Presenters:
Chris Jackson, Statistic Canada
Stephanie L. Howell and J. Stephen Landefeld, Bureau of Economic Analysis
Duncan Ironmonger, University of Melbourne

J. Stephen Landefeld

There are some things that some people do not want in the gross domestic product (GDP), so we should not revamp the GDP but use "satellite accounts" as complements to it. Time-use and money satellite accounts are important in moving forward, but there are trade-offs to be considered.

Using T-accounts and budgets adds up to a very large number. (e.g., If you add all costs of inputs in health care, it's bigger than our current determination of GDP.) Also, there is the issue of willingness to pay; so we need to put together a complete time-use budget considering dollars as well as minutes and new focused surveys.

Chris Jackson:

"The valuation of unpaid work, a Canadian perspective"

This is a summary of the first generation of studies in this area.

Background

Six studies have been done since the mid-1970s. Some of the refinements include:

Population coverage increased and improved: Added Yukon/NWT in 1981, and added children age 15 or older and non-family persons in family household in 1986.

In 1981, sources of time-use data improved in Halifax and Toronto. A National Pilot Time Use Study was done with 2,686R, 14 locations, and was seasonal. A General Social Survey, Cycle II, was done in 1986 with 9,744R, national, and also seasonal. A General Social Survey, Cycle VII, in 1992, was done with 8,996R, national and annual. Also, we went back and revised earlier ones. The next one is due in 1998.

In the early 1960s, the first estimate of the value of unpaid work (VUW) was used. The definition of unpaid work (UW) is largely unchanged, but now it includes volunteer work. In 1992, we added care to household adults and dropped medical and dental care services.

The unpaid work produced by all persons in the counts is not of practical use for several reason. It is not a full geographic coverage; the age criterion is limited (15 and over in private households broken down by various categories), and it is non-comparable across countries and counties.

The valuation methods differed across studies. Only the Replacement Costs-Specialization (RC-S) method was used in all of them.

How do the estimates compare over time? The most recent study revised all past estimates to standardize as far as possible, in order to look at the trends from 1961 to 1992. A comparable coverage breakdown of the population and a comparable definition breakdown of UW, plus voluntary work, was done. Comparable valuation methods, plus SS contributions, were factored in.

Whose unpaid work counts?

In principle, all resident persons count. In practice, TUS sample eligibility rules restrict coverage by area/location, age criteria with lower and upper age limits, and by household.

Imputations are made to extend the coverage. Time-use data for Yukon/NWT is imputed. We do not impute UW for children under age 15, and we do nothing for persons in collective dwellings.

Current coverage includes persons age 15 or older in private households in Canada. This is broken down by province, sex, family status, labor force status, number of children, and, if any, by the age of the youngest child. Early studies covered only 75% of the population now covered.

What counts as unpaid work?

This is difficult to count in practice. For current definition, see Table A, p.4, in his paper, "The Valuation of Unpaid Work — A Canadian Perspective." It includes domestic chores, help and care of household members, shopping, services and household management, transport and travel related to household work, and other unpaid work, such as voluntary work and related travel.

The SNA guidelines look at the "third person criterion" (TPC) to distinguish between non-market production and other activity, but the interpretation and application of TPC varies widely.

In practice, only primary activity is taken into account. This is the case with the majority of studies on VUW, including STC. One would like to do something with secondary activity, but it is unclear how to approach it. The last ABS study did take secondary activity into account.

How is it measured?

The General Social Survey uses random digit dialing with a stratified sample of about 10,000. One person per selected household is chosen. A one-day recall with a 24-hour, yesterday diary and primary activity only is used. Collection is done throughout the year.

We compute average time on primary UW activity broken down by group (52) and activity (22). Multiply by 365/60 to convert to minutes per day to hours per year. Multiply by population count, by group (52) and activity (22) to obtain aggregate annual hours.

The 1961/71 data is based on average annual change 1981-92, projected back for the 52 groups and 22 activities. Bounding was done to limit deviation from the 1981 benchmarks.

How is it "valued?"

SNA recommends valuation of the non-marketed output at price of equivalent market goods or services as first best. The valuation at cost of different inputs such as labor, capital, and intermediate inputs is second best.

The focus in the past was often on labor inputs. More recently, effort has been directed to accounting for all inputs. We currently use two variants of opportunity cost (OC): 1. Before tax, including employer’s contributions; 2. After tax, excluding employee’s contributions. OC varies only by sex and province.

We use two variables of replacement cost (RC) to average hourly earnings for different types of unpaid work. (See Table B.) Specialist is based on single best match, full year/full time (FY/FT) earnings. Generalist is matched to personal services. This includes domestic employees. RC varies by province and activity.

There have been many criticisms of both OC and RC. In the future, the Replacement Cost-Generalist (housekeeper approach), with refinements, will be the principal value imputation method for SNA.

Some results:

There is a wide range of estimates from 32% to 54% of GDP in 1992. VUW declines steadily against GDP from 1961-86 in all methods, but the "trend" reversed in 1992. GDP grows 3.9% annually, 1961-92. This is 0.6%-0.8% more than GDP + VUW, depending on the method used. (i.e., GDP is growing faster than GDP plus VUW.) With a 2% annual gain in household productivity, GDP and GDP + VUW both grow at the same rate.

There were 15 billion hours on UW in 1961, and 25 billion in 1992, which is 23% more than SNA hours of paid work. The average hours per year fall from 1,223 in 1961 to 1,108 in 1986, but are up to 1,164 in 1992. Women are spending less time on UW, from an average of 1,663 hours per year in 1961 to 1,482 in 1992. However, men are spending more, from 787 to 831 hours per year.

There are changes in activity distribution. From example, meal preparation goes from 28% of UW to 23%. Help and care of household members goes from 16% to 11%, while management and shopping goes up from 12% to 15%. There are also changes in division of labor by sex. For example, women’s share of meal preparation goes from 83% to 76%, and their share of transport and travel goes from 51% to 58%. Women do about two-thirds of the UW, but their share declines slightly from 68% to 65%.

What lies ahead?

A review of methods, definitions, sources and concepts
A design of a conceptual framework for households (HSA)
A database of indicators for household production
A flow of consumption services from consumer durables
A survey of volunteer work and a monetary value assigned to it for Canada in 1998
An input/output (I/O) model of household production—prototype II
A review and assessment of the output approach all need to be done and compared with the previous estimates. 1998 will couch estimates within a full household satellite account.

Stephanie L. Howell

Current coverage of household production in the existing US National Income and Product Accounts (NIPA) and the 1993 SNA:

The NIPA are fairly good measures. They provide an aggregate organizing framework for the nation’s economic statistics and have been described as the pre-eminent tools for macroeconomic analysis. They record economic production, income, consumption, accumulation and wealth by sector, industry, type of income, type of product and region. However, their focus is on observable market transactions. NIPA include non-market production of goods for own-use, such as food and fuel produced and consumed on farms, but they exclude non-market production of services except for owner-occupied housing. SNA include non-market production of goods for own-use, but exclude services other than owner-occupied housing, major repairs and storage.

Thus there are limitations in the consistency and usefulness of the existing accounts and GDP. GDP is affected by institutional arrangements and may change even if economic activity has not changed. For example, consider the housekeeper who marries her employer. Before the marriage, her output was recorded as part of the GDP, but after the marriage, it is not, because her work is now "unpaid."

Market boundary affects the magnitude and comparisons of GDP across time and across countries in what types of production are purchased in the market. If non-market production is not included, shifts from non-market to market production over time overstates GDP growth, and the difference in proportions of market to non-market work overstates difference in GDP across nations. Such changes in the US in recent decades have resulted in overstating real growth in GDP.

These limitations of the market production boundary have long been recognized. There is consensus that non-market output should be included in GDP, but the problems are in measurement of this output. Economic accountants focus on: 1. Uncertainty — there are no observable transactions by which to measure non-market production; 2. If we expand GDP from production concepts to welfare concepts, this introduces subjectivity in determining what does and does not contribute to welfare as well as how to value that welfare.

The use of Satellite Accounts (SA) will allow exploration of alternative production boundaries or valuation methods without introducing unreasonable uncertainty into existing accounts. We can maintain basic consistency between new accounts and existing accounts by using the market to value the non-market output that we wish to include. This will make the accounts directly comparable and thus allow us to use them together. The focus of many studies is solely on the valuation of time spent working in the home. A thorough analysis requires a full production account for households, accounting for all time use. An I/O would be the ideal framework because it could show the inputs, labor and capital associated with given categories of household outputs. There is some data in the US, such as purchased goods and services and consumer durables, but time-use data is infrequent.

Their illustrative SA for household production goes beyond just the value of work-time by including other factors of production, but it does not consider all time use, such as leisure, and does not identify inputs with associated outputs in detail.

Three structural features of SA for household production:

1. Conceptual basis: Use market- rather than welfare-based accounts to maintain basic consistency with existing accounts and to avoid subjectivity.
2. Limited scope: Use market values and proxies and a market-like definition of household production (TPC).
3. Range of values: Use different valuation methods to allow evaluation of methods, source data and usefulness.

There are three approaches to imputing values for outputs of households. (See Table 1 of their paper, "Accounting for Nonmarketed Household Production within a National Accounts Framework.")

1. Replacement cost ("housekeeper approach"): Use average compensation for housekeepers multiplied by the number of hours worked in households.
2. Specialist approach: Assume those working in households are specialists at household production; this produces the highest valuation.
3. Opportunity cost: Value hours worked in the household at the wage that could have been earned in the market.

When we add the contribution of household production, GDP and GNP increase by 40-50%; labor's share of income increases by 52-58%; and the savings rate increases 6-12%. (See page 20 and Table 2 in their paper.)

(Please note error in the earlier paper, page 20: Robert Eisner's estimate of total unpaid household work time is 25 hours per week per person and their estimate is 26 hours, not 106 and 99.)

Duncan Ironmonger

"National accounts of household productive activities"

We actually have 2 markets:

1. GMP: Gross Market Product
2. GHP: Gross Household Product

Why we need these accounts, how these two markets interact together, the productive activities done by and in households using household capital, what is regarded as goods and services for use by own or other households including volunteer time formally or informally done without payment, and more, is illustrated in his paper, "National Accounts of Household Productive Activities."

What should the accounts cover? The complete picture, both what the market and non-market produces, so we can see how things move back and forth from each sector/market. We should disaggregate different types of households.

The accounts should be in the form of input-output tables. After we have prepared for one or two base years, then we can estimate for the current year or quarter. Actual tables will lag a year, as is true today.

Need three types of nationwide household surveys:

Household Time-Use Surveys (TUS)
Household Expenditure Surveys (HES)
Household Output Surveys (HOS).

The estimates from these surveys will need to be combined on a common population base and common time frame. Statistics Finland is working on account standards guidelines for household accounts.

The 1993 SNA bible has multiple activities classified according to the principal activity which is its greatest value. We must move the value of owner-occupied housing into the household-capital category because the household’s greatest asset is accommodations (living space), and capital goods are defined as goods that last more than a year.

Four modes or methods of production:

1. Use both market capital and market labor (e.g., hotel accommodation)
2. Household capital and market labor (e.g., housekeeper)
3. Market capital and household labor (e.g., the Laundromat)
4. Household capital and labor (e.g., home meal preparation)

Consider the outputs from each of these modes and focus on the change from inputs to outputs. Ironmonger wants to have two balloons (i.e., markets) not just one enlarged to include the other.

Discussant: Jan W. van Tongeren: These questions come out of her perspective from development of the SNA.

1. Relating well-being to GDP is a problem. Some say GDP does not measure well-being. GDP was not meant to measure well-being. In the SNA, "net disposable income of the household sector" is the closest thing to a measure of well-being, and while not a perfect measure, it comes much closer than GDP. Note that we can measure production in some household activities, but for some, like leisure, there is no market counterpart.
2. Is the production orientation appropriate for measuring well-being? We can change well-being without changing production.
3. Which elements to add and which to deduct? Not all market activities lead to output and therefore do not all add to GDP. Only final consumption should be added to GDP. Is all production of non-market activities just intermediate consumption? For example, writing software at home and then using it in your job or transportation to work should these not be counted if they are part of the costs of final production.
4. She agrees with Ironmonger. He is convincing because he classifies non-market activities in the same way as market activities. This is a more useful approach analytically. But the household is considered a super-establishment with lots of secondary production, i.e., activities and products being produced. We may not be able to connect inputs with outputs, so there is a question of double-counting. For example, where does transport fall?
5. There is much focus on adding things in monetary terms which could be in real (physical) terms. It is useful to look at things in physical terms, too, not just in monetary terms.

Group Discussion and Comments

Ironmonger: There is a problem of how to allocate things to final consumption, but we can handle this by, for example, investigating where electricity goes — for meals or other activities, etc. This may require subsidiary studies, but it can be solved. We may need to do some subsidiary studies to find out the arrangements.

We are currently measuring production, not well-being, but we can extend this measurement of production. (Points out some figures from tables) We do not have output estimates in the tables.

van Tongeren: There tends to be a downward bias in the NIPA.

Timothy Smeeding: Says he did not come here to tell the BLS to collect data on well-being but rather to say that this data is very important for many other reasons. There are many uses for the data that will be collected, both at the aggregate and micro levels. He is interested in the micro-data, the heterogeneity.

Landefeld: Agrees with Smeeding.

Robert A. Pollak, Washington University in St. Louis: 1) Do we want to think of an index as production- or utility-based? 2) An input measure? Take meals, for example: If we look at restaurant meals, we do not ask about quality, but about what they cost. It is important regarding a household meal to ask what restaurant meal is its equivalent.

Luisella Goldschmidt-Clermont, France: We ask households how home-prepared meals compare to meals they eat out (1-5 scale). Those who eat out can answer, but those who do not cannot make that comparison. To measure the value of household income in non-market terms, we must ask: What do households appreciate? What do they value and how does it compare in quality? Households should be at the center of our thinking. We should be more modest and ask households what they think. We have input-output tables for the market, but if we value housework in the present with wages in time, we get a different value. There is consumer surplus/profit there that we might want to measure.

Andrew Harvey, Saint Mary's University: We know that meal production at home is a very inefficient process. The third-person criterion is not very good because it excludes education and leisure. Consider a separation between production and consumption: produce something now and get a stream of benefits from it later, e.g., education, fitness. On the other hand, some things are consumed now.

Ironmonger: Considers volunteering and education as non-market household activities. Volunteering is something you do for another household. Education you do for yourself, but it is productive. These have to go in the time-use calculation somewhere. As Harvey says, we can count education as household input time, sleeping as output time, cleaning as input time, etc. All of these have process benefits, and we do not know how to value these.

Robert Michael, University of Chicago: Referenced Mitchell’s "Backward art of spending money and other essays." Noted the importance of how to use time inputs as a measure of output. There is a distinction between input and output which is critical to getting at technology, i.e., how to measure inputs and outputs of a meal in the household.

Ironmonger: We have counts of meals by person in the time-use diaries. We can count the time a person is being cared for.

Harvey: We can get the number of nights spent at home. With respect to measuring child care: output in one hour spent caring for four kids is four child-care hours, not just one.

Aletha Huston, University of Texas at Austin: As a developmental psychologist, she is not going to talk about inputs and outputs. People’s time does not divide itself into easy categories. Simultaneous activities are more the rule than the exception.

Landefeld: That is a common problem, and one reason to focus on the output.

Huston: But as Smeeding said, there are a lot of reasons to look at other things.

Session ended approximately at the scheduled time.

back to top