Summaries of Workshop Papers
The Decision to Allocate Time Between Market and Nonmarket Activities
Time Use by and for Older Adults
Accounting for Nonmarketed Household Production Within a National Accounts Framework
Methodological Features of the Time Diary Method
An International Perspective to Collecting Time-Use Data
TIME-USE DATA: ANALYTIC FRAMEWORK, DESCRIPTIVE FINDINGS, AND MEASUREMENT ISSUES
F. Thomas Juster
University of Michigan
This paper provides an overview of time-use research. It starts with a brief description of an ambitious social accounting system that has time use as its core, moves to a description of important scientific and policy issues that can be examined with time-use data, explores differences among countries in two key areas of time-use research (labor/leisure choices and investments in children), examines alternative methods of measuring time use and concludes with a discussion of the optimum sample design for the collection of data on time use.
The unifying analytical framework of a time-use based social accounting system is the notion that the ultimate constraints on individual and societal change can be found in the availability of human time and the stock of wealth inherited from the past. Human time can be allocated to the market; to nonmarket production (cooking, cleaning, child care, etc.); to leisure activities (television viewing, socializing, etc.); or for biological maintenance functions (eating, sleeping, etc.). The outputs associated with these inputs of time are of various sorts: command over market goods and services is the output of time spent working for pay; nonmarket outputs such as orderly houses, well- or ill-behaved children, and gourmet meals are the outputs of nonmarket work in the household; improved health, skills, or stocks of information are additional nonmarket outputs; and direct enjoyments or satisfactions from the activities themselves are the final outputs of the system.
The role of capital stock in this view of the generation of well-being is crucial and has a rather unconventional flavor in comparison with the usual economic meaning of capital. Capital stock refers to a very broad range of settings that have the effect of conditioning the outcomes from the use of time in particular activities. This capital stock includes not only the effect of tangible assets such as factories, machinery, houses, cars, and other consumer durables, but also such tangible and intangible factors as human skills and knowledge, networks of personal associations, environmental assets like climate and water quality, sociopolitical assets such as the representational or the judicial systems, etc. In short, this framework regards capital stock as an appropriate term to describe a broad range of factors that condition the results of applying human time to various activities.
The paper highlights at least five distinct areas in which time-use data can make a substantial contribution to our understanding of the way in which economic and social systems function: (1) improving our understanding of
the change over time in work and leisure activities and of the distribution of work and leisure between household members; (2) assessing the level of overall well-being; (3) understanding differences in economic systems characterized by different institutional arrangements; (4) examining a much broader definition of societal investments in the future, particularly investments devoted to the education and training of children; and (5) achieving a better understanding of distribution issues.
One of the most obvious benefits from the availability of time-use data is that it enables us to analyze a much more useful concept of work and leisure hours than the usual analysis of total available hours less paid work hours. Work hours as defined by a time-use study not only includes work for pay in the market, but also unpaid household work, commuting time to and from work, and work hours masquerading as leisure hours (shop talk over dinner) and leisure hours masquerading as work hours (long coffee breaks at work discussing fishing and golf).
The paper discusses methods for collecting time-use data and their strengths and limitations. The paper also discusses some sample design issues for time diary studies. It is ordinarily true that sample design issues are relatively straightforward and constitute the least problematic of survey design issues generally. But in the case of studies with time dairies, sample design issues are critically important and extremely complex. There is substantial room for disagreement about characteristics of an optimum sample design. The choices basically consist of the following:
Select a random sample of households, select a random person within the household (in a multiperson household), and collect a single time diary from the randomly selected person within the randomly selected household on a randomly selected day of the year.
Select a random sample of households, but in multiperson households, collect a time diary from all eligible persons in the household on a randomly selected day of the year.
Select a random sample of households, collect time diaries from all eligible persons in that household, and collect multiple time diaries for each eligible person: an obvious choice for the number of diaries to be collected for each eligible person would be four—two weekdays, one Saturday, and one Sunday, spread randomly over the course of the year.
Which of these designs is most appealing depends in part on the analytic objectives of the study, but also on considerations of statistical noise, interclass correlation, and relative cost. The only systematic study of this topic concluded that the optimum design for maximizing the effective sample size for a given budget was a design that included two weekdays for each respondent, plus one Saturday and one Sunday, and also included spouses of re-
spondents in married-couple households. In this analysis, no account is taken of any advantage accruing to analysis of microlevel data because the multiple observations per respondent design reduces the level of statistical noise in the time diary estimates. Thus, the conclusion that multiple observations per person are better than single observation from the perspective of minimizing sampling error is even more attractive if we take into account the reduction in statistical noise by collecting multiple diaries per respondent.
Robert A. Pollak
Washington University, St. Louis
This paper provides a theoretical framework for estimating structural and behavioral relationships with time-use data. Four components of this framework are identified and discussed: technology, preferences, intrahousehold allocation and marriage market sorting. Starting with Becker’s (1965) theory of household production, where households allocate their time (towards both market and nonmarket activities) to produce commodities that are then consumed and yield “utility,” the paper describes these four components. Households use the technology available to them to produce the commodities. Individual preferences of household members also play a role in determining what a household produces, how the household spends its time producing, and how much is produced. The theoretical framework also covers how households allocate time towards producing commodities and how the commodities are allocated to household members. Finally, the paper describes a theoretical framework for how individuals sort into marriage (and form a multiperson household).
The paper also discusses implications of the joint production of commodities to the theoretical framework. An example of joint production is cooking a meal, which produces not only the meal, but also, if a person enjoys cooking, “process benefits” (intrinsic rewards, to use Juster’s terminology). If so, then two goods are produced—the meal and the increase in satisfaction from spending time cooking the meal. This has implications for the theoretical assumptions of the production process.
Unlike the original theoretical framework laid out by Becker, this paper argues that households do not have preferences and utility functions, but that individuals within households have preferences and utility functions. The time allocated to household production and the benefits from household production are allocated within the household based on bargaining between household members and each member’s preferences. The paper discusses the implications of this under different types of household preferences, such as
interdependent preferences, when one household member’s preferences are dependent on another household member’s preferences, time use, and consumption, and preferences that differ across gender.
The paper also examines an application of the household production theory. The application discussed is how parental time spent with children translates into the production of a child’s well-being and future well-being. The paper highlights the difficulty in empirically estimating this relationship, arguing that it is unlikely that a model could control for all determinants of the child’s well-being (however measured) to estimate the relationship between the time parents spend with their children and the production of the child’s well-being.
The paper briefly discusses how simultaneous activities (doing two things at once) should be treated in time-use studies. The paper proposes that activities be recorded as compound activities, so that time spent reading a book while flying in a plane is one category of activity (as opposed to two categories—reading and flying). As the paper points out, the problem with compounding activities is that the number of activities grow rapidly out of control. However, it is argued that a limited number of compounded activities that are particularly relevant for policy purposes could be accounted for in such manner on a survey. The example of being on call for child care duties (such as cooking while caring for a sleeping child) may be one such compounded activity.
THE DECISION TO ALLOCATE TIME BETWEEN MARKET AND NONMARKET ACTIVITIES
Linda J. Waite and Mark Nielsen
Center on Parents, Children and Work and Alfred P. Sloan Working Families Center University of Chicago
This paper examines the allocation of time by individuals and households between work in the market and nonmarket and household work. It traces labor force participation for men and women, married and single, over time. The paper also examines reasons for nonwork given by those who are not working and their reported sources of financial support. Joint labor force and nonmarket work statuses for married couples are also presented. In addition, the paper simulates husbands’ nonmarket work activities in response to their wives working more hours and in response to working more hours themselves. Finally, the paper examines differences in lifetime market work decisions.
The paper begins with a summary of labor force participation rates for
both men and women between 1962 and 1997. In this period, the percentage of men not in the labor force tripled, but remained modest (under 10 percent). However, labor force participation for women increased dramatically, from 42 percent in 1962 to 73 percent in 1997. The paper also describes the self-reported reasons for not working by those who were not working in the market. One noteworthy result is the increase in the number of nonemployed men who reported they are not working for pay in order to care for their family or home: in 1962, it was 1 percent; in 1997 it was 8 percent. For those who are out of the labor force, sources of support were also described. Of single women who are not working for pay, most reported that they received welfare or worker’s compensation as a means of supporting themselves. Of single men who did not participate in the labor force, most reported receiving Social Security or worker’s compensation as their means of support.
The paper also describes the number of hours that men and women devote to market and nonmarket work per week, using data from the National Survey of Families and Households. Women reported spending 18 hours per week (on average) in market work, while men reported spending 32 hours per week in market work. The average hours per week spent in nonmarket work is almost reversed by gender: women reported working 33 hours per week, and men reported working 18 hours per week.
The time that married couples spend in market and nonmarket work is also explored in the paper. A simulation is conducted of the responsiveness of husband’s time spent doing five household chores (shopping, dishwashing, laundry, cooking, and cleaning) as the amount of time his wife spends working for pay increases. The simulation examines how the amount of time the husband spends in these five household chores varies as his wife’s hours of time spent in the labor force go from zero to over 60 hours per week (holding the husband’s time spent working for pay constant at the average for all men). Similarly, the responsiveness of husband’s time spent doing household chores as his own time spent working for pay varies from zero to over 60 hours per week (holding constant the time his wife spends working for pay at the average for all women) is also simulated. Results of this simulation find that the amount of time husbands spend in nonmarket work changes less as the amount of time he spends working for pay varies, than when the amount of time his wife spends working for pay varies.
Finally, the paper discusses time spent in the labor market over a lifetime. The study finds that men over the age of 65 are working less now in comparison with the 1960s. The paper argues that further research on lifetime work decisions should be conducted.
FAMILY READING TO YOUNG CHILDREN: SOCIAL DESIRABILITY AND CULTURAL BIASES IN REPORTING
University of Michigan
This paper compares measures of the time parents spend reading to their child gathered from single-item (stylized) questions and the time reported from time diaries filled out by the parents for the child. The importance of reading to young children in promoting language proficiency and literacy has been documented in many studies. Some data has also shown that children with more educated parents are read to more often than children with less educated parents. The paper further explores parental reports of reading to children, specifically, how reports gathered through single-item questions compare with those gathered through time diaries. The paper hypothesizes that parental reports of time spent reading to children is exaggerated through single-item questions relative to time diary reports. Furthermore, the paper hypothesizes that because more educated parents are likely to be more aware of the benefits of reading to their child, any social-desirability bias in reporting will be stronger for more educated parents. The paper also hypothesizes that there will be racial and ethnic differences in reports of time spent reading to children.
The paper uses data from the Child Development Supplement of the 1997 Panel Study of Income Dynamics (PSID). The supplement collected data on 2,394 children between the ages of 0 and 12, randomly selected from PSID respondents. The survey collected time diaries of the children’s activities over a specified 24-hour period. The parent or caretaker of the child filled out the diary. One weekend day diary and one weekday diary was filled out for each child. The survey also used stylized questions on how often the parents read to the child: “How often do you read to (child)? Would you say never, several times a year, several times a month, about once a week, a few times a week, or every day?” The study compares responses to stylized questions on time spent reading to children with responses from time diaries on time spent reading to children.
Results show that 47 percent of children aged 3-5 were read to on a daily basis according to the stylized questions, but only 29 percent of the children were read to on a weekday or a weekend day according to the time diary reports. Holding the time diary as the standard, the paper concludes that parents exaggerate the number of times they read to their children on stylized questions. However, if reading reported through the diary on either a weekend day or a weekday is counted, 42 percent of children were read to, which is still under the amount reported from stylized measures.
The paper also regresses many demographic characteristics of the parents
on an indicator for whether or not the parents reported reading to their child on a daily basis. This logistic regression includes a measure for whether or not the child was read to on a diary day as reported in the time diaries. The education level of the mother had a strong and positive association with reporting daily reading, which was given as evidence to support the hypothesis that more educated parents were more likely to exaggerate how often they read to their children.
TIME USE BY AND FOR OLDER ADULTS
Martha S. Hill, A. Regula Herzog, and F. Thomas Juster
Institute for Social Research
University of Michigan
This paper identifies key research issues concerning the time use of older adults and the time others spend caring for older adults. Key issues include: (1) how older adults spend their time, (2) how much time family members and relatives spend caring for older adults, (3) paid and unpaid work by older adults, (4) the activities of older adults and the impact of those activities on health, well-being, and mortality, and (5) the scheduling of activities of older adults. For each of these issues, the paper summarizes the implications for data collection. The paper also describes several existing data sets that have time-use data on older adults. They include time diary studies in the United States and elsewhere as well as longitudinal surveys containing time-use information.1
One of the key issues in time-use data collection for older adults is accounting for intergenerational transfers of time. Traditionally, research on intergenerational transfers of resources focuses on transfers of money and goods. However, there are also often transfers of care from the adult children to their dependent parents. Of policy interest is the degree to which labor market decisions are entwined with decisions to provide care for an older adult. Also of interest is the degree to which care from family members or resource sharing with family members (sharing housing, for example) substitutes for public assistance or for nonfamily private care providers. For researching these issues, extensive data are needed about different generations
of extended families, including not only their time use, but also their sharing of other resources and their characteristics, needs and levels of resources.
Another issue relevant for time-use studies of older adults is the amount of time older adults spend in paid and unpaid work. Although many older adults are officially retired, they may still spend time working for pay. Older adults may also spend time in unpaid work, including household work. More broadly, it may also include time spent in volunteer activities and time spent caring for others. Older adults in good health often provide care for grandchildren. They may also be the chief providers for a spouse in need of care. The authors argue that a combination of stylized questions and time diary data are needed to study the work of older adults. The potentially sporadic and irregular nature of this work means that time diaries tend to yield more accurate measures at the aggregate level while stylized questions asking about work over a typical day or week allow more accuracy at the individual level. The open format of a time diary reveals potentially unknown types of activities; hence, stylized questions could be guided by time diary findings to focus on a wide range of work activities.
The paper also highlights the importance of understanding how the activities of older adults are associated with health, well-being, and mortality. The implication for data collection are that measures over time of both time use and well-being are needed. Time diaries are preferred over stylized diaries because they facilitate gathering contextual information on location and social partners in activities, as well as avoiding having to pre-specify the type of activity.
Finally, the paper hypothesizes that older adults’ schedules may be of both individual and public importance. Older adults usually have more flexibility in scheduling activities and so can take advantage of off-peak hours for such things as grocery shopping, use of roadways and public transport, and obtaining personal and household services. Older adults that are home during the day can also provide a sort of neighborhood watch. To explore these hypotheses, data are needed concerning the types and location of activities at different times of the day, on different days of the week, and in different seasons of the year.
In sum, the paper shows that the research issues regarding time use by and for older adults are diverse in their data needs. Some issues require stylized questions asking for typical amounts of time spent in pre-specified types of activities. Others require time diaries assessing not only the amount of time in activities but also social partners, location, and time of day. Some issues require data on all adults as well as older adults. Many require measures in addition to time use; some require panel data. The authors stress that it is important to match the methodological approach to the research issue and to consider further methodological development.
ACCOUNTING FOR NONMARKETED HOUSEHOLD PRODUCTION WITHIN A NATIONAL ACCOUNTS FRAMEWORK
J. Steven Landefeld and Stephanie H. McCulla
Bureau of Economic Analysis
This paper describes how time-use data can be used to produce estimates that account for nonmarket household production in the national income and product accounts. The paper discusses the structural framework of satellite accounts and how nonmarket production fits into that framework. It then covers valuation issues for including household production in national accounts. Estimates of outputs and inputs of household production are given, and their impact on gross domestic product over time is also estimated. Finally, the paper offers some areas of further research and discusses the policy implications of the estimates.
The national income and product accounts (NIPAs) record the present value of the amount, composition, and distribution of income generated from market transactions in the U.S. economy. The accounts do not include for household production or unpaid housework, partly because of the difficulty in measuring and valuing nonmarket production. Satellite accounts offer a way to show measures of production that are not included in the standard set of national accounts while maintaining consistency with them. In the satellite accounts presented in this paper, estimates of household production were incorporated into measures of gross domestic product (GDP) from 1946 to 1997. Some of the resulting impacts on GDP are highlighted here.
The adjusted measures show a slower overall growth in GDP over the time period—7.1 percent annually instead of 7.3 percent annually—as the adjustments raise GDP by 43 percent in 1946 but by only 36 percent in 1997. The decreasing impact of the adjustments is a result of more women working in the labor force and spending less time working at home in 1997 than in 1946. Conversely, including consumer durables as investment raises GDP by 5 percent in 1946 and by 8 percent in 1997, reflecting the increased reliance on improved technology and household appliances as labor shifted from the home to the marketplace. Savings measures also increase, due to the reclassification of consumer durables; as a result, there is a slowdown in the fall of the personal savings rate, from 7 percent to 2 percent over the period. These and other results, taken together, suggest that the pecuniary tradeoffs between market production and home-based production have been positive and that recent concerns over falling savings and investment rates, especially in relation to other countries, may be exaggerated.
The paper also conducts an input/output analysis of the household for 1992. This analysis allows a more detailed look at the composition of household production. For instance, households, as reflected by personal con-
sumption expenditures, have historically made up the largest portion of GDP. However, the expanded view in this paper reveals that households actually contribute more than just their final consumption. For instance, such expenditures on “cleaning, storage, and repair of clothing and shoes” is only $11 billion in 1992, while household laundry output is valued at $89 billion. The paper also reclassifies many household expenditures to illustrate that $2,596 billion (62 percent) of the conventional estimate for personal consumption expenditures of $4,209 billion in 1992 was actually spent on intermediate goods used in the household production process, and $471 billion was actually investment in consumer durables. In fact, only $524 (or 12 percent) of the conventional estimate of final consumption expenditures is actually final consumption.
METHODOLOGICAL FEATURES OF THE TIME DIARY METHOD
John P. Robinson
University of Maryland
This paper describes the diary method of collecting time use data. It describes the key features of the diary method, key uses of data collected from diaries, procedures to analyze time diary data, and the methodological properties of time diaries. Summaries of previous time diary studies and of alternatives to time diary studies are also given in the paper.
The paper argues that a key feature of the diary method of collecting time use information is that it acts like a “social microscope” into human behavior. The open-ended nature of a time diary allows respondents to describe in their own words what they were doing throughout the day. Diaries are typically collected for the entire day, so that the entire day’s activities are accounted for, in contrast to other methods, such as stylized questions or experiential sampling studies, that do not collect information on the entire day. With time diaries, respondents are also able to designate what the most important activity was during a time frame if the respondent was simultaneously engaged in more than one activity. Although time diaries are often only collected for a single day, when they are accumulated across a large representative sample, aggregate accounts of how a population uses time can be estimated.
A limitation of time diary studies is that respondents report only what they want to report. As a result, some activities of a sensitive nature may not be reported accurately (e.g., sexual activity or reports of drug use). Recall error may also be an issue for time diary studies since respondents are typically asked to recall what they did one day ago. However, in comparison with time-use collected through single stylized questions (e.g., How much time
each week do you spend cooking?), data collected through a time diary may be less prone to recall error.
The paper also describes reliability and validity features of the time diary method. Past studies of the reliability of time diary data are summarized. The paper concludes that time diary methods produce rather reliable accounts of time use at the aggregate level. The paper also summarizes validity studies of time-use data collected from time diaries. These studies have compared data collected from time diaries to data collected through a “random hour” technique (where respondents are contacted randomly throughout the day and asked to report what they were doing); through use of the “with whom” questions on the time diaries to cross-check spouses’ reports of what they were doing at certain times; through the shadow technique, in which respondents are followed throughout a day and their activities cross-checked with self-reports of time use; and through direct observational studies, such as television camera monitoring studies. The results of these validity studies are reported in the paper.
Also summarized in the paper are procedures for analyzing time-use diary data. The usual way to analyze the data is to focus on the primary activity. However, time diary data have also been analyzed using the location of the activity, whom the respondent was with during the activity, the day of the week or the time of the day, and reported secondary activities.
EXPERIENCE SAMPLING METHOD: CURRENT AND POTENTIAL RESEARCH APPLICATIONS
University of Waterloo
This paper gives an overview of the experience sampling method (ESM) of studying time use. The method monitors how respondents spend their time by using a pager, beeper, programmable wristwatch, or palm-top computer to randomly “beep” the respondent throughout the day. Once the respondent has been beeped, he or she records what he or she is doing at the time and records other items about the activity. This paper describes the method, compares its strengths and limitations to other methods of obtaining data on time use and highlights the potential applications of using it to study time pressure, psychological stress, and health.
The experience sampling method was developed mainly by psychologists interested in understanding behavior and states of consciousness throughout the day. As described above, the method randomly beeps individuals through-
out the day, at which time, respondents, through a self-report form, answer a series of questions about what they were doing at the time. There are typically a core set of questions on the self-report forms, such as what the person was doing at the time, where he or she was, who he or she was with, and what were his or her thoughts at the time of the beep. Self-report forms usually include a series of questions about the experiential, emotional, cognitive, and motivational aspects of the activity the respondent is engaged in when beeped. Experience sampling studies have been used to measure qualitative aspects of daily life, to study leisure experiences, to study psychological and health disorders and behavioral and experiential correlates to them, and to study organizational behavior in the workplace.
The paper also discusses the general strengths and limitations of these studies in comparison with other methods for collecting time-use data, specifically, time diaries and stylized questions. One strength of the experience sampling method is that it can be used to study psychological states during activities throughout the day. In addition, there are no recall issues since respondents fill out the self-reports as soon as they are beeped. There is also likely to be less “normative editing” in these studies. The method can be used to study variations in psychological states across different activities, in different locations, with different individuals, etc. It can also be used to study human behavior in a natural context, as the respondents go about their daily lives, outside of a laboratory. Experience sampling studies can also be used to study interpersonal relationships, such as the “unmutual togetherness” that was described in the main body of this report. Finally, data collected through these studies can be analyzed as either a cross-section or as panel data, since multiple reports are collected from study respondents.
Limitations of experience sampling studies listed in the paper are that they can be intrusive, they may be subject to self-selection bias in who agrees to participate in the studies, they are not standardized, they do not completely cover the day (unlike time diaries, which cover the entire day), and they are more expensive than other methods.
The last section of the paper describes how these studies can be used to better understand the link between time pressure, stress, and health. The paper reports results of a preliminary analysis of experience sampling data that correlates respondent’s daily moods and feelings of being pressured for time. Results show that being pressed for time (reported in a recall questionnaire) are negatively correlated with the respondent’s sense of well-being as reported in the experience sampling method study, but positively correlated with reports of anxiety. The paper argues that such data can be used for further analyses of the dynamics of time pressure and emotional and behavioral conditions across weeks to understand uses of time and mental health.
AN INTERNATIONAL PERSPECTIVE TO COLLECTING TIME-USE DATA
University of New South Wales
This paper summarizes some key theoretical and policy concerns for which time-use data can be used and discusses the implications for designing a time-use study. In doing so, the paper describes approaches used in time-use studies across different countries. The primary emphasis of the paper is that the best method for collecting time-use data depends on how the data will be used; the ultimate use of the data should be the guiding principle for choosing methodologies.
The paper describes two broad theoretical motives for collecting time-use data. The first is interest in conditions of economic progress. This includes changes in quality of life, social welfare considerations, and time spent caring for others. Quality of life is typically measured through national income accounts. However, national income measures count only goods and services produced and transacted through the market, not goods and services produced for a household’s own consumption. This is especially important for measuring quality of life in less developed countries. Furthermore, national income accounts do not measure unpaid household work or transactions in the informal sector of the economy. The paper contends that distribution of free time in society should also be considered to better understanding social welfare. Finally, there are policy concerns to understanding time spent caring for others. The second theoretical motive for collecting time-use data is to better understand social changes. Specific social changes highlighted are changes in the organization of the household, in the division of labor within the household, and in the increased participation of women in the labor force.
These theoretical motives have implications for data collection. The first implication discussed in the paper is who should be sampled. The paper emphasizes that for informing welfare distributional questions, the household is the key sampling unit. The typical approach is to randomly select households and to collect time-use diaries from all household members. This method has been used in all three Australian time-use surveys and in the Eurostat pilot study. The age limits of sample selection are also relevant considerations. Most studies now do not have an upper age limit on sample members. Lower age limits are more common. The Eurostat pilot study collected diaries from those at least 11 years old. A 1989 Italian study collected time diaries for children aged 3 and older, and a Bulgarian study in 1988 had no lower age limit. (For younger children, diaries are filled out for the child by a care-giving adult.)
The theoretical motives also have implications for diary design. One
design issue is the time span for which data are collected. Most studies collect data for a single day. The Australian study collected data for two days. Attempts have also been made to collect data for an entire week. Another issue is collecting information on secondary activities. If only primary activities are recorded, many activities of respondents will not be counted, especially those that tend to be “background” activities, for example, passive child care (being on call), listening to the radio, and conversation with others. In understanding care arrangements, the Eurostat pilot study asked not only who the respondent was with during the survey, but for whom an activity was conducted. The question was confusing, however, and coding costs increased. As a result, guidelines for the full Eurostat study do not call for inclusion of this question. Finally, the paper describes theoretical implications of designing questionnaires to be included with time diary studies. The paper suggests that to understand nonmarket work, it is important to collect information on household stocks of capital and on consumption of market services that may substitute for household labor. Also emphasized is the importance of collecting information on child care. A child care module was designed as part of Australia’s last two national time-use studies.