The Importance of Time-Use Data
There is a wide range of potential uses of data on how Americans spend their time, including understanding the effects of public policies on individual behavior. For example, low-income workers are sometimes eligible to receive subsidized child care. Time-use data can help in understanding how these policies affect the amount of time that parents spend working at home or outside the home and how much time they spend with their children. Aside from public policy uses, time-use data can improve our understanding of individual and household behavior, especially with respect to time allocation decisions and in improving our knowledge of the well-being of the nation. In this chapter, important functions of time-use data for informing public policy and for better understanding of behavior and well-being are discussed.
TIME-USE DATA FOR PUBLIC POLICY
This chapter focuses on five potential ways in which time-use data can be used for public policy: (1) to expand the national economic accounts; (2) to understand the transition from work to unemployment (and vice versa) and from work to retirement and the time spent working for pay during “retirement”; (3) to document time spent in market, nonmarket, and leisure activities; (4) to document and understand decisions that individuals make about how much time they spend caring for children and for other family members; and (5) to understand the effects of recent major changes in social welfare programs. Linda Waite, Thomas Juster, Sandra Hofferth, and Steven
Landefeld presented papers that covered these issues. This section looks first at the national income accounts, issues of work and retirement, child care, and welfare reform and then considers issues of well-being more broadly.
Augmented National Economic Accounts
A primary public policy use of time-use data is to enhance the coverage of National Income and Product Accounts (NIPA). The NIPA accounts provide measures of economic activity for the nation and are the principal means of measuring growth in the nation’s economy over time and in comparing income and production across countries. The NIPA accounts almost exclusively measure only market production and, hence, do not take into account goods and services—such as the production and consumption of meals at home—that a household produces for its own private consumption or for any household production that is not traded in the formal market but is consumed by others, such as giving a neighbor fresh tomatoes from your garden.
A goal of the accounts is to comprehensively tabulate all economic activity in the nation (Landefeld and McCulla, 1999). The exclusion of nonmarket production has been noted for many years.1 Nonmarket production has not been included in part because there are conceptual and practical issues in measuring these activities. Conceptual issues include classifying a nonmarket activity as a productive activity, valuing the output produced, and valuing the time inputs needed to produce it. (These issues are discussed further in the Chapter 3 of the report.) A practical issue that is a barrier to measuring nonmarket output in national accounts is the lack of consistently and regularly produced data on how much time is spent in nonmarket activities. Landefeld and McCulla (1999:27) argue that the “greatest barrier to constructing a consistent set of time series on the value of household production is the lack of consistent data on time-use, both for current and earlier periods.”
What policy questions could be better informed if time-use data were available to use in the NIPA accounts? Participants at the workshop emphasized the following areas:
What have been the trends in the number of hours worked? Are data on hours for which pay is received a good reflection of true hours worked? Are measures of labor productivity reliable?
How much of the historic increase in the U.S. gross domestic product
(GDP) is due to the increased labor force participation of women and how much was in fact offset by a reduction in nonmarket activity?
How much household time use should be classified and measured as “investment,” such as time spent helping a child with homework as an investment in the child’s future well-being, and how does that factor into national accounting?
How do the national income levels of less developed countries compare with those of more developed countries when nonmarket time is accounted for?
How do tax policies and other government policies, such as subsidized child care and education loan and tax breaks, affect labor market and nonmarket time use?
Significant efforts to include nonremunerated work into national income accounts are under way in other countries. Estimates of household production output and the inputs used for the outputs have been made in Australia, Canada, and three Scandinavian countries—Finland, Sweden and Norway (Ironmonger, 1997). The Bureau of Economic Analysis in the U.S. has also produced household output and input tables (Landefeld and McCulla, 1999), but these estimates are not currently used in “core” GDP figures. Rather, the estimates are used in satellite accounts to the national income accounts, which measure production typically not included in the standard set of national accounts.
Results from the Landefeld and McCulla study provide a valuable example of how time-use data can be used. Using data from the 1980s, the study adjusts GDP from 1946 to 1997 by including measures of nonmarket household production and by counting household expenditures on consumer durables as investment. One particularly interesting result is that the growth in total nominal output for 1946-1997 would be estimated at a 7.1 percent annual rate instead of the official 7.3 percent annual rate when nonmarket production is included and household durables are treated as investment in GDP. The authors argue that the lower growth figures reflect a decrease in nonmarket production over the five decades largely due to an increase in female labor force participation. The GDP in 1946 increases by 43 percent when household production is included, but only by 24 percent in 1997. When expenditures on household durables are treated as investment, GDP increases by 5 percent in 1946 and by 8 percent in 1997.
In the future, more countries will be producing such satellite accounts. Perhaps the biggest boost to these efforts will come from the Eurostat harmonized time-use survey involving 18 countries, which will be used to produce such tables. As more and more countries develop the data for measuring nonmarket production, methods for dealing with some of the conceptual
issues in measuring nonmarket production will evolve. In addition, the value of the data for making cross-country comparisons will increase.
Work and Retirement
Time Use at Work
Time-use data can be used to improve measures of how time is spent at work or while working for pay and to understand the effects of public policy on labor market and job outcomes. Technological gains are allowing more work to be done away from the office and have contributed to the blurring of the lines between work in the market, nonmarket work, and leisure. Time spent at the “workplace” may not entirely consist of time spent in market work; it may also include time spent in nonmarket work or leisure. Likewise, time spent away from the “workplace” or at home may include market work time. Conventional measures of time spent at work—which are usually collected through recorded hours of work from company or organizational payroll records or through stylized questions asking the amount of time respondents typically spend at work—are unlikely to fully capture the blurring of these lines and do not provide detail of what “work” is being done. Such information is important because better data on time use while on the job can help improve productivity measures and can contribute to understanding how technological innovations have affected productivity.
The Transition from Work to Unemployment and Retirement
Another area of public policy for which time-use data can be used to better inform policy making is in understanding transitions between paid work and nonmarket work, volunteerism, unemployment, and retirement. This includes an individual’s choice of how many hours to spend in paid work and other activities and how public policy affects individuals’ use of time, in contrast to measures of the aggregate levels discussed in the previous section. Some of the policy questions raised in the previous section apply here as well. For example, one question is how tax and employment policies, such as the Earned Income Tax Credit, family leave policies, and subsidized child care, affect household decisions regarding how much to work in the home or in the market for pay.
The tradeoff between household work and market work has been studied by economists, sociologists and policy analysts for a long time and there is an extensive literature on the topic. There are, however, other alternatives to household work (besides leisure) for those who do not work full time for pay, as workshop participants Thomas Juster and Linda Waite discussed. One of these is volunteerism, which is of particular interest for understanding how
older and retired adults spend their time (Hill et al., 1999). Of policy interest here are how wage, tax, Social Security, and other policies affect volunteer time, how volunteer time translates into the production of goods and services, the social or community capital that is built from volunteerism (see Juster, 1999; Smeeding, 1997; and Hill et al. 1999), the personal satisfaction and health benefits that accrue to the volunteers, and how policies might affect the amount of volunteering.
Nonmarket activity also includes educational or training activities, either fulltime or while working. Public education from kindergarten through high school (K-12), subsidized public colleges, and federal and state grant and loan programs for college students and educational tax credits are all publicly funded investments to develop skills and to train current and future employees. How policies affect time spent in these activities is an important question on an individual level. Discussant Suzanne Bianchi also noted that examining how elementary and secondary school children spend their time in educational activities is a key to understanding educational outcomes, both on an individual level and at the school level. For example, a study of children’s time use at school could provide information on how different classroom settings and schedules affect the cognitive development of children. Measures of time spent in learning and training activities outside the formal educational system could also prove useful. The productivity outcomes of time spent in on-the-job training might be used to better inform learning and training policies for workers. Each of these kinds of studies would require longitudinal data on time use and extensive data on the characteristics of respondents. Throughout the entire formal educational system, it is also important to understand how time investments pay off in terms of future earnings and productivity gains on an aggregate level, as Dale Jorgenson pointed out in his discussion. Time-series data on time use in educational activities are required for this type of research.
People who are not in the paid work force may be unemployed and looking for work. Time-use data can help researchers and policy makers understand how much time unemployed persons spend looking for jobs, how much they work in the informal sector, and in general, how they spend their time while unemployed. Of policy interest is how unemployment benefits and the timing of benefit coverage affect these decisions.
The time use of those who are disabled is also of policy interest, as Joseph Altonji pointed out in his discussion. The types of policy questions discussed include:
Has the Americans with Disabilities Act increased work time for the disabled?
Have public accommodations for the disabled increased their work participation?
How do those who receive disability payments or Supplemental Security Income (SSI) payments spend their time?
Each of these policy questions could be informed with data on the time use of disabled people.
Time-use data are also of policy interest for understanding transitions to retirement and how retirees spend their time. As Americans live longer and as some Americans retire earlier, it becomes important to know what retirees do with their time (Waite and Nielsen, 1999). When an individual retires, he or she may still work for pay or engage in unpaid work activities. Retirement for many may actually mean a change in a career, a move to part-time work for pay, or work as a volunteer. How social security, Medicare, and other policies for older adults affect these decisions is of particular interest for policy development.
Child and Family Care
For many people, a primary component of nonmarket work is time spent caring for others. This is especially true for parents with young children. For many elderly couples, one spouse often needs assistance or care that is often provided by the other spouse. Likewise, the children of elderly parents often provide care for their parents. Workshop participants discussed several policy considerations that could be informed by data on time spent in care-taking or care-receiving activities. One example is how subsidized child care for low-income families affects parents’ time spent working for pay. Another example is how child outcomes (educational achievement or test scores, for example) are affected by the time parents spend with children (Hofferth, 1999).
Also important to policy is the degree to which individuals substitute their own time caring for relatives with the time of market-provided care-givers and what factors determine how much time is spent caring for a relative or spouse. A related issue is whether the health and general well-being outcomes of those who are receiving the care are better when the care is provided by a relative as opposed to when care is given by a market provider. In her discussion, Rebecca Blank identified several other issues. For example, the question of how the care that older adults receive from their spouse or another relative interact with policies of the health care system is relevant for policy discussions about giving tax breaks for those who care for relatives. Similarly, as different types of health care systems are debated, it would be useful to know how different health institutional structures support family care.
The wide-sweeping changes to the main federal cash assistance program to low income families with children in 1996 have many implications for how recipients of cash assistance spend their time. In contrast to the old system, the new system requires most of those receiving assistance to engage in work or work-related activities in order to receive assistance. How current recipients, former recipients, and potential recipients spend their time and how this relates to their outcomes (earnings, program participation, and well-being) and their children’s outcomes are major policy questions. Of particular interest is what these families are doing to support themselves in terms of market and nonmarket activities.
An ethnographic study by Edin and Lein (1997) indicates that under the old program rules, many welfare recipients worked both in the formal market (although earnings from such work would reduce benefits) and informally (where work income was supposed to be, but was not always, reported). The incentives for devoting time in the formal labor market have changed under the new program rules. The entitlement for cash assistance was eliminated and most recipients must now participate in work or work-related activities to receive assistance. Time-use data can help policy analysts understand behavioral responses to the new incentives. As Rebecca Blank stated at the workshop, an important part of evaluating welfare reform is understanding how the new work requirements affect a segment of the population that previously received public assistance and worked less than they now do.
TIME-USE DATA FOR UNDERSTANDING WELL-BEING
Collecting time-use data on how retirees spend their time and on how much time Americans spend in educational activities or in volunteer activities need not be justified only for policy purposes. It is also important to know how Americans use time in order to have a better understanding of the well-being of the nation, including the degree to which people feel time-crunched or experience stress due to having too little time to do the things they want to do. This section discusses how time-use data can improve understanding of well-being in the United States.
Recently there has been some debate on whether Americans are working more hours than they have previously and whether Americans are spending fewer years in the work force than they did in the past. Part of the reason this debate has not been settled is that there are inadequate data on the number of hours that people work for pay (Smeeding, 1997). Data on how Americans use their time are not produced regularly and have not been produced since
1985. Interesting questions could be addressed with such data. For example, as the American population ages, it would be interesting to track changes in time use. If Americans are spending fewer years in the labor force, is it because people retire earlier, as some workshop participants suggested, or is it because more and more young adults are going to college, delaying full-time entry into the paid labor force for several years, and because many middle-aged Americans are going back to college or receiving additional training, which takes them out of the full-time labor force, as other participants suggested?
Collecting time-use data will augment knowledge on these types of trends. Time-use data will also help researchers and policy makers understand what retirees do when they leave the labor force and can provide data on the educational and training activities in which individuals participate. It will also be useful to know how time use varies over business cycles as the unemployment rate rises and falls. Measures of time use on the job would be extremely useful for research on labor productivity. Most broadly, time-use data will be valuable for describing what people do when they are not at work.
Time-use data can provide interesting assessments of noneconomic measures of well-being. In past time-use studies, respondents have been asked to describe their satisfaction levels from different activities and their emotional states during those activities. These subjective measures of intrinsic satisfaction associated with time spent in different activities can be used to better understand well-being. Time diary studies and experiential sampling method (ESM) studies (both are described below) have both been used to better understand subjective satisfaction from work, leisure, and other activities.
The hours during which work activities take place may also affect well-being and have implications for quality of life. Daniel Hamermesh, in his discussion, showed results from his recent study examining the hours of the day during which people work and how that has changed over the past 25 years (Hamermesh, 1999). In comparison with 25 years ago, he found that workers are working more during the middle of the day (between 6 a.m. and 6 p.m.), than during evening or nighttime hours. Many would argue that a movement away from working during the middle of the night towards working standard hours is an improvement in the quality of life. This finding is one example of how time-use data can improve the richness of measures of well-being.
The growing disparity in income and earnings across the population has received a great deal of attention in policy and research communities. One aspect of well-being that is not usually a part of these discussions is whether there is a large disparity in the amount and timing of leisure. For example, one person may be “money rich, but time poor,” a phrase used to describe those who are monetarily wealthy but have little time away from work to devote to leisure. Another person may be “time rich, but money poor,” a
term used for those who have more leisure time and relatively less money.2 While standard economic measures of well-being would classify the first individual as better off, if differences in leisure time are counted, the first individual may not look as well off. Time-use data can at least help describe who has leisure time and what hours they have it.
Related to the issue of well-being is the issue of the “time crunch.” The time crunch is generally used to describe the condition of those who feel as if they do not have enough time to do the things they need and want to do. For many individuals, the time crunch raises the level of stress in their lives, which may in turn have negative effects on their physical and mental health, work performance, or family relations. Others may not experience such tangible negative effects of the time crunch, but may instead feel that their quality of life is not as good as it was or could be. Workshop participants argued that understanding these trends in how people feel about their use of time or lack of time for leisure activities can give a more complete picture of the quality of life and changes in the quality of life in the United States. Time-use data with information on perceptions of time use (for example, how quickly time passes or how stressful an activity was) and on time spent in activities can help analysts decipher what activities or schedules make people feel time crunched or what activities are taking up time.
These are some of the major uses of time-use data for informing public policy and for furthering knowledge of the well-being of the nation. There are, of course, many other public policy questions that could benefit from time-use data, as well as many private uses of time-use data. For example, marketers would like to know when certain demographic groups are watching television or listening to the radio. Businesses may want to keep on top of consumer shopping trends (e.g., the time of day or the day of the week consumers typically shop, by demographic category). Also, as Rebecca Blank pointed out, employers could use detailed time-use data on how their employees use their time when setting personnel policies.
Workshop participants emphasized the potential applications of time-use data for understanding behavior and well-being and for informing public policy. There are, however, limitations of time-use data for these purposes. For example, time has shortcomings as a metric as it is not easy to put a value on time.3 Individual skills, ambition, and intelligence determine how productive a person is in different activities. Measuring time spent in activities is subject to these differences in productivity (i.e., some people are more pro-
ductive than others in a given period for a given activity). Furthermore, classifying activities in which people spend their time can be difficult, and it is often difficult to classify, measure, and value the outputs of these activities. These limitations present conceptual and measurement challenges for time-use researchers. These challenges arose during the workshop discourse and are discussed in the next chapter of the report.