2
Accounting and Data Foundations

A large portion of this report is devoted to working through the many conceptual challenges that clutter the path toward construction of nonmarket satellite accounts. Another obstacle to this achievement, and one that will be expensive to clear, is the lack of data to support quantification and valuation of nonmarket activities in an acceptable fashion. As noted above, the new American Time Use Survey (ATUS) will provide rich information on the most important input to nonmarket production—the time people devote to nonmarket activities. Other inputs to nonmarket production commonly are purchased in markets, so that the challenges associated with their measurement, while not trivial, should be similar in nature to those routinely encountered in the construction of the National Income and Product Accounts (NIPAs). Considerable work will be required to develop the data needed for independent measurement of nonmarket outputs.

In this chapter, we begin, for reference purposes, with an overview of the NIPAs—the prominent U.S. model for national economic accounting. Many of the principles used in the construction of the NIPAs also carry over in a natural way to the construction of nonmarket economic accounts. We next discuss the role of time-use data in nonmarket accounting and describe the new ATUS. Another cross-cutting data development need is that for a coherent, readily accessible demographic database. The data needed to support measurement of nonmarket outputs are diverse; development of these output measures will be the house-to-house combat of nonmarket accounting.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States 2 Accounting and Data Foundations A large portion of this report is devoted to working through the many conceptual challenges that clutter the path toward construction of nonmarket satellite accounts. Another obstacle to this achievement, and one that will be expensive to clear, is the lack of data to support quantification and valuation of nonmarket activities in an acceptable fashion. As noted above, the new American Time Use Survey (ATUS) will provide rich information on the most important input to nonmarket production—the time people devote to nonmarket activities. Other inputs to nonmarket production commonly are purchased in markets, so that the challenges associated with their measurement, while not trivial, should be similar in nature to those routinely encountered in the construction of the National Income and Product Accounts (NIPAs). Considerable work will be required to develop the data needed for independent measurement of nonmarket outputs. In this chapter, we begin, for reference purposes, with an overview of the NIPAs—the prominent U.S. model for national economic accounting. Many of the principles used in the construction of the NIPAs also carry over in a natural way to the construction of nonmarket economic accounts. We next discuss the role of time-use data in nonmarket accounting and describe the new ATUS. Another cross-cutting data development need is that for a coherent, readily accessible demographic database. The data needed to support measurement of nonmarket outputs are diverse; development of these output measures will be the house-to-house combat of nonmarket accounting.

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States OVERVIEW OF THE NATIONAL INCOME AND PRODUCT ACCOUNTS The purposes of this section are to present an overview of the NIPAs, the standard to which we refer often throughout this report, and to provide a basis of comparison for concepts and data needs for the nonmarket accounts described in subsequent chapters. National income and product accounting is the centerpiece of national economic accounting in the United States. The NIPAs show the real and nominal value of output, the composition of output, and the distribution across types of income generated in its production. There are three other major branches of national economic accounting—capital finance accounting, balance sheet accounting, and input-output accounting. The capital finance accounts are better known as the flow-of-funds accounts. They show the role of financial institutions and instruments in transforming saving into investment and the associated changes in assets and liabilities. These changes occur as monetary flows over time, resulting in an increase or depreciation in the stock (the accumulated amount) of the asset. Balance-sheet accounts display these assets and liabilities at particular points of time. Input-output accounts trace the flow of goods and services among industries in the production process, and show the value added by each industry and the detailed commodity composition of output. Input-output matrices, particularly the benchmark input-output matrices, are a foundation for the NIPAs. Other accounts, specifically the international and the regional accounts, are also important sources of information for the national accounts (Bureau of Economic Analysis, 1985). These other accounts are mentioned only briefly in this chapter. In the NIPAs’ double-entry accounting system, domestic output can be measured by either gross domestic product (GDP) or gross domestic income (GDI). GDP is measured as the market value of goods and services produced by labor and property located in the United States. GDI is measured as the costs incurred and the incomes earned in the production of GDP. Business purchases from other businesses are netted out so that domestic output is an unduplicated total. In theory, nominal GDP should equal nominal GDI; because the two sides of the accounts are measured using independent and imperfect data, however, the aggregates typically do not match. The statistical discrepancy, which is recorded as an income component, is equal to nominal GDP less nominal GDI. Within a production function framework, GDP represents the economy’s output and GDI represents the capital and labor inputs used in its production, plus taxes on production (such as sales and excise taxes) and the surplus of government enterprises. In the United States, GDP is measured using an expenditure approach. GDP, as shown in Table 2-1, is equal to the sum of personal consumption expenditures, gross private domestic investment, net exports of goods and services (exports minus imports), and government consumption expenditures and gross invest-

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States TABLE 2-1 Gross Domestic Product and Gross Domestic Income, 2001 (billions of dollars) Compensation of employees, paid 5,881.0 Personal consumption expenditures 6,987.0 Taxes on production and imports less subsidies 659.8 Gross private domestic investment 4,109.9 Net exports of goods and services −348.9 Net operating surplus 2,329.3 Government consumption expenditures and gross investment 1,858.0 Consumption of fixed capital 1,329.3 Gross Domestic Income 10,199.4   Statistical discrepancy −117.3 Gross Domestic Product 10,082.1 Gross Domestic Product 10,082.1 SOURCE: Mayerhauser et al. (2003, p. 10). ment. The change in private inventories is included in the gross private domestic investment component of GDP. Imports are subtracted in the calculation of GDP, as expenditures for both consumption and investment include imported goods and services that are not a part of domestic output. GDI is equal to the sum of compensation of employees, taxes on production and imports (minus subsidies), net operating surplus, and consumption of fixed capital; a close synonym for consumption of fixed capital is economic depreciation. Operating surplus is one of several new concepts introduced into the NIPA framework as part of the December 2003 comprehensive revision. Net operating surplus is a profits-like measure that shows business income after subtracting the costs of compensation of employees, taxes on production and imports (minus subsidies), and consumption of fixed capital from gross product (or value added), but before subtracting such financing costs as net interest and business transfer payments (Mayerhauser et al., 2003). GDI does not include capital gains on assets as these do not arise from current production. Consumption of fixed capital is an entry that represents a cost of production, as it is a charge for capital used up during the accounting period. The double-entry nature of the NIPAs is not obvious from Table 2-1, as there are entries with different titles under GDP and GDI, but if an entry appears on one side of the accounts, a corresponding amount must appear someplace on the other side. For example, if an individual buys an automobile with his wage income, this amount appears as labor compensation under GDI and as a personal consumption expenditure under GDP. If the individual saves part of his wage income, that saving appears as investment under GDP. By definition, gross investment is equal to gross saving plus the statistical discrepancy, though the relationship between GDI and GDP is less obvious in the savings case than in the automobile purchase case.

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States Although nominal GDP should in theory be equal to nominal GDI, there is no expectation that, in practice, real GDP will equal real GDI. In fact, multifactor productivity is typically estimated by the difference in the rate of growth of output compared to the rate of growth of inputs. The NIPAs as currently formulated do not allow direct estimation of multifactor productivity, but the Bureau of Labor Statistics estimates multifactor productivity by major sector largely with data from the NIPAs (Fraumeni et al., 2004) and the international System of National Accounts (Commission of the European Communities et al., 1994) may be modified to call for the estimation of capital inputs within a national income accounting framework. Beyond pure monetary-based calculations, the NIPAs include a number of imputations, amounting to approximately 15 percent of GDP (Moulton, 2002). Although it is difficult to define rigorously what an imputation is, Moulton (2002, p. 3) offers the following: An imputation in national accounts refers to a flow that must be estimated by the national accountant because there is no directly related monetary transaction that is recorded in the books of a party to the transaction. Imputations generally arise for one of two reasons: (a) own-account production that takes place within the production boundary of the system, such as the services produced by owner-occupied dwellings, or (b) transactions that are not directly associated with an exchange of money between the transacting parties because the transactions involve barter, transactions in kind, or bundling the provision of a service with a financial transaction, such as depositing funds in a bank. As Table 2-2 shows, four categories account for the majority of the value of imputations in the NIPAs: services provided by owner-occupied housing, which accounts for about 6 percent of GDP; employment-related imputations, which account for about 3 percent of GDP; unpriced services provided by financial TABLE 2-2 Imputations in the NIPAs, 2001 (billions of dollars) Gross domestic product 10,082.2 Imputations 1,549.5 Owner-occupied housing 648.5 Rental value of nonresidential fixed assets owned and used by nonprofit institutions serving individuals 61.2 Services furnished without payment by financial intermediaries except life insurance carriers 288.9 Employment-related imputations 354.8 Farm products consumed on farms 0.2 Margins on owner-built housing 8.2 Consumption of general government fixed capital 187.7 Excluding imputations 8,532.6 SOURCE: Moulton (2002, p. 11).

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States intermediaries, which account for about 3 percent of GDP; and services from general government fixed capital, which amounts to something less than 2 percent of GDP. This outline of the NIPAs focuses on measures for the economy as a whole. Sectors covered in the accounts include the business, household, and government sectors. The Bureau of Economic Analysis (BEA) also has produced tables showing information for nonprofit institutions serving individuals, which represent the vast majority of nonprofits; in the NIPAs these nonprofits are lumped together with households. Acknowledging the imputations in and sector coverage of the NIPAs highlights the fact that the accounts do not exclude all nonmarket activities or the services from all nonmarket assets. Many government and nonprofit activities register in the NIPAs but, in general, only goods and services that involve payments (whether market or below market) are included. For example, payments for services rendered under the Medicare program are included under GDP and GDI, but no value is imputed for volunteer services in nonprofit hospitals, because they are provided without remuneration. Services from owner-occupied housing and from general government capital are imputed and included in the NIPAs. Valuation of the services from owner-occupied housing is linked to the terms of observed rental transactions. BEA acknowledges that the services from general government capital are underestimated in the accounts; uncertainty about the appropriate net return to such capital led to the adoption of an assumption that the net rate of return was zero (National Research Council, 1998; Bureau of Economic Analysis, 1995). In addition to the summary domestic income and product accounts shown in Table 2-1, the NIPA summary accounts include six others: the private enterprise account, the personal income and outlay account, the government receipts and expenditures account, the foreign transactions current account, the domestic capital account, and the foreign transactions capital account. Together these summary accounts, from which approximately 150 NIPA tables and the fixed-asset estimates (which underlie NIPA measures of consumption of fixed capital and consumer durable investment) are derived, provide a wealth of information for users (see BEA website). The satellite accounts envisioned in this report would build primarily on the domestic income and product accounts, which include information on the production costs that constitute GDI and on the values of final output that constitute GDP. In the satellite accounts we contemplate, the nonmarket analogs are values for the inputs used in nonmarket production and values for the output that is produced. MEASURING TIME USE Time is the most quantitatively significant input to both market and nonmarket production. One cannot begin to understand economically oriented nonmarket

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States activity without knowing how a population spends its time. Data appropriate for measuring time devoted to nonmarket activities require the recording of information on people’s activities away from their jobs. Just as market transactions are recorded directly by measuring expenditures rather than relying on people’s recollections of what they produced or purchased, it is also important to measure what people actually do with their time, not what they recollect having done with it long afterwards. The vehicle for collecting such information is a time-budget survey—a study in which a large sample of individuals keep a diary of their activities over one or several days. In a time-budget survey, respondents describe the various activities in which they have engaged and these are then coded into a set of categories. A well-designed time budget survey forces the aggregate of time devoted to all activities to equal 1,440 minutes per day for each person. Previous Collections of Time-Use Data Most classifications of household time use have followed the precedent set by Sandor Szalai (1973) and his collaborators, who organized time-use surveys in the 1960s to record people’s activities in a number of countries. Time budgets have been collected in the United States since at least the 1930s (Sorokin and Berger, 1939). There have been a few American time-budget surveys in the post-World War II period, including surveys in 1965-1966 and 1975-1976 (with a small extension in 1981), conducted at the University of Michigan (Juster and Stafford, 1985). Time-budget surveys were conducted at the University of Maryland in 1985, 1992-1994, and 1995 (Robinson and Godbey, 1999); a more recent study focusing particularly on child care activities also was done at the University of Maryland (Bianchi, 2001). While all of these studies were funded by federal agencies, none was designed or conducted by any part of the federal statistical apparatus. There have been differences in sample design across the surveys, in the ages of respondents to which the sampling frame applied, in the number of diary days sampled per individual, in the number of activities categorized, and in the categorization of activities. Perhaps most important for our purposes, the samples have typically been quite small—never more than 8,000 diary-days (1992-1994)—and they have been conducted on an irregular basis. Internationally, interest in time-use studies as a tool to estimate the quantity and, in turn, the value of nonmarket work has grown considerably over the last 20 years, and a number of countries are now committed to this task. In 1995, the U.N. Fourth World Conference on Women called for national and international statistical organizations to measure unpaid work and estimate its value in satellite accounts to the GDP. Other countries’ statistical agencies have conducted much larger-scale time-budget surveys in recent years, albeit only on a quinquennial or decennial bases, with several using the resulting data in the construction of nonmarket accounts. Efforts to date that are most directly relevant to the United

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States States are those underway in three other large English-speaking countries—Australia, Canada, and the United Kingdom. Canada first administered a national time-use survey in 1981 and subsequently established it as a regular component of its General Social Survey. Additional questions on unpaid housework and child care and elder care were asked on the 1996 census. Statistics Canada has used time-budget data to construct estimates of the value of households’ unpaid work. The most recent estimates, constructed using a replacement cost approach, put the value of households’ unpaid work in 1992 at about 34 percent of GDP (Statistics Canada, 1995, p. 42). Australia administered a pilot national time-use survey in 1987, with expanded versions in 1992 and 1997. The Australian Bureau of Statistics (ABS) published estimates of the value of unpaid work based on the 1997 survey. Relying on several different measures of replacement cost, ABS calculated that unpaid work amounted to about 48 percent of Australian GDP in that year (Australian Bureau of Statistics, 2000, p. 5). The divergence in the Canadian and Australian estimates of the amount of unpaid work likely reflects differences in the methodologies used in their time-use surveys, including differences in the categorization of activities, but no detailed comparisons or explanations have been offered to date. The United Kingdom administered a time-use survey in 2000, but in developing measures of nonmarket output the U.K. Office for National Statistics has applied an output-based method of valuation, rather than simply assigning a replacement value to time devoted to unpaid work (Holloway et al., 2002). The experimental accounts that have resulted are focused on several different outputs of the household sector: provision of housing, transport, nutrition, clothing and laundry, child care, adult care, and volunteer activity. Perhaps because of the emphasis on methodological development, the office did not provide an estimate of the size of the country’s household sector relative to standard measures of GDP (Holloway et al., 2002). Pioneering academics have engaged in lonely but important data collection efforts to collect time-budget data for the United States. Mainly because of the absence of continuing federal funding for this activity, however, it seems fair to conclude that the United States has until very recently been in the derrière garde worldwide in the collection of such data. The American Time Use Survey: Overview In January 2003, the Bureau of Labor Statistics initiated the monthly American Time Use Survey (ATUS). This study originated in part out of research interest in valuing women’s time in the household. Concerns that women’s contributions were being undervalued by the exclusion of household production activities prompted the initial Bureau of Labor Statistics (BLS) efforts to develop and test the collection of time-use data, leading to a pilot study in 1997 and full-

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States scale field testing in 2002 (see Horrigan and Herz, 2005). Once estimates have been subjected to appropriate scrutiny and reasonably well verified, the data from this survey will be a crucial input into the creation of nonmarket accounts. Recommendation 2.1: The American Time Use Survey, which can be used to quantify time inputs into productive nonmarket activity, should underpin the construction of supplemental national accounts for the United States. To serve effectively in this role, the survey should be ongoing and conducted in a methodologically consistent manner over time. The sampling frame for the ATUS is that of the monthly Current Population Survey (CPS)—the actual ATUS samples are taken randomly from households just completing their eighth month in the CPS sample. For example, a household that had been included in the CPS in January through April 2002 (waves 1-4) and January through April 2003 (waves 5-8) might have been included shortly thereafter in the ATUS (a new wave 9). Households are chosen based on a variety of stratifications (including race/ethnicity and the presence of children of various ages), all designed to reduce the sampling variance of the time-use statistics that cover smaller subsets of the U.S. population. One randomly selected (by BLS) adult member in each household chosen for participation in the ATUS is asked to complete a time diary. The diary is to be completed for the previous day, with a telephone interviewer leading the respondent through his or her activities over the 24-hour period that began at 4 a.m. on that day. Ten percent of the diary days are assigned to each weekday (Monday through Friday), 25 percent are assigned to Saturday, and 25 percent are assigned to Sunday. Respondents list their activities, showing when each new activity began and describing it in their own words. “Secondary” activities, undertaken simultaneously with the listed activities, are recorded if the respondent volunteers that they occurred. The respondent also lists where each activity was undertaken (e.g., at home, at the workplace, elsewhere) and who else was present (e.g., nobody else, spouse/partner, child/children, friends, coworkers). A crucial issue for our purposes is the classification of the respondents’ verbal descriptions of activities into categories that are useful for accounting and analysis. While the coding system created by the Szalai group has underlain the sporadic U.S. time-budget surveys, the ATUS has gone far beyond this. Beginning with a three-tier six-digit coding system, the basic codes are aggregated into 17 top-level categories: personal care activities (mainly sleep); household activities; caring for and helping household members; caring for and helping non-household members; work and work-related activities; education;

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States consumer purchases (e.g., food shopping); purchasing professional and personal care services (e.g., doctors’ visits); purchasing household services; obtaining government services and civic obligations; eating and drinking; socializing, relaxing, and leisure; sports, exercise, and recreation; religious and spiritual activities; volunteer activities; telephone calls; and traveling. This categorization appears to accord well with the construction of supplemental accounts along the lines discussed in this report. In addition to completing the time diaries, ATUS respondents again answer most of the questions that they had been asked in the CPS, providing updates on their work behavior, demographic characteristics, earnings and (bracketed) family income. These additional data keep economic information current, which is important for allowing the full ATUS sample to be used in valuing time in household production and for more accurate construction of estimation weights. BLS had expected to sample roughly 2,800 households per month in 2003 and to obtain a 70 percent response rate. But the response rate from the diaries taken by telephone was only 59 percent, and the response rate for the small number taken in person (from households without telephones) was only 34 percent. Because some of the 2003 sample was supported with funding that had been carried over from the survey development period, fewer households can be surveyed on a continuing basis. About 1,800 households per month were to be surveyed beginning in January 2004, with actual responses expected from individuals in about 1,200 households. Thus, the number of individual observations available for 2003 is about 21,000; roughly 14,000 individuals are expected to complete diaries in 2004 and each year thereafter. Significant nonresponse in a sample already trimmed by dropouts from participation in the CPS certainly has the potential to adversely affect the quality of the ATUS data, a matter to which we return below. As a large-scale and on-going time-budget survey, the ATUS is unique worldwide. Several other countries’ time-budget data sets are large enough to generate reliable measures of time allocation and include enough economic and demographic information to allow values to be attached to the hours they spend in productive activities. Other countries thus can construct statistically meaningful point-in-time supplemental national accounts, and some have done so. No other country, however, currently has the ability to produce satellite nonmarket accounts that can be continuously updated. The size of the underlying samples in the ATUS soon will be the largest in the world, but what makes the survey particu-

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States larly valuable for the purposes of creating regularly published nonmarket accounts is that its information will be provided every year, not just at intervals. The American Time Use Survey: Problems—and Nonproblems During its deliberations, the panel heard from a number of people who have been involved in the development and fielding of time-use surveys. Several of them raised questions about certain features of the ATUS design, questions which raise legitimate concerns. Not all of these concerns, however, are directly relevant to the potential usefulness of the ATUS for constructing nonmarket economic accounts. Despite the tremendous step forward that the ATUS represents, and its prospective role as a linchpin of nonmarket economic accounts for the United States, there are real concerns about its reliability for this purpose that need to be addressed. They can be summarized by the questions: Who? When? and What? The most important, the “who” concern, is engendered by the lower-than-expected response rates in the ATUS. Response rates on the precursor studies also have been low (Egerton et al., 2004), but the 59 percent response rate on the ATUS was even lower, and far below the 90-plus percent response rates in the CPS. Because most of the ATUS nonrespondents provided CPS responses the month before, a good deal is known about their demographic and economic characteristics. It is a relatively simple matter to reweight the sample averages of time allocations to account for differential nonresponse rates across groups with different observable characteristics. The difficulty is that there is no reason to assume that nonresponse is random (relative to the CPS sampling frame) across unobservable characteristics that may be correlated with time allocations. Although there is no way to know for sure until the data can be carefully examined, it seems plausible that busier individuals, or those with more irregular schedules, might simply be less likely to participate in the survey, meaning that the survey estimates could be distorted. Additionally, these biases in the ATUS are compounded with any biases in the sampling frame related to survivorship from the CPS. There is no simple way to adjust for nonrandom nonresponse related to unobservables. The BLS is fully aware of this difficulty and has explored various means of boosting survey response rates (Horrigan and Herz, 2005). Nonetheless, low response rates—and the possible resulting nonresponse bias—are the biggest concern with using ATUS data to construct satellite nonmarket accounts. Efforts to assess the extent of any possible bias in the survey responses—and, if necessary, to address that bias by raising response rates or making appropriate adjustments to the estimates—should be a priority. Recommendation 2.2: The Bureau of Labor Statistics should commit resources adequate to improve response rates in the American Time Use Sur-

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States vey and to investigate the effects of lower-than-desirable response rates on survey estimates. The “when” question stems basically from the nonresponse problem, and it has two aspects. First, are the ATUS responses representative of activities in which people engage on the days of the week for which they complete time budgets? As with the “who” question, it is a simple matter to reweight completed diaries so that each day of the week receives the same weight in the aggregated survey estimates. But even if BLS received diaries from an apparently (based on observable characteristics) random sample of individuals across days of the week, one cannot know whether the nonrespondents are distributed differently by day of the week in relation to their unobservable characteristics. One might, for example, believe that those whose time is relatively more valuable on weekends, given their observable characteristics, would be less likely to respond if asked to complete a time budget for a Saturday or Sunday. Difficulties with response rates generate difficulties with inferring the true weekly distribution of time allocations from the distribution of completed time-budgets. This underscores the importance of Recommendation 2.2. A second “when” issue is seasonal and is qualitatively similar to the possibility of nonrandom response propensities by day of the week: Might nonresponse rates across weeks of the year also be nonrandom in the unobserved characteristics of the individuals who are sampled? Among observationally identical individuals who are sampled in, say, July, those who respond may tend to be those with relatively less active schedules. Again, there is a risk that response probability may be correlated with patterns of time allocation. Raising response rates also is likely to reduce this problem. The “what” question of particular relevance to the panel’s purpose is the extent to which the ATUS records activities that are performed simultaneously (e.g., secondary, multiple, or standby activities).1 BLS has made a concerted effort—far more intensive than in the precursor studies—to ensure uniformity in the coding of respondents’ descriptions of their primary activities and has created the most detailed set of basic codes ever used in a time-budget survey. One limitation of the data produced by the survey is that they track “primary” activities, but not secondary ones; in other words, the data are coded to show people engaged in just one activity at a time. The survey does include separate questions designed to learn about time devoted to child care activities, which empirically is by far the most important “secondary” activity reported by respondents in other time-use surveys. Still, more complete information about secondary activities could prove to be important for monitoring time devoted to productive nonmarket activities that may occur simultaneously with other tasks or pastimes. A related 1   The issue of standby and secondary activities is thoroughly discussed in Pollak (1999).

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States question is whether activities that typically require only a few minutes at a time—for example, moving a load of laundry from the washer to the dryer—will be reported consistently enough to support good estimates of time devoted to them. Cost and burden considerations led BLS to decide against trying to elicit information about secondary activities generally; it records them only if the respondent volunteers that he or she was engaged in something else simultaneously with the primary activity, and it does not now code this information. While BLS is examining potential ways to obtain more information on secondary activities, the present lack of comprehensive information about them is a significant limitation for using the data to construct nonmarket accounts. Recommendation 2.3: The Bureau of Labor Statistics should continue to study and, where possible, obtain information on secondary activities that are to be covered in satellite nonmarket accounts. There are other potential concerns about the ATUS that the panel would consider secondary to those described above. One is that the ATUS obtains information only from one person in each household. While most of its American predecessors were no different, the 1975-1976 time-use survey conducted at the University of Michigan did obtain diaries from both spouses (in married-person households), and that practice is increasingly common in other countries (e.g., recent surveys in Australia, Germany, and Korea). For some kinds of academic research—for example, work examining household bargaining and marital sorting—having diaries from both spouses is crucial, since idiosyncratic spousal interactions are central to understanding this behavior. For purposes of measuring how a population allocates its time, however, the fact that the ATUS does not collect data from multiple family members seems less important. In constructing nonmarket accounts, we are interested primarily in how much time individuals in different demographic groups and with different economic characteristics allocate to the various activities whose valuation might be included in the accounts. Spousal interactions may be important, as spousal complementarities probably generate substantial additional value, but including the value of this intrahousehold capital is a more difficult and subtler matter than the central one of obtaining good estimates of how people spend their time. With higher levels of funding, it might be desirable to attempt to obtain time budgets from several (or even all) household members. One reason for this is that, as it now stands, ATUS cannot be used to calculate how much combined time parents spend with children in two-adult households. The total time spent with a child depends on whether mom’s time and dad’s time are positively or negatively correlated, and ATUS provides no information about this. This would be a problem, for example, for a “family care and human capital” account. Whether such a redesign would be feasible within the existing ATUS framework is an open question; Canadian experience suggests that the fraction of households from which multiple household responses could be collected for a single day would be

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States significantly lower than the fraction from which a single response could be collected. In any event, in the presence of limited resources, it is the panel’s judgment that this is not a top priority. It would be useful to conduct some split-sample experiments comparing the alternatives in each case—data collected from multiple adults in the household versus a single adult, and data collected for multiple days versus a single day—in order to have a firmer basis for assessing the benefits that might be associated with the multiple adult or multiple day approaches. A second “minor” concern is that the ATUS asks each respondent for a time budget for only one day. Only the 1975-1976 U.S. survey obtained diaries on multiple (four) days, although the leading international time-budget surveys have typically obtained data for two days. Having time budgets for multiple days would allow researchers to address a variety of interesting issues, particularly those related to variation in the timing of activities. Obtaining diaries on multiple days also carries the potential to reduce sampling costs. On the other hand, because respondents are drawn from the CPS outgoing rotations, any potential advantage with respect to sampling costs may be less for the ATUS than for a time-budget survey that requires a fresh sample. Perhaps more importantly, asking respondents to report for multiple days could have the serious disadvantage of depressing survey response rates still further. In the end, the relative value of having multiple reports from particular respondents as opposed to single reports from a larger number of respondents may depend on whether interday or inter-personal heterogeneity in time allocation is greater. For purposes of constructing nonmarket accounts, the gains from having data on time use for multiple days seem quite small. It is true that the activities reported by any particular individual on any particular day may be unrepresentative of how they typically spend their time, but such anomalies should average out across the population of respondents. Another limitation of the ATUS from the nonmarket accounting perspective is that data are collected only for people aged 15 and older. The exclusion of children and young teens means that other data will be needed to quantify the time spent in school or school-related pursuits, as would be required to construct an education satellite account. A final point here is that different sorts of time-use data may be needed to examine the tradeoffs that households face when weighing unpaid production and market substitute options. Learning about these tradeoffs would require information about how individual households combine time with purchased goods and services to produce the various things they need and enjoy in daily life. Analyses based on ATUS data might create a few demographic cells (e.g., age and education), construct estimates of average time expenditures for those cells, and link them to average goods expenditures from a consumer expenditure survey. Such analyses could be informative with regard to how individuals, on average, combine goods and time (see, e.g., Gronau and Hamermesh, 2003), but they would miss all of the idiosyncrasies inherent in marital matching and household behav-

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States ior. It might be valuable to have a one-time survey, perhaps for a sample of households that had previously responded to the Consumer Expenditure Survey, that would provide time budgets and consumer expenditure information for the same people. The ATUS is not perfect for purposes of constructing nonmarket accounts: it could not be, given budget constraints and the conceptual and measurement difficulties inherent in obtaining time-budget data. We understand that there were good operational reasons for the decisions made in designing the ATUS. There was evidence, for example, that, had the survey been designed to collect time-use information from multiple members of responding household on a particular day, survey response rates would have been much lower. Similarly, testing carried out during the development period raised serious concern that probing systematically for secondary activities in which respondents might have been engaged would have greatly increased the perceived survey response burden and thus adversely affected response rates. And BLS is well aware of the potential for nonresponse bias and has planned research to assess its significance. Still, as work proceeds on the ATUS and on time-use data collection more generally, the limitations and potential biases in the data currently being collected for nonmarket accounting purposes should be kept in mind, and efforts to improve the data pursued. The criticisms of this section notwithstanding, the ATUS is a tremendous step forward for the federal statistical system. Indeed, without something like the ATUS, one could not seriously contemplate the creation of nonmarket accounts for the United States. DEMOGRAPHIC DATA Time-use and demographic data must be combined to provide a firm foundation for nonmarket accounts. Time-use data can be used to answer questions about what individuals with given characteristics are doing with their time; demographic data describe the distribution of these individual characteristics in the population. Time use varies significantly across population subgroups. For example, in general, individuals with young children have less time for certain activities (e.g., traveling, work, going out at night) than adults without children. In addition, the value of time spent may vary with an individual’s characteristics. A higher value may be placed on time spent completing 4 years of college than on time spent completing 4 years of high school, for example, because of the greater value of forgone earnings for someone who already has completed high school. Detailed demographic data are needed to estimate differences in time allocation patterns across various socioeconomic subgroups of the population. There are several reasons that a comprehensive demographic database is not available for the United States. First, in our decentralized statistical system, agencies commonly specialize in producing certain types of data, and these efforts typically are not coordinated. For example, the National Center for Education

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States Statistics is a source of information on preschool and early grade enrollment by mother’s education and employment status, and the National Center for Health Statistics is a source of information on health care visits by age, sex, and race. Second, statistical and methodological revisions frequently are not carried back through time. For example, the Current Population Survey used the 1980 census population controls from 1981 through 1993 before switching to 1990 census population controls in October 1994. This procedure may significantly affect estimates of the absolute numbers of school enrollees and the comparability of the enrollment series across time; other measures, such as enrollment rates, may not be affected (U.S. Census Bureau, 2002, p. 131). Additionally, the available information may not be cross-classified by all dimensions of interest. An attempt is made to minimize respondent burden and, accordingly, a given survey may collect information on only a few demographic variables. A project that attempts to link administrative record information contained in establishment surveys and household surveys is being undertaken at the Census Bureau (Abowd et al., 2004); such projects may facilitate the creation of a unified set of demographic statistics. Although we recognize that creation of better coordinated demographic data would require significant effort by statistical agencies and other suppliers of data, the development of such data is an important goal. Recommendation 2.4: A consistent and regularly updated demographic database should be assembled as an input to nonmarket accounts. The database should include information on the population by age, sex, school enrollment, years of education and degrees completed, occupation, household structure, immigrant status, employment status and, possibly, other dimensions. It would also be ideal to have information on health in this demographic database, but this may not be a realistic goal. Household structure data should include demographic information on members of a household, including relationships if any (e.g., divorced or married), and numbers and ages of children and unrelated individuals, as well as information on children not living within the household. The fact that high-quality demographic data already are collected by the Census Bureau and other agencies makes this recommendation feasible to implement. Data currently collected by the decennial census and by the Current Population Survey provide much of the necessary input. What is needed is to have key data sources linked together, made consistent over time, and published at intervals to support accounting efforts effectively. Many critical census projections are made annually, but these do not include all of the variables that are needed. Most of the essential demographic data fields are included in the census long form but in the past some have been presented only once every 10 years. The Census Bureau’s new American Community Survey will (contingent on funding) be continuous, though rolling geographically, and will advance the effort to produce a more fluid demographic description of the population. Given the budget constraints of the statistical agencies, the demographic database should be

OCR for page 39
Beyond the Market: Designing Nonmarket Accounts for the United States built, to the maximum extent possible, using existing or already planned information, minimizing the extent of any new data collection. OTHER DATA NEEDS In addition to the above-described data, which relate mainly to labor inputs, a complete nonmarket account must include values of nonlabor inputs. For example, a home production account must include data on the capital services, materials, and energy inputs that complement unpaid labor in generating home-produced outputs. Purchases of materials used in home production already are included in the NIPAs, as consumer goods on the production side and as returns to capital, labor, and other inputs on the income side. The NIPAs also include spending on consumer durables, such as refrigerators and washing machines, though the annual flow of services associated with the stock of consumer durables does not correspond on a year-by-year basis with spending on purchases of consumer durables in the same year (see Fraumeni and Okubo, 2001). In accounting for household production, it is the flow of services from these durables that is relevant and for which data are required. Full development of nonmarket accounts also will require further research and data development to advance understanding of age-old questions relating to the definition and measurement of output. What are the outputs of the various nonmarket activities? Zvi Griliches observed that “in many service sectors it is not exactly clear what is being transacted, what is the output, and what services correspond to the payments made to their providers” (Griliches, 1992, p. 7). This observation is especially pertinent for many of the areas of interest here that are dominated by services—and services difficult to measure at that—such as education, health, social services, culture and the arts, and recreation. The need for development of better measures of nonmarket outputs can be illustrated with reference to education and health. Frequently, in difficult-to-measure sectors, the value of output is set equal to the aggregate value of the inputs used in its production. Accordingly, little is known about growth, quality improvements, or productivity in these sectors. In recent years, alternative approaches have been developed for estimating educational output more directly. Examples of these approaches include indicator (e.g., test-score based) approaches, incremental earnings approaches, and housing value approaches. Similarly, for a health account, data on the population’s health status, of the sort now being developed in disease state and health impairment research, hold promise of providing direct measures of the output of the health sector. The data that will be needed to create these output measures, as well as the data required to construct defensible measures of other sorts of nonmarket production, are discussed at the appropriate points in Chapters 3 through 8.