Income Trajectories in Later Life: Longitudinal Evidence from the Health and Retirement Study

Abstract

We track low-income respondents in the longitudinal Health and Retirement Study for 23 years, to observe how their financial situations unfolded as they aged. We document that (a) real incomes remained relatively stable as individuals entered retirement and progressed through their later years; and (b) labor force participation declined and thus earnings became less important with age, while Social Security and retirement savings rose as a proportion of annual income. Low-income people near retirement also tended to fare poorly during retirement.

During the transition from work to retirement, peoples’ earnings decline while reliance on Social Security, pensions, and personal saving rises. Some key questions associated with this life course transition are: 61To evaluate these queries and to assess how individuals fare financially as they move into retirement, this study follows individuals surveyed in the initial Health and Retirement Study (HRS) for over two decades. While there is considerable evidence from cross-sectional surveys illustrating age/income differences across population subgroups, longitudinal data on specific individuals is required to track actual changes in income as individuals enter retirement.

The HRS is the preeminent source of information for following the financial fortunes of older Americans. Using this dataset, we track individuals initially age 51–61 in 1992, to age 75–85 in 2016. In particular, we examine whether and how much older Americans’ incomes varied as they transitioned from work to retirement. We devote special attention to work, saving, health, and other measurable factors.

In what follows, we first describe how we identify low-income older adults, and how we follow them over time. The Original HRS cohort was inducted into the study in 1992 when respondents were age 51–61, so first we report on the factors associated with being a low-income respondent in this cohort, at baseline. Next, we estimate age/income profiles and explore the factors associated with income changes among the same cohort, in later years. A subsequent section discusses how results change when we include the annuitized value of wealth in older persons’ financial resources. In an extension, we compare the Original HRS group with the War Babies cohort (WBB) first included in 1998, and the Early Baby Boomer cohort (EBB) first included in 2004. We follow all three cohorts though their 2016 interviews.¹ Throughout, all dollar values are provided in real $2019 terms, and all results are weighted using sample weights provided by the HRS.

To this end, for each household we combine HRS reports on earnings from employment and self-employment, income from pensions or annuities, income from Social Security, unemployment and worker compensation benefits, and household capital income.² In order to compare households of different sizes, we adjust these money income values using the conventional household equivalency metric used by both the Congressional Budget Office and the Organization for Economic Cooperation & Development.³ This adjustment divides total household money income by the square root of the number of people in the household to obtain the household’s Adjusted Money Income. In a subsequent analysis, we also compute Adjusted Full Income measures over time, where this measure includes the annuitized value of the respondent’s net wealth.⁴

Our main findings are as follows:

1. Factors closely associated with being in the lowest income quartiles at baseline included being Black or Hispanic, female, having less education, being nonmarried, not working for pay, being disabled, and having children under age 18 at home. Additionally, respondents who resided in the South were systematically more likely to be in the lowest income quartile.

2. Respondents initially found in the two lowest income quartiles at baseline remained at about the same real income levels throughout retirement. This real income stability occurred as Social Security and retirement saving replaced declining earnings. The bottom quartile group remained close to poverty across all waves.

3. People in the highest baseline groups saw large improvements in income as they aged, ending up with values 30% higher than at baseline.

4. Including the annuitized share of wealth when measuring peoples’ access to resources improves our measure of many elderly respondents’ financial conditions. Nevertheless, even after taking household wealth into account, Blacks and Hispanics, women, the least educated, disabled persons, nonmarrieds, residents of the US South, and those having underage children and health problems still fared relatively worse in later life.

While there have been prior studies of peoples’ financial status in retirement, there are relatively few that track different cohorts of individuals as they age. For instance, Brown et al. (2020) examined only two waves of the HRS, 1994 and 2014, beginning with people initially age 57–62. By contrast, we follow respondents from three separate birth cohorts, and we evaluate their status every two years from their baseline wave through 2016. Accordingly, we have a far more detailed and granular perspective of the factors associated with financial conditions at older ages, compared to prior research. The primary advantage of having longitudinal data is that we can trace peoples’ financial status as they underwent both temporary and permanent shocks as they aged. In addition, for this sample, we can document changes in their sources of income over time.

Methodology

We begin by examining the Original HRS cohort, where we first focus on a sample of 9,955 individuals initially age 51–61 at baseline in 1992. For this sample, we have 13 waves of data enabling us to assess their financial status every other year from 1992 to 2016. For this analysis, we collected each household’s money income (e.g., labor earnings, pension benefits, Social Security benefits, disability and welfare benefits), withdrawals from accounts (e.g., IRAs, bank accounts), self-employment income, consulting income, and any other income (see Online Appendix Table 1 for descriptive statistics).⁵ Next, we adjusted total household money income by family size to obtain an individual’s Adjusted Money Income (these measures of income are identical in single person households). In separate robustness analysis, we also added in the annuitized value of household wealth including net financial and nonfinancial assets; we then adjusted household wealth by the same equivalency measure and computed the annuity value of this wealth if the respondent were to convert his or her share to an income flow in retirement.⁶ This second measure we call Adjusted Full Income, also summarized in Online Appendix Table 1.

The HRS is a very rich dataset, as it contains information on each respondent’s age, sex, education, race/ethnicity, current and past marital/partnered status, labor force status, self-reported health (limitations of daily living, the respondent’s depression score, self-assessed chance that the respondent will live to some future age, high blood pressure, diabetes, cancer, lung disease, heart disease, had suffered a stroke, psychiatric disease, arthritis, ulcer, cognition score, numeracy score), and whether the respondent had health insurance status (none, private, public).

Figure 1 depicts the distribution of the Adjusted Money Income values at baseline for the Original HRS cohort first surveyed in 1992 (in $2019). According to this metric, baseline median Adjusted Money Income for respondents age 51–61 was around $44,795 (in $2019), with 1% having no or negative earnings, and 4.1% earning over $150,000. Next, we split the baseline sample into four equal sized Adjusted Money Income quartiles, shown at the bottom of Figure 1.⁷ The lowest group, Q1, had annual median Adjusted Money Income of $11,411; Q2 had $30,770; Q3 had median income of around $53,504; and Q4, the highest income group, had median income of around $94,050. Figure 1 also indicates that there were about 2,400 respondents per quartile at baseline, and the quartile cutoffs for Q2, Q3, and Q4 were, respectively, $21,024, $41,596, and $68,345.⁸

Figure 1. Adjusted Money Income for Original HRS Baseline (in $2019).

Note: The sample analyzed includes all HRS respondents age 51–61 having adjusted money income at baseline; see text. Data weighted.

Before turning to additional findings, two comments are in order regarding the use of longitudinal HRS data. First, there is the potential for non-random attrition in the HRS over the long period we track people, which could lead to biased estimates for coefficients of interest (Bound et al. 2001).⁹ This possibility was previously evaluated in the HRS by Kapteyn et al. (2006), Michaud et al. (2011), and Meijer et al. (2010), and their analyses reported little evidence for bias due to selective attrition. It is still conceivable though, that over the 23 years we track Original HRS respondents, that attrition could selectively sort against the low-income. Indeed, we do find that people in the bottom two quartiles experienced higher mortality compared to the top two quartiles.¹⁰

Second, the HRS income and wealth variables (as well as the independent variables) could be measured with error, potentially biasing our estimates of the factors associated with income transitions (Bound et al. 2001). We leave for future work an analysis of this possibility, because Meijer and Karoly (2016), Meijer et al. (2010), and Sierminska et al. (2008) have compared income and wealth self-reports for HRS subsamples with administrative records from the Social Security Administration, the Survey of Consumer Finances, and the Panel Study of Income Dynamics. In general, they found that income and wealth measures in the HRS suffered less from measurement error than the other surveys.¹¹ Moreover, it is likely that measurement error is heterogenous across the older HRS sample, as suggested by Mazzona and Peracchi’s (2021) finding that older HRS respondents suffering cognitive declines were also more likely to experience drops in wealth.

Results: Lowest Income Recipients in the Original HRS Cohort at Baseline

To evaluate the factors associated with being in the lowest adjusted money income quartile at baseline, we focus initially on the Original HRS respondents first surveyed when they were age 51–61. The factors associated with being in the lowest income quartile, Q1, are derived from a multivariate logit regression analysis with estimated marginal effects reported in Table 1. Here, the first Column uses an abbreviated set of controls, while Column 2 includes additional variables such as health, insurance coverage, and region controls. All variables were measured at baseline.

Table 1.

Probability of Adjusted Money Income Being in Lowest Quartile at Baseline (Logit Marginal Effects Reported): Original HRS cohort (in $2019)

	Adj. Money Income in Q1 (0/1)
Age 52	0.002		0.007
	(0.019)		(0.019)
Age 53	0.006		0.010
	(0.018)		(0.019)
Age 54	0.017		0.031
	(0.019)		(0.021)
Age 55	0.020		0.024
	(0.020)		(0.020)
Age 56	0.010		0.019
	(0.019)		(0.020)
Age 57	0.031		0.045	*
	(0.020)		(0.021)
Age 58	0.039		0.050	*
	(0.021)		(0.022)
Age 59	0.031		0.053	*
	(0.021)		(0.024)
Age 60	0.031		0.058	*
	(0.021)		(0.024)
Age 61	0.064	*	0.089	**
	(0.026)		(0.029)
Male	−0.029	**	−0.035	**
	(0.007)		(0.008)
Black	0.097	**	0.074	**
	(0.015)		(0.015)
Race, others	0.036		0.015
	(0.026)		(0.024)
Hispanic	0.119	**	0.080	**
	(0.023)		(0.022)
Education years	−0.029	**	−0.021	**
	(0.002)		(0.002)
Married	−0.276	**	−0.233	**
	(0.014)		(0.016)
#Marriages	−0.008		−0.018	**
	(0.006)		(0.006)
#Children<18 yr	0.058	**	0.053	**
	(0.008)		(0.008)
Working for pay	−0.204	**	−0.106	**
	(0.012)		(0.013)
Disabled	0.186	**	0.081	**
	(0.033)		(0.027)
Poor health			0.063	**
			(0.014)
CESD score			0.003
			(0.002)
#Health problems			0.008	*
			(0.004)
Prob live to 75			−0.017
			(0.015)
Covered by fed. Govt HI			−0.012
			(0.012)
Covered by priv. HI			−0.211	**
			(0.018)
Covered by ER HI			−0.052	**
			(0.011)
Census region, northeast	0.016		0.027
	(0.018)		(0.018)
Census region, midwest	0.015		0.025
	(0.016)		(0.017)
Census region, south	0.060	**	0.052	**
	(0.015)		(0.015)
N	9,608		9,608
Pseudo R-sq	0.27		0.33
Dep. var. mean	0.22		0.22
Dep. var. st. dev.	0.41		0.41

Note: Adjusted Money Income includes all sources of adjusted household income; see text. Reference levels: age 51, white, west census region. Data weighted.

^*p≤0.05

^**p≤0.01.

Robust standard errors and clustered on HH

We see from Column 1 that Blacks and Hispanics, women, the least educated, and nonmarried persons were more likely to be found in Q1, as were disabled persons and people with underage children at home. Nonworking persons were also more likely to be in Q1, as were residents of the US South. These findings are robust to the inclusion of additional controls, as is evident from Column 2. That is, the magnitude and statistical significance of the estimated coefficients for Blacks and Hispanics, nonmarried, lower-educated, and women were similar after additional controls were included.

Having underage children also continues to predict Q1 Adjusted Money Income. In addition, we see that those in poor health and with health problems were more likely to have Q1 Adjusted Money Income than their counterparts, as were people without health insurance. Those working for pay were 20 percentage points less likely to have Q1 Adjusted Money Income in Column 1, though the effect halves after health and health insurance were controlled in Column 2.

One finding from Column 2 is that the probability of a respondent being in the Q1 Adjusted Money Income group at baseline was higher for older persons. That is, Column 2 shows that people age 56 and younger were half as likely to be in the lowest quartile at baseline, compared to people age 57–61. These predictors of low income at baseline conform to the findings of economic studies conducted in the 1990s as well as more recent research. Some of these findings could have been anticipated, but some were not, which speaks to the importance of both multivariate analysis and following people over time. As noted below, they also speak to the importance of social insurance policies.

Table 2 provides additional detail on the probability of being in any of the lowest three Adjusted Money Income quartiles (Q1, Q2, or Q3), versus being in the reference or highest quartile (Q4) at baseline. Marginal effects from multinomial Logit models are reported, and most of the results gleaned from Table 1 are reconfirmed here. For instance, age effects were mostly not significant up to about age 58. Thereafter, older people were more likely to be found in Q1 than in the higher income quartiles, and this effect was strongest for those over the age of 58. As before, men were less likely than women to be in any of the bottom three quartiles and more likely to have top quartile Adjusted Money Income.

Table 2.

Probability of Adjusted Money Income Being in Q1, Q2, or Q3 (vs. Q4) at Baseline (Logit Marginal Effects Reported): Original HRS cohort (in $2019)

	Quartiles of Adj. Money Income
	Q1 vs. Q4		Q2 vs. Q4		Q3 vs. Q4		Q1 vs. Q4		Q2 vs. Q4		Q3 vs. Q4
Age 52	0.003		0.014		−0.014		0.008		0.010		−0.015
	(0.020)		(0.025)		(0.023)		(0.021)		(0.025)		(0.023)
Age 53	0.007		0.038		−0.046	*	0.013		0.039		−0.047	*
	(0.019)		(0.025)		(0.022)		(0.020)		(0.026)		(0.022)
Age 54	0.019		0.010		−0.032		0.034		0.006		−0.038
	(0.020)		(0.025)		(0.023)		(0.022)		(0.026)		(0.024)
Age 55	0.024		0.040		−0.058	**	0.027		0.034		−0.058	*
	(0.021)		(0.026)		(0.022)		(0.022)		(0.027)		(0.023)
Age 56	0.013		0.040		−0.029		0.024		0.044		−0.034
	(0.020)		(0.026)		(0.023)		(0.022)		(0.027)		(0.024)
Age 57	0.036		0.016		−0.043		0.051	*	0.013		−0.050	*
	(0.022)		(0.026)		(0.023)		(0.023)		(0.027)		(0.023)
Age 58	0.045	*	0.048		−0.049	*	0.058	*	0.040		−0.052	*
	(0.022)		(0.028)		(0.024)		(0.024)		(0.028)		(0.025)
Age 59	0.038		0.070	*	−0.055	*	0.061	*	0.058	*	−0.064	**
	(0.023)		(0.028)		(0.024)		(0.026)		(0.029)		(0.024)
Age 60	0.038		0.048		−0.048	*	0.068	**	0.042		−0.061	*
	(0.022)		(0.028)		(0.024)		(0.026)		(0.028)		(0.024)
Age 61	0.073	**	0.049		−0.054	*	0.100	**	0.041		−0.067	*
	(0.028)		(0.031)		(0.027)		(0.031)		(0.032)		(0.027)
Male	−0.032	**	−0.039	**	0.046	**	−0.039	**	−0.033	**	0.043	**
	(0.008)		(0.009)		(0.009)		(0.009)		(0.011)		(0.010)
Black	0.111	**	0.027		−0.057	**	0.087	**	0.035		−0.050	**
	(0.015)		(0.017)		(0.017)		(0.016)		(0.019)		(0.018)
Race, others	0.047		0.004		0.012		0.021		−0.001		0.028
	(0.028)		(0.036)		(0.039)		(0.027)		(0.038)		(0.041)
Hispanic	0.153	**	0.048		−0.113	**	0.114	**	0.060	*	−0.102	**
	(0.024)		(0.027)		(0.023)		(0.024)		(0.028)		(0.025)
Education years	−0.036	**	−0.034	**	0.007	**	−0.027	**	−0.033	**	0.004
	(0.002)		(0.002)		(0.003)		(0.002)		(0.003)		(0.003)
Married	−0.295	**	−0.042	**	0.152	**	−0.254	**	−0.063	**	0.146	**
	(0.014)		(0.015)		(0.013)		(0.016)		(0.016)		(0.014)
#Marriages	−0.011		−0.008		0.003		−0.022	**	−0.009		0.010
	(0.007)		(0.009)		(0.009)		(0.007)		(0.009)		(0.010)
#Children<18 yr	0.072	**	0.054	**	−0.035	**	0.068	**	0.056	**	−0.035	**
	(0.008)		(0.011)		(0.012)		(0.008)		(0.012)		(0.012)
Working for pay	−0.219	**	0.004		0.106	**	−0.117	**	−0.018		0.062	**
	(0.013)		(0.013)		(0.012)		(0.014)		(0.016)		(0.015)
Disabled	0.209	**	−0.014		−0.102	**	0.096	**	−0.009		−0.056
	(0.035)		(0.036)		(0.038)		(0.029)		(0.043)		(0.046)
Poor health	0.015		0.002		−0.017		0.027		0.000		−0.024
	(0.019)		(0.023)		(0.023)		(0.020)		(0.024)		(0.024)
CESD score	0.015		0.014		0.012		0.026		0.009		0.005
	(0.017)		(0.022)		(0.022)		(0.018)		(0.022)		(0.022)
#Health problems	0.064	**	−0.025		−0.006		0.056	**	−0.028		−0.002
	(0.016)		(0.020)		(0.020)		(0.017)		(0.021)		(0.021)
Prob live to 75							0.074	**	0.040	*	−0.043	*
							(0.015)		(0.018)		(0.018)
Covered by fed. Govt HI							0.005		0.007	*	0.000
							(0.002)		(0.003)		(0.003)
Covered by priv. HI							0.010	*	0.012	*	−0.013	*
							(0.004)		(0.006)		(0.006)
Covered by ER HI							−0.019		−0.010		−0.025
							(0.016)		(0.023)		(0.022)
Census region, northeast							−0.022		−0.093	**	0.054	*
							(0.013)		(0.021)		(0.026)
Census region, midwest							−0.238	**	−0.019		0.139	**
							(0.019)		(0.019)		(0.018)
Census region, south							−0.055	**	−0.001		0.045	**
							(0.012)		(0.013)		(0.012)
N					9,608						9,608
Pseudo R-sq					0.16						0.19
Dep. var. mean					2.59						2.59
Dep. var. st. dev.					1.11						1.11

Note: See Table 1.

^*p≤0.05

^**p≤0.01.

Blacks and Hispanics were also more likely to be in Q1 than Q4, as were the least educated and nonmarried persons. Being in poor health or disabled, not having health insurance, and not working for pay are all factors clearly associated with worse economic standing, as before. In terms of quantitative magnitudes, not working for pay was associated with a 12–22 percentage point higher chance of being in the lowest compared to the highest quartile. Residents of the US South were roughly as likely to be in the lowest quartile as are respondents having underage children. Overall, the baseline results tell a consistent story about the directionality of the factors predictive of low incomes at baseline, when most of our sample was nearing retirement. In other words, the factors that contribute to financial vulnerability during the working life also contribute to vulnerability in retirement.

Results: Age/Income Profiles Over Time for the Original HRS Cohort

In this section, we ask how real income changed with age for this Original HRS cohort, and whether these changes differed across respondents in the quartiles of the income distribution. That is, we examine how Adjusted Money Income changed over time for the respondents first observed at baseline in 1992 when they were age 51–61. To this end, we classified each respondent as before according to his or her Adjusted Money Income quartile at baseline, and we then followed the respondents in every wave observed thereafter. To trace age/income trajectories, we regressed (ln) Adjusted Money Income on a set of age controls, with age 51 as the reference category.¹² Other controls also included are all of the socio-demographic factors in Table 1, and controls for the year of interview. As before, all dollar values are in $2019.¹³

A plot of the results appears in Figure 2, which illustrates the average percentage change in Adjusted Money Income by respondent age and initial quartile.¹⁴ As is clear from the figure, Adjusted Money Incomes across the full set of respondents (black line) remained relatively stable in real terms from ages 52 to 82. Interestingly, people whose baseline Adjusted Money Incomes were initially in the lowest two quartiles (Q1 red line, Q2 blue line) experienced the largest fluctuations in their Adjusted Money Incomes with age. Nevertheless, incomes for those in Q1 continued to hover right around the poverty line over time, without much improvement (see Online Appendix Table 4). This age pattern reflects the fact that many were disabled prior to age 62 (see Online Appendix Table 5), plus the fact that numerous Q1 respondents began receiving Social Security and pension benefits around that age (see Online Appendix Table 6).¹⁵ An additional consideration might be that people reporting the lowest annual incomes at baseline (below $10,000/year) could have experienced a recent income shock, and then their subsequent Adjusted Money Income might have reverted to more normal levels thereafter (Hudomiet 2015). This is not evident in our data, however.¹⁶

Figure 2. Percentage Changes in Adjusted Money Income by Age in the Panel and Quartile of Baseline Adjusted Money Income Quartile: Original HRS at Baseline Followed Over Time (in $2019).

Notes: Quartiles of Adjusted Money Income determined at baseline (Q1–4); see Figure 1. The profiles represent estimated age effects from a regression of ln(adj. money income) on age, demographic factors, and interview year. Data weighted.

Individuals in Q2 had the relatively best experience during the sample period, as their Adjusted Money Income remained higher than at baseline throughout retirement. Turning to the two top quartiles at baseline, Q3–4, Figure 2 indicates that they experienced rather different trajectories. After about age 62, the Q3 (yellow line) group experienced a steady decline in their Adjusted Money Incomes until age 82. At that point, the Original HRS cohort in Q3 at baseline had 35% less Adjusted Money Income than they did at age 52, indicating that upper middle-income households were not necessarily able to maintain their preretirement standard of living late in retirement. Money incomes of the top quartile, Q4 (green line), actually fell from about age 53 onwards, perhaps reflecting the relatively lower replacement rates paid by Social Security as well as monetary limits on employer pension plans.

Our analysis of the original HRS cohort pattern over time illustrates that, over the 23-year period, real household income remained relatively stable; however, income fluctuations differed across income quartiles measured at baseline. In sum, those initially in the lowest Adjusted Money Incomes at baseline had relatively stable (and low) incomes, while those in higher baseline quartiles saw their Adjusted Money Incomes decline in real terms. Such an analysis of income fluctuations is, of course, only possible using longitudinal data. The HRS provides the longest such data for older households, and our findings yield a clear pattern of age/income profiles as individuals moved into and through retirement.

Results: Explaining the Stability of Real Income

The age/income patterns observed above occurred as the proportion of respondents working for pay in each quartile fell rapidly with age. Approximately 80% of individuals in the top three quartiles were working for pay in their early 50s, while by age 65 only about 40% were working for pay; by age 82, virtually all of the respondents had left the labor force (see Figure 3 and Online Appendix Table 6).¹⁷ Despite the rapid decline in the probability of working, real income remained relatively constant between ages 62–72. The probability of working for Q1 respondents was much lower at baseline than those in other quartiles, and members of Q1 continued to have lower labor force participation rates up to age 82.

Figure 3. Proportion of Respondents in Each Quartile Working for Pay at Each Age.

Note: Percent working for pay by baseline Adjusted Money Income quartile; see Figure 1.

Of course, as individuals leave the labor force, money from earnings declines; therefore, if real income is to remain constant, other sources of income must increase. To examine the changing contribution of various income sources, we calculated the share of annual income for each respondent attributable to earnings, Social Security, unemployment and worker compensation, pensions and annuities, and capital income. Figure 4 shows the decline in the share of income for the entire cohort due to earnings falling from 75–80% in the mid-50s to essentially zero by age 80. Over the same ages, the share of income due to Social Security rose from less than 5% to over 60%, with smaller increases in income shares for pensions and capital income.¹⁸

Figure 4. Shares of Adjusted Money Income by Age: Total and by Quartile A.

Notes: See Figure 1.

Even more interesting are the changes in income shares for the four quartiles. Beginning with respondents in the lowest quartile, we observe that earnings initially represented a good share of total income (about 50%), and then they declined to under 10%. By contrast, the share of income due to Social Security rose from around 30% when respondents were in their 50s, to over 80% when they reached their late 60s. The relatively high rate of Social Security receipt prior to age 62 suggests that these individuals were drawing income from either the Disability or Supplemental Security Income program, or received benefits as surviving spouses caring for young children. The dotted line indicates that over 60% of these low-income households received 90% or more of their income from Social Security. Given that Social Security benefits are indexed for inflation, it is easy to see how the income of those in the lowest quartile remained relatively constant in real terms.

Similar changes in income shares occurred for respondents in the second and third quartiles. For respondents in the highest quartile, capital income and pensions were more important, with each representing about 20% of total income. Interestingly, even for individuals in the highest quartiles, benefits from Social Security represented over 40% of annual income.

Results: Integrating Respondent Wealth Into A Comprehensive Measure for the Original HRS Cohort

Previous sections focused only on money income to trace peoples’ financial fortunes over time. In this section, as described above, we also incorporate the baseline level of each household’s net wealth by converting its baseline wealth into an equivalent income stream. Our goal is to establish what each respondent could have obtained, if his or her share of household wealth had been converted to an annuity at baseline. To derive this measure, we first divide baseline household wealth for an individual living in a multi-person household by the number of (adult) co-residents, if any. Next, we apply an appropriate age/sex annuity factor (Academy of Actuaries 2012) to the resulting wealth allocation, to determine what the annuitized total income of that respondent would be, and we add it to Adjusted Money Income. In what follows, we call this Adjusted Full Income, which is the sum of Adjusted Money Income plus the adjusted baseline value of annuitized household wealth.

Figure 5 reports the distribution of Adjusted Full Income at baseline. A few people had negative wealth and no money income (most of these lived with other persons); about 37% had Adjusted Full Incomes of under $40,000 per year; and about 15% had measured Adjusted Full Incomes of over $100,000 per year. The Figure also reports quartiles of Adjusted Full Income for the Original HRS cohort at baseline, labeled as EWQs to distinguish them from the Adjusted Money Income quartiles (Q1-Q4) in the discussion above. At the bottom of Figure 5, we see that the median Adjusted Full Income was $13,569 for the lowest quartile (EWQ1), and for EWQ2/3/4, respectively, $36,400, $63,045, and $114,361. Average Adjusted Full Income was 23% above average Adjusted Money Income ($68,391 versus $55,726), and median Adjusted Full Income exceeded median Adjusted Money Income by 18% ($53,047 versus $44,795), In other words, a comparison of Figures 1 and 5 confirms that all Adjusted Full Income quartiles indicate greater access to resources than the Adjusted Money Income measures, though the Q1 group was far less wealthy than were the other quartiles.

Figure 5. Adjusted Full Income: Original HRS at Baseline (in $2019).

Note: The sample analyzed includes those having Full Income adjusted for household size, computed by summing adjusted income plus annuitized wealth. Data weighted.

Figure 6 tracks the percentage changes in Adjusted Full Incomes for each baseline quartile as respondents aged. Interestingly, the average across all quartiles (black line) traced a gradual but steady upward trajectory from age 62 onward, ending up with Adjusted Full Income about 30% higher than at the outset. This assessment of economic conditions is more positive than the impression gleaned from focusing only on Adjusted Money Income in the earlier Figures.¹⁹

Figure 6. Percentage Changes in Adjusted Full Income by Age in the Panel and Quartile of Initial Full Income: Original HRS Followed Over Time (in $2019).

Note: Quartiles of Adjusted Full Income (adjusted income plus annuitized wealth at baseline) are determined at baseline (EWQ1–4). The profiles represent estimated age effects from a regression of ln(Adj. Full Income) on age, demographic factors, and interview year. Baseline quartile cutoffs appear in Figure 5. Data weighted.

Note: Percent working for pay by baseline Adjusted Full Income quartile.

We also see from Figure 6 that the lowest EWQ1 baseline quartile (red line) experienced some small improvements in its Adjusted Full Income after age 62 up to age 75, whereas those in the highest (EWQ4) quartile had a steadily rising Adjusted Full Income throughout the sample period. While the first pattern replicates what was seen in Figure 2, the improvement in top quartile Adjusted Full Income is much more strongly positive.²⁰ By contrast, those initially in the second (blue line) and third quartiles (yellow line) experienced relatively little upward or downward movement in Adjusted Full Incomes with age.

Table 3 reports marginal effects from multivariate logit models of the probability that an Original HRS respondent was in the lowest Adjusted Full Income quartile at baseline (EWQ1). Here there are no statistically significant age effects, a result that differs from Table 1, indicating that including incorporating annuitized wealth contributes to less inequality by age.²¹ Nevertheless, as before, we again find that men were much less likely to be in the lowest Adjusted Full Income quartile, while Blacks and Hispanics, the least educated, and nonmarried persons were more likely, as were Southerners, the disabled, and those with underage children. As before, people lacking health insurance were also more likely to be in the lowest Adjusted Full Income quartile. Those still working for pay were 8 percentage points less likely to have the lowest (EWQ1) Adjusted Full Income when health and health insurance are controlled in Column 2.

Table 3.

Probability of Adjusted Full Income Being in Lowest Quartile (EWQ1) at Baseline (Logit Marginal Effects Reported): Original HRS at Baseline (in $2019)

	Adj. Full Income in EWQ1 (0/1)
Age 52	−0.006		−0.002
	(0.017)		(0.017)
Age 53	−0.001		0.004
	(0.017)		(0.017)
Age 54	−0.020		−0.011
	(0.015)		(0.016)
Age 55	−0.014		−0.013
	(0.016)		(0.016)
Age 56	−0.011		−0.004
	(0.016)		(0.017)
Age 57	−0.004		0.006
	(0.017)		(0.017)
Age 58	0.003		0.010
	(0.018)		(0.018)
Age 59	−0.025		−0.008
	(0.016)		(0.017)
Age 60	−0.025		−0.005
	(0.016)		(0.017)
Age 61	−0.004		0.014
	(0.020)		(0.021)
Male	−0.032	**	−0.046	**
	(0.007)		(0.008)
Black	0.131	**	0.107	**
	(0.016)		(0.016)
Race, others	0.053		0.032
	(0.028)		(0.027)
Hispanic	0.118	**	0.076	**
	(0.023)		(0.022)
Education years	−0.031	**	−0.023	**
	(0.002)		(0.002)
Married	−0.290	**	−0.237	**
	(0.015)		(0.016)
#Marriages	0.002		−0.007
	(0.006)		(0.006)
#Children<18 yr	0.061	**	0.057	**
	(0.008)		(0.008)
Working for pay	−0.176	**	−0.082	**
	(0.012)		(0.012)
Disabled	0.204	**	0.086	**
	(0.036)		(0.028)
Poor health			0.060	**
			(0.013)
CESD score			0.007	**
			(0.002)
#Health problems			0.008	*
			(0.004)
Prob live to 75			−0.035	*
			(0.015)
Covered by fed. Govt HI			−0.011
			(0.012)
Covered by priv. HI			−0.249	**
			(0.019)
Covered by ER HI			−0.017
			(0.011)
Census region, northeast	0.006		0.014
	(0.016)		(0.017)
Census region, midwest	−0.004		0.005
	(0.015)		(0.015)
Census region, south	0.067	**	0.060	**
	(0.015)		(0.015)
N	9,608		9,608
Pseudo R-sq	0.30		0.37
Dep. var. mean	0.21		0.21
Dep. var. st. dev.	0.41		0.41

Note: See Table 1.

^*p<0.05

^**p<0.01.

Table 4 extends the analysis of Table 3 using a multinomial Logit model to evaluate the probability of appearing in each of the bottom three quartiles, versus being in the top adjusted Full Income quartile (EWQ4) at baseline. As in Table 3, age was not significantly related to the chances of being in a low-income quartile. Also, as before, men were less likely to be in the bottom three quartiles and were more likely to have top quartile Adjusted Full Incomes. Results for Blacks and Hispanics were significant across the board, and they confirm that these two population subgroups were always least likely to be in the highest Adjusted Full Income group.

Table 4.

Probability of Adjusted Full Income Being in EWQ1, EWQ2, or EWQ3 (vs. EWQ4) at Baseline (MLogit Marginal Effects Reported): Original HRS at Baseline (in $2019)

	Quartiles of Adj. Full Income
	EWQ1 vs. EWQ4		EWQ2 vs. EWQ4		EWQ3 vs. EWQ4		EWQ1 vs. EWQ4		EWQ2 vs. EWQ4		EWQ3 vs. EWQ4
Age 52	−0.005		0.017		−0.008		−0.002		0.014		−0.009
	(0.018)		(0.025)		(0.023)		(0.018)		(0.026)		(0.023)
Age 53	−0.002		0.010		−0.041		0.004		0.011		−0.042
	(0.017)		(0.025)		(0.022)		(0.018)		(0.025)		(0.022)
Age 54	−0.021		0.036		−0.048	*	−0.010		0.035		−0.053	*
	(0.016)		(0.026)		(0.023)		(0.017)		(0.026)		(0.023)
Age 55	−0.014		0.035		−0.049	*	−0.014		0.028		−0.047	*
	(0.017)		(0.026)		(0.023)		(0.017)		(0.027)		(0.023)
Age 56	−0.011		0.019		−0.029		−0.003		0.022		−0.034
	(0.017)		(0.025)		(0.023)		(0.019)		(0.026)		(0.024)
Age 57	−0.004		−0.011		−0.007		0.006		−0.015		−0.011
	(0.018)		(0.026)		(0.024)		(0.019)		(0.027)		(0.025)
Age 58	0.004		0.021		−0.038		0.013		0.013		−0.039
	(0.019)		(0.027)		(0.025)		(0.020)		(0.028)		(0.025)
Age 59	−0.023		0.052		−0.041		−0.007		0.041		−0.047
	(0.017)		(0.028)		(0.025)		(0.019)		(0.028)		(0.025)
Age 60	−0.025		0.022		−0.035		−0.004		0.016		−0.043
	(0.017)		(0.027)		(0.025)		(0.019)		(0.028)		(0.025)
Age 61	0.000		0.054		−0.043		0.019		0.046		−0.052
	(0.021)		(0.031)		(0.027)		(0.023)		(0.031)		(0.027)
Male	−0.035	**	−0.026	**	0.035	**	−0.051	**	−0.021		0.037	**
	(0.008)		(0.009)		(0.009)		(0.008)		(0.011)		(0.010)
Black	0.147	**	0.035	*	−0.075	**	0.123	**	0.043	*	−0.069	**
	(0.016)		(0.017)		(0.017)		(0.017)		(0.018)		(0.018)
Race, others	0.065	*	0.006		0.012		0.041		0.006		0.027
	(0.030)		(0.035)		(0.039)		(0.030)		(0.037)		(0.041)
Hispanic	0.147	**	0.057	*	−0.109	**	0.104	**	0.074	*	−0.097	**
	(0.024)		(0.027)		(0.024)		(0.024)		(0.029)		(0.026)
Education years	−0.039	**	−0.036	**	0.012	**	−0.030	**	−0.036	**	0.009	**
	(0.002)		(0.003)		(0.003)		(0.002)		(0.003)		(0.003)
Married	−0.308	**	−0.031	*	0.152	**	−0.257	**	−0.052	**	0.140	**
	(0.014)		(0.015)		(0.014)		(0.016)		(0.016)		(0.015)
#Marriages	0.000		−0.010		0.008		−0.011		−0.012		0.014
	(0.006)		(0.009)		(0.009)		(0.007)		(0.010)		(0.010)
#Children<18 yr	0.074	**	0.049	**	−0.033	**	0.070	**	0.052	**	−0.033	**
	(0.008)		(0.012)		(0.013)		(0.008)		(0.012)		(0.013)
Working for pay	−0.188	**	0.024		0.080	**	−0.088	**	0.010		0.029
	(0.012)		(0.013)		(0.013)		(0.013)		(0.016)		(0.015)
Disabled	0.229	**	0.002		−0.095	*	0.105	**	0.011		−0.033
	(0.037)		(0.038)		(0.038)		(0.030)		(0.045)		(0.048)
Poor health							0.070	**	0.042	*	−0.043	*
							(0.014)		(0.019)		(0.018)
CESD score							0.008	**	0.009	*	−0.004
							(0.002)		(0.004)		(0.003)
#Health problems							0.010	*	0.016	**	−0.006
							(0.004)		(0.006)		(0.006)
Prob live to 75							−0.039	*	−0.016		−0.003
							(0.015)		(0.023)		(0.023)
Covered by fed. Govt HI							−0.016		−0.055	*	0.027
							(0.013)		(0.023)		(0.026)
Covered by priv. HI							−0.273	**	0.017		0.153	**
							(0.020)		(0.019)		(0.017)
Covered by ER HI							−0.017		0.002		0.030	*
							(0.011)		(0.013)		(0.012)
Census region, northeast	0.006		0.059	*	−0.041		0.016		0.057	*	−0.050	*
	(0.017)		(0.025)		(0.023)		(0.018)		(0.025)		(0.023)
Census region, midwest	−0.004		0.079	**	0.006		0.006		0.074	**	−0.002
	(0.016)		(0.023)		(0.022)		(0.016)		(0.024)		(0.022)
Census region, south	0.074	**	0.022		−0.023		0.067	**	0.021		−0.021
	(0.016)		(0.021)		(0.020)		(0.016)		(0.022)		(0.021)
N					9,608						9,608
Pseudo R-sq					0.17						0.20
Dep. var. mean					2.61						2.61
Dep. var. st. dev.					1.11						1.11

Note: See Table 1.

^*p<0.05

^**p<0.01.

Having more education and being married did reduce the chances of being in the bottom quartile, while having more underage children, being in poor health, and not working for pay were associated with worse economic standing, as was residing in the US South. The quantitative impact of working for pay was attenuated in the second panel compared to the first, suggesting that the additional controls in the second panel – including being in good health and having private health insurance -- were stronger influences than working per se. Overall, these results continue to tell a consistent story about the directionality of the factors predictive of poor financial conditions at baseline, even after taking household wealth into account.

Robustness: Comparing the Original HRS with Subsequent Cohorts

With the support of the National Institutes of Health and the Social Security Administration, the HRS has included new cohorts of older Americans every six years since 1992 (when the original baseline group entered the study). For two of these additional cohorts, sufficient additional waves have now been fielded to enable a comparison with the original HRS cohort examined above. Specifically, in this section, we compare the original HRS cohort with the War Babies (WBB), age 51–56 in 1998, and the Early Baby Boomers (EBB) who turned age 51–56 in 2004. Both of these additional cohorts were surveyed every two years until the year 2016 (see Online Appendix Table 8 for descriptive statistics).²²

Figure 7 pools these three cohorts for a sample size of 14,180 into a single figure using the Adjusted Money Income measure (as before, in $2019). This expanded dataset has fewer respondents with zero or negative income, and a higher fraction with income over $150,000. Nevertheless, the distribution is similar in the mid-range of adjusted total income. To generate these comparisons, we utilize the same dollar amounts for the maximum amount cap for Q1-Q4 based on Adjusted Money Income for the Original HRS cohort at baseline.

Figure 7. Adjusted Money Income for Three Cohorts of Respondents: Original HRS, WBB, and EBB at their Baseline (in $2019).

Note: The sample analyzed includes three cohorts of respondents (original HRS, WBB, EBB) age 51–61 at baseline; see text. Quartiles are defined by same Adjusted Money Income thresholds in HRS Baseline (see Fig 1). Data weighted.

Note: Quartile thresholds defined as for Original HRS cohort at baseline and applied to subsequent cohorts, all in $2019.

To track these individuals through time, Figure 8 includes all three cohorts and traces the percentage changes in Adjusted Money Incomes by age and baseline quartile. As we saw in Figure 2, the overall average remained fairly constant (black line), but evidently the lowest Adjusted Money Income quartile (Q1) did somewhat better between age 52 and 72 (red line).²³ The Q2 group (blue line) here is similar to that in Figure 2, rising toward older ages. The top two quartiles’ Adjusted Money Incomes eroded somewhat with age, as in Figure 2.²⁴

Figure 8. Percentage Changes in Adjusted Money Income by Age in the Panel and Quartile of Adj. Money Income for Three Cohorts: Original HRS, WBB, and EBB Followed Over Time (in $2019).

Note: The profiles represent estimated age effects from a regression of ln(Adj. Money Income) on age, demographic factors, interview year. Baseline quartile cutoffs are the same as those defined for HRS Original cohort (see Figure 1). Data weighted.

Tables 5 and 6, respectively, report the probability of the pooled sample falling into the lowest Adjusted Money Income at baseline, which may be compared with Tables 1 and 2. As in Table 1, older individuals (age 57+) in Table 5 were more likely to be found in the lowest group, but the age effects are attenuated in Table 6 (as in Table 2). One of the most robust findings is that men were less likely to fall into Q1 in both Tables 5 and 6, supportive of the earlier findings. And once again, Blacks, Hispanics, the least educated, nonmarried, the disabled, Southern residents, and those with underage children were more at risk for falling into the lowest income group. As before, those in poor health and those lacking health insurances were also at greater risk. Those working for pay were also less likely to be found in the lowest quartile. Overall, then, a consistent story emerges from all three HRS cohorts examined. In general, the cohort dummy variables in the regressions are not significantly different from zero in the models with the additional controls. This implies that our results for the age/income profiles are robust across cohorts as well. Accordingly, the age pattern of real income in each of these HRS cohorts follows the same basic pattern.

Table 5.

Probability of Adjusted Money Income Being in Lowest Quartile at Baseline (Logit Marginal Effects Reported): Original HRS cohort, EBB, and WBB (in $2019)

	Adj. Money Income in Q1 (0/1)
Age 52	−0.001		0.000
	(0.013)		(0.013)
Age 53	0.006		0.004
	(0.013)		(0.013)
Age 54	0.013		0.018
	(0.014)		(0.014)
Age 55	0.031	*	0.029	*
	(0.015)		(0.014)
Age 56	0.014		0.020
	(0.015)		(0.015)
Age 57	0.029		0.038	*
	(0.018)		(0.018)
Age 58	0.036		0.043	*
	(0.019)		(0.019)
Age 59	0.028		0.047	*
	(0.019)		(0.021)
Age 60	0.028		0.051	*
	(0.018)		(0.020)
Age 61	0.058	*	0.076	**
	(0.023)		(0.025)
Male	−0.017	**	−0.026	**
	(0.006)		(0.006)
Black	0.085	**	0.060	**
	(0.011)		(0.011)
Race, others	0.037	*	0.021
	(0.019)		(0.018)
Hispanic	0.111	**	0.071	**
	(0.018)		(0.017)
Education years	−0.028	**	−0.020	**
	(0.001)		(0.001)
Married	−0.259	**	−0.216	**
	(0.011)		(0.012)
#Marriages	−0.006		−0.015	**
	(0.005)		(0.005)
#Children<18 yr	0.048	**	0.045	**
	(0.006)		(0.006)
Working for pay	−0.208	**	−0.106	**
	(0.011)		(0.011)
Disabled	0.166	**	0.070	**
	(0.025)		(0.020)
Poor health			0.057	**
			(0.010)
CESD score			0.003
			(0.002)
#Health problems			0.006	*
			(0.003)
Prob live to 75			−0.029	*
			(0.012)
Covered by fed. Govt HI			−0.005
			(0.010)
Covered by priv. HI			−0.201	**
			(0.015)
Covered by ER HI			−0.048	**
			(0.009)
WBB cohort	−0.022	*	−0.015
	(0.011)		(0.011)
EBB cohort	0.017		0.002
	(0.011)		(0.011)
Census region, northeast	0.009		0.018
	(0.014)		(0.014)
Census region, midwest	0.014		0.024
	(0.013)		(0.013)
Census region, south	0.053	**	0.043	**
	(0.012)		(0.012)
N	14,180		14,180
Pseudo R-sq	0.28		0.34
Dep. var. mean	0.21		0.21
Dep. var. st. dev.	0.41		0.41

Note:

^*p<0.05

^**p<0.01.

See Table 1.

Table 6.

Probability of Adjusted Money Income Being in Q1, 2, and 3 (versus 4) at Baseline (Logit Marginal Effects Reported): Original HRS cohort, EBB, and WBB (in $2019)

	Quartiles of Adj. Money Income
	Q1 vs. Q4		Q2 vs. Q4		Q3 vs. Q4		Q1 vs. Q4		Q2 vs. Q4		Q3 vs. Q4
Age 52	−0.001		0.000		−0.006		0.001		−0.001		−0.006
	(0.014)		(0.017)		(0.017)		(0.014)		(0.018)		(0.017)
Age 53	0.007		0.019		−0.016		0.005		0.021		−0.015
	(0.014)		(0.018)		(0.017)		(0.014)		(0.018)		(0.017)
Age 54	0.015		0.000		0.005		0.020		−0.004		0.003
	(0.014)		(0.018)		(0.018)		(0.015)		(0.018)		(0.018)
Age 55	0.036	*	0.022		−0.032		0.033	*	0.018		−0.031
	(0.016)		(0.019)		(0.018)		(0.016)		(0.019)		(0.018)
Age 56	0.018		0.021		−0.019		0.026		0.024		−0.024
	(0.016)		(0.020)		(0.019)		(0.017)		(0.021)		(0.019)
Age 57	0.033		0.003		−0.025		0.044	*	0.003		−0.031
	(0.019)		(0.023)		(0.022)		(0.020)		(0.024)		(0.022)
Age 58	0.043	*	0.036		−0.027		0.051	*	0.031		−0.031
	(0.020)		(0.024)		(0.023)		(0.021)		(0.025)		(0.024)
Age 59	0.037		0.057	*	−0.034		0.056	*	0.049		−0.043
	(0.020)		(0.025)		(0.023)		(0.022)		(0.025)		(0.023)
Age 60	0.036		0.036		−0.027		0.061	**	0.034		−0.040
	(0.020)		(0.024)		(0.023)		(0.022)		(0.025)		(0.023)
Age 61	0.069	**	0.038		−0.032		0.089	**	0.035		−0.044
	(0.025)		(0.027)		(0.027)		(0.027)		(0.028)		(0.027)
Male	−0.019	**	−0.035	**	0.030	**	−0.031	**	−0.031	**	0.030	**
	(0.006)		(0.007)		(0.007)		(0.007)		(0.008)		(0.008)
Black	0.100	**	0.049	**	−0.055	**	0.076	**	0.055	**	−0.050	**
	(0.012)		(0.014)		(0.014)		(0.012)		(0.015)		(0.015)
Race, others	0.048	*	0.033		−0.008		0.027		0.027		0.002
	(0.020)		(0.027)		(0.028)		(0.020)		(0.028)		(0.029)
Hispanic	0.145	**	0.055	*	−0.115	**	0.104	**	0.067	**	−0.104	**
	(0.020)		(0.022)		(0.019)		(0.019)		(0.023)		(0.020)
Education years	−0.036	**	−0.034	**	0.002		−0.027	**	−0.034	**	−0.001
	(0.001)		(0.002)		(0.002)		(0.001)		(0.002)		(0.002)
Married	−0.280	**	−0.058	**	0.135	**	−0.232	**	−0.075	**	0.126	**
	(0.011)		(0.012)		(0.011)		(0.013)		(0.013)		(0.012)
#Marriages	−0.009		−0.011		0.003		−0.019	**	−0.015		0.008
	(0.005)		(0.007)		(0.008)		(0.005)		(0.008)		(0.008)
#Children<18 yr	0.060	**	0.050	**	−0.018	*	0.057	**	0.052	**	−0.019	*
	(0.006)		(0.008)		(0.009)		(0.006)		(0.008)		(0.009)
Working for pay	−0.224	**	−0.005		0.096	**	−0.120	**	−0.022		0.052	**
	(0.011)		(0.011)		(0.011)		(0.012)		(0.013)		(0.013)
Disabled	0.186	**	−0.006		−0.086	**	0.080	**	−0.006		−0.047
	(0.026)		(0.028)		(0.029)		(0.022)		(0.032)		(0.035)
Poor health							0.070	**	0.046	**	−0.031	*
							(0.011)		(0.015)		(0.015)
CESD score							0.004	*	0.009	**	−0.003
							(0.002)		(0.003)		(0.003)
#Health problems							0.008	*	0.011	*	−0.006
							(0.003)		(0.005)		(0.005)
Prob live to 75							−0.031	*	−0.005		−0.012
							(0.013)		(0.018)		(0.019)
Covered by fed. Govt HI							−0.016		−0.082	**	0.034
							(0.011)		(0.017)		(0.022)
Covered by priv. HI							−0.237	**	−0.035	*	0.124	**
							(0.016)		(0.016)		(0.015)
Covered by ER HI							−0.045	**	−0.001		0.039	**
							(0.010)		(0.011)		(0.010)
WBB cohort	−0.027	*	−0.032	*	−0.023		−0.018		−0.028		−0.028
	(0.011)		(0.015)		(0.016)		(0.012)		(0.016)		(0.016)
EBB cohort	0.017		−0.057	**	−0.006		0.001		−0.058	**	0.003
	(0.012)		(0.014)		(0.016)		(0.012)		(0.015)		(0.016)
Census region, northeast	0.007		0.002		−0.007		0.019		0.002		−0.015
	(0.014)		(0.018)		(0.019)		(0.015)		(0.019)		(0.020)
Census region, midwest	0.014		0.005		0.012		0.025		0.002		0.005
	(0.013)		(0.017)		(0.018)		(0.014)		(0.018)		(0.018)
Census region, south	0.057	**	−0.012		−0.005		0.047	**	−0.014		−0.001
	(0.012)		(0.016)		(0.017)		(0.013)		(0.016)		(0.017)
N					14,180						14,180
Pseudo R-sq					0.17						0.20
Dep. var. mean					2.66						2.66
Dep. var. st. dev.					1.12						1.12

Note:

^*p<0.05

^**p<0.01.

See Table 1.

Figure 9 reports the findings using Adjusted Full Income values at baseline for the pooled cohorts, where we see that the larger sample size helps smooth the volatility evident in earlier figures using a single cohort. Again, the cutoffs for each of the EWQ thresholds are the same as for the Original HRS cohort at baseline. In the pooled sample at baseline, fewer had zero Adjusted Full Income and more respondents had Adjusted Full Incomes above $150,000.

Figure 9. Adjusted Full Income for Three Cohorts of Respondents: Original HRS, WBB, and EBB at their Baselines (in $2019).

Note: Full income computed by summing Adjusted Money Income plus annuitized wealth. Quartiles defined by same Adjusted Full Income thresholds in HRS Baseline (see Fig 5). Data weighted.

Note: Quartiles measured for Original HRS Baseline cohort and applied to subsequent cohorts as well, all in $2019.

Figure 10 traces the Adjusted Full Income paths of the pooled sample by age, and interestingly, there is an upward trajectory in the lowest quartile’s Adjusted Full Income from baseline onward (red line). The two middle quartiles (blue and yellow lines) fared less well but ended up above where they started out, while the group that initially had the highest Adjusted Full Income (green line) entered its 80’s with 60% more Adjusted Full Income compared to its baseline.²⁵

Figure 10. Percentage Changes in Adjusted Full Income by Age and Initial Full Income Quartile for Three Cohorts: Original HRS, WBB, and EBB Followed Over Time (in $2019).

Note: Quartiles of Adjusted Full Income determined at baseline (EWQ1–4). The profiles represent estimated age effects from a regression of ln(Adj. Full Income) on age, demographic factors, and interview year. Quartile cutoffs appear in Figure 5. Data weighted.

In Table 7, we report marginal effects estimated using a multinomial Logit regression model of the factors associated with the chances of someone being in the lowest Adjusted Full Income Quartile at baseline, but now using the pooled dataset with all three cohorts; results may be compared with Table 3. As we saw when we focused on the Original HRS cohort alone, there was no statistically significant age effect associated with the chance of being in the bottom (Q1) Adjusted Full Income quartile at baseline. As in our earlier analysis, we see that men were least likely to be found in Q1, while the most vulnerable groups included Blacks, Hispanics, the least educated, nonmarried, persons with underage children, the disabled, those without health insurance, Southern residents, and the nonemployed.

Table 7.

Probability of Adjusted Full Income Being in EQW1 at Baseline (Logit Marginal Effects Reported): Original HRS cohort, EBB, and WBB (in $2019)

	Adj. Full Income in EWQ1 (0/1)
Age 52	0.003		0.004
	(0.012)		(0.012)
Age 53	0.005		0.003
	(0.012)		(0.012)
Age 54	−0.008		−0.005
	(0.012)		(0.012)
Age 55	0.018		0.016
	(0.013)		(0.013)
Age 56	0.005		0.011
	(0.014)		(0.014)
Age 57	0.007		0.014
	(0.016)		(0.016)
Age 58	0.012		0.018
	(0.017)		(0.017)
Age 59	−0.015		0.001
	(0.015)		(0.016)
Age 60	−0.015		0.002
	(0.015)		(0.016)
Age 61	0.005		0.019
	(0.019)		(0.020)
Male	−0.020	**	−0.035	**
	(0.006)		(0.006)
Black	0.121	**	0.094	**
	(0.012)		(0.013)
Race, others	0.056	**	0.038
	(0.021)		(0.020)
Hispanic	0.116	**	0.074	**
	(0.019)		(0.017)
Education years	−0.030	**	−0.021	**
	(0.001)		(0.001)
Married	−0.272	**	−0.218	**
	(0.011)		(0.012)
#Marriages	0.002		−0.007
	(0.005)		(0.005)
#Children<18 yr	0.052	**	0.050	**
	(0.005)		(0.006)
Working for pay	−0.181	**	−0.079	**
	(0.010)		(0.010)
Disabled	0.189	**	0.079	**
	(0.026)		(0.021)
Poor health			0.062	**
			(0.010)
CESD score			0.005	**
			(0.002)
#Health problems			0.008	**
			(0.003)
Prob live to 75			−0.037	**
			(0.011)
Covered by fed. Govt HI			−0.004
			(0.010)
Covered by priv. HI			−0.238	**
			(0.016)
Covered by ER HI			−0.019	*
			(0.009)
WBB cohort	−0.023	*	−0.014
	(0.010)		(0.011)
EBB cohort	0.007		−0.009
	(0.010)		(0.009)
Census region, northeast	−0.002		0.007
	(0.013)		(0.013)
Census region, midwest	−0.006		0.003
	(0.012)		(0.012)
Census region, south	0.054	**	0.044	**
	(0.012)		(0.011)
N	14,180		14,180
Pseudo R-sq	0.30		0.38
Dep. var. mean	0.21		0.21
Dep. var. st. dev.	0.41		0.41

Note:

^*p<0.05

^**p<0.01.

See Table 1.

Table 8 shows marginal effects of the same model used in Table 4, again with the pooled sample, but now the dependent variable is the probability of being in any of the lowest three Adjusted Full Income quartiles, versus being in the reference or highest Adjusted Full Income quartile (EWQ4) at baseline. Marginal effects from multinomial Logit models are reported in Table 8. Our earlier results in Table 4 are again confirmed, in that age was not usually statistically significant. Consistent with the earlier findings, men were least likely to be found in the lowest Adjusted Full Income quartile. Our previous results are also supported in that the chance of being in the lowest quartile was greatest for Blacks and Hispanics, Southerners, nonmarried persons, the least educated, people with health problems, those with underage children, the disabled, and those not working for pay. An important finding throughout our analysis is that the cohorts follow similar age/income life patterns, as only four of 12 cohort dummy variables were significant at the 5% level.

Table 8.

Probability of Adjusted Full Income Being in EQW1, 2, 3 versus (EQ4) (MLogit Marginal Effects Reported): HRS Original, WBB, EBB at Baseline

	Quartiles of Adj. Full Income
	EWQ1 vs. EWQ4		EWQ2 vs. EWQ4		EWQ3 vs. EWQ4		EWQ1 vs. EWQ4		EWQ2 vs. EWQ4		EWQ3 vs. EWQ4
Age 52	0.002		−0.009		−0.012		0.004		−0.010		−0.012
	(0.013)		(0.017)		(0.017)		(0.013)		(0.018)		(0.017)
Age 53	0.004		−0.005		−0.023		0.002		−0.003		−0.022
	(0.013)		(0.017)		(0.017)		(0.013)		(0.018)		(0.017)
Age 54	−0.007		0.015		−0.009		−0.004		0.012		−0.011
	(0.013)		(0.018)		(0.018)		(0.013)		(0.019)		(0.018)
Age 55	0.021		−0.004		−0.018		0.017		−0.009		−0.015
	(0.014)		(0.018)		(0.018)		(0.014)		(0.019)		(0.018)
Age 56	0.006		−0.001		−0.033		0.012		0.001		−0.038	*
	(0.015)		(0.019)		(0.019)		(0.016)		(0.020)		(0.019)
Age 57	0.007		−0.029		0.007		0.013		−0.032		0.002
	(0.017)		(0.022)		(0.023)		(0.017)		(0.023)		(0.023)
Age 58	0.014		0.001		−0.023		0.019		−0.005		−0.026
	(0.018)		(0.023)		(0.023)		(0.018)		(0.024)		(0.024)
Age 59	−0.012		0.031		−0.025		0.002		0.021		−0.033
	(0.016)		(0.024)		(0.023)		(0.017)		(0.025)		(0.023)
Age 60	−0.015		0.001		−0.020		0.003		−0.002		−0.030
	(0.016)		(0.023)		(0.023)		(0.017)		(0.024)		(0.023)
Age 61	0.011		0.031		−0.026		0.024		0.025		−0.036
	(0.020)		(0.027)		(0.026)		(0.021)		(0.028)		(0.026)
Male	−0.021	**	−0.020	**	0.021	**	−0.040	**	−0.017	*	0.022	**
	(0.006)		(0.008)		(0.007)		(0.006)		(0.009)		(0.008)
Black	0.139	**	0.054	**	−0.075	**	0.113	**	0.063	**	−0.070	**
	(0.013)		(0.014)		(0.014)		(0.013)		(0.015)		(0.015)
Race, others	0.069	**	0.038		−0.005		0.047	*	0.034		0.004
	(0.022)		(0.027)		(0.028)		(0.021)		(0.028)		(0.029)
Hispanic	0.145	**	0.040		−0.091	**	0.102	**	0.056	*	−0.078	**
	(0.020)		(0.022)		(0.020)		(0.019)		(0.023)		(0.022)
Education years	−0.038	**	−0.038	**	0.007	**	−0.029	**	−0.038	**	0.005	*
	(0.002)		(0.002)		(0.002)		(0.001)		(0.002)		(0.002)
Married	−0.292	**	−0.047	**	0.137	**	−0.235	**	−0.064	**	0.123	**
	(0.011)		(0.012)		(0.011)		(0.012)		(0.013)		(0.012)
#Marriages	0.001		−0.008		0.006		−0.010		−0.012		0.012
	(0.005)		(0.007)		(0.008)		(0.005)		(0.008)		(0.008)
#Children<18 yr	0.064	**	0.047	**	−0.019	*	0.062	**	0.051	**	−0.020	*
	(0.006)		(0.008)		(0.009)		(0.006)		(0.009)		(0.009)
Working for pay	−0.194	**	0.010		0.081	**	−0.087	**	−0.002		0.030	*
	(0.011)		(0.011)		(0.011)		(0.011)		(0.013)		(0.013)
Disabled	0.210	**	−0.005		−0.076	*	0.091	**	−0.001		−0.024
	(0.027)		(0.029)		(0.030)		(0.023)		(0.033)		(0.036)
Poor health							0.074	**	0.037	*	−0.034	*
							(0.011)		(0.015)		(0.015)
CESD score							0.007	**	0.011	**	−0.006	*
							(0.002)		(0.003)		(0.003)
#Health problems							0.010	**	0.016	**	−0.001
							(0.003)		(0.005)		(0.005)
Prob live to 75							−0.040	**	−0.014		−0.003
							(0.012)		(0.018)		(0.019)
Covered by fed. Govt HI							−0.009		−0.050	**	0.021
							(0.010)		(0.019)		(0.022)
Covered by priv. HI							−0.270	**	−0.009		0.143	**
							(0.017)		(0.016)		(0.015)
Covered by ER HI							−0.015		0.004		0.030	**
							(0.009)		(0.011)		(0.010)
WBB cohort	−0.027	*	−0.027		−0.024		−0.016		−0.022		−0.033	*
	(0.011)		(0.016)		(0.016)		(0.011)		(0.017)		(0.016)
EBB cohort	0.006		−0.035	*	−0.028		−0.010		−0.032	*	−0.021
	(0.011)		(0.015)		(0.015)		(0.010)		(0.016)		(0.016)
Census region, northeast	−0.002		0.048	*	−0.033		0.009		0.049	*	−0.042	*
	(0.013)		(0.020)		(0.019)		(0.014)		(0.021)		(0.019)
Census region, midwest	−0.006		0.059	**	0.005		0.005		0.056	**	−0.003
	(0.012)		(0.018)		(0.018)		(0.013)		(0.019)		(0.018)
Census region, south	0.060	**	0.028		−0.016		0.050	**	0.028		−0.014
	(0.012)		(0.017)		(0.017)		(0.012)		(0.017)		(0.017)
N					14,180						14,180
Pseudo R-sq					0.17						0.21
Dep. var. mean					2.66						2.66
Dep. var. st. dev.					1.12						1.12

Note:

^*p<0.05

^**p<0.01.

See Table 1.

Additional Considerations

As noted above, the reported income profiles traced for older persons over time may be influenced by differential mortality experiences across cohorts and demographic groups. For instance, if higher income persons had higher survival rates than lower income persons, survival bias could skew observed changes in financial conditions with age. Hungerford (2019) recently explored this question in the same dataset we use, the HRS, focusing on the Original HRS Baseline cohort, and he concluded that differential mortality by education, sex, and race did not account for rising income inequality as the population aged. In addition, above we noted that Michaud et al. (2011) and Meijer et al. (2010) found little evidence for bias due to selective attrition.²⁶ Nevertheless, we do see some evidence of higher disability and mortality rates for persons initially in the lowest income quartile, suggesting that future work should explore selective mortality as well as morbidity in the HRS more thoroughly.²⁷

Conclusions

A key finding of this analysis is that the real incomes of individuals in the lowest HRS baseline quartile remained relatively stable between 1992 and 2016, as the cohort aged from 51–61 to 74–85. In other words, it was not the case that most older persons lived on “fixed incomes,” notwithstanding frequent media reports to the contrary (e.g. LTLA 2018). One key reason is that Social Security benefits are indexed to inflation, so they are clearly not “fixed” in the economic sense. Overall, the relatively constant real incomes of the lowest quartile income group resulted from changing income sources as they aged, moving from labor earnings to Social Security, pensions, and personal savings. We also illustrated how these findings differed from other income quartiles, and we showed that people in the highest quartiles initially remained relatively well-off as they aged. Adding annuitized wealth strengthens these findings.

Another contribution of our paper is to identify the key factors associated with poor financial status in retirement by undertaking a granular analysis of three HRS cohorts followed over time. We point to several factors systematically associated with being in the lowest quartile of what Adjusted Money Income distribution in their early 50’s, including age, in that people in age 51–56 typically had more income than did their older counterparts in their late 50’s and early 60’s. This was mainly because the latter group was more likely to have left paid employment. Multivariate analysis confirmed that Blacks, Hispanics, people with underage children, those in poor health, and people lacking private/employer-provided health insurance, were also more likely to be in the lowest Money Income quartile when observed at baseline. Another group systematically at risk was respondents living in the South of the US. The better-educated and those who continued to work for pay at older ages were most likely to be in higher Adjusted Money Income quartiles.

We also traced how peoples’ financial fortunes changed with age and identified the factors associated with these trajectories. Interestingly, we found that individuals whose baseline Adjusted Money Incomes were initially the lowest tended to receive Social Security benefits well before the full retirement age (most likely due to disability and child support due to widow(er)hood). People initially located in the top two income quartiles experienced a downward trend in real money incomes as they moved through retirement.

When we incorporated wealth into computations of peoples’ financial resources at older ages, this measure tracked upward from age 62, on average, providing evidence of a more positive trajectory than one might glean from looking only at Adjusted Money Income. In other words, including the annuitized share of wealth when measuring peoples’ access to resources made many elderly respondents’ financial conditions appear substantially better, particularly for those initially in the higher income quartiles. As a result, making it easier for older people to access their net home equity and other assets could enhance their financial positions in old age, though it is not necessarily easy to do so, and reverse mortgages have not been very popular to date (Mayer and Moulton forthcoming). Moreover, helping the poorest access locked up assets would do little to enhance their position as they have few assets. Even using the Adjusted Full Income measure, we again confirmed that Blacks and Hispanics, the least educated, disabled, and nonmarried persons, fared relatively worse in later life, as well as those having underage children. Health problems also played an important role; enhancing lower paid workers’ health conditions and health insurance could improve their retirement wellbeing. Additionally, since many older persons in fragile economic circumstances were also likely to be caring for underage children, identifying ways to ease this burden could substantially enhance wellbeing for many in later life. Finally, we showed that residents of the Southern US states were consistently more likely to be in the lowest income quartile on the verge of retirement, and they continued to be in this group throughout retirement, even after controlling on many other socioeconomic factors. While the HRS data do not permit a fuller examination of this last result, other analysts (e.g., Henderson 2019) have pointed to substandard educational levels, a dearth of job training and skills, and the paucity of well-paying jobs in Southern states as potential explanations for this finding.

Online Appendix for Income Trajectories in Later Life: Longitudinal Evidence from the Health and Retirement Study

Appendix Table 1.

Baseline Descriptive Statistics: Original HRS Cohort ($2019)

A. Using Adjusted Money Income
	N	Mean	Std. Dev.	Min	Median	Max
Adj. Money Income Q1 Bline	9,608	0.22	0.41	0	0	1
Quartiles of Adj. Money Income	9,608	2.59	1.11	1	3	4
Age (yr)	9,608	55.56	3.20	50	55	61
Male	9,608	0.47	0.50	0	0	1
Black	9,608	0.10	0.31	0	0	1
Race, others	9,608	0.04	0.18	0	0	1
Hispanic	9,608	0.07	0.25	0	0	1
Education years (yr)	9,608	12.29	3.04	0	12	17
Currently married	9,608	0.76	0.42	0	1	1
#Marriage	9,608	1.30	0.71	0	1	13
#Children≤18 yr	9,608	0.22	0.61	0	0	15
Working for pay	9,608	0.69	0.46	0	1	1
Disabled	9,608	0.04	0.19	0	0	1
Poor health	9,608	0.20	0.40	0	0	1
CESD score	9,608	2.17	1.96	0	2	8
#Health problem	9,608	1.25	1.20	0	1	7
Prob live to 75	9,608	0.64	0.28	0	0.70	1
Covered by fed. govt. HI	9,608	0.12	0.32	0	0	1
Covered by priv. HI	9,608	0.78	0.41	0	1	1
Covered by ER HI	9,608	0.48	0.50	0	0	1
Census region, northeast	9,608	0.22	0.41	0	0	1
Census region, midwest	9,608	0.24	0.43	0	0	1
Census region, south	9,608	0.35	0.48	0	0	1
Census region, west	9,608	0.19	0.39	0	0	1
B. Using Adjusted Full Income
	N	Mean	Std. Dev.	Min	Median	Max
Adj.Full Income Q1 Bline	9,608	0.21	0.41	0	0	1
Quartiles of Adj. Full Income	9,608	2.61	1.11	1	3	4
Age (yr)	9,608	55.56	3.20	50	55	61
Male	9,608	0.47	0.50	0	0	1
Black	9,608	0.10	0.31	0	0	1
Race, others	9,608	0.04	0.18	0	0	1
Hispanic	9,608	0.07	0.25	0	0	1
Education years (yr)	9,608	12.29	3.04	0	12	17
Currently married	9,608	0.76	0.42	0	1	1
#Marriage	9,608	1.30	0.71	0	1	13
#Children≤18 yr	9,608	0.22	0.61	0	0	15
Working for pay	9,608	0.69	0.46	0	1	1
Disabled	9,608	0.04	0.19	0	0	1
Poor health	9,608	0.20	0.40	0	0	1
CESD score	9,608	2.17	1.96	0	2	8
#Health problem	9,608	1.25	1.20	0	1	7
Prob live to 75	9,608	0.64	0.28	0	0.70	1
Covered by fed. govt. HI	9,608	0.12	0.32	0	0	1
Covered by priv. HI	9,608	0.78	0.41	0	1	1
Covered by ER HI	9,608	0.48	0.50	0	0	1
Census region, northeast	9,608	0.22	0.41	0	0	1
Census region, midwest	9,608	0.24	0.43	0	0	1
Census region, south	9,608	0.35	0.48	0	0	1
Census region, west	9,608	0.19	0.39	0	0	1

Note: Analysis sample includes all those with Adjusted Total Money Income; see text. Data weighted.

Note: Analysis sample includes all those having Adjusted Money Income, and Adjusted Full Income computed by summing household adjusted income plus annuitized wealth, all in $2019. Data weighted.

Appendix Table 2.

Sample Retention by Wave and Source of Attrition

A. Original HRS Cohort
Longitudinal step	1	2	3	4	5	6	7	8	9	10	11	12	13
Core interview obtained	9,608	8,662	8,196	7,771	7,312	7,001	6,686	6,312	5,992	5,527	5,137	4,586	3,936
Died this wave		166	208	210	266	332	232	308	308	478	342	416	484
Died previous wave		0	166	374	584	850	1,174	1,404	1,710	2,018	2,496	2,838	3,254
Dropout (Non-interview)		780	1,038	1,253	1,446	1,425	1,516	1,584	1,598	1,585	1,633	1,768	1,934
Total	9,608	9,608	9,608	9,608	9,608	9,608	9,608	9,608	9,608	9,608	9,608	9,608	9,608
B. Original HRS, WBB, and EBB Cohorts
Longitudinal step	1	2	3	4	5	6	7	8	9	10	11	12	13
Core interview obtained	14,180	12,771	12,170	11,597	10,968	10,511	9,997	7,724	7,307	6,708	5,137	4,586	3,936
Died this wave		212	271	287	345	429	316	351	360	552	342	416	484
Died previous wave		0	212	483	770	1,115	1,536	1,595	1,944	2,304	2,496	2,838	3,254
Dropout (Non-interview)		1,197	1,527	1,813	2,097	2,125	2,331	1,870	1,929	1,976	1,633	1,768	1,934
Total	14,180	14,180	14,180	14,180	14,180	14,180	14,180	11,540	11,540	11,540	9,608	9,608	9,608

Source: Authors’ computations using HRS interview status variable R*IWSTAT. The Original HRS cohort entered in 1992 and followed for 13 waves (26 years); the WBB entered in 1998 and followed for 10 waves (or 20 years); and the EBB entered in 2004 and followed for 7 waves (14 years).

Appendix Table 3.

Cumulative Attrition Percentages by Wave and Baseline Adjusted Money Income Quartile

A. Original HRS Cohort
	Wave
Adj. Money Income	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	0%	13%	20%	24%	29%	33%	36%	42%	45%	51%	55%	61%	68%
Q2	0%	11%	15%	19%	25%	28%	32%	35%	39%	44%	48%	54%	61%
Q3	0%	7%	13%	19%	23%	26%	28%	32%	35%	39%	43%	49%	57%
Q4	0%	7%	12%	15%	18%	22%	26%	29%	32%	36%	40%	45%	51%
B. Original HRS, WBB, and EBB Cohorts
	Wave
Adj. Money Income	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	0%	13%	19%	22%	28%	32%	36%	40%	44%	50%	55%	61%	68%
Q2	0%	10%	15%	19%	24%	27%	30%	34%	38%	43%	48%	54%	61%
Q3	0%	8%	12%	17%	21%	24%	27%	31%	35%	39%	43%	49%	57%
Q4	0%	8%	12%	15%	19%	22%	26%	28%	31%	35%	40%	45%	51%

Source: Authors’ computations using HRS interview status variable R*IWSTAT. See also Notes to Appendix Table 2.

Appendix Table 4.

Median Adjusted Money Income by Wave and Baseline Quartile ($2019)

A. Original HRS Cohort
	Baseline quartiles
Wave	Q1	Q2	Q3	Q4
1	$11,411	$30,770	$53,504	$94,050
2	$13,829	$32,444	$53,823	$84,892
3	$15,303	$33,220	$53,037	$83,329
4	$15,228	$32,846	$49,291	$81,395
5	$15,070	$31,198	$48,117	$73,506
6	$15,540	$28,718	$43,426	$62,459
7	$14,985	$28,134	$40,188	$62,049
8	$14,408	$26,919	$37,248	$58,731
9	$14,506	$25,635	$36,966	$54,659
10	$14,700	$23,543	$33,905	$49,414
11	$14,148	$22,460	$30,568	$45,890
12	$14,391	$22,018	$31,510	$46,471
13	$12,982	$20,296	$28,553	$42,698
B. Original HRS, WBB, and EBB Cohorts
	Baseline quartiles
Wave	Q1	Q2	Q3	Q4
1	$11,078	$31,067	$53,974	$96,720
2	$13,777	$33,568	$54,617	$88,452
3	$15,214	$33,777	$54,641	$87,591
4	$15,143	$32,812	$51,094	$82,707
5	$14,954	$31,715	$49,518	$76,877
6	$14,642	$29,340	$45,204	$67,889
7	$14,892	$28,539	$41,113	$65,462
8	$14,469	$26,984	$38,353	$58,749
9	$14,421	$26,225	$37,612	$55,729
10	$14,531	$24,145	$34,336	$50,196
11	$14,148	$22,460	$30,568	$45,890
12	$14,391	$22,018	$31,510	$46,471
13	$12,982	$20,296	$28,553	$42,698

See also notes to Appendix Table 2.

Appendix Table 5.

Percent Self-reporting as Disabled by Wave and Baseline Quartile

A. Original HRS Cohort
	Wave
Adj. Money Income	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	12.0%	10.8%	14.7%	14.8%	13.1%	9.9%	6.0%	4.3%	3.8%	2.6%	2.0%	1.6%	1.5%
Q2	2.3%	3.1%	3.0%	4.2%	3.4%	2.9%	1.2%	1.0%	0.5%	0.5%	0.5%	0.4%	0.5%
Q3	1.1%	1.3%	1.5%	2.0%	2.6%	1.7%	1.3%	0.9%	0.4%	0.5%	0.0%	0.3%	0.1%
Q4	0.7%	0.9%	1.1%	0.8%	0.9%	0.7%	0.6%	0.5%	0.3%	0.2%	0.1%	0.3%	0.0%
Total	3.7%	3.6%	4.5%	4.8%	4.4%	3.3%	2.0%	1.5%	1.1%	0.8%	0.5%	0.6%	0.4%
B. Original HRS, WBB, and EBB Cohorts
	Wave
Adj. Money Income	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	13.5%	12.9%	15.2%	14.4%	12.6%	9.4%	6.1%	4.4%	4.0%	2.9%	2.0%	1.6%	1.5%
Q2	2.6%	3.6%	3.3%	4.1%	3.5%	3.1%	1.2%	1.1%	0.5%	0.7%	0.5%	0.4%	0.5%
Q3	1.4%	1.6%	1.9%	2.1%	2.4%	1.8%	1.3%	0.8%	0.4%	0.4%	0.0%	0.3%	0.1%
Q4	0.9%	0.9%	1.2%	0.9%	0.8%	0.9%	0.4%	0.5%	0.3%	0.2%	0.1%	0.3%	0.0%
Total	4.1%	4.1%	4.6%	4.6%	4.1%	3.2%	1.9%	1.4%	1.0%	0.9%	0.5%	0.6%	0.4%

Appendix Table 6.

Components of Adjusted Money Income by Wave and Baseline Quartile (%)

A. Original HRS Cohort
Social Security
	Wave
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	22.2	28.5	37.5	45.4	48.0	59.1	65.3	69.9	73.4	75.5	77.1	77.3	79.5
Q2	5.7	8.6	15.7	24.0	30.6	42.3	49.0	54.4	58.5	64.4	65.4	65.8	69.2
Q3	1.8	4.6	10.0	16.6	22.3	31.6	39.6	45.2	48.5	54.7	57.4	57.0	59.6
Q4	0.6	2.3	5.7	10.6	15.7	25.3	30.7	35.8	38.0	43.8	45.0	46.6	47.7
Earnings
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	43.0	36.9	29.7	26.3	24.5	18.9	14.0	11.1	8.0	5.3	4.4	3.7	2.5
Q2	73.1	63.0	53.3	45.0	38.1	29.6	21.7	16.0	12.8	9.5	7.6	5.9	4.8
Q3	81.7	68.3	55.9	46.9	37.5	29.9	23.2	16.4	13.1	9.3	7.4	6.1	5.0
Q4	77.1	66.0	51.6	44.0	34.9	28.0	22.3	16.8	13.3	10.6	9.0	6.9	5.3
Unemployment and worker compensation benefits
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	3.0	1.9	1.6	1.1	1.1	0.3	0.5	0.2	0.1	0.2	0.1	0.0	0.0
Q2	2.0	2.0	1.3	1.2	0.7	0.7	0.3	0.3	0.1	0.3	0.1	0.1	0.1
Q3	0.9	1.7	1.0	1.1	0.6	0.3	0.3	0.2	0.2	0.2	0.1	0.1	0.1
Q4	0.3	1.0	0.6	0.6	0.5	0.3	0.3	0.1	0.1	0.2	0.1	0.1	0.1
Pensions or annuities
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	6.4	7.4	7.1	6.9	7.7	7.5	7.6	7.9	7.9	6.7	7.8	8.7	7.8
Q2	8.1	10.4	10.6	11.9	13.7	13.2	15.0	15.6	15.2	14.1	15.4	15.6	15.4
Q3	5.4	10.2	11.5	13.9	17.6	19.3	20.1	21.1	21.9	19.9	21.4	21.6	21.3
Q4	3.7	9.8	12.1	14.9	18.6	20.5	22.3	23.5	22.7	22.9	25.8	26.1	26.0
Capital income
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	8.1	10.2	12.2	10.6	10.4	8.3	6.6	6.0	5.5	5.6	4.7	4.4	4.4
Q2	7.4	11.2	14.3	13.5	12.2	10.7	10.4	10.4	10.0	7.6	7.4	8.3	6.8
Q3	6.9	11.6	17.5	16.6	17.0	13.9	12.1	12.5	12.0	11.2	9.4	10.3	9.2
Q4	12.2	16.5	24.6	24.7	25.0	21.0	18.5	18.6	20.6	17.1	14.9	15.5	15.7
B. Original HRS Cohort, WBB, and EBB Cohorts
Social Security
	Wave
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	22.8	28.0	35.2	42.0	45.7	56.8	63.2	69.2	72.9	75.1	77.1	77.3	79.5
Q2	5.5	7.8	13.7	20.8	27.2	37.3	46.0	53.4	58.2	63.6	65.4	65.8	69.2
Q3	1.6	3.8	8.2	13.4	19.2	28.1	37.9	44.5	48.1	54.2	57.4	57.0	59.6
Q4	0.6	1.8	4.1	7.7	12.3	20.1	27.8	35.3	37.8	43.8	45.0	46.6	47.7
Earnings
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	42.5	37.5	32.4	29.2	26.8	20.1	15.8	11.8	8.3	5.6	4.4	3.7	2.5
Q2	73.8	65.0	57.5	49.2	43.0	34.6	25.6	17.5	13.7	10.1	7.6	5.9	4.8
Q3	81.5	70.4	60.8	52.3	43.5	35.4	27.7	17.6	14.4	10.1	7.4	6.1	5.0
Q4	76.4	67.7	58.4	51.4	43.0	35.9	28.4	19.4	15.1	11.5	9.0	6.9	5.3
Unemployment and worker compensation benefits
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	2.8	1.8	1.4	1.3	1.1	0.6	0.5	0.3	0.1	0.2	0.1	0.0	0.0
Q2	2.0	1.9	1.3	1.6	0.9	0.7	0.6	0.3	0.1	0.3	0.1	0.1	0.1
Q3	0.8	1.4	0.9	1.3	0.8	0.4	0.4	0.2	0.2	0.1	0.1	0.1	0.1
Q4	0.3	0.9	0.5	0.7	0.6	0.3	0.4	0.2	0.1	0.1	0.1	0.1	0.1
Pensions or annuities
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	5.4	6.8	6.2	6.2	7.1	6.9	7.0	7.9	8.0	7.0	7.8	8.7	7.8
Q2	6.7	8.6	8.9	10.3	11.9	12.0	13.6	15.2	14.6	14.1	15.4	15.6	15.4
Q3	4.5	8.4	9.8	11.8	15.5	17.4	17.9	20.9	21.6	20.1	21.4	21.6	21.3
Q4	2.8	7.3	9.7	12.3	15.4	18.2	19.2	22.7	22.5	23.0	25.8	26.1	26.0
Capital income
Baseline Quartile	1	2	3	4	5	6	7	8	9	10	11	12	13
Q1	9.2	10.9	12.2	10.2	9.7	8.2	6.5	5.7	5.4	5.5	4.7	4.4	4.4
Q2	8.6	11.6	13.6	13.1	11.8	10.7	9.9	9.9	9.6	7.6	7.4	8.3	6.8
Q3	8.5	12.1	16.1	15.9	15.5	13.2	11.3	12.1	11.5	10.7	9.4	10.3	9.2
Q4	14.5	18.2	22.7	23.0	23.3	20.6	18.5	17.5	19.1	16.1	14.9	15.5	15.7

Appendix Table 7.

Adjusted Money Income (AMI) and Adjusted Household Wealth by Wave and Baseline Quartile ($2019)

A. Original HRS Cohort
	Q1		Q2		Q3		Q4
Wave	Median AMI	Median Adj. HH wealth	Median AMI	Median Adj. HH wealth	Median AMI	Median Adj. HH wealth	Median AMI	Median Adj. HH wealth
1	$11,411	$22,168	$30,770	$81,527	$53,504	$137,093	$94,050	$271,331
2	$13,829	$28,982	$32,444	$92,433	$53,823	$168,690	$84,892	$320,597
3	$15,303	$29,382	$33,220	$92,557	$53,037	$175,838	$83,329	$349,954
4	$15,228	$29,974	$32,846	$94,406	$49,291	$180,107	$81,395	$393,551
5	$15,070	$34,012	$31,198	$105,080	$48,117	$205,641	$73,506	$434,305
6	$15,540	$35,253	$28,718	$112,860	$43,426	$207,009	$62,459	$460,878
7	$14,985	$38,409	$28,134	$113,310	$40,188	$228,371	$62,049	$491,616
8	$14,408	$37,238	$26,919	$127,969	$37,248	$251,607	$58,731	$537,917
9	$14,506	$33,213	$25,635	$120,780	$36,966	$247,155	$54,659	$521,916
10	$14,700	$40,530	$23,543	$107,551	$33,905	$203,519	$49,414	$440,132
11	$14,148	$31,524	$22,460	$97,010	$30,568	$196,200	$45,890	$410,577
12	$14,391	$32,230	$22,018	$101,187	$31,510	$197,160	$46,471	$427,983
13	$12,982	$23,460	$20,296	$96,647	$28,553	$189,688	$42,698	$411,112
B. Original HRS, WBB and EBB Cohorts
	Q1		Q2		Q3		Q4
Wave	Median AMI	Median Adj. HH wealth	Median AMI	Median Adj. HH wealth	Median AMI	Median Adj. HH wealth	Median AMI	Median Adj. HH wealth
1	$11,078	$19,198	$31,067	$75,240	$53,974	$131,734	$96,720	$266,852
2	$13,777	$25,176	$33,568	$83,311	$54,617	$159,325	$88,452	$308,921
3	$15,214	$26,347	$33,777	$89,322	$54,641	$165,197	$87,591	$339,029
4	$15,143	$26,533	$32,812	$89,738	$51,094	$169,244	$82,707	$376,343
5	$14,954	$29,430	$31,715	$94,187	$49,518	$187,327	$76,877	$410,758
6	$14,642	$28,482	$29,340	$101,562	$45,204	$193,589	$67,889	$439,601
7	$14,892	$32,265	$28,539	$104,814	$41,113	$207,147	$65,462	$454,387
8	$14,469	$34,143	$26,984	$119,761	$38,353	$242,467	$58,749	$491,669
9	$14,421	$31,315	$26,225	$116,278	$37,612	$237,234	$55,729	$498,192
10	$14,531	$37,229	$24,145	$107,640	$34,336	$202,650	$50,196	$424,742
11	$14,148	$31,524	$22,460	$97,010	$30,568	$196,200	$45,890	$410,577
12	$14,391	$32,230	$22,018	$101,187	$31,510	$197,160	$46,471	$427,983
13	$12,982	$23,460	$20,296	$96,647	$28,553	$189,688	$42,698	$411,112

Appendix Table 8.

Baseline Descriptive Statistics: Original HRS, WBB, and EBB Cohorts

A. Using Adjusted Money Income
	N	Mean	Std. Dev.	Min	Median	Max
Adj. Money Income Q1 Bline	14,180	0.209	0.407	0	0	1
Quartiles of Adj. Money Income	14,180	2.662	1.122	1	3	4
Age (yr)	14,180	54.76	3.09	50	54	61
Male	14,180	0.50	0.50	0	0	1
Black	14,180	0.11	0.31	0	0	1
Race, others	14,180	0.05	0.21	0	0	1
Hispanic	14,180	0.07	0.26	0	0	1
Education years (yr)	14,180	12.61	3.02	0	12	17
Currently married	14,180	0.75	0.43	0	1	1
#Marriage	14,180	1.31	0.72	0	1	13
#Children≤18 yr	14,180	0.26	0.65	0	0	15
Working for pay	14,180	0.72	0.45	0	1	1
Disabled	14,180	0.04	0.20	0	0	1
Poor health	14,180	0.21	0.41	0	0	1
CESD score	14,180	1.97	1.98	0	1.00	8
#Health problem	14,180	1.22	1.20	0	1	8
Prob live to 75	14,180	0.64	0.28	0	0.7	1
Covered by fed. govt. HI	14,180	0.12	0.32	0	0	1
Covered by priv. HI	14,180	0.79	0.41	0	1	1
Covered by ER HI	14,180	0.50	0.50	0	1	1
Census region, northeast	14,180	0.20	0.40	0	0	1
Census region, midwest	14,180	0.25	0.43	0	0	1
Census region, south	14,180	0.36	0.48	0	0	1
Census region, west	14,180	0.19	0.39	0	0	1
HRS	14,180	0.70	0.46	0	1	1
WBB	14,180	0.15	0.36	0	0	1
EBB	14,180	0.15	0.36	0	0	1
B. Using Adjusted Full Income
	N	Mean	Std. Dev.	Min	Median	Max
Adj. Full Income Q1 Bline	14,180	0.207	0.405	0	0	1
Quartiles of Adj. Full Income	14,180	2.662	1.119	1	3	4
Age (yr)	14,180	54.76	3.09	50	54	61
Male	14,180	0.50	0.50	0	0	1
Black	14,180	0.11	0.31	0	0	1
Race, others	14,180	0.05	0.21	0	0	1
Hispanic	14,180	0.07	0.26	0	0	1
Education years (yr)	14,180	12.61	3.02	0	12	17
Currently married	14,180	0.75	0.43	0	1	1
#Marriage	14,180	1.31	0.72	0	1	13
#Children≤18 yr	14,180	0.26	0.65	0	0	15
Working for pay	14,180	0.72	0.45	0	1	1
Disabled	14,180	0.04	0.20	0	0	1
Poor health	14,180	0.21	0.41	0	0	1
CESD score	14,180	1.97	1.98	0	1.00	8
#Health problem	14,180	1.22	1.20	0	1	8
Prob live to 75	14,180	0.64	0.28	0	0.7	1
Covered by fed. govt. HI	14,180	0.12	0.32	0	0	1
Covered by priv. HI	14,180	0.79	0.41	0	1	1
Covered by ER HI	14,180	0.50	0.50	0	1	1
Census region, northeast	14,180	0.20	0.40	0	0	1
Census region, midwest	14,180	0.25	0.43	0	0	1
Census region, south	14,180	0.36	0.48	0	0	1
Census region, west	14,180	0.19	0.39	0	0	1
HRS	14,180	0.70	0.46	0	1	1
WBB	14,180	0.15	0.36	0	0	1
EBB	14,180	0.15	0.36	0	0	1

Appendix Table 9.

Probit Models Comparing HRS Survivors to Final Wave with Nonsurvivors (Marginal Effects Reported)

	Original HRS		Original HRS, WBB, & EBB
Age 52	−0.033		−0.033
	(0.023)		(0.018)
Age 53	−0.041		−0.044	*
	(0.023)		(0.018)
Age 54	−0.033		−0.030
	(0.023)		(0.018)
Age 55	−0.030		−0.026
	(0.023)		(0.018)
Age 56	−0.037		−0.037
	(0.024)		(0.020)
Age 57	−0.040		−0.043
	(0.024)		(0.023)
Age 58	−0.091	**	−0.095	**
	(0.023)		(0.023)
Age 59	−0.125	**	−0.132	**
	(0.023)		(0.023)
Age 60	−0.124	**	−0.131	**
	(0.023)		(0.022)
Age 61	−0.145	**	−0.154	**
	(0.025)		(0.025)
Male	−0.113	**	−0.103	**
	(0.012)		(0.010)
Black	0.014		0.021
	(0.016)		(0.013)
Race, others	−0.030		0.003
	(0.031)		(0.023)
Hispanic	0.120	**	0.103	**
	(0.023)		(0.019)
Education years	0.008	**	0.007	**
	(0.002)		(0.002)
Married	0.045	**	0.036	**
	(0.016)		(0.013)
#Marriages	−0.027	**	−0.020	**
	(0.009)		(0.007)
#Children<18 yr	0.021	*	0.016	*
	(0.009)		(0.008)
Working for pay	0.048	**	0.057	**
	(0.015)		(0.013)
Disabled	−0.052		−0.034
	(0.033)		(0.026)
AMI ($100K)	0.009		0.005
Poor health	−0.112	**	−0.107	**
	(0.016)		(0.014)
CESD score	0.010	**	0.009	**
	(0.003)		(0.003)
#Health problems	−0.037	**	−0.032	**
	(0.006)		(0.005)
Prob live to 75	0.088	**	0.087	**
	(0.021)		(0.018)
Covered by fed. Govt HI	0.020		0.024
	(0.020)		(0.017)
Covered by priv. HI	0.041	*	0.038	*
	(0.018)		(0.016)
Covered by ER HI	0.012		0.007
	(0.015)		(0.013)
AMI, Q1	−0.046		−0.038
	(0.026)		(0.021)
AMI, Q2	−0.054	**	−0.039	*
	(0.021)		(0.017)
AMI, Q3	−0.033		−0.017
	(0.018)		(0.015)
WBB cohort			0.161	**
			(0.014)
EBB cohort			0.260	**
			(0.013)
Census region, NE	−0.006		−0.012
	(0.019)		(0.016)
Census region, Mwest	0.016		0.013
	(0.018)		(0.015)
Census region, S	−0.027		−0.031	*
	(0.016)		(0.014)
N	9,608		14,180
Pseudo R-sq	0.06		0.09
Dep. var. mean	0.42		0.49
Dep. var. st. dev.	0.49		0.50

Note:

^*p<0.05

^**p<0.01

Robust standard errors

Reference levels: age 50/51, white, west census region, AMI Q4