Asymmetries and interdependencies in time use between Italian parents

Abstract

In recent decades, changes in parents’ attitudes towards the importance of spending time with children to optimize their future behaviour and cognitive development have greatly affected patterns of time allocation among both working and nonworking parents in all developed countries. We compare the two waves of the Italian Time Use dataset (1988 and 2002) to analyse how family time allocation changed over time in a country that was undergoing a marked increase in female employment rate and a continuous decline in total fertility rate. We focus especially on how parents’ time with their children depends on their employment status and household characteristics. Using a simultaneous sequential approach, we consider links among the different time uses of individuals and correlations with spouses’ decisions. We find that wives’ time at work time strongly influences the time spent by both spouses with their children in 2002, but not in 1988. Fathers were much more involved in childcare and rearing in 2002 than in 1988. In general, as women's work time increased, substitutes for their childcare time were found within the household (fathers or other co-resident adults).

Keywords:

• time use
• parents
• children
• mother's work

JEL Classification::

• D13
• J13
• J22

Notes

¹ According to research by the Fondazione Rodolfo Debenedetti, 35% of Italian households think that very young children are better staying at home with relatives than going to a creche (Boeri et al., 2005).

² See OECD (2008).

³ Ichino and Sanz de Galdeano (2004), Mencarini and Tanturri (2004), Anxo et al. (2007), Tanturri and Mencarini (2009) and Bloemen et al. (2010).

⁴ They instrument childcare time using the previously predicted working time.

⁵ Leisure is the residual time category.

⁶ Given the types of wealth controls available in our dataset, we believe that Y directly affects only the amount of time dedicated to paid work. It is, in fact, unlikely that families rely on wealth to pay for housework or child care substitutes. Generally, the decision to use housework or childcare services depends more on income, i.e., mainly paid work. Nevertheless, we run a sensitivity analysis including the wealth controls in all our equations. Results are robust and the wealth variables included in Y were never significant in the nonwork time use equations.

⁷ In the empirical analysis, we disregard the price of market substitutes for home production, since they are not known.

⁸ The time diary of very young children was completed by parents.

⁹ The oversampling of weekend diaries was a deliberate choice of the data collector (ISTAT).

¹⁰ For example, someone may be cooking and watching television or cooking and looking after the children. It is the respondent who chooses which activity is the main one and which is the secondary one.

¹¹ For example, in the 1988 survey, almost no one reported childcare as a secondary activity, whereas in 2002 parents often did.

¹² In 2002, we were also able to exclude couples in which the mother is on compulsory maternity leave.

¹³ It is likely that in these types of household, in fact, time allocation process works differently.

¹⁴ 69 observations in 1988 and 288 in 2002.

¹⁵ The effect of the sample selection on the variables most relevant in the analysis is not presented. It is available from the authors upon request.

¹⁶ An individual is classified as ‘working’ when he/she declares to be employed.

¹⁷ A well-known problem in time use studies is that the time diary reflects a one day time allocation and it is possible that individuals do not engage that day in some activities they normally do. For example, the ratio of employed individuals who declare a positive amount of working hours is less than 100% for both men and women.

¹⁸ We tried to control also for the presence of sick adults within the household. In general, sick adults play a competing role with children for the wife (but also for the husband) time, since they need care for themselves, and their care might also require additional expenses. However, in 1988 the sickness status is based on a question that asks if the individual is chronically ill, while in 2002 is based on a self-reported variable on the individual health status, with possible answers that varies from very good to very bad. Therefore, even if our results were robust to the inclusion of the dummy ‘sick adults’, we were not convinced by the comparability of the two definitions and therefore we preferred not to include the variable in the estimates presented here.

¹⁹ Since it is a self-reported variable, it depends crucially on individual beliefs and it is likely to be downward bias and centred around the mean (as it is). Nevertheless, we think that those individuals who report to be poor or really poor are likely to be families that suffer from some kind of real economic constraints.