Data Descriptions

III. Data Descriptions

The main data used in this paper comes from the 2003-2010 American Time Use Survey (ATUS). The participants in ATUS, which include individuals over the age of 15, are drawn from the existing sample of the CPS, and they are surveyed approximately 3 months after completion of the final month of the CPS survey.⁶ The ATUS asks the participants to recall the activities done in the previous day, so it allows one to measure the amount of time people spend doing various activities, such as paid work, childcare, socializing, etc.⁷

6 The number of household sampled in the ATUS is around 13000 each year, though in 2003 there are around 20000 households in the sample.

It also ask questions about the specific person(s) around when doing each activity, so in our case, we can infer which child the parents are taking care of in each childcare-related activity.

7 The “previous” day is defined as from 4am on the previous day to 4am on the survey day.

6

In this paper, we follow Guryan, Hurst, and Kearney (2008) and define children to be family members below 18. We consider four types of childcare time: basic care, education, playing, and traveling and waiting, and the total time spent on these four types of activities constitute the total childcare time. Descriptions of the childcare activities covered are detailed in table A1 in the appendix.

Table 1 summarizes the time spent on each son or daughter among parents. Here, our sample only includes parents who have at least one child aged 18 or less within the household.

As we can see from table 1, mothers on average spend much more time (around 137 minutes per day) on childcare then fathers (around 75 minutes per day), and the amount of time they spent on basic care, education, and traveling and waiting basically doubles that of fathers. The sole exception is time on playing, where we see little difference between mothers and fathers.

Table 1 also gives information on the time parents spent on children of different genders. As we can see, in total fathers significantly spend more time on sons than daughters, where the difference arises from time on basic care and playing. However, mothers’ childcare time does not seem to differ among sons and daughters.

Table 2 further summarizes childcare time among parents of different education attainments. Here, as in previous studies such as Guryan, Hurst, and Kearney (2008) and Ramsey and Ramsey (2010) we find a positive relationship between parents’ education attainments and childcare time, and this positive relationship holds among both mothers and fathers, and also among all four types of childcare activities. Again, we find that, except for time on playing, mothers’ childcare time doubles fathers’ childcare time.

As this paper is to study the relationship between a state’s industrial structure and parental childcare time, we combine employment information from the 2000 Census and the occupational characteristics from the occupational information network (O*NET) to calculate

7

the share of labor intensive jobs in each state.⁸

The O*NET provide information on the importance and the requirement of different abilities in each occupation, and here we specifically focus on the physical abilities required in occupations.

As described earlier in the introduction, we focus on labor intensive jobs because these jobs are more “localized” in the sense that they are usually held by local low skilled incumbent residents, who have been found in the literature to have a much smaller migration rate. Therefore, we hypothesize that parents living in states with more labor intensive jobs are less adamant about their children’s education outcomes because they know that even if their children do not do well in school, they still can easily find local labor-intensive jobs.

9

From the first column in table 3, we find large difference in industry structure across states. Specifically, we find the three states (Arkansas, Wyoming, and West Virginia) with the largest share of labor intensive jobs have over 35% of their jobs been classified as labor intensive, while the three states (District of Columbia, Maryland, and New Jersey) with the smallest share of labor intensive jobs only have around 25% of their jobs been classified as labor intensive. In the second and third columns in table 3, we generally find that the ordering of states in terms of its industry structure does not change when we redefine labor intensive jobs as those occupations with top 1/4 or 1/5 scores in physical requirement. Hence, in our empirical study we simply focus on the result where labor-intensive jobs are defined as occupations with top 1/3 physical requirement scores.

As the O*NET identifies 9 types of physical abilities and give scores on each of them, we take the simple average of the scores to represent the physical requirement for each occupation. Based on these scores, we identify the occupations with the top 1/3 scores and label them as labor-intensive jobs. We then calculate the share of labor intensive jobs in each state and the results are given in table 3.

8 Census Information is obtained from the Integrated Public Use Micro data Series (IPUMS). The O*NET information is obtained directly from its website: www.onetonline.org.

9 There are four major categories of abilities: cognitive, physical, psychomotor, and sensory.

8