Measuring Growth of the Nonprofit Sector: The Choice of Indicator Matters

Deciding Which Organizations to Measure

Although the debate about nonprofit sector definition is an old one among nonprofit scholars, the question of what to include in this definition is not yet settled (see Defourny, Grønbjerg, Meijs, Nyssens, & Yamauchi, 2016; Salamon & Sokolowski, 2016). In many countries, the nonprofit sector is not a distinct sector in official statistics (United Nations, 2003, p. 1). This leads to relatively poor data, few reporting regulations for the organizations and much less knowledge of the sector compared to the for-profit sector. Consequently, researchers often do not have much choice about the indicators in use. In many countries substantial information deficits exist concerning the nonprofit sector, and the datasets available will usually not have been created for the sake of scientific research.

For some countries, it is possible to use nonprofit institutions serving households (NPISH) as a sector definition. This is part of the System of National Accounts (United Nations Statistics Division, 2009), which has the advantage that data are collected by internationally established standards and are available across many countries for a long period of time. However, this only captures one part of the nonprofit sector and excludes nonprofit organizations that are more business-like and rely more heavily on sales revenue to cover expenditure (Y. H. Kim & Kim, 2016).

Counting Nonprofit Organizations

The starting point in analyzing size variations is counting the number of nonprofit organizations. This is the most commonly used indicator in this line of research (Corbin, 1999; Grønbjerg & Paarlberg, 2001; J. Kim, 2000; M. Kim, 2015; Lecy & Van Slyke, 2013; Lee, 2010; Liu, 2017; Luksetich, 2008; Marcuello, 1998; Matsunaga & Yamauchi, 2004; Van Puyvelde & Brown, 2016). Using the number of nonprofit organizations is attractive, because it is a relatively easy to obtain and easy to understand measure. Depending on the data source, it may be a very comprehensive measure, taking into account all nonprofit organizations irrespective of their activity field or degree of professionalization. However, it omits the aspect of organization size and thus market composition. Is it accurate to talk about stagnation in a case where only the existing nonprofits grow (e.g., in employed staff, offered services, income or assets), but no new nonprofits are being founded?

When counting organizations, it is useful to contrast active and registered nonprofits. Casey (2016, p. 189), for example, reports on a decrease of 200,000 nonprofits in the United States between 2010 and 2013 that can be explained with a change in reporting regulations as most deregistered organizations were inactive. Conversely, not all active nonprofits will necessarily have a reporting or registration requirement; often reporting thresholds by income will exclude smaller organizations. The difference between the population of organizations and those included in the data can be quite substantial. Furthermore, changing reporting thresholds within countries can make analysis over time problematic. Harrison and Laincz (2008) report that only about 38% of existing nonprofit organizations in the United States are included in the IRS data, with others falling below the threshold or failing to report. Within the United Kingdom, charities in England do not have to register if their income is below £5,000; while in Scotland all organizations must register to operate as a charity. Austrian administrative data on nonprofit organizations only contain organizations that employ at least one paid person. According to estimations, only around 9% of all existing nonprofit organizations have paid staff and are consequently required to report (Pennerstorfer, Schneider, & Badelt, 2013). The small organizations that are left out in many databases typically are self-help or voluntary grassroots initiatives that represent a specific part of the nonprofit sector (Toepler, 2003).

Measures that focus on the flow, rather than the stock, of organizations can capture sector dynamics (Saxton & Benson, 2005; Yu, 2016). While the inflow of new registrations may be well-recorded in administrative data, it can be more difficult to distinguish organizations in the outflow that cease to operate, miss filing deadlines, merge, or stop being required to register.

Measuring Financial Characteristics

Compared with for-profit organizations, nonprofit organizations have very diverse funding sources ranging from earned income to donative income and membership fees. Consequently, we also find different income measures that could serve as a size indicator. Example income measures include donations per GDP (Pryor, 2012) and total revenue (Backus & Clifford, 2013; Chikoto & Neely, 2014; Clifford, 2017; Paarlberg & Gen, 2009).

When using only one income source as an indicator, it is important to think about whether the income structure is comparable over time or space. In Austria, for example, the funding mode—and consequently the income structure—of many nonprofit organizations has changed over the past 20 years. Many services delivered by nonprofits were previously funded with lump-sum government subsidies. This changed as public service contracting became more popular and now income from service contracting is recorded as earned income and not as a government grant (Pennerstorfer et al., 2013). Measuring organizational growth with income from sales, in this case, overestimates growth as only the income structure changed without a corresponding increase in services. Conversely, relying only on a measure of total income does not capture the diversity of income sources. Organizations reliant on a small number of income sources rather than diversifying are likely to experience more volatile fluctuations in their growth (Carroll & Stater, 2009; Mayer, Wang, Egginton, & Flint, 2014).

Expenditure variables can be a proxy for output, especially useful when services are delivered below market prices or free of cost to service users, rendering income an incomplete measure. However, it is usual to only register activity that actually produces expenses. Services that are (co-)produced by (wage-free) volunteer work are not registered (fully) and thus expenditure underestimates the true size of an organization. In addition, improvements in efficiency will underestimate growth. Various output measures such as number of clients or housing units are often used when particular nonprofit activity fields are the focus. It is, however, not possible to use this sort of indicator when measuring the whole nonprofit sector, because there is no single meaningful output measure for all organizations.

Assets are also used as a proxy for size (Chikoto & Neely, 2014; Chikoto-Schultz & Neely, 2016; Harrison & Laincz, 2008; Liu, 2017; Twombly, 2003). In a sector with fluctuations in income due to government policy and funding arrangements, assets might arguably provide a longer term measure of the size of the sector. Harrison and Laincz (2008) use this indicator and also discuss its problems. They mention that some nonprofits have large endowments compared with the volume of provided services so that these organization have more weight in estimates of sector growth (Harrison & Laincz, 2008, p. 20).

Measuring Nonprofit Personnel

A distinctive feature of the nonprofit sector is a workforce that consists both of paid employees and volunteers. Several studies use measures such as total employment (Bae & Sohn, 2018; Ben-Ner & Van Hoomissen, 1992) or volunteer and paid employment as a percentage of nonagricultural employment (Matsunaga, Yamauchi, & Okuyama, 2010; Pryor, 2012; Salamon & Anheier, 1998; Salamon, Sokolowski, & Anheier, 2000). Paid employment can be measured by person-head count, full-time equivalents, or employment relative to the population. These measures only capture nonprofit organizations which employ paid staff, giving greater weight to industries such as social and health services that heavily rely on personnel input.

Volunteers are a very important and sometimes the only input factor in many nonprofits. Harrison and Laincz (2008, p. 20) cite that for the US neglecting volunteer work and using only paid employment would underestimate the economic weight of the nonprofit sector by approximately 40%. Consequently, the unpaid workforce should ideally be included in a measure that concentrates on personnel. However, the lack of a market transaction in volunteering means that it often leaves little trace in the data. Usually, volunteers are not part of the organizations’ accounting system (Sajardo & Serra, 2011, p. 875) and there may be no official reporting requirements on volunteer volume, so that organizations without a professionalized volunteer management might not even be able to make a good estimation of their total volunteer volume (Mook, Handy, & Quarter, 2007).

Measuring Nonprofit Sector Composition

Nonprofit sector growth can result from two sources: the increase in the number of organizations (market entry) and/or the increase in organization size (organization growth). Both Tucker and Sommerfeld (2006) as well as Backus and Clifford (2013) consider the composition of the sector. The former analyze change in establishment sizes in U.S. social service organizations; measuring size with employment (head) count. The latter have a similar endeavor, analyzing the composition of English and Welsh charities and differences in growth rates of organizations of different size (Backus and Clifford, 2013). In their article, size is measured by the income of the charity (p. 764).

We have seen that the choice of measures could be determined by data availability, data quality, applicability to the (sub)sector of interest; and to the research question being answered. But importantly, choosing alternative measures may lead to different conclusions and so an appreciation of the sensitivity to choice of measure is critical in evaluating a research finding. We now go on to illustrate this point empirically using nonprofit data from both Scotland and Austria.

Data on the Austrian Nonprofit Sector

As in many other countries, quantitative information on the Austrian nonprofit sector is scarce. No statistic covers the entire sector (e.g., Neumayr, Pennerstorfer, Vandor, & Meyer, 2017). The only administrative data that are available come from the Austrian pay slip and sales tax statistics provided by the Austrian Statistical office. They are available for the years 2003 to 2014 and incorporate only nonprofit organizations that had at least one paid employee. As a size measure we use the number of organizations in each year, the income from sales, yearly payroll expenses, and number of issued pay slips per year. All monetary values are deflated using the Austrian consumer price index with the year 2000 serving as the base year.

All Austrian employers have to report gross wages paid to employees in a given calendar year. Wages, on average, account for around 70% of total expenses in the Austrian nonprofit sector with paid employment (Pennerstorfer, Schneider, & Reitzinger, 2015). A pay slip is a sort of employment head count. It is issued yearly for each employee, but also each time the employment status of an employee changes. The number of pay slips therefore includes two dimensions—possible “real” employment growth, but also employee turnover. Sales income can consist both of (privately) earned income and income from service-contracting with public authorities. Although not all organizations have sales revenues, it is the most important income category for Austrian nonprofit organizations. On average, sales revenues account for around 72% of total income for an Austrian nonprofit organization that has paid employees (Pennerstorfer et al., 2015). Almost 70% of all sales income comes from public contractors, although there is a great variability between activity fields (Pennerstorfer et al., 2015).

Data on the Scottish Nonprofit Sector

Nonprofit organizations in Scotland that wish to operate as charities must register with the Office of the Scottish Charity Regulator. Charitable status allows organizations to access a range of tax benefits in return for increased reporting requirements. Detailed organizational characteristics are collected at registration and through annual returns. Reliable information is available on charity finances, drawn from organizations’ annual accounts, and broken down into standard subcategories. However, unlike Austria, there is very little reliable information on the numbers of staff employed by organizations. Here we use a dataset of all active charities in Scotland between 2008 and 2014 that have submitted annual returns (McDonnell & Rutherford, 2018). All monetary values are deflated using the U.K. consumer price index with the base year as 2000.

Comparing the Growth Measures

For the empirical example, we present two different perspectives of growth—aggregate sector growth and organizational growth within the sector. Starting with the sectoral perspective, Figure 1 depicts yearly nonprofit sector growth rates for Austria and Scotland. Between the years 2003 and 2009, all Austrian measures register growth of the nonprofit sector although growth rates differ. After 2009, some measures register a contraction of the sector in some years. The year 2010 stands out as a year with an actual decline of aggregated sales income, coinciding with the economic crisis and consequent welfare retrenchment measures, while growth in the total number of organizations went up in this year. Between 2004 and 2010, growth of sales income was smaller than growth of aggregated payroll expenses, but this reversed between 2011 and 2013.

OPEN IN VIEWER

Figure 1.

Yearly nonprofit sector growth rates for Austria and Scotland.

A comparison of the aggregated values and the number of organizations contributing to these values enables us to make a statement about the composition of the sector. While growth in the number of organizations with income from sales has slowed since 2009 to an actual contraction after 2012, growth in aggregated sales income has risen between 2010 and 2013, meaning that the average sales income for an organization that receives sales income has increased.

Looking at the growth rates of income and expenditure in Scotland, we find a similar pattern to Austria with positive growth rates prior to 2010 and negative growth rates until 2013. Income and expenditure growth are fairly stable and mirror each other. For Scotland, we include two measures of assets: “Current Assets” are the liquid cash available to nonprofit organizations, while “Total Funds” is a measure of all assets, including property and long-term investments. Current asset growth is a much noisier measure; negative pre-2010 when the other measures were positive, and declining again in 2014 when income and expenditure growth are at their peak. Total Funds growth is smoother, but appears negatively correlated with income and expenditure. Current assets dropped dramatically in 2011, without a corresponding drop in Total Funds. While our initial enquiries established that this measure had not changed definition across the time period, a more in-depth examination of the historical reporting forms showed that the wording on other measures of assets had been altered slightly in that year. This appears to have led a number of large charities to reclassify their assets out of the Net Current Assets measure. These assets are still included in the Total Funds measure. This explains the apparent large drop in the current assets measure and is an example which highlights the potential for measures to be strongly influenced by the way in which data are collected, even if the wording of the specific variable being analyzed remains constant.

Growth in the number of charities peaked in 2009, falling negative from 2013. The five measures clearly show different patterns, most pronounced in 2010 to 2012 when the number of organizations was growing, income and expenditure was stable or falling slightly, liquid assets were falling sharply, while total funds grew.

Table 2 presents correlations between the growth rates of the various indicators. Values for Austria are reasonably correlated, with the highest correlation between growth of aggregated payroll expenses and growth of issued payslips. Aggregated sales income and the number of organizations exhibit the lowest correlation. Correlation values for Scotland make evident that the use of income and expenditure are quite similar, while there is virtually no correlation between income/expenditure and assets as well as the number of organizations in the nonprofit sector.

OPEN IN VIEWER

Table 2.

Correlations Between Growth Indicators.

Austria
Growth of	Sales income	Payslips	Payroll expenditure	N
sales income	1
payslips	0.53	1
payroll expenditure	0.60	0.95	1
N	0.43	0.80	0.75	1
N (income)	0.58	0.76	0.73	0.88
Scotland
Growth of	Income	Expenditure	Current assets	Total funds
income	1
expenditure	0.96	1
curr’t assets	0.22	0.19	1
total funds	−0.26	−0.38	−0.73	1
N	−0.10	0.05	−0.18	−0.26

Source. Austrian and Scottish NPS data, own calculation.

Next we examine the nonprofit sector composition, following Backus and Clifford (2013). In their article, the authors ask if big charities are increasingly becoming more dominant, using data for English and Welsh charities for the years 1997 to 2008. They use income as a size variable to describe changes in organization size and thus market composition, and concentration. In a cross-sectional perspective, two ratios are calculated to examine changes in the concentration of income: the 100-charity concentration ratio and the top 1% share (Backus & Clifford, 2013, p. 764f.). In the cross-sectional perspective for England and Wales, the authors find “relatively stable or slightly declining levels of concentration” (p. 766). We replicate this example for Austria and Scotland, using not only income as a variable but also expenses and assets, to analyze whether the answer to the question differs when choosing a different indicator.

Here we present the top 1% share, defined as the share of total income (expenditure, assets) captured by the largest 1% of organizations each year (Backus & Clifford, 2013, p. 764). The ratio takes into account that the number of organizations changes each year. The ratio is calculated as

c k t = ∑ i = 1 N δ i t y i t ∑ i = 1 N y i t ,

where c k t represents the top 1% share (k = 1) and y i t is the income (expenditure, assets) of organization i = 1 , … , N . δ i t is a dummy equaling 1 if the organization is among the top 1% of organizations in each year, and 0 otherwise.

Figure 2 shows the two concentration measures for Austria and Scotland. Turning to the income variables, we find an increase of income concentration between the years 2004 and 2009 in Austria when using the top 1% followed by a decline to levels that are very close to the year 2004. Using expenditure or employment instead of income, we find an increase in concentration. Using expenditure or employment as size variables, we could say that big organizations were becoming more dominant in the sample, while we would not have made this conclusion using the income variable.

OPEN IN VIEWER

Figure 2.

Changes in sector composition.

Results for Scotland are even more striking. In Scotland, we find a rising income concentration for the years 2008 to 2014. However, asset concentration sharply declines across the period. Like Austria, we would also draw different conclusions about the dominance of big charities if we used assets instead of income as our measure.