Provided is a discussion of the cumulative effects of the measurement alternatives described in the three previous articles: considering family income of persons rather than aged units, using administrative data in place of survey reported data, and switching the data source from CPS to SIPP. The current-methodology CPS statistic of 17.9 percent of beneficiary aged units receiving all of their income from Social Security in 1996 falls to a substantially smallerestimated 4.5 percent of elderly beneficiary persons based on family income when using the SIPP and Social Security administrative data.
Financial adequacy in retirement largely depends on Social Security, pensions, and savings—commonly referred to as the "three-legged stool" of retirement income. Correspondingly, the elderly who receive all of their income from Social Security benefits are recognized as being economically vulnerable. Income of the Population 55 or Older, 2004, produced by the Social Security Administration (SSA 2006), reported that 21 percent of beneficiary aged units 65 or older received all of their income from Social Security. Three other articles in this issue of the Bulletin examine how the estimate of the percentage of elderly beneficiaries receiving most or all of their income from Social Security changes depending on the unit of observation and the source of the data. This article presents the combined effects of the unit of observation and the data source and evaluates the relative impact of each measurement option.
The unit of observation is an important methodological choice related to what constitutes an economic resource. The measure of the relative importance of Social Security income that SSA has been using in its data series includes only income received by an aged unit (aged nonmarried person or aged couple). In addition, each married couple counts as one unit, just as one nonmarried person counts as one unit. Different units of observation have been used for other measures of economic vulnerability; for example, the official poverty measure includes income received by all members of the family and counts each person as a unit when determining poverty rates. This article constructs statistics similar to official poverty measures for comparison with the aged-unit statistics produced by SSA.
The publicly available Annual Social and Economic (ASEC) Supplement to the Current Population Survey (CPS), referred to here as the March Supplement, is used to produce both Income of the Population 55 or Older and Income of the Aged Chartbook. One alternative to using the CPS as the data source would be to generate statistics from the Survey of Income and Program Participation (SIPP). Another option would be to combine administrative data on Social Security benefits and Supplemental Security Income (SSI) with reported data from the CPS or the SIPP. This article uses data on annual income for 1996 from both surveys and administrative data in its analysis.
The article illustrates the effects of these measurement alternatives in combination. The cumulative effect of switching from the CPS to the SIPP, using administrative data in place of survey-reported data, and considering family income of persons rather than aged units is striking. The published statistic of 17.9 percent of elderly beneficiary aged units being completely reliant on Social Security in 1996 falls to an estimated 4.8 percent of elderly beneficiary persons based on family income using the SIPP and administrative data.
The Social Security Administration produces two series on the income of the elderly and near-elderly: Income of the Population 55 or Older (since 1976) and Income of Aged Chartbook (since 1990). Both publications primarily provide income data for aged units, which consist either of nonmarried persons or married couples.
Although these SSA publications focus on aged units, researchers may choose to use different units of observation, such as persons, families, or households, based on the resources they wish to measure. One prominent example is the official poverty statistic produced by the U.S. Census Bureau, which includes family income when determining whether a person is in poverty. To demonstrate the effect of the unit of observation, this article compares estimates of the relative importance of Social Security using aged units with estimates using the family income of aged persons.
Statistics for aged units treat each marital unit (married couple or nonmarried individual) as one unit. A nonmarried individual has only his or her own income and demographic attributes. In both of SSA's data series cited earlier, aged units classified as "65 or older" are nonmarried persons 65 or older or married couples in which the husband is 65 or older or the husband is younger than 55 and the wife is 65 or older. All other demographic characteristics for a married couple, including the sample statistical weight, are that of the husband. Income for the married couple is the sum of both spouses' income; if either spouse has income from a specific source, the married couple is considered to be a recipient unit. The aged unit focuses on the income of the aged, whether they live with other family members or not, but counts married couples as a single unit with shared resources. Aged-unit statistics exclude the income of nonspouse members and hence may not provide a complete picture of the resources available to the unit. Interpreting aged-unit statistics in the same was as person statistics will emphasize the economic well-being of nonmarried persons relative to that of married persons.
Statistics for the family income of persons are based on the demographic attributes (age, sex, race, and Hispanic origin) of each person. Total income from all family members (related through blood, marriage, or adoption) is treated as another attribute of the person. If any person in the family has income from a specific source, the aged person is considered to be in a recipient family.
Income of the Population 55 or Older and Income of the Aged Chartbook are derived from the Annual Social and Economic Supplement to the Current Population Survey, which is conducted annually by the Census Bureau. Alternative data sources, which have different strengths and weaknesses, are also available. The Survey of Income and Program Participation is strong at measuring small or infrequently received sources of income but is not conducted at regular intervals. SSA's administrative records are more accurate than survey data but lack demographic information and supply data for only a few income sources. This analysis uses income data for 1996 because the SIPP match rate to administrative data declined considerably in the 2001 panel.1
Surveys may differ in many ways, and these differences may affect the statistics produced from the resulting data. The following descriptions of the CPS and SIPP briefly describe some of the broader differences between the surveys that may influence differences in the statistics produced using their data, such as questionnaire detail and the frequency of interviews.
Current Population Survey. The Annual Social and Economic (ASEC) Supplement to the CPS is conducted annually in March.2 The survey, also known as the March Supplement, collects data on income received during the previous calendar year for approximately 35 cash and in-kind sources. Official estimates of income and poverty in the United States are based on this survey (U.S. Census Bureau 2007).
This article uses the 1997 March Supplement to the CPS, which collected data for annual income received in 1996. Persons are designated as "65 or older" if they were at least age 65 as of the interview in March 1997. Aged units classified as "65 or older" are nonmarried persons 65 or older (in March 1997) or married couples in which the husband is 65 or older or the husband is younger than 55 and the wife is 65 or older (in March 1997).
Survey of Income and Program Participation. The SIPP is a longitudinal panel survey conducted by the Census Bureau; panels begin periodically with durations ranging from 2½ to 4 years. The survey is designed to collect data on sources and amounts of income to provide improved statistics on the distribution of income in the United States. Data are collected on approximately 70 cash and in-kind sources of income (U.S. Census Bureau 2005 and 2007).
This article uses the 1996 SIPP panel for income received during 1996; income is reported for individual months and summed over the entire year.3 Persons classified as "65 or older" were at least age 64 as of the interview in March 1996. Aged units classified as "65 or older" were nonmarried persons 64 or older (in March 1996) or married couples in which the husband was 64 or older or the husband was younger than 54 and the wife was 64 or older (in March 1996).4
The Social Security Administration administers two cash benefit programs—Social Security, or Old-Age, Survivors, and Disability Insurance (OASDI), and Supplemental Security Income (SSI), the separate program for low-income aged and disabled persons.
In this article, statistics based on administrative data refer to statistics based on a combination of survey-reported and administrative information. Self-reported data were replaced with Social Security's administrative data for matched survey records.5 Self-reported data from the survey were used for unmatched survey respondents. A slightly greater proportion of observations in the 1996 SIPP were matched with Social Security administrative records (85 percent of those aged 64 or older present in March 1996) than in the March 1997 Supplement of the CPS (77 percent of those aged 65 or older).
Social Security (OASDI). Survey-based Social Security benefit amounts have been replaced with the administrative amount of the Social Security benefit paid to the beneficiary plus the beneficiary's Medicare Part B premium (when the latter is applicable) in both the SIPP and the CPS.6
Supplemental Security Income (SSI). Survey-based SSI payments have also been replaced by administrative payment amounts.7 The process is somewhat more complicated for the SSI program because there are federal and state components to SSI payments. The SIPP and the CPS treat this differently in their questionnaires: the SIPP asks specifically for federal payments and state payments separately; the CPS asks respondents for a single, combined SSI payment amount.
For states with federally administered state SSI payments, both federal and state SSI payment amounts were taken directly from administrative payment data files and were used to replace reported SSI payments for matched observations for both the SIPP and the CPS. For states without federally administered SSI payments, the procedure was different for processing the SIPP and the CPS. For the SIPP, survey-collected federal payments were replaced by administrative data, and survey-based state payments were not changed. For the CPS, the survey-based SSI payment (combined state and federal) amount was replaced by administrative information.8
This analysis compares the estimates of the relative importance of Social Security when changes to the unit of observation and data sources are made singly or in combination. Table 1 highlights the effects of incorporating various combinations of changes. The numbers were calculated according to the methodology currently used in Income of the Population 55 or Older and Income of the Aged Chartbook. Because the results are similar for beneficiaries receiving almost all (90 percent) of their income from Social Security and beneficiaries receiving all of their income from Social Security (see Table 1), only the latter statistics are discussed.
Unit of observation | 90 percent or more from Social Security |
100 percent from Social Security |
||
---|---|---|---|---|
CPS | SIPP | CPS | SIPP | |
Aged unit | ||||
Survey data only | 30.4 | 19.9 | 17.9 | 8.5 |
Survey and administrative data | 30.0 | 20.5 | 17.3 | 8.4 |
Family income of persons | ||||
Survey data only | 22.3 | 13.2 | 11.4 | 4.8 |
Survey and administrative data | 22.2 | 13.7 | 11.2 | 4.8 |
SOURCES: Author's calculations using the March 1997 Current Population Survey, the 1996 Survey of Income and Program Participation, and Social Security administrative records. | ||||
NOTE: CPS = Current Population Survey; SIPP = Survey of Income and Program Participation. |
Taken alone, shifting the unit of observation from aged units causes the estimate of 17.9 percent of beneficiary aged units receiving all of their income from Social Security to change to 11.4 percent of the elderly in beneficiary families receiving all of their family income from Social Security. There are two reasons why the estimate for aged units is larger: (1) the aged-unit statistic excludes nonspouse family income, which is included in the measure for persons with family income, and (2) an aged unit may be a nonmarried person or a married couple, in which case one unit consists of two persons. If an aged-unit statistic is interpreted as being based on persons, it places extra emphasis on nonmarried persons, who tend to be less well off economically;9 in this case it corresponds to a higher percentage of beneficiaries/beneficiary units receiving all of their income from Social Security.10
Switching only the survey from the CPS to the SIPP causes the estimate of elderly beneficiary aged units receiving all of their income from Social Security to drop by more than half, from 17.9 percent to 8.5 percent. This result is due in part to the SIPP's frequent interviews and questionnaire, which were designed to pick up small or infrequently received sources of income. A more in-depth look at the differences in income reported in the two surveys suggests that underreporting of the receipt of asset income, and possibly pensions, in the CPS leads to discrepancy in the two estimates.11
Social Security's administrative data from the OASDI and SSI programs can be used to replace the income amounts reported by respondents in surveys such as the CPS and the SIPP. However, the impact of using administrative data for program income amounts has a relatively small effect when compared with either the effects of using statistics with various definitions of the unit of observation or various surveys chosen for analysis. The percentage of all beneficiary units receiving all of their income from Social Security decreases slightly, from 17.9 percent using only reported survey data in the CPS to 17.3 percent when administrative data are matched to the CPS to correct program income amounts. This decrease in the CPS estimates because of the use of administrative data indicates that either some nonbeneficiary units are reporting Social Security benefits that they have not received, some units receiving SSI are not reporting SSI, or both.12 The change in the estimates from switching reported data for administrative data in the CPS (0.6 percentage points) is smaller than the change that occurs when switching from CPS-reported data to SIPP-reported data (9.4 percentage points) partly because administrative records are used in place of reported data only for Social Security benefits and SSI, not for all sources of income.
Depending on the combination of choices of data sources and units of observation, 4.8 percent to 17.9 percent of elderly beneficiaries are estimated to be completely dependent on Social Security for their income. The lowest figure, 4.8 percent, was calculated for the person based on family income using the SIPP. The highest figure, 17.9 percent, was calculated for the aged unit using only reported data from the CPS—the methodology currently used for the statistics published in both Income of the Population 55 or Older and Income of the Aged Chartbook. The majority of this difference is attributable to use of the SIPP rather than the CPS because the SIPP better captures sources of income other than Social Security, particularly asset income.
The purpose of these exercises was to demonstrate how key measurement choices affect our understanding of the resources available to the elderly. Depending on the unit of observation and underlying data, the percentage of elderly beneficiaries receiving all of their income from Social Security ranges from 4.8 percent to 17.9 percent. Given this broad range of results, a natural next question is whether any of the measurement options described in this article should be adopted when measuring the relative importance of Social Security for elderly beneficiaries. As expected, there are tradeoffs associated with each option.
In the case of choosing a unit of observation, the cost of changing from one unit to another is the discontinuity it would introduce into a nearly 30-year time series. Further, some researchers want to include only income of the elderly marital unit, while others want information on income of the entire family. To better accommodate the needs of users of Income of the Population 55 or Older, SSA has revised and expanded the publication beginning with an expanded 2004 edition that will be available online only. The new structure provides more information on aged units and the family income of elderly persons. An electronic version of the new publication for 2004 data is expected to be available in 2008, and electronic and paper copies of the new format for 2006 data are also expected to be available in 2008.
The selection of survey is highly influential on the results presented in this article, but there are tradeoffs among data sources. The CPS is timely and is used by other agencies for official statistics, but it underestimates whether certain income types were received. The SIPP does a better job of ascertaining whether or not certain income types were received, but it may be more likely to underestimate the amounts of that income.13 Also, new SIPP panels have not started at regular intervals, which makes it difficult to produce a consistent time-series publication.14 Other surveys are available but have not been matched to Social Security administrative records. Some surveys provide data on pension withdrawals, savings, and wealth not captured in the CPS, which would provide a broader picture of economic well-being, but these data are not necessarily available at individual and family levels. In the case of Income of the Population 55 or Older and Income of the Aged Chartbook, availability of data at regular intervals and the ability to work with person- and family-level data have led to the selection of the CPS as the underlying source of data.15
The Social Security Administration has agreements in place to match its administrative records to the SIPP and CPS, but only after a significant time lag and not to all persons in the surveys.16 The choice here is one of timeliness versus accuracy. Even if matched data were to be used, the question then would be whether to use only matched respondents and data or matched data with reported data for respondents unable to be matched. Using only matched respondents would introduce bias if matched respondents differ systematically in any way from unmatched respondents; either choice may lead to false movements in statistics over time if the proportions of the sample that are matched change over time. Although it may not be desirable to publish a statistical series using administrative data, these data are useful for assessing the accuracy of self-reported data.
Another issue that has not been addressed in this article is the definition of income, which can differ by agency and usage. The definition of income in both Income of the Population 55 or Older and Income of the Aged Chartbook is the Census Bureau definition of total money income. The definition of total money income excludes capital gains, 401(k) and individual retirement account (IRA) withdrawals that are not regularly scheduled, and noncash benefits. These items also provide resources for the elderly, which are not captured under SSA's current measure of the relative importance of Social Security. The appendix addresses some of the effects of including noncash benefits as income. In addition, SSA is currently working with Census to gather and analyze data on nonannuitized withdrawals from pensions and IRAs.
Regardless of any changes that may be implemented in Income of the Population 55 or Older or Income of the Aged, it is important to keep in mind the issues outlined here when interpreting the data.
Just as family income may be considered a resource, noncash benefits like energy assistance or food stamps may also be a resource for many of the elderly. Noncash benefits are not included as income for official statistics such as the poverty rate, but they can be an important resource to recipients. Among its list of recommendations for changes to the measurement of poverty, the National Research Council (1995) has suggested that noncash benefits be considered when determining the resources available to an individual or family. A workshop on experimental poverty measures by the National Research Council (2005, 4) reported that
the omission of these [taxes and near-money government benefits] from the official definition of income has become increasingly serious in recent years because government transfers are now concentrated in benefits that are not considered part of families' gross cash income—such as housing subsidies, child care subsidies, and the EITC—rather than cash welfare assistance. The unfortunate result is that the current official poverty measure no longer captures either people's perceptions of poverty or the effect of various policies on poverty.
Various noncash benefit programs provide food, energy, housing, medical care, transportation, or child care to persons who qualify.
Inclusion of the value of noncash benefits as income for poverty measures has been controversial and difficult to measure.17 Regardless of the dollar values assigned to food stamps or the like, noncash benefits are a resource. This Appendix presents data on the noncash benefits received by aged units (or persons) receiving all of their income from Social Security and recalculates reliance on Social Security with noncash benefits counted as a resource. These calculations do not require dollar values to be assigned to the noncash benefits received.
The analysis below incorporates food, energy, and housing benefits for 1996. Food benefits in the Current Population Survey comprise Food Stamps; in the SIPP, they comprise the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and Food Stamps.18 Energy assistance in both surveys includes vouchers, direct payments to utilities, and checks to the household for energy. Housing assistance in both surveys is indicated by a rent subsidy or residence in public housing. Medical benefits in the form of Medicare Part A and Medicaid were not included.19 Transportation assistance data were unavailable in the 1996 SIPP panel until 1998 (wave 8), and child care benefits are unlikely to be a significant source of noncash income for the elderly.
According to the SIPP, 24.6 percent of beneficiary aged units reporting all of their cash income from Social Security received housing, energy, or food assistance in 1996, and 5.3 percent of elderly aged units reporting all of their cash income from Social Security received assistance from more than one source (Table A-1). Receipt of noncash benefits is similar in the CPS, with 20.9 percent of beneficiary aged units reporting all of their cash income from Social Security receiving housing, energy, or food assistance and 4.8 percent receiving assistance from more than one source.
Type of noncash benefit |
All aged units | Beneficiary aged units reporting all income from Social Security |
||
---|---|---|---|---|
CPS | SIPP | CPS | SIPP | |
Energy | 3.3 | 3.4 | 5.3 | 6.6 |
Food | 5.6 | 5.5 | 10.5 | 11.8 |
Housing | 5.5 | 5.3 | 10.1 | 11.8 |
SOURCES: Author's calculations using the March 1997 Current Population Survey and the 1996 Survey of Income and Program Participation. | ||||
NOTE: CPS = Current Population Survey; SIPP = Survey of Income and Program Participation. |
As expected, the inclusion of noncash benefits leads to a decline in the percentage of aged units reporting complete dependence on Social Security. In the SIPP, inclusion of food, energy, and housing assistance results in a drop of 2.1 percentage points in 100 percent reliance on Social Security benefits; in the CPS, inclusion of these benefits causes a decrease of 3.7 percentage points (Table A-2).
Type of noncash benefit | CPS | SIPP |
---|---|---|
None | 17.9 | 8.5 |
Energy | 16.9 | 7.9 |
Food | 16.0 | 7.5 |
Housing | 16.1 | 7.5 |
Energy and food | 15.4 | 7.1 |
All of the above | 14.2 | 6.4 |
SOURCES: Author's calculations using the March 1997 Current Population Survey and the 1996 Survey of Income and Program Participation. | ||
NOTE: CPS = Current Population Survey; SIPP = Survey of Income and Program Participation. |
Table A-3 highlights the effect of including the value of noncash benefits with the combinations of changes in the article. The estimates were calculated according to the methodology currently used in Income of the Population 55 or Older and Income of the Aged Chartbook.
As with the poverty measure, exclusion of the value of noncash benefits understates the resources available to an individual and thus overstates reliance on Social Security benefits. However, data on the values of noncash benefits are not necessarily available in all surveys.
Unit of observation | Excluding noncash benefits | Including noncash benefits | ||
---|---|---|---|---|
CPS | SIPP | CPS | SIPP | |
Aged unit | ||||
Survey data only | 17.9 | 8.5 | 14.2 | 6.4 |
Survey and administrative data | 17.3 | 8.4 | 13.8 | 6.4 |
Person with family income | ||||
Survey data only | 11.4 | 4.8 | 9.0 | 3.4 |
Survey and administrative data | 11.2 | 4.8 | 8.9 | 3.5 |
SOURCES: Author's calculations using the March 1997 Current Population Survey, the 1996 Survey of Income and Program Participation, and Social Security administrative records. | ||||
NOTE: CPS = Current Population Survey; SIPP = Survey of Income and Program Participation. |