Nonprofit Financial Response to Immigration

Abstract

Nonprofit organizations are important actors in local communities, providing services to vulnerable populations and acting as stewards for charitable contributions from other members of the population. An important question is whether nonprofits spend or receive additional revenues in response to changes in the populations they serve. Because immigrant populations both receive and contribute to nonprofit resources, changes in immigrant numbers should be reflected in changing financial behavior of local nonprofits. Using data from the National Center for Charitable Statistics and the American Community Survey, we study whether nonprofit financial transactions change in response to changes in the local immigration population, the nature of the change, and the degree to which these changes vary by nonprofit type. Findings suggest that nonprofit financial behavior changes with growth and decline in immigrant populations underscoring the importance of nonprofits as service providers and contribute to an understanding of how organizations respond to external forces.

Introduction

Nonprofit organizations are vital actors in many local communities, filling important gaps in the provision of services left by government and private firms (Garkisch et al., 2017; Weisbrod, 1988). In the United States, they provide the majority of services in the human service sector (Lipsky & Smith, 1989; Smith & Grønbjerg, 2006), including critical services and resources for immigrants (Cordero-Guzmán et al., 2008; Garkisch et al., 2017; Hung, 2007; Mayblin & James, 2019). Nonprofits also receive large shares of U.S. charitable contributions for education, health services, arts, cultural activities, and other causes (Havens et al., 2006). Research on nonprofit responses to environmental change has largely focused on competition for limited government financial assistance and nonprofit response to dynamic regulatory environments. (Mason & Fiocco, 2017; Smith & Phillips, 2016). Despite the important role of nonprofits in local communities and in service to vulnerable populations, we know little about how they respond financially to changes in those populations. In particular, it is unclear whether nonprofits receive additional revenues, spend more, or otherwise change their financial behavior when their target populations change. It is also unclear whether the response varies across types of nonprofits.

Growth and decline in immigrant populations likely represent an important impetus for changes to the resources of local nonprofits. Over the past four decades, the U.S. foreign-born population has grown dramatically, and the proportion of the population composed of immigrants reached near-record highs by 2017 (Radford, 2019). Latino, Asian, and multiracial groups are expected to double by 2060 (Alba, 2020; Frey, 2018). Immigrant populations depend on social services such as education (Dondero & Muller, 2012; Kandel & Parrado, 2006), housing aid (Mayblin & James, 2019), and job assistance (Guo, 2014; Handy & Greenspan, 2009); and local nonprofit budgets may expand (e.g., through government grants or donations) to meet growing immigrant needs (Bloemraad, 2005; Cordero-Guzmán, 2005). Immigrants are also important donors to nonprofits, many of which assist newer members of their ethnic group as they integrate into the host country (Havens et al., 2006; Weng & Lee, 2016). Because immigrants rely on and support nonprofits, nonprofit response to changing immigrant populations are key indicators of how well these growing populations will transition to the United States in the short term and how effectively they will integrate over time.

In this article, we study nonprofit responsiveness to changing immigration patterns in local communities. The article addresses how and when nonprofits adapt to changes in the demand for their services and also explores the factors associated with the availability of services for immigrants. We have three objectives. First, we aim to document whether nonprofit financial transactions change in response to changes in the size of the local immigrant population. Second, we explore how nonprofits change financially in response to shifts in the immigrant population. In particular, we study whether nonprofits receive additional revenues, from different sources, and whether they spend additional funds when their target populations change. Third, we explore how these processes vary by the organization’s sector (i.e., whether some nonprofits are more nimble than others) and by the nativity of the immigrant population (i.e., whether the response differs for Asian vs. Latino vs. other immigrants). This final question provides insight into the broader issue of how and when organizations respond to demand for their services by disentangling which organizations respond and by highlighting whether the response reflects organizational traits, immigrant traits, or both. We explore these issues using a unique compilation of data from the Urban Institute’s National Center for Charitable Statistics (NCCS) and the American Community Survey (ACS), enabling us to provide unique estimates of nonprofit financial behavior that include significant detail about conditions in each organization’s local area. We find that increased immigration at the county level is associated with increased nonprofit revenues, and that this result holds across revenue types and nonprofit sector. Furthermore, increases in immigration from both Latin America and Asia are associated with higher nonprofit expenditure and receipts. Our findings contribute to research on organizational change, the readiness of nonprofits to respond to demand for their services, and the ability of local resources to meet the needs of vulnerable populations.

Resource Dependence and Nonprofit Organizations

The resource dependence literature provides a useful starting point for understanding how nonprofit organizations respond to changes in the local immigrant population. It is well established that managers draw on their organization’s unique knowledge and capabilities (Dutton & Duncan 1987; Tichy 1983) while avoiding dependence on outside entities and resources (Mizruchi & Stearns 1994). Resource dependence thus suggests that managers avoid relying on critical external resources to preserve the organization’s autonomy and to increase chances of its survival, particularly in uncertain environments (Mintz & Schwartz 1985; Pfeffer & Salancik 2003).

Three ideas from the resource dependence literature are particularly relevant to understanding the behavior of nonprofits in areas with significant growth in foreign-born populations. First, organizations seek organizational autonomy, particularly from public resources. With shifts in the political and regulatory environment in the United States over time, nonprofits increasingly provide social services and frequently receive government funding and contracts to offer these services, which would otherwise be provided by government agencies (Froelich, 1999; Gazley & Brudney, 2007; Lu, 2013, 2015; Slyke, 2003). Given their close relationship with government agencies, nonprofits often face acute tension between obtaining public funding to maintain operations and preserving their own autonomy (Froelich, 1999; Jung & Moon, 2007; Pfeffer & Salancik, 2003; Verschuere & De Corte, 2014). Reliance on government grants and contracts can reduce nonprofit flexibility in serving immigrants or specific clientele because governments may compel nonprofits to deliver services to a broader group (Verschuere & De Corte, 2014). Thus, with autonomy and flexibility in mind, nonprofits seek to diversify their revenue sources (Froelich, 1999; Hodge & Piccolo, 2005; Moulton & Eckerd, 2012).

Second, organizations seek to decrease uncertainty and eliminate volatility. Diversification of revenue sources, particularly away from government funds and toward individual and corporate donations, can induce large year-on-year changes in overall income, subjecting nonprofits to unpredictability in budgeting and operations (Froelich, 1999; Grønbjerg, 1993). To reduce uncertainty, nonprofits may attempt to please government donors, which could provide them with a steady, predictable revenue flow and increased legitimacy (Froelich, 1999). In doing so, however, organizations can incur large costs. For example, they may alter their bureaucratic structure to maintain complex systems of financial compliance, reporting, and accountability in line with public interest in maintaining transparency and resource stewardship (Jung & Moon, 2007; Verschuere & De Corte, 2014; Verbruggen et al., 2011; Webb & Waymire, 2016).

Third, organizations seek to align the source of their resources with their core values (Moulton & Eckerd, 2012). Whereas for-profit organizations are more likely to respond to market demand, nonprofits go through an iterative, dynamic process in which current funding and potential future funding interact with and shape organizational goals and values. Organizations that successfully implement funded work may be more likely to obtain future resources from the same source, which could lead to additional adjustments to the organization’s mission and values (Jung & Moon, 2007; Pfeffer & Salancik, 2003). For example, nonprofits that receive public funds are more likely to report political engagement, whereas those that rely on individual donations are more likely to report social capital building (Moulton & Eckerd, 2012). Consistent with this, organizations are likely to react to and reflect their environments in a bid to increase legitimacy (Hillman et al., 2000). In line with goals to increase autonomy, nonprofits with strong community ties (having local employees, local clients, and broad community influence) can vary their revenue sources and reduce dependence on their most prominent funder (Provan, 1980; Provan et al., 1980).

Nonprofit Revenues: Sources and Challenges

Nonprofits typically derive their revenues from two sources: earned income and contributions. Earned income includes fees for goods and services from private sources, government sources (including contracts), and membership dues. Earned income tends to be the largest source of revenue for nonprofits, accounting for about 75% of total receipts for the average organization (Roeger et al., 2012). Contributions—including private donations, government grants, and other contributions—are certainly important, but these revenues account for only about 20% of total receipts for the typical nonprofit (Roeger et al., 2012).

Revenue volatility results from year-on-year inconsistencies in private donations and commercial activity. This produces instability in planning and service delivery and difficulty in budgeting for future years (Froelich, 1999). Mission drift—a shift from an organization’s initial focus as a result of their revenue sources—threatens organizational autonomy, in part, by forcing nonprofits to align their goals with their funding source, which may not reflect their core values (Froelich, 1999; Jones, 2007). Although less susceptible to volatility, government funding can engender goal displacement of this sort.

Nonprofit Response to Changing Foreign-Born Populations

Resource dependence theory suggests several predictions regarding the association between changes in foreign-born populations and nonprofit finances. First, areas with a more entrenched immigrant community are likely to have more established immigrant-serving organizations that have relationships with the local governments that may fund them (de Graauw et al., 2013; Hung, 2007). However, because nonprofits seek autonomy from public funds, they are likely to pursue funding from both individual and public funds in an environment where both are rising (Froelich, 1999). Second, existing work on immigrant nongovernmental organizations (NGOs) shows that government funding is key to legitimizing and promoting the growth of the organization over time (Bloemraad, 2005), and that government grants and contracts are the main source of funding for immigrant organizations (Cordero-Guzmán, 2005). Because government funding is more susceptible to causing mission drift (Froelich, 1999), these sorts of community-based organizations may resist being solely dependent on government money and seek to diversify. Although ample evidence suggests that public funding can crowd out other funding sources (Lu, 2015), other research has shown that it attracts external funding (Schatteman & Bingle, 2017), particularly at low levels of public funds (Brooks, 2000), in part by providing legitimacy (Jung & Moon, 2007). Lastly, organizations focused on immigrant well-being address an important local need (Cordero-Guzmán, 2005) and thus seek out resources required. In this case, funding may come from individual donations from members of the community with interests in providing support (Hung, 2007); federal, state, or municipal government funds (Bloemraad, 2005); or fees for service delivery (Brown, 2018). One of the key drawbacks of private donations, however, is revenue volatility. However, this can be overcome by increasing funding levels overall (Carroll & Stater, 2009). For these reasons, we expect the following:

Hypothesis 1A (H1A): Nonprofits will raise additional revenues from all sources in response to changes in immigration in the local community.

An increase in local immigrant populations is likely to result in two simultaneous changes:

1. increased government funding—government funding to nonprofits has shifted from grants (a contribution) to contracts and fees (forms of earned income) in recent decades (Pettijohn et al., 2013; Roeger et al., 2012; Wing et al., 2008);

2. increased donations—although donations are becoming a less important source of nonprofit funding (Roeger et al., 2012; Wing et al., 2008), witnessing a growth in arrivals of immigrants from the same home country is a powerful incentive for prior generations of immigrants to make resources available to address the needs of the new arrivals and to support related causes such as the preservation of national heritage through arts and culture (Hung, 2007; Weng & Lee, 2016).

Hence, the next hypothesis follows:

Hypothesis 1B (H1B): Increasing immigration is associated with increases in both contributions and earned income.

Nonprofit organizations’ financial responses are also likely to differ depending on the nativity of the immigrant group in question. Two immigrant communities are particularly relevant to contemporary U.S. nonprofits: (a) immigrants from Asia, the fastest growing group of recent arrivals; and (b) immigrants from Latin America, the largest immigrant group in the United States. Chinese Americans are the largest Asian subgroup (at 23.2%) and 76% of Chinese Americans are foreign born (Hoeffel et al., 2012; Pew Research Center, 2012; Rumbaut, 2008; Walters & Trevelyan, 2011). Many Chinese immigrants are highly selected on education and socioeconomic status (SES) and thus have the means to make private donations to organizations they support (Borelli & Keister, 2015; Keister et al., 2016; Pew Research Center, 2012). Latin American immigrants, however, are likely to have lower SES, less able to make donations (Pew Research Center, 2014), and more likely to form and contribute to service agencies (e.g., those that provide English classes or health services, which are typically less donation based; Hung, 2007). Thus, we expect the following:

Hypothesis 2 (H2): Contributions to nonprofits will increase in response to immigration from all countries, but increases will be largest in response to immigration from Asian countries.

Because nonprofits vary considerably in their purpose, structure, and financial well-being, their response to changes in the local foreign-born population likely varies by sector. Research on resource dependence suggests that to ensure their financial success and reduce uncertainty, nonprofits are likely to seek funding that is attainable and available, and thus continue to seek funding from the same types of sources they have successfully secured in the past (Carroll & Stater, 2009; Froelich, 1999).

In the arts sector, 57% of nonprofit revenue comes from contributions, compared with 35% from earned income (Roeger et al., 2012). Indeed, arts, culture, and humanities nonprofits are unique in their reliance mainly on contributions, 80% of which are private donations (Roeger et al., 2012). We therefore expect an influx of immigrants to provide increased revenue via contributions to arts but not as much to service-based organizations in the education, health, and human service sectors because of historical reliance on contributions in the arts sector.

Nonprofits in education, health, and human services are service based, providing tangible services such as English classes, clinic visits, or youth programs. The need for such services is likely to expand with increases in immigrant populations. Nonprofits in these sectors rely more heavily on earned income from fees for delivery and government contracts for local services (Roeger et al., 2012). Furthermore, these nonprofit sectors (especially education and human services) have historically been developed mainly by Latin American immigrants (Hung, 2007), who experience higher levels of poverty (Pew Research Center, 2014) and are, therefore, less likely to make large financial contributions. Therefore, we expect education, health, and human service sector nonprofits to see an especially large increase in earned income, compared with arts nonprofits.

Hypothesis 3A (H3A): As immigrant populations increase, all nonprofit sectors will receive additional earned income and contributions.

Hypothesis 3B (H3B): The relationship between immigration and contributions received will be strongest for arts nonprofits, whereas the relationship between immigration and earned income will be strongest for service sector nonprofits (education, health, and human service sectors).

Changing immigrant numbers are also likely to encourage changes in nonprofit expenditures. Government funding can force nonprofits to deliver services to constituents who are not members of the changing immigrant group but who have needs that change with a growth in immigration (Verschuere & De Corte, 2014). Furthermore, as nonprofits seek to diversify their revenue, they are likely to incur expenses. For example, attracting more private and corporate donors requires organizing campaigns and fundraisers, and attracting public funds increases monitoring and compliance expenditures to comply with government reporting (Froelich, 1999; Grønbjerg, 1993; Webb & Waymire, 2016; Young, 1998). Therefore, to serve local populations, comply with the wishes of the organization’s main donor, and diversify resources, nonprofits are likely to spend additional resources when the local immigrant population changes significantly. We anticipate that spending will be equal from all sources:

Hypothesis 4 (H4): Nonprofits will spend additional resources from all sources in response to increased immigration in the local community.

Analytic Strategy

Data

We use data from two sources to study these ideas empirically. First, to obtain county-level estimates of immigration, we use the ACS prepared by Integrated Public Use Microdata Series (IPUMS) (Ruggles et al., 2020), which includes the most populous U.S. counties since 2005. We keep only counties with data for all years from 2005 to 2013 with nonprofit financial data, leaving us with 273 counties per year. To construct percentage foreign born per county, we use person-level weights multiplied by counts of individuals and counts of foreign-born individuals to obtain population counts and immigrant counts, respectively. Although only the 273 most populous counties are retained due to the availability of data (home to 56% of the U.S. population), we find that 75% of all foreign born are captured in our sample and 18% of our sample is foreign born whereas this number is only 7% for excluded counties. Individuals in the included counties are also reasonably representative of the U.S. population. Although individuals are slightly poorer on average in the full sample, this difference is only around 6% of a standard deviation of our poverty measure (description below). For employment, we find a difference in means of about a third of 1% on average between our sample and the full sample. Similarly, for median income, the differences hover around US$5,500, only about 7% of a standard deviation for that variable.

Second, we use the Urban Institute’s NCCS Core Fiscal Year Trend Files for public charities to derive our financial measures (Urban Institute, National Center for Charitable Statistics, 2005–2013). Specifically, we use the Core Fiscal Year Trend Files relating to 501(c)(3) public charity organizations, tax-exempt nonprofit organizations that can receive tax-deductible contributions, have restricted lobbying activities, and cannot direct earnings to benefit shareholders or individuals (Internal Revenue Service, 2020). These data are made available as trend files up until the year 2015. Trend files provide a longitudinal source of data, which is essential to the main goal of this article, which studies changes over time. Because organizations may take up to 3 years to file their taxes, 2013 is the most recent year with complete data in the NCCS Core Fiscal Year Trend Files.

The NCCS data draw in part on Internal Revenue Service (IRS) sources that provide financial data for organizations that file IRS Forms 990, 990-EZ, or 990-PF; tax-exempt nonprofits file these documents annually. Organizations with gross receipts below US$25,000 (before 2010) and below US$50,000 (after 2010) are not obliged to file a Form 990. Therefore, we retain only nonprofit-years in which the nonprofit’s reported gross receipts exceeded these limits. We align all financial data to the calendar year, retain only observations included in our county panel, and restrict our sample further to a balanced panel across all years in our analysis: 2005 to 2013. The final sample contains 601,227 nonprofit-years, or 66,803 nonprofits.

Measures

We use four dependent variables from the NCCS Core Fiscal Year Trend Files. Contributions sum all gifts, grants, and similar amounts received, including both government grants and private donations. Earned income is total program service revenues, including service fees, government fees and contracts, and membership dues and assessments. Gross receipts are the sum of contributions, earned income, and other net income. Expenditures are total nonprofit expenses: grants made to other organizations, compensation, fees paid for services, benefits to members, promotion activities, and other expenses (see, for example, Internal Revenue Service, 2005). All financial variables are measured in U.S. dollars and are converted to 2013 real dollars using the Consumer Price Index (U.S. Bureau of Labor Statistics, 2017). Our nonprofit financial measures are highly variable, a result of their distribution’s long right tail. Furthermore, we see negative revenues for around 1% of organizations, indicating a reported loss in a small fraction of charities. To normalize the distribution for our regression models, we shift all financial measures by the lowest negative reported value and take the log base 10.

We include control variables from the NCCS data for the nonprofit sector: arts, culture, and humanities (hereafter referred to simply as arts); education; health; human services; and other. These sector classifications are based on the National Taxonomy of Exempt Entities Core Codes, which classifies nonprofits based on their primary activities. The inclusion of all five sectors allows us to accomplish two goals: (a) Understand the relationship between revenue type (contributions vs. earned income) and sector and (b) understand the relationship between immigration levels and sector, given evidence that sectors respond differently to different types of immigration. Per Hung (2007), Asian Americans tend to support arts and cultural organizations and Latin Americans tend to support social sector organizations in the United States.

We also include a number of independent variables from the ACS. Percentage immigrant is the percentage of the county’s total foreign-born population. Percentage new arrivals is the percentage of the county that is foreign born and arrived within the 5 years before the survey year. Percentages Latin American, Asian, and other are, respectively, the percentage of the county that was born outside the United States in a Latin American, Asian, or other region; these percentages sum to 100% of the foreign born in a county. Change in percentage immigrant is the percentage foreign born in the county in year t minus the percentage foreign born in year t − 3. Change in percentage new arrivals is the percentage foreign born who arrived to the United States in the past 5 years in year t minus that in year t − 3. Change in percentage from Latin America and change in percentage from Asia are, respectively, the percentage Latin American–born and Asian-born in the county in year t minus that in year t − 3.

In order to ensure that our findings are not merely an artifact of an economic change that occurred over time in counties in our sample, we include three additional measures as controls: (a) median household income, calculated by taking the median value (using household weights) of household income in the county from the ACS; (b) proportion employed, estimated by taking the weighted number of individuals employed over the number reported to be in the labor force, at the county level; and (3) poverty measure, a measure calculated by the ACS, which gives family income as a multiple of the poverty threshold. For example, if a family earns 3 times the poverty threshold, all members of that family will receive a “300” for that measure. It is top coded at 501 (5× the poverty threshold) and can never be negative.

Table 1 provides descriptive statistics for the main financial and county-level variables used in the study. Descriptives of financial variables show average expenditures ranging from US$7.62 million in 2013 to US$9.99 million in 2012. Average gross receipts range from US$10.92 to US$14.40 in 2012. Indeed, average gross receipts are high due to the long right tail. To ensure that our sampling choices did not affect the distribution of financial variables, we compare these figures with official values released by the NCCS from 2015, which found 5.3% of organizations in 2015 had more than US$10 million in gross receipts (McKeever, 2018), whereas we find that the figure is 6.4% for 2012, and 5.3% for 2013, the two most recent years in our data set. We deem this comparable and conclude that averages appear high because some of the financial variables are large.

Table 1.

Summary Table of Nonprofit Financial Variables and County-Level Migration Rates Across Years (Standard Deviations in Parentheses).

	Nonprofit-level variables (yearly N = 66,803)
	2008	2009	2010	2011	2012	2013
Gross receipts (millions of dollars)	12.86 (139.85)	13.08 (152.55)	13.68 (198.95)	13.63 (192.73)	14.40 (207.86)	10.92 (154.55)
Contributions (millions of dollars)	2.20 (27.56)	2.15 (20.82)	2.22 (20.6)	2.23 (21.88)	2.56 (65.55)	1.98 (62.13)
Earned income (millions of dollars)	6.39 (66.87)	6.79 (71.93)	6.98 (73.6)	7.18 (77.72)	7.62 (82.72)	5.83 (74.38)
Expenditure (millions of dollars)	8.73 (75.21)	9.13 (79.75)	9.29 (79.54)	9.43 (83.43)	9.99 (89.73)	7.62 (78.22)
	County-level variables (yearly N = 273)
	2008	2009	2010	2011	2012	2013
% immigrant in given year	10.16 (7.94)	10.26 (7.75)	10.67 (8.15)	10.86 (8.08)	10.71 (8.17)	10.87 (8.17)
% recent arrival in given year	1.83 (1.44)	1.83 (1.44)	1.81 (1.44)	1.74 (1.43)	1.61 (1.38)	1.61 (1.34)
% from Latin America in given year	4.65 (5.48)	4.70 (5.41)	4.95 (5.7)	4.99 (5.63)	4.90 (5.61)	4.93 (5.62)
% from Asia in given year	2.67 (2.57)	2.79 (2.55)	2.91 (2.75)	2.97 (2.83)	3.01 (2.84)	3.08 (2.83)
% from other regions in given year	2.84 (1.96)	2.78 (1.9)	2.81 (1.95)	2.89 (1.94)	2.81 (1.95)	2.85 (1.92)
Change in % immigrant over previous 3 years	0.17 (1.31)	0.13 (1.16)	0.46 (1.27)	0.70 (1.32)	0.45 (1.16)	0.20 (1.13)
Change in % recent arrival over previous 3 years	−0.37 (0.89)	−0.22 (0.72)	−0.08 (0.71)	−0.09 (0.74)	−0.22 (0.67)	−0.19 (0.76)
Change in % from Latin America over previous 3 years	0.13 (0.87)	0.03 (0.81)	0.23 (0.92)	0.34 (0.86)	0.21 (0.8)	−0.02 (0.83)
Change in % from Asia over previous 3 years	0.06 (0.55)	0.15 (0.54)	0.24 (0.58)	0.30 (0.68)	0.22 (0.63)	0.18 (0.48)
Change in % from other regions over previous 3 years	−0.03 (0.82)	−0.05 (0.65)	−0.01 (0.66)	0.06 (0.77)	0.02 (0.66)	0.04 (0.72)
Median household income (in 000s of dollars)	72.53 (19.89)	71.92 (20.03)	69.58 (19.36)	70.91 (19.39)	72.71 (20.16)	75.75 (21.53)
Poverty measure (multiple of poverty threshold × 100)	309.13 (42.89)	302.61 (42.74)	295.74 (43.12)	294.13 (43.05)	294.08 (42.92)	295.74 (42.71)
Proportion employed	93.79 (2.06)	90.33 (2.82)	89.54 (3.01)	90.3 (2.9)	90.97 (2.9)	91.88 (2.46)

Note. Standard deviations in parentheses.

Although we capture charities that are slightly larger due to their location in more populous counties, there is relatively little difference between included organizations and the overall means. For example, in 2012, 27.3% of our sample organizations had less than US$100,000 in expenditures, compared with 29.6% in the total sample. Our sample had 6.25% of organizations with expenditures of US$10 million or more, compared with 5.3% in the total.

On average, populations in the sample counties are 10% immigrant; half of these immigrants are Latin American and approximately 30% are Asian. Most groups grew over the period of study, with the exception of Latin American born in 2013 and recent arrivals. The latter indicates that immigrants who arrived to the country at least 5 years before are moving to sample counties and/or that more recent immigrants are moving out. We find relatively stable county-level economic measures. The median household income hovers around US$70,000 a year, families have income of around 3 times the poverty threshold, and the employment rate is around 90% across years.

Method

To examine how changing immigration is related to nonprofit financial indicators, we propose a mixed model with observed time points (years) nested within nonprofits, which are in turn nested within counties. A distinct advantage of this model over ordinary least squares county-level fixed-effects models is the ability to include county-level covariates: in this case, levels and change in percentage of foreign born. In Equation 1, we show a model for one of four dependent variables Y —gross receipts, earned income, contributions, and expenditures—in standard mixed model notation.

$$\text{Log}\left(Y_{ij}\right)={\beta }_0+b_{0j}+c_{0i}+{\beta }_1{\mathit{\text{year}}}_{ij}+b_{1j}{\mathit{\text{year}}}_{ij}+{\beta }_2\Delta \%{\mathit{\text{immigrant}}}_i+{\beta }_3\%{\mathit{\text{immigrant}}}_i+{\beta }_4{\mathit{\text{sector}}}_{ij}+{\beta }_5{\mathit{\text{median income}}}_i+{\beta }_6{\mathit{\text{poverty}}}_i+{\beta }_7{\mathit{\text{employment}}}_i+{ϵ}_{ij},$$ (1)

$$\left(b_{0j},b_{1j}\right)~N\left(0,\left[\begin{array}{cc}{\tau }_{00}^2& {\tau }_{00}{\tau }_{11}\\ \rho {\tau }_{00}{\tau }_{11}& {\tau }_{11}^2\end{array}\right]\right),$$

$$c_{0i}~N\left(0,{\omega }_{00}^2\right),$$

$${ϵ}_{ij}~N\left(0,{\sigma }^2\right),$$

where Y is the financial outcome variable, i indexes the county, and j indexes the nonprofit. Both b_0j and b_1j are drawn from a standard normal distribution and have the covariance matrix seen in the equation. b_0j is a nonprofit-level random intercept and b_1j year_ij is a random slope for nonprofit by year, which allows us to model annual trends within each nonprofit. We assume that the intercept and year-on-year trend within nonprofits are related. We recode year_ij as years since 2008 (e.g., 2005 is −3; 2013 is 5).

Our analytical strategy proceeds in two steps. First, we run the mixed models proposed in Equation 1, with three models for each financial outcome for our independent variables: all immigrants, recent immigrants, and immigrant origins (Asia, Latin America, or other). To test whether there is a relationship between change in immigrant percentage in a county and nonprofit finances (H1A and H1B), we look at the β₂ coefficient and its standard error.

We compare the coefficients of Asian-born immigrants and Latin American–born immigrants to test whether nonprofit contributions will be largest in response to immigration from Asian countries (H2). To do so, we run a full model with controls for change in percentage and percentage for Asian, Latin American, and other, and compare the coefficients as follows. We first define the contrast matrix such that

$${\beta }_{\mathit{\text{Asia}}}-{\beta }_{\mathit{\text{Latin America}}}=0.$$

We then use the multcomp package in R (Hothorn, 2020) to conduct a general linear hypothesis test with the null hypothesis that the coefficients are the same.

Second, we run a series of separate models based on Equation 1 for each sector (arts, health, human services, and education). To test whether the relationship between financial variables and immigration differs by sector (H3A and H3B), we run aggregated models interacting sector with all other fixed-effect controls: Δ%immigrant_i, %immigrant_i, and year_ij as well as economic controls. Then, using the general linear hypothesis test, we perform pairwise comparisons of the coefficients on the interaction of Δ%immigrant_i and each sector. For example, we make the following comparisons for the arts sector:

$${\beta }_{\mathit{\text{Arts}}\times \Delta \%{\mathit{\text{immigrant}}}_i}-{\beta }_{\mathit{\text{Education}}\times \Delta \%{\mathit{\text{immigrant}}}_i}=0,$$

$${\beta }_{\mathit{\text{Arts}}\times \Delta \%{\mathit{\text{immigrant}}}_i}-{\beta }_{\mathit{\text{Health}}\times \Delta \%{\mathit{\text{immigrant}}}_i}=0,$$

$${\beta }_{\mathit{\text{Arts}}\times \Delta \%{\mathit{\text{immigrant}}}_i}-{\beta }_{\mathit{\text{Human Services}}\times \Delta \%{\mathit{\text{immigrant}}}_i}=0.$$

We test the hypothesis that the coefficients are the same—that is, that outcomes do not differ by sector.

Results

Revenues Increase When Immigration Increases

Our results provide strong and consistent evidence for H1A, which suggests that an increase in the local foreign-born population is associated with an increase in all forms of revenue. Table 2 shows results from mixed models for gross receipts, earned income, and contributions. Controlling for nonprofit sector, percentage foreign born in the county, and county-level economic variables, we find that a 1% increase in foreign born from all destinations leads to an increase of 0.008 in logged gross receipts or a 1.95% increase. This is equivalent to US$16,725 from the random-effects null-model intercept of US$857,642—a sizable year-on-year change for an organization. This finding is consistent with the predictions that nonprofits will seek and have access to funding through different channels following a local immigration influx. As governments seek to support local needs through the conferral of service provision contracts and grants (Lipsky & Smith, 1989; Lu, 2015), nonprofits will simultaneously look for other sources of funding to maintain their autonomy from government sources (Jung & Moon, 2007; Verschuere & De Corte, 2014). By this logic, community-focused nonprofits will seek donations from local immigrant groups, both to mirror their core values in their funding sources and to increase overall funds so as to reduce volatility (Moulton & Eckerd, 2012; Provan, 1980; Provan et al., 1980).

Table 2.

Multilevel Models of the Relationship Between Change in Immigration and Nonprofit Receipts, Organized by Revenue Type (Values in 2013 Real Logged Dollars).

	Gross receipts			Earned income			Contributions
	All	New	Birth region	All	New	Birth region	All	New	Birth region
(Intercept)	10.132*** (0.101)	10.244*** (0.100)	9.795*** (0.103)	7.449*** (0.033)	7.484*** (0.033)	7.328*** (0.034)	10.470*** (0.421)	10.772*** (0.418)	9.985*** (0.431)
Change in % immigrants over previous 3 years	0.008*** (0.000)			0.002*** (0.000)			0.009*** (0.001)
% immigrants in county	−0.007*** (0.000)			−0.002*** (0.000)			−0.003† (0.001)
Change in % recent arrivals over previous 3 years		0.006*** (0.001)			0.002*** (0.000)			0.005* (0.002)
% recent arrivals in county		−0.004*** (0.001)			−0.001*** (0.000)			−0.002 (0.003)
Change in % from Asia over previous 3 years			0.013*** (0.001)			0.004*** (0.000)			0.013*** (0.003)
% from Asia in county			−0.010*** (0.001)			−0.003*** (0.000)			−0.006† (0.003)
Change in % from Latin America over previous 3 years			0.011*** (0.000)			0.003*** (0.000)			0.013*** (0.002)
% from Latin America in county			−0.009*** (0.001)			−0.003*** (0.000)			−0.002 (0.002)
Change in % from other regions over previous 3 years			0.073 (0.058)			−0.043* (0.019)			0.043 (0.241)
% from other region in county			−0.002* (0.001)			−0.000 (0.000)			−0.001 (0.003)
Median household income in county	−0.826*** (0.023)	−0.870*** (0.023)	−0.756*** (0.024)	−0.217*** (0.008)	−0.230*** (0.008)	−0.192*** (0.008)	−0.957*** (0.098)	−1.031*** (0.097)	−0.855*** (0.100)
Proportion employed in county	−0.628*** (0.016)	−0.599*** (0.016)	−0.617*** (0.016)	−0.183*** (0.005)	−0.177*** (0.005)	−0.180*** (0.005)	−1.084*** (0.066)	−1.046*** (0.068)	−1.064*** (0.066)
Poverty measure (divided by 100)	0.079*** (0.007)	0.084*** (0.007)	0.072*** (0.007)	0.026*** (0.008)	0.028*** (0.008)	0.024*** (0.008)	0.065* (0.027)	0.067* (0.027)	0.056* (0.028)
Controls	Sector, year	Sector, year	Sector, year	Sector, year	Sector, year	Sector, year	Sector, year	Sector, year	Sector, year
N	400,818	400,818	400,818	400,818	400,818	400,818	400,818	400,818	400,818
AIC	−24,326	−23,835	−24,591	−856,346	−856,000	−856,648	1,074,590	1,074,632	1,074,597
BIC	−24,151	−23,661	−24,373	−856,172	−855,826	−856,430	1,074,764	1,074,807	1,074,815
Log likelihood	12,179	11,933	12,315	428,189	428,016	428,344	−537,279	−537,300	−537,278

Note. Nonprofit financial variables and household income variable are reported in logged real 2013 dollars. Poverty measure transformed by dividing by 100. Standard errors in parentheses. For model fit, AIC = Akaike information criterion; BIC = Bayesian information criterion.

^†p < .1.

^*p < .05.

^**p < .01.

^***p < .001.

Models for earned income and contributions show that both forms of revenue increase in response to growth in local immigrant populations (H1B). The coefficients for all immigrants in the contributions model show that, net of nonprofit sector and percentage foreign born in the county, an increase in immigrants from all destinations by 1% is associated with an increase in logged contributions of 2.01%. For earned income, the coefficient for all foreign born is equivalent to an increase of only 0.52%. This result mirrors findings that immigrants and refugees give back to their new communities (Handy & Greenspan, 2009; Hung, 2007; Weng & Lee, 2016) and that inter-ethnic solidarity is reinforced by charitable giving (Vallejo, 2015).

Revenues and Contributions See Similar Responses From All Regions

We next turn to a test of H2, which suggests that contributions to nonprofits will increase in response to immigration from all countries but that increases will be largest in response to immigration from Asian countries. Coefficients in Table 2 show that, indeed, contributions are positively associated with all immigrant groups. The final model shows that an increase in Asian immigration is associated with an increase in nonprofit contributions by 3.02%. This increase is similar (and statistically indistinguishable) to that associated with Latin American immigration, 2.98%. This is surprising, both because of established differences between immigrant subgroups in terms of the types of nonprofits they support and their ability to contribute financially (Borelli & Keister, 2015; Keister et al., 2016; Pew Research Center, 2012, 2014). However, in models without county economic controls (not shown), coefficients on Asian immigration are significantly higher than those for Latin American immigration. This lends evidence to the idea that it is Asian immigrants’ financial means and higher incomes that allows them to give back to their community.

Positive Response Seen Across Nonprofit Sectors, With Differing Magnitudes by Sector

Models shown in Table 3 test H3A, which predicts that all nonprofit sectors will receive additional earned income and contributions as foreign-born populations increase. We include earned income models disaggregated by the four main sectors. We find positive relationships: 0.66% for education, 0.37% for human services, 0.78% for health, and 0.30% for arts (p < .001 for all).

Table 3.

Multilevel Models of Relationship Between Change in Immigration and Nonprofit Receipts, Organized by Revenue Type and Sector (Values in 2013 Real Logged Dollars).

	Earned income				Contributions
	Education	Services	Health	Arts	Education	Services	Health	Arts
Intercept	8.393*** (0.084)	7.249*** (0.050)	8.357*** (0.126)	7.166*** (0.063)	13.578*** (0.973)	9.612*** (0.770)	7.575*** (1.101)	9.850*** (0.856)
Change in % immigrants over previous 3 years	0.003*** (0.000)	0.002*** (0.000)	0.003*** (0.000)	0.001*** (0.000)	0.007* (0.003)	0.012*** (0.002)	0.005† (0.003)	0.006* (0.003)
% immigrants in county	−0.002*** (0.000)	−0.001*** (0.000)	−0.003*** (0.000)	−0.000 (0.000)	−0.006* (0.003)	−0.005* (0.002)	0.004† (0.003)	0.000 (0.002)
Controls	Year, median income, employment, poverty	Year, median income, employment, poverty	Year, median income, employment, poverty	Year, median income, employment, poverty	Year, median income, employment, poverty	Year, median income, employment, poverty	Year, median income, employment, poverty	Year, median income, employment, poverty
N	67,716	60,384	144,798	41,388	67,716	60,384	144,798	41,388
AIC	−139,270	−341,835	−75,356	−129,919	176,938	414,779	170,219	789,62
BIC	−139,160	−341,717	−75,248	−129,815	177,048	414,897	170,327	79,066
Log likelihood	69,647	170,929	37,690	64,971	−88,457	−207,377	−85,097	−39,469

Note. Nonprofit financial variables and household income variable are reported in logged real 2013 dollars. Poverty measure transformed by dividing by 100. Standard errors in parentheses. For model fit, AIC = Akaike information criterion; BIC = Bayesian information criterion.

^†p < .1.

^*p < .05.

^**p < .01.

^***p < .001.

We next turn to a test of H3B, which predicts that the relationship between immigration and contributions received will be strongest for arts nonprofits but that the relationship between immigration and earned income will be strongest for service sector nonprofits (education, health, and human service sectors). Looking at Table 4, we find that there is no statistically significant difference between education and health sector nonprofits in the association between change in earned income. Relative to education, change in immigration is associated with a decrease in earned income for arts and human service nonprofits.

Table 4.

Comparing Sectors Using Contrasts From Multilevel Models of Relationship Between Change in Immigration and Nonprofit Receipts, by Revenue Type.

	Earned income				Contributions
	Reference category				Reference category
	Education	Human services	Health	Arts	Education	Human services	Health	Arts
Education		+	=	+		−	=	=
Human services	−		−	=	+		=	+
Health	=	+		=	=	=		=
Arts	−	=	=		=	−	=

Note. Models from Table 4 are run with interactions between a categorical nonprofit category variable and all independent variables. This table reports whether the coefficient on % change in immigration × nonprofit sector was larger (+, p < .05), smaller (−, p < .05), or not statistically significantly different (=) than the reference category.

Importantly, the models in Table 4 also suggest that nonprofits in the largest sectors (education and health) are more likely to be recipients of earned income as a result of local need than human service nonprofits. One potential explanation is that immediate needs in terms of education and health are a higher priority for local and state governments than human service nonprofits, such as youth charities. Alternatively, given their size and visibility, hospitals and schools may experience the shift in funding across the board from government grants (contributions) to contracts (earned income) compared with other nonprofits (Gazley & Brudney, 2007; Lipsky & Smith, 1989; Lu, 2015). This finding, however, is unsurprising and in line with our expectation that arts nonprofits rely the least on earned income.

All sectors see a statistically significant increase in contributions as immigration increases (p < .1): 1.66% in education, 2.90% in human services, 1.26% in health, and 1.45% in arts. The final columns of Table 4 show that a 1% change in immigration is associated with increased contributions to human service nonprofits relative to the arts sector (p < .05). This is a surprising result: We expected arts nonprofits to see the highest effect for contributions. Human service nonprofits receive a majority of their income from fees for services and goods from private and government sources, so we would expect these fees to grow in response to immigration. However, a nontrivial proportion of nonprofit funding for the human service sector (43% in 2010) comes from contributions (Roeger et al., 2012). As immigration increases, these charities may embark on fundraising or grant writing to solicit individual donations or government grants.

Expenditures Increase When Immigration Increases

To understand whether nonprofits will spend additional resources from all sources in response to changes in immigration in the local community (H4), we run mixed models on nonprofit expenditures, shown in Table 5. Controlling for nonprofit sector and percentage foreign born in the county, we find that a 1% increase in foreign born from all destinations leads to an increase in logged expenditures of 0.001 (0.29%), a US$58,981 increase from the intercept in the null model (US$20,076,777). This figure is 0.21%, or a growth of US$41,900, for new immigrants. In addition, increased expenditure is associated with immigrant population growth among both Latin American and Asian groups. These findings align with H4. Increases in the number of immigrants, particularly from the two largest immigrant groups in the country, is likely to spur further need and prompt nonprofits to spend more resources (Bloemraad, 2005).

Table 5.

Multilevel Models of the Relationship Between Change in Immigration and Nonprofit Expenditure (Values in 2013 Real Logged Dollars).

	All	New	Birth region
Intercept	7.874*** (0.019)	7.889*** (0.019)	7.817*** (0.019)
Change in % immigrants over previous 3 years	0.001*** (0.000)
% immigrants in county	−0.001*** (0.000)
Change in % recent arrivals over previous 3 years		0.001*** (0.000)
% recent arrivals in county		−0.001*** (0.000)
Change in % from Asia over previous 3 years			0.002*** (0.000)
% from Asia in county			−0.001*** (0.000)
Change in % from Latin America over previous 3 years			0.002*** (0.000)
% from Latin America in county			−0.001*** (0.000)
Change in % from other regions over previous 3 years			0.005 (0.011)
% from other region in county			−0.000** (0.000)
Median income	−0.110*** (0.004)	−0.117*** (0.004)	−0.098*** (0.004)
Proportion employed	−0.112*** (0.003)	−0.108*** (0.003)	−0.110*** (0.003)
Poverty measure	0.013*** (0.001)	0.014*** (0.001)	0.012*** (0.001)
Controls	Sector, year	Sector, year	Sector, year
N	400,818	400,818	400,818
AIC	−1,337,496	−1,337,166	−1,337,676
BIC	−1,337,321	−1,336,991	−1,337,458
Log likelihood	668,764	668,599	668,858

Note. Nonprofit financial variables and household income variable are reported in logged real 2013 dollars. Poverty measure transformed by dividing by 100. Standard errors in parentheses. For model fit, AIC = Akaike information criterion; BIC = Bayesian information criterion.

^*p < .05.

^**p < .01.

^***p < .001.

Conclusion and Discussion

This article explored nonprofit organization responsiveness to changing immigration patterns in local communities. Our general underlying focus was on how and when nonprofits are able to pivot as demand for their services changes over time, generating three broad sets of findings. First, we documented that nonprofit financial transactions change in important ways in response to changes in the local immigrant population. Although we cannot offer a causal interpretation of financial responses to immigration, we found strong associational evidence that nonprofit gross receipts, earned income, contributions, and expenditures increase in response to increased local immigration.

Second, we found nonprofits’ financials respond to new arrivals in terms of gross receipts, earned income, contributions, and expenditures. After controlling for economic factors, we find no difference in response to increases in Latin American and Asian immigration. However, in models without controls, it appears Asian immigration is associated with greater increases in contributions. We interpret this as evidence that more contributions in areas with more Asian immigration is associated with higher SES of Asian immigrants, particularly Chinese immigrants (Pew Research Center, 2012; Rumbaut, 2008).

Finally, we found positive, statistically significant relationships between earned income, contributions, and immigration for all sectors. Furthermore, we found that the relationship between immigration and earned income is stronger for education and health nonprofits than for human services and arts. The relationship between immigration and contributions, however, is highest for human service charities. Although it is curious that human service nonprofits have a different relationship with immigration than, say, education nonprofits, this could be due to the relatively larger portion of private donations received by human service nonprofits compared with health and education nonprofits (Roeger et al., 2012).

Some questions remain about what nonprofit fundraising and budgeting processes underlie our findings. First, we expect that nonprofits that cater explicitly to immigrants, which we cannot identify given our data, will respond more strongly to increased immigration than the average nonprofit. However, as our analysis shows, a wide range of nonprofit sectors, from arts to human services, respond to local demographic change; thus, our broad analytical approach was appropriate to gain a better understanding of nonprofit response in the aggregate. Given that we could not include smaller community organizations in our analysis, we likely undercounted local immigrant nonprofits (Gleeson & Bloemraad, 2013). The effect of this potential undercount on our estimates is unclear. On one hand, our analysis probably excluded immigrant organizations that are likely to see the largest response to immigration. On the other hand, larger nonprofits may see increased year-on-year changes given the nature of their financial support (large donations or government contracts). Further research using alternative data sources would shed light on the effects of including smaller organizations on average outcomes.

Second, the nature of the nonprofit financial data from the NCCS provides broad financial categories that do not allow for a more granular exploration of organizational revenues and spending. For example, the contributions category combines both grants and private donations, making it difficult to disentangle whether individuals or governments are the sources of additional funding when immigration rises. Understanding how different sources of contributions and earned income are affected by demographic change is an important next step in this line of research.

Third, there are data limitations that encourage caution when interpreting our results. The Trend Files from the NCCS do not have full data beyond 2013. This means that the associations found are not necessarily valid for later years. However, we expect that nonprofit financial response to immigration will continue to grow as immigration and second-generation individuals will be the main drivers of population growth over the next 50 years and nonprofit provision of social services continues to expand (Budiman, 2020; Smith & Phillips, 2016). Also, we chose the county as the geographic level of analysis for the article due to availability of data at this level. However, nonprofits may serve immigrant populations across county lines, as has been shown (de Graauw et al., 2013). Future work is needed to address this limitation, either by considering other geographic units, or using different approaches to study nonprofit links to local immigrant communities, regardless of county.

Given that nonprofits fill important gaps in the provision of services left by government and private firms (Garkisch et al., 2017; Lipsky & Smith, 1989; Weisbrod, 1988), the social services that nonprofits provide in response to community changes will be available even as governments vary in their commitments to meeting community needs. We build on research suggesting that nonprofits react to their funding and competition environments (Mason & Fiocco, 2017; Smith & Phillips, 2016) and show that they also respond to shifting demographic environments. Understanding their responses to local immigration is of utmost importance, especially in light of increasing numbers of immigrants across the country (Alba, 2020; Frey, 2018).

Biographies

Claire Le Barbenchon is a PhD candidate in public policy and sociology at Duke University where she is also pursuing a master’s in statistical science. Her work is at the intersection of immigration, economic sociology including organizational sociology, and social networks.

Lisa A. Keister is professor in the Department of Sociology and Sanford School of Public Policy at Duke University. She is also an affiliate of the Duke Network Analysis Center. Her research spans topics related to economic sociology including organizational structure and strategy, wealth ownership and accumulation, and inequality and poverty.