As the ‘price’ of the output or the value of having a clean house and neat/clean clothing is missing from these input demand equations, we expect differences in this valuation to be incorporated in the residuals. To the extent that these preferences are correlated with any of the covariates, the coefficient estimates obtained for these covariates will be biased as they will capture both the true association and the effect via their correlation with preferences. We find no particular reason to expect a correlation between preferences and our price measures or other covariates. However, to the extent that such preferences do differ among households, we expect that those who value domestic services more will, all else equal, be likely to use more inputs causing the residuals to be positively correlated across equations.
Productivity in home production that is not already captured by the covariates (such as education) will also be incorporated in the residuals. Individuals with lower productivity in home production have to spend more time to produce the same output compared to others with identical characteristics, and hence may have higher residuals in the time‐use equations. If this is the case, we would expect to see strong positive correlation between the residuals from the weekend and weekday equations for the same individuals. Significant positive correlations for a single partner may also be indicative of that individual’s (as opposed to the household’s) valuation of domestic services. It is not possible to distinguish here between individual preferences for domestic services and individual productivity.
Analysis sample criteria
The data for this analysis are drawn from two countries, France and the UK. The general social structure is similar between these countries, justifying comparative analysis to check robustness. OECD statistics indicate that the labour force participation rate of women aged 25–59 was 69.6% in the UK and 73.1% in France in the year 2000. The use of maids services and availability of appliances are similar between the UK and France, as is confirmed by descriptive evidence in this paper. The surveys also have a similar design. Obtaining similar results using these two samples will, we believe, make these findings substantially more credible.
The primary source of the French data is the 1998–9 French Time Use Survey (Enquête Emploi du Temps, henceforth EDT), carried out by the National Statistical Office (INSEE). The primary source of the British data is the 2000–1 UK Time Use Survey (UKTUS) (Ipsos‐RSL and Office for National Statistics 2003). The EDT samples 8186 representative households; the UKTUS samples 6414 households.
Each of these surveys collected three types of questionnaires: household questionnaires with household‐specific information such as household composition and location; individual questionnaires with individual‐specific information such as age, education and employment; and individual‐specific 24‐hour time diaries. For the time diaries, individuals were asked to use their own words to complete a written diary of their activities for each of 144 ten‐minute intervals. These activities were then recoded into approximately 140 standardized activities. One advantage of both of these surveys is that time diaries were collected for each adult in the household and, for the most part, all household members filled out diaries for the same day. In the case of the French data, diaries were collected for only one day—either a weekday or a weekend day. In the case of the British data, diaries were collected for both a weekday and a weekend day.
Both samples are restricted to heterosexual couple households. The British sample is further restricted to exclude those households residing in Northern Ireland and those households that do not have individual‐level surveys from both partners. These UK‐specific restrictions are substantial (causing a 25% drop in the sample size), but necessary to identify key covariates such as education. These restrictions yield samples of 5287 households in France and 2893 households in the UK. Using individual information, we restrict the sample to couples in which both partners are between the ages of 20 and 59 inclusive, and in which neither partner reports being in school full‐time, in the military, on disability or retired. This yields samples of 3405 households in France and 1782 households in the UK. Finally, the sample is restricted to households that provide information on purchased services, households for which sufficient information is available to impute prices and households for which both partners complete at least 23 hours and report at least five different activity spells for each possible time diary on days that are not deemed ‘unusual’.b Our goal is to obtain reliable diary information for a normal day.a Our final sample consists of 2924 households in France and 1295 households in the UK.a
A key input to domestic services is household housework time. Our focus is on the time spent on routine tasks that individuals do not generally enjoy, that maids provide and that all households perform. These surveys are quite remarkable in that preference data on various housework tasks are available.a Figure 1 summarizes these data by country, gender and activity type. Panel A presents the results for the UK, and Panel B presents the results for France. The fraction that enjoys an activity in France or enjoys it very much in the UK is illustrated in white, while the fraction that is indifferent to an activity is illustrated in grey, and the fraction that dislikes an activity is illustrated in black. Those who enjoy an activity a little in the UK are captured in light grey. For both countries we have information on preferences pertaining to cooking everyday meals, shopping for food, cleaning, ironing, cooking for special occasions, gardening and home repair. For France we also have information on preferences for dishwashing. For the UK we have information for laundry and shopping for non‐food items.
Open in figure viewerPowerPoint
Preferences regarding housework activities by country, gender and activity type.
The results indicate that substantially fewer people derive much pleasure from cleaning or ironing as compared to cooking for special occasions, gardening and home repair. Less than 40% of British men and less than 50% of British women report enjoying the former tasks, whereas between 60% and 80% report enjoying the latter. This distinction is even clearer from the French data. Less than 20% of the French report enjoying cleaning or ironing, while over 70% report enjoying cooking for special occasions, gardening and home repair. Preferences for dishwashing, available only from the French data, and for laundry, available only for the British data, indicate that these tasks are about as enjoyable as cleaning and ironing. Preferences for everyday cooking suggest that British individuals rather enjoy it. While it would be difficult to say that French men and women enjoy everyday cooking, they are more than twice as likely to say that they enjoy it as they are to say that they enjoy cleaning, dishwashing, or ironing. Neither the British nor the French report enjoying food shopping, but Figure 1 shows that British women enjoy ‘other shopping’. As the French time‐use data do not distinguish between food and other shopping, we exclude shopping from our analysis and focus on cleaning, dishwashing, laundry and ironing.a
This definition of domestic services is further justified as these are tasks that maids services generally provide. The British survey includes information on the type of tasks for which households hire market labour, distinguishing among ‘food preparation’, ‘cleaning, tidying up’, ‘ironing’, ‘shopping or errands’ and ‘household accounts’. The vast majority of purchased aid is for cleaning/tidying up activities, with ironing the second most common. None of the other services is purchased by even 1% of the sample. Meals can instead be purchased ready‐made, significantly altering preparation time, and indeed evidence from the UK Family Expenditure Survey 2000–1 suggests that the vast majority of couple households do take advantage of such opportunities. Information on this alternative input to meal preparation is not available in the data used here, providing further support for excluding meal preparation from this analysis.a
Other activities are excluded because they are not performed in every household. The additional housework necessitated by the presence of children is reflected in the demand for inputs to domestic services. This increased demand should be accounted for in our model as we control for the presence of children of different ages. Other childcare (such as feeding, dressing and playing) is excluded from our analysis because this task is child‐dependent, because there is evidence that the determinants of childcare and housework are quite different (Kimmel and Connelly 2007) and finally because private childcare services are not captured in our measure of maids services. Activities contingent on home ownership (such as lawn care, home repair and gardening) and pet ownership (pet care) are excluded as well, though we do conduct some sensitivity analysis along this dimension. Cleaning, dishwashing, laundry and ironing constitute the focus of our analysis and in fact correspond to activities that are often defined as ‘routine’ or ‘compulsory’ housework by sociologists (Presser 1994)