Planning ahead: Consumer expenditure patterns in retirement

Monthly Labor Review

Volumn 125, Issue 7, 2002, Pages 38-54

  Paulin, Geoffrey D ^a   Duly, Abby L ^a  

^a BUREAU OF LABOR STATISTICS   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 11244318206     PISSN: 00981818     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (33)

1. See 65+ in the United States, Current Population Reports, Special Studies (U.S. Bureau of the Census, 1996), pp 23-190.
2. Geoffrey D. Paulin, "Expenditure patterns of older Americans, 1984-97," Monthly Labor Review, May 2000, pp. 3-28.
3. Rose M. Rubin and Kenneth Koelin, "Elderly and nonelderly expenditures on necessities in the 1980s," Monthly Labor Review, September 1996, pp. 24-31.
4. Mohammed Abdel-Ghany and Deanna L. Sharpe, "Consumption Patterns Among the Young-Old and Old-Old," Journal of Consumer Affairs, Summer 1997, pp. 90-112.
5. The reference person is the first person mentioned by the survey respondent when asked: "Start with the name of the person or one of the persons who owns or rents this home."
6. Nancy E. Schwenk, "Trends in the Economic Status of Retired People," Family Economic Review, 1994, 7(2), pp. 19-27.
7. Ibid., pp. 24-25.
8. Frankie N. Schwenk, "A Comparison of Households Headed by Persons 55 to 65 Years of Age: Retired and Employed," Family Economic Review, 1990, 3(3), pp. 19-25.
9. Ibid., pp. 22, 24.
10. A consumer unit is defined as members of a household related by blood, marriage, adoption, or other legal arrangement; a single person living alone or sharing a household with others but who is financially independent; or two or more persons living together who share responsibility for at least two out of three major types of expenses - food, housing, and other expenses. In this article, consumer unit and household are used interchangeably.
11. Thomas Moehrle, "Expenditure patterns of the elderly: workers and nonworkers," Monthly Labor Review, May 1990, pp. 34-41.
12. Ibid., p. 34.
13. Ibid., p. 36.
14. Rose M. Rubin and Michael Nieswiadomy, "Expenditure patterns of retired and nonretired persons," Monthly Labor Review, April 1994, pp. 10-21.
15. A complete income reporter is a consumer unit that provides values for at least one of the major sources of its income, such as wages and salaries, self-employment income, and Social Security income. A complete reporter may not provide a full accounting of all income from all sources, however.
16. Rubin and Nieswiadomy, "Expenditure patterns...," p. 36.
17. The 1996-97 Consumer Expenditure Survey 2-year report notes that the "Interview survey collects detailed data on an estimated 60 to 70 percent of total household expenditures. In addition, global estimates - that is, expense patterns for a 3-month period - are obtained for food and other select items. These global estimates account for an additional 20 to 25 percent of total expenditures." Source: Bureau of Labor Statistics, Consumer Expenditure Survey, 1996-97, Report 935 (U.S. Department of Labor, September 1999), p. 256.
18. It is important to note that some retirees in our sample may be "retired due to disability." However, in the Consumer Expenditure Survey, there is no way to identify those who are both retired and disabled. Respondents may select only one of the categories - "retired" or "not working due to disability."
19. The other possible occupational statuses for the spouse are "working without pay" or "not working" because they are either going to school or doing something else (that is, not working for a reason not already described).
20. Beth Harrison highlights the differences in expenditure levels and shares between these two age groups from the 1984 Consumer Expenditure Interview Survey, finding them to be distinct in most major categories. (See "Spending patterns of older persons revealed in expenditure survey," Monthly Labor Review, October 1986, p. 15-17.) In addition, in a study following up on Harrison's findings, Pamela Hitschler (p. 3) finds that "consumer units in the younger group spent, on average, a significantly larger amount on every major expenditure category except housing and healthcare in both years [1980 and 1990 are compared]." ("Spending by older consumers: 1980 and 1990 compared," Monthly Labor Review, May 1993, pp. 3-13.)
21. Beth Harrison highlights the differences in expenditure levels and shares between these two age groups from the 1984 Consumer Expenditure Interview Survey, finding them to be distinct in most major categories. (See "Spending patterns of older persons revealed in expenditure survey," Monthly Labor Review, October 1986, p. 15-17.) In addition, in a study following up on Harrison's findings, Pamela Hitschler (p. 3) finds that "consumer units in the younger group spent, on average, a significantly larger amount on every major expenditure category except housing and healthcare in both years [1980 and 1990 are compared]." ("Spending by older consumers: 1980 and 1990 compared," Monthly Labor Review, May 1993, pp. 3-13.)
22. This "age effect" is assumed to include differences by age in health status. Although health status can be an important influence on the expenditures of older consumers, there are no concrete measures of health status available in the Consumer Expenditure Survey.
23. As described previously, the definition of "preretired" and "retired" depends on the occupational status of the reference person and spouse, in the case of married couples. It is possible that one of the parents owns the home, and is therefore the reference person, but the child moves back in with them to provide economic support.
24. Even eliminating families with children does not guarantee that the couple is childless. College students, when living in university-sponsored housing, are considered to be separate consumer units from their parents. Additionally, children may have reached the age of majority, and may have moved out to establish consumer units of their own. However, the survey does not collect information in such a way as to allow the selection of singles or couples who do not have children at all. Therefore, although it is possible that some of these families purchase items for their children that those without children would not, it is not possible to identify those families with the possibility of such additional spending.
25. Recall that the definition of retirement in this study is based on the self-reported occupation of the reference person. Thus, it is possible to retire from one's life-long work and to pursue other avenues of employment. The "retiree" may choose to work for pay in a field that was previously a hobby, or perhaps may seek a low-wage job to keep active, but not for income, per se.
26. While it is true that some "expenditures," such as mortgage principal, may be considered an "investment" in some cases, most homeowners do not own their home solely for investment purposes, as they might a stock or bond; they also consume the housing services the home provides. Similarly, some consumers may own life insurance policies that pay annuities at some point; however, the policy is not merely a savings vehicle; it is primarily a purchase of protection of one's estate in case of unexpected events.
27. Calculated by dividing the average expenditure for the whole group ($372 for preretirees, and $224 for retirees) by the percent reporting, shown in table 6 (89.6 percent for preretirees, and 73.4 percent for retirees).
28. See the Technical Notes section for a detailed explanation of the regression methodology, including the model specifications.
29. See Milton Friedman, A Theory of the Consumption Function (Princeton, NJ: Princeton University Press, 1957).
30. There are also empirical reasons for using "permanent" income in this case. Respondents do not always provide information on "current" income, and even those who do may not provide a full accounting of all income from all sources. Furthermore, data regarding assets and liabilities are only collected on a limited basis in the Interview survey. However, the primary goal of the Interview Survey is to collect expenditures.
31. One possible solution is to use four complete quarters for each consumer unit, rather than treat each quarter independently as is done in this article. However, even this solution does not provide a balanced treatment of medical expenditures and reimbursements. For example, a reimbursement reported in the second interview (the first interview during which these data are collected) will have no matching expenditure because that expense would have been incurred by the consumer unit prior to its participation in the survey. Likewise, a medical expenditure reported in the fifth and final interview may very well be reimbursed afterward, when the consumer unit is no longer a survey participant. There is no way to capture these prior expenses or future reimbursements.
32. Because the logit models share the same specification, and because they predict the probability of an expenditure occurring, nearly all of them have the same number of observations as the sample size for the group under study. The exception is the set of healthcare less insurance models. The logit models have fewer observations than the sample for the group under study in this case, because the negative Healthcare outlays are omitted from the sample before running the regression. (For single men, the total is 470 observations; for single women, it is 1,260 observations; and for married couples, it is 2,515 observations.)
33. At first glance, the predicted value for out-of-town trips may appear low, but there are at least two reasons for this. First, out-of-town trips are defined "in the survey either as trips that last at least overnight for recreation purposes, or "day trips" in which the participant travels at least 75 miles from home. Therefore, they may be short in duration and not costly. Second, this phenomenon may be due to the econometrics underlying the model. The specification may be inaccurate due to omitted variables, improper transformation of the dependent or independent variables, or other reasons. However, the standard errors of the relative coefficients are wide enough to encompass an extremely large range of predicted values. This is because, as noted, E(InY) is the predicted value resulting from the regression, and exp[E(InY)] is the predicted value for the expenditure. A very small deviation in E(InY) can lead to a very large difference in exp[E(InY)]. For example, as shown in the table, the current predicted value for preretirees is $98. This is based on E(InY) of approximately 4.58. However, if E(InY) increases by 1 to 5.58, exp[E(InY)] increases to $265. Even at the 90-percent confidence level, an estimate of 5.58 is plausible; if all relevant parameters are evaluated at the lowest level in the 90-percent confidence interval, E(InY) is approximately -3.88; if all are evaluated at the highest level in the 90-percent confidence interval, E(InY) is approximately 12.99. The same reasoning applies to travel expenditures for single women. Applying the confidence intervals to their parameters yields an estimated range from 0.51 to 9.59 for E(InY).