Spending patterns of families receiving public assistance.
Expenditures of families receiving assistance vary widely, depending on the employment status of adults, the marital status of the household head, and the presence of childrenHow do households that receive public assistance spend their income? And, how are employment and demographic characteristics of these families related to their expenditure levels?
Data from the Bureau of Labor Statistics Consumer Expenditure Survey can provide some answers to these and other questions about the demographic characteristics of consumer units, or households, receiving public assistance and the ways in which they allocate their expenditure dollar among categories of goods and services. This article presents an analysis of these data, with special emphasis on employment status and family structure as possible influences on spending patterns. Tables 1 through 3 show the demographic characteristics, types of assistance received, and average annual expenditures of households receiving any of the six forms of assistance for which data are available.
[TABULAR 1-3 DATA OMITTED]
General household characteristics
Households on public assistance average just over three persons, with about 1-1/4 children under 18. (See table 1.) The reference person is about 46 years old, slightly younger than the average of 47.9 years for all other households. This runs counter to the perception that families receiving public assistance are significantly younger than the population at large. While it is true that assistance income in the form of Aid to Families with Dependent Children goes primarily to families who are younger than the average family in the expenditure survey sample, sources such as Supplemental Security Income are targeted toward the elderly. Many of the households receiving public assistance have working members and own a vehicle: the average number of earners is 0.9, while the average number of vehicles is 1.
About two-thirds of the households rent their living quarters, while the roughly 30 percent who are homeowners are equally split between those paying off a mortgage and those with no mortgage. For every household with a reference person aged 65 or older, there are almost four with a younger reference person. About 60 percent of the households are headed by women, while 3 in 10 have a black reference person. Husband-wife households with any combination of additional members comprise just over one-third of the sample of households receiving some form of public assistance. Single-person units make up one-quarter of the sample, single-parent units another one-fifth, and all other types of units the remaining fifth.
Households can receive one or more forms of public assistance. (See table 2.) A wide range of program combinations are reported by households in the sample, with no one combination dominant. Among the possible combinations of public assistance received, the most common is medicaid only, reported by just under 15 percent of sample households. The combination of welfare, food stamps, and medicaid is the next most frequent at 11 percent. Households receiving food stamps and medicaid and those receiving food stamps only each represent about 9 percent of the sample. No other combination was reported by more than about 6 percent of the households.
Households receiving public assistance spent $15,304 on average during 1992-93. (See table 3.) About three-fifths of their total spending was allocated to food and housing. Transportation took up the next largest share, about 15 percent of the total. The shares of total spending apportioned to entertainment, apparel, health care, and personal insurance and pensions hover between 4 and 5percent. The remaining major categories--alcoholic beverages, tobacco, personal care, reading, education, cash contributions, and other miscellaneous expenditures--combine to make up the remainder, just over 6 percent.
Spending patterns of subgroups
By work status. The expenditure patterns and demographic characteristics of households that receive public assistance in which there are no working members are compared with those having one or more working members also receiving such income. For purposes of this comparison, each group is limited to households with a reference person under age 65. Households headed by reference persons aged 65 or older are less likely to be in the work force due to retirement, illness, and other factors. As such, these households are found primarily in the nonworking group, and their distinctive spending patterns would unduly affect the distribution of expenditures of that group, particularly for health care and housing.
The definition of a "working" member of a household is primarily based On, but is not identical to, the definition used by BLS to describe the "working poor." A household member who is over 15 years of age and who has been employed full time or part time for at least 27 weeks over the previous 12 months is considered "working." Whether he or she is paid for working is not relevant to this definition.
The nonworking group provided data from 2,275 interviews, while the working group completed 2,669 interviews. (See table 1.) The demographic characteristics of the groups demonstrate interesting similarities and differences. The average age of the reference person is almost identical--38.9 years for the nonworking group and 38.3 years for the working. Households in the working group average 3.8 persons, almost 1 person more than those in the nonworking group (2.9 persons). Yet the mean number of children under 18 is about the same in both groups, at 1.5 in the nonworking and 1.6 in the working.
By design of the groups, the average number of earners is markedly different between the groups. (See the box note on pages 26-27, for information on the difference between workers and earners, as defined for this study.) Perhaps as noteworthy as the magnitude of the difference--an average of 0.3 earners in the nonworking group versus 1.7 earners in the working group--is the fact that so many of the households in the working group have more than one earner.
A female reference person is found in almost 70 percent of the households in the nonworking group, while the working group shows a slight tilt toward a male reference person (53 percent to 47 percent). The ratio of white and other to black reference persons is 3 to 2 in the nonworking group and rises to 3 to 1 in the working group. Renters predominate among the households in the nonworking group, with 80 percent reporting such living arrangements. Renting their living quarters is also the most common practice among working group units (62 percent); however, more than one-third of the households in this group are homeowners.
The higher percentage of female-headed households is reflected in the distribution of family types in the nonworking group. Two in five households are made up of a single parent with at least one child under 18. An additional 25 percent are single-person units. Fewer than 20 percent contain a husband and wife. By contrast, the composition of households in the working group varies widely. More than one-half of the working group households are headed by a husband and wife. Single-parent units with at least one child under 18 are, marginally, the largest remaining fraction of this group at 16 percent. (The "All households" category is technically the largest, but this category consists of a collection of family types generally containing too few units to stand alone for tabulation or analysis.)
There also are differences among the most frequent combinations of forms of public assistance that each group collects. (See table 2.) Within the nonworking group, the combination of welfare, food stamps, and medicaid tops the list, with just under 19 percent of the units reporting it. About 9 percent of households in the working group receive that combination, ranking it fourth in frequency. On the other hand, medicaid only is received by the largest portion of the working group households with just over one-fifth so reporting, compared with fewer than 5 percent in the nonworking group.
Two other combinations of assistance round out the top three among the nonworking group. Food stamps and medicaid are reported by more than 8 percent of this group, and a combination of welfare, government housing support other than public housing, food stamps, and medicaid is received by 7.5 percent. Food stamps plus medicaid is also the second most frequent combination for working group households, with more than 11 percent reporting it. A package consisting of welfare, government housing support other than public housing, food stamps, and medicaid is reported by only 2 percent of the working group. In this group, receipt of food stamps only holds third place among the assistance combinations received, with 11 percent reporting it. Assistance consisting of food stamps only ranks fourth, at 7 percent, in the nonworking group. The 10 most common forms of assistance among households in the nonworking group accounted for less than 80 percent of all reports, while for the working group, almost 90 percent of units were represented by the 10 most frequent forms of assistance received.
The effect of additional earners can be seen in the striking disparity in total expenditures between the groups. (See table 3.) Households in the working group report average total expenditures more than double those of units in the nonworking group. The allocation of the expenditure dollar among major categories shows that the nonworking group, with lower overall expenditures, apportions a greater share to necessities, such as food and housing, than does the working group. Over 43 percent of total spending goes to housing and over 28 percent to food for households with no working members. In contrast, the working group spends only 34 cents of every expenditure dollar on housing and 19 cents on food. The residual 18 percent, which the nonworking group devotes to food and housing, is allocated to the transportation and personal insurance and pensions categories by the working group. Expenditure shares for other major categories of spending are very similar for the two groups, varying by little more than l percentage point in most cases.
Households in the working group devote an expenditure share to transportation that is twice as large as that reported by their nonworking counterparts. They direct 19 percent of total expenditures to transportation, with over 9 percent going to vehicle purchases, including finance charges, alone. Only 9.5 percent of total spending is allocated to transportation in the nonworking group, and less than 3 percent goes to vehicle purchases, including finance charges. (Finance charges, a component of "other vehicle expenses," average $177 (2.0 percent) for the working group and $22 (0.2 percent) for the nonworking group.)
Personal insurance and pensions occupy a minor position among expenditures of the nonworking group, registering a share of just over I percent. Not so among the working group, for whom almost 8 percent of total spending is found in this category. The difference in share between the groups could be explained almost entirely by the presence of workers in one group and the absence of workers in the other, because the lion's share of expenditures in this category is directed to retirement and pensions. Typically, these are payroll deductions for Social Security and pension plan contributions--outlays contingent on one's being employed.
By family structure. It has been suggested that the composition or structure of a family affects its poverty status. Comparing single-parent households, irrespective of the sex of the parent, receiving public assistance with dual-parent units receiving such assistance permits an assessment of the ways in which spending patterns differ between these groups. Differences in socioeconomic characteristics are also presented.
Households in these groups consist only of parents and their own children. No extended family members, such as grandparents, aunts, or cousins, nor unrelated individuals are included. While the preponderance of single-parent units are headed by women, this group includes single-parent households headed by men as well.
These groups are smaller than the working-nonworking groups from the previous section, numbering 1,386 interviews in the single-parent category and 1,123 interviews in the dual parent group. (See table 1.) According to the demographic breakdown, single-parent households are headed by much younger reference persons (31.9 years) than are dual-parent units (37.5 years). Dual-parent households average one more person than single-parent units (4.2 persons compared with 3.2 ). However, single-parent households average slightly more children under 18 (2.1) than their dual-parent counterparts (just under 2). The presence of two parents suggests that the average number of earners in the dual-parent group would be larger than that in the single-parent group. This is indeed the case, with dual-parent units averaging about 1.4 earners, and those with single parents, about 0.6 earners.
Single-parent households are overwhelmingly headed by women, with 95 percent of the group so reporting. About 46 percent of the reference persons are black. Renting is by far the major housing tenure of single-parent households:--86 percent rent their living quarters, while fewer than 10 percent are homeowners. Three in five dual-parent households also rent, although 1 in 4 in this group is a homeowner with a mortgage and another 10 percent own their homes outright. Younger children tend to be the rule in terms of the age distribution of own children in the households in these groups. In about 30 percent of the units in both groups, all children are under 6 years of age. In addition, in about 1 in 5 units in both groups, the oldest child is between 6 and 11 years of age and there is at least one other child under age 6. Nineteen percent of the single-parent group have at least two children, with the oldest aged 12 to 17, and at least one under age 12. Fifteen percent of the dual-parent group is similarly composed.
Welfare, food stamps, and medicaid are common components of the three most frequently reported combinations of assistance received by single-parent households. (See table 2.) About 24 percent of this group receive these three alone; about 12 percent receive them in tandem with government housing support other than public housing; for another 9 percent, they are combined with public housing. Only the first combination--welfare, food stamps, and medicaid--applies to a significant portion of dual-parent households, ranking third at about 12 percent. The combinations including public housing and government housing support other than public housing are relatively rare among dual-parent units; fewer than 1 percent of this group report the former, and about 2.5 percent the latter.
More than 19 percent of dual-parent households receive medicaid only from among the possible combinations of assistance; fewer than 4 percent of single-parent units receive similar assistance. Food stamps plus medicaid is the next most frequent combination received by households in the dual-parent group, with more than 16 percent so reporting. Only 7 percent of single-parent units get such a combination.
Not only does the dual-parent group average more than twice as many earners as the single-parent group, but it also reports almost twice as much in total expenditures--$22,281, compared with $11,948. (See table 3.) As in the working nonworking comparison, single-parent households, with lower total expenditures, direct a larger share of spending to food and housing than do their dual-parent counterparts. In fact, the shares devoted to these expenditure categories by the single-parent group closely match the shares devoted by the nonworking group discussed earlier. About 28 percent of total spending goes toward food and 41 percent goes toward housing for this group. In like manner, the dual-parent group allocates percentages of total tending to food and housing that are similar to those reported above for the working group--20 percent and 34 percent, respectively.
While dual-parent households allocate smaller budget shares to food and housing, they spend proportionately more (19 percent) on transportation than do their single-parent counterparts (just over 10 percent). Larger shares devoted to the purchase of cars and trucks, including finance charges, account for about 60 percent of the difference in transportation share between the groups. (Finance charges averaged $183 (0.8 percent) for dual-parent households and $34 (0.3 percent) for single-parent households.) Purchases of used cars and trucks account for about 40 percent of the total difference. Other vehicle operating expenses--insurance, gasoline and motor oil, and maintenance and repair--make up most of the remaining 40-percent difference in share.
On the other hand, single-parent households allot a greater expenditure share to apparel than do dual-parent units. Over 7 percent of their total spending goes for clothes, footwear, other apparel products, and apparel services, versus under 5 percent in the dual-parent group. Much of the difference in share is concentrated in apparel for women and girls. The share of spending for men's and boys' apparel is virtually identical in the two groups. Expenditures on apparel for women and girls typically outstrip such expenditures for men and boys as a share of total spending in the expenditure survey. In addition, upkeep of women's apparel usually costs more than that of men's apparel.
As in the working-nonworking comparison, shares for personal insurance and pensions differ between single-parent and dual-parent households. With more earners in the dual-parent group, the share of spending allocated to retirement and pensions, mainly deductions for Social Security and pension plan contributions, tops 6 percent, compared with just over 2 percent in the single-parent group. Shares in the remaining expenditure categories do not exhibit notable variation between the groups.
EXAMINATION OF THE SPENDING BEHAVIOR and demographic characteristics of families identified by their receipt of selected forms of public assistance reveals this group to be a heterogeneous one. Families who have at least one working member not only spend more than their counterparts with no working members, but allocate their expenditures differently among major item categories. The composition of families receiving some form of assistance also has an impact on the levels and distribution of expenditures. In particular, the spending profile of dual-parent families does not look like that of single-parent families.
Further research is indicated to determine what causes these families to make the budget choices that they do. Are expenditure levels linked to the number of earners or number of children in the family? Does the level of education or type of assistance received have any influence? Regression analysis and analysis of variance suggest themselves as ways to answer these and other questions.
RELATED ARTICLE: Scope and method of the study
The sample
The focus of this study is consumer units, or households, that participate in a government program which provides general income maintenance or specific assistance for food, housing, or medical care, rather than "poor families," as defined on income or expenditure criteria. The Consumer Expenditure Survey data base contains variables positively identifying households participating in many of these programs. A sample consisting of households with incomes under a poverty line would exclude some units with higher incomes that participate in the programs. The processing of income variables in the expenditure survey also introduces limitations on the use of any income threshold in this analysis: because values are not imputed for missing responses to income questions, it is not possible to get a complete accounting of income for all households in the expenditure survey data base. Consequently, one cannot determine conclusively whether some households are below or above a poverty threshold. While it is possible to create a poverty line estimate based on expenditures that is not subject to this problem, neither a sample of the "income-poor" nor one of the "expenditure-poor" households would necessarily include all units receiving program assistance. To avoid these complications, it was decided to simplify the sample selection task by choosing households that definitively reported receiving program assistance.
The sample constructed for this study consists of 6,307 interviews from qualifying households. A consumer unit can qualify for inclusion in the sample in any of three ways. The first is by reporting the amount of assistance received from any of the following programs: welfare or public assistance, defined as including Aid to Families with Dependent Children, general assistance, emergency assistance, Cuban Refugee Assistance, Indian assistance, Alaska Longevity Pay, or job training grants, such as for Job Corps participation; Supplemental Security Income; or food stamps. The second is by residing in public housing or receiving housing subsidy payments from programs such as the Federal Housing Administration (FHA) Section 236 program, the FHA rent supplement program, the Section 202 program, and the "below market interest rate" program. The final way is by reporting any member of the household as enrolled in medicaid.
The programs itemized above are not an exhaustive list of assistance programs available to American households. For example, expenditure survey respondents are not asked whether they benefit from the Earned Income Tax Credit. Participation in needs-based food assistance programs such as the school lunch and school breakfast programs, which provide free or reduced-price meals, the summer food service program for children, the child and adult care food program, and the special milk program is not checked in the expenditure survey instruments. No questions are included to cover assistance programs geared to veterans, such as medicare for veterans without service-connected disability; pensions for needy veterans, dependents, and survivors; and dependency, indemnity, and death compensation for parents of veterans.
Data source
The source of expenditure, income, and demographic characteristic data for this study is the Quarterly Interview Survey component of the Consumer Expenditure Survey. The objective of the Quarterly Interview Survey is to obtain 1 year's worth of expenditure data from each sample household. To achieve this, an extensive questionnaire is administered by an interviewer from the Bureau of the Census, which conducts the Consumer Expenditure Survey under contract to BLS. Each sample household undergoes five interviews. In the first, demographic characteristic data for each household member are collected, along with characteristics of the housing unit and inventory data on major appliances and vehicles. Expenditure data are collected over a 1-month recall period. These data are used for bounding purposes only, and not for estimation. In each of the remaining four interviews, conducted at 3-month intervals, households report expenditures that have been made since the last interview. During the second and fifth interviews, data on sources and levels of income are collected for household members and for the unit as a whole for the 12-month period prior to the interview. Information on assets and liabilities over the previous year also is collected during the fifth interview. Overall, about 90 to 95 percent of all expenditure; are covered by the Interview Survey instrument. (Expenditures for postage, housekeeping supplies, personal car' products, and nonprescription drugs are among the 5 per cent not included in the Interview Survey.)
Households are interviewed using a rotational sampling procedure. Each month, new households begin the interviewing cycle, replacing units that have completed their participation. The Interview Survey is designed so that 20 percent of the sample is replaced every 3 months.
Reference period
While BLS has carried out some form of expenditure survey approximately every 10 years since the turn of the century, the survey in its current form has operated continuously since 1980. The analysis in the accompanying article covers data from interviews completed from January 1992 through March 1994. Selecting these interviews essentially provides data from calendar years 1992 and 1993. Two years' worth of data is considered desirable if one is to obtain a sample of qualifying households large enough for analysis. At the same time, expenditures reported over the 1992-93 period do not manifest significant shifts in spending behavior that would render combining data from the 2 years suspect.
Variables
The data base created for this analysis consists of a wide range of demographic, income, and expenditure variables. The demographic variables include number of persons in the household; age, race, sex, and educational attainment of the reference person (usually the household head); housing tenure; number of children under age 18; composition of the household (classified by the relationship of other household members to the reference person and by the age of the oldest "own" child); age composition of own children; number of persons over age 64; number of workers; number of earners; number of owned vehicles; and types of assistance received. These variables reflect the characteristics of the household at the time of each interview.
Expenditure variables in this analysis correspond to categories presented in the standard tables of Consumer Expenditure bulletins and reports published periodically by BLS. These variables identify total expenditures and expenditures for food, food at home, food away from home, housing, shelter, shelter for owned dwellings, shelter for rented dwellings, shelter in other lodging, utilities, other household expenses, apparel, transportation, purchases of new cars and trucks, purchases of used cars and trucks, gasoline and motor oil, other vehicle expenses, health care, health insurance, medical services, prescription drugs, entertainment, and personal insurance and pensions. Variables representing alcoholic beverages, personal care, reading, education, tobacco, miscellaneous expenditures, and cash contributions are combined into a category of "all other expenses."
Four of the variables above deserve some elaboration. The first is "earner," which is more loosely defined than is "worker," for purposes of this article. To be an earner, one had to be over 14 years old and have worked any number of weeks, for pay, over the preceeding 12 months. Second is the category of shelter for owned dwellings, which includes mortgage interest and charges, property taxes, maintenance and repair expenses, insurance, and management fees. The portion of mortgage payments that is principal is not included, being considered repayment of a loan, and thus a reduction of liabilities. And finally, the variables for purchases of new and used automobiles and trucks contain the net outlay for the purchase of a vehicle--that is, the amount paid after trade-in allowance and any costs paid by an employer. Finance charges paid on a vehicle loan are included in "other vehicle expenses."
Expenditure variables aggregate all purchases made for their component items during the 3-month reference period covered by the interview. Expenditure variables that contain 3-month estimates are multiplied by 4 to annualize them prior to computation of weighted means.
Weights
In this study, each interview is treated as an independent observation in the data set. Each household in the sample is assigned a population weight such that the sum of the weights for interviews completed over a calendar quarter equals the national population. These weights can change from quarter to quarter for any household, depending on the characteristics of the units interviewed during that quarter. The means and percent distributions presented in this article are calculated using these weights.
It is important to note that mean estimates are often inadequate for use in comparing expenditure patterns of different subgroups of this sample, because large differences in expenditure levels between subgroups tend to hide variations in the allocation of the expenditure dollar. Consequently, mean expenditure estimates have been converted to shares of total spending to capture these effects.
William D. Passero is an economist in the Division of Consumer Expenditure Surveys, Bureau of Labor Statistics.
 Printer friendly
 Cite/link
 Email
 Feedback
| |
| Author: | Passero, William D. |
|---|---|
| Publication: | Monthly Labor Review |
| Date: | Apr 1, 1996 |
| Words: | 4470 |
| Previous Article: | The erosion of police and firefighter wage parity. |
| Next Article: | Access to customers. |
| Topics: | |