The Government and Private Nonprofit Sectors
Measuring the value of public- and nonprofit-sector activities has long been recognized as a challenge. As with the other nonmarket areas considered in this volume, a common and fundamental difficulty is the absence of market prices, but there are also other conceptual issues that need to be resolved.
We examine the government and nonprofit sectors together because they are the loci for the production of outputs having sizable public goods elements, and it is these elements that pose a number of central measurement and valuation issues.1 A key topic in this chapter is volunteer labor, which constitutes a large nonmarket input, especially in the nonprofit sector, but in government as well. This chapter does not address all of the ways in which measurement of governmental and nonprofit sector activity might be improved for specific types of applications. For example, we do not address in any detail the issue of how outputs that offset what would otherwise be negative effects on welfare should be accounted for—e.g., governmental resources going into added police services in response to an increase in crime.
Outputs of the government and nonprofit sectors commonly are given away, not sold. Examples include the provision of national defense and basic medical research by government, the research and dissemination activities of nonprofit universities, and the cultural and species preservation work of nonprofit museums and zoos. To finance such public goods (also sometimes termed collective goods),
governments typically levy taxes. The activities of nonprofit organizations are financed through some combination of private and government contributions of money, time (volunteer labor), and goods; tax subsidies; user fees; and assorted ancillary revenue-generating activities (e.g., university bookstores, museum shops, or hospital-operated commercial fitness centers).
In keeping with standard practice, the outputs of the government and nonprofit sectors in the National Income and Product Accounts (NIPAs) are valued by summing the market value of the inputs purchased for their production (plus, for government, a capital depreciation component). Measured in this limited way—largely in terms of expenditures on labor and structures—output of federal, state, and local governments included in gross domestic product (GDP) was estimated to be $1.96 trillion of the total $10.49 trillion for national output in 2002 (Bureau of Economic Analysis, 2004). Over 22 million people are employed by government at all levels. This represents about 17 percent of all employment—a share that has held fairly steady over the past decade (U.S. Office of Management and Budget, 2003). According to Bureau of Economic Analysis (BEA) data, these government employees received more than $1.1 trillion in wages, salaries, and benefits in 2002 (Bureau of Economic Analysis, 2004).
Currently collected data also provide some sense of the importance of nonprofit institutions in the economy. Most of what is known relates to organizations that serve the public and are classified under section 501(c) of the tax code. These organizations are eligible to receive tax-deductible donations and are large enough (over $25,000 in revenues) to be required to file annual reports with the Internal Revenue Service (IRS). Other organizations of particular interest, such as religious congregations and organizations with revenues of less than $25,000 per year, are not required to report to the IRS and, as a result, information about them is lacking (Boris, 1998).
Under law, there are many classes of tax-exempt “nonprofit” organizations. 501(c)(3) public charities include, among others, most nonprofits involved in the arts, education, health care, human services, and community service. 501(c)(3) private foundations are primarily grant-making organizations, such as the Ford Foundation or the Pew Charitable Trusts, that make grants to other nonprofit organizations. Other exempt organizations registered with the IRS include trade unions, business leagues, social and recreational clubs, and veterans associations classified under varying parts of section 501(c)(4) of the IRS code (Urban Institute and National Center for Charitable Statistics, 2000).2
Data published in the New Nonprofit Almanac (Independent Sector and the Urban Institute, 2001) provide an indication of the role that these organizations play in the economy. The Almanac focuses on groups it calls independent-sector
For full definitions of these kinds of organizations, see the IRS web page on the topic, available at http://www.irs.gov/charities/index.html [accessed October 14, 2004].
organizations, which include those operating under sections 501(c)(3) and 501(c)(4) of the Internal Revenue Code, plus religious organizations. The total number of independent-sector organizations grew from 793,000 in 1982 to 1.2 million in 1998. In 1998, these organizations employed an estimated 10.9 million paid workers—about 7.1 percent of working Americans—and paid approximately $258 billion in wages (Independent Sector and the Urban Institute, 2001).
In the NIPAs, government activity is aggregated across industries, which serves to highlight its overall size as well as changes over time. While government purchases have long been aggregated to constitute a standard “final use” component of the national accounts, nonprofit activity is scattered across a number of industries. This has made it difficult to aggregate the data required to determine the size of and trends in the sector, even if the intent is simply to quantify market-based activity.
The BEA has historically maintained data on income and outlays of nonprofit institutions as part of the personal sector, which by convention includes nonprofit organizations serving households. Young (1993) provides a detailed description of how and where nonprofit organization activity is treated in the NIPAs and in the detailed input-output tables that underlie GDP. Recently, BEA has disaggregated the personal sector to provide separate estimates of expenditures, income, and savings for households and nonprofit institutions serving households. The new tables cover a wide range of institutions—religious, social services, medical care, education and research, recreation and some personal business—and these organizations’ receipts and expenditures were reconciled with statistics on tax exempt organizations from the Internal Revenue Service. The BEA’s preliminary estimates indicate that net current receipts of nonprofit institutions serving households—from sales of goods and services; transfer payments from businesses, governments, and households; and rental, dividend, and interest income—amounted to roughly $742.4 billion in 2001. This represents approximately 8.5 percent of total personal income, a share that has remained fairly constant since 1992 (Mead et al., 2003, p. 16). Nonprofits that serve business—credit unions, financial institutions, chambers of commerce, and trade associations, to name a few—are not included in the new tables (Mead et al., 2003, p. 13).
The government component of the national accounts has been improved in recent years. To take an important example, since their 1996 revision, the U.S. national accounts have distinguished between consumption and investment expenditures by governments. The accounts, however, still do not provide measures of the output of the government sector that are independent of the inputs, something that is essential to confronting a number of basic research questions related to the efficiency of government operations and other matters (National
Research Council, 1998, p. 2) What is the return on capital investment by the government? What is the return on government-supported research and development? Are government workers growing more or less productive over time? Do government enterprises—such as utilities, airports, transit services, schools, and hospitals—operate efficiently? How much aggregate savings occurs in the United States and how does government policy influence the savings rate? Meaningful answers to all of these questions will require independent measures of the output of the government sector.
Recent work to improve accounting for the nonprofit sector has been shaped by guidelines set down in the Handbook on Nonprofit Institutions in the System of National Accounts (United Nations, 2004). The stated goal of the U.N. project, and the presumed objective of a U.S. satellite account for the nonprofit sector, is to improve the ability to monitor the scope, size, structure, and financing of nonprofit activities and their effects on the economy. The U.N. work has focused primarily on measuring inputs—particularly volunteer labor—in a consistent manner across countries. The Handbook pays relatively little attention to the valuation of output produced by nonprofit organizations, discussing the matter but offering few recommendations.
It is easy to see why accounting for activity in the nonprofit sector is important and how it might be improved. The nonprofit sector is a large and growing component of many of the world’s economies—accounting for perhaps 8-12 percent of paid nonagricultural employment in most developed countries and even larger shares in developing regions (United Nations and Johns Hopkins University, 2003, p. 3)—but the value it creates is not well represented in core national accounts. Additionally, provision of services by nonprofit organizations sometimes is viewed as a substitute for the services provided by governments, which means that government activity cannot be fully understood without also understanding activity in the nonprofit sector.
Nonprofit organizations share many common characteristics—nonprofit status, public-goods production, revenue structures that rely on donations of time and money, heavy use of volunteer staffing, and tax and legal treatment (United Nations and Johns Hopkins University, 2003). These characteristics argue for the analytic value of coherent data on the nonprofit sector as a whole, but both the System of National Accounts (SNA) and the NIPAs distribute the sector’s expenditures and revenues across the government, corporate, and nonprofit institutions (serving households) categories (United Nations and Johns Hopkins University, 2003), making it difficult to obtain a unified picture of its activities.
Similar to their treatment of government, national accounting structures generally do not measure the output of nonprofit organizations independently from the inputs used in producing that output, something that should be a goal for a nonprofit satellite account. Furthermore, it may be beneficial to develop measurement concepts and classification structures for the nonprofit sector that are different from those used in the conventional NIPAs, and a satellite account would
offer a vehicle for experimentation with alternatives. In both the government and the nonprofit sectors, the key challenges are how to define and measure “outputs” and then how to value them, recognizing that different objectives and uses imply different valuation approaches, so that there is no single “correct” approach.
Recommendation 7.1: Measurement of government and nonprofit economic activity should be strengthened by developing satellite accounts that include and attempt to value nonmarketed inputs and outputs.
Because they would maintain both market and nonmarket components, these satellite accounts would not necessarily be fully compatible with the existing NIPAs. They would, however, expand the information base for studying changes in government and nonprofit activity over time and, possibly, making comparisons across regions and countries.
Experimental government and nonprofit sector accounting work should focus on developing new and quite possibly multiple, measures of output. For the nonprofit sector, the initial focus should be on developing an account for charitable organizations that are exempt from taxation—specifically those operating under section 501(c)(3), and perhaps some operating under 501(c)(4), of the Internal Revenue Code. These are essentially the organizations that are providers of public goods, rather than providers of private outputs to their members.
Nonprofit organizations that use mainly purchased inputs are similar to government entities in that the prices and quantities of these inputs register in the national accounts. Although not sufficient to answer all of the questions of interest, summing these up offers one measure of the organizations’ contribution to GDP. For the satellite nonprofit sector account, a higher priority should be given to classes of organizations—particularly those in IRS subsection 501(c)(3)—whose activities are most seriously underrepresented in the national accounts because they involve large amounts of volunteer labor or rely heavily on in-kind donations.
The first step toward strengthening the measurement of nonprofit economic activity is to identify and organize the relevant market elements from the NIPAs, as the BEA has begun to do. These market elements then should be augmented with information on the nonmarket activities that are integral to the sector but are currently unmeasured or measured in ways that may be unsuitable for certain purposes. An important issue is how to value inputs that, like volunteer labor, are rationed by nonprice mechanisms, which in turn raises the issue of how to value the outputs produced by such inputs. Consumers generally are not free to acquire as much of a nonprofit- or government-provided good or service as they would like at the notional dollar price. This transaction price, which is often zero, thus may not accurately indicate the consumer’s marginal valuation of the output.
Many of the inputs to the production of government and nonprofit organization production are purchased in markets: a university buys a computer, a church buys a new roof, or the Department of Defense buys new defense hardware.
These types of intermediate goods purchases and capital investments, already captured in the NIPAs, also should be included in any new satellite account that aggregates nonprofit sector activities. In addition, it is important to include in the satellite account information on government and nonprofit inputs that are not purchased in markets and, thus, not counted in the NIPAs. Examples include pharmaceuticals given by manufacturers to government and nonprofit health programs, groceries supplied for free or at reduced cost to soup kitchens, food provided by restaurants to volunteers who are donating their time, and computer hardware donated to schools. Most of these same goods are also sold in private markets and so, though they are unmeasured in the NIPAs, in principle they have observable prices. These prices may or may not provide a good indication of the value of items that have been donated. Consider, for example, donations of computer hardware or software to a school or donations of pharmaceuticals to a nonprofit health clinic. It would be questionable to value such donations at the prices at which the school or the clinic could have, but did not, buy the products. The observed market prices would probably overstate the recipient’s willingness (and ability) to pay. Nonetheless, satellite accounts that include in-kind donations to government and nonprofit organizations could shed light on currently unmeasured economic inputs.
Recommendation 7.2: Donations of labor and goods to government and nonprofit organizations should be characterized and described in quantitative terms, and approaches based on market comarisons should be developed for estimating their value as inputs.
Because some donated goods are first purchased in markets, the corresponding links with and effects on the NIPAs should be documented.
Volunteer labor is the principal input to government- and nonprofit-sector activities that goes unmeasured in the NIPAs. There is considerable variation in measures of volunteer activity across surveys. Bureau of Labor Statistics (BLS) data from a supplement to the Current Population Survey (CPS) indicate that about 63.8 million people (age 16 and older) performed volunteer work from September 2002 to September 2003 in the United States. This translates into a volunteer rate of 28.8 percent among the civilian noninstitutionalized population. The median amount of time that people reported spending in volunteer activity for the period was 52 hours per year. For this work, volunteers are defined as “persons who did unpaid work (except for expenses) through or for an organization” (U.S. Bureau of Labor Statistics, 2003). Independent Sector and the Urban Institute (2004), based on data from their 2001 Giving and Volunteering Survey, report that 83.9 million Americans (age 18 and older) volunteered in 2000 (a volunteer rate of 44 percent), contributing an average of 3.6 hours a week, figures
much larger than those based on the CPS data. Both of these estimates rely on answers to retrospective questions about activity over a 12-month period. The figures from the two sources have not been reconciled. When they become available, data on volunteer activity from the new American Time Use Survey (ATUS) should be more reliable than any that currently exist, but the ATUS data surely will not alter the conclusion that volunteer activity is significant in scope and magnitude.
According to the CPS data for the 2002-2003 period, the major types of organizations for which individuals volunteered were religious (34.6 percent), educational or youth-service related (27.4 percent), social or community service (11.8 percent), and hospitals or other health related (8.2 percent). The major activities performed included fundraising (28.8 percent), coaching, refereeing, tutoring, or teaching (28.6 percent), collecting, preparing, distributing, or serving food (24.9 percent), providing information such as by serving as an usher, greeter, or minister (22.0 percent), and general labor (21.8 percent).
Other more specialized examples of volunteer employment include accountants who help low-income people prepare their income tax returns and Earned Income Tax Credit applications, lawyers who provide pro bono legal services, and corporate executives who serve on the boards of directors of nonprofit organizations. In contrast to their counterparts on for-profit organization boards, even in the same industry, nonprofit board members often receive a negative wage in the sense that they are typically expected to give donations to the organization in return for the honor of board membership.
While survey data disclose important information on hours of volunteering and about the industries to which it is supplied, the meaning of the data is subject to interpretation. Even when asked specifically about volunteering in connection with an organization, for example, some respondents may perceive their participation in informal activities, such as a private quilting group or a poker club, as a volunteer activity. Examples of informal, nonmarket groups abound, some providing external benefits that may be of sizable consequence—e.g., “neighborhood watch” and community youth literacy groups—and others, such as local garden and investment clubs, providing benefits that are limited to members. These different cases suggest that multiple options exist for defining the “value” of organizations and for establishing the boundary of what is considered the output produced by volunteer labor. Furthermore, in some surveys, a great amount of what people report as “volunteering” may not be connected with an organization at all. In all of these cases, observed transaction prices of zero often mask a complex set of barter arrangements that yield explicit monetary prices of zero but that understate the private and social values being created (see below).
These definitional issues highlight the question of what should be covered in a nonprofit satellite account. Focusing attention in a satellite account on information about volunteering through or for an organization is defensible on practical grounds for initial forays into nonmarket accounting, both because of the interest
that has been manifest in the activities of nonprofit organizations and because there is a reasonably clear boundary that can be drawn around this set of activities. At this time, the panel would prioritize valuation of volunteering to organizations, particularly those recognized as nonprofits by the IRS. As a long-term goal, a satellite account could be expanded to cover volunteer activity more broadly. To understand how the economy is functioning and changing over time requires measuring and valuing all productive unpriced labor time. Time spent helping others informally is arguably as or more important than formal volunteering. As the population ages, we might expect an increase in the reliance on friends, family, and neighbors to help out with grocery shopping, yard work, and other activities of daily life. Some of these activities fall conceptually between household production and volunteer work. The practical problems of measuring and valuing volunteer labor activity not connected with a formal organization may be severe; still, we want to highlight its omission from traditional economic accounts and, because such activity clearly contributes to real output, encourage attention to it.
We would also note, again, the potential for overlap. If a comprehensive volunteer labor account were developed, it would cover activities that also factor into other accounts proposed in this report. For example, time spent volunteering at a hospital is an input to health sector output, and time spent volunteering at a school is an input to education sector output. This potential overlap should not be avoided, but it means that analysts and users of data from the separate satellite accounts should be cautious to avoid double-counting of outputs.
Whatever accounting scheme is pursued, one must know the number of volunteers and the number of hours they worked in order to derive an estimate of total volunteer hours. One also needs to know the nature of volunteer activity and the values to be assigned to the hours of each type of work performed. It is reasonable to expect that good estimates of the number of hours worked by volunteers can be obtained—as already noted, the ATUS should be enormously helpful in this regard.
The appropriate valuation to be attached to an hour of volunteer labor depends on the purpose of the measure being constructed. An hour of volunteer labor implicitly enters the national income accounts with a price of zero, just as other inputs enter at the prices paid for them. If the goal is to measure productivity or economic welfare, however, attributing a value of zero to volunteer labor has little appeal. If, for example, the supply of volunteers increases in response to shrinking opportunities in the paid labor market, a zero valuation of volunteer time would bias measures of change in economic output. The NIPAs would reflect the decrease in paid employment, but they would not capture any increase in production associated with the increase in unpaid volunteer work. Thus, a shift of workers between paid and nonpaid employment would, other things equal, alter measured national income and product in a manner that distorts the true change in economic welfare and economic growth. It would exaggerate the
decline in economic welfare if the economy is contracting and labor shifts from the paid labor market to the volunteer market, and it would exaggerate the gain in economic welfare if the economy expands and there is a shift in the opposite direction. This is very much like the well-recognized distortion in national income trends accompanying shifts from household production, such as home-prepared meals, to market production, such as restaurant meals, or vice versa.
Valuing the contributions of volunteer labor to government or nonprofit activity at zero, as in the national accounts, masks a great deal of complexity. The total contribution of this volunteer labor to economic output, and to welfare, clearly is not zero, and, indeed, even its marginal value very likely is positive. If a cost-minimizing organization—government, nonprofit, or for profit—can obtain as much of an input as it likes at a zero price, it can be expected to increase utilization of the input until its marginal productivity falls to zero. But the supply of volunteer labor is not unlimited—organizations typically cannot obtain as much of it as they wish. Thus, the observed market price of zero likely understates the marginal productivity of volunteer hours. In this case, valuing volunteer time at its observed price produces downward-biased estimates of its contribution to economic output.
If one is interested in the full welfare contribution of volunteer work, a different approach that accounts not only for the value of volunteer services to the receiving organization but also for the utility that volunteers derive from the activity may be appropriate. Indeed, in addition to providing experiential benefits, some volunteer activities (e.g., pro bono legal work) may enhance a person’s professional status and possibly raise his or her future earnings. People who obtain utility from volunteering will be willing to work at a monetary wage of zero despite having a positive opportunity cost of time. This is not so different from the market case in which an individual, confronted with two job choices, accepts the lower paying option because it is expected to provide more nonmonetary benefits.
Interestingly, paying a wage of zero gives the “employer” an implicit exemption from minimum wage requirements. If an organization were to seek to increase the amount of labor available to it through volunteer channels by offering an explicit wage greater than zero, it would face a significant discontinuity, as it would be required to jump from paying a zero wage to paying at least the minimum wage.
There are at least two possible goals that an analyst might have in mind when calculating the value of volunteer labor. The goal that is chosen will affect the approach adopted. One goal might be to measure only the contribution of the labor input toward the production of organizational goods and services, however those are valued. Alternatively, the goal could be to capture the full welfare effect of volunteer activity, including any utility flowing to the volunteer in addition to the benefit to the organizations to which time has been donated. The latter is certainly a more ambitious goal, and one that introduces the need for additional assumptions and information.
Adopting an approach to the valuation of volunteer labor that is consistent with the approach we have recommended for the household production account (see Chapter 3) would imply that one is interested only in the contribution of volunteer labor to the production of goods and services by the organization to which the volunteer labor is supplied. In this case, it seems natural to value volunteer labor at the cost of hiring someone to perform the same tasks, adjusting for possible productivity differentials between paid and volunteer labor, rather than at the opportunity cost of the person providing the volunteer services.
Applying the replacement cost method for valuing volunteer labor requires the identification of employment categories from which paid replacements could be drawn. In the past, this commonly has been done using very blunt wage information—e.g., the average hourly wage rate in the economy. A more targeted and useful approach would use the wage earned by workers who are hired to supply a similar kind of labor (e.g., using the wage of paid teacher aides to value the time of classroom volunteers). For volunteers who donate services that they also sell in commercial markets—such as accountants and lawyers who provide pro bono hours—these individuals’ market wage rates might be a reasonable estimate of the relevant replacement cost.
For the household production account, the panel advocates a modified replacement cost approach (see Chapter 3), and there is considerable appeal to using the same approach for the almost parallel situation of volunteer activity. The idea of adjusting the wage rate to reflect systematic differences in the skill and productivity of paid versus volunteer labor is conceptually akin to the idea of adjusting market wage rates to account for the relative productivity of home and market producers. Volunteers may not have the same level of specialized skills as paid workers; the work environment for volunteers may be quite different (e.g., less formal) from that of paid workers, even within the same organization; and, perhaps most importantly, we do not know the extent to which volunteers and paid workers with similar job titles actually perform the same tasks. For example, do volunteer fundraisers, firemen, teacher aides, and hospital “candy-stripers” have the same levels of responsibility, skill, and productivity as their paid counterparts? If there are reasons to believe that volunteers are generally less productive than their paid counterparts, the wages of paid labor in a particular job category would be an upper bound on the market value of volunteer time. The legal minimum wage would represent a lower bound on organizations’ cost to replace volunteers with paid labor, although there is no assurance that a user of volunteer labor would be willing to pay even the minimum wage for the quantity of volunteer labor being used.
It is worth noting that the Handbook on Nonprofit Institutions in the System of National Accounts supports an approach to valuing volunteer labor that is similar to this panel’s recommendations relating to unpaid work in the household (United Nations, 2003, p. 73):
The recommended procedure for the NPI satellite account is … a form of the replacement cost approach that ideally uses as the shadow wage for volunteers the average gross wage for the occupational activities in which the volunteers are involved, taking account of known large discrepancies in the skill levels of paid employees and volunteers. Since this requires more detail on the activities in which volunteers engage than is likely to be available in most countries, however, we recommend a fall-back approach that assigns to volunteer hours the average gross wage for the community, welfare, and social service occupation category.
The SNA choice of the fall-back occupation is rightly conservative—the wage rate for the “community, welfare, and social service” occupation category is below the average wage rate in most countries. But using this approach as a basis for valuing volunteer labor, while operationally appealing, has little conceptual justification. Moreover, it assumes that volunteers represent a cross-section of types of labor utilized in paid jobs, a matter about which little is known.
The welfare-based valuation of volunteer time can be examined in the framework of a model in which a person seeks to maximize utility (U) by allocating time among three uses: leisure, paid work, and volunteering:
U = U(L, E, V)
where L is hours of leisure, E is hours in paid employment, and V is hours of volunteer work. Time uses L and V are positively related with utility (∂U/∂L > 0 and ∂U/∂V > 0); hours in paid employment typically are assumed to have a negative relationship with utility (∂U/∂E < 0), though some people certainly may enjoy their market work effort. The three uses of time must all be non-negative and must sum to total discretionary time available. In equilibrium, a person seeks to equate the marginal value of time in all three uses, subject to the non-negativity constraint. The marginal value of time devoted to leisure or volunteer work is just the marginal utility associated with the activity; the value of market work equals the wage that is earned by working less the disutility, if any, associated with the performance of that work. A person may choose to perform no paid labor, which is typical of retired people, or to engage in no volunteer work, which characterizes about 70 percent of all adults.
For persons who work and also volunteer, which describes some 30 percent of paid workers (U.S. Bureau of Labor Statistics, 2003), the equilibrium values of an hour of both V and L should be equal—to each other and (in the absence of marginal utility or disutility from time spent in market work) to the marginal wage, W, in the paid labor market. W is observable and provides a basis for imputing a value, at the margin, to volunteer time.
In this framework, volunteering contributes to economic welfare in two ways—as a labor input to production and as a provider of utility to the volunteer.
A full accounting for the contribution of volunteer labor to economic welfare would identify, measure, and sum these values. In the simplest case, the value to the individual of participating in volunteer activity is just the foregone wage. If volunteers actually receive a negative wage by incurring costs (of travel, meals, uniforms, requirements to donate, and so on) in return for the opportunity to engage in volunteer activity, then the volunteer is deriving utility that exceeds the market wage that is foregone.3 Viewed in this way, the total contribution of an hour of volunteer labor to economic welfare is the sum of the positive value accruing to the recipient organization and the potentially even larger positive value accruing to the volunteer.
Conceptually, there is merit to the method that reflects the value the volunteer places on the opportunity to participate in volunteer activity. Examining the activity as a barter arrangement, in which the organization derives a marginal product of labor and volunteers derive utility, may reveal information about the value of the volunteer time to the employing organization as well as to the volunteer. At this time, however, the methodology to produce such estimates is not developed, and, while this direction for future research should be explored, it is difficult to envision how reliable data for this approach would be produced.
Recommendation 7.3: Volunteer labor should be valued, at least initially, by finding the appropriate market analogues—wages paid to employees working in similar occupations. To the extent possible, this replacement cost approach should be modified to reflect skill and effort differences between volunteer and paid labor.
If productivity adjustments cannot be made, we suspect that the replacement labor wage generally will represent an upper bound on the value of volunteer time inputs. This is a subject on which additional information certainly would be welcome.
Recognizing that there are many uses of this kind of data and that no one approach is best for all of them, we take the position that multiple approaches should ultimately be accommodated. For example, given that both a replacement cost and an opportunity cost approach may be useful, the satellite account could contain two or more sets of value data. Additionally, the panel favors the publication of information that presents quantities (person-hours) of volunteering, by type of work performed and by type of provider organization (specific industry, government, or nonprofit), separate from the values, so that each may be used as needed for a particular application.
Labor is not the only input that is donated to nonprofit or public organizations and that, as a result, is unpriced and unmeasured in the NIPAs. Many of the issues discussed above in the context of volunteer labor also apply to donations of goods. Estimating replacement costs to recipient organizations is an option for valuing donated goods, just as it is for valuing volunteer labor. It should be noted, however, that just as in the discussion of how to value volunteer time, it is by no means clear that recipient organizations would, in the absence of the donations, purchase the donated goods at anything close to the observed market prices. Thus, market prices are likely to be upper bounds on the values to recipient organizations of donated goods. In contrast with the volunteer labor analysis, however, it is less obvious that corporate donors, or perhaps even individual donors, derive utility from the act of donating goods. For corporate donors, such donations could represent investments in goodwill and future sales, and the return on these investments would eventually appear on the donating companies’ books as market transactions.
Scant attention has been given to the magnitudes or values of goods donated to nonprofit and public organizations. The values of donated goods as reported on itemized individual tax returns seem likely to be overestimates of market values. For example, a recent report by the U.S. General Accounting Office (2003) looked at donations of automobiles to charities and concluded that the amounts deducted on itemized tax returns by donors substantially exceeded the amount of money realized by the charities when the cars were later sold. The values of corporate donations of goods also may well be upward biased insofar as they reflect corporate tax laws that permit deductions based either on market prices for the donated products or on average production costs even when marginal costs are far lower. Thus, the market price of goods donated to public or nonprofit organizations is an overestimate of what would be purchased at that price, and allowable tax deductions for donated goods are also overestimates of the marginal cost of producing the donated goods. Still, if these measures of the value of donated goods could be obtained from the IRS or other sources, they would constitute at least a first step toward extending the coverage of satellite accounts to include goods donated to nonprofit and public organizations. More refined measures can and should be pursued over time.
MEASURING AND VALUING OUTPUT
What are the outputs of the government and nonprofit sectors? These sectors are dominated by services—and difficult ones to measure at that—education, health, social services, culture and the arts, recreation, and others. As noted elsewhere in this report, a major challenge in measuring the output of such services is simply to define the unit of output that is being transacted (Griliches, 1992).
Government spending represents about 18 percent of U.S. GDP. These expenditures go toward provision of goods and services ranging from defense to the judicial system to education. GDP excludes the transfer payment components of government spending (social security, welfare benefits, unemployment benefits, and so on).
Statistical agencies have made some attempts to measure government output directly rather than indirectly. Jenkinson (Organisation for Economic Co-operation and Development, 2003) offers a summary of international efforts, most of which aspire to the SNA recommendation to pursue direct volume measures, as opposed to the traditional input-based methodology. Atkinson (2004) provides an interesting review of the United Kingdom’s efforts in this regard. Sectors typically covered as countries begin to develop direct measures of government output include education—where measures such as number of pupils and pupil hours are calculated—and health—where indicators include such things as number of patients and hospital treatments. But valuation of these outputs is difficult. Observed prices, even when they are positive, as with hospital charges to health care insurers, may well be of limited usefulness. In the case of health care, for example, because patients confront far lower marginal prices than do insurers but also confront a variety of nonprice rationing mechanisms, the marginal value to the patient of treatment provided could be either lower or higher than its cost. In work discontinued in the mid-1990s, The BLS developed measures of output for selected federal government agencies as part of its government productivity measurement program. These output measures typically relied on quantity indicators of one sort or another (Fisk and Forte, 1997). Work to measure government output for the United States is currently under way at the BEA (see Fraumeni et al., 2004); the goal of this work is monetary valuations.
In a similar spirit, the SNA categorizes physical output measures for 11 nonprofit sector groups (United Nations, 2003):
culture and recreation;
education and research;
development and housing;
law, advocacy, and politics;
philanthropic intermediaries and volunteerism promotion;
business and professional associations, unions.
Each of these groups contains numerous fields, subfields, and target physical output measures. Several countries—including Australia, Belgium, Canada, Italy,
Mozambique, the Netherlands, the Philippines, South Africa, and Sweden—have begun to develop new accounts intended to estimate the size of the nonprofit sector and to support international comparisons of nonprofit activity. Much of this work is progressing within the framework provided by the Handbook on Nonprofit Institutions in the System of National Accounts guidelines.
In some situations, outputs of governments or nonprofit organizations are not sold at a positive price because they are difficult or impossible to sell—as is the case with collective goods such as national defense, basic medical research, and environmental protection. In other cases, when outputs could be sold (e.g., education), a social judgment has been made not to do so. In still other cases, outputs are sold, but at prices not covering their full costs—e.g., service on some mass-transit systems and, in the nonprofit sector, college education even for the typical student but especially for those with scholarships. There is nothing necessarily inefficient about such pricing policies, but the social valuation of the associated outputs remains a vexing problem.
When the output is a purely collective good, it is by definition infeasible or inefficient to exclude particular individuals from its consumption. There is no practical way, for example, to exclude any person from the benefits of national defense, species preservation, or locally, clean air. In some cases, such exclusion may be technologically feasible, as with public radio or public television. The Public Broadcasting System could institute pay-per-view charges for particular programs, as sometimes is done by commercial vendors. Even when feasible, however, such pricing may be economically inefficient because there is no marginal cost of allowing an additional person to have access to the service.
The Zero Price Problem
How should one think about a price of zero? A price of zero could imply a marginal value of zero, as is surely the case for a vast number of potential outputs that are not provided at all because there is little or no demand for them. But a price of zero for pure public goods is another matter. Though observed prices are, and to be efficient should be, equal to zero, to conclude that such outputs have no value to society does little to inform research in most policy areas. Likewise, the often-invoked convention of valuing output on the basis of prices paid for inputs, whether positive or zero, is largely arbitrary. Under conditions characterized by positive “congestion costs”—when consumption by one person imposes external costs on others (as on crowded roads and in air-polluted areas)—efficient prices would be greater than zero. The absence of estimates of these values is a current shortfall of the NIPAs, and more attention to them and to their changes over time and across areas would be useful.
The key issue here is the relationship between the valuation of outputs at their market prices and the economic welfare interpretation of those values. Market prices (and the associated quantities) have general utility as an indication
of the value that purchasers attach to the marginal output. When explicit prices are not charged for output, as is common in the government and nonprofit sectors, buyers’ valuations of the outputs are not available.
It should be noted that, even when market prices are available, they reflect buyers’ willingness to pay for a marginal unit of output, not the willingness to pay for the entire value to them of the good or service. The difference between the two values, a matter discussed in earlier chapters, is attributable to the inclusion of consumer surplus in the latter but not in the former. For aggregate welfare measurement purposes, and to measure changes in welfare over time and across countries, information on consumer surplus is conceptually both relevant and important; the inclusion of consumer surplus in a satellite account for government or the nonprofit sector, however, would make that account incompatible with the NIPAs.
Even with a narrower focus on marginal valuations, problems associated with outputs having prices of zero remain. One alternative is to value such outputs by using the prices paid for the inputs to their production. When outputs are not sold, it is nonetheless true that their production typically requires purchased inputs of labor, raw materials, and capital (in addition to such donated inputs as volunteer labor). Aggregating market input values is, in fact, the methodology used to value the output of most government as well as nonprofit-sector services but, as we explain above, this procedure can be thought of as a pragmatic compromise when the alternatives are either to value those outputs at zero because they are not sold in the market or to value them at some positive, imputed, amount.
Asserting that outputs have values equal to the market values of the inputs with which they are produced imposes a strong, controversial assumption about output beneficiaries’ willingness (and ability) to pay. Whether incremental expenditures on national defense, basic research, charity services to the poor, or roads are “worth” their cost—not more, not less—remains a subject of active debate.
Ruggles (1983) defends the idea of measuring government output by the market value of inputs. The claim is that, in a democratic society, government will supply goods and services in amounts that reflect citizens’ willingness to pay taxes for their provision. It would follow that the total value of output is at least equal to the cost of the inputs. This perspective deserves further attention, but it is by no means unproblematic. It assumes that voters are well informed about the values of public-sector outputs, have the choice to vote on specific “earmarks,” and do not engage in strategic behavior, depending on the specific tax mechanism expected to be used to finance the activity. It also assumes consumers’ knowledge of that finance mechanism. In addition, this perspective relies on consumers’ (voters’) assessments of the total value of the service output, not the marginal values as would be captured by private market prices.
Whatever the merit of Ruggles’s political-economic rationale for valuing government output by the cost of the inputs used to produce it, essentially the same arguments and limitations apply to the valuation of nonprofit-sector activity. Instead of relying on majoritarian voting processes, for the nonprofit sector, one
would rely on donors’ willingness to contribute and, hence, to pay for inputs purchased in ordinary markets, as indicative of the value of outputs to them. If this approach were adopted, one would also need to account somehow for output financed by profits generated through user fees and auxiliary activities, which vary greatly across industries. Also, “free-rider” behavior, leading to suboptimal donations, can be expected to generate expenditures on inputs that fall short of efficient levels, although tax subsidies to donee organizations (exemptions from taxes on property, sales, and profits) and the deductibility of donations on personal income tax returns exert a countervailing effect. It is likely that, in the nonprofit sector, the observed levels of output are short of those at which the aggregate willingness to pay for marginal output equals the marginal supply cost, but this is not certain.
What, then, are the options for measuring the value of nonmarketed outputs or, more generally, outputs that, if not provided at a price of zero, are offered at subsidized prices? Even when market prices are available as indicators of value, the appropriate measure depends—as we note throughout this report—on the use to which an output measure is to be put. Information needed for benefit-cost analyses is not the same as that needed for assessing growth of output or of economic welfare over time. Both differ from the data captured in national accounts that are based primarily on transactions prices and that in many cases value inputs and outputs at zero when no explicit prices are paid. To see this, consider cases in which public policy involves decisions about the efficiency of a governmental action—a plan to construct a dam, a new environmental pollution regulation, or an increase in funding for basic science research through the National Institutes of Health or the National Science Foundation. The benefit component of benefit-cost analysis requires, conceptually, valuation of the aggregate willingness to pay for the proposed outputs, amounts that encompass the total area under all beneficiaries’ demand functions, not simply estimates of the marginal value to society of the dam, the environmental protection, or the research.
Acknowledging the difference between measurement for benefit-cost analysis and measurement for consistency with national economic accounting is critical for both the government and nonprofit sectors. Benefit-cost analysis addresses an explicitly normative question of whether a particular output is “worth” the cost of producing it, which requires measuring the total value of a specified quantity of a public- or nonprofit-sector activity for comparison with the associated cost. To this end, the value of public goods can be estimated by the sum of consumers’ willingness to pay as derived from surveys that would in principle, but with great complexity in practice, capture consumer surplus, or by some imputation from observed behavior.4 By contrast, measurement of the output of public goods that
is consistent with NIPA concepts requires imputation of marginal willingness to pay for the quantities being supplied.
All of these options for valuing government and nonprofit-sector nonmarket outputs are problematic. Nonetheless, there is no justification, except on grounds of convenience and consistency with the use of observed market prices, for placing values of zero on nonmarket activities. If this approach were followed, the experimental satellite account would provide no new information relative to the NIPAs, and there would be little point to pursuing its construction.
In the government and nonprofit sectors some outputs are sold, while others are not. In the hospital and nursing home industries, for example, as well as in private higher education, outputs are sold to some consumers, and at multiple prices, but given away to others. In such cases it is standard practice under generally accepted accounting principles (GAAP) for nonprofit organizations, in their filing of IRS Form 990 returns, to “gross up” the charity services of health care or education that they have provided. That is, organizations report not the realized revenue but the revenue that would have been realized had the free care or education instead been sold. This accounting practice, involving the imputation of revenue to outputs not actually generating revenue, assumes implicitly that the value of the charity care (or low-cost education) to the consumers who do not pay for the services they receive should be estimated by the value of similar services provided to the consumers who do pay. It is clear, of course, that the charity patients (or scholarship students) are typically not, in fact, willing to pay—the willingness to pay concept encompasses the ability to pay as well—the market prices. In that sense the imputed values overstate market values. The GAAP convention on imputation of charitable health care and education activities has been developed with respect to private nonprofit organizations, yet the fundamental issue it addresses is far broader. The grossing up of revenue, or more generally of output, to value unpriced output poses conceptual problems of value imputation at both the private accounting and national accounting levels.
Imputation is not a process that is used consistently in the NIPAs, but neither is it unprecedented. As is noted throughout this volume, some values are imputed—for example, unobserved values (annual flows) of owner-occupied housing services. The potential for making greater use of imputations remains. For nonprofit-sector accounts, the usefulness of imputing values to outputs exchanged at prices of zero, and the alternative mechanisms for making the imputations, deserve more attention. For the government accounts, the imputations for services of capital, which now are assumed to equal depreciation, might in the future include a net return (which separates out depreciation). For governmental and nonprofit organizations, the research knowledge base is currently insufficient for identifying a best approach for imputing values to unpriced outputs—in part
because there are multiple uses for such imputations. Priority should be given to research that would lead to professional consensus on a small number of approaches for imputing values to unpriced governmental and nonprofit-sector outputs, and to developing methodologies for implementing each.
The BEA is already using a broad array of data sources to account for income and outlays of nonprofit institutions that serve households. Data on expenditures of nonprofit institutions originate from the Census Bureau’s quinquennial economic census and the annual economic surveys. Additional data sources are used to fill gaps for labor, political, religious, and educational organizations. The Urban Institute’s National Center for Charitable Statistics disseminates information about the reports that tax-exempt institutions file with the IRS concerning their income, expenditures, and activities. Other data come from the American Association of Fundraising Counsel’s Trust for Philanthropy and surveys of charitable contributions to religious and other nonprofit organizations conducted by the Independent Sector.
To push research further, we note five areas in which further data development activity would be helpful. First, for documenting activity in the nonprofit sector, improved access is needed to existing administrative records—including forms that organizations file with the IRS, as well as other financial statements and records. The Urban Institute’s recent initiative has made the IRS Form 990 information easily accessible for anyone with an Internet connection, but there is other useful information that remains inaccessible. As an example, we support permitting access by researchers, under appropriate safeguards, to the confidential Form 990T data on commercial activity by nonprofit organizations that is “not substantially related” to its tax-exempt mission.5
Second, coordination of and access to data maintained by various nonprofit and government organizations can be improved. A good example of a situation that should be addressed is the lack of coordination between the BLS and the Census Bureau business lists. The Confidential Information Protection and Statistical Efficiency Act of 2002 (CIPSEA) has facilitated some information sharing among the statistical agencies for statistical purposes. Unfortunately, until companion legislation that modifies existing IRS statutes is passed to allow the Census Bureau to share information based on tax records with other statistical agencies, much useful work will be stalled.
Third, improvement is needed in the coverage of giving and volunteering in household surveys. The private surveys that have been conducted generally have small sample sizes and relatively low response rates, and they are not conducted on a regular basis. Even the CPS supplement data cited earlier may be subject to bias related to the length of the recall period for which respondents are asked to report their volunteer activities. The ATUS should be a boon to researchers wanting to compile data on time spent in volunteer activities.
Fourth, new surveys of nonprofit institutions could provide much valuable information. Two types of additional information would be particularly useful. On the input side, existing data on volunteering have been derived almost exclusively from household surveys. It would be valuable to have data on volunteering from the perspective of the recipient nonprofit and government organizations. How much volunteer time is used by organizations of various sizes and in various industries? Would the organizations use more volunteer time if it were available—that is, are they supply constrained? What would it cost to hire people to perform the work now done by volunteers? Would they hire the replacements if the volunteers were not available? What is the most they would pay if they had to hire them? Turning from volunteers to other unpriced inputs, what kinds, and how much, of other donated inputs do they receive—food, equipment, office or production space, and so on? On the output side, how do organizations measure their performance? How do they, or would they, estimate the value of the outputs? Whether the typical nonprofit organization would be able to answer all of these questions is uncertain, but there would be value in attempting to learn what information they can provide.
Fifth, it would be useful to identify and develop data that create options for valuing hours spent in volunteer activities. Shadow wages based on similar paid occupations would be a sensible starting point, but research to assess the relative productivity of paid and volunteer labor performing similar tasks also would be worthwhile.
In considering how to progress with research to develop government and nonprofit satellite accounts, it is important to keep clear the distinction between what is optimal conceptually versus what is the best that can be done operationally. This chapter has only touched on the complications inherent in the construction of these satellite accounts. Pieces of the puzzle are within grasp. This report, and others, have described how a nation’s volunteer labor inputs could be counted and valued. Such data would be useful to research in a number of policy areas—for example, that aimed at improving provision of public services and defining the role of the state—even if it is missing the output side valuation needed for a fully specified income and product account. It is realistic to believe that an account can be organized to provide a more comprehensive picture of the market
inputs and outputs associated with legally defined private nonprofit organizations. As noted earlier, BEA is already working to produce new aggregate tables of nonprofit-sector activity. Such data will improve policy makers’ ability to track the role of nonprofit institutions in providing services in specific areas such as education and health.
At this point, however, full and independent (non-input based) valuation of goods and services produced by government and by the economy’s nonprofit institutions remains a long way off. For the foreseeable future, national accountants and other data producers will have to choose from more standard options: valuing output at exchange prices (often zero); valuing output based on the cost of inputs; or imputing values based on GAAP (the generally accepted accounting principals), which attempts to approximate the prices at which the outputs of nonprofit organizations in the education and health sectors could have been sold. In some cases, practitioners may choose simply to provide physical indicators of output.
One obstacle to producing a double-entry accounting of the nation’s nonprofit activities is that they defy easy categorization. Nonprofit organizations and volunteer labor operate conspicuously and independently across many sectors of the economy. Volunteer labor is offered not just to nonprofit institutions, but also to government and other organizations, and nonprofit as well as governmental organizations utilize both paid and volunteer workers. All of this makes it problematic to prescribe a unified approach.
Much of the activity of the public and nonprofit sectors involves preservation and development of human capital. Conceptually, the output associated with public and nonprofit inputs to health should be valued in ways consistent with the methodology prescribed for the health account. A different methodology would be used for education and each of the other areas in which government or nonprofit institutions are active. In a real sense, development of the experimental health and education accounts is a prerequisite to the development of more comprehensive government or nonprofit satellite accounts.
The principal contribution of a satellite government account is the opportunity to make a bolder and more experimental stab at independent valuation of government output. As just indicated, work to construct such an account necessarily will draw from counting and valuation methodologies used in the accounts for the various sectors in which there is a strong government presence. As output valuation methods improve in health, education, and other services, the ability to estimate government output independently by definition also will improve.
A nonprofit-sector satellite account would require, at a minimum, a selective rearrangement of data for education, health, and other social services, and would entail major overlap not only with satellite accounts for activities in these areas but with market accounts as well. As with the government satellite account, a nonprofit account could embody numerous valuation methodologies. This fact underscores the panel’s view that there is no single right way to present input and
output data. In principle, data could be presented at an “atomic” level, such that the atoms could be aggregated in any way the user wished. As a practical matter, decisions must be made about which aggregations are worth making, as has been done historically for the NIPAs and will need to be done for the satellite accounts. Categorization by industry has proven useful, as has the consumption and investment delineation and the separate presentation of information for government and private entities. In advocating the development of a nonprofit satellite account, we are in essence making the case for expansion of the ownership criterion to include separate designation of the nonprofit form. For some purposes, one may wish to know, for example, the size and activity levels of the hospital and education industries, and their changes over time. For other purposes one may wish to know the size and activity level of the government, nonprofit, and for-profit sectors across industries. Viewed in this way, detailed information about industries does not make the government and nonprofit accounts redundant with other nonmarket accounts, such as health and education.