrect, this limitation is appropriate. The alternative is to impose a specific functional form, a process that will reveal more of the key parameters required for policy analysis, on the assumption that the structure that is imposed is correct. Otherwise, the imposition of too much structure will create bias in the estimated coefficients and will foster misleading policy analysis. Because of the continuing debate about the behavior underlying the savings decision, the extent to which these models should impose a particular structure continues to be a subject of disagreement (Venti and Wise, 1993; Engen and Gale, 1993).

In some sense those who do not impose a great deal of structure on their empirical estimates have a different methodological perspective. Especially when investigating the effects of well-defined policy changes, they are applying a quasi-experimental approach in which their major aim is to distinguish the effects of that policy, such as the adoption of rules permitting Individual Retirement Accounts (IRAs) or 401(k) savings plans. They are not, however, attempting to isolate the effects of the components of the policy. Often those who have a tight structural model in mind do not estimate the full structural model but a reduced form of one type or another. When that is the case, the parameter estimates obtained may be subject to a number of interpretations. Nevertheless, they may

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

provide useful information even if the full structural model is incorrectly specified.

Research Arguing that Full Optimization is Unlikely

Thaler (1994) argues that a number of findings in the savings literature are inconsistent with the predictions from leading models (see also Deaton, 1991; Zeldes, 1989a, 1989b). This in turn leads him to suggest that the leading models are probably not accurately describing behavior. Among the reasons is the difficulty of the assumed optimization. The assumed optimization requires the solution to a dynamic programming problem that when specified to include the array of risks that are encountered, requires a supercomputer to solve and is too hard even for as if behavior, especially because the behavior is not repeated but occurs only once in the lifetime. Moreover, because the problem is so complex, easy rules of thumb do not bring us close to the right answer. Bounded rationality and lack of self-control lead Thaler to suggest an approach to savings based on mental accounts, in which substitution among types of savings meant for different purposes is highly imperfect. Bernheim (1993) expresses doubts that the population is sufficiently economically literate, and he has discussed the importance of providing adequate information through the Social Security system. There is not sufficient understanding of the limitations in our computational abilities to predict the effects of providing potential retirees with increased information, as the new Social Security Administration initiative to inform individuals as to their entitlements will do.

Continuing Controversy in the Savings Literature

A major question in the savings area is how to reconcile the findings and models that have been developed with different behavioral motivations in mind. A number of the approaches to retirement savings summarized above are fundamentally inconsistent with one another. In addition, the relevant facts remain in dispute.12 When policy innovations arise, such as the availability of IRAs and 401(k)s, either new savings is generated or it is not. (For discussion of the continuing debate on this topic, see Venti and Wise, 1987; Poterba, Venti, and Wise, 1993, 1994; Gale and Scholz, 1994; Engen, Gale, and Scholz, 1994. For a critique of the last study, see Bernheim, 1994.) Either those with pension or housing wealth reduce their holdings of financial assets proportionately, or they do not.13 Without resolving these questions, which requires resolving ongoing controversies in the literature, we are in no position to judge the effects of a number of important policies that will affect savings.

The difficulties that modelers have faced in explaining these facts are widely appreciated. Some argue that it is a matter of integrating the various motivations and constraints into a single framework (see Hubbard, Skinner, and Zeldes, 1994a,

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

1994b). Some suggest that many of the outcome differences may be due to unmeasurable differences in taste and that selection issues will be difficult to unravel (Bernheim, 1994). Others suggest that in analyzing savings behavior, we are at the limits of the usefulness of a model that assumes full information and complete and rational decision making (Thaler, 1994). The question of the importance of each of these motivations for savings has yet to be resolved.

With regard to the analysis of savings, a fundamental task is to reconcile the competing explanations for the observed behavior. The puzzling differences in savings outcomes within the population, the failure to explain low savings rates and the counterintuitive findings on the substitution among different types of non-tax favored savings, tax-favored savings vehicles such as IRAs, 401(k)s, and pensions, and other savings vehicles, provide an opportunity to reconcile these results. A basic question in the savings literature is whether it will be possible to explain observed behavior with a significantly modified life-cycle model that integrates some of the other leading explanations for savings behavior. The weights given to these various motivations are unclear and at times appear inconsistent. It is possible that a different mix of theories may be required at different points in the income distribution to explain the variety of behaviors observed for those with different incomes. Alternatively, it may be necessary to pay much greater attention to the difficulties of making life-cycle calculations, incorporating rules of thumb and importing ad hoc or nonconventional explanations for imperfect substitutability among various types of savings instruments, the very high apparent rate of time preference among younger workers, the wide heterogeneity in savings, and the high frequency of zero savers.

Pensions and Social Security

At the Level of the Individual

At the level of the individual's decision, Social Security and pension income are the result of joint choices determining labor supply and benefit acceptance. The provisions of Social Security and pensions are taken as exogenous to the choices the individual will make. The pension benefit coverage, pension formulas, and earnings histories of each individual determine potential pension incomes in retirement, conditional on the choice of retirement date and on the timing of pension and Social Security acceptance (see Burkhauser, 1979, and Rust and Phelan, 1993, among others).

It is a straightforward procedure to apply retirement models to explain the effects of changes in Social Security and pension policies on retirement outcomes and on incomes from pensions and Social Security in retirement.14 However, there are some questions that arise from findings that suggest both that individuals may not fully understand the rules governing Social Security and that they

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

may be liquidity constrained; the second possibility has been recognized but is not incorporated in most dynamic life-cycle models of labor supply.15

Analysis of the incentives from pension plans suggests that if the pension formula is not actuarially fair, this may have a significant effect on the ultimate level of pension wealth. Changes in pensions and Social Security may induce further changes in savings behavior. But to predict the second-order effects of these changes will require a greater understanding of savings behavior than we now have.

Pension Plan Determination at the Level of the Firm

A basic building block in the conventional model explaining the demand for private pensions is the tax-favored status of the pension. Even the simplest models of why Firms adopt pensions predict that workers like to substitute the tax-favored savings available under pension plans for private savings and that it is in the interest of firms to accommodate this demand (see, e.g., Woodbury, 1983; Woodbury and Huang, 1991). Early studies of the reaction between pensions and savings suggested greater savings by those who were covered by pensions (Cagan, 1965; Katona, 1965). A survey by Munnell and Yohn (1992) of studies conducted in the 1970s and 1980s suggests that there is substitution but that it is imperfect. More recent studies using actual rather than expected pension amounts or coverage suggest that the substitution is weak or nonexistent. If lifetime income is controlled for, those with higher employer-provided pension assets do not exhibit lower personal retirement saving.16

There is a substantial literature arguing that the firm has other motivations for pensions. Given its goal of maximizing profits subject to constraints from the production technology and factor supply curves, and operating within the implicit contract, the firm is hypothesized to choose parameters of pension plans to allocate compensation optimally among wages, pensions, and other fringes to influence employment, worker productivity, and other dimensions of costs.

Undoubtedly the tax-favored status of pensions contributed importantly to their spread in the post-World War II period (Ippolito, 1986). It is therefore quite surprising that the evidence is ambiguous that those with pensions reduce their savings in other forms. There are other questions about the behavioral mechanisms driving the determination of the coverage, terms, and amounts of pension savings, choice of plan types, explanations for pension backloading, and relationships of pensions to wages, turnover, and other dimensions of labor quality. Many of the human resource motivations that are said to underlie or buttress the demand for pensions are not consistent with the data.

Consider the questions that may be raised about the elaborate models of implicit contracts in which pensions are used as a tool of human resource policy to screen workers on the basis of unmeasured ability and to prevent shirking. Elements of these models that are not consistent with the data include the pre-

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

sumption that pensions provide a strong incentive against mobility in the years following hire and initial training (the incentive is really quite weak), that mobility is lower from pension-covered jobs because of pension backloading (mobility is lower from pension-covered jobs whether the pension is defined benefit and backloaded or defined contribution and not significantly backloaded), and that workers are inhibited from shirking because they are afraid they will lose their backloaded benefits if fired from a job offering a defined benefit plan (workers have only a modest understanding, if that, of the incentives in their pension plans). (For discussions of these and related predictions from conventional models of pensions that attribute the attraction of firms to pensions to their human resource properties, see Gustman and Steinmeier, 1989; Gustman and Mitchell, 1992; Gustman, Mitchell, and Steinmeier, 1994; and Gustman and Steinmeier, 1995a.)

Anomalies appear in studies with overidentifying restrictions. In turnover models we find that constraints requiring identical coefficients for wage and pension terms are violated and that the estimated coefficients differ by orders of magnitude. Within models of compensating wage differentials for pensions, the findings are not robust. And it is apparent from examining empirical work that identification has sometimes been forced, for example, by instrumenting on what are clearly endogenous variables—using some pension characteristics as instruments in a model that is designed to explain pension and wage outcomes (for details, see Gustman, Mitchell, and Steinmeier, 1994).

The conventional models also generate other predictions that do not accord with the data, raising concern for our ability to predict how private pensions will respond to changes in pension policies, in tax policies, or in the Social Security system.

Among the empirical regularities that pension studies attempt to explain are the basic result that compensation accrual and productivity do not correspond in each year of attachment and the existence of other unique labor market institutions, such as mandatory retirement provisions, which were commonplace before they were banished by law. Researchers have also tried to understand why defined benefit pensions are backloaded (that is accrue more in later than in earlier years), why union pensions are underfunded relative to nonunion pensions, why workers are less likely to leave pension-covered jobs than jobs without pensions, why firms grant post-retirement benefit increases, and other puzzles (see, e.g., Lazear, 1979, 1983; Ippolito, 1983, 1985a, 1985b, 1986, 1987). They have also attempted to generate predictions from models of long-term worker attachment and to test those predictions (see, e.g., Hutchens, 1986, 1987; Stern and Todd, 1993). Other efforts have described the differences in pension outcomes among demographic, firm, and industry groups and by unionization and other factors. Among the major differences in pension outcomes are differences in plan types, that is, whether the plan is defined benefit or defined contribution.17 Still other studies have focused on documenting the trends in pension outcomes

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

and explaining the reasons for these trends. (Relevant studies are reviewed in Gustman, Mitchell and Steinmeier, 1994.) After rising during the 1960s and 1970s, pension coverage ceased to grow in the middle 1980s. Now there is evidence that the upward trend in coverage, especially for defined contribution plans, may have resumed in just the last year or two (Employee Benefit Research Institute, 1993).18

Despite all of the efforts to explain the regularities in pension-related outcomes, most testing looks for partial relationships in the context of multivariate single-equation models. There has been no structural analysis of firms analogous to that available for analyzing behavior and policies from the perspective of the individual. This greatly limits our ability to use the available empirical work on pensions to predict the effects of pension regulations and policies on income and wealth outcomes. A basic reason there has been no structural analysis of firms is that the data are not available at the level of the firm. (For a discussion of the data that are available, the shortcomings in the data, and what would be required to support policy analysis, see Gustman and Mitchell, 1992.) And with all of the remaining questions about the importance of the competing motivations determining pension design, it is premature to impose a comprehensive model.

In sum, the pension literature is very far from generating empirical estimates of a reliable behavioral model. We certainly have no model with sufficient structure that it can be used to predict the effects of pension policies on the basic pension outcomes, including pension amounts, plan characteristics, insurance features, or other outcomes that would be useful in understanding how pension policies will affect retirement incomes and wealth. Nor can we predict the effects of these policies on wages.

Research on Family Structures and Transfers

There is some research linking earnings, wealth, and savings behavior to family structure. Earnings equations consistently find that earnings are higher among married individuals. With regard to savings and wealth, Smith (1994), for example, finds that married couples have higher assets and savings than unmarried individuals, even after standardizing for differences in incomes and using longitudinal data to distinguish the effects of selectivity of marital state, in which lower income families are more likely to dissolve. Thus, the need for help in old age is likely to be affected by the history of marital status. And, of course, the availability of help in old age is going to depend on whether unmarried individuals were ever married and whether they had children.

Living arrangements in old age are different from those at younger ages, reflecting not only the course of the life cycle as children leave home, but also the effects of mortality. Women are more likely to survive than men, so older households are more likely to include a single woman than a single man. When the survivor is a woman, however, the household is likely to be poorer. (See

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Hurd and Wise, 1989, and Burkhauser, Holden, and Feaster, 1988, for studies addressing the economic status of widows.) Nevertheless, increases in real income over time have resulted in a halving of the fraction of older women living with relatives (Hurd, 1990b).

Work on family structure and transfers analyzes a range of decisions by families to transfer assets among members, from bequests through transfers from children or siblings to their parents. Moreover, decisions for multiple generations to live together affect decisions about the need to purchase (and save or insure for) different types of care outside the home, such as nursing home care. The basic structure of behavioral models ranges from Becker's seminal work (1991), including matching models that underlie the decision to form and leave a family and intergenerational models of transfers that have been used to explain bequests and caregiving to patents, to econometric models of household dissolution (Borsch-Supan, 1989, 1990; and Borsch-Supan, McFadden, and Schnabel, 1993), to analysis of housing demand by the elderly (Feinstein and McFadden, 1989), to models of demand for nursing homes (Garber and MacCurdy, 1990).

Integrated structural models have not yet been estimated. (For reduced form analyses using the HRS and PSID, see McGarry and Schoeni, 1994). HRS and AHEAD are going to provide excellent data for testing the integrated models of family relationships that are currently being refined. (For summaries and recent research, see Soldo and Hill, 1993.) These data are going to tightly constrain explanations, forcing researchers to integrate explanations based on intergenerational insurance, altruism, and bargaining models. But researchers have a considerable way to go before we have structural models of the type needed to simulate the effects of changes in tax policies, income and wealth testing of benefits, or health benefit policies, on the full array of outcomes that may be generated, including not only living arrangements but an understanding of the feedback on the income and wealth of the older family unit.

We are only beginning to explore the questions related to family structure and transfers, and their linkage to labor supply, savings, and pension determination. Living arrangements are a first outcome that will be investigated with more sophisticated models. We have mentioned the rising interest in models of joint labor supply behavior. Structurally, this is going to involve the introduction of bargaining models. Bargaining models are also a natural for trying to understand savings behavior. But to learn a lot more about the relation of family prospects to these dimensions of behavior is going to require specification of a credible mechanism and careful measurement of the threat price. These models will need to be extended to an intergenerational setting to resolve some of the continuing controversy about the motivations for bequests. We have yet to determine the motivation for dividing bequests and whether bequests are treated as a form of insurance. Analogous issues also arise about the transfer of time and money from children to parents, including the issue of how and why responsibilities get divided among children.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

It appears that intergenerational linkages may be important in explaining the differences in savings behavior among countries (see Poterba, 1994). Where appropriate credit markets are not well developed, generations are more likely to be living together, and intergenerational transfers will be more of an everyday occurrence. There are models of the family as annuity market (Kotlikoff and Spivak, 1981). These behavioral models and changes in behavior with the rise of Social Security suggest that intergenerational linkages are no longer as important as they once were, but that family linkages should nevertheless be explicitly considered when analyzing the determination of labor market and savings behavior.

Research on Housing

Housing wealth peaks for older households between 55 and 70. About four out of five older households own a home. At least until they reach their early 70s, home owners do not draw down on their housing wealth. Although it may be argued that stickiness in housing wealth reflects the fixed costs of location, older home owners do not adjust housing equity even when they move (Venti and Wise, 1989). There is some evidence that individuals over 70 do draw down on their housing wealth, but not at a rate that would be suggested by life-cycle consumption (Sheiner and Weil, 1992). This suggests that housing wealth may be useful for a bequest motive or that it is an asset that is held to meet precautionary motives in old age. In the latter case we may find that the median older person is not downsizing, but that a person in the bottom of the income distribution who is in a bad state is. Still, it does not appear that housing wealth is a close substitute for other forms of wealth, which are presumably held to meet similar goals.19 Moreover, while people do spend a portion of windfall gains to their housing assets, they incorporate changes in housing wealth in their other asset holdings only when housing assets decline in value (Engelhardt, 1994). Nor are the elderly enthusiastic about accepting reverse mortgages. In fact, Venti and Wise (1990) calculate that drawing down housing wealth through reverse mortgages would supplement the incomes of older families by only 10 percent. This suggests that housing equity could not substantially supplement the incomes of older Americans.

Large transaction costs make it difficult to isolate the relation of housing wealth to models of savings discussed above. Hurd (1990b) cites evidence on changes in housing wealth among those who turn over their housing that he feels is consistent with the life-cycle model. Nevertheless, he prefers testing the life-cycle model by using wealth data that exclude housing wealth. All of these findings leave us with a collection of facts, but raise a number of questions about how housing demand fits in with other dimensions of savings and wealth behavior.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

RECONCILING RESEARCH ON LABOR SUPPLY, SAVINGS, AND PENSIONS

At least as disturbing as the formidable array of questions about our leading models within each of the separate areas of inquiry—labor supply, savings, and pension determination—is the inconsistency among these three areas of inquiry. Research in each area of behavior ignores findings from the other areas of behavior. For example, many models of savings assume that retirement is fixed, so insurance against unexpected earnings risk afforded by a flexible retirement date is ignored. Even when related behavior is not ignored, it is oversimplified. In structural retirement models, for example, savings is assumed either to be motivated by life-cycle savings, with savings choices made in the context of a perfectly operating capital market, or to occur in a world where income and consumption are assumed to be identical.20

Mechanically, with three major behavioral models, there are six linkages that we would like to understand. We would like to understand the linkages from labor supply to savings behavior and pension plan determination, from savings behavior to labor supply and pension plan determination, and from pension plan determination to labor supply and savings.

We are not talking here about effects that have a secondary impact on the mode of behavior under examination. Thus, for example, consider the consequence when most structural retirement models greatly oversimplify the motivations for savings and assume either that earnings are being reallocated to finance consumption in accordance with a simple life-cycle motivation in a perfectly operating capital market21 or that lending and borrowing is not at all possible (see, e.g., Rust and Phelan, 1993). As is well known, if those approaching retirement with no liquid assets are overannuitized, they may time their retirement decision to regulate not only the amount of leisure, but also the path of consumption over time.22 If that is the case, then parameters estimated in models that assume perfectly operating capital markets may be misleading, perhaps significantly so. On the other hand, we know that savings behavior is heterogeneous. Certainly those who approach the retirement date with some liquid assets are less likely to be overannuitized, and thus should not be influenced by the age 62 early retirement provisions of Social Security. Thus, the assumption that all retirees in a certain class are liquidity constrained is also likely to lead to bias, perhaps significantly so.

Moreover, policy changes that are viewed as benign in standard retirement models may have significant behavioral consequences if liquidity constraints play an important role, as they do in some models of precautionary savings. For example, according to most structural retirement work, moving the early retirement age under Social Security from 62 to 65 should have little effect, so long as the present values are not disturbed. Under the current system, benefits adjustments for work after 62 are actuarially fair. Even if liquidity constraints would

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

normally inhibit borrowing, there should be little effect from having to wait for Social Security benefits that late in the life cycle, since all life cycle savings should have been completed. Yet if significant numbers of people enter retirement lacking financial wealth, as appears to be the case, then as suggested by Rust and Phelan's work, moving the retirement age from 62 to 65 will indeed have the effect of raising the retirement age. Consequently, the effects of certain policies will be misunderstood in retirement models that do not incorporate the effects of savings behavior.

The difficulties in understanding savings behavior thus have implications not only for the determination of income from nonpension savings, but also for understanding retirement behavior and thus for the path of income from earnings.

Consider next the implications of retirement research for the models of savings behavior. From the perspective of the precautionary model of savings, certain motivations cannot be insured by changing labor supply behavior. But certainly some variations in wages can be insured against by postponing retirement. Indeed, over the life cycle, it may also be possible to adjust to certain shocks by changing the allocation of work within the family or changing hours of work once the shock is resolved. Nevertheless, most savings models, even models of precautionary demand, take labor supply and retirement behavior as fixed (see, however, Samwick, 1994). Assuming fixed retirement dates can have adverse consequences for savings research for other reasons. As pointed out early in the savings literature, those who have preferences that favor early retirement, or who are working for firms that encourage early retirement, may be observed to have higher savings rates. To the extent that these considerations are ignored, it will appear that they will have higher life-cycle savings. In fact, the amount of life-cycle savings may be lower for early retirees once their shorter period of attachment to the labor force is taken into account.

Pension research also has implications both for retirement research and for savings research. With regard to retirement research, the issue of the possible endogeneity of pensions may be raised. Pensions are determined by the firm with the preferences of covered workers in mind. This raises the question of whether the effects of pension incentives on retirement are exaggerated. It is possible to argue that those now approaching retirement age could not have foreseen the sharp changes in pensions over the past two decades and thus did not select their jobs on the basis of a taste for early retirement.23 Nevertheless, it still is in the firm's interest to shape the pension to accord with worker preferences. Thus a relationship between early retirement provisions in pension plans and retirement outcomes may to some extent reflect the firm's efforts to accord with their own workers' tastes, leading to an overstatement of the effect of pension incentives on retirement.

With regard to savings research, one cannot help but be puzzled by recent findings that those with pensions do not reduce their saving correspondingly. If we correctly hold constant factors associated with differences in tastes and stan-

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

dardize for incomes, it is hard to conceive of a model in which pensions should not substitute for financial savings, or at least for that portion of financial savings meant to finance retirement. Not only are pensions a tax-favored form of savings, but the defined benefit plans carry one of the few opportunities to purchase an annuity with a price that is not substantially increased by the effects of adverse selection. Moreover, the penalties are small enough that many forms of pension savings can meet the demands for other types of savings, such as saving to pay for the children's college educations. Somewhere in the process, one suspects that preferences are being changed as firms and unions provide information about the importance of retirement savings or for other reasons. This information must in turn be affecting the demand for financial savings, creating an unmeasured linkage between pensions and financial savings.

We have already seen that savings research has implications for pension research. The recent evidence that substitution between pensions and savings is weak or nonexistent is not consistent with the standard tax-based explanation for pensions.24 It is not that we doubt that the favorable tax treatment of pensions underlies much of the growth of pensions (Ippolito, 1986). It does appear, however, that any effort to fit a model of pension demand must go beyond the mechanical substitution of pension for nonpension savings and provide an explanation for the empirical findings in available studies of pensions.

There are no simple fixes for these problems. But they do provide an important agenda for future research.

MODELS AND DATA NEEDED TO UNDERSTAND THE INCOMES AND WEALTH OF OLDER AMERICANS

In each of the separate areas of behavior, the direction that research is taking is appropriate. The basic outlines of dynamic retirement models are established. The models of savings behavior require further integration of competing motivations into a single framework. Models of pensions will require a further understanding of the motivation of the demand for pensions by workers as well as the behavior underlying human resource policies of the firm and their importance in shaping pensions. The question is how long it will take and how much progress will be required before we are in good enough shape to analyze the effects of the detailed policies specified in the next section.

The question facing those who would wish to use structural modeling as a basis for policy analysis in the more immediate future is, do we know enough about behavior to be comfortable imposing a model with sufficient structure to allow analysis of the effects of detailed changes in policy? A basic test of any model, and of the structure it imposes, is whether the model can explain major features of the data. If the structure is wrong in some fundamental way, we should be able to find important characteristics of the data that the model cannot explain. In the case of models of savings, retirement, and pensions that would be

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

used for policy analysis, there are much data that could and should be brought to bear before policy analysis based on the model is taken seriously. Approaches that create complex models by borrowing parameters from various sources face significant problems. Presumably to avoid the kind of specification error that occurs when parameters are estimated in the context of a partially specified set of behavioral equations, we should eventually expect structural equations of more complex models to be estimated directly from data. Only in the area of retirement modeling have we come close to meeting this criterion.

There may be classes of models that achieve a balance, providing sufficient structure to analyze effects of complex policies while resisting any temptation to overparameterize, thus avoiding potential errors of the type we have been discussing. To the extent that the investigator has sufficiently good intuition, the policy analysis that emerges may not be badly biased by the failure to specify all dimensions of behavior fully. It is difficult to determine exactly when this goal has been achieved, but there are sufficient data to put any model through its paces.

To the extent that the three dimensions of behavior we have focused on most strongly, labor force participation, savings, and pensions, are separable, this careful balancing act is likely to be most successful. Otherwise, structural models incorporating the three types of behavior will be required to analyze the array of policies affecting incentives to retire, to save, and to form pensions with different characteristics. Even if behavior in each of the three areas were separable, the task ahead is formidable. Our descriptions of the motivations governing retirement, savings, and pensions must accurately depict the relevant behaviors, and as the discussion has indicated, a great deal of work is required before we reach that goal.

If retirement, savings, and pension plan determination are not separable modes of behavior, then before we can generate a set of equations useful for analyzing the types of structural changes that major policy innovations create, it will be necessary to make progress on all three fronts. Available models fail to consider the relation between the behavior being analyzed and the other leading modes of behavior that have been noted here. This leaves open the possibility that the parameters we have are biased and will provide misleading predictions about the likely effects of major policy initiatives.

Thus, in deciding on priorities for future research, we must answer questions such as the following: If the motivation for savings takes us well beyond the simple life-cycle model, what are the implications for parameter estimates obtained in most of our life-cycle labor supply models? If pensions are designed to meet the preferences of the work force, how serious is endogeneity of the pension incentives for our measurement of worker preferences? If the motivation for pensions is more complicated than the simple substitution of a tax-favored form of savings for one that is not (and many human resource explanations for pension characteristics are inconsistent with the data), then how can we integrate the

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

findings from savings and pension theory to better understand what is driving pension coverage and the choice and structure of pension plans?

These efforts will be aided by the availability of data from HRS and AHEAD. HRS was designed with dynamic, structural retirement models in mind and provides the depth of information that will help us to improve our understanding of retirement behavior at the level of the individual.25 HRS and AHEAD were designed to provide the basic information needed to model retirement and savings behavior. Special efforts were made to reduce measurement error. Where possible, data have been matched from administrative records and employer documents. For each individual we will have a record of Social Security earnings history and thus an objective record of employment that is not subject to recall bias. Incentives from pensions will be measured from pension data provided by the employer. As we have described, innovations are being employed to reduce bias in measuring wealth. Moreover, there will be unique data for measuring the experiences of those with disabilities and of those with health problems in general, and for measuring the financial and nonfinancial support provided by family structures. Once the panel data are in, we will have two related sources of data that are capable of supporting the next generation of behavioral research. Having the data in these surveys is not sufficient to overcome the serious difficulties we face in isolating the true model, but these data will be a significant help.

What we do not have are comparable data for improving our understanding of the behavior of the firm that determines pensions and their characteristics. There are some efforts under way at the Bureau of Labor Statistics to make available establishment data collected in the Employee Benefit Survey and the Employment Cost Index, but we are a long way from meeting the requirements for release of a data set that can support the types of behavioral models that are required for policy analysis.

SELECTED POLICIES AFFECTING RETIREMENT INCOMES AND WEALTH

To conclude the paper, we return to the box labeled D in Figure 2-1, and briefly outline an array of policies changes that have the potential of significantly affecting the incomes and wealth of older Americans. These policies, if adopted, would change the incentives for retirement, savings, and pensions.

A wide variety of Social Security reforms have been recommended in the past and may be considered again in the future. One suggestion pertains to accelerating the reforms already scheduled under the 1983 Social Security amendments. Under an accelerated schedule, the normal retirement age would be raised to 67 and the delayed retirement credit increased to 8 percent immediately. The effect would be to decrease Social Security benefits for the transition generation (as they will be for all cohorts that face an age 67 retirement age under current law) and to increase the incentive to delay retirement age. A related suggestion is

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

to abolish the retirement earnings test immediately.26 After the recent increase in the portion of Social Security benefits subject to income tax, there have been suggestions to means-test Social Security benefits. It has also been continually suggested that changes be made in the treatment of spouse benefits under Social Security.

A more sweeping recommendation that continues to surface would allow some sort of privatization of the Social Security system. One possibility is to allow those who save in a tax-favored vehicle to opt out of the system, as is now possible in Britain. (For an analysis of potential enrollments in a privatized United States Social Security system, see Gustman and Steinmeier, 1995b.) Here the likely effects would incorporate fundamental reactions based on dimensions of behavior affecting labor supply, savings, and pensions. Such a program, for example, would probably encourage firms and workers to cash out defined benefit pensions to provide benefits in a form that would allow covered workers who found it to their advantage to opt out of the Social Security system to do so.27

We have recently witnessed a major health care reform debate. The role of retiree health insurance has received a good deal of attention in the course of that debate. The debate on policies that are required to balance the budget also may have major effects on health insurance. Resulting policies may change the age of eligibility for Medicare or otherwise break the linkage between work and health insurance. (For conflicting analyses of the effects of retiree health insurance on retirement, sec Gustman and Steinmeier, 1994a, and Karoly and Rogowski, 1994.) With further pressure on the Medicare system, proposals are also surfacing for means-testing Medicare benefits.

Potential pension reforms continue to be formulated with at least five goals in mind: to increase coverage, to reduce revenue losses through tax deductibility, to protect the implicit pension contract, to increase the incentive to postpone retirement, and to level the distribution of benefits among high- and low-income employees.28 Potential changes in policies could change the tax treatment of pensions, mandate the availability of pensions, alter a variety of eligibility and vesting rules, further change treatment of spouses under pensions, further change funding rules including minimum and maximum funding levels, change treatment of retiree health benefits, change acceptable actuarial assumptions and procedures, alter discrimination rules (including further regulation of matching provisions under 401(k)s), change rules governing Social Security offsets, change the rules governing pensions of highly paid employees, regulate the backloading of pensions, require that payments to terminated vested employees be based on projected earnings or some related mechanism rather than on the last few years of nominal earnings, adjust rules affecting the returns to invested pension assets by further regulating these investments or mandating certain types of investment, change rules governing asset investments for defined contribution plans, regulate or mandate post-retirement benefit increases, regulate rollovers from pension plans, and introduce an array of policies that would affect the terms of Pension

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Benefit Guaranty Corporation insurance. (For further discussions of pension policies, see Congressional Budget Office, 1987, and Ippolito, 1983, 1986.)

One of the more controversial proposals that would reduce taxes on the income from savings is to resurrect the IRA in its initial form, as favored in the past by Treasury Secretary Bentsen. Such a reform would again allow full initial deductibility of IRA contributions up to specified limits, with eligibility not contingent on lack of pension coverage. On the other hand, proposals are now floating to reduce the eligibility for what is left of IRAs. Among the more sweeping tax reforms that would have major effects on incentives to save is adoption of a consumption tax.

A variety of training, employment, and regulatory programs in the labor market have also been suggested to foster increased employment of older individuals and to raise the reward to work.

To fully understand the effects of these policies, we will need a more complete understanding of behavior than we currently have. We will need a better understanding not only of the behaviors of labor supply, savings, and pension plan determination, but of the relationships among these modes of behavior.

APPENDIX TABLES AND FIGURES

The tables and figures begin on page 42.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-A1 Net Worth by Family Composition, HRS Data

Family Composition

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Net Worth Less

 

 

 

 

 

 

 

Housing Equity

 

 

 

 

 

 

 

Couples

5,229

197.9

2.0

13.0

53.0

166.0

426.0

Single men

741

124.9

0

1.0

16.0

81.8

229.0

Single women

1,632

55.0

-0.7

0

7.0

45.0

140.0

Total

7,602

163.0

0

6.1

37.1

130.0

355.0

Housing Equity

 

 

 

 

 

 

 

Couples

5,229

90.9

0

27.0

60.5

110.0

197.0

Single men

741

42.0

0

0

6.0

56.0

126.0

Single women

1,632

42.4

0

0

16.5

58.0

122.0

Total

7,602

76.9

0

10.0

50.0

100.0

177.5

Total Net Worth

 

 

 

 

 

 

 

Couples

5,229

288.8

15.0

56.0

132.8

285.0

602.0

Single men

741

166.9

0

4.0

43.0

157.0

320.5

Single women

1,632

97.5

0

1.2

36.4

115.5

259.3

Total

7,602

239.9

1.1

32.4

101.0

239.0

512.0

TABLE 2-A2 Net Worth by Ethnicity, HRS Data

Ethnicity

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Net Worth Less

 

 

 

 

 

 

 

Housing Equity

 

 

 

 

 

 

 

Black

1,424

45.4

-0.9

0

5.0

29.0

91.0

Hispanic

716

59.2

0

0.1

4.5

29.0

95.0

Other

5,462

189.7

1.3

11.7

50.0

160.0

415.9

Total

7,602

163.0

0

6.1

37.1

130.0

355.0

Housing Equity

 

 

 

 

 

 

 

Black

1,424

38.6

0

0

17.0

50.0

89.0

Hispanic

716

45.2

0

0

22.0

59.0

129.0

Other

5,462

85.4

0

20.0

58.0

107.0

191.0

Total

7,602

76.9

0

10.0

50.0

100.0

177.5

Total Net Worth

 

 

 

 

 

 

 

Black

1,424

84.1

0

0.4

30.5

86.5

170.1

Hispanic

716

104.1

0

1.7

34.0

95.5

224.5

Other

5,462

275.1

7.5

48.3

129.9

275.0

586.0

Total

7,602

239.9

1.1

32.4

101.0

239.0

512.0

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-A3 Net Worth by Income Group, HRS Data

Income

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Net Worth Less

 

 

 

 

 

 

 

Housing Equity

 

 

 

 

 

 

 

< 10K

854

48.6

-0.7

0

0.4

15.0

48.8

10–25K

1,696

66.2

-0.3

0.9

8.2

46.0

148.0

25–50K

2,405

96.7

1.2

9.5

34.0

97.0

224.2

50–100K

2,077

195.4

7.5

25.5

72.4

193.6

423.0

>100K

570

665.4

32.0

90.0

240.0

695.0

1,792.2

Total

7,602

163.0

0

6.1

37.1

130.0

355.0

Housing Equity

 

 

 

 

 

 

 

< 10K

854

29.3

0

0

0

35.0

80.0

10–25K

1,696

50.3

0

0

27.0

65.0

120.0

25–50K

2,405

65.3

0

17.0

48.0

85.0

150.0

50–100K

2,077

93.3

1.0

37.0

70.0

120.0

197.0

> 100K

570

181.7

29.0

67.0

125.0

225.0

393.0

Total

7,602

76.9

0

10.0

50.0

100.0

177.5

Total Net Worth

 

 

 

 

 

 

 

< 10K

854

78.8

-0.5

0

5.8

54.0

169.6

10–25K

1,696

116.5

0

7.3

43.5

120.8

269.0

25–50K

2,405

162.0

8.0

39.0

92.0

188.0

350.0

50–100K

2,077

288.6

35.0

83.7

165.0

313.1

590.0

> 100K

570

847.1

100.3

198.0

405.0

923.4

2,166.0

Total

7,602

239.9

1.1

32.4

101.0

239.0

512.0

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-A4 Total Net Worth by Family Composition and Age, AHEAD Data

Family Composition

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Age 70–74

 

 

 

 

 

 

 

Couples

946

276.3

17.0

57.0

140.9

309.0

578.8

Single men

223

201.7

0

10.0

65.0

167.0

367.8

Single women

825

178.1

0

4.1

51.4

132.5

295.0

Total

2,004

209.2

0.3

26.6

90.6

315.0

462.5

Age 75–79

 

 

 

 

 

 

 

Couples

581

274.1

11.0

50.7

118.0

264.0

610.0

Single men

203

164.3

1.0

14.0

71.0

171.0

319.0

Single women

812

105.1

0

3.7

47.8

116.0

236.0

Total

1,596

177.0

0.1

16.2

71.0

168.8

364.0

Age 80–84

 

 

 

 

 

 

 

Couples

371

242.2

10.0

43.5

114.2

236.0

495.3

Single men

154

148.6

0.5

10.0

51.0

132.0

289.0

Single women

702

87.5

0

3.5

40.5

101.5

188.5

Total

1,227

142.7

0.1

10.2

60.0

143.0

290.0

Age 85+

 

 

 

 

 

 

 

Couples

155

162.1

1.0

21.0

75.6

216.0

415.0

Single men

145

121.5

0

1.0

25.9

101.0

287.6

Single women

595

85.8

0

1.5

30.0

100.0

203.0

Total

895

104.7

0

2.0

37.0

111.0

247.0

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-A5 Net Worth Excluding Home Equity by Family Composition and Age, AHEAD Data

Family Composition

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Age 70–74

 

 

 

 

 

 

 

Couples

946

176.6

1.0

11.0

55.7

190.0

420.0

Single men

233

140.3

0

2.1

17.0

89.0

255.4

Single women

825

68.3

0

0.1

7.5

50.0

160.5

Total

2,004

129.6

0

2.2

25.9

109.2

323.0

Age 75–79

 

 

 

 

 

 

 

Couples

581

176.8

0.8

10.0

38.0

150.0

417.0

Single men

203

80.2

0

2.0

15.5

83.0

205.0

Single women

812

53.0

0

0.2

7.0

39.5

125.0

Total

1,596

103.6

0

1.3

15.8

76.3

247.0

Age 80–84

 

 

 

 

 

 

 

Couples

371

161.3

1.0

8.0

37.0

139.0

370.3

Single men

154

107.5

0.1

1.2

19.5

66.8

224.0

Single women

702

42.7

0

0.2

5.1

40.0

112.0

Total

1,227

87.2

0

1.0

12.2

64.1

191.0

Age 85+

 

 

 

 

 

 

 

Couples

155

104.8

0

2.0

20.0

125.0

271.0

Single men

145

74.8

0

0.4

7.0

45.0

114.5

Single women

595

48.1

0

0

5.0

33.2

113.0

Total

895

62.2

0

0.2

6.7

50.0

140.0

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-A6 Total Net Worth by Ethnicity and Age, AHEAD Data

Ethnicity

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Age 70–74

 

Black

309

50.4

0

0

25.0

69.0

152.0

Hispanic

130

72.8

0

0

25.5

67.0

181.2

White plus other

1,565

238.5

5.5

45.5

116.2

262.2

522.0

Total

2,004

209.2

0.3

26.6

90.6

215.0

462.5

Age 75–79

 

Black

241

51.6

0

0.8

25.0

56.2

113.0

Hispanic

90

42.6

0

0

1.0

43.0

99.0

White plus other

1,265

200.1

2.0

33.2

94.0

202.0

409.0

Total

1,596

177.0

0.1

16.2

71.0

168.8

364.0

Age 80–84

 

Black

184

61.7

0

0.6

31.0

62.0

130.0

Hispanic

71

64.5

0

0

15.0

50.0

101.5

White plus other

972

159.0

0.8

20.0

77.7

166.5

332.0

Total

1,227

142.7

0.1

10.2

60.0

143.0

290.0

Age 85+

 

Black

126

39.5

0

0

5.0

45.0

122.0

Hispanic

44

31.9

0

0

0.3

62.8

75.0

White plus other

725

116.4

0

5.5

46.2

131.0

270.0

Total

895

104.7

0

2.0

37.0

111.0

247.0

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

TABLE 2-A7 Net Worth Excluding Home Equity by Ethnicity and Age, AHEAD Data

Ethnicity

Sample Size

Mean Value ($000)

Percentiles ($000)

10

25

50

75

90

Age 70–74

 

Black

309

20.3

0

0

1.0

16.0

60.0

Hispanic

130

25.5

0

0

0.5

8.0

48.8

White plus other

1,565

150.3

0.5

7.0

41.2

150.0

365.0

Total

2,004

129.6

0

2.2

25.9

109.2

323.0

Age 75–79

 

Black

241

20.9

-0.5

0

0.8

10.0

40.0

Hispanic

90

11.9

-0.6

0

0

2.0

20.3

White plus other

1,265

119.0

0.3

5.3

26.0

101.0

297.0

Total

1,596

103.6

0

1.3

15.8

76.3

247.0

Age 80–84

 

Black

184

23.0

0

0

1.0

13.1

45.5

Hispanic

71

27.7

0

0

0.3

5.2

35.0

White plus other

972

99.9

0

2.4

20.4

83.3

224.0

Total

1,227

87.2

0

1.0

12.2

64.1

191.0

Age 85+

 

Black

126

14.9

0

0

0.2

9.5

40.0

Hispanic

44

7.9

0

0

0

2.5

30.0

White plus other

725

70.8

0

1.0

10.5

61.0

179.0

Total

895

62.2

0

0.2

6.7

50.0

140.0

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

FIGURE 2-A1 Relationship of health status and mean net worth for married-couple households. NOTE: Health status is respondent's subjective rating. SOURCE: Tabulations of the Health and Retirement Survey.

FIGURE 2-A2 Relationship of health status and median net worth for married-couple households. NOTE: Health status is respondent's subjective rating. SOURCE: Tabulation of the Health and Retirement Survey.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

FIGURE 2-A3 Relationship of health status and mean income for married-couple house-holds. NOTE: Health status is respondent's subjective rating. SOURCE: Tabulations of the Health and Retirement Survey.

FIGURE 2-A4 Relationship of health status and median income for married-couple households. NOTE: Health status is respondent's subjective rating. SOURCE: Tabulations of the Health and Retirement Survey.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

NOTES

1.  

In examining behavior from the perspective of the individual, we will take the wage offer at a given age and tenure as exogenous to the individual's decision in older age.

2.  

For recent studies elaborating on the dynamic structure of the dependent variable, see Blau (1994) and Peracchi and Welch (1994). For models of retirement using a rich set of dynamic flows in the context of a dynamic programming model, see, for example, Rust (1990) and Berkovec and Stern (1991). For an analysis that incorporates the effects of incomplete annuity markets, borrowing constraints, and incomplete markets for health insurance on labor market behavior, see Rust and Phelan (1993). For inclusion in the opportunity set of the discontinuities from the option value of the pension, see Stock and Wise (1990a, 1990b) and Lumsdaine, Stock, and Wise(1990). For family retirement models, see Hurd (1990a) and Gustman and Steinmeier (1994b). For models that include further information on nonwage aspects of employment, see Gustman and Steinmeier (1986a) and Hurd and McGarry (1993a).

3.  

In order to make estimation tractable, researchers have made a number of simplifications reducing the complexity of the dependent variable specification, the detail of the opportunity set, the complexity of the dynamic decisions allowed, and the richness of the econometric specification. Some inadequacies in models estimated to date have reflected limitations in the available data sets. One fundamental choice has been between nationally representative data sets that include only respondent-provided descriptions of a few elements of the pension plan, such as studies based on data from the Retirement History Study or the National Longitudinal Study of Older Men, and data sets with highly detailed employer-provided descriptions of pensions, but with very narrow samples, including the employees of only a few firms. Thus, studies using employer-provided plan descriptions have been confined to analyzing the decision to leave the firm from which the pension data had been obtained; subsequent employment experience was not observed and could not be analyzed. Nor in the studies focusing on the analysis of behavior in a few firms have there been data on family characteristics and labor market activity or detailed information on the health status of the worker. Health is simply included under the random term. Only recently have we begun to see retirement analysis based on nationally representative data that include information from employer-provided pension plan descriptions (Samwick. 1993a). But the model specification remains highly simplified compared with some of the structural analyses now available in the literature.

4.  

Anderson, Gustman, and Steinmeier (1994) attribute about a quarter of the trend to the sharp lowering of early and normal retirement dates for pension eligibility and to the reduced effect of continued work on the Social Security benefit in the 1980s as compared with the incentives in the early 1970s. This type of analysis does not answer why the pension incentives were changed as they were and leaves unexplained the other three-fourths of the change in retirement behavior.

5.  

One issue that has recently been subject to disagreement is the question of whether the trends in retirement are due to trends in incentives and unexpected wealth effects from the start-up and revisions of pension plans and Social Security, as implied by the work of Ippolito (1990) and others, or to trends in wages differentials and the forces underlying these trends, as implied by Peracchi and Welch (1994). If trends in retirement among older cohorts may be attributed to the effects of unexpected changes in Social Security wealth, then this raises a question. Shouldn't we expect to see the trend cease, as we did, and then strongly reverse itself as the Social Security system matured and Congress ceased granting important benefit increases after the 1970s? (Perhaps the flattening of the trend in recent years is just the beginning of a strong reversal; however, it has been 8 years since the trend ceased and no strong reversal is yet apparent.) Moreover, if changes in the wage structure due to the decline in relative and real wages of the unskilled account for the trend to earlier retirement, as Peracchi and Welch (1994) argue, why do we see the same trend to earlier retirement in the last two decades in Britain, where despite a decline in relative wages for the unskilled, real wages for the unskilled continued to increase? If higher wages are responsible for the trend to earlier retirement, why did the trend begin only after the 1930s in the United States? And more generally, why aren't these wage effects picked up by structural retirement models?

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

6.  

The studies with the best pension measures, those obtained directly from firm-provided data, have no information on health status, for example, Fields and Mitchell (1984), Stock and Wise (1990a, 1990b), and Lumsdaine, Stock and Wise (1990, 1992a, 1992b).

7.  

Savers may also be motivated by the need to finance various expenditures, such as their children's college education or the purchase of a home, or by other motivations.

8.  

Hurd (1990b) also argues that consumption paths are consistent with the life-cycle model.

9.  

HRS is designed to permit direct measurement of individuals' attitudes toward risk as well as their time preference and expectations regarding particular risks to, among other, the economy and to Social Security and to provide the opportunity to explore heterogeneity in these measures within the population. For a study based on the new HRS data, see Barsky et al. (1993).

10.  

Smith (1994) also finds that the bequest motive is more likely to be operating among those with higher incomes. He finds much smaller effects of a belief in a bequest motive when using median repressions than when using mean regressions.

11.  

As a test of the importance of the bequest motive, Hurd examines the wealth path of families with and without children and concludes that the higher rate of decumulation of wealth by those with children is evidence against the importance of the bequest motive.

12.  

Hamermesh (1984) argues that asset decumulation is too rapid to sustain consumption after retirement. On the other hand, it has frequently been argued that decumulation is too slow for life-cycle motivations and that bequests will result (see Kotlikoff and Summers, 1981).

13.  

Those with a pension save less than those without a pension (Smith, 1994), but there is a question about the rates at which these assets are substituted. See, for example, Bernheim and Scholz (1993) and Samwick (1994) on the relation between pension holding and nonpension wealth. See Hoynes and McFadden (1994) for evidence on the nonhousing wealth of those who do and do not own houses.

14.  

One set of results suggests that although Social Security benefit changes adopted in 1983 will reduce retirement incomes by about 14 percent by the time they are fully phased in, half of the reduction will be offset by induced increases in earnings as retirement is postponed in response to the change in incentives (Gustman and Steinmeier, 1985). Another set of results suggests that accelerating the 1983 Social Security reforms to eliminate penalties from continuing work after reaching Social Security normal retirement age, or equivalently abolishing the retirement earnings test, will have only a small effect on retirement in the affected cohorts (Gustman and Steinmeier, 1991). Simulations with models fit to payroll data from selected firms (Stock and Wise, 1990a, 1990b; Lumsdaine and Wise, 1994: Wise and Woodbury, 1994) suggest that retirement incentives from Social Security are much weaker than are incentives from pensions. Other recent work focuses on the proper modeling of Social Security effects on retirement (Reimers and Honig, 1993a, 1993b; Rust and Phelan, 1993). These studies suggest that simple, mechanical models of a response to the benefit formula and earnings test are unsatisfactory, and that liquidity constraints, knowledge of the earnings test, and the nonwage characteristics of jobs (Hurd and McGarry, 1993b; Gustman, Mitchell, and Steinmeier, 1994) are important.

15.  

The suggestion that workers may not fully understand the Social Security rules comes from a finding of a spike in the participation rate at the disregard amount of the Social Security earnings test (Burtless and Moffitt, 1984). The rules allow any benefit that is lost to the earnings test by someone between the ages of 62 and 65 to be recovered in future years on a roughly actuarially fair basis, so that the spike at the disregard amount would not make sense for those under 65 if they understood the rules. Analogously, incentives from Social Security should not create a spike in the retirement hazard at age 62, in that any loss to the earnings test can be recovered in future years (Hurd, 1990a). Nevertheless, in some studies there is a suggestion at age 62 of a spike in the retirement hazard that seems to be associated with Social Security. Specifically, while findings in Gustman and Steinmeier (1986b) suggest that spikes in retirement hazards can be completely attributed to actuarial incentives of pensions, Social Security, and mandatory retirement, without considering any effects of liquidity constraints, Hurd (1990b), using data from Kotlikoff and Wise (1985, 1987), finds a spike in the

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

   

retirement hazard at age 62 for workers with a pension plan that does not generate any retirement incentive at age 62. This leads Hurd to infer that liquidity effects associated with Social Security are responsible, a finding confirmed in subsequent analysis by Wise and his colleagues. Nalebuff and Zeckhauser (1985) explain why liquidity effects may also arise from pension plans that are designed for a heterogeneous work force.

16.  

See Venti and Wise (1993) for a discussion and evidence of imperfect substitution between personal retirement savings and a number of other savings instruments. Venti and Wise include assets in 401(k) plans under the heading of personal targeted retirement assets, as opposed to employer-provided pension assets. See also the evidence in Samwick (1994).

17.  

Studies of pension outcomes discuss the effects of certain plan differences; for example, Green (1985) and Bodie (1990) analyze the differential risk of defined benefit and defined contribution plans. Most of the empirical studies of the choices of pension outcomes are descriptive, relating pension outcomes to demographic and employment characteristics.

18.  

Recent articles examining trends in pension coverage include Parsons (1991, 1993), Bloom and Freeman (1992), Turner and Beller (1992), and Even and Macpherson (1994). Other studies have tried to describe and explain trends in plan type (Clark and McDermed, 1990; Gustman and Steinmeier, 1992; Ippolito, 1995: Kruse, 1995) and to understand the forces shaping the trends in provision of pensions (Mitchell and Luzadis, 1988; Luzadis and Mitchell, 1991; Mitchell, 1992).

19.  

Skinner (1993) found a negative association between housing wealth and nonhousing savings in the PSID, measuring housing values by the individual's self-report of the housing value in two periods. Hoynes and McFadden (1994) used data on housing prices in metropolitan areas and found either a very small positive association between changes in housing prices and the savings rate, or a larger one, depending on specification. When Engelhardt (1994) reestimated the results in Hoynes and McFadden using median regression, reducing the weight on outliers, he found a negative and significant effect of housing wealth on savings about the same size as the effect found by Skinner.

20.  

In either case such phenomena as the bunching of retirees at the Social Security earnings test maximum are difficult to explain (see Burtless and Moffitt, 1984).

21.  

Rust (1989, 1990) developed an influential model of the joint determination of savings and retirement, but that model has not been estimated. Otherwise, all of the structural models mentioned above are estimated using procedures that ignore information about savings behavior.

22.  

Nalebuff and Zeckhauser (1985), Hurd (1990b), Rust and Phelan (1993), and others note that in the absence of assets, liquidity-constrained individuals will retire at 62 to obtain access to their Social Security benefits.

23.  

The normal retirement age for pension covered workers in the Retirement History Study average 64.2. The early retirement age averaged 61. According to 1989 Survey of Consumer Finances data, the normal retirement age averaged 61.7 and the early retirement age averaged 54.3 (Anderson, Gustman, and Steinmeier, 1994). For data on the changes in early retirement provisions of pension plans over the past two decades, see Ippolito (1990) and Mitchell (1992).

24.  

Venti and Wise (1993) find no evidence of any offset at all between pension wealth and personal financial assets, with the coefficient on the relationship being positive but insignificant. See also Bernheim and Scholz (1993), who relate savings to a qualitative indicator of pension coverage and find an indication of some substitution for those with college degrees. The analysis of the relation between pensions and savings is at an earlier stage than some other lines of research. Nevertheless, the work to date raises important questions about the substitutability between pensions and other forms of savings.

25.  

These data sets and their potential use in research on retirement, savings, health, family linkages, and related issues are described in detail in a forthcoming issue of the Journal of Human Resources.

26.  

These suggestion were made by Dorcas Hardy, the former head of the Social Security Administration, and continue to be made by others. For an analysis, see Gustman and Steinmeier (1991).

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

27.  

We have not emphasized the disability system, but clearly changes in disability rules will affect incentives to save for precautionary reasons. Moreover, disability provisions may interact with early retirement provisions of pension programs affecting retirement incentives. Because the roll of the disability system has grown rapidly in recent years and the finances of that system have deteriorated, there have been a number of suggestions for change. Such changes might, for example, include tightening eligibility requirements, reducing benefits, and limiting the duration of benefits.

28.  

Pension discrimination rules attempt to equalize benefits within firms. Special pension regulations pertaining to high-income employees limit the absolute size of benefits.

REFERENCES

Abel, A.B. 1985. Precautionary savings and accidental bequests. American Economic Review 75(4):777–791.

Allen, S.G., R.L. Clark, and A.A. McDermed 1993. Pensions, bonding and lifetime jobs. Journal of Human Resources 28(3):463–481.

Anderson, P.M., A.L. Gustman, and T.L. Steinmeier 1994. The Trend to Earlier Retirement Among Males. Report to the Social Security Administration.

Ausink, J.A., and D.A. Wise 1993. The Military Pension, Compensation, and Retirements of U.S. Air Force Pilots. NBER Working Paper #4593. Cambridge, Mass.: National Bureau of Economic Research.


Barro, R.J. 1974. Are governments bonds net wealth? Journal of Political Economy 82(6):1095–1117.

Barsky, R., M. Kimball, M. Shapiro, and F.T. Juster 1993. Experimental Measures of Preferences in the HRS. Paper presented at Health and Retirement Survey Early Results Workshop, Survey Research Center, University of Michigan.

Becker. G.S. 1991. A Treaties on the Family. Cambridge, Mass.: Harvard University Press.

Berkovec, J.C., and S. Stern 1991. Job exit behavior of older men. Econometrica 59(1):189–210.

Bernheim. B.D. 1988. Social Security benefits: An empirical study of expectations and realizations. Pp. 312–345 in R.R. Campbell and E. Lazear, eds., Issues in Contemporary Retirement. Stanford, Calif.: Hoover Institution.

1989. The timing of retirement: A comparison of expectations and realizations. Pp. 335–355 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

1991. How strong are bequest motives? Journal of Political Economy 99(5):899–927.

1993. Personal Saving, Information, and Economic Literacy: New Directions for Public Policy. Unpublished manuscript. Department of Economics, Princeton University.

1994. Do saving incentives work?: A discussion of Engen, Gale and Scholz. Brookings Papers on Economic Activity 1994(1):152–156.

Bernheim, B.D., and J.K. Scholz 1993. Private saving and public policy. Pp. 73–110 in J. Poterba, ed., Tax Policy and the Economy Vol. 7. Cambridge, Mass.: MIT Press.

Berheim, B.D., Shleifer, and L. H. Summers 1985. The strategic bequest motive. Journal of Political Economy 96(6):1045–1076.

Blau, D.M. 1994. Labor force dynamics of older men. Econometrica 62(1):117–156.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Bloom, D.E., and R.B. Freeman 1992. The fall in private pension coverage in the United States American Economic Review, Papers and Proceedings 82(2):539–545.

Bodie, Z. 1990. Pensions as retirement income insurance. Journal of Economic Literature 28(1):28–49.

Borsch-Supan, A. 1989. Household dissolution and the choice of alternative living arrangements among elderly Americans. Pp. 119–146 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

1990. A dynamic analysis of household dissolution and living arrangement transitions by elderly Americans. Pp. 89–114 in D.A. Wise, ed., Issues in the Economics of Aging. Chicago, Ill.: University of Chicago Press.

Borsch-Supan, A., D. McFadden, and R. Schnabel 1993. Living Arrangements: Health and Wealth Effects. Unpublished paper. Department of Economics, University of California, Berkeley.

Borsch-Supan, A., and K. Stahl 1991. Life cycle savings and consumption constraints: Theory, empirical evidence and fiscal implications. Journal of Population Economics 86:3–27.

Brown, C. 1993. Early Retirement Windows: Windows of Opportunity? Defenestrations? (and Even Refenestrations…). Paper presented at Health and Retirement Survey Early Results Workshop, Survey Research Center, University of Michigan.

Burkhauser, R.V. 1979. The pension acceptance decision of older workers. Journal of Human Resources 14(1):63–75.

Burkhauser, R.V., K.C. Holden, and D. Feaster 1988. Incidence, timing, and events associated with poverty: A dynamic view of poverty in retirement. Journal of Gerontology 43(2):S46–S52.

Burtless, G., and R.A. Moffitt 1984. The effect of Social Security benefits on the labor supply of the aged. Pp. 135–174 in H.J. Aaron and G. Burtless, eds., Retirement and Economic Behavior. Washington, D.C.: Brookings Institution.

1985. The joint choice of retirement age and post-retirement hours of work. Journal of Labor Economics 3(2):209–236.


Cagan, P. 1965. The Effect of Pension Plans of Aggregate Saving: Evidence from a Sample Survey. National Bureau of Economic Research Occasional Paper 95. New York: Columbia University Press.

Carroll, C.D. 1992. The buffer-stock theory of saving: Some macroeconomic evidence. Brookings Papers on Economic Activity 2:61–135.

Carroll, C.D., and A.A. Samwick 1994. The Nature of Precautionary Wealth. Manuscript, National Bureau of Economic Research, Cambridge, Mass.

Carroll, C.D., and L.H. Summers 1991. Consumption growth parallels income growth: Some new evidence. Pp. 305–343 in D.B. Bernheim and J.B. Shoven, eds., National Saving and Economic Performance. Chicago, Ill.: University of Chicago Press.

Clark, R.L., and A.A. McDermed 1990. The Choice of Pension Plans in a Changing Regulatory Environment Washington, D.C.: American Enterprises Institute.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Congressional Budget Office 1987. Tax Policy for Pension and Retirement Savings. Washington, D.C.: U.S. Government Printing Office.

Davies, J.B. 1981. Uncertain lifetime, consumption, and dissaving in retirement. Journal of Political Economy 89(3):561–576.

Deaton, A. 1991. Savings and liquidity constraints. Econometrica 59(5):1221–1248.


Employee Benefit Research Institute 1993. Pension coverage and participation growth: A new look at primary and supplemental plans. EBRI Issue Brief No. 144. Washington, D.C.: Employee Benefit Research Institute.

Engelhardt, G.V. 1994. House Prices and Home Owner Saving Behavior. Unpublished manuscript. Dartmouth College.

Engen, E.M., and W. Gale 1993. IRAs and Saving in a Stochastic Life-Cycle Model. Unpublished manuscript. University of California, Los Angeles.

Engen, E.M., W.G. Gale, and J.K. Scholz 1994. Do saving incentives work? Brookings Papers on Economic Activity 1994(1):85–180.

Even, W.E., and D.A. Macpherson 1994. The pension coverage of young and mature workers. Report to the U.S. Department of Labor. Pp. 85–106 in Pension Coverage Issues for the 90's. U.S. Department of Labor, Pension and Welfare Benefits Administration.


Feinstein, J., and D. McFadden 1989. The dynamics of housing demand by the elderly: Wealth, cash flow and demographic effects. Pp. 55–86 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

Fields, G.S., and O.S. Mitchell 1984. Retirement, Pensions and Social Security. Cambridge, Mass.: MIT Press.


Gale, W.G., and J.K. Scholz 1994. IRAs and household saving. American Economic Review 84(5):1233–1260.

Garber, A.M., and T.E. MacCurdy 1990. Predicting nursing home utilization by high risk elderly. Pp. 173–200 in D.A. Wise, ed., Issues in the Economic of Aging. Chicago, Ill.: University of Chicago Press.

Grad, S. 1994. Income of the Population 55 or Older, 1992. Washington, D.C.: U.S. Department of Health and Human Services, Social Security Administration.

Green, J. 1985. The riskiness of private pensions. Pp. 357–375 in D.A. Wise, ed., Pensions, Labor, and Individual Choice. Chicago, Ill.: University of Chicago Press.

Guiso, L., T. Jappelli, and D. Terlizzese 1992. Earnings uncertainty and precautionary saving. Journal of Monetary Economics 30:307–337.

Gustman, A.L., and O.S. Mitchell 1992. Pensions and labor market activity: Behavior and data requirements. Pp. 39–87 in Z. Bodie and A.H. Munnell, eds., Pensions and the Economy: Sources, Uses, and Limitation of Data. Philadelphia, Pa.: Pension Research Council Publications and University or Pennsylvania Press.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Gustman, A.L., O.S. Mitchell, and T.L. Steinmeier 1994. The role of pensions in the labor market. Industrial and Labor Relations Review 47(3):417–438.

Gustman, A.L., and T.L. Steinmeier 1985. The 1983 Social Security reforms and labor supply adjustments of older individuals in the long run . Journal of Labor Economics 3:237–253.

1986a. A disaggregated structural analysis of retirement by race, difficulty of work and health. Review of Economics and Statistic 67(3):509–513.

1986b. A structural model. Econometrica 54(3):555–584.

1989. An analysis of pension benefit formulas, pension wealth and incentives from pensions. Pp. 53–106 in R. Ehrenberg, ed., Research in Labor Economics 10. Greenwich, Conn.: JAI Press.

1991. Changing the Social Security rules for work after 65. Industrial and Labor Relations Review 44(4):733–745.

1992. The stampede towards defined contribution pension plans: Fact or fiction? Industrial Relations 31(2):361–369.

1993. Pension portability and labor mobility: Evidence from the survey of income and program participation. Journal of Public Economics 50:299–323.

1994a. Employer-provided health insurance and retirement behavior. Industrial and Labor Relations Review 48(1):124–140.

1994b. Retirement in a Family Context: A Structural Model for Husbands and Wives. NBER Working Paper #4629. Cambridge, Mass.: National Bureau of Economic Research .

1995a. Pension Incentives and Job Mobility. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research.

1995b. Privatizing Social Security: First Round Effects of a Generic, Voluntary, Privatized U.S. Social Security System. NBER Working Paper #5362. Cambridge, Mass.: National Bureau of Economic Research.


Hall, R.E., and F.S. Mishkin 1982. The sensitivity of consumption to transitory income: Estimates from panel data on households. Econometrica 50(2):461–482.

Hamermesh, D.S. 1984. Consumption during retirement: The missing link in the life cycle. Review of Economics and Statistics 66(1):1–7.

Hoynes, H., and D. McFadden 1994. The Impact Demographics on Housing and Non-Housing Wealth in the United States. NBER Working Paper #4666. Cambridge, Mass.: National Bureau of Economic Research.

Hubbard, R.G., J. Skinner, and S.P. Zeldes 1994a. Expanding the life-cycle model: Precautionary saving and public policy. The American Economic Review, Papers and Proceedings 84(2):174–179.

1994b. The importance of precautionary motives in explaining individual and aggregate saving. Carnegie-Rochester Conference Series On Public Policy 40(June):59–126.

1995. Precautionary saving and social insurance. Journal of Political Economy 103(2):360–399.

Hurd, M. 1987. Savings of the elderly and desired bequests. American Economic Review 77(3):298–312.

1990a. The joint retirement decisions of husbands and wives. Pp. 231–254 in D.A. Wise, ed., Issues in the Economics of Aging. Chicago, Ill.: University of Chicago Press.

1990b. Research on the elderly: Economic status, retirement and consumption and saving. Journal of Economic Literature 28(2):565–637.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

1993. The Effect of Labor Market Rigidities on the Labor Force Behavior of Older Workers. NBER Working Paper #4462. Cambridge, Mass.: National Bureau of Economic Research.

Hurd, M., and K. McGarry 1993a. Evaluation of Subjective Probability Distributions in the HRS. NBER Working Paper #4560. Cambridge, Mass.: National Bureau of Economic Research.

1993b. The Relation Between Job Characteristics and Retirement. NBER Working Paper #4558. Cambridge, Mass.: National Bureau of Economic Research.

Hurd, M., and D.A. Wise 1989. The wealth and poverty of widows: Assets before and after the husband's death. Pp. 177–200 in D. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

Hutchens, R. 1986. Delayed payment contracts and a firm's propensity to hire older workers. Journal of Labor Economics 4(4):439–457.

1987. A test of Lazear's theory of delayed payment contract. Journal of Labor Economics 5(4)Part 2:S153–S170.


Ippolito, R.A. 1983. Public policy toward private pensions. Contemporary Policy Issues, a supplement to Economic Inquiry 3:53–76.

1985a. The economic function of underfunded pension plans. The Journal of Law and Economics 28(3):611–651.

1985b. The labor contract and true economic pension liabilities. American Economic Review 75(5):1031–1043.

1986. Pensions, Economics and Public Policy. Homewood, Ill.: Dow Jones-Irwin.

1987. The implicit pension contract: Developments and new directions. Journal of Human Resources 22(3):441–467.

1990. Toward explaining earlier retirement after 1970. Industrial and Labor Relations Review 43(5):556–569.

1995. Toward explaining the growth of defined contribution pension plans. Industrial Relations 34(1):1–20.


Juster, F.T., and F.P. Stafford 1991. The allocation of time: Empirical findings, behavioral models, and problems of measurement. Journal of Economic Literature 29:471–522.


Karoly, L.A., and J.A. Rogowski 1994. The effects of health insurance on the decision to retire. Industrial and Labor Relations Review 48(1):103–123.

Katona, G. 1965. Private Pension and Individual Saving. Ann Arbor, Mich.: Survey Research Center, University of Michigan.

Kimball, M.S. 1990. Precautionary saving in the small and in the large. Econometrica 58(1):53–73.

1993. Standard risk aversion. Econometrica 61(3):589–611.

Kotlikoff, L.J. 1988. Intergenerational transfers and savings. The Journal of Economic Perspectives 2(2):41–58.

Kotlikoff, L.J., and A. Spivak 1981. The family as an incomplete annuities market. Journal of Political Economy 89(2):372–391.

Kotlikoff, L.J., and L. Summers 1981. The role of intergenerational transfers in capital accumulation. Journal of Political Economy 89(4):706–732.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Kotlikoff, L.J., and D.A. Wise 1985. Labor compensation and the structure of private pension plans: Evidence for contractual vs. spot labor markets. Pp. 55–85 in D.A. Wise, ed., Pensions, Labor, and Individual Choice. Chicago, Ill.: University of Chicago Press.

1987. The incentive effects of private pension plans. Pp. 283–336 in Z. Bodie, J.B. Shoven, and D.A. Wise, eds., Issues in Pension Economics. Chicago, Ill.: University of Chicago Press.

Kruse, D.L. 1995. Pension substitution in the 1980's: Why the shift toward defined contribution pension plans? Industrial Relations 34(2):218–241.

Lazear, E.P. 1979. Why is there mandatory retirement? Journal of Political Economy 87(6):1261–1284.

1983. Pensions as severance pay. Pp. 57–89 in Z. Bodie and J.B. Shoven, eds., Financial Aspects of the United States Pension System. Chicago, Ill.: University of Chicago Press.

Lumsdaine, R., J. Stock, and D.A. Wise 1990. Efficient windows and labor force reduction. Journal of Public Economics 43:131–159.

1992a. Pension Plan Provisions and Retirement: Men & Women, Medicare, and Models. NBER Working Paper #4201. Cambridge, Mass.: National Bureau of Economic Research.

1992b. Three models of retirement: Computational complexity versus predictive validity. Pp 19–57 in D. Wise, ed., Topics in the Economics of Aging. Chicago, Ill.: University of Chicago Press.

Lumsdaine, R., and D.A. Wise 1994. Aging and labor force participation: A review of trends and explanations. Pp. 7–41 in D. Wise, ed., Aging in the United States and Japan: Economic Trends. Chicago, Ill.: University of Chicago Press.

Luzadis, R.A., and O.S. Mitchell 1991. Explaining pension dynamics. Journal of Human Resources 26(4):679–703.


McGarry, K., and R.F. Schoeni 1994. Transfer Behavior: Measurement and the Redistribution of Resources Within the Family. Paper presented at Health and Retirement Survey Early Results Workshop, Survey Research Center, University of Michigan. Revised.

Mitchell, O.S. 1988. Worker knowledge of pension provisions. Journal of Labor Economics 6(1):28–39.

1992. Trends in pension benefit formulas and retirement provisions. Pp. 177–216 in J.A. Turner and D.J. Beller, eds., Trends in Pensions 1992. Washington, D.C.: U.S. Department of Labor, Pension and Welfare Benefits Administration.

Mitchell, O.S., and R.A. Luzadis 1988. Changes in pension incentives through time. Industrial and Labor Relation Review 42(1):100–108.

Modigliani, F. 1988. The role intergenerational transfers and life cycle saving in the accumulation of wealth. The Journal of Economic Perspectives 2(2):15–40.

Munnell, A.H., and F.O. Yohn 1992. What is the impact of pensions on savings? Pp. 115–139 in Z. Bodie and A.H. Munnell, eds., Pensions and the Economy: Sources, Uses, and Limitations of Data. Philadelphia, Pa.: Pension Research Council Publications and University of Pennsylvania Press.


Nalebuff, B., and R.J. Zeckhauser 1985. Pensions and the retirement decision. Pp. 283–316 in D.A. Wise, ed., Pensions, Labor, and Individual Choice. Chicago, Ill.: University of Chicago Press .


Parsons, D.O. 1991. The decline in private pension coverage in the United States. Economics Letters 36:419–423.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

1993. The Contraction in Pension Coverage. Unpublished manuscript. Ohio State University.

Peracchi, F., and F. Welch 1994. Trends in labor force transitions of older men and women. Journal of Labor Economics 12(2):210–242.

Poterba, J.M., ed. 1994. Public Policies and Household Savings. Chicago, Ill.: University of Chicago Press.

Poterba. J.M., S.F. Venti, and D.A. Wise 1993. Do 401(k) Plans Crowd Out Other Retirement Savings? NBER Working Paper #4391. Cambridge, Mass.: National Bureau of Economic Research.

1994. Targeted retirement saving and the net worth of elderly Americans. American Economic Review 84(2):180CH:150–185.


Reimers, C., and M. Honig 1993a. The perceived budget constraint under Social Security: Evidence from reentry behavior. Journal of Labor Economics 11(1)Part 1:184–204.

1993b. Responses to Social Security by Men and Women: Myopic and Far-Sighted Behavior. Paper presented at the 1993 meeting of the Association for Public Policy Analysis and Management.

Ruhm, C.J. 1992. Secular Changes in the Work and Retirement Patterns of Older Men. Unpublished manuscript. University of North Carolina, Greensboro.

Rust, J. 1989. A dynamic programming model of retirement behavior. Pp. 359–398 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

1990. Behavior of male workers at the end of the life cycle: An empirical analysis of states and controls. Pp. 317–379 in D.A. Wise, ed., Issues in the Economics of Aging. Chicago, Ill.: University of Chicago Press.

Rust, J., and C. Phelan 1993. How Social Security and Medicare Affect Retirement Behavior in a World of Incomplete Markets. Unpublished manuscript. Department of Economics, University of Wisconsin.


Samwick, A.A. 1993a. The Joint Effect of Social Security and Pensions on the Timing of Retirement: Some New Evidence. Unpublished manuscript. Dartmouth College.

1993b. Wage Risk Compensation Through Employer-Provided Pensions. Unpublished manuscript. Dartmouth College.

1994. The Limited Offset Between Pension Wealth and Other Private Wealth: Implications of Buffer Stock Saving. Unpublished manuscript. Dartmouth College.

Sheiner, L., and D.N. Weil 1992. The Housing Wealth of the Aged. NBER Working Paper #4115. Cambridge, Mass.: National Bureau of Economic Research.

Skinner, J. 1988. Risky income, life-cycle consumption and precautionary savings. Journal of Monetary Economics 22:237-255.

1993. Is Housing Wealth a Sideshow? NBER Working Paper #4552. Cambridge, Mass.: National Bureau of Economic Research.

Smith, J.P. 1994. Marriage, Assets, and Savings. Unpublished manuscript. The RAND Corporation, Santa Monica, Calif..

Soldo, B.J., and M.S. Hill 1993. Family Structure and Transfer Measures in the HRS: Background and Overview. Paper presented at Health and Retirement Survey Early Results Workshop, Survey Research Center, University of Michigan.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×

Stern, S., and P. Todd 1993. A Test of Lazear's Mandatory Retirement Model. Unpublished manuscript. University of Virginia.

Stock, J.H., and D.A. Wise 1990a. The pension inducement to retire: An option value analysis. Pp. 205–224 in D.A. Wise, ed., Issues in the Economics of Aging. Chicago, Ill.: University of Chicago Press.

1990b. Pensions, the option value of work, and retirement. Econometrica 58(5):1151–1180.


Thaler, R.H. 1994. Psychology and savings policies. American Economic Review 84(2):186–192.

Turner, J.A., and D.J. Beller, eds. 1992. Trends in Pensions 1992. Washington, D.C.: U.S. Department of Labor, Pension and Welfare Benefits Administration.


Venti, S.F., and D.A. Wise 1987. IRAs and saving. Pp. 7–52 in M. Feldstein, ed., The Effects of Taxation on Capital Accumulation. Chicago, Ill.: University of Chicago Press.

1989. Aging, moving and housing wealth. Pp. 9–48 in D.A. Wise, ed., The Economics of Aging. Chicago, Ill.: University of Chicago Press.

1990. But they don't want to reduce housing equity. Pp. 13–29 in D.A. Wise, ed., Issues in the Economics of Aging. Chicago, Ill.: University of Chicago Press.

1993. The Wealth of Cohorts and the Changing Assets of Older Americans. NBER Working Paper #4600. Cambridge, Mass.: National Bureau of Economic Research.


Wise, D.A., and R.G. Woodbury 1994. Policy toward the aged. Pp. 741–780 in M. Feldstein, ed., American Economic Policy in the 1980s. Chicago, Ill.: University of Chicago Press.

Woodbury, S.A. 1983. Substitution between wage and nonwage benefits. American Economic Review 73(1):166–182.

Woodbury, S.A., and W. Huang 1991. The Tax Treatment of Fringe Benefits. Kalamazoo, Mich.: W.E. Upjohn Institute for Employment Research..


Zeldes, S.P. 1989a. Consumption and liquidity constraints: An empirical investigation. Journal of Political Economy 97(2):305–346.

1989b. Optimal consumption with stochastic income: Deviations from certainty equivalence. Quarterly Journal of Economics 104:275–298.

Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 11
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 12
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 13
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 14
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 15
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 16
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 17
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 18
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 19
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 20
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 21
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 22
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 23
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 24
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 25
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 26
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 27
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 28
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 29
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 30
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 31
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 32
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 33
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 34
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 35
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 36
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 37
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 38
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 39
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 40
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 41
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 42
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 43
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 44
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 45
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 46
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 47
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 48
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 49
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 50
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 51
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 52
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 53
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 54
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 55
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 56
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 57
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 58
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 59
Suggested Citation:"2 INCOME AND WEALTH OF OLDER AMERICAN HOUSEHOLDS: MODELING ISSUES FOR PUBLIC POLICY ANALYSIS." National Research Council. 1996. Assessing Knowledge of Retirement Behavior. Washington, DC: The National Academies Press. doi: 10.17226/5367.
×
Page 60
Next: 3 FACTORS AFFECTING LABOR SUPPLY DECISIONS AND RETIREMENT INCOME »
Assessing Knowledge of Retirement Behavior Get This Book
×
Buy Paperback | $55.00 Buy Ebook | $43.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This book brings together in one volume what researchers have learned about workers, employers, and retirees that is important for formulating retirement income policies. As the U.S. population ages, there is increasing uncertainty about the solvency of the Social Security and Medicare systems and the adequacy of private pensions to provide for people's retirement needs. The volume covers such critical behaviors as workers' decisions to retire, people's choices of saving over consumption, and employers' decisions about hiring older workers and providing pension and health care benefits. Also covered are trends in mortality, health status, and health care costs that are key to projecting the likely costs and effects of alternative retirement income security policies and a strategy for combining data and research knowledge into a policy modeling framework.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!