Abstract
Although anecdotal evidence of older parents postponing retirement to financially support their grown children is common, the empirical evidence is scarce. In this paper, we use data from the 1992 to 2010 waves of the Health and Retirement Study to identify a broad set of pivotal events in the lives of adult children. First, we determine whether these events affect subsequent financial transfers from parents to children over multiple years. Next, we determine whether those events that result in subsequent transfers also shift parental retirement expectations. Finally, we quantify the impact of the unexpected children’s events on retirement realizations, moving beyond the correlational analyses in prior literature. Our findings show that a child’s move out of a parental home decreases both expectations and realizations of working after age 65. The magnitude of this effect is similar to that of an own health shock experienced during pre-retirement years.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
For instance, Van Bavel and De Winter (2013) show that births of grandchildren affect the retirement decisions of women but not men.
Other studies find strong links between children’s events and contemporaneous parental transfers (i.e., McGarry 2016; McGarry and Schoeni 1995). Our paper differs in that we focus on how children’s events affect subsequent retirement decisions. In an effort to detect long-term impact of events, we study transfers that occur up to 4 years after the events take place.
More recently, McGarry (2016) examined a broad set of children’s events (also analyzed in this paper) and showed that parental probability to give transfers increases in response to most of the important children events, including marriage, divorce, and college graduation. In addition to helping children in times of need, the author showed that the parents may also give transfers to mark a happy event in children’s lives.
Altonji et al. (1992) as well as Hayashi et al. (1996) rejected full risk-sharing within families using earlier data on food consumption. Charles et al. (2014) employed the same dataset as was used in the prior studies, the Panel Study of Income Dynamics (PSID), but utilized the latest available records as well as a more comprehensive measure of consumption within extended families.
For example, see McGarry and Schoeni (1995).
Studies have shown that parents give more financial assistance to children who have lower incomes or experience negative wealth shocks. See McGarry (2016) for additional discussion of the altruism model.
In addition, we conduct a parallel analysis with an indicator for whether respondents were employed full-time after they reached age 62.
HRS family data also include the total number of children in school as well as the total number of home-owning children. Unfortunately, data on children’s schooling are only available for 37% of our sample due to missing records. We do not utilize data on the number of home-owning children in our baseline analysis due to difficulties with interpretation. It is not clear whether a decrease in the number of home-owning children indicates higher or lower financial need, as loss of homeownership could imply children’s financial ruin or signal decreased need due to elimination of mortgage payments. When included in the analysis, coefficients on variables for loss or gain of homeownership among children are not statistically significant and do not change our baseline results.
Although it is possible for our variable to pick up cases where parents are the ones who are moving out of their children’s homes, the data suggest that such cases are unlikely. About 87% of our respondents are homeowners at the baseline, and no respondents indicate that any of their children were on their home deed prior to the move.
Because the HRS family dataset does not include data on the age of each child, we utilize information from a supplementary file, HRS Respondent-Kid file (see Section 6.5 for more details). Using that dataset, we ensure that children’s events in our analysis reflect only those events that are experienced by children ages 18 and over (events that are experienced by children under the age of 18 are recoded as zero’s). Because events that are experienced by children under 18 could have a systematically different impact on parental retirement expectations and realizations than the events that are experienced by adult children, the age restriction simplifies interpretation of our coefficients.
In 1994 and 1995, significant transfers were defined as $100 or more.
We also examined the female spouses of the respondents; however, our analysis was constrained by a much smaller sample. As women typically retire earlier than men, the average age of spousal retirement is 54 years in our data. At the baseline, 34% of spouses are already not in the labor force, and we observe only 66% past age 65. Our findings suggested that there is some adjustment to children’s events taking place on the spousal labor supply margin; however, the small sample prevents us from drawing any definitive conclusions.
Following Hurd et al. (2004), we define individuals to be in the labor force if they report working full-time, part-time or are unemployed.
Our main sample size of 974 workers is a result of the following series of sample restrictions. Starting with 4959 men in the 1931–1941 birth cohort, we exclude all workers who were retired before ages 58/59 and who were not in the labor force at that age, leaving 3424 people. We further exclude workers who were not observed during preretirement ages of 58 to 61 or were not observed after age 65, which reduces our sample to 1709 people. Conditioning on having a child further reduces our sample to 1613 people. While missing data on children has very little effect on sample size, yielding 1566 parents, missing data on parental retirement expectations and health of parent and spouse reduces our sample further to 974 people.
In addition, we conduct a parallel analysis examining expectations and realizations of full-time work past age 62.
The analysis with FT(65)i is estimated using a probit model.
We regress P(65) reported at ages 60/61 on P(65) reported at the baseline rather than estimating a first difference model in order to avoid restricting the coefficient on baseline P(65) to equal 1 since our results in Table 5 show the coefficient on baseline P(65) to be significantly less than 1. However, if we run the expectations analysis in first differences, the coefficient on child moving out remains very similar in magnitude and statistically significant at 10% level.
Our findings are robust to excluding P(62) i,58/59.
Two labor force status indicators differentiate workers who have part-time jobs or are unemployed, with the omitted category being a full-time worker. We include indicators for whether respondent has completed some college or has a college degree. Race is reflected via indicators for black, other race, and Hispanic.
Primary earner is an indicator for having higher annual earnings than spouse during respondent’s ages 56–59.
The omitted health category is excellent/very good health. Indicator for good health reflects good health. Indicator for poor health captures fair and poor health.
We include the self-reported expectation of living to age 75 reported at ages 58/59 in our control set throughout the analysis since past studies have found mortality expectations to affect actual retirement timing (Hurd et al. 2004). Our results are robust to excluding this control.
Financial controls include respondent’s annual earnings, spouse’s annual earnings, total household’s financial wealth (including net value of checking and savings accounts, stocks, bonds, and other saving tools), and non-financial wealth (including the value of primary residence, vehicles, and businesses). In addition, financial controls include indicators for reporting defined benefit, defined contribution, or both types of pensions.
Our data do not allow us to distinguish child’s layoff from voluntary job leave as the loss of employment is constructed from changes in the number of employed children between waves.
McGarry (2004) uses HRS data to show that self-reported health changes have large effects on retirement expectations, even relative to changes in financial variables.
All children’s events including divorce and moving into the parental home are included in the regressions unless stated otherwise. The coefficients on these two events are typically small in magnitude and not statistically significant.
A subset of children’s events is shown in Table 4 for brevity. None of the omitted coefficients on children’s events are statistically significant.
We further examined the effect of birth of grandchild based on whether or not the parent is a primary or secondary earner. While the interaction of birth of a grandchild with secondary earner indicator is not significant in the full sample, we find that when we restrict the sample to secondary earners, the birth of a grandchild has a statistically significant negative effect on probability of working full-time past age 65. However, our sample of non-primary earners is only 160 people, and our estimates are noisy. Results are available upon request.
RAND prepares two longitudinal versions of HRS Family data: one with respondent-kid observations and one containing summary measures on all of the respondent’s children. The respondents in the two versions of the datasets are the exact same individuals. For our main analysis, we utilize the summary data on all of the respondent’s children; however, the summary measures alone do not allow us to identify characteristics of the children who moved out of the parental home. Thus, we use the other file, the HRS Respondent-Kid File, for the supplemental analysis.
The main limitation of using Respondent-Kid File in the analysis is the pervasiveness of missing records for many of the children’s characteristics. For instance, data on children’s income range is missing for over half of the children who moved out of parental home in our sample.
We find similar results when we add annual financial transfers received between ages of 60 and 65 to the retirement realizations equations in Table 6: the coefficient on the child moving out remains unchanged in magnitude and significance.
References
Altonji JG, Hayashi F, Kotlikoff LJ (1992) Is the extended family altruistically linked? Direct tests using micro data. Am Econ Rev 82(5):1177–1198
Benitez-Silva H, Dwyer DS (2005) The rationality of retirement expectations and the role of new information. Review of Economics and Statistics 87(3):587–592
Bernheim BD (1989) The timing of retirement: a comparison of expectations and realizations. In: Wise D (ed) The economics of aging. University Press, Chicago, p 1989
Charles KK, Danziger S, Li G, Schoeni R (2014) The intergenerational correlation of consumption expenditures. Am Econ Rev 104(5):136–140
Coile C, Gruber J (2007) Future social security entitlements and the retirement decision. Rev Econ Stat 89(2):234–246
Couch KA, Tamborini CR, Reznik G, Phillips JWR (2013) Divorce, Women’s earnings and retirement over the life course. In: Couch KA, Daly MC, Zissimopoulos JM (eds) Lifecycle events and their consequences: job loss, family change, and declines in health. Stanford University Press, Stanford, pp 133–157
Cox D, Way M (2011) Intergenerational transfers and the great recession. Paper, Boston College Working
Damman M, Henkens K, Kalmijn M (2011) The impact of midlife educational, work, health, and family experiences on Men's early retirement. J Gerontol Ser B Psychol Sci Soc Sci 66((5):617–627
Goda GS, Shoven JB, Slavov SN (2011) Does stock market performance influence retirement expectations? NBER working paper, October 2011
Hayashi F, Altonji J, Kotlikoff L (1996) Risk-sharing between and within families. Econometrica 64(2):261–294
Hurd M, Smith JP, Zissimopoulos JM (2004) The effects of subjective survival on retirement and social security claiming. J Appl Econ 19(6):761–775
Johnson RW, Penner RG, Toohey D (2008) Do out-of-pocket health care costs delay retirement?" retirement policy discussion paper 08–02
Leukhina O, Santoro M (2011) "An in-depth look at intergenerational flows." Boston College Center for Retirement Research Working Paper, No. 2011–11
McGarry K (2004) Health and retirement: do changes in health affect retirement expectations? J Hum Resour 39(3):624–648
McGarry K (2016) Dynamic aspects of family transfers. J Public Econ 137:1–13
McGarry K, Schoeni RF (1995) Transfer behavior in the health and retirement study: measurement and the redistribution of resources within the family. J Hum Resour 30:S184–S226
Merrill Lynch Wealth Management and Age Wave (2013) “Family & retirement: the elephant in the room.” http://www.agewave.com/research/FamilyAndRetirementStudyReport.pdf
Szinovacz ME, DeViney S, Davey A (2001) Influences of family obligations and relationships on retirement: variations by gender, race, and marital status. J Gerontol 56B(1):S20–S27
Van Bavel J, De Winter T (2013) Becoming a grandparent and early retirement in Europe. Eur Sociol Rev:1–14
Wightman PD, Patrick ME, Schoeni RF, Schulenberg JE (2013) Historical trends in parental financial support of young adults. Population Studies Center Research Report:13–801
Acknowledgements
We thank seminar participants at the University of California, San Diego, for their valuable feedback. We especially would like to thank Julie Berry Cullen for many helpful discussions as well as Gordon B. Dahl and Roger Gordon for their thoughtful comments. In addition, we thank three anonymous referees for helpful feedback and suggestions. The views expressed here should not be interpreted as those of the Congressional Budget Office or the Social Security Administration.
Funding
This study was not funded.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Responsible editor: Alessandro Cigno
Appendix
Appendix
Rights and permissions
About this article
Cite this article
Miller, M., Tamborini, C.R. & Reznik, G.L. Parental retirement timing: the role of unanticipated events in the lives of adult children. J Popul Econ 31, 747–781 (2018). https://doi.org/10.1007/s00148-018-0698-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00148-018-0698-8