Income in Retirement

 

At a Glance

Veterans who served during the Vietnam War—including those who were deployed to combat zones and those who served elsewhere—constitute the last cohort of service members that was subject to a draft. More than 6 million of the nearly 9 million people who served on active duty during the war are still living.

The Congress and those Veterans themselves have expressed concern about the lifelong effects of that military service, but little is known about their financial security now that most have left the labor force. The Congressional Budget Office compared the income of male Vietnam Veterans with the income of male nonVeterans the same ages. (Very few Vietnam Veterans were women.) CBO found that:

  • On average, Vietnam Veterans in 2018 had roughly the same income as nonVeterans their ages: $63,300 and $65,000, respectively. For Veterans and nonVeterans age 71—the modal, or most common, age of Veterans­—average income was also about the same.
  • About 1.3 million Vietnam Veterans, nearly 25 percent, collected disability compensation from Department of Veterans Affairs (VA) in 2018; their average annual payment was $18,100. Those payments boosted average income for all Vietnam Veterans by $4,300. When those payments are excluded, Veterans’ income averaged $59,000, 9 percent less than nonVeterans’ income.
  • The gap between the average income of Veterans and nonVeterans was largest for men in their mid-60s. It was smaller for older men: Veterans older than 71 had, on average, more income than nonVeterans of the same age. That was true whether or not disability payments from VA were included in income.
  • In general, Vietnam Veterans received more money from Social Security and retirement plans than nonVeterans; nonVeterans had more earnings and more investment income. Those differences probably arose from differences in the types of employers and jobs that members of each group had over their working lives.
  • Income was distributed more evenly among Vietnam Veterans than among nonVeterans. In other words, the percentages of Vietnam Veterans in the highest and lowest income categories were smaller, and the percentages in the middle categories were larger, than those for nonVeterans.

Summary

In this report, the Congressional Budget Office looks at the income of male Vietnam Veterans now that most of them have reached retirement age.1 (Very few Vietnam Veterans were women.) More than 6 million of those men—some of whom were drafted and others who volunteered—are still alive. Previous research showed that Veterans earned less for a decade after the war ended because of their military service but caught up to nonVeterans by the early 1990s.2 But since then, as Veterans have left the labor force, their sources of income have changed, including compensation and benefits from the Department of Veterans Affairs (VA), and little is known about Veterans’ current financial status. Lawmakers and others have expressed concern about their well-being.

What Did CBO Find?

In 2018, the average income of Vietnam Veterans and nonVeterans was roughly comparable: For Veterans, who were 63 to 78 years old at that point, it was $63,300, slightly less than the $65,000 average for nonVeterans. The Veterans’ average includes the disability compensation that some receive from VA. With that disability compensation excluded, Veterans’ average income was $59,000, 9 percent less than nonVeterans’ average income.

The income gap between Veterans and nonVeterans was largest for men in their mid-60s; on average, Vietnam Veterans who were age 72 or older in 2018 had more income than nonVeterans, whether or not VA’s disability compensation was included. For Veterans and nonVeterans age 71—the modal, or most common, age of Veterans—there was little or no gap in average income (see Figure 1).

Source: Congressional Budget Office, using data from the American Community Survey and the Department of Veterans Affairs.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

The Department of Veterans Affairs provides tax-free disability compensation to Veterans with medical injuries or conditions that were incurred or aggravated during active-duty military service. Other income sources may be subject to taxation.

 

VA = Department of Veterans Affairs.

Veterans age 72 or older had more income than younger Veterans because those older Veterans probably earned more during their working years. Older Veterans generally had spent more years in the military, had higher levels of education, and probably differed in other ways that are harder to measure—such as skills learned in the military or sense of purpose. Those characteristics could affect their income in retirement.

In general (and at age 71), Veterans received more income from Social Security and retirement plans than nonVeterans, and nonVeterans had higher earnings and more income from investments. The differences probably arose from the types of jobs Veterans and nonVeterans held.

With VA’s disability compensation excluded, income was distributed more equally among Vietnam Veterans than among nonVeterans. In comparison with nonVeterans, a smaller share of Veterans’ income was in either the lowest or highest quintile (fifth) of income, and a larger share was in the middle three quintiles, for all men ages 63 to 78. For Veterans who received VA’s disability payments in addition to their other income, the average annual payment was $18,100. Those disability payments made their income higher than other Veterans’ income, on average.

Whose Income Did CBO Examine?

CBO looked at male Veterans who served on active duty during the Vietnam War and were between the ages of 63 and 78 in 2018, about 5.4 million Veterans. That group included most of the Veterans who served during the peak years of the war. Only those members of the National Guard and reserves who were activated during the war—roughly 25,000 men—were considered Vietnam Veterans. CBO excluded women from its analysis because they were a very small share of Vietnam Veterans in 2018 (4 percent). Veterans who did not serve during the Vietnam War were excluded from the samples that CBO analyzed.

Of the nearly 9 million people who served, 3.4 million were deployed to Vietnam or to other countries in Southeast Asia where the war was waged. (The rest were located in the United States or on overseas bases outside the war zone.) Less than one-quarter of those who served were drafted, but a disproportionate share of draftees served in Southeast Asia. In many aspects, the makeup of the military reflected the young male population in the United States at that time: Most service members were White and had a high school education.

How Did CBO Analyze Income?

CBO used data from the Census Bureau’s 2018 American Community Survey (ACS) to estimate income from different sources for Vietnam Veterans and nonVeterans ages 63 to 78. The ACS is one of the biggest surveys that the Census Bureau administers, reaching roughly 1 in 40 U.S. households each year. CBO relied on the ACS because it samples a large number of Veterans, and its data on income information are about as accurate as data from other national surveys that also report Veteran status.

CBO examined the selected group of Veterans and nonVeterans at two points in time: 2008 and 2018. In 2008, most were still working but near the end of their careers. CBO calculated earnings in that year because earnings are closely linked to retirement income. At the second point, 2018, most of the group were no longer working. CBO did not quantify the potential effects of different factors such as education and work experience that strongly influence the amount and sources of income.

At the second point in time, 2018, CBO focused on four sources of income common to Veterans and nonVeterans—earnings, Social Security, investments, and retirement plans. For Veterans, the agency also calculated income with and without disability compensation paid by VA. In 2018, about 1.3 million Vietnam Veterans ages 63 to 78 received that compensation because of medical conditions or injuries incurred during their military service. CBO’s estimates of VA’s disability payments relied on data from both the ACS and VA.

What Are the Limitations of the Report?

CBO’s study has limitations that are common to any analysis based on survey data. Some income is reported incorrectly, and CBO used statistical methods to improve the accuracy of results. In addition, the report is an incomplete picture of Veterans’ overall finances because CBO did not examine all types of income. But in describing Veterans’ regular sources of income in retirement, it contributes to a greater understanding of their financial security and can inform Congressional decisions about support for Veterans.

The Vietnam War and the U.S. Military

Although America was involved in Vietnam for more than 10 years, the conflict itself is often considered to run from August 1964 to January 1973.3 On August 7, 1964, the Congress passed the Gulf of Tonkin Resolution, which allowed the President “to take all necessary steps, including the use of armed force” to prevent further attacks against the United States. On January 27, 1973, the Paris Peace Accords were signed, including a cease-fire and a timetable for the withdrawal of U.S. troops. Within 90 days of that agreement, U.S. ground troops had withdrawn from Vietnam. South Vietnam surrendered to North Vietnam in April 1975, and the last Americans were evacuated.

More than 8.5 million men, or about one-third of those eligible for military service, served in the U.S. armed forces during the Vietnam War.4 The first ground troops were sent to Vietnam in March 1965; their number peaked in 1968 at nearly 550,000. In total, 3.4 million men were deployed to Southeast Asia on combat tours that typically lasted for one year.

As military operations escalated, the annual number of new enlisted personnel roughly doubled in the first two years of the war (see Figure 2). In all, more than 6 million service members joined the military during the Vietnam War.

Source: Congressional Budget Office, using data from the Department of Defense and the Selective Service System.

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973. The Department of Defense estimates the number of new enlisted personnel (accessions) it will need each year to maintain specific force levels. The values shown here apply to new enlisted personnel without prior military service.

Less than 25 percent of the personnel who served during the Vietnam War were drafted; the others were volunteers. Service members generally reflected America’s young male population. In 1965 and thereafter, a majority of the armed forces were White males with high school diplomas, age 35 or younger. Men who had joined the military before the war began and served during the Vietnam conflict were likely to be more educated than the average service member and to complete a military career.

Composition and Quality of the Military

As in previous eras, men had to meet physical, moral, and mental standards to enter the military. Performance on aptitude tests and high school attainment were the two primary ways the Department of Defense (DoD) measured the quality of the enlisted force—the mental attributes that contribute to productivity and capability in the military. Aptitude tests predicted whether recruits would complete training and receive promotions in the service, and educational attainment was shown to predict later job performance.

Accessions (new enlisted personnel) were at neither the top nor the bottom of their generation in terms of measurements of ability, although the need to pass a medical examination meant that they were healthier. Fewer accessions scored among the top 10 percent on aptitude tests than young male adults generally, and the military did not accept applicants who scored in the bottom 10 percent. About 70 percent of accessions held a high school diploma or higher, compared with slightly more than 75 percent of all men ages 20 to 29.5 Accessions were likely to be more physically fit than nonVeterans their age; in calendar year 1970, for instance, nearly 10 percent of potential volunteers who completed a comprehensive physical examination were disqualified for medical reasons.6

The overall quality of the enlisted forces differed depending on when they had joined the military. In 1965, about half of all active-duty personnel had already completed at least four years of service and were more likely to stay in the military for a full career (20 years) than wartime accessions. Members of the former group received more training than new enlisted personnel over their careers, had more qualifications, and probably differed in other ways as well, such as in their desire to serve in the military. For instance, in 1965, about 82 percent of all enlisted personnel (careerists and new personnel) held at least a high school diploma, higher than the average for accessions.

The officer corps was both more educated and better paid than enlisted men: All officers commissioned during the war held high school diplomas, and the vast majority had earned college degrees. However, the corps was much smaller than the enlisted force: Fewer than 500,000 officers were commissioned during the war, CBO estimates.

The Draft

About 1.9 million men were drafted during the Vietnam War, less than one-quarter of all those who served and less than one-half of accessions. Most draftees went into the Army and were deployed to Southeast Asia. The most common length of service was two years.

In the early years of the war, the draft followed the same basic system used since World War II, relying on local draft boards to select candidates. Men who registered for the draft were typically granted deferment if they had children or other dependents, were in college, or held jobs that were in the national interest. Exemptions (disqualification for medical or other reasons) were common.7 Draftees were 20 to 21 years old, on average.

As the population of young men grew in the early 1960s (because of the post-World War II baby boom), the use of deferments expanded: By 1968, one-third of men who registered for the draft received deferments, up from just 13 percent in 1958. The largest category of deferment (4.1 million) was for men with dependents, followed by those who were enrolled in college (1.8 million).8

As the war escalated, opposition to the draft intensified and concern about the fairness of deferments and exemptions increased. Educational deferments for college, in particular, were thought to favor the affluent.

A national commission in 1966 recommended replacing the existing system with a nationwide lottery, among other changes. The first lottery, in December 1969, was for men born between January 1, 1944 and December 31, 1950 (ages 19 to 25). In the early 1970s, occupational, agricultural, new-paternity, and new-student deferments were largely eliminated. Nevertheless, an individual’s chance of being drafted declined, both because draft calls decreased substantially in the 1970s and because later lotteries only considered men who would turn 20 years old in their enlistment year. In December 1972, the Selective Service System held its final lottery; on July 1, 1973, legal authority to draft men into the military expired.

Volunteers

Volunteers made up the remaining three-quarters of service members during the war. Some of those who enlisted during the war did so to avoid being drafted: Unlike draftees, volunteers could choose a service branch that might reduce their risk of being deployed (although volunteering also increased the length of an enlistment). Others joined in part because the pay and benefits were reasonable compared with civilian options. For enlisted careerists, pay was typically about 85 percent of the median pay (the midpoint value in the pay range) of White high school graduates of comparable ages for most of the war. In the 1960s, junior enlisted personnel were paid much less than most of them could have earned elsewhere, but pay raises in the early 1970s brought their income into rough parity with the private-sector wages of young high school graduates.9 Officers’ pay was consistently above the median pay of White college graduates. The military also offered additional pay to some men with special skills or who were in occupations with shortages.

Nor was income the only inducement to serve. The military offered occupational training and educational benefits. Additional benefits that were typically not found in the private sector included free medical care, low-cost groceries and merchandise from commissaries and exchanges, and on-base bowling alleys, movie theaters, and gyms. Other, less quantifiable aspects of military service—such as patriotism—also attracted volunteers.

Characteristics of Vietnam Veterans in 2018

Various characteristics—including age, education, and health—influence the type and amount of income adults have during their working years and in retirement. Vietnam Veterans differed from nonVeterans in several ways.

CBO examined male Veterans ages 63 to 78 (mostly born between 1940 and 1955) who served during the Vietnam War, about 85 percent of the more than 6 million Veterans of that era who were still alive in 2018. Veterans born before 1940 probably spent a substantial portion of their careers serving in the peacetime era before that war, and those born after 1955 would have been part of the all-volunteer force. Members of the National Guard and reserves were not considered Vietnam Veterans unless they served on active duty during the war. Veterans who did not serve during the Vietnam War were excluded from the samples that CBO analyzed. Women were excluded because they composed just 4 percent of Vietnam Veterans in 2018. Because the ACS does not ask whether the Veterans were drafted or where they served, CBO did not incorporate those factors in its analysis.

The average Vietnam Veteran was more likely than the average nonVeteran to have a high school degree, to be White, to be a U.S. citizen, and to report a serious impairment in his ability to function (see Table 1). Such functional disabilities did not necessarily qualify Veterans to receive disability compensation from VA.

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Data for Vietnam Veterans were weighted to match the age distribution of nonVeteran men. The median age of Vietnam Veterans in 2018 was 70 and of nonVeterans was 68.

  1. Includes men who completed grade 12 but received no diploma.
  1. Includes men who received a general equivalency diploma or alternative credential.
  1. Includes anyone who reported Hispanic ancestry; the other three categories exclude that group.
  1. Men were considered functionally disabled if they reported serious difficulty hearing, seeing, remembering, moving (such as walking or climbing stairs), or problems with self-care or independent living.

Age

Vietnam Veterans were, on average, older than men in the same age group who did not serve. Veterans’ age distribution was bell-shaped because a large number of 18- to 23-year-olds entered military service during the height of the war (1966 to 1969). Fewer service members joined before 1965 or later in the war, so Veterans in 2018 were most commonly in their late 60s and early 70s (see Figure 3). Veterans’ modal age was 71. Nearly 45 percent of all men age 71 were Veterans.

Source: Congressional Budget Office, using data from the American Community Survey.

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

In contrast to Veterans’ age distribution, the distribution of nonVeterans was highest at age 63 and tailed off. Through 2000, the mortality rate for all Vietnam Veterans in the sample was similar to the mortality rate for nonVeterans, according to the limited evidence available.10

Education

Vietnam Veterans had a different pattern of educational attainment than nonVeterans. Although enlisted personnel were a little less likely than nonVeterans to have a high school diploma when they entered military service, by 2018 Vietnam Veterans were much more likely to have completed a high school education or some college. That was partly because DoD and VA offered financial assistance to help service members and Veterans further their education. For instance, more than 200,000 soldiers in the Army earned a high school diploma or its equivalent (a general educational development credential), many under the Pre-discharge Education Program.11 In addition, almost five million Vietnam Veterans used the GI Bill for further education or training.12 However, a greater share of nonVeterans than Veterans completed a postsecondary degree (36 and 27 percent, respectively). The Veterans’ share was even smaller for those younger than 72. Although some evidence shows that many men enrolled in postsecondary programs partly in an effort to avoid military service, the draft had little effect on whether they completed those programs.13

Ethnicity and Geographic Distribution

Vietnam Veterans were more likely than nonVeterans to be White and less likely to be Hispanic. Roughly equal percentages of Veterans and nonVeterans were Black. However, a slightly larger share of Black men than White men joined the enlisted ranks in the early 1970s.14 (In today’s military, the share of non-Whites who enlist is bigger than in the Vietnam era, partly because a larger share of U.S. youth are not White.) Like Veterans of other eras, Vietnam Veterans have tended to live and retire in the South, where large active-duty populations are located. Only a small share of all men, both Vietnam Veterans and nonVeterans, lived in rural communities.

Health

Vietnam Veterans were more likely than nonVeterans to report functional disabilities—impairments that restrict someone’s ability to work or undertake daily activities.15 Although Veterans would have had to meet the military’s health and physical fitness standards, serious injuries and medical impairments may be more common among them because of intense physical training or experiences during deployment and combat. Research has confirmed that Vietnam Veterans have more health problems and functional disabilities than nonVeterans, including hearing loss, hepatitis C, and post-traumatic stress disorder.16 Some evidence suggests that those differences in health are increasing as the population ages. It is not certain, however, to what extent health problems are the result of military service.17

Just over 40 percent of the Veterans who reported functional disabilities on the ACS received disability payments from VA, which may be awarded to Veterans with a medical condition that developed or worsened during their service.18 VA determines whether Veterans qualify for such service-connected disability payments and the amount of payment they receive. Service-connected disabilities are not necessarily functional disabilities—they may not substantially affect a Veteran’s ability to work or perform day-to-day activities.

Compared with other men the same age in 2018, Vietnam Veterans were less likely to work. In a recent study, researchers at CBO and the Urban Institute determined that older workers with little education, poor health, other income sources (besides earnings), or with health insurance were the most likely to retire early.19 Several of those categories apply to many Veterans.

CBO’s Approach to Analyzing Veterans’ Income

To quantify the income of Vietnam Veterans and nonVeterans, CBO analyzed ACS data from 2008 and 2018.20 The ACS is among the largest of the Census Bureau’s surveys, reaching about 2 million households each year, and is designed to represent the entire U.S. population. Households in the ACS survey provide information on demographics, employment status, education, disabilities, and military service, among other topics. More than 75,000 male Veterans who said they served during the Vietnam War were interviewed for the 2018 ACS.21 (CBO used the ACS’s survey weights to estimate data for all male Vietnam Veterans and nonVeterans.)

CBO chose the ACS after examining the Current Population Survey and the Survey of Income and Program Participation. The ACS covers more Veterans than those surveys and is about as accurate, CBO concluded.

To determine VA’s disability payments, CBO analyzed administrative data from VA, partly because VA’s disability compensation is mingled with several other types of income in a residual category on the ACS.

CBO examined five sources of regular income: earnings, Social Security, retirement plans, investment income, and disability payments from VA. Onetime payments such as inheritances or home sales were excluded.22 The first four income sources CBO studied were common to all men; the fifth, disability compensation, could only be received by Veterans.

  • Earnings. Earnings were defined as money an employee received in the form of wages and salaries (including tips, commissions, and bonuses) and self-employment income from a business, professional practice, or farm after subtracting business expenses from gross receipts. (No distinction was made between incorporated and unincorporated businesses.)
  • Social Security. ACS included retirement, disability, and survivors’ benefits, and payments made by the U.S. Railroad Retirement Board. (Railroad workers who qualify for retirement benefits do not participate in the Social Security program.)
  • Investments. The category included income from assets: interest, dividends, royalties, rents, and income from estates or trusts.
  • Retirement Plans. ACS defined retirement plans as pensions in the form of defined benefit and defined contribution plans from companies, unions, and federal, state, and local governments (including the military); individual retirement accounts; Keogh plans; and any other type of pension, retirement account or annuity. Pensions paid to survivors and disability pensions were included.
  • Disability Compensation from VA. That compensation went to Veterans with medical conditions or injuries that were incurred or aggravated during active-duty military service.

CBO did not include income from other sources reported in the ACS, such as alimony, public assistance, Supplemental Security Income, or VA’s pensions for low-income wartime Veterans. Those sources accounted for less than 5 percent of total income for the men in this analysis. For instance, less than 3 percent of Vietnam Veterans collected pensions from VA, according to that agency. In addition, the accuracy of reporting on those other sources of income was unclear.

CBO focused on broad trends and did not quantify the net effects of education, race, ethnicity, workforce experience, and other characteristics that exert a strong influence on the amount and sources of income people receive. Similarly, CBO did not examine other financial measures such as debt or expenses. Other studies have examined Veterans’ wealth and financial well-being.23

Survey data vary from year to year, so reporting of income fluctuates annually. However, CBO conducted a sensitivity analysis that indicated the results for 2017 and 2018 did not differ substantially from each other.

Vietnam Veterans’ Earnings in 2008

To help understand Vietnam Veterans’ income in retirement, CBO examined their earnings in 2008, when well over half of all men in that group were still working (as measured by the percentage who reported wage, salary, or self-employment income). Earnings before retirement are closely linked to income after exiting the labor force. For instance, employer-sponsored pensions are based on a percentage of the employee’s annual pay and the number of years worked at that employer.

In 2008, Vietnam Veterans ranged in age from 53 to 68; CBO examined earnings only for men 53 to 65 because a majority of those older than 65 no longer worked. CBO looked at earnings by age because earnings can differ substantially by age. In addition, comparing Veterans to nonVeterans by age accounted for the disparity in the two groups’ age distribution.

Vietnam Veterans earned an average of $50,000 in 2008, 20 percent less than nonVeterans in the same age range, who earned an average of $62,200.24 Veterans younger than 63 earned less than nonVeterans their ages, but Veterans age 63 or older earned about the same. At age 61, Veterans’ modal age in 2008, the difference was smaller, $6,500. Average earnings at that age were $52,600 for Veterans and $59,100 for nonVeterans (see Figure 4, top panel).

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Vietnam Veterans (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

Earnings refer to wages, salaries, and self-employment income. Working individuals are those with nonzero earnings.

 

CBO did not adjust the data to account for characteristics other than age and Veteran status.

When CBO looked only at men with wages, salaries, or self-employment earnings in 2008, Vietnam Veterans earned, on average, $71,200, 12 percent less than nonVeterans, who averaged $80,700. At age 61, Veterans who worked earned 9 percent less, on average, than nonVeterans.

Veterans’ earnings may have been lower than nonVeterans’ earnings for several reasons: because their military experience did not fully substitute for the labor market experience it replaced; because they chose jobs with lower pay but more generous retirement benefits; because they had difficulty transitioning to civilian life; or because they had fewer job opportunities in the civilian sector, and that disadvantage might have persisted throughout their working years.

Other factors, such as employers’ preferences and legislative actions to assist Veterans, surely played a part in Veterans’ work history and income, although the magnitude of those effects is unknown. Employers’ views of Veterans could have influenced their decisions about hiring and wages. Some employers may have believed that Veterans had physical or emotional difficulties as a result of military service; others that Veterans had special skills or attributes such as a sense of leadership, teamwork, or commitment. Federal legislation may have been a factor as well. Federal law prohibits discrimination in employment on the basis of past, current, or future military service.25 In addition, a law enacted in 1944 (and amended many times since) grants a hiring preference to Veterans who apply for jobs in the federal government. Vietnam Veterans might have had the occupational skills to fill many positions in the federal civilian workforce.

Compared with nonVeterans, a markedly smaller share (5 percent smaller) of Vietnam Veterans younger than 60 worked in 2008 (see Figure 4, bottom panel). The reason that a different share of Vietnam Veterans and nonVeterans worked is unclear. Higher rates of functional disability among those Veterans may be part of the explanation, although 43 percent of Veterans with those disabilities worked.26 Some researchers have suggested that disability payments from VA to some Veterans allowed them to leave the labor market early.27 Among men who held jobs, about the same share of Veterans (88 percent) worked full time as nonVeterans (89 percent).

The sector of the economy in which workers were employed affected their retirement benefits. A larger share of Vietnam Veterans (7 percent) than nonVeterans (2 percent) worked for the federal government. Veterans (of all wars) made up 25 percent of the federal workforce in 2008, and 40 percent of DoD’s civilian employees. Similar shares of Veterans and nonVeterans worked for state or local governments (13 percent for both) and the private sector (64 percent of Veterans and 63 percent of nonVeterans). However, Veterans were more likely than nonVeterans to have blue-collar jobs in fields with a significant union presence, such as transportation and construction. In addition, fewer Veterans, 16 percent, were self-employed, compared with more than 20 percent of nonVeterans.

Most government workers and two-thirds of union members in the private sector had access to a traditional pension (a defined benefit plan), unlike many other workers. Those who were self-employed probably only had access to individual retirement accounts. Although it is difficult to compare the earnings of Veterans and nonVeterans in different sectors because of differences in jobs and employee qualifications, the types of jobs Vietnam Veterans held in 2008 meant that many could expect a steady stream of income from a retirement plan.

The gap between the earnings of Veterans and nonVeterans, and likelihood of being employed, differed by race. Among all working men ages 53 to 65, White Vietnam Veterans earned an average of $73,700, about 17 percent less than White nonVeterans. The same was true to a much smaller degree for Black Vietnam Veterans, who on average earned $51,200, about $2,200 (4 percent) less than Black nonVeterans. Black men—whether or not they were Veterans—were less likely to work than White men. For example, 72 percent of White male Veterans age 61 worked, compared with 55 percent of Black male Veterans that age; results were similar for White and Black nonVeterans.

The earnings gap between Veterans and nonVeterans also differed by educational level. There was almost no difference (1 percent, or $600) between the average earnings of Vietnam Veterans and nonVeterans who held high school diplomas. At higher levels of education, however, earnings differed. Veterans with college degrees earned less on average (11 percent, or $13,600) than nonVeterans of the same age with college degrees.

Vietnam Veterans’ Income in 2018

By 2018, most Vietnam Veterans and nonVeterans were no longer working. At that point, CBO measured four sources of income for men ages 63 to 78 that probably composed the vast majority of income for most men: earnings, Social Security benefits, investments, and retirement plans. For Veterans, the agency also considered disability compensation from VA. That compensation, however, was paid only to certain Veterans. Therefore, CBO calculated total income for Veterans two ways: by averaging all five sources of income and by averaging four sources of income, excluding disability payments. CBO then compared both those measures with nonVeterans’ total income.

The disability compensation VA provides is unlike other sources of income because it is only available to Veterans, who may have faced special risks in the course of their military service. That compensation can be viewed as a work-related benefit, a lifetime indemnification that the federal government owes to Veterans with a medical condition that was incurred or worsened while they were in the military. If those disabled Veterans are out of the workforce for a long time, their Social Security benefits will be less than they otherwise would be, and they might not accumulate much personal savings. VA’s compensation mitigates such effects. The amount of that compensation can be sizable: CBO calculated that for Vietnam Veterans who received it, average annual disability compensation was nearly as much as they received in Social Security benefits.

Total Income

Veterans’ average income, including disability compensation, was $63,300, whereas the average for nonVeterans was $65,000. CBO found that disability compensation from VA increased Veterans’ income by $4,300, on average, bringing it close to nonVeterans’ income. Excluding disability income from VA, Vietnam Veterans’ average income was $59,000.

Differences by Age. When including disability compensation, Veterans at the modal age (71) had an average income of $65,500, $200 less than the average for all nonVeterans that age ($65,700). Whether or not disability income was included in income, Veterans younger than 71 had less income, on average, than nonVeterans, and Veterans older than 71 had more income, on average, than nonVeterans (see Figure 5, top panel).

Source: Congressional Budget Office, using data from the American Community Survey and the Department of Veterans Affairs.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Vietnam Veterans (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

Except where noted, average income is from five sources: earnings (wages, salaries, and self-employment); Social Security; investments; retirement plans; and disability compensation from VA. Median income excludes VA disability compensation. That disability compensation is untaxed and is provided to Veterans with medical injuries or conditions that were incurred or aggravated during active-duty military service.

 

CBO did not adjust the data to account for characteristics other than age and Veteran status.

VA = Department of Veterans Affairs.

Certain dissimilarities between younger and older Vietnam Veterans make it doubtful that younger Veterans will ever have the same amount of income as older Veterans. In effect, there were two sets of Vietnam Veterans, rather than a single group. Older Veterans probably earned more than younger Veterans while working. They entered the military before the war escalated in 1965, were less likely to have been drafted, and were more likely to have had careers in the military and to collect military retirement pay after a 20-year career. As previously observed, a greater share of older Veterans had college degrees than younger Veterans. In addition, the share of older Veterans with low test scores upon entry into the military was smaller than the share of younger Veterans.28 They also probably differed in other ways that affected their income in retirement.

Income Distribution. To exclude extremely high and low values that might have skewed income findings, CBO also examined median income (the midpoint value in the income range) for Veterans and nonVeterans. Excluding disability payments, at age 71, median income for Veterans ($44,200) was higher than for nonVeterans ($41,800).29 (See Figure 5, bottom panel.)

Comparing Veterans’ and nonVeterans’ income distribution (excluding VA’s disability payments) confirmed that fewer Veterans had low or high income (see Figure 6). When income for all men ages 63 to 78 was grouped into quintiles, only 15 percent of Veterans fell into the bottom quintile, compared with 22 percent of nonVeterans. Even in the bottom quintile, Veterans’ income was 10 percent higher than nonVeterans’ income. In addition, disability compensation and other benefits like free health care and special pensions for poor Veterans (neither of which were included in CBO’s calculations) probably helped support the neediest Veterans. On the opposite end of the income distribution, 17 percent of Vietnam Veterans and 21 percent of nonVeterans had income in the highest quintile; nonVeterans in that quintile had average annual income that was more than 10 percent higher than Veterans’ income.30

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Includes income from earnings (wages, salaries, and self-employment); Social Security; investments; and retirement plans. Excludes income from disability compensation from the Department of Veterans Affairs.

 

Bars for Veterans add up to 100 percent, as do bars for nonVeterans.

Common Components of Income

Although Vietnam Veterans and nonVeterans relied on many of the same sources of income, some sources were more important for Veterans. On average, Vietnam Veterans had a greater share of annual income from Social Security and retirement plans than nonVeterans, who remained more reliant on earnings even in old age. That was also true when men at Veterans’ modal age of 71 were examined (see Table 2). CBO looked at income by age because the sources and amounts of income differ by age.

Source: Congressional Budget Office, using data from the American Community Survey and Department of Veterans Affairs.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Income refers to income from five sources: earnings (wages, salaries, and self-employment); Social Security; investments; retirement plans; and disability compensation from VA. Disability compensation is untaxed and is provided to Veterans with medical injuries or conditions that were incurred or aggravated during active-duty military service.

 

VA= Department of Veterans Affairs; n.a. = not applicable

Earnings. On average, working Vietnam Veterans ages 63 to 78 earned $52,800 annually, 23 percent less than other working men ($68,700). For men at age 71, the difference was smaller: Veterans earned $50,500, and nonVeterans earned $62,200 (see Figure 7, top panel). The earnings gap was larger for younger men than for older men.

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Veterans who served in the Vietnam War (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

Earnings refer to wages, salaries, and self-employment income. Working individuals are those with nonzero earnings.

 

CBO did not adjust the data to account for characteristics other than age and Veteran status.

The gap between Veterans and nonVeterans’ earnings for those who worked occurred partly because a large share of less-educated nonVeterans—who probably had lower-wage jobs—had left the workforce by 2018. Their departure raised the average wage for nonVeterans who kept working. In addition, just half of working Veterans were employed full time, compared with 62 percent of working nonVeterans.

Because most men between the ages of 63 and 78 were not working, the average earnings among all of them were much lower: $14,900 for Veterans and $28,200 for nonVeterans. For men age 71, the difference was smaller: those averages were $12,400 and $18,800, respectively. However, earnings were becoming a smaller share of total income for all men in retirement.

Part of the difference in average earnings arose because there was a gap in the share of Veterans and nonVeterans who were employed at age 63 or older (see Figure 7, bottom panel). The gap in employment was largest at age 63, indicating that Veterans left the workforce at younger ages than nonVeterans. Veterans may have stopped working earlier for several reasons, including their higher disability rates or the possibility that VA’s disability compensation enabled them to leave the labor market earlier.

Among men with self-employment income in 2018, nonVeterans were a little more likely than Vietnam Veterans to report such income, and on average they reported more: $47,800, compared with $36,600 for Veterans. Self-employment was the main source of earnings for about 30 percent of all men with earnings income.31

Social Security. Participation in Social Security was extremely high for all men CBO studied. Almost all Vietnam Veterans (94 percent) and nonVeterans (90 percent) received payments at age 71 (see Figure 8, bottom panel).32 The share of Veterans that collected Social Security benefits grew with each year of age until age 71 and then remained flat.

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Veterans who served in the Vietnam War (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

Social Security payments include retirement, disability, and survivors’ benefits, as well as payments made by the U.S. Railroad Retirement Board.

 

CBO did not adjust the data to account for characteristics other than age and Veteran status.

Veterans’ average annual Social Security payment was $17,900, much higher than nonVeterans’ $13,900 average payment. That is generally explained by the fact that more Veterans than nonVeterans received Social Security benefits. A larger share of Veterans (88 percent) than nonVeterans (69 percent) had turned 66 by 2018, the age at which they were eligible for full Social Security benefits.33 In addition, many Veterans claimed Social Security benefits earlier than nonVeterans (in general, however, claiming benefits early decreases the monthly payment for that recipient).

When only men who received Social Security payments were considered, a small difference persisted, but only at certain ages and levels of income. Veterans who received payments collected, on average, $20,800; nonVeterans collected $20,000, about $800 less.

At age 71, Social Security income was about the same for Veterans and nonVeterans (see Figure 8, top panel). Veterans younger than 71 collected slightly less in Social Security benefits, and Veterans older than 71 collected more. Strikingly, even those older Veterans with lower earnings in 2008 collected more Social Security benefits than nonVeterans, on average.

CBO examined special earnings credits toward Social Security that Vietnam Veterans received. But those credits probably do not explain older Veterans’ bigger Social Security payments. (Those additional credits were not added to Veterans’ Social Security payments; rather, they increased the total amount of income Social Security considered when setting monthly benefits.) Veterans on active duty from 1957 through 1977 were credited with $300 in additional earnings for each calendar quarter in which they received basic pay. For Veterans who continued serving after 1977, the calculation for the credits changed but the maximum remained $1,200 per year. (In January 2002, the extra earnings credits ceased.)

CBO concluded that the formula used to calculate benefits was the most likely explanation for the difference between average Social Security payments for older Veterans and nonVeterans.34 NonVeterans were more likely to have had very low or very high earnings than Veterans during their working lives, which might have generated a disparity in average Social Security benefits. Benefit amounts for very high earners—more of whom were nonVeterans—replace a much smaller share of earnings than for others; benefits for low earners—more of whom were also nonVeterans—more closely reflected their lower earnings. Thus, the group of nonVeterans qualified for a slightly smaller amount of Social Security benefits, on average.

Investments. About 30 percent of Veterans and nonVeterans received investment income, and it was a small share of total income for all recipients except the wealthiest. The share of all men with investment income increased from less than 10 percent in the lowest income quintile to more than 50 percent for those in the highest quintile. Men in the lowest income group received an average of less than $500 annually; those with total annual income in the top quintile collected roughly $30,000.35

A slightly smaller share of Vietnam Veterans younger than 71 reported investment income compared with nonVeterans. Among men 71 and older, a bigger share of Veterans had such income than nonVeterans. Among all recipients, average income from investments was $19,900 for Veterans and $26,800 for nonVeterans. At age 71, Veteran recipients reported $19,200 in investment income and nonVeteran recipients $28,000, a difference of $8,800 (see Figure 9).

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Veterans who served in the Vietnam War (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

Investment income is from assets: interest, dividends, royalties, rents, and income from estates or trusts.

 

CBO did not adjust the data to account for characteristics other than age and Veteran status.

Average investment income was smaller when the average includes people who did not receive such income. It was less, on average, for Veterans than for nonVeterans. For 71-year-olds, investment income averaged $6,400 for Veterans and $9,200 for nonVeterans. In general, amounts were greater for older men regardless of their Veteran status. Investment income for older households—generally, those with at least one person age 65 or older—has been declining since the 1990s because fewer and fewer adults own financial instruments.

Retirement Plans. Employer-sponsored retirement plans were an important source of income for older Americans. On average, Vietnam Veterans had about one-third more income from retirement plans in 2018 than nonVeterans, ($20,300 and $15,300, respectively), probably in part because Veterans were much more likely to have held jobs that offered such plans. At age 71, Veterans had, on average, more income from retirement plans than nonVeterans: $21,000 and $18,000, respectively (see Figure 10, top panel).

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Veterans who served in the Vietnam War (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

Retirement plans include defined-benefit and defined-contribution plans from companies, unions, and federal, state, and local governments (including the military); individual retirement accounts; Keogh plans; and any other type of pension, retirement account, or annuity.

 

CBO did not adjust the data to account for characteristics other than age and Veteran status.

A much bigger share of Veterans than nonVeterans received payments from retirement plans. The gap was about 30 percentage points for the youngest men but shrank to about 17 percentage points by age 66 (see Figure 10, bottom panel). The finding suggests that, in addition to having greater access to retirement plans, Vietnam Veterans began drawing income from their retirement plans earlier than other men.

As CBO’s 2008 analysis of earnings showed, more Veterans than nonVeterans were likely to have held jobs that included a defined benefit or defined contribution plan. For example, about 20 percent of Vietnam Veterans had a government job, compared with 15 percent of nonVeterans.36

Federal workers have access to both defined benefit and defined contribution plans. State and local government workers commonly have access to only a defined benefit plan, and private-sector workers typically have access only to a defined contribution plan. Access to and participation in any type of retirement benefit is much greater for government workers; in 2008, about 85 percent of government employees participated in a retirement plan compared with about 50 percent of private industry workers. In addition, more nonVeterans than Veterans were self-employed and were less likely to have defined benefit retirement plans.

The military offers separate retirement plans (including a defined benefit plan that is usually available after 20 years of service). In 2018, a little over 10 percent of Vietnam Veterans collected a military pension, CBO estimates. Although relatively few Vietnam Veterans qualified for retirement benefits from the military, those who did would typically have been in their late 30s or early 40s when they separated from service; they could have joined the civilian labor force and had second careers, adding to their retirement income.

When income from retirement plans is averaged only among those who received it rather than among all men, average annual payments were a bit smaller for Vietnam Veterans ($26,800) than for nonVeterans ($28,200). At the modal age of 71, Veterans collecting retirement income also received less ($27,500) than nonVeterans ($29,500). Vietnam Veterans older than 73, though, received about $1,500 more, on average, than nonVeterans. The payment differences can probably be explained by the nature of the jobs each group held while working.

Disability Compensation from the Department of Veterans Affairs

In addition to the four sources of retirement income that could be collected by all men, some Veterans also collected disability payments from VA for injuries or medical conditions they received or developed during active-duty service. (See Box 1 for detail on VA’s disability program and other benefits for Veterans.) The average payment for recipients in 2018 was $18,100, so VA’s disability payments were a substantial part of income for some Veterans. About 24 percent of Vietnam Veterans received disability payments, CBO estimates.

For Vietnam Veterans ages 63 to 78, including those who did not receive disability payments, the average annual disability payment was $4,300; the average payment at age 71 was $4,900 (see Figure 11). Those payments make up much of the average difference in other income between Veterans and nonVeterans, increasing Veterans’ average income by about 7 percent to $63,300.

Source: Congressional Budget Office, using data from the American Community Survey.

 

The Vietnam War ran from August 1964 (fiscal year 1965) to January 1973.

 

Shaded area denotes the largest three-year cluster of Veterans who served in the Vietnam War (one-third of total). Veterans in that cluster probably served from 1966 to 1969, when forces were largest.

 

The Department of Veterans Affairs provides tax-free disability compensation to Veterans with medical injuries or conditions that were incurred or aggravated during active-duty military service.

 

CBO did not adjust the data to account for characteristics other than age.

Vietnam Veterans who received disability payments from VA were likely to be in the lower half of the income distribution for all men in the age range that CBO analyzed. In addition, they had less income from other sources than Veterans who did not receive VA payments, mainly because they were less likely to be employed and because those who were employed had lower earnings. However, on average, their disability payments made up that difference, increasing the average total income of Veterans who received those payments above the average income of Veterans who did not.

The incidence and amount of disability payments from VA varied by age. A greater share of Vietnam Veterans who were 68 to 71 years old in 2018 received disability payments (28 percent) than Veterans of other ages (22 percent). Furthermore, partly because those 68- to 71-year-olds were also likely to have higher disability ratings (70 percent or more), CBO estimates that the average payments for that age group were about $19,400 a year, $1,300 more than the $18,100 average payment for all Vietnam Veterans receiving VA disability compensation.

Because Vietnam Veterans ages 68 to 71 were likely to have served in the mid- to late 1960s, when deployments to Vietnam and surrounding countries were largest, they may have been more likely to experience combat-related injuries and medical conditions, some of which did not show up immediately. Although Veterans claim far more disabilities than the number of combat injuries reported at the time (about 150,000), many medical conditions can take years to develop (such as limitations on joint movement) or may not be identified at the time (such as mental illnesses).

In addition to such late-onset or previously unrecognized conditions, VA has declared that several medical conditions can be presumed to have been caused by the service member’s deployment to Vietnam. Those presumptive conditions substantially increase the number of Veterans who receive disability payments.37 Among such presumptive conditions are diabetes, certain cancers, and other problems that may have arisen from exposure to Agent Orange or other herbicides during the war. A sizable share of Vietnam Veterans receive VA disability payments for presumptive conditions. For instance, VA administrative data showed that nearly 375,000 Vietnam Veterans received disability compensation for diabetes in 2018; nearly all of those cases were considered presumptive conditions.

Limitations of CBO’s Study

CBO’s study has limitations that are common to any analysis based on survey data. Survey data may be inaccurate because of misreporting. Some people may not answer certain questions or underreport the amount they receive from particular sources. Both problems have been extensively noted.38

The Census Bureau uses a statistical approach to estimate the values of missing responses. CBO also used statistical methods to assign Social Security and retirement-plan income to some respondents. About 95 percent of men ages 60 or older in the ACS correctly responded that they received Social Security benefits, according to administrative data. However, nearly 20 percent underreported the amount of Social Security benefits they received. CBO therefore imputed additional Social Security income to those who reported that they had received it. Some of those who receive income from retirement plans do not report it at all, so CBO imputed such income to some respondents. For a discussion of CBO’s approach to imputing retirement-plan income, see the appendix.

CBO did not adjust ACS’s data for earnings or investment income. Information about wages and salaries is generally reported correctly. The exception is self-employment income, which is subject to nonresponse and underreporting.39 Likewise, investment income is subject to nonresponse and underreporting. For those income sources, there is no consensus on how to account for those problems. However, the consequences of nonresponse and underreporting in those two categories probably had a minimal effect on this report: In the United States, self-employment and investments are a small portion of total annual income for most retired men.40

CBO also used VA’s data to estimate the amount of disability compensation Vietnam Veterans received. That was necessary because ACS data only included the number of people receiving VA disability compensation, not the amount of that compensation. CBO adjusted ACS’s profile for those Veterans on the basis of VA’s aggregate data. It then used VA’s data on total spending and average payments by disability rating to estimate the average amount of payments to Vietnam Veterans by age.

Appendix: How CBO Imputed Income from Retirement Plans

The quality and usefulness of survey data depend on the accuracy of responses from the people questioned. Misreporting is a common problem—typically, nonresponse (people not answering a question) and underreporting (people reporting less of an amount than they actually receive). A large body of literature discusses the quality of household surveys and the type and extent of measurement errors they contain.1

The reporting of income from retirement plans is a particular problem. According to researchers at the Census Bureau, respondents to the 2013 American Community Survey (ACS) who were age 65 or older failed to report income from retirement plans 44 percent of the time.2 (In this Congressional Budget Office report, retirement plans encompass defined benefit and defined contribution plans from companies, unions, and federal, state, and local governments [including the military]; individual retirement accounts; Keogh plans; and any other type of pension, retirement account or annuity, including survivors’ pensions and disability pensions.)

There are a number of reasons that income from retirement plans in surveys is subject to greater measurement error than other income sources, and some methods have been developed to account for such errors.3 Recipients of such income might perceive it as a form of deferred earnings rather than contemporaneous income. Survey respondents may also find it difficult to recall irregular income payments over a long reference period.

To account for misreporting of retirement income, CBO took steps to impute income from retirement plans and validate those results.4

Imputing Income from Retirement Plans

For its analysis, CBO used a regression-based method with three steps. First, CBO developed targets for the total number of people receiving income from retirement plans by age group and by year, using administrative data. Second, it selected respondents for imputation by using a regression-based method while matching the targets set in the first step. Third, CBO used predictive mean matching (PMM) for the selected data: The agency imputed income amounts by borrowing values from donors (respondents who were observationally similar to those who did not report retirement-plan income).

Developing Administrative Targets

CBO developed targets using administrative data from the Internal Revenue Service (IRS). Those data showed the number of people who received retirement income, by survey year and age group. The two age-based targets for receipt of income from retirement plans were ages 55 to 64 and age 65 or older. Both were derived from estimates reported in a working paper published in 2017 by the Census Bureau’s Social, Economic, and Housing Statistics Division. The authors linked respondents in several national datasets, including the 2013 ACS, to Forms 1099-R filed with the IRS in 2012 by banks and other entities to report disbursements they made from retirement accounts.5 About 60 percent of respondents 65 or older received a distribution reported on a Form 1099-R.

By using the linked Current Population Survey and Survey of Income Program Participation, authors Adam Bee and Joshua Mitchell showed that the share of respondents to those surveys who received retirement income had remained roughly constant for that age group since 2004. Therefore, CBO estimated that 60 percent of respondents 65 or older received income from retirement plans in 2013 and 2018. For respondents ages 55 to 64, only 26 percent received 1099-R distributions in 2012. Bee and Mitchell did not provide estimates of how that proportion had changed over time. CBO estimated that the share of respondents in that age group who received retirement-plan income was constant at a rate of 26 percent.

Imputing Receipt of Income

CBO used the targets it developed to derive the number of respondents in the survey for which imputation was needed. To allocate those imputations, CBO used a probit regression to predict each respondent’s probability of receiving income from retirement plans.6 Separate regressions were estimated for two age groups (55-to-64-year-olds and those 65 or older) and two survey years (2013 and 2018). The regressions tended to assign higher probabilities to individuals who were male, older, more educated, White, and Veterans.

Each person’s predicted probability of receiving income was compared with a random number (uniformly distributed between zero and 1). If the respondent’s probability of receiving income exceeded the random number assigned to that probability, CBO designated that person to receive an imputation. The random-number comparison was repeated for each individual in the survey until the sum of the respondents with retirement income and the respondents designated to receive an imputation matched the administrative target. A random component was included in the procedure to incorporate heterogeneity in the imputed data. A feature of that method is the assumption that the distribution of respondents in the survey data for whom imputation was needed was similar to the distribution of other respondents, which may not be accurate.

Imputing Amounts of Income

CBO assigned income values to respondents using PMM.7 Under that method, respondents who had been designated to receive retirement income were matched with observationally similar cases (donors), and the amount of income the respondents was assigned was determined by the amount the matching donors reported. PMM has a number of desirable properties. In contrast to standard regression imputation, which is deterministic, PMM includes a random component to introduce heterogeneity into the imputed values. Further, PMM’s semi-parametric imputation process is less vulnerable to model misspecification than a fully parametric imputation model. Relative to hot-deck imputation within adjustment cells, PMM allows for more predictors and for the use of continuous variables as predictors.8

Using PMM involved several steps. First, a linear regression model was used to predict the amount of income from retirement plans for each person (including those who would receive an imputation). Retirement income was regressed on the same covariates used in the probit model four times, using different combinations of the two age groups and the two survey years. Second, predicted values were computed by using a set of coefficients randomly drawn from a posterior predictive distribution. That procedure aimed to introduce sufficient variability in the predicted values by incorporating the residual variance from the regression model. (Without that adjustment, the range of the predicted values would probably have been smaller than the true range.) For those respondents with income values of zero who were selected for imputation, five donors were selected who reported positive income from retirement plans and whose predicted income values were nearest to the target. One of those respondents was then randomly selected to serve as the donor. The previous two steps, obtaining predicted values and selecting donors, were repeated four times to produce five sets of imputed values. The final assigned amount was the average of those five values.

Validating Imputations

Only imputations for respondents to the 2013 ACS who were 65 or older were validated. There are no comparable administrative benchmarks to evaluate CBO’s estimates for 2018 or to evaluate its estimates for the 55-to-64-year-old age group. CBO’s results were similar to Bee and Mitchell’s.

Validating Imputed Recipients

Differences between the characteristics of CBO’s imputed sample and those of Bee and Mitchell’s linked ACS sample were small (see Table A-1). CBO’s procedure imputed shares of all men receiving income from retirement plans and shares of Veterans receiving that income that were slightly larger than those in Bee and Mitchell’s sample—by 2 percentage points and 3 percentage points, respectively.

Source: Congressional Budget Office, using data from the ACS and from C. Adam Bee and Joshua W. Mitchell, Do Older Americans Have More Income Than We Think? Working Paper SEHSD-WP2017-39 (Census Bureau, July 2017), Appendix Table 7, https://go.usa.gov/xfzx7.

 

To create its sample, CBO imputed retirement-plan income to some ACS respondents. The ACS asks respondents to report income from defined benefit and defined contribution plans from companies, unions, federal, state, and local governments (including the military); individual retirement accounts; Keogh plans; and any other type of pension, retirement account, or annuity.

 

For the Census sample, Bee and Mitchell linked ACS data with IRS Form 1099-Rs, which record gross distributions that were made from employer-sponsored plans and individual retirement accounts and that were permanently withdrawn from tax-preferred retirement plans. (Distributions to people who do not file tax returns are not included.) Bee and Mitchell excluded certain kinds of distributions that the authors did not consider income, primarily direct rollovers and conversions, which move money from one tax-preferred retirement plan to another.

 

ACS = American Community Survey; IRS = Internal Revenue Service.

Validating Imputed Amounts

CBO compared aggregate totals in the ACS with the tabulations from the IRS’s Statistics of Income (SOI) division (see Table A-2). The SOI division publishes estimates of aggregate amounts of income from retirement plans by age.9 Those estimates include taxable distributions from individual retirement accounts, taxable and nontaxable pensions, and annuity income.10 CBO’s imputed values for 2018 matched (for ages 65 or older) or exceeded (for ages 55 to 64) SOI’s taxable amount of income but were below the aggregate amount of income. CBO also found that the ACS’s aggregate income from retirement plans in 2013 for those 65 or older was greater than the SOI taxable amount and smaller than the aggregate amount; similar results were reported by Bee and Mitchell. Those results may be partly explained by differences in the types of income included in the two sources of data. For example, the ACS does not include nontaxable direct rollovers, which are included in the aggregate retirement amount reported by SOI; the ACS includes income from people who receive retirement distributions but did not file tax returns, which is not included in the SOI tabulations.11

Source: Congressional Budget Office, using data from the ACS and the IRS’s Statistics of Income—2017 Individual Income Tax Return.

 

The ACS data understates retirement-plan income because many people fail to report it. CBO’s sample includes retirement-plan income both for those who reported such income and those to whom CBO imputed income. The IRS’s taxable and aggregate income figures do not include distributions to people who did not file returns. The IRS’s aggregate income figure includes distributions such as direct rollovers and conversions, which move money from one tax-preferred retirement plan to another.

 

ACS = American Community Survey; IRS = Internal Revenue Service.

  1. a.The IRS publishes taxable income and aggregate income from retirement plans annually in its Statistics of Income—2017 Individual Income Tax Return. Because the IRS data for 2018 had not been published when CBO began its analysis, the agency projected taxable and aggregate income from retirement plans for that year by using IRS data from 2007 to 2017. To make the IRS data more comparable to data reported in the ACS, CBO averaged its projection with IRS data from 2017.

Unlike Bee and Mitchell’s IRS data, CBO’s imputed income amounts were subject to intensive margin underreporting; in other words, among those who correctly reported receiving income from retirement plans, the amounts listed were smaller than the amounts actually received.12 Research has shown that although underreporting in the ACS was primarily driven by the extensive margin (that is, respondents failing to report any retirement income when such income was in fact present in the administrative records), some intensive margin underreporting was also present.13 Because the PMM procedure borrows income values from other respondents in the data, which were also underreported, the imputed values were underreported as well.14 Given that most underreporting was along the extensive margin, and lacking an appropriate target, CBO did not make further adjustment for intensive margin underreporting.15

About This Document

This Congressional Budget Office report was prepared at the request of the Ranking Member of the House Committee on Veterans’ Affairs. In accordance with CBO’s mandate to provide objective, impartial analysis, the report makes no recommendations.

Heidi Golding and Elizabeth Bass prepared the report with guidance from David Mosher and Edward G. Keating. Jimmy Chen (formerly of CBO) and John Kerman contributed to the analysis. William Carrington, Xinzhe Cheng, Molly Dahl, Bilal Habib, Rebecca Heller, Nadia Karamcheva, Noah Meyerson, David Newman, and Julie Topoleski provided useful comments.

David S. C. Chu of the Institute for Defense Analyses and analysts at the Census Bureau provided helpful comments on the draft. (The assistance of external reviewers implies no responsibility for the final product, which rests solely with CBO.)

Jeffrey Kling, John Skeen, and Robert Sunshine reviewed the report. The editor was Elizabeth Schwinn, and the graphics editor was Jorge Salazar. An electronic version of the report is available on CBO’s website (www.cbo.gov/publication/56679).

CBO continually seeks feedback to make its work as useful as possible. Please send any comments to This email address is being protected from spambots. You need JavaScript enabled to view it.">This email address is being protected from spambots. You need JavaScript enabled to view it..

Source

{jcomments on}