RESEARCH NOTE. Metropolitan Nonprofit Sectors Bielefeld. Findings From NCCS Data. Wolfgang Bielefeld Indiana University

Transcription

1 Metropolitan Nonprofit Sectors Bielefeld RESEARCH NOTE Metropolitan Nonprofit Sectors: Findings From NCCS Data Wolfgang Bielefeld Indiana University Data from the National Center for Charitable Statistics (NCCS) and other secondary sources was used to examine the nonprofit sectors of nine metropolitan regions. The results indicate that nonprofit sectors vary widely in terms of the numbers of organizations in them and the proportions of different types of providers. Moreover, the findings showed complex and intriguing relationships between nonprofit sectors and political culture, generosity, wealth, poverty, and heterogeneity. Traditionalistic sites had sectors with the opposite characteristics. The sectors in individualistic sites lay between these two patterns. Wealthier sites had larger, better-supported and secure sectors. Sites with higher poverty had less well supported sectors with smaller human service components. The most and least heterogeneous sites had the largest and smallest nonprofit sectors respectively. These findings bolster confidence in the use of NCCS data. The National Center for Charitable Statistics (NCCS) has recently made its IRS Return Transaction Files (hereafter designated the NCCS Core Files) available to researchers. These files contain a considerable amount of information from the Form 990s of 501(c)(3) operating organizations with $25,000 or more in annual gross receipts. In addition, the NCCS has verified portions of the files, classified the organizations according to the National Taxonomy of Exempt Entities (NTEE), and added Federal Information Processing Standard (FIPS) county codes for each organization. It is hoped that these files will prove to be useful to nonprofit researchers and practitioners because there have been few sources of reliable secondary data on nonprofit organizations. 1 Consequently, considerable effort and money has been expended on preparing these data files for distribution to the nonprofit community at large. It is important evaluate the extent to which this information can be used. This is the purpose of this Note: This is a revised version of a paper presented at the 1998 ARNOVA conference in Seattle on November 5-7. I wish to thank two anonymous reviewers and Steve R. Smith for helpful comments on an earlier draft. Nonprofit and Voluntary Sector Quarterly, vol. 29, no. 2, June Sage Publications, Inc. 297

2 298 Bielefeld intermetropolitan analysis. The results will enhance our understanding of the factors influencing metropolitan nonprofit sectors and demonstrate the utility of NCCS data files for comparative analysis. This analysis will use the NCCS 1996 Core File to explore the nonprofit sectors of nine metropolitan regions. The sites include Boston, Dallas/Ft. Worth, Indianapolis, Memphis, Minneapolis/St. Paul, Orlando, Pittsburgh, Portland (Oregon), and San Diego. These sites vary in 1994 population from 1.1 million (Memphis) to 5.5 million (Boston). More importantly, they vary on a number of factors that have been held to be relevant to nonprofit organizations. Most research assumes and has found that needs and resources (e.g., income or poverty) will influence the number and distribution of nonprofit providers. For example, Wolch and Geiger (1983) studied the distribution of social welfare and community service nonprofits across municipalities in Los Angeles County and found it to be related to community needs and resources. This research leads to the expectation that resources such as income or wealth will foster larger nonprofit sectors. In addition, given that nonprofits provide service to middle- and upper-income patrons, it would be expected that wealthier communities will have a larger proportion of nonprofits providing services for them (so-called amenity services), for example education or arts (Wolpert, 1993b). Nonprofits, however, also provide welfare-type services for needy clients, and it is expected that the relative presence of these types of providers (e.g., human service nonprofits) will be positively associated with measures of poverty and needs. An additional factor that has been found to influence the presence of nonprofits is community heterogeneity (Weisbrod, 1988). As an example, each of the minority ethnic groups in a diverse community might well desire a specialized version of some service, such as education. Government, with its mandate to serve the desires of a majority of the voters, will not fill this heterogeneous demand for education. The demand will instead be filled by private nonprofit organizations. This leads to the expectation that more ethnically diverse communities will have more nonprofit organizations. Several other factors have been investigated over and above the demographic characteristics mentioned. Of particular interest for this analysis are generosity and political culture. In a series of studies, Julian Wolpert (Wolpert, 1989, 1993a, 1993b) examined levels of generosity in U.S. cities using various measures of public and private generosity. Public generosity was indicated by the level of local and state government support for welfare services, and private generosity included gifts and grants to nonprofits, the proportion of these going to human service providers, and support for a variety of local fundraising initiatives such as the United Way. Wolpert s work clearly established that communities varied widely in terms of generosity and that in general,

3 Metropolitan Nonprofit Sectors 299 higher generosity is positively related to income and negatively related to population and distress levels. The political scientist Daniel Elazar (1984) developed a typology of political culture that has been used extensively. The typology is based on political and social attitudes in the original American colonies and subsequent migration and settlement patterns. His framework consists of three political cultures that measure, among other factors, citizen attitudes about the proper role of government. In a moralistic political culture (originating in the New England colonies), citizens believe that government s proper role is the general well-being of the community and the redistribution of wealth. This can be contrasted to an individualistic political culture (originating in the middle-atlantic colonies), where the belief is that government s proper role is the support and promotion of the market. Finally, in a traditionalistic political culture (originating in the Southern colonies), citizens view the government s proper role as the support of tradition and the status quo. Elazar s (1984) typology has been related to philanthropy and the makeup of nonprofit sectors (Bielefeld & Corbin, 1996; Schneider, 1996). In addition, the typology can be expected to have implications for the types of generosity Wolpert considered. The connection between these two important formulations has not received much attention, and the data to be used in this analysis will allow us to examine this. It can be expected that communities with moralistic political cultures will have larger nonprofit sectors that will contain relatively more human service providers. We would also expect to find a high level of public generosity as well as private generosity for human services. Communities with individualistic political cultures, on the other hand, should have nonprofit sectors more oriented toward individual entrepreneurial activities, containing, for example, relatively more education providers. In terms of generosity, we would expect lower levels of public generosity and high levels of private generosity in terms of general nonprofit support and support for the United Way (based on its ties to corporate activity). Finally, communities with traditionalistic political cultures should have nonprofit sectors providing traditional and conservative services such as health, some kinds of human services, or the arts. Public generosity should be low, and private generosity should focus on traditional nonprofit activity. The expectations derived in the paragraphs above are plausible given theory and previous research. The research has been limited, however, by the lack of data, and the expectations need to be more systematically investigated on a widespread basis. The goal of this analysis is to address this need by examining the degree to which the factors outlined earlier explain intermetropolitan variation in the extent of nonprofit sectors, the mix of services offered in them, and selected characteristics of their service providers.

4 300 Bielefeld DATA, VARIABLES, AND METHODS METROPOLITAN DATA AND VARIABLES The sites included in this analysis are the metropolitan regions of Boston, Dallas/Ft. Worth (DFW), Indianapolis, Memphis, Minneapolis/St. Paul (MSP), Orlando, Pittsburgh, Portland (Oregon), and San Diego. These sites were chosen because they varied on a number of factors, outlined earlier, that have been held to be important for nonprofit functioning. It has been assumed that regions of the United States vary on a host of factors, and previous nationwide studies of nonprofits (see Salamon, 1987) have guarded against regional concentration. This study, likewise, contains sites from a number of regions. Boston and Pittsburgh are selected from the North and East region; Memphis and Orlando are selected from the South; MSP, Indianapolis, and DFW are from the central region; and Portland and San Diego are from the West. Wealth has been assumed to influence the size and composition of nonprofit sectors. Wolpert (1993a) classified cities on a Woods and Poole wealth index for 1990, and these classifications will be used in this analysis. The index is constructed from income from wages, rents, and dividends and excludes income from transfer payments. Population change and poverty in the center city will be used as measures of needs. Bureau of the Census data were used to measure these. Population change was the percentage change in the population of the central city in each site from 1990 to Declines were seen as indicators of population flight from blighted areas. Poverty was measured by the percentage of individuals below the poverty line in Heterogeneity has been used as an indicator of the desire for diverse nonprofit services. It will be measured in this analysis by census data on the number of racial groups comprising 2% of the population. Generosity is taken to indicate resources available for nonprofits. Wolpert (1989, 1993b) constructed a variety of measures of generosity for U.S. cities for This analysis will use several of the measures he computed and reported in his work. The state maximum share of Aid to Families with Dependent Children (AFDC) will be used as a measure of public generosity. This measure was constructed by dividing the maximum state AFDC payment by the federal matching rate. Private generosity will be measured by: (a) the per capita dollar value of gifts and grants to metropolitan nonprofit organizations, 2 (b) the proportion of these gifts and grants that went to nonprofits providing human services, and (c) United Way contributions per employee. Political culture has been used to measure citizen attitudes about the proper role of government and nonprofits. Kincaid (1980) operationalized Elazar s (1984) typology of political culture at the city level, and his classifications will be used in this analysis. Cities were classified as having moralistic,

5 Metropolitan Nonprofit Sectors 301 individualistic, or traditionalistic political cultures. Cities can be classified as pure types or as a mixture of two types (e.g., moralistic-individualistic). If mixed, the first type indicates the dominant political culture. NONPROFIT DATA AND VARIABLES Data on the nonprofit sectors in the study sites were obtained from the NCCS Core File for This file contains IRS Form 990 information from all 501(c)(3) operating organizations with $25,000 or more in annual gross receipts that filed Form 990 returns in In addition, the NCCS has verified portions of the files and added other useful information to them (such as NTEE and FIPS county codes). Further information about these files can be obtained from the NCCS, and extensive discussion and documentation is contained on their Web site ( The 1996 Core File contains information on more than 200,000 organizations. FIPS county codes were used to select the nonprofits in each of the counties contained in the metropolitan areas in the analysis. Variables to be used in the analysis include the total number of nonprofits in each site as well as the number of nonprofits per 10,000 people in each site. These will represent the size of the nonprofit sector in each site. In addition, the analysis will examine the distribution of nonprofits providing various types of services according to the NTEE classification system. The 10 NTEE major categories will be used to examine the distribution of nonprofits types, and the proportion of those providing arts/culture/humanities, education, health, and human services will be examined in detail. A number of financial variables will also be considered. Organizational revenue will be used as another indicator of sector size in this case, whether the sector is composed of relatively large or small organizations. The proportion of revenues coming from dues will be used to indicate the degree to which the sector is participatory and relies on grassroots support. 3 A measure of vulnerability will be used to assess the financial health of each sector and its potential for future financial problems. The ratio of net worth (assets minus liabilities) to expenses will measure the extent to which organizations can continue to provide services in the absence of new income (Pollak & Pettit, 1998). The larger this ratio is, the longer the organization could function. FINDINGS CITY CHARACTERISTICS Table 1 presents the characteristics of the metropolitan areas included in the analysis. The table shows that the metropolitan areas varied widely. Population ranged from 5.5 million (Boston) to 1.1 million (Memphis). These sites

6 302 Bielefeld also represented the extremes on the wealth index (127.9 for Boston and 93.3 for Memphis). In terms of center city characteristics, population change between 1990 and 1994 varied from 4.6% (Boston) to +3.7% (San Diego). Declining center cities included Boston, MSP, and Pittsburgh; stable sites included Memphis and DFW; and growing sites included Indianapolis, Portland, San Diego, and Orlando. In addition, the percentage below the poverty line varied from 12.5% (Indianapolis) to 23% (Memphis). Finally, the number of racial groups comprising 2% of the population (heterogeneity) ranged from two (Memphis) to five (MSP). In terms of political culture, Portland and MSP are classified as moralistic, Pittsburgh and Indianapolis are classified as individualistic, and Memphis and Orlando are classified as traditionalistic. Three sites have mixed political cultures: Boston is individualistic-moralistic, DFW is traditionalisticindividualistic, and San Diego is moralistic-traditionalistic. The generosity measures also varied widely between cities. Per capita gifts and grants to nonprofits varied from $1.15 (Boston) to $0.21 (San Diego). United Way contributions per employee varied from $51 (MSP) to $18 (Orlando). The percentage of gifts and grants to nonprofits that went to human service providers varied from.27 (San Diego) to.08 (Boston). Finally, the state maximum share of AFDC varied from 14.4% (San Diego) to 5.2% (Memphis). Much of the analysis to follow will use city rankings on variables and rank order correlations between variables. This is necessary given the limited number of cases. Another consequence of the limited sample size and the means by which sites were chosen (selectively rather than randomly) is that the results will be presented as descriptive and exploratory. That is, inferential techniques such as measures of statistical significance will not be used. The findings, therefore, cannot be taken as rigorous statistical tests of the expectations outlined earlier. Nevertheless, they can provide evidence of patterns in this set of sites that are either consistent or inconsistent with the expectations. Although not definitive, this is a valuable first step. Table 2 shows the political culture of sites and site rankings on the generosity measures. For the rankings, lower numbers indicate higher generosity. The information in Table 2 was considered in two, more detailed ways to shed further light on the relations between political culture and generosity. A number of expectations were presented earlier regarding the relations between political culture and generosity. It was held that generosity should vary systematically with political culture, and the data confirm this. 4 Table 3 shows the average rankings on generosity for the sites with pure (unmixed) political cultures. We would expect that sites with moralistic cultures would be most favorable to both public and private assistance to the needy (welfare services). This is shown in the data in that these sites had the highest percentage of nonprofit gifts/grants going to human service providers and the highest state AFDC share. Individualistic sites, on the other hand, should be most oriented toward individual action. The table shows that these sites had the

7 Table 1. Metropolitan Characteristics of Study Sites Dallas/ Minneapolis/ Portland San Boston Ft. Worth Indianapolis Memphis St. Paul Orlando Pittsburgh (Oregon) Diego 1994 metro population (millions) a Metropolitan wealth index (1990) b Center city characteristics (1994) a [D] c [M] c Population change ( percentage change) Poverty (percentage below poverty line) Heterogeneity (number of 2% groups) Political culture d IM TI I T M T I M MT Generosity measures (1989) b Gifts/grants to nonprofits ($ per cap) United Way $ per employee Percentage gifts/grants to human services State max share of Aid to Families with Dependent Children (percentage) a. Metro and center city demographics (census data). b. Generosity and wealth (Wolpert, 1989, 1993b). c. Center city data for Dallas and Minneapolis only. d. Political culture (Kincaid, 1980). M = moralistic, I = individualistic, and T = traditionalistic. 303

8 304 Bielefeld Table 2. Metropolitan Political Culture and Rank a on Generosity Measures United Percentage State Max Aids Gifts/ Way Gifts/Grants to Families Political Grants to $ per to Human With Dependent Culture Nonprofits Employee Services Children Share Minneapolis/St. Paul M b Portland M Indianapolis I Pittsburgh I Memphis T Orlando T San Diego MT Boston IM Dallas/Ft. Worth TI a. Ranks were assigned such that 1 = most generous and 9 = least generous. b. Political culture (Kincaid, 1980). M = moralistic, I = individualistic, and T = traditionalistic. highest level of private giving to nonprofits in general and the highest level of giving to the United Way. Traditionalistic sites should be conservative in outlook, favoring limited and traditional nonprofit activities. The table shows that these sites were the least generous in terms of giving to nonprofits in general, United Way contributions, and state AFDC share. They were, however, fairly generous in terms of giving to human services, a traditional role for the nonprofit sector. Table 4 looks at the interrelations among the generosity measures and their relationship to wealth. Wealth is related to higher giving to nonprofits in general (consistent with the findings of previous research) and higher United Way contributions but lower generosity to human services providers. This is consistent with the observation that much of the support of middle- and upperincome donors goes to causes favored and used by them as opposed to welfare-type services. This is more clearly brought out by the large negative correlation (.84) between gifts and grants in general and the percentage going to human service providers. The negative correlation (.23) between United Way support and the percentage of gifts going to human services could be interpreted in several ways. It could represent the fact that in communities that strongly support the United Way, this support is seen as adequate to deal with community needs. Alternatively, the United Way could be viewed as a fairly conservative, middle-class agency, and in communities that favor it, there is little desire to give to welfare services. Finally, the data show that private support for the needy (gifts/grants to human services) is positively associated with public support (state AFDC share). NONPROFIT SECTORS Table 5 shows the numbers of nonprofits in the metropolitan areas included in the analysis and the percentages in the 10 NTEE major categories. It should

9 Metropolitan Nonprofit Sectors 305 Table 3. Political Culture a and Average Ranks b on Generosity Percentage State Max Aids to Gifts/Grants to United Way Gifts/Grants to Families With Nonprofits $ per Employee Human Services Dependent Children Share Moralistic Individualistic Traditionalistic a. The three sites with combined political cultures were excluded from this table. b. Ranks were assigned such that 1 = most generous and 9 = least generous. Table 4. Spearman Rank Order Correlations: Wealth and Measures of Generosity Percentage State Max Aids to Families Gifts/Grants to United Way Gifts/Grants to With Dependent Nonprofits $ per Employee Human Services Children (AFDC) Share Metropolitan wealth Gifts/grants to nonprofits United Way $ per employee Percentage gifts/ grants to human services.28 State max AFDC share be remembered that these are the numbers of 501(c)(3) organizations filing IRS 990 forms in 1995 and not the total number of nonprofits in these sites. This is discussed further in the final section of the article. The table shows that the numbers of nonprofits varied between the sites. The nonprofit sector in Boston, with about 6,000 nonprofit filers, is significantly larger than the nonprofit sectors in the other sites. This is followed by medium-sized sectors including DFW and MSP, which have about 2,700 nonprofits each; and Pittsburgh, Portland, and San Diego, which have between 1,700 and 2,000 nonprofits each. The smallest sites are Indianapolis, which has about 1,300 organizations, and Memphis and Orlando, which only have about 700 organizations each. In terms of NTEE major categories, the percentages across sites are fairly consistent for arts (7.6% to 11.0%), environment and animal (1.8% to 3.6%), international (.3% to 1.7%), mutual and member benefit (.1% to 1.1%), and unknown (3.7% to 7.0%). The percentages vary widely, however, for education (12.4% to 20.6%), health (12.7% to 21.7%), human services (27.2% to 39.3%), public benefit (7.5% to 13.2%), and religion (1.9% to 11.8%).

10 306 Table 5. Number of Nonprofits and Percentages in National Taxonomy of Exempt Entities (NTEE) Major Categories Dallas/ Minneapolis/ Portland San Category Description (major group a ) Boston Ft. Worth Indianapolis Memphis St. Paul Orlando Pittsburgh (Oregon) Diego Total number of nonprofits 6,079 2,661 1, , ,962 1,741 1,669 I Arts, Culture, Humanities (A) II Education (B) III Environment and Animals (C, D) IV Health (E, F, G, H) V Human Services (I, J, K, L, M, N, O, P) VI International, Foreign Affairs (Q) VII Public, Societal Benefit (R, S, T, U, V, W) VIII Religion Related (X) IX Mutual/Membership Benefit (Y) X Unknown (Z) Total percentage a. NTEE major groups are subdivisions within NTEE major categories.

11 Metropolitan Nonprofit Sectors 307 INTERMETROPOLITAN ANALYSIS Table 6 shows the nonprofit variables that will be related to city characteristics. They include: sector size expressed as the number of nonprofits per 10,000 people; the percentages of sector nonprofits in the NTEE major categories of arts, education, health, and human services; and a number of financial variables including average revenue, average proportion of income from dues, and vulnerability measured by net worth divided by expenditures. Table 7 shows the relations between political culture and the nonprofit variables. Political culture theory would lead one to expect that communities with moralistic political cultures would have the largest and most supported nonprofit sectors, with a relatively large proportion of organizations providing human service, welfare-type services. The data are consistent with these expectations. The table shows that the moralistic sites ranked highest on sector size and the percentage of income from dues (showing grassroots participation and support) and lowest on financial vulnerability (having the financially healthiest nonprofits). In addition, they ranked highest on the proportion of the sector providing human services. Communities with traditionalistic political cultures could be expected to have the smallest and weakest nonprofit sectors with a relatively low diversity of organization types and relatively few human service providers. Table 7 shows that the traditionalistic sites ranked lowest on nonprofit sector size and the proportion of income from dues. These sites also ranked highest on financial vulnerability. In addition, they ranked the lowest on the proportion of human service providers as well as the lowest on the proportions of arts and education providers. They did, however, rank highest on the proportion of health providers, consistent with a focus on traditional nonprofit services. Communities with individualistic political cultures could be expected to fall between these two extremes, with an additional focus on activities promoting individual action and enterprise. The data, again, bear out these expectations. It shows that the individualistic sites ranked between the other types on nonprofit sector size, financial vulnerability, proportion of income from dues, and proportions of human service, health, and arts providers. They ranked highest, however, on the proportion of education providers, consistent with an individualistic focus. Table 8 shows the relations between ethnic heterogeneity and nonprofit sector size. Our expectation was that increased heterogeneity will lead to more nonprofits. This expectation is partially supported in the data. The table shows that size of the sector (measured by the number of nonprofits per 10,000 people) is the second largest in the site with the largest number of ethnic groups comprising 2% or more of the population (MPS) and smallest in the site with the fewest number of ethnic groups (Memphis). The sector sizes in the sites with three and four ethnic groups, however, appear to overlap considerably. These findings may be due in large part to the way heterogeneity was measured for this analysis. The rather small number of ethnic groups

12 308 Table 6. Nonprofit Variables Used in Intermetropolitan Analysis Dallas/ Minneapolis/ Portland San Boston Ft. Worth Indianapolis Memphis St. Paul Orlando Pittsburgh (Oregon) Diego Number of nonprofits 6,079 2,661 1, , ,962 1,741 1,669 Nonprofits per 10,000 population Percentage of nonprofits classified as: Arts, Culture (NTEE a Category I) Education (NTEE Category II) Health (NTEE Category IV) Human Services (NTEE Category V) Financial measures site averages Total revenue (in $1,000s) 4,052 2,559 3,023 4,353 2,801 3,447 5,907 2,226 2,437 Dues/total revenue Net worth/expenditures a. NTEE = National Taxonomy of Exempt Entities

13 Metropolitan Nonprofit Sectors 309 Table 7. Political Culture and Average Rank a on Sector Size, Vulnerability, Percentage Dues Income, and Types of Nonprofits Percentage of Sector Nonprofits Classified as Dues as Nonprofits Percentage Human per 10,000 Vulnerability b of Income Services Health Education Arts Moralistic Individualistic Traditionalistic Note: The three sites with combined political cultures were excluded from this table. a. Smaller numbers signify higher rankings. b. Vulnerability measured by net worth divided by expenditures. considered by the census in computing these proportions may limit the utility of this measure for the analysis being conducted. It might be expected that heterogeneity would be better examined with more fine-grained measures of ethnic categories and their incumbents, for instance, using the more extensive number of categories found in other census publications. This would allow the computation of more sophisticated measures of concentration such as herfindahl indexes. Finally, it should also be noted that the measure here refers to ethnic heterogeneity. Other researchers have suggested using the heterogeneity of other population characteristics, particularly income (see Bielefeld, Murdoch, & Waddell, 1997, for a discussion of these measures). Table 9 shows the relationship between metropolitan wealth, poverty, and generosity and the nonprofit variables. As expected, metropolitan wealth is associated with larger nonprofit sectors, sectors with more financially secure organizations in them (being less financially vulnerable), relatively more education and arts providers, and relatively fewer health providers. There does not seem to be a relation between wealth and the number of human service providers. Consistent with Wolpert (1989, 1993a, 1993b), the level of gifts and grants to nonprofits is associated with larger sectors and more financially secure organizations. It is also associated with relatively more arts and health providers and relatively fewer human service providers. The level of United Way support is also associated with larger nonprofit sectors and (slightly) more financially secure organizations. In addition, United Way support is positively associated with relatively more human service and health providers and negatively associated with relatively fewer education and arts providers. The percentage of gifts and grants going to human service providers is, not surprisingly, associated with the presence of relatively more of these types of organizations. In addition, however, this percentage is associated with relatively smaller and less financially secure sectors and sectors with fewer health and education providers. This is consistent with the findings in Table 4 that

14 310 Bielefeld Table 8. Heterogeneity and Sector Size Number of Number of Site 2% Ethnic Groups Nonprofits per 10,000 Minneapolis Boston Portland San Diego Dallas Orlando Indianapolis Pittsburgh Memphis Table 9. Spearman Rank Order Correlations: Sector Size, Vulnerability, and Types of Nonprofits by Wealth and Generosity Percentage of Sector Nonprofits Classified as Nonprofits Human per 10,000 Vulnerability a Services Health Education Arts Metropolitan wealth Gifts/grants to nonprofits United Way $ per employee Percentage gifts/grants to human services State Aid to Families with Dependent Children Share a. Vulnerability measured by net worth divided by expenditures. showed the human service gift/grant percentage was higher in less wealthy metropolitan areas where nonprofits will also be less financially secure and where there is less money available for amenity services such as education or health. Finally, public generosity (state maximum AFDC share) in this data set is associated with larger and more secure nonprofit sectors, suggestive of the notion of a partnership between nonprofits and government in service provision (Salamon, 1987). This is further supported by the positive association for human services (an area of government responsibility). Interestingly, the table also shows a strong positive association for the relative number of arts organizations, possibly attesting to the influence of public funding for the arts. Table 10 looks more closely at the relationships between poverty and metropolitan wealth, generosity, and the number of human service providers. It

15 Metropolitan Nonprofit Sectors 311 Table 10. Spearman Rank Order Correlations: Poverty by Wealth, Human Service Nonprofits, and Generosity Central City Poverty Metropolitan Poverty a Metropolitan wealth Percentage human service providers Percentage gifts/grants to human services United Way $ per employee State Aid to Families with Dependent Children share a. Metropolitan Poverty results need to be interpreted cautiously because one site had missing data and the remaining eight sites had relatively little variation (three were tied on one rank and two were tied on another). shows that sites with relatively high central city poverty are also sites with relatively low metropolitan wealth, relatively low public or private support for human service providers and welfare, and relatively few human service providers. This seems to indicate a gap in needs and the institutions designed to alleviate them. The sites with relatively high poverty, however, do show relatively high support for the United Way. Wolpert s observation that generosity is negatively related to distress is, therefore, only partially supported in this data. The expectation holds for public generosity and private gifts and grants to human service providers. It does not hold for generosity to the United Way, consistent with its role of providing services to the needy as well as amenity services. It must be remembered, however, that poverty was measured in the center city and generosity was measured for the entire metropolitan area. Whereas poverty was also measured for the entire metropolitan areas, it showed relatively little variation at this level, making conclusions based on the use of rank order correlations very tenuous. SUMMARY AND CONCLUSION Taken together, the results of this analysis are encouraging. The expectations generated from previous research and theory were generally supported by the findings. The findings give us further insights into the relations between metropolitan characteristics and nonprofit functioning as well as confidence that the NCCS Core File data can be useful for the analysis of nonprofit sectors. The findings showed a number of complex and intriguing relationships between metropolitan political culture, generosity, wealth, poverty, and nonprofit sectors. To summarize the major findings, moralistic sites were found to have the largest and most financially secure nonprofit sectors. In addition, they were characterized by the most generosity to and had the highest proportion of human services providers. Traditionalistic sites, on the other hand, had

16 312 Bielefeld the smallest and least financially secure sectors and provided the least generosity to and had the smallest proportion of human service providers. A third pattern was found for individualistic sites, which were the most generous overall and whose nonprofit sector size, financial security, and human service components were between those of the other two types of political culture. Wealthier sites had larger, more financially secure sectors, were more generous overall, had more education and arts organizations, and provided lower support for services for the needy. Sites with high poverty, on the other hand, were less wealthy, provided low support for and had few human service providers, but had relatively high United Way support. In terms of heterogeneity, the sites with highest and lowest measures had among the largest and smallest sectors, respectively, as predicted. The sector sizes of sites with intermediate levels of heterogeneity, however, showed an inconsistent pattern. The analysis was subject to a number of limitations. Chief among them is the statistical techniques employed. The limited number of sites made regression analysis at the intermetropolitan level impossible and led to the use of rank order correlations. This is a relatively weak statistical measure and limits the analysis to bivariate relations. It was not possible to assess the effects of the independent variables in isolation from others. In addition, the nonrandom site selection limits the use of statistical results to description only. As a consequence, the results should be viewed with caution and considered suggestive only and not be taken as rigorous statistical tests of the hypotheses proposed. This limitation could be alleviated with the inclusion of more, randomly chosen sites in future analyses. In addition, in some cases, there was a mismatch between center city and metropolitan-wide measures. This could be rectified by obtaining comparable data for both levels and more clearly identifying where providers are located. There are likely to be different nonprofit patterns and interactions in center cities as opposed to the suburbs, and this analysis did not address this. The nonprofits should be assigned to more detailed geographic units, such as municipalities, zip codes, or census tracts, and the relationships examined within these alternative units (Bielefeld et al., 1997). It would be interesting and useful to examine the relative utility of intrametropolitan versus intermetropolitan analysis. The variables used in this article as well as others can be measured within cities as well as between them. Their use at the intrametropolitan level may, in fact, provide additional or more detailed explanations of nonprofit sector dynamics. In addition, these more fine-grained types of studies are likely to be useful to practitioners, providing them with valuable information about their local environments, which could assist them in planning. Finally, the implications of excluding the nonprofits that did not file Form 990s are not clear. This is an important issue and will, moreover, be quite difficult to address given that little, if any, reliable secondary data exists on these organizations. Most of the organizations that are excluded from the NCCS Core File are small, making less than $25,000 a year in gross income, but there

17 Metropolitan Nonprofit Sectors 313 are a large number of them in the Business Master File (more than half of the organizations contained in the Master File). Although it is likely that the larger organizations included in the NCCS Core File generate most of the financial activity of the sector as a whole, a number of questions remain. One question would be the degree to which nonfillers are concentrated in any particular service areas, perhaps biasing the NCCS file. In addition, their joint impact on the amount of nonprofit and philanthropic activity over and above finances in a given region should be assessed relative to the impact of the organizations in the NCCS data set. In addition, religiously based organizations that are not required to file with the IRS and who choose not to do so are not included in the NCCS data. The data may, consequently, exclude a number of religiously based organizations that provide human services. To take these complications into account, a more complete community analysis would be based on samples drawn from nonprofit lists that are gathered from several sources. Kirsten Gronbjerg (1989, 1998) investigated this issue extensively. These limitations are not overwhelming and should be seen as challenges for future researchers. It is hoped that this analysis has demonstrated the utility of the NCCS data and provided suggestions for its further use in helping researchers move to more widespread and sophisticated analysis. Notes 1. The IRS provides another source of nonprofit data through its Statistics of Income (SOI) sample. This is detailed data from all large nonprofits ($10 million plus in assets) and a sample of smaller organizations. One problem with the use of this data is that a count of the numbers of organizations in a given geographical region is not possible. Whereas the National Center for Charitable Statistics (NCCS) data do not provide a complete census of organizations, it does provide a census of 990 filers. 2. When using giving to metropolitan nonprofit organizations as a measure of metropolitan generosity, it is important to remember that whereas most giving to nonprofits comes from the local level (Bielefeld, Murdoch, & Waddell, 1997; Wolpert, 1993a), not all of it does so. Those seeking to do a more detailed analysis of particular communities should assess whether those communities contain large organizations that might get a significant amount of funding from extralocal sources. If such organizations are found, the amounts given to them should be adjusted to reflect local giving. 3. The conceptualization and measurement of dues income involves several complications. For this analysis, the amount recorded as dues on Form 990 was used. This, however, reflects only amounts given to the organization for which tangible benefits are received. In other words, if someone gave $20 and received a premium valued at $5, only the $5 would be counted as dues. If nothing tangible was received in return, no dues would be recorded. The number of contributors an organization had would be a better measure of grassroots support, but this number is not contained on the Form Another factor that could account for variations in gifts and grants to nonprofits is differences in religiously based giving. Denominations may differentially prefer using congregations or independent nonprofits to provide human services. More detailed comparative analyses should examine the denominational profiles of the included communities and, if variations are found, the types of service provision favored by those denominations that dominate.

18 314 Bielefeld References Bielefeld, W., & Corbin, J. (1996). The institutionalization of nonprofit human service delivery: The role of political culture. Administration and Society, 28, Bielefeld, W., Murdoch, J., & Waddell, P. (1997). The influence of demographics and distance on nonprofit location. Nonprofit and Voluntary Sector Quarterly, 26, Elazar, D. (1984). American federalism: A view from the states. New York: Thomas Y. Crowell. Gronbjerg, K. (1989). Developing a universe of nonprofit organizations: Methodological considerations. Nonprofit and Voluntary Sector Quarterly, 18, Gronbjerg, K. (1998). Mapping small religious nonprofit organizations: An Illinois profile. Nonprofit and Voluntary Sector Quarterly, 27, Kincaid, J. (1980). Political culture and the urban quality of life. Publius, 10, Pollak, T., & Pettit, K. (1998). The finances of operating public charities. Unpublished paper, Center on Nonprofits and Philanthropy, The Urban Institute, Washington, D.C. Salamon, L. (1987). Partners in public service: The scope and theory of government-nonprofit relations. In W. W. Powell (Ed.), The nonprofit sector: A research handbook (pp ). New Haven, CT: Yale University Press. Schneider, J. C. (1996). Philanthropic styles in the United States: Toward a theory of regional differences. Nonprofit and Voluntary Sector Quarterly, 25, Weisbrod, B. A. (1988). The nonprofit economy. Cambridge, MA: Harvard University Press. Wolch, J., & Geiger, R. K. (1983). The urban distribution of voluntary resources: An exploratory analysis. Environment and Planning A, 15, Wolpert, J. (1989). Key indicators of generosity in communities. In V. A. Hodgkinson, R. W. Lyman, & Associates (Eds.), The future of the nonprofit sector (pp ). San Francisco: Jossey-Bass. Wolpert, J. (1993a). Decentralization and equity in public and nonprofit sectors. Nonprofit and Voluntary Sector Quarterly, 22, Wolpert, J. (1993b). Patterns of generosity in America: Who s holding the safety net? New York: The Twentieth Century Fund Press. Wolfgang Bielefeld is an associate professor in the School of Public and Environmental Affairs at Indiana University. His research interests include the relations between nonprofit organizations and their environments, the dynamics of nonprofit sectors, the development of human service delivery systems, and the social impacts of government human service policies and spending patterns.