Health and Housework in Later Life: A Longitudinal Study of Retired Couples

Health and Housework in Later Life: A Longitudinal Study of Retired Couples Abstract Objectives To examine how changes in wives’ and husbands’ health influenced housework time and domestic outsourcing in retired couples. Method We estimated fixed-effects models to test hypotheses about the gendered influence of health declines on absolute and relative measures of time spent on routine and nonroutine housework as well as the probability of outsourcing housework. The data were obtained from 23 waves of the German Socio-Economic Panel Study, comprising N = 25,119 annual observations of N = 3,889 retired couples aged 60–85 years. Results Wives’ and husbands’ housework time declined with health status, but these effects were large only for serious health problems. We found evidence for within-couple compensation of spouses’ health declines, a mechanism that was limited to indispensable tasks of routine housework. The probability of getting paid help from outside the household increased with declining health, and this increase was more strongly tied to wives’ health declines than to husbands’ health declines. Discussion The results demonstrate the relevance of health status for the performance of housework in retired couples. The evidence attests to the resilience of couples during later-life stages in which health issues may severely inhibit domestic productivity. Division of labor, Family dynamics, Family economics, Gender roles, Outsourcing In modern societies, retired women and men spend many more years of life in good health and remain productive well beyond retirement age (Gonzales, Matz-Costa, & Morrow-Howell, 2015; Leinonen, Martikainen, & Myrskylä, 2018). Productivity in older age, including grandparenting, volunteering, and domestic labor, has been linked to “successful aging” and reduced mortality in older people (Glass, de Leon, Marottoli, & Berkman, 1999; Liffiton, Horton, Baker, & Weir, 2012; Menec, 2003). Most productive activities are conditional on health (Gimenez-Nadal & Molina, 2015; Podor & Halliday, 2012). This is particularly true for home production defined as “the production of goods and services by the members of a household . . . using their own capital and their own unpaid labor” (Ironmonger, 2001, p. 6934). Ample research has shown that health is an important precondition for various kinds of daily routines in later life (Bennett, 1998; Menec, 2003; Rogers, Meyer, Walker, & Fisk, 1998). Yet, a gap of knowledge concerns the importance of women’s and men’s health for their participation in housework. The few existing studies on the relationship between health and housework performance are limited in scope (Caltabiano, Campolo, & Di Pino, 2016; Geist & Tabler, 2018; Gimenez-Nadal & Molina, 2015; Gimenez-Nadal & Ortega-Lapiedra, 2013; Hank & Jürges, 2007; Podor & Halliday, 2012; Stich & Hess, 2014). These studies have relied either on cross-sectional or on short-term longitudinal data to examine health effects on housework. Moreover, factors that are crucial to understanding the role of health and gender for housework performance have not been considered. These include the couple context, measures of housework that capture different and gendered housework tasks, and measures of health that capture more than two broad categories. In light of these limitations, the current understanding of the relationship between health and housework remains cursory and inconclusive, especially when it comes to late life stages in which health issues may severely inhibit domestic productivity. In the present study, we addressed these limitations to contribute to the literature in three ways. First, we considered the two main theoretical models of research on housework—gender and productivity—to derive hypotheses about the relevance of health for women’s and men’s participation in housework. Second, we used large-scale and long-running household panel data (1992–2015) from the German Socio-Economic Panel Study (Wagner, Frick, & Schupp, 2007) to test these hypotheses, providing a comprehensive longitudinal investigation of the dynamics of health and housework in retired couples. Third, we used a broader set of relevant outcomes than previous studies, including absolute and relative measures of time spent on female-typed and male-typed domestic activities as well as a measure for domestic outsourcing. Background Previous Studies The few cross-sectional studies that have explicitly addressed the relationship between health status and housework have shown mixed evidence. In U.S. studies, poorer health was associated with less housework time (Geist & Tabler, 2018; Podor & Halliday, 2012). In European studies, better health was associated with less time devoted to housework (Gimenez-Nadal & Molina, 2015; Gimenez-Nadal & Ortega-Lapiedra, 2013). Furthermore, Hank and Jürges (2007) showed that the division of housework in older European couples was responsive to health limitations of women, but not of men. Additional evidence has been offered by longitudinal studies that have controlled for health in models of housework performance in later life, although their prime research interest was in factors other than health. For Italian women and men, no effect of health on unpaid work was found (Caltabiano et al., 2016). German women and men spent less time on routine housework, such as cooking, cleaning, or doing the laundry, and nonroutine housework, such as repairs or gardening, if they were in poorer health (Stich & Hess, 2014). This study also showed that men responded to poor health of their partners by decreasing time spent on household maintenance and slightly increasing time spent on routine tasks. Women, in contrast, did not respond to poor health of their partners. The only studies that have addressed outsourcing as a reaction to health declines found that poorer health increased the likelihood of hiring paid help from outside the household (Baxter, Hewitt, & Western, 2009; Spitze, 1999). For Australia, the finding further indicated that men tended to compensate for their wives’ health declines and the associated loss of household productivity with a higher probability of outsourcing. For women, neither their own nor their spouse’s health was a significant predictor of outsourcing (Baxter et al., 2009). Theoretical Perspectives In line with the notion of “health [as] a resource for everyday life” (World Health Organization, 1998, p. 1), economic resource theories consider health as a predictor of time allocation (Gimenez-Nadal & Molina, 2015; Podor & Halliday, 2012). According to this view, women and men are seeking utility from market or home-produced goods and leisure. Time allocation is based on individuals’ health influencing productivity in different activities. Reduced to the sphere of housework—as market work is no longer relevant in our target group of retired couples—higher productivity and efficiency, for instance in cooking, cleaning, or home improvement, reduces the price of home production. Health declines imply productivity declines, increases in prices, constraints in time allocation, and call for compensation, given that there is a baseline amount of housework that has to be covered. Thus, women and men in couples do not decide on their time allocation independent from their spouse, but on the basis of their productivities and the productivities of their partner. The major alternative to this economic account of housework is gender construction theories. The relevance of this perspective for our study is twofold. First, it stresses the importance of symbolic exchange and identity formation for the division of housework in couple households (Berk, 1985). In the context of housework, “doing gender” (West & Zimmerman, 1987) refers to the performance of specific activities that affirm the traditional female gender identity (e.g., doing routine tasks such as cooking) and the traditional male gender identity (e.g., doing nonroutine tasks such as repairs). Gender display is considered a major determinant of women’s and men’s time allocation, enforcing and reinforcing “sluggish” patterns of a traditional division of labor over the life course, largely independent from other processes such as labor market participation and relative productivities. Routine housework is “women’s work”—an essential component of women’s doing gender. According to this view, women are more attached to domestic tasks than men, and this difference increases over the life course as women perform the lion’s share of housework in earlier and middle stages of coupled life (Bianchi, Sayer, Milkie, & Robinson, 2012). It is interesting to note that this notion, although highlighting a different mechanism, is equivalent to economic specialization theory (Becker, 1991) in terms of empirical predictions, as women who do gender in the home have higher domestic productivity than men. Second, the gender construction perspective stresses that housework consists of different and gendered housework tasks. A basic distinction is between routine and nonroutine chores. As Barnett and Shen (1997) have outlined, routine housework activities such as cooking, cleaning, or doing the laundry are necessary, less voluntaristic, and usually cannot be postponed for longer periods. These “low-schedule-control tasks” are considered as “female tasks.” Nonroutine or “high-schedule-control tasks” such as doing repairs and caring for the car or the garden are considered as “male tasks.” In contrast to (routine) female tasks, (nonroutine) male tasks are optional and may often be considered leisure-like activities. From the perspective of gender construction, women’s doing gender is more strongly attached to routine housework, whereas men refrain from these activities to avoid dissonance in their gender identity (Berk, 1985). Hypotheses Drawing on both of these theoretical perspectives, we propose three sets of hypotheses for (a) within-individual effects of health on housework, (b) within-couple compensation in terms of health and housework, and (c) domestic outsourcing in response to women’s and men’s health declines. Within-Individual Effects Given the productivity declines associated with poorer health, we expected that health declines reduce both women’s and men’s housework time (Hypothesis 1). Furthermore, as routine tasks are indispensable for household maintenance whereas nonroutine tasks are dispensable, we expected weaker effects of health declines on routine tasks and stronger effects on nonroutine tasks (Hypothesis 2). In addition to these gender-neutral hypotheses, the gender construction perspective suggests differences in the consequences of health declines for routine and nonroutine tasks. Given that the performance of routine tasks affirms women’s gender identities, we expected women to be more resistant to reducing time on these tasks in response to health declines; conversely, we expected men to be more resistant to reducing time on nonroutine tasks in response to health declines. This implies that the effect of health declines on routine housework is weaker for women than for men (Hypothesis 3a) and the effect of health declines on nonroutine housework is weaker for men than for women (Hypothesis 3b). Within-Couple Compensation Health declines of at least one partner in a household call for some type of replacement, as the loss of commodity production needs to be compensated for. Consequently, we hypothesize that partners’ health declines lead to increases in housework time (Hypothesis 4). Again, the effect of partner’s health declines should be stronger for indispensable routine housework than for dispensable nonroutine tasks (Hypothesis 5). In addition to these gender-neutral hypotheses, the gender construction perspective suggests differences in women’s and men’s reactions to their partners’ health declines. This applies to the routine tasks requiring compensation. Given that these tasks conform with the traditionally female identity and conflict with the traditionally male identity, we expected the effect of a partner’s health decline on routine housework to be stronger for women (Hypothesis 6a) and weaker for men (Hypothesis 6b). Domestic Outsourcing Finally, we considered domestic outsourcing as an alternative route to compensate for housework time lost to health declines. In line with the recent empirical evidence reviewed above, we expected that the probability of getting outside help for housework increases with health declines of both women and men (Hypothesis 7). As women’s loss of productivity creates a stronger need for compensation, we further expected this effect to be stronger for women’s health declines than for men’s health declines (Hypothesis 8). Method Data and Sample Our analysis was based on data from the German Socio-Economic Panel Study (SOEP, version 32.1; Wagner et al., 2007), one of the world’s largest and longest running surveys of households and individuals. We used annual data from the period between 1992 and 2015 (except for the 1993 wave in which no measure of health was available), for a total of 23 panel waves. Apart from the large sample size and the extensive observation window ideally suited for longitudinal analysis, the SOEP data offered three important benefits for the purposes of our study. First, each individual aged 17 and older living in a SOEP household answers the annual questionnaire, allowing us to study women and men jointly as couples. Second, the SOEP comprises longitudinal data about hours spent on different types of domestic labor including routine and nonroutine tasks as well as a measure of domestic outsourcing, allowing us to examine a broad set of relevant outcome measures. Third, the SOEP collects data about self-rated health since 1992 and annually since 1994. Given our research focus, we restricted the sample to observations of older, nonworking, heterosexual couples living together in a household. To define the analytic sample accordingly, we selected all annual observations of couples in which both partners were aged 60–85 and neither partner worked for pay. Moreover, we kept only observations in which valid data about health and housework were available. These conditions were met by 3,889 couples comprising 25,119 observations (couple-years). In Table 1, we show descriptive statistics and information about the measurement of all analytic variables. Table 1. Description and Coding of Analytic Variables     M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.      M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.  Note: SOEP 1992–2015, release 2017. N = 3,889 married couples observed at N = 25,119 occasions (couple-years). View Large Table 1. Description and Coding of Analytic Variables     M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.      M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.  Note: SOEP 1992–2015, release 2017. N = 3,889 married couples observed at N = 25,119 occasions (couple-years). View Large Outcomes Housework time was measured by annual survey questions about hours spent on different domestic activities. The data about housework time were based on so-called stylized measures of time use on a typical weekday, which are common in multipurpose surveys such as the SOEP. Stylized measures may lead to overestimates of housework hours (Kan, 2008; Lee & Waite, 2005), but they adequately reflect relative gender gaps and changes within individuals and couples over time (Juster, Ono, & Stafford, 2003; Kan & Pudney, 2008). We used two outcome measures for housework time, one pertaining to hours spent on routine tasks (washing, cooking, cleaning) and one pertaining to hours spent on nonroutine tasks (repairs and garden work). For each of these outcomes, we used absolute measures for wife’s and husband’s hours and relative measures for the wife’s and the husband’s shares of a couple’s total hours in either domain. All measures were available at every panel wave. Note that the sample sizes for the models for the wife’s and the husband’s shares are slightly lower because shares are undefined if a couple reports zero hours of housework. For routine housework, this was the case for 13 couples (0.3%); for nonroutine housework, this was the case for 336 couples (8.6%). In addition to the measures of housework time, we used a measure of domestic outsourcing. This dichotomous measure (available in 1994 and annually from 1999 until 2015) was based on the SOEP household questionnaire and indicated whether a couple regularly or occasionally paid a person to provide household help. Predictors Our main predictors of interest were wife’s and husband’s self-rated health measured on a 5-point scale ranging from very good to bad. This health measure is widely used in social science research. Its benefits include (a) data availability, as self-rated health is often included in the core questionnaires of panel studies, (b) validity, as self-rated health correlates with current and future health problems and mortality (Idler & Benyamini, 1997; Mossey & Shapiro, 1982), and (c) life-course coverage, as self-rated health captures health differences across the entire life span (Willson, Shuey, & Elder, 2007). Controls We estimated the effects of changes in wife’s and husband’s health net of age effects on the outcomes by adding controls for wife’s and husband’s age, measured in 5-year intervals. In the models for the wife’s and husband’s share of housework time, we additionally controlled for the couple’s total hours of routine housework and repairs/garden work, respectively. Model To examine the effects of health on housework time and within-couple shares of housework time, we used linear fixed-effects panel regression models (Allison, 2009). Changes in the dichotomous outcome measure of domestic outsourcing were estimated by a fixed-effects linear probability model. Fixed-effects models focus only on within-person or within-couple change over time, relating temporal variation in the outcome measures only to temporal variation in the independent variables. Specifically, our models linked changes over time in (a) housework hours, (b) individual shares of a couple’s total hours, and (c) domestic outsourcing to changes over time in the wife’s and in the husband’s self-rated health. As only characteristics that vary over time enter the fixed-effects model, all time-constant variables drop out of the equation. As a result, time-constant heterogeneity (observed and unobserved) is rendered inconsequential (Allison, 2009). We estimated a total of nine models. The first set of four models pertained to routine tasks, estimating change in (a) wife’s hours, (b) husband’s hours, (c) wife’s share, and (d) husband’s share; the second set of four analogous models pertained to nonroutine tasks; and the final model pertained to domestic outsourcing. In all models, we used dummy coding for changes in wife’s and husband’s self-rated health (reference: very good health) to capture nonlinear effects of these variables. We also examined interactions between both spouses’ levels of health, but found no meaningful effects of health constellations within couples above and beyond what was already captured by the two individual indicators. Therefore, we opted for a parsimonious model without these interactions. We also examined whether skewed distributions of the outcomes affected our results. Particularly, not only the husband’s hours of routine tasks, but also the wife’s hours of nonroutine tasks and the corresponding indicators for within-couple shares were right-skewed. However, logging these outcomes (adding 1 for zero hours and shares) did not improve model fit or influence our results on the key predictors of interest. The same was true for estimating a fixed-effects conditional logit model instead of a fixed-effects linear probability model for the dichotomous outcome of domestic outsourcing. Finally, note that case numbers are slightly lower in the models for shares of housework hours. Results We present the main results of our analysis in Figure 1 (routine tasks), Figure 2 (nonroutine tasks), and Figure 3 (outsourcing). The figures show marginal effects of changes in the outcomes associated with changes in the wife’s self-rated health (black curves) and the husband’s self-rated health (gray curves) along with 95% confidence intervals. In Table 2, we present the nine models on which the estimates shown in the figures are based. For calculating the marginal effects, all covariates (including the spouse’s self-rated health) were fixed at their means. Table 2. Fixed-Effects Models for Changes in Routine Tasks, Nonroutine Tasks, and Domestic Outsourcing   Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687    Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687  Note: SOEP 1992–2015, release 2017. SE in parentheses. See Table 1 for details on the measures. ***p < .001, **p < .01, *p < .05. View Large Table 2. Fixed-Effects Models for Changes in Routine Tasks, Nonroutine Tasks, and Domestic Outsourcing   Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687    Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687  Note: SOEP 1992–2015, release 2017. SE in parentheses. See Table 1 for details on the measures. ***p < .001, **p < .01, *p < .05. View Large Figure 1. View largeDownload slide Changes in wives’ and husband’s hours and shares of routine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 1. View largeDownload slide Changes in wives’ and husband’s hours and shares of routine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 2. View largeDownload slide Changes in wives’ and husband’s hours and shares of nonroutine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 2. View largeDownload slide Changes in wives’ and husband’s hours and shares of nonroutine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 3. View largeDownload slide Changes in domestic outsourcing. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 3. View largeDownload slide Changes in domestic outsourcing. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. In Figures 1 and 2, we illustrate changes in hours of routine tasks and hours of nonroutine tasks in the upper plots and corresponding changes in shares of a couples’ total hours of routine tasks and total hours of nonroutine tasks in the lower plots. The results shown in these figures allow us to evaluate our first two sets of hypotheses about within-individual effects and within-couple compensation. Our first and most general expectation was that housework time declines with health (Hypothesis 1). This expectation was supported by the data. Declines in housework time were found for wives and husbands, and this applied to routine tasks and nonroutine tasks. The relationship between declines in health and declines in housework time was not linear across the five categories of self-rated health. It emerged most clearly for the bottom category of bad health, whereas declines from very good and poor health were associated with only minor declines in housework time. Notably, women’s hours of routine tasks declined only if health declined to the bottom level. Our second expectation was that the effects of health declines are stronger for time on nonroutine tasks and weaker for time on routine tasks (Hypothesis 2). This expectation was only supported when looking at relative declines in wives’ housework time. Wives’ time spent on routine tasks declined by approximately 15% from 3.5 to 3 hr if their health declined from good to bad; wives’ time spent on nonroutine tasks declined by approximately 25% from slightly more than 1 hour to three quarters of an hour. In absolute terms, however, wives’ time on routine tasks declined more strongly than their time on nonroutine tasks. For husbands, we found that relative declines when moving from good to bad health were larger for routine tasks (–35%) than for nonroutine tasks (–25%); absolute declines were similar for both types of tasks, estimated at approximately half an hour. Our third hypothesis posited that the effects of health declines on routine tasks are weaker for wives than for husbands (Hypothesis 3a) and that the effect of health declines on nonroutine tasks are weaker for husbands than for wives (Hypothesis 3b). These expectations were supported only when looking at relative declines in time spent on routine tasks. Although absolute declines across the health categories were similar in magnitude for wives and husbands, relative declines were smaller in wives (–15%) than in husbands (–35%). Moreover, wives performed many more hours of routine tasks than husbands, and this difference was especially notable when comparing the estimates for wives in bad health (3 hr) and husbands in very good health (1.2 hr). Looking at nonroutine tasks, we did not find support for our hypothesis, as husbands’ absolute declines were larger than wives’ absolute declines, and relative declines were similar in both spouses. Moving to our second set of hypotheses about within-couple compensation, we expected, first, that partners’ health declines lead to increases in housework time (Hypothesis 4) and, second, that this compensation effect is stronger for routine tasks than for nonroutine tasks (Hypothesis 5). Both expectations were supported. For routine tasks, we observed compensation effects particularly in cases in which the spouse’s health declined to the bottom level (Figure 1). For this scenario, the husband’s hours of routine tasks increased by approximately half an hour, an amount that was equivalent to the wife’s concurrent decline. As a result, the husband’s share of the couple’s total hours was estimated to almost double from 20% to 35% (Figure 1, lower plot), although it fell short of equal division even when the wife was in bad health. If the husband’s health declined to the bottom category, wives’ hours and shares of routine tasks also went up, albeit to a lesser extent. In view of the latter result, it is clear that our expectations about a gendered compensation of routine tasks in couples were not supported. Contrary to our hypotheses (Hypotheses 6a and 6b), husbands were not more reluctant than wives to increase their hours spent on these tasks in response to their spouses’ health declines. In Figure 3, we illustrate changes in the probability of domestic outsourcing. The results shown in this figure allow us to evaluate our third set of hypotheses about outsourcing as an alternative route to compensate for housework time lost to health declines. We expected, first, that the probability of getting outside help with housework generally increases with spouses’ health declines (Hypothesis 7) and, second, that this effect is stronger for women’s than for men’s health declines (Hypothesis 8). Both expectations were supported. The probability of outsourcing domestic labor was estimated less than 10% in cases of good health. It increased slightly but not significantly (Table 2) with declines to satisfactory and poor health. Similar to the results described above, larger effects were limited to the bottom category of bad health. As expected, the wife’s declines rather than the husband’s declines to bad health were associated with a rise in the probability of outsourcing. Although this probability increased by 5 percentage points or approximately 50%, the absolute probability remained low, as five in six couples did not get paid help from outside the household even if the wife was in bad health. Discussion This study addressed several gaps of knowledge about the importance of health for wives’ and husbands’ participation in housework. Previous research has mainly relied on cross-sectional or short-term longitudinal data and ignored the couple context of gendered time allocations as well as gendered housework tasks. In light of these limitations, our understanding of the relationship between health and housework remained cursory and inconclusive. In the present study, we used large-scale panel data from Germany covering an extensive window of observation to provide a comprehensive longitudinal investigation of the dynamics of health and housework in retired couples. Our data included a broad set of relevant outcomes, including absolute and relative measures of time spent on “female-typed” routine tasks and “male-typed” nonroutine tasks as well as a measure of domestic outsourcing. Fixed-effects models for health-related changes in these outcomes yielded five main findings. First, wives’ and husbands’ housework time declined with health status, but these effects were large in magnitude only for the bottom level indicating serious health problems. Second, declines in housework time concerned not only nonroutine activities but also routine tasks, although wives—the main performers of routine housework—reduced their time on these tasks only by a sixth even after declines to bad health. Third, we found evidence for a gendered response in terms of “female-typed” activities, as effects of health declines on relative reductions of time spent on routine tasks were smaller for wives than for husbands. The evidence in this regard was only partial, however, as the expected reverse picture was not found for “male-typed” nonroutine tasks. Fourth, we found robust evidence for within-couple compensation of spouses’ health declines. As expected, this mechanism was limited to indispensable tasks of routine housework. Contrary to our expectations, however, within-couple compensation was not gendered, as both husbands and wives increased their time on routine housework in response to spousal health declines. Finally, we found the expected patterns for compensation outside the household. The probability of getting paid help from outside the household increased with health declines and this increase was more strongly tied to wives’ health declines than to husbands’ health declines. This gendered response indicates that wives rather than husbands perform tasks that are indispensable to the household maintenance of retired couples. If health issues render wives unable to perform these tasks, a stronger need for outsourcing arises. Taken together, our evidence demonstrates the relevance of health status for the performance of housework in retired couples. The findings attest to the resilience of couples during later-life stages in which health issues may severely inhibit domestic productivity. Three aspects about this resilience are notable. First, spouses compensated for each other’s health declines by increasing housework time in response to health-related declines in their partners. This response was less gendered than it could be expected on the basis of gender construction theories, as husbands substantially increased their time on “female-typed” routine tasks if their wives were in bad health. Second, housework time declined substantially only for declines to the bottom category of self-rated health. Housework performance in retired couples was hardly affected by declines even from good to poor health. Moreover, even in cases of bad health, wives’ contribution to routine tasks remained substantial. As wives in bad health still accounted for two thirds of a couple’s total time spent on routine housework, these findings also speak to the path dependency and persistence of a gendered division of housework even in later-life stages in which tradeoffs with market work are no longer relevant. Third, our findings show that domestic outsourcing was an alternative route to compensate for wives’ health-related declines in housework time. Yet, the prevalence of these arrangements remained low even after wives moved to the bottom category of bad health. This low prevalence could be explained by different factors that we could not examine in more detail with our data. These include gender display, as wives may still be reluctant to relinquish activities that are strongly linked to their gender display even while in bad health. Alternatively, budget constraints may prevent couples from getting the outside help that they required. Couples may also receive assistance from children, relatives, or friends who are not paid directly for their support. This type of help from outside the household was not captured by the outcomes used in our study. In closing, we note three limitations of our study that warrant future investigation. First, we interpreted the association between health and housework only in one direction, whereby changes in housework time were implied to result from changes in health. Although this direction is pertinent, it is important to note that housework may also impact on health. Research, particularly in the field of epidemiology, has shown positive and negative effects of housework on health (Everard, Lach, Fisher, & Baum, 2000; Menec, 2003). In this regard, our findings direct attention to wives performing the lion’s share of routine housework even when in bad health. This could imply further negative health effects if these women are overburdened by domestic labor that they are unable to, or unwilling to, relinquish. Remaining active in the housework even in the face of health limitations, however, may also prevent or slow down further declines. Second, although our main predictor of interest—self-rated health—improved on the measures used in previous studies on the association between health and housework (Caltabiano et al., 2016; Geist & Tabler, 2018; Gimenez-Nadal & Molina, 2015; Podor & Halliday, 2012; Stich & Hess, 2014), it is still subject to important limitations. For example, previous research has found that self-ratings of health may be affected by factors other than health changes, such as self-esteem (Borawski, Kinney, & Kahana, 1996; Ferraro & Feller, 1996). Moreover, self-rated health as a holistic measure of health status does not provide direct information about functional limitations that may impact on the performance of housework. For example, the measure of self-rated health is sensitive to the onset of chronic conditions, although highly prevalent conditions such as high blood pressure and diabetes are not directly relevant to the performance of housework. A measure of functional limitations may better capture relevant health changes. Furthermore, objective indicators such as grip strength may also be more strongly linked to housework than the general measure used in the present study. As data about these indicators are increasingly included in large-scale multipurpose panel surveys, longitudinal data will soon be available to examine the relationship between health and housework on the basis of more detailed and more objective health measures. Third, although our outcome measures on hours spent on routine and nonroutine housework represent standard indicators in the literature on housework, these measures have limitations specific to a study focus on later life and declining health. Most notably, housework time captures only the input in terms of hours spent on housework rather than the output in terms of completed household tasks. If health declines reduce domestic productivity, our outcome measures may not fully capture these reductions because people may take more time to complete less housework. This suggests that health declines may have affected domestic productivity even at levels above the bottom category, although our outcome measures of housework time did not pick up these effects. To gain more insight into these possibilities, future research should complement the outcomes used in the present study by indicators that allow measuring domestic productivity as a ratio of output to input. These limitations of our predictor and outcome measures call for research that examines the association between health and housework in more detail. As longevity and productivity in later life are still on the rise, questions about this association will remain relevant in future research on housework and the gender division of labor. Funding This study was supported by the German Research Foundation (grant number SCHU 3081/1-1). Conflict of Interest None reported. Acknowledgments Replication files to this article are available at the authors’ websites: www.thomasleopold.eu and www.floschulz.de. References Allison, P. D. ( 2009). Fixed effects regression models . Thousand Oaks, CA: Sage. Google Scholar CrossRef Search ADS   Barnett, R. C., & Shen, Y.-C. ( 1997). Gender, high- and low-schedule-control housework tasks, and psychological distress. Journal of Family Issues , 18, 403– 428. doi: 10.1177/019251397018004003 Google Scholar CrossRef Search ADS   Baxter, J., Hewitt, B., & Western, M. ( 2009). Who uses paid domestic labor in Australia? Choice and constraint in hiring household help. Feminist Economics , 15, 1– 26. doi: 10.1080/13545700802248989 Google Scholar CrossRef Search ADS   Becker, G. S. ( 1991). A treatise on the family . Cambridge, MA: Harvard University Press. Bennett, K. M. ( 1998). Gender and longitudinal changes in physical activities in later life. Age and Ageing , 27( Suppl 3), 24– 28. doi: 10.1093/ageing/27.suppl_3.24 Google Scholar CrossRef Search ADS PubMed  Berk, S. F. ( 1985). The gender factory . New York, NY: Plenum Press. Google Scholar CrossRef Search ADS   Bianchi, S. M., Sayer, L. C., Milkie, M. A., & Robinson, J. P. ( 2012). Housework: Who did, does or will do it, and how much does it matter? Social Forces , 91, 55– 63. doi: 10.1093/sf/sos120 Google Scholar CrossRef Search ADS   Borawski, E. A., Kinney, J. M., & Kahana, E. ( 1996). The meaning of older adults’ health appraisals: Congruence with health status and determinant of mortality. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 51, S157– S170. doi: 10.1093/geronb/51B.3.S157 Google Scholar CrossRef Search ADS   Caltabiano, M., Campolo, M. G., & Di Pino, A. ( 2016). Retirement and intra-household labour division of Italian couples: A new simultaneous equation approach. Social Indicators Research , 128, 1217– 1238. doi: 10.1007/s11205-015-1076-5 Google Scholar CrossRef Search ADS   Everard, K. M., Lach, H. W., Fisher, E. B., & Baum, M. C. ( 2000). Relationship of activity and social support to the functional health of older adults. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 55, S208– S212. doi: 10.1093/geronb/55.4.S208 Google Scholar CrossRef Search ADS   Ferraro, K. F., & Feller, J. R. ( 1996). Self and age differences in defining health situations: A comparison of measurement strategies. Research on Aging , 18, 175– 201. doi: 10.1177/0164027596182002 Google Scholar CrossRef Search ADS   Geist, C., & Tabler, J. ( 2018). Somebody has to DUST! Gender, health, and housework in older couples. Journal of Women & Aging , 30, 38– 48. doi: 10.1080/08952841.2016.1259442 Google Scholar CrossRef Search ADS PubMed  Gimenez-Nadal, J. I., & Molina, J. A. ( 2015). Health status and the allocation of time: Cross-country evidence from Europe. Economic Modelling , 46, 188– 203. doi: 10.1016/j.econmod.2014.12.034 Google Scholar CrossRef Search ADS   Gimenez-Nadal, J. I., & Ortega-Lapiedra, R. ( 2013). Health status and time allocation in Spain. Applied Economics Letters , 20, 1435– 1439. doi: 10.1080/13504851.2013.818205 Google Scholar CrossRef Search ADS   Glass, T. A., de Leon, C. M., Marottoli, R. A., & Berkman, L. F. ( 1999). Population based study of social and productive activities as predictors of survival among elderly Americans. BMJ , 319, 478– 483. doi: 10.1136/bmj.319.7208.478 Google Scholar CrossRef Search ADS PubMed  Gonzales, E., Matz-Costa, C., & Morrow-Howell, N. ( 2015). Increasing opportunities for the productive engagement of older adults: A response to population aging. The Gerontologist , 55, 252– 261. doi: 10.1093/geront/gnu176 Google Scholar CrossRef Search ADS PubMed  Hank, K., & Jürges, H. ( 2007). Gender and the division of household labor in older couples: A European perspective. Journal of Family Issues , 28, 399– 421. doi: 10.1177/0192513X06296427 Google Scholar CrossRef Search ADS   Idler, E. L., & Benyamini, Y. ( 1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior , 38, 21– 37. doi: 10.2307/2955359 Google Scholar CrossRef Search ADS PubMed  Ironmonger, D. ( 2001). Household production. In N. J. Smelser & P. B. Baltes (Eds.), International encyclopedia of the social & behavioral sciences  (pp. 6934– 6939). Amsterdam, The Netherlands: Elsevier. doi: 10.1016/B0-08-043076-7/03964-4 Google Scholar CrossRef Search ADS   Juster, F. T., Ono, H., & Stafford, F. P. ( 2003). An assessment of alternative measures of time use. Sociological Methodology , 33, 19– 54. doi: 10.1111/j.0081-1750.2003.t01-1-00126.x Google Scholar CrossRef Search ADS   Kan, M. Y. ( 2008). Measuring housework participation: The gap between “stylized” questionnaire estimates and diary-based estimates. Social Indicators Research , 86, 381– 400. doi: 10.1007/s11205-007-9184-5 Google Scholar CrossRef Search ADS   Kan, M. Y., & Pudney, S. ( 2008). Measurement error in stylized and diary data on time use. Sociological Methodology , 38, 101– 132. doi: 10.1111/j.1467-9531.2008.00197.x Google Scholar CrossRef Search ADS   Lee, Y.-S., & Waite, L. J. ( 2005). Husbands’ and wives’ time spent on housework: A comparison of measures. Journal of Marriage and Family , 67, 328– 336. doi: 10.1111/j.0022-2445.2005.00119.x Google Scholar CrossRef Search ADS   Leinonen, T., Martikainen, P., & Myrskylä, M. ( 2018). Working life and retirement expectancies at age 50 by social class: Period and cohort trends and projections for Finland. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 73, 302–313. doi: 10.1093/geronb/gbv104 Liffiton, J. A., Horton, S., Baker, J., & Weir, P. L. ( 2012). Successful aging: How does physical activity influence engagement with life? European Review of Aging and Physical Activity , 9, 103– 108. doi: 10.1007/s11556-012-0098-0 Google Scholar CrossRef Search ADS   Menec, V. H. ( 2003). The relation between everyday activities and successful aging: A 6-year longitudinal study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 58, S74– S82. doi: 10.1093/geronb/58.2.S74 Google Scholar CrossRef Search ADS   Mossey, J. M., & Shapiro, E. ( 1982). Self-rated health: A predictor of mortality among the elderly. American Journal of Public Health , 72, 800– 808. doi: 10.2105/AJPH.72.8.800 Google Scholar CrossRef Search ADS PubMed  Podor, M., & Halliday, T. J. ( 2012). Health status and the allocation of time. Health Economics , 21, 514– 527. doi: 10.1002/hec.1725 Google Scholar CrossRef Search ADS PubMed  Rogers, W. A., Meyer, B., Walker, N., & Fisk, A. D. ( 1998). Functional limitations to daily living tasks in the aged: A focus group analysis. Human Factors , 40, 111– 125. doi: 10.1518/001872098779480613 Google Scholar CrossRef Search ADS PubMed  Spitze, G. ( 1999). Getting help with housework. Household resources and social networks. Journal of Family Issues , 20, 724– 745. doi: 10.1177/019251399020006001 Google Scholar CrossRef Search ADS   Stich, D., & Hess, M. ( 2014). Renteneintritt und Hausarbeit  [Retirement and Household Labor] (SOEP Paper No. 683). Berlin: DIW. doi: 10.2139/ssrn.2501245 Wagner, G. G., Frick, J. R., & Schupp, J. ( 2007). The German Socio-Economic Panel Study (SOEP)—Scope, evolution and enhancements. Schmollers Jahrbuch , 127, 139– 169. doi: 10.2139/ssrn.1028709 West, C., & Zimmerman, D. H. ( 1987). Doing gender. Gender & Society , 1, 125– 151. doi: 10.1177/0891243287001002002 Google Scholar CrossRef Search ADS   Willson, A. E., Shuey, K. M., & Elder, G.-H., Jr. ( 2007). Cumulative advantage processes as mechanisms of inequality in life course health. American Journal of Sociology , 112, 1886– 1924. doi: 10.1086/512712 Google Scholar CrossRef Search ADS   World Health Organization.( 1998). Health promotion glossary . Geneva, Switzerland: Author. Retrieved November 10, 2017, from http://apps.who.int/iris/bitstream/10665/64546/1/WHO_HPR_HEP_98.1.pdf © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journals of Gerontology Series B: Psychological Sciences and Social Sciences Oxford University Press

Health and Housework in Later Life: A Longitudinal Study of Retired Couples

Loading next page...
 
/lp/ou_press/health-and-housework-in-later-life-a-longitudinal-study-of-retired-LXTLuSeiKP
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
1079-5014
eISSN
1758-5368
D.O.I.
10.1093/geronb/gby015
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives To examine how changes in wives’ and husbands’ health influenced housework time and domestic outsourcing in retired couples. Method We estimated fixed-effects models to test hypotheses about the gendered influence of health declines on absolute and relative measures of time spent on routine and nonroutine housework as well as the probability of outsourcing housework. The data were obtained from 23 waves of the German Socio-Economic Panel Study, comprising N = 25,119 annual observations of N = 3,889 retired couples aged 60–85 years. Results Wives’ and husbands’ housework time declined with health status, but these effects were large only for serious health problems. We found evidence for within-couple compensation of spouses’ health declines, a mechanism that was limited to indispensable tasks of routine housework. The probability of getting paid help from outside the household increased with declining health, and this increase was more strongly tied to wives’ health declines than to husbands’ health declines. Discussion The results demonstrate the relevance of health status for the performance of housework in retired couples. The evidence attests to the resilience of couples during later-life stages in which health issues may severely inhibit domestic productivity. Division of labor, Family dynamics, Family economics, Gender roles, Outsourcing In modern societies, retired women and men spend many more years of life in good health and remain productive well beyond retirement age (Gonzales, Matz-Costa, & Morrow-Howell, 2015; Leinonen, Martikainen, & Myrskylä, 2018). Productivity in older age, including grandparenting, volunteering, and domestic labor, has been linked to “successful aging” and reduced mortality in older people (Glass, de Leon, Marottoli, & Berkman, 1999; Liffiton, Horton, Baker, & Weir, 2012; Menec, 2003). Most productive activities are conditional on health (Gimenez-Nadal & Molina, 2015; Podor & Halliday, 2012). This is particularly true for home production defined as “the production of goods and services by the members of a household . . . using their own capital and their own unpaid labor” (Ironmonger, 2001, p. 6934). Ample research has shown that health is an important precondition for various kinds of daily routines in later life (Bennett, 1998; Menec, 2003; Rogers, Meyer, Walker, & Fisk, 1998). Yet, a gap of knowledge concerns the importance of women’s and men’s health for their participation in housework. The few existing studies on the relationship between health and housework performance are limited in scope (Caltabiano, Campolo, & Di Pino, 2016; Geist & Tabler, 2018; Gimenez-Nadal & Molina, 2015; Gimenez-Nadal & Ortega-Lapiedra, 2013; Hank & Jürges, 2007; Podor & Halliday, 2012; Stich & Hess, 2014). These studies have relied either on cross-sectional or on short-term longitudinal data to examine health effects on housework. Moreover, factors that are crucial to understanding the role of health and gender for housework performance have not been considered. These include the couple context, measures of housework that capture different and gendered housework tasks, and measures of health that capture more than two broad categories. In light of these limitations, the current understanding of the relationship between health and housework remains cursory and inconclusive, especially when it comes to late life stages in which health issues may severely inhibit domestic productivity. In the present study, we addressed these limitations to contribute to the literature in three ways. First, we considered the two main theoretical models of research on housework—gender and productivity—to derive hypotheses about the relevance of health for women’s and men’s participation in housework. Second, we used large-scale and long-running household panel data (1992–2015) from the German Socio-Economic Panel Study (Wagner, Frick, & Schupp, 2007) to test these hypotheses, providing a comprehensive longitudinal investigation of the dynamics of health and housework in retired couples. Third, we used a broader set of relevant outcomes than previous studies, including absolute and relative measures of time spent on female-typed and male-typed domestic activities as well as a measure for domestic outsourcing. Background Previous Studies The few cross-sectional studies that have explicitly addressed the relationship between health status and housework have shown mixed evidence. In U.S. studies, poorer health was associated with less housework time (Geist & Tabler, 2018; Podor & Halliday, 2012). In European studies, better health was associated with less time devoted to housework (Gimenez-Nadal & Molina, 2015; Gimenez-Nadal & Ortega-Lapiedra, 2013). Furthermore, Hank and Jürges (2007) showed that the division of housework in older European couples was responsive to health limitations of women, but not of men. Additional evidence has been offered by longitudinal studies that have controlled for health in models of housework performance in later life, although their prime research interest was in factors other than health. For Italian women and men, no effect of health on unpaid work was found (Caltabiano et al., 2016). German women and men spent less time on routine housework, such as cooking, cleaning, or doing the laundry, and nonroutine housework, such as repairs or gardening, if they were in poorer health (Stich & Hess, 2014). This study also showed that men responded to poor health of their partners by decreasing time spent on household maintenance and slightly increasing time spent on routine tasks. Women, in contrast, did not respond to poor health of their partners. The only studies that have addressed outsourcing as a reaction to health declines found that poorer health increased the likelihood of hiring paid help from outside the household (Baxter, Hewitt, & Western, 2009; Spitze, 1999). For Australia, the finding further indicated that men tended to compensate for their wives’ health declines and the associated loss of household productivity with a higher probability of outsourcing. For women, neither their own nor their spouse’s health was a significant predictor of outsourcing (Baxter et al., 2009). Theoretical Perspectives In line with the notion of “health [as] a resource for everyday life” (World Health Organization, 1998, p. 1), economic resource theories consider health as a predictor of time allocation (Gimenez-Nadal & Molina, 2015; Podor & Halliday, 2012). According to this view, women and men are seeking utility from market or home-produced goods and leisure. Time allocation is based on individuals’ health influencing productivity in different activities. Reduced to the sphere of housework—as market work is no longer relevant in our target group of retired couples—higher productivity and efficiency, for instance in cooking, cleaning, or home improvement, reduces the price of home production. Health declines imply productivity declines, increases in prices, constraints in time allocation, and call for compensation, given that there is a baseline amount of housework that has to be covered. Thus, women and men in couples do not decide on their time allocation independent from their spouse, but on the basis of their productivities and the productivities of their partner. The major alternative to this economic account of housework is gender construction theories. The relevance of this perspective for our study is twofold. First, it stresses the importance of symbolic exchange and identity formation for the division of housework in couple households (Berk, 1985). In the context of housework, “doing gender” (West & Zimmerman, 1987) refers to the performance of specific activities that affirm the traditional female gender identity (e.g., doing routine tasks such as cooking) and the traditional male gender identity (e.g., doing nonroutine tasks such as repairs). Gender display is considered a major determinant of women’s and men’s time allocation, enforcing and reinforcing “sluggish” patterns of a traditional division of labor over the life course, largely independent from other processes such as labor market participation and relative productivities. Routine housework is “women’s work”—an essential component of women’s doing gender. According to this view, women are more attached to domestic tasks than men, and this difference increases over the life course as women perform the lion’s share of housework in earlier and middle stages of coupled life (Bianchi, Sayer, Milkie, & Robinson, 2012). It is interesting to note that this notion, although highlighting a different mechanism, is equivalent to economic specialization theory (Becker, 1991) in terms of empirical predictions, as women who do gender in the home have higher domestic productivity than men. Second, the gender construction perspective stresses that housework consists of different and gendered housework tasks. A basic distinction is between routine and nonroutine chores. As Barnett and Shen (1997) have outlined, routine housework activities such as cooking, cleaning, or doing the laundry are necessary, less voluntaristic, and usually cannot be postponed for longer periods. These “low-schedule-control tasks” are considered as “female tasks.” Nonroutine or “high-schedule-control tasks” such as doing repairs and caring for the car or the garden are considered as “male tasks.” In contrast to (routine) female tasks, (nonroutine) male tasks are optional and may often be considered leisure-like activities. From the perspective of gender construction, women’s doing gender is more strongly attached to routine housework, whereas men refrain from these activities to avoid dissonance in their gender identity (Berk, 1985). Hypotheses Drawing on both of these theoretical perspectives, we propose three sets of hypotheses for (a) within-individual effects of health on housework, (b) within-couple compensation in terms of health and housework, and (c) domestic outsourcing in response to women’s and men’s health declines. Within-Individual Effects Given the productivity declines associated with poorer health, we expected that health declines reduce both women’s and men’s housework time (Hypothesis 1). Furthermore, as routine tasks are indispensable for household maintenance whereas nonroutine tasks are dispensable, we expected weaker effects of health declines on routine tasks and stronger effects on nonroutine tasks (Hypothesis 2). In addition to these gender-neutral hypotheses, the gender construction perspective suggests differences in the consequences of health declines for routine and nonroutine tasks. Given that the performance of routine tasks affirms women’s gender identities, we expected women to be more resistant to reducing time on these tasks in response to health declines; conversely, we expected men to be more resistant to reducing time on nonroutine tasks in response to health declines. This implies that the effect of health declines on routine housework is weaker for women than for men (Hypothesis 3a) and the effect of health declines on nonroutine housework is weaker for men than for women (Hypothesis 3b). Within-Couple Compensation Health declines of at least one partner in a household call for some type of replacement, as the loss of commodity production needs to be compensated for. Consequently, we hypothesize that partners’ health declines lead to increases in housework time (Hypothesis 4). Again, the effect of partner’s health declines should be stronger for indispensable routine housework than for dispensable nonroutine tasks (Hypothesis 5). In addition to these gender-neutral hypotheses, the gender construction perspective suggests differences in women’s and men’s reactions to their partners’ health declines. This applies to the routine tasks requiring compensation. Given that these tasks conform with the traditionally female identity and conflict with the traditionally male identity, we expected the effect of a partner’s health decline on routine housework to be stronger for women (Hypothesis 6a) and weaker for men (Hypothesis 6b). Domestic Outsourcing Finally, we considered domestic outsourcing as an alternative route to compensate for housework time lost to health declines. In line with the recent empirical evidence reviewed above, we expected that the probability of getting outside help for housework increases with health declines of both women and men (Hypothesis 7). As women’s loss of productivity creates a stronger need for compensation, we further expected this effect to be stronger for women’s health declines than for men’s health declines (Hypothesis 8). Method Data and Sample Our analysis was based on data from the German Socio-Economic Panel Study (SOEP, version 32.1; Wagner et al., 2007), one of the world’s largest and longest running surveys of households and individuals. We used annual data from the period between 1992 and 2015 (except for the 1993 wave in which no measure of health was available), for a total of 23 panel waves. Apart from the large sample size and the extensive observation window ideally suited for longitudinal analysis, the SOEP data offered three important benefits for the purposes of our study. First, each individual aged 17 and older living in a SOEP household answers the annual questionnaire, allowing us to study women and men jointly as couples. Second, the SOEP comprises longitudinal data about hours spent on different types of domestic labor including routine and nonroutine tasks as well as a measure of domestic outsourcing, allowing us to examine a broad set of relevant outcome measures. Third, the SOEP collects data about self-rated health since 1992 and annually since 1994. Given our research focus, we restricted the sample to observations of older, nonworking, heterosexual couples living together in a household. To define the analytic sample accordingly, we selected all annual observations of couples in which both partners were aged 60–85 and neither partner worked for pay. Moreover, we kept only observations in which valid data about health and housework were available. These conditions were met by 3,889 couples comprising 25,119 observations (couple-years). In Table 1, we show descriptive statistics and information about the measurement of all analytic variables. Table 1. Description and Coding of Analytic Variables     M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.      M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.  Note: SOEP 1992–2015, release 2017. N = 3,889 married couples observed at N = 25,119 occasions (couple-years). View Large Table 1. Description and Coding of Analytic Variables     M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.      M  SD  Range  N (observations)  Description/survey question  Age  Wife  68.96  5.66  60–85  25,119      Husband  71.26  5.75  60–85  25,119    Self-rated health          25,119  “How would you describe your current health?”    Wife  0.02    0–1    Very good      0.21    0–1    Good      0.44    0–1    Satisfactory      0.25    0–1    Poor      0.08    0–1    Bad    Husband  0.02    0–1    Very good      0.23    0–1    Good      0.45    0–1    Satisfactory      0.22    0–1    Poor      0.08    0–1    Bad  Housework time            “What does a typical weekday look like for you? How many hours per day do you spend on the following activities?”  Routine housework            Washing, cooking, cleaning; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  3.48  1.59  0–12  25,119  Hours on a typical weekday  Wife’s share  0.78  0.22  0–1  24,901  Share of a couple’s total hours    Husband’s hours  1.04  1.12  0–12  25,119  Hours on a typical weekday    Husband’s share  0.22  0.22  0–1  24,901  Share of a couple’s total hours  Repairs and garden work            Repairs on and around the house, car repairs, garden work; “does not apply” recoded to 0 hr; top coded at 12 hr    Wife’s hours  0.96  1.14  0–10  25,119  Hours on a typical weekday  Wife’s share  0.31  0.28  0–1  20,842  Share of a couple’s total hours    Husband’s hours  1.86  1.64  0–12  25,119  Hours on a typical weekday    Husband’s share  0.69  0.28  0–1  20,842  Share of a couple’s total hours  Outsourcing    0.10    0–1  22,104  “Do you regularly or occasionally pay someone to provide household help?” 1 = yes, 0 = no.  Note: SOEP 1992–2015, release 2017. N = 3,889 married couples observed at N = 25,119 occasions (couple-years). View Large Outcomes Housework time was measured by annual survey questions about hours spent on different domestic activities. The data about housework time were based on so-called stylized measures of time use on a typical weekday, which are common in multipurpose surveys such as the SOEP. Stylized measures may lead to overestimates of housework hours (Kan, 2008; Lee & Waite, 2005), but they adequately reflect relative gender gaps and changes within individuals and couples over time (Juster, Ono, & Stafford, 2003; Kan & Pudney, 2008). We used two outcome measures for housework time, one pertaining to hours spent on routine tasks (washing, cooking, cleaning) and one pertaining to hours spent on nonroutine tasks (repairs and garden work). For each of these outcomes, we used absolute measures for wife’s and husband’s hours and relative measures for the wife’s and the husband’s shares of a couple’s total hours in either domain. All measures were available at every panel wave. Note that the sample sizes for the models for the wife’s and the husband’s shares are slightly lower because shares are undefined if a couple reports zero hours of housework. For routine housework, this was the case for 13 couples (0.3%); for nonroutine housework, this was the case for 336 couples (8.6%). In addition to the measures of housework time, we used a measure of domestic outsourcing. This dichotomous measure (available in 1994 and annually from 1999 until 2015) was based on the SOEP household questionnaire and indicated whether a couple regularly or occasionally paid a person to provide household help. Predictors Our main predictors of interest were wife’s and husband’s self-rated health measured on a 5-point scale ranging from very good to bad. This health measure is widely used in social science research. Its benefits include (a) data availability, as self-rated health is often included in the core questionnaires of panel studies, (b) validity, as self-rated health correlates with current and future health problems and mortality (Idler & Benyamini, 1997; Mossey & Shapiro, 1982), and (c) life-course coverage, as self-rated health captures health differences across the entire life span (Willson, Shuey, & Elder, 2007). Controls We estimated the effects of changes in wife’s and husband’s health net of age effects on the outcomes by adding controls for wife’s and husband’s age, measured in 5-year intervals. In the models for the wife’s and husband’s share of housework time, we additionally controlled for the couple’s total hours of routine housework and repairs/garden work, respectively. Model To examine the effects of health on housework time and within-couple shares of housework time, we used linear fixed-effects panel regression models (Allison, 2009). Changes in the dichotomous outcome measure of domestic outsourcing were estimated by a fixed-effects linear probability model. Fixed-effects models focus only on within-person or within-couple change over time, relating temporal variation in the outcome measures only to temporal variation in the independent variables. Specifically, our models linked changes over time in (a) housework hours, (b) individual shares of a couple’s total hours, and (c) domestic outsourcing to changes over time in the wife’s and in the husband’s self-rated health. As only characteristics that vary over time enter the fixed-effects model, all time-constant variables drop out of the equation. As a result, time-constant heterogeneity (observed and unobserved) is rendered inconsequential (Allison, 2009). We estimated a total of nine models. The first set of four models pertained to routine tasks, estimating change in (a) wife’s hours, (b) husband’s hours, (c) wife’s share, and (d) husband’s share; the second set of four analogous models pertained to nonroutine tasks; and the final model pertained to domestic outsourcing. In all models, we used dummy coding for changes in wife’s and husband’s self-rated health (reference: very good health) to capture nonlinear effects of these variables. We also examined interactions between both spouses’ levels of health, but found no meaningful effects of health constellations within couples above and beyond what was already captured by the two individual indicators. Therefore, we opted for a parsimonious model without these interactions. We also examined whether skewed distributions of the outcomes affected our results. Particularly, not only the husband’s hours of routine tasks, but also the wife’s hours of nonroutine tasks and the corresponding indicators for within-couple shares were right-skewed. However, logging these outcomes (adding 1 for zero hours and shares) did not improve model fit or influence our results on the key predictors of interest. The same was true for estimating a fixed-effects conditional logit model instead of a fixed-effects linear probability model for the dichotomous outcome of domestic outsourcing. Finally, note that case numbers are slightly lower in the models for shares of housework hours. Results We present the main results of our analysis in Figure 1 (routine tasks), Figure 2 (nonroutine tasks), and Figure 3 (outsourcing). The figures show marginal effects of changes in the outcomes associated with changes in the wife’s self-rated health (black curves) and the husband’s self-rated health (gray curves) along with 95% confidence intervals. In Table 2, we present the nine models on which the estimates shown in the figures are based. For calculating the marginal effects, all covariates (including the spouse’s self-rated health) were fixed at their means. Table 2. Fixed-Effects Models for Changes in Routine Tasks, Nonroutine Tasks, and Domestic Outsourcing   Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687    Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687  Note: SOEP 1992–2015, release 2017. SE in parentheses. See Table 1 for details on the measures. ***p < .001, **p < .01, *p < .05. View Large Table 2. Fixed-Effects Models for Changes in Routine Tasks, Nonroutine Tasks, and Domestic Outsourcing   Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687    Routine housework  Nonroutine housework      Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Wife’s hours  Husband’s hours  Wife’s share  Husband’s share  Outsourcing  Wife’s self-rated health (ref. very good)                     Good  0.12 (0.07)  0.02 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.01 (0.05)  0.02 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  0.15* (0.07)  0.03 (0.05)  0.01 (0.01)  −0.01 (0.01)  −0.03 (0.05)  0.04 (0.07)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Poor  0.12 (0.07)  0.12* (0.05)  −0.01 (0.01)  0.01 (0.01)  −0.10 (0.05)  0.00 (0.07)  −0.01 (0.01)  0.00 (0.01)  0.02 (0.01)   Bad  −0.39*** (0.08)  0.53*** (0.06)  −0.13*** (0.01)  0.13*** (0.01)  −0.29*** (0.06)  −0.12 (0.08)  −0.07*** (0.02)  0.07*** (0.02)  0.07*** (0.01)  Husband’s self-rated health (ref. very good)                     Good  −0.02 (0.06)  −0.09* (0.05)  0.02** (0.01)  −0.02** (0.01)  0.02 (0.05)  −0.06 (0.06)  0.01 (0.01)  −0.01 (0.01)  0.01 (0.01)   Satisfactory  −0.01 (0.07)  −0.10* (0.05)  0.02** (0.01)  −0.02* (0.01)  0.03 (0.05)  −0.12 (0.06)  0.01 (0.01)  −0.02 (0.01)  0.01 (0.01)   Poor  −0.03 (0.07)  −0.15** (0.05)  0.03*** (0.01)  −0.03*** (0.01)  0.04 (0.05)  −0.23*** (0.06)  0.04** (0.01)  −0.04** (0.01)  0.01 (0.01)   Bad  0.12 (0.08)  −0.47*** (0.06)  0.09*** (0.01)  −0.09*** (0.01)  0.01 (0.05)  −0.63*** (0.07)  0.13*** (0.02)  −0.13*** (0.02)  0.02 (0.01)  Age (ref. 60–64)                     65–69  −0.18*** (0.02)  0.02 (0.02)  −0.02*** (0.00)  0.01*** (0.00)  −0.04* (0.02)  −0.16*** (0.03)  0.02*** (0.00)  −0.01 (0.01)  0.01* (0.01)   70–74  −0.29*** (0.03)  0.09*** (0.02)  −0.03*** (0.00)  0.03*** (0.00)  −0.10*** (0.02)  −0.37*** (0.03)  0.04*** (0.01)  −0.02*** (0.01)  0.03*** (0.01)   75–79  −0.49*** (0.04)  0.13*** (0.03)  −0.06*** (0.00)  0.05*** (0.00)  −0.18*** (0.03)  −0.59*** (0.04)  0.05*** (0.01)  −0.04*** (0.01)  0.05*** (0.01)   80–85  −0.79*** (0.05)  0.19*** (0.03)  −0.09*** (0.01)  0.07*** (0.01)  −0.33*** (0.04)  −0.93*** (0.04)  0.04*** (0.01)  −0.06*** (0.01)  0.11*** (0.01)  Couple’s total hours      −0.02*** (0.00)  0.02*** (0.00)      0.00 (0.00)  −0.00 (0.00)    Number of couple-years  25,119  25,119  20,901  20,901  25,119  25,119  20,842  20,842  22,104  Number of couples  3,889  3,889  3,876  3,876  3,889  3,889  3,553  3,553  3,687  Note: SOEP 1992–2015, release 2017. SE in parentheses. See Table 1 for details on the measures. ***p < .001, **p < .01, *p < .05. View Large Figure 1. View largeDownload slide Changes in wives’ and husband’s hours and shares of routine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 1. View largeDownload slide Changes in wives’ and husband’s hours and shares of routine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 2. View largeDownload slide Changes in wives’ and husband’s hours and shares of nonroutine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 2. View largeDownload slide Changes in wives’ and husband’s hours and shares of nonroutine tasks. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 3. View largeDownload slide Changes in domestic outsourcing. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. Figure 3. View largeDownload slide Changes in domestic outsourcing. Marginal effects are shown, and covariates are fixed at their means. Estimates based on the models shown in Table 2. In Figures 1 and 2, we illustrate changes in hours of routine tasks and hours of nonroutine tasks in the upper plots and corresponding changes in shares of a couples’ total hours of routine tasks and total hours of nonroutine tasks in the lower plots. The results shown in these figures allow us to evaluate our first two sets of hypotheses about within-individual effects and within-couple compensation. Our first and most general expectation was that housework time declines with health (Hypothesis 1). This expectation was supported by the data. Declines in housework time were found for wives and husbands, and this applied to routine tasks and nonroutine tasks. The relationship between declines in health and declines in housework time was not linear across the five categories of self-rated health. It emerged most clearly for the bottom category of bad health, whereas declines from very good and poor health were associated with only minor declines in housework time. Notably, women’s hours of routine tasks declined only if health declined to the bottom level. Our second expectation was that the effects of health declines are stronger for time on nonroutine tasks and weaker for time on routine tasks (Hypothesis 2). This expectation was only supported when looking at relative declines in wives’ housework time. Wives’ time spent on routine tasks declined by approximately 15% from 3.5 to 3 hr if their health declined from good to bad; wives’ time spent on nonroutine tasks declined by approximately 25% from slightly more than 1 hour to three quarters of an hour. In absolute terms, however, wives’ time on routine tasks declined more strongly than their time on nonroutine tasks. For husbands, we found that relative declines when moving from good to bad health were larger for routine tasks (–35%) than for nonroutine tasks (–25%); absolute declines were similar for both types of tasks, estimated at approximately half an hour. Our third hypothesis posited that the effects of health declines on routine tasks are weaker for wives than for husbands (Hypothesis 3a) and that the effect of health declines on nonroutine tasks are weaker for husbands than for wives (Hypothesis 3b). These expectations were supported only when looking at relative declines in time spent on routine tasks. Although absolute declines across the health categories were similar in magnitude for wives and husbands, relative declines were smaller in wives (–15%) than in husbands (–35%). Moreover, wives performed many more hours of routine tasks than husbands, and this difference was especially notable when comparing the estimates for wives in bad health (3 hr) and husbands in very good health (1.2 hr). Looking at nonroutine tasks, we did not find support for our hypothesis, as husbands’ absolute declines were larger than wives’ absolute declines, and relative declines were similar in both spouses. Moving to our second set of hypotheses about within-couple compensation, we expected, first, that partners’ health declines lead to increases in housework time (Hypothesis 4) and, second, that this compensation effect is stronger for routine tasks than for nonroutine tasks (Hypothesis 5). Both expectations were supported. For routine tasks, we observed compensation effects particularly in cases in which the spouse’s health declined to the bottom level (Figure 1). For this scenario, the husband’s hours of routine tasks increased by approximately half an hour, an amount that was equivalent to the wife’s concurrent decline. As a result, the husband’s share of the couple’s total hours was estimated to almost double from 20% to 35% (Figure 1, lower plot), although it fell short of equal division even when the wife was in bad health. If the husband’s health declined to the bottom category, wives’ hours and shares of routine tasks also went up, albeit to a lesser extent. In view of the latter result, it is clear that our expectations about a gendered compensation of routine tasks in couples were not supported. Contrary to our hypotheses (Hypotheses 6a and 6b), husbands were not more reluctant than wives to increase their hours spent on these tasks in response to their spouses’ health declines. In Figure 3, we illustrate changes in the probability of domestic outsourcing. The results shown in this figure allow us to evaluate our third set of hypotheses about outsourcing as an alternative route to compensate for housework time lost to health declines. We expected, first, that the probability of getting outside help with housework generally increases with spouses’ health declines (Hypothesis 7) and, second, that this effect is stronger for women’s than for men’s health declines (Hypothesis 8). Both expectations were supported. The probability of outsourcing domestic labor was estimated less than 10% in cases of good health. It increased slightly but not significantly (Table 2) with declines to satisfactory and poor health. Similar to the results described above, larger effects were limited to the bottom category of bad health. As expected, the wife’s declines rather than the husband’s declines to bad health were associated with a rise in the probability of outsourcing. Although this probability increased by 5 percentage points or approximately 50%, the absolute probability remained low, as five in six couples did not get paid help from outside the household even if the wife was in bad health. Discussion This study addressed several gaps of knowledge about the importance of health for wives’ and husbands’ participation in housework. Previous research has mainly relied on cross-sectional or short-term longitudinal data and ignored the couple context of gendered time allocations as well as gendered housework tasks. In light of these limitations, our understanding of the relationship between health and housework remained cursory and inconclusive. In the present study, we used large-scale panel data from Germany covering an extensive window of observation to provide a comprehensive longitudinal investigation of the dynamics of health and housework in retired couples. Our data included a broad set of relevant outcomes, including absolute and relative measures of time spent on “female-typed” routine tasks and “male-typed” nonroutine tasks as well as a measure of domestic outsourcing. Fixed-effects models for health-related changes in these outcomes yielded five main findings. First, wives’ and husbands’ housework time declined with health status, but these effects were large in magnitude only for the bottom level indicating serious health problems. Second, declines in housework time concerned not only nonroutine activities but also routine tasks, although wives—the main performers of routine housework—reduced their time on these tasks only by a sixth even after declines to bad health. Third, we found evidence for a gendered response in terms of “female-typed” activities, as effects of health declines on relative reductions of time spent on routine tasks were smaller for wives than for husbands. The evidence in this regard was only partial, however, as the expected reverse picture was not found for “male-typed” nonroutine tasks. Fourth, we found robust evidence for within-couple compensation of spouses’ health declines. As expected, this mechanism was limited to indispensable tasks of routine housework. Contrary to our expectations, however, within-couple compensation was not gendered, as both husbands and wives increased their time on routine housework in response to spousal health declines. Finally, we found the expected patterns for compensation outside the household. The probability of getting paid help from outside the household increased with health declines and this increase was more strongly tied to wives’ health declines than to husbands’ health declines. This gendered response indicates that wives rather than husbands perform tasks that are indispensable to the household maintenance of retired couples. If health issues render wives unable to perform these tasks, a stronger need for outsourcing arises. Taken together, our evidence demonstrates the relevance of health status for the performance of housework in retired couples. The findings attest to the resilience of couples during later-life stages in which health issues may severely inhibit domestic productivity. Three aspects about this resilience are notable. First, spouses compensated for each other’s health declines by increasing housework time in response to health-related declines in their partners. This response was less gendered than it could be expected on the basis of gender construction theories, as husbands substantially increased their time on “female-typed” routine tasks if their wives were in bad health. Second, housework time declined substantially only for declines to the bottom category of self-rated health. Housework performance in retired couples was hardly affected by declines even from good to poor health. Moreover, even in cases of bad health, wives’ contribution to routine tasks remained substantial. As wives in bad health still accounted for two thirds of a couple’s total time spent on routine housework, these findings also speak to the path dependency and persistence of a gendered division of housework even in later-life stages in which tradeoffs with market work are no longer relevant. Third, our findings show that domestic outsourcing was an alternative route to compensate for wives’ health-related declines in housework time. Yet, the prevalence of these arrangements remained low even after wives moved to the bottom category of bad health. This low prevalence could be explained by different factors that we could not examine in more detail with our data. These include gender display, as wives may still be reluctant to relinquish activities that are strongly linked to their gender display even while in bad health. Alternatively, budget constraints may prevent couples from getting the outside help that they required. Couples may also receive assistance from children, relatives, or friends who are not paid directly for their support. This type of help from outside the household was not captured by the outcomes used in our study. In closing, we note three limitations of our study that warrant future investigation. First, we interpreted the association between health and housework only in one direction, whereby changes in housework time were implied to result from changes in health. Although this direction is pertinent, it is important to note that housework may also impact on health. Research, particularly in the field of epidemiology, has shown positive and negative effects of housework on health (Everard, Lach, Fisher, & Baum, 2000; Menec, 2003). In this regard, our findings direct attention to wives performing the lion’s share of routine housework even when in bad health. This could imply further negative health effects if these women are overburdened by domestic labor that they are unable to, or unwilling to, relinquish. Remaining active in the housework even in the face of health limitations, however, may also prevent or slow down further declines. Second, although our main predictor of interest—self-rated health—improved on the measures used in previous studies on the association between health and housework (Caltabiano et al., 2016; Geist & Tabler, 2018; Gimenez-Nadal & Molina, 2015; Podor & Halliday, 2012; Stich & Hess, 2014), it is still subject to important limitations. For example, previous research has found that self-ratings of health may be affected by factors other than health changes, such as self-esteem (Borawski, Kinney, & Kahana, 1996; Ferraro & Feller, 1996). Moreover, self-rated health as a holistic measure of health status does not provide direct information about functional limitations that may impact on the performance of housework. For example, the measure of self-rated health is sensitive to the onset of chronic conditions, although highly prevalent conditions such as high blood pressure and diabetes are not directly relevant to the performance of housework. A measure of functional limitations may better capture relevant health changes. Furthermore, objective indicators such as grip strength may also be more strongly linked to housework than the general measure used in the present study. As data about these indicators are increasingly included in large-scale multipurpose panel surveys, longitudinal data will soon be available to examine the relationship between health and housework on the basis of more detailed and more objective health measures. Third, although our outcome measures on hours spent on routine and nonroutine housework represent standard indicators in the literature on housework, these measures have limitations specific to a study focus on later life and declining health. Most notably, housework time captures only the input in terms of hours spent on housework rather than the output in terms of completed household tasks. If health declines reduce domestic productivity, our outcome measures may not fully capture these reductions because people may take more time to complete less housework. This suggests that health declines may have affected domestic productivity even at levels above the bottom category, although our outcome measures of housework time did not pick up these effects. To gain more insight into these possibilities, future research should complement the outcomes used in the present study by indicators that allow measuring domestic productivity as a ratio of output to input. These limitations of our predictor and outcome measures call for research that examines the association between health and housework in more detail. As longevity and productivity in later life are still on the rise, questions about this association will remain relevant in future research on housework and the gender division of labor. Funding This study was supported by the German Research Foundation (grant number SCHU 3081/1-1). Conflict of Interest None reported. Acknowledgments Replication files to this article are available at the authors’ websites: www.thomasleopold.eu and www.floschulz.de. References Allison, P. D. ( 2009). Fixed effects regression models . Thousand Oaks, CA: Sage. Google Scholar CrossRef Search ADS   Barnett, R. C., & Shen, Y.-C. ( 1997). Gender, high- and low-schedule-control housework tasks, and psychological distress. Journal of Family Issues , 18, 403– 428. doi: 10.1177/019251397018004003 Google Scholar CrossRef Search ADS   Baxter, J., Hewitt, B., & Western, M. ( 2009). Who uses paid domestic labor in Australia? Choice and constraint in hiring household help. Feminist Economics , 15, 1– 26. doi: 10.1080/13545700802248989 Google Scholar CrossRef Search ADS   Becker, G. S. ( 1991). A treatise on the family . Cambridge, MA: Harvard University Press. Bennett, K. M. ( 1998). Gender and longitudinal changes in physical activities in later life. Age and Ageing , 27( Suppl 3), 24– 28. doi: 10.1093/ageing/27.suppl_3.24 Google Scholar CrossRef Search ADS PubMed  Berk, S. F. ( 1985). The gender factory . New York, NY: Plenum Press. Google Scholar CrossRef Search ADS   Bianchi, S. M., Sayer, L. C., Milkie, M. A., & Robinson, J. P. ( 2012). Housework: Who did, does or will do it, and how much does it matter? Social Forces , 91, 55– 63. doi: 10.1093/sf/sos120 Google Scholar CrossRef Search ADS   Borawski, E. A., Kinney, J. M., & Kahana, E. ( 1996). The meaning of older adults’ health appraisals: Congruence with health status and determinant of mortality. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 51, S157– S170. doi: 10.1093/geronb/51B.3.S157 Google Scholar CrossRef Search ADS   Caltabiano, M., Campolo, M. G., & Di Pino, A. ( 2016). Retirement and intra-household labour division of Italian couples: A new simultaneous equation approach. Social Indicators Research , 128, 1217– 1238. doi: 10.1007/s11205-015-1076-5 Google Scholar CrossRef Search ADS   Everard, K. M., Lach, H. W., Fisher, E. B., & Baum, M. C. ( 2000). Relationship of activity and social support to the functional health of older adults. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 55, S208– S212. doi: 10.1093/geronb/55.4.S208 Google Scholar CrossRef Search ADS   Ferraro, K. F., & Feller, J. R. ( 1996). Self and age differences in defining health situations: A comparison of measurement strategies. Research on Aging , 18, 175– 201. doi: 10.1177/0164027596182002 Google Scholar CrossRef Search ADS   Geist, C., & Tabler, J. ( 2018). Somebody has to DUST! Gender, health, and housework in older couples. Journal of Women & Aging , 30, 38– 48. doi: 10.1080/08952841.2016.1259442 Google Scholar CrossRef Search ADS PubMed  Gimenez-Nadal, J. I., & Molina, J. A. ( 2015). Health status and the allocation of time: Cross-country evidence from Europe. Economic Modelling , 46, 188– 203. doi: 10.1016/j.econmod.2014.12.034 Google Scholar CrossRef Search ADS   Gimenez-Nadal, J. I., & Ortega-Lapiedra, R. ( 2013). Health status and time allocation in Spain. Applied Economics Letters , 20, 1435– 1439. doi: 10.1080/13504851.2013.818205 Google Scholar CrossRef Search ADS   Glass, T. A., de Leon, C. M., Marottoli, R. A., & Berkman, L. F. ( 1999). Population based study of social and productive activities as predictors of survival among elderly Americans. BMJ , 319, 478– 483. doi: 10.1136/bmj.319.7208.478 Google Scholar CrossRef Search ADS PubMed  Gonzales, E., Matz-Costa, C., & Morrow-Howell, N. ( 2015). Increasing opportunities for the productive engagement of older adults: A response to population aging. The Gerontologist , 55, 252– 261. doi: 10.1093/geront/gnu176 Google Scholar CrossRef Search ADS PubMed  Hank, K., & Jürges, H. ( 2007). Gender and the division of household labor in older couples: A European perspective. Journal of Family Issues , 28, 399– 421. doi: 10.1177/0192513X06296427 Google Scholar CrossRef Search ADS   Idler, E. L., & Benyamini, Y. ( 1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior , 38, 21– 37. doi: 10.2307/2955359 Google Scholar CrossRef Search ADS PubMed  Ironmonger, D. ( 2001). Household production. In N. J. Smelser & P. B. Baltes (Eds.), International encyclopedia of the social & behavioral sciences  (pp. 6934– 6939). Amsterdam, The Netherlands: Elsevier. doi: 10.1016/B0-08-043076-7/03964-4 Google Scholar CrossRef Search ADS   Juster, F. T., Ono, H., & Stafford, F. P. ( 2003). An assessment of alternative measures of time use. Sociological Methodology , 33, 19– 54. doi: 10.1111/j.0081-1750.2003.t01-1-00126.x Google Scholar CrossRef Search ADS   Kan, M. Y. ( 2008). Measuring housework participation: The gap between “stylized” questionnaire estimates and diary-based estimates. Social Indicators Research , 86, 381– 400. doi: 10.1007/s11205-007-9184-5 Google Scholar CrossRef Search ADS   Kan, M. Y., & Pudney, S. ( 2008). Measurement error in stylized and diary data on time use. Sociological Methodology , 38, 101– 132. doi: 10.1111/j.1467-9531.2008.00197.x Google Scholar CrossRef Search ADS   Lee, Y.-S., & Waite, L. J. ( 2005). Husbands’ and wives’ time spent on housework: A comparison of measures. Journal of Marriage and Family , 67, 328– 336. doi: 10.1111/j.0022-2445.2005.00119.x Google Scholar CrossRef Search ADS   Leinonen, T., Martikainen, P., & Myrskylä, M. ( 2018). Working life and retirement expectancies at age 50 by social class: Period and cohort trends and projections for Finland. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 73, 302–313. doi: 10.1093/geronb/gbv104 Liffiton, J. A., Horton, S., Baker, J., & Weir, P. L. ( 2012). Successful aging: How does physical activity influence engagement with life? European Review of Aging and Physical Activity , 9, 103– 108. doi: 10.1007/s11556-012-0098-0 Google Scholar CrossRef Search ADS   Menec, V. H. ( 2003). The relation between everyday activities and successful aging: A 6-year longitudinal study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 58, S74– S82. doi: 10.1093/geronb/58.2.S74 Google Scholar CrossRef Search ADS   Mossey, J. M., & Shapiro, E. ( 1982). Self-rated health: A predictor of mortality among the elderly. American Journal of Public Health , 72, 800– 808. doi: 10.2105/AJPH.72.8.800 Google Scholar CrossRef Search ADS PubMed  Podor, M., & Halliday, T. J. ( 2012). Health status and the allocation of time. Health Economics , 21, 514– 527. doi: 10.1002/hec.1725 Google Scholar CrossRef Search ADS PubMed  Rogers, W. A., Meyer, B., Walker, N., & Fisk, A. D. ( 1998). Functional limitations to daily living tasks in the aged: A focus group analysis. Human Factors , 40, 111– 125. doi: 10.1518/001872098779480613 Google Scholar CrossRef Search ADS PubMed  Spitze, G. ( 1999). Getting help with housework. Household resources and social networks. Journal of Family Issues , 20, 724– 745. doi: 10.1177/019251399020006001 Google Scholar CrossRef Search ADS   Stich, D., & Hess, M. ( 2014). Renteneintritt und Hausarbeit  [Retirement and Household Labor] (SOEP Paper No. 683). Berlin: DIW. doi: 10.2139/ssrn.2501245 Wagner, G. G., Frick, J. R., & Schupp, J. ( 2007). The German Socio-Economic Panel Study (SOEP)—Scope, evolution and enhancements. Schmollers Jahrbuch , 127, 139– 169. doi: 10.2139/ssrn.1028709 West, C., & Zimmerman, D. H. ( 1987). Doing gender. Gender & Society , 1, 125– 151. doi: 10.1177/0891243287001002002 Google Scholar CrossRef Search ADS   Willson, A. E., Shuey, K. M., & Elder, G.-H., Jr. ( 2007). Cumulative advantage processes as mechanisms of inequality in life course health. American Journal of Sociology , 112, 1886– 1924. doi: 10.1086/512712 Google Scholar CrossRef Search ADS   World Health Organization.( 1998). Health promotion glossary . Geneva, Switzerland: Author. Retrieved November 10, 2017, from http://apps.who.int/iris/bitstream/10665/64546/1/WHO_HPR_HEP_98.1.pdf © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

The Journals of Gerontology Series B: Psychological Sciences and Social SciencesOxford University Press

Published: Apr 11, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off