Private Provision of Public Goods: Evidence from the Effect of... Water Quality Laura E Grant

advertisement
Private Provision of Public Goods: Evidence from the Effect of Environmental Groups on
Water Quality
Laura E Grant – grantle@uwm.edu – Department of Economics & School of Freshwater
Science, University of Wisconsin, Milwaukee
Christian Langpap – christian.langpap@oregonstate.edu – Department of Applied Economics,
Oregon State University
Abstract: A large literature has focused on explaining why individuals contribute to the
provision of public goods. However, there are no direct tests of the effectiveness in
actually providing a public good. This paper provides such a test. Specifically, we use a
common form of environmental organization, watershed groups, to test if increased
presence of and activity by these groups improves water quality. We find that increased
presence of water groups results in fewer impaired water bodies in a watershed – roughly
five fewer listed water bodies every two years, even after conditioning on group revenues
and expenditures. The results give insight into the role that private groups may play in
mitigating environmental problems, including promoting compliance with environmental
regulations. Finally, our results provide empirical evidence of private provision of a
public good by nonprofit organizations.
Key Words: Public goods, Private Provision, Nonprofits, Water Quality, Environmental
groups.
JEL Codes: H41, Q25, Q53
1. Introduction
The private sector makes significant donations to nonprofit organizations with the intention of
providing public goods. In 2011 total giving in the US amounted to almost $300 billion, or 2
percent of GDP, of which 73 percent came from individual contributions. There were over one
million 501(c)(3) nonprofits with gross receipts of at least $5,000. The average donation per
household in 2006 was $2,213, with 65.5 percent of households making contributions (The
Center on Philanthropy at Indiana University 2010; 2012).
A large literature in economics focuses on explaining why individuals contribute to the
provision of public goods. Theoretical work establishes that donations may be motivated by pure
altruism and warm glow, as well as social approval, public prestige, or a desire to signal income
1 (Bergstrom et al. 1986; Andreoni 1989; 1990; Holländer 1990; Glazer and Konrad 1996;
Harbaugh 1998; Ribar and Wilhelm 2002). Empirical studies examine the determinants of
donations to nonprofits, identifying the influence of household characteristics, fundraising
expenditures, and the possibility of crowding out by public funding (Kingma 1989; Posnett and
Sandler 1989; Smith et al. 1995; Payne 1998; Okten and Weisbrod 2000; Khanna and Sandler
2000; Kotchen and Moore 2007; Andreoni and Payne 2011).
In this literature individuals express their demand for public goods through their
donations, and hence the private provision of public goods is modeled through contributions to
nonprofit organizations (Harbaugh 1998). The prevailing implicit assumption is that a donation
to a nonprofit organization is equivalent to, or directly translates into, supply of the public good
itself. Provision is measured through the inputs to the production process rather than actual
outputs with little attention paid to how the public good is generated by the nonprofits receiving
private donations. In both theoretical and empirical approaches the public good is usually
measured as the sum of all individual donations to nonprofits, with applications that include
public radio (Kingma 1989), rural health care (Smith et al. 1995), green electricity programs
(Kotchen and Moore 2007), or charitable giving broadly defined (Posnett and Sandler 1989;
Payne 1998; Okten and Weisbrod 2000; Khanna and Sandler 2000, Andreoni and Payne 2011).
To the extent that a desire to increase provision of public goods motivates private donors,
donations are only meaningful if they increase the supply of the public good (Duncan 1999). Yet,
it is not clear whether and in what proportion donations to nonprofits or the amounts spent by
these organizations translate into provision of public goods. There is a growing literature in
economics that highlights several reasons why greater donations and increased numbers of
2 nonprofit groups may not necessarily lead to significant increases in the provision of public
goods.
First, a key characteristic of non-profit organizations is that they operate under a nondistribution constraint, which means that no one has a legal claim over the organization’s
residual earnings. Nonprofits lack the traditional ownership structure and governance framework
of for-profit firms, are highly dependent on non-contractible donations, and have managers
whose objectives do not inherently match those of donors or society as a whole. These
characteristics can create incentives for managers to distribute residual earnings through
perquisite consumption expenditures that increase the utility of managers and staff, but reduce
spending on program operations (Rose-Ackerman 1996; Glaeser and Shleifer 2001; Glaeser
2003; Desai and Yetman 2006; Castaneda et al. 2008; Werker and Ahmed 2008). Second, nonprofit organizations compete for donors. This competition may increase fundraising and other
non-program expenses, and thereby reduce the potential production of the public good.
Specifically, competition may lead to excessive fundraising and thus less time and resources
devoted to programs and lower productivity of funds in program output (Rose-Ackerman 1982;
Castaneda et al. 2008; Aldashev and Verdier 2010).1 Finally, compared to costumers of for-profit
firms operating in conventional market settings, the beneficiaries of public goods supplied by
nonprofits cannot rely on market forces to penalize or reward these organizations (Aldashev and
Verdier 2010). As a consequence of these characteristics, nonprofits may not face adequate
incentives to efficiently use donations to provide public goods.
Given the predominant focus in the literature on understanding donations and the
possibility that these contributions may not translate into significant public good provision, it is
1
Competition may also affect perquisite consumption, but this effect could be positive or negative. Aldashev and
Verdier (2010) argue that, by lowering the productivity of funds in program output, additional fundraising reduces
3 important to examine the effectiveness of nonprofits in supplying public goods. A few empirical
papers attempt to evaluate the efficacy of nonprofits in the context of international development
non-governmental organizations (NGO) providing foreign aid. For example, Wane (2004)
examines the effect of the design of foreign aid on aid quality and effectiveness. To measure
project performance, he uses ratings by the World Bank that are based on development impact
and likelihood of sustainability. Nunnenkamp and Öhler (2012) examine how the structure of
financing and the degree of competition among NGOs affect the efficiency of aid delivery.
Following the general literature in donor motivations, they measure efficiency in terms of the
share of administrative and fundraising expenses in the NGO’s budget, rather than using a direct
measure of efficacy in delivery of aid. Henderson and Lee (2011) analyze how different
organizational structures for funding and implementing agencies affect the quality of aid
delivered in the context of disaster relief following the 2004 tsunami in Indonesia. They measure
effectiveness of NGOs in terms of delivery of “hard” aid by using survey data on quality of
construction of housing provided by aid agencies to villagers and failure rate of fishing boats
given to fishermen. However, to the best of our knowledge there have been no direct tests of the
efficacy of nonprofits outside of the foreign aid context. In this paper we take a first step towards
filling this gap by conducting a direct empirical test of the effectiveness of nonprofits in
providing a public good.
To conduct this empirical test we focus on environmental nonprofit organizations.
Environmental groups are potential vehicles for voluntary provision of green public goods such
as endangered species conservation, climate change mitigation, or improvements in the quality of
water resources. Theory and anecdotal evidence suggest environmental groups are important in
providing and protecting environmental amenities (Baik and Shogren 1994; Heyes, A.G. 1997;
4 Cronin and Kennedy 1999; Sundberg 2006; Langpap 2007; Albers et al. 2008). However, as is
the case with nonprofits in general, little is known regarding how presence of and spending by
these groups translate into supply of environmental public goods. Here we provide a direct,
large-scale statistical test of the effectiveness of these organizations in providing a public good.
Specifically, we use a common form of environmental organization, watershed groups, to test if
increased presence of and activity by these groups improves water quality. Over the last decade
and a half the number of groups across the US that act as stewards of local rivers has more than
doubled, from 600 groups in 1992 to 1400 in 2008. Despite this extraordinary increase in the
number and level of activity of environmental nonprofit groups, we know of no empirical
investigation of the extent of their role in encouraging public good provision and mitigating
environmental problems.
Our analysis combines data on water quality, presence of watershed groups, and revenues
and spending by these groups in the US over the period 1996 to 2008. We assess the impact of
these nonprofits on water quality assessments through time. Specifically, we link spatially
explicit information on bi-annual listing of streams as “impaired waters” as required by the Clean
Water Act with data on watershed groups located near those streams, including the
organizations’ characteristics and finances, such as funds raised and watershed-specific spending
by year. We then examine whether these watershed organizations are effective at helping
improve water quality.
This paper makes important contributions to various strands of literature. By focusing on
an output resulting from revenue of nonprofits, we make a contribution to the literature on
private provision of public goods. By examining the impact of these groups on water quality we
contribute to the growing literature on the effectiveness of nonprofit organizations. We also
5 provide the first empirical investigation of the role environmental groups play in mitigating
environmental problems. Finally, this paper contributes to a growing literature on private
enforcement of environmental regulations, which to date has focused almost exclusively on the
role played by environmental nonprofits through citizen suits.
In the following section we describe the data used in our analysis. Section 3 explains our
empirical approach and identification strategy. We present our key results in section 4, and
discuss robustness checks in section 5. In section 6 we discuss implications of our results.
2. Data
We focus on water quality as a measure of output for a privately provided public good. A key
step in constructing an empirical measure of any public good is to note the correct spatial scale.
To account for the relevant factors affecting water quality, the appropriate scale for our research
is land areas designated as watersheds. The definition of a watershed can be ambiguous because
drainage basin sizes vary, and watersheds can be further subdivided into smaller areas defined by
the forks of main rivers.2 Hence, we rely on US Geological Survey (USGS) methodology, which
divides the entire country into successively smaller units embedded within larger units and
identified by unique hydrologic unit codes (HUC). USGS names these units with two to fourteen
digits based on the size class (Seaber et al. 1987). We use the 8-digit HUC (HUC8) designation
as the definition of a watershed. We chose this approach for two main reasons. First, drainage
basins correspond to the natural boundary for surface water flow. Second, nonprofit groups
targeting water quality generally operate at a local scale and the HUC8 scale corresponds to the
2
Watersheds are delineated by geographic topography, defined by an outlet point on a river or into a lake/sea and
outlined by a ridge of land such that precipitation falling within those boundaries that remains as surface water
(rather than ground water) eventually flows through the outlet.
6 smallest area a single group will likely affect (a radius averaging 15 miles).3 There are 2,264
HUC8 watersheds in the US, averaging 700 mi2 in land cover size.
To construct a measure of water quality, we use information generated as a result of
regulatory requirements established by the Clean Water Act (CWA). Section 303(d) of the CWA
mandates states to identify bodies of water (e.g. stream or river segments, lakes) for which
current pollution controls are not sufficient to attain applicable water quality standards (impaired
waters) or have declining quality trends (threatened waters). The US Environmental Protection
Agency (EPA) requires each state to submit a list of all threatened and impaired waters (known
as the “303(d) list”) to during even-numbered years (EPA 2009).4 We use these biannual listings
of impaired water bodies, obtained from the EPA, (http://www.epa.gov/waters/ir/) and count the
number of listed water bodies in each HUC8 watershed for every even year during the 1996 –
2008 period5. This is our measure of water quality and thus of public good provision. Figure 1
shows the locations of 303(d)-listed rivers and streams for our dataset; number listed is quite
heterogeneous by state.
Our unique data on water-focused nonprofit groups comes from several sources. An
initial search in Guidestar,6 an organization that gathers information about nonprofits, yielded a
comprehensive list of groups working in the area of “Water Resource, Wetlands Conservation,
and Management.” The information retrieved includes date of incorporation, location, type of
watershed group and, critically, the federal tax identifier, the Employer Identification Number
3
Personal communications with water group directors and researchers at USGS confirm that these groups carry out
projects and engage the community within a relatively small area.
4
States must also prioritize waters on these lists and develop Total Maximum Daily Loads (TMDLs) for relevant
pollutants. 5
When a water body flows across HUC8 boundaries, we add fractions to the count of listed water bodies in each of
the watersheds involved. For example, if a listed stream flows across three HUC8s, the listing count in each
watershed goes up by one third.
6
The search was conducted with the advanced search function in guidestar.org.
7 (EIN).7 We cross-referenced and supplemented this information with lists from EPA and River
Network, a national group assisting over 2,000 regional and local organizations whose primary
mission is protecting water resources. The two latter data sets do not contain the EIN number, so
we manually searched for and added any missing numbers to our final list. The EIN number was
then used to link this list to a database from the US Internal Revenue Service (IRS) and obtain
tax return data for each group, including yearly revenues and expenditures. Figure 2 shows the
location of the water groups in our data set.
With these data, we can measure public good provision by assessing the impact of
watershed groups on water quality in two ways.8 First, we assess the effect of the total number of
groups that are active in a HUC8 watershed in a given year. Additionally, conditional on the
number of groups, we examine the impact of total revenue and expenditures (net of fundraising
outlays) in a watershed for each two-year period. Watershed group presence and level of activity
can have two effects on the number of listed water bodies. More groups with higher revenues
and expenditures could play a “watchdog” role by helping to identify additional impaired water
bodies, thus leading to further listings. Additionally, through this oversight over time as well as
through direct (e.g. river cleanups, habitat restoration) and indirect (e.g. advocacy, education)
actions, increased presence and activity of watershed groups could lead to improvements in
water quality and therefore reductions in listings. We view both listing and delisting streams as
processes in provision of public goods – the former makes visible the lack of water quality and
the latter remedies it.
7
The EIN number assures us of the legitimacy of a groups’ existence and its longevity.
Watershed groups provide an appropriate context for assessing the effectiveness of environmental groups because
geographical boundaries define a watershed rather than arbitrary political ones, and hence oversight by watershed
groups takes place within the correct spatial scale to account for relevant factors affecting water quality.
8
8 We control for additional factors that could have an impact on water quality at the
watershed level. First, we account for state and federal enforcement of the Clean Water Act by
including the total number of discharge permit violations in a watershed in each two-year period.
Information on violations and location of facilities was obtained from the EPA’s Enforcement
and Compliance History Online (ECHO) database, then aggregated biannually to the HUC8
level. In a related paper, Grant and Grooms (2013) measure the annual effects of groups on
numerous government enforcement actions.
Land use is another factor affecting water quality. Agriculture is a leading cause of
impairment of surface waters (Baker 1992). In urban areas, impervious surfaces contribute to
increased pollutant loads and surface runoff of contaminants and sediment, as well as larger
variances in stream flow and temperature (Tang et al. 2005). We control for these effects by
including measures of urban and rural land cover from the Multi-Resolution Land
Characterization Consortium (MRLC). Using maps of the contiguous U.S. at a 30 by 30 m.
resolution, we derive proportion of each type of land in each HUC8 for each year.
Another determinant of water quality is precipitation. Even relatively small amounts of
precipitation feed pollutants into water bodies through runoff. Conversely, relatively large
amounts of precipitation will accelerate dilution of pollutants. We account for this by including
mean precipitation in a watershed for each year. The data used to construct this variable were
obtained from the PRISM Climate Group (2013), which provides point measurements of
precipitation for the entire U.S. in a continuous 4 km. grid.
Finally, we control for a number of demographic characteristics that can have an impact
on water quality. Specifically, we include population density, per capita income, and percentage
of population with a high school degree. This information was obtained at the census tract level
9 from the US Census for 1990, 2000, and 2010, interpolated for intra-census years, and
aggregated to the HUC8 level based on the proportions of census tracts contained in a watershed.
We construct a panel data set of these variables for 1,860 HUC8 watersheds in 39 states,
with observations for every even year during the period 1996 – 2008. Summary statistics are
presented in Table 1.9 Number of listings per watershed grows through time, from around 8 to
over 38. The growth corresponds to increasing numbers of and spending by water groups per
watershed, implying a negative relationship between groups and water quality. However, other
characteristics like the growth in population density and urban footprint may be driving both. Or
perhaps water groups simply locate where water quality is worse. We overcome these confounds
through empirical estimation.
3. Estimation
3.1 Empirical Models
For number of listings in watershed i in year t, our basic regression model is
Listingsit = α 0 + α1 Water Group Activity it −1 + X itα 2 + ε it
(1)
We consider three alternative specifications, in which Water Group Activityit-1 is measured using
lagged total number of groups, total revenues, or expenditures (net of fundraising) in watershed i,
respectively. When using either revenues or expenditures, we continue to include the number of
groups to reflect both total and per-group effects. The matrix Xit contains the explanatory
variables discussed above (with lagged violations) as well as year fixed effects, state fixed
effects, and trans-boundary fixed effects. We include state fixed effects because water body
listing decisions are made at the state level. We also include trans-boundary fixed effects to
account for watersheds that extend across state lines in the most flexible way possible. These are
9
States omitted due to insufficient data are UT, VT, PA, NY, NE, MS, MN, HI, CT, CO, AK. We also drop data
from California for earlier than 1998 and New Hampshire for earlier than 2000 because of poor data quality.
10 additional fixed effects for the combinations of states included in these trans-boundary
watersheds. We discuss the robustness of our results to alternative ways of accounting for such
watersheds in section 5.
3.2 Identification
A possible concern when estimating model (1) is the potential endogeneity of the number of
water groups, revenue, and expenditure variables due to simultaneity. Summary statistics and
Figures 1 and 2 suggest that water groups tend to be located where water bodies are impaired.
This could be because these groups help identify imperiled water bodies, leading to their listing,
or because water groups choose to locate where water quality is poorer to begin with, which may
also lead to more donations and higher expenditures. The resulting bias in OLS estimates can
understate or even entirely reverse any potential positive impacts on water quality (negative
effects on the number of listings). Therefore, we use instrumental variables estimation to
establish a causal connection between the water group activities and the number of listings.
To construct instruments we rely on a growing body of evidence suggesting that culture,
social relationships, and in particular social capital are associated with voluntary private
provision of public goods and, more specifically, with pro-environmental behavior, conservation
of resources, and environmental protection by local communities. Anderson et al. (2004) show
that common measures of social capital, including attitudinal and behavioral trust, as well as
measures of participation in voluntary activities such as membership in voluntary groups and
attendance at religious services, have a significant impact on contribution levels in a public
goods experiment. Owen and Videras (2007) use survey data from 14 OECD regions and find
evidence that measures of social capital, such as involvement in religious groups and
membership in civil or social organizations, positively influence pro-environmental behaviors
11 and attitudes. Bouma et al. (2008) find evidence that a measure of social capital generated using
a trust game is associated with investments in soil and water conservation in rural villages in
India. Videras et al. (2012) use survey data from the US to examine how social relationships are
related to pro-environmental behaviors, and find that individuals with “green” ties are more
likely to engage in behaviors such as donating to environmental organizations, volunteering to
environmental projects, recycling, or working with other community members to solve
environmental problems. The basic intuition underlying the relationship between social capital
and pro-environmental behavior is that that social bonds and norms are important for
communities, and that social capital lowers the transaction costs of working together, thereby
facilitating investment in collective activities (Glaeser et al. 2002, Pretty 2003). Hence local
environmental groups, and water groups in particular, may be more likely to form and effectively
provide environmental public goods in locations where there is more social capital.
Based on this evidence, we use two measures of social capital as instruments for the level
of activity of private water groups. Our first instrument is the home ownership rate. DiPasquale
and Glaeser (1999) argue that home ownership encourages investment in local amenities and
social capital because it creates incentives to improve community and creates barriers to
mobility. They use data from the US and Germany to show that home ownership is strongly
correlated with social capital variables such as membership in nonprofessional organizations and
involvement in local politics. We construct this instrument using census-tract level data on home
ownership from the US Census for 1990, 2000, and 2010, interpolating for intra-census years,
and aggregating to the HUC8 level.
Our second instrument is the rate of membership in religious congregations and churches.
Religious participation is one of the most commonly used measures of social capital, as
12 participation in religious groups is credited with promoting the creation of communication
networks and fostering trust and reciprocity among members (Anderson et al. 2004, Gruber
2005, Owen and Videras 2007, Gerber et al. 2008, Liu et al. 2009). Robert Putnam, a leading
proponent of the importance of social capital for societal performance, argues that religious
organizations “are arguably the single most important repository of social capital in America”
(Putnam 2000). We construct this instrument using county-level data on congregation
membership rates for 1990, 2000, and 2010 (Association of Religion Data Archives 2013),
interpolating for the remaining years, and aggregating to the HUC8 level.10
Home ownership and church membership rates are plausible instruments. These variables
should not have a direct correlation with water quality, and hence arguably satisfy exclusion
restrictions. As measures of social capital, we expect them to be highly correlated with private
water group activity. The sign of this correlation, however, is undetermined a priori. Higher
home ownership and religious congregation membership rates reflect more social capital, and
thus should lead to greater private water group presence and activity. However, more social
capital will lead to greater overall presence and activity of non-profit groups of all kinds, not just
environmental or water quality-related groups. If there are any substitution effects between other
non-profits and water groups, the net effect of higher social capital on water group activity is
uncertain and could potentially be negative.
Furthermore, both home ownership and church membership rates are higher in rural than
in urban areas, whereas nonprofit groups tend to locate more frequently in urban areas. To verify
this we use population density as a measure of the degree of urbanization and run regressions of
10
The data were downloaded from the Association of Religion Data Archives, www.TheARDA.com, and were
collected by the Association of Statisticians of American Religious Bodies in1990 and 2000, and by principal
investigators Clifford Grammich, Kirk Hadaway, Richard Houseal, Dale E. Jones, Alexei Krindatch, Richie Stanley,
and Richard H. Taylor in 2010.
13 the number of water groups on population density, population density on the home ownership
rate, and population density on the church membership rate. These regressions include a time
trend as well. The coefficients and t-statistics, presented in Table 2, confirm a positive and
significant relationship between presence of watershed groups and population density, and
negative and significant relationships between population density and home ownership and
church membership rates.11 This could also lead to a negative correlation between the
instrumental variables and water group activity. Hence, the expected sign of the instruments in
first-stage regressions is undetermined.
4. Results
4.1. Presence of water groups
We start by examining the effect of the number of water groups in a watershed on listed water
bodies. We estimate a two-stage least squares (2SLS) model with the home ownership rate as the
instrument for the number of water groups.
4.1.1. Determinants of number of water groups: First stage regression
The results for the first-stage regression of the number of water groups on the instrument and the
exogenous variables are presented in Table 3. The estimate on home ownership is negative and
significant. This suggests that the social capital effect is dominated by possible substitution
effects from other types of non-profits, as well as by the opposite correlations between increasing
urbanization, home ownership rates, and location of water groups. The Stock-Yogo (2005) F statistic on the excluded instrument is 47.19, indicating a relevant instrument.
4.1.2. Effect of presence of water groups on water quality: Second stage regression
11
The negative and significant relationship between population density and home ownership and church
membership holds even when these variables are conditioned on each other.
14 Table 4 presents our first key result for the effect of water groups on water quality and hence for
the private production of a public good. For reference we report ordinary least squares (OLS)
estimates, which do not instrument for the number of water groups, in the first column of the
table. The second column shows second stage regression results for 2SLS. Estimated coefficients
on the control variables generally are statistically significant and have the expected sign.
Watersheds with more previous violations and a higher proportion of urban land have more
impaired water bodies, whereas watersheds with more rural land and a higher proportion of high
school graduates have fewer impairment listings. Watersheds where per capita income is higher
also have more listed water bodies, possibly because environmental quality is a normal good and
also because higher income areas tend to be concentrated in more urban watersheds. Finally,
precipitation is positively correlated with the number of impairment listings, which suggests that
on average precipitation contributes more to delivering pollutants into water bodies than to
diluting them.
The estimated coefficient on the number of water groups, the key regressor, is positive
and significant for OLS. This highlights the confounding effects of the two-way causality of
water group location decisions and the water quality impacts of these groups. Once instrumented,
the 2SLS estimate is negative and significant, indicating a positive net impact of the presence of
water groups on water quality in a watershed.12 Specifically, the estimated coefficient suggests
that on average an additional water group active in a watershed reduces the number of listed
impaired water bodies in that watershed by almost eleven.
12
We also estimated an overidentified model by using both home ownership and the church membership rate as
instruments for the number of groups. The Stock – Yogo F – statistic for the instruments is 26.65, but church
membership does not have a statistically significant effect on the number of groups. The second-stage results are
consistent with those presented in Table 4: the coefficient for the number of water groups is negative and significant
(p = 0.067) and of comparable magnitude.
15 To provide some context for the magnitude of this effect, we note that the average
watershed has less than one water group (0.73) and that, conditional on the existence of at least
one group, the average number of groups in a watershed is only three. Hence, considering the
effect of one additional group does not provide a meaningful context. Instead, we measure the
impact of an additional 0.1 groups in a watershed because this is the average number of groups
added every two years (see Table 1). The estimated coefficient for the number of groups
indicates that an additional 0.1 groups in a watershed reduce the number of listings by about 1.1.
In other words, as new groups become established in a watershed, the number of impaired water
bodies decreases by about one every two years. This represents a 5% reduction in listings for the
average watershed.
4.2. Activity of water groups: Revenues and Expenditures
Identifying the effect of the presence of these groups on water quality provides some initial
evidence that they are effective at providing the public good. However, there is considerable
heterogeneity among water groups in size, budget, scope, and types of activities carried out.
Hence, next we conduct a more nuanced analysis of public good provision by these groups. We
examine the impact of their level of activity, conditional on the effect of their presence as
identified in the previous section. We use groups’ revenues and expenditures as measures of their
level of activity.
4.2.1. Determinants of water group activity: First stage regressions
The estimated coefficients for the first-stage regressions of revenues and expenditures on the
instruments and exogenous variables are shown in Table 5. The coefficient on home ownership
is negative and significant in all three models. The coefficient on church membership is negative
and significant in the revenue and expenditure models, but not significant in the regression for
16 the number of water groups. As in the case of home ownership, the negative sign suggests that
the opposite correlations between increasing urbanization, church membership, and water group
location, as well as substitution effects, dominate the impact of this instrument on social capital.
Stock-Yogo (2005) F – statistics are 26.93, 45.61, and 44.69, respectively, which indicates that
weak instruments are not a concern in either of the models.
4.2.2. Effect of water group activity on water quality: Second stage regressions
Table 6 shows estimated coefficients for the effects of total water group revenues and program
expenditures on water quality, conditional on the number of water groups present in a
watershed.13 As before, for reference we present both OLS and 2SLS estimates for each of the
two models. When using OLS the coefficients for revenue, expenditures, and number of water
groups are positive and significant. With 2SLS the coefficient on the number of water groups is
negative and significant in both models, whereas revenue and expenditures have positive and
significant coefficients.
Interpreting these results is challenging because the number of groups in a watershed is
highly positively correlated with total revenue and expenditures in that watershed, yet estimated
coefficients measure impacts conditional on all other covariates remaining constant. The
estimated coefficients for revenue and expenditures indicate that, given the number of water
groups in a watershed, a 10% increase in revenues for those groups leads to 1.3 additional water
bodies listed, and a 10% increase in their expenditures results in 1.5 additional listings. This
suggests that the major impact of additional funding for and expenditures by water groups may
come through activities, such as monitoring to identify water quality issues, lobbying, advocacy,
or legal action, that result in additional water bodies being listed as impaired.
13
We also estimated the model using total expenditures (including fundraising). The results are consistent with those
presented here.
17 The estimated coefficients for number of water groups indicate that, keeping revenue and
expenditures constant, an additional 0.1 groups reduce the number of impairment listings by
roughly five water bodies, a 24% decrease for the average watershed. Alternatively, as more
groups become active, the number of listed water bodies decreases by about five every two
years. Hence, even after accounting for the effects of funds raised and program expenditures on
the number of listings, additional groups in a watershed have a positive net impact on water
quality. This suggests that a significant part of these groups’ activities may not be reflected in
their revenues or expenditures. For instance, most of these groups have a relatively small staff
and rely heavily on volunteers whose activities can have direct impacts on water quality, such as
cleanup and restoration, or longer-term effects through outreach and education. The presence of
additional groups in a watershed may also raise the overall level of awareness about water
quality issues in the community, which could lead to indirect positive impacts on water quality.
5. Sensitivity
5.1. Identification
We repeat our analysis with a reduced form regression of the number of listings in a watershed
on the excluded instruments and the exogenous explanatory variables. Reduced form OLS
estimates are unbiased, and the coefficients on the excluded instruments are proportional to the
causal effect of interest (Chernozhukov and Hansen 2008, Angrist ansd Pischke 2009). Home
ownership is negatively correlated with number of groups, revenues, and expenditures. The
negative effect of number of groups on impairment listings dominates the positive impact of
revenue or expenditures, so we expect home ownership to have a positive sign in the reduced
form. Church membership is negatively correlated with revenue and expenditures, which in turn
have a positive impact on impairment listings. Hence we expect church membership to have a
18 negative sign in the reduced form. The results of the reduced form regression are shown in Table
7. The coefficients on both instruments are statistically significant and have the expected signs,
which supports our causal interpretations.
5.2. Model specification – Poisson
Our dependent variable is the count of listed water bodies in a watershed. Although the range of
this variable (0 to 1,158) is sufficiently wide to justify use of a linear IV model, we assess the
sensitivity of our results to estimation using a Poisson IV specification. The corresponding
results are shown in the second and third columns of Table 7. The results are consistent with
those from the liner IV model: The coefficient for number of water groups is negative and
significant, while the coefficients for revenue and expenditures are positive and significant.
5.3. Trans-boundary watersheds
Within the dataset, 74 percent of the watersheds are fully within the boundaries of a single state.
However, the remaining 415 overlap two or more states. Our models include trans-boundary
fixed effects to account for watersheds that extend across state lines. We verify that our results
are robust to two alternative ways of accounting for such watersheds. First, we drop observations
from watersheds that cross borders with states that are not in our sample. Second, we update
trans-boundary fixed effects into state fixed effects based on the state responsible for the listing.
For example, a trans-boundary fixed effect for a watershed that crosses the Oregon-Idaho border
is added to the Oregon fixed effect if Oregon was responsible for the impairment listing.14 2SLS
estimates for models of revenue and expenditures conditional on number of groups are presented
in Table 8. The results are consistent with those from our preferred specification.15
6. Discussion and conclusions
14
The EPA provides information on which state is responsible for listing when water bodies cross state lines.
We also performed this sensitivity analysis for the model that only includes number of groups. The results are
consistent with those presented here as well.
15
19 The economics literature on private provision of public goods by nonprofit organizations has
traditionally focused on donations, which are inputs to the public good production process, rather
than on the output of the public good itself. However, little is known about the extent to which
contributions to nonprofit organizations translate into increased supply of public goods. This
paper provides the first large scale empirical assessment of the effectiveness of nonprofits at
providing public goods.
Our analysis indicates that increased presence of water groups results in fewer impaired
water bodies in a watershed, even after conditioning on group revenues and expenditures.
Specifically, we find that as the number of groups grows impairment listings decrease at a rate of
roughly five fewer listed water bodies every two years. These findings reflect significant direct
positive impacts of these groups on water quality. Additionally, we find that, conditional on the
number of water groups, additional group revenues and expenditures results in further listings.
This finding reflects the potential indirect effect of these groups on water quality through their
monitoring role.
By showing that watershed groups significantly contribute to improvements in water
quality, our results provide some initial insight into the role that private groups may play in
mitigating environmental problems. These findings also suggest that the role of the private sector
in promoting compliance with environmental regulations may extend beyond litigation through
citizen suits. Finally, and most importantly, our results provide empirical evidence of private
provision of a public good by nonprofit organizations.
20 Table 1. Sample Means by HUC 8 Watershed
Variable
Number of Impairment listings
Number of water groups
Number of water groups
conditional on at least one group
Total revenues of water groups
(1000s $)
Expenditures by water groups (net
of fundraising – 1000s $)
Violations
Percent rural land
Percent urban land
Population density (persons/mi2)
Per capita income (1000s $)
Percent of population with high
school degree
Mean precipitation (1000s mm)
Home ownership rate
Church membership rate
1996
1998
2000
2002
2004
2006
2008
7.88
0.58
14.85
0.65
15.87
0.71
24.93
0.78
29.63
0.83
33.94
0.87
38.27
0.88
2.90
3.01
3.07
3.19
3.26
3.32
3.33
290.93
476.03
769.61
1231.41
774.05
999.79
81.87
973.12
0.07
0.33
0.05
54.57
15.10
838.62
0.07
0.31
0.06
55.76
16.12
1518.29
0.09
0.29
0.06
56.95
17.16
1641.42
0.10
0.27
0.06
57.86
18.33
1704.50
0.19
0.25
0.07
58.76
19.54
2107.32
0.20
0.23
0.07
59.67
20.76
200.74
0.15
0.21
0.08
60.58
21.99
0.37
98.37
0.61
0.47
0.36
100.94
0.61
0.53
0.35
83.00
0.61
0.59
0.35
88.11
0.61
0.57
0.34
94.30
0.60
0.55
0.34
85.85
0.60
0.53
0.33
89.48
0.59
0.51
Table 2. Relationship between Population Density and Number of Water Groups, Home
Ownership, and Church Membership
Coefficient
Regression: Number of
Water Groups on Population
Density
0.006
Regression: Population
Density on Home
Ownership Rate
-42.93
Regression: Population
Density on Church
Membership Rate
-204.24
t - Statistic
53.70
-4.52
-11.15
21 Table 3. First – Stage Regression: Determinants of
Number of Water Groups
Explanatory Variables
Home Ownership t – 1
Dependent Var.: Number
of Water Groups
-1.452***
(0.212)
Violations t – 1
-0.049
(0.092)
Fraction Rural Land
-0.182***
(0.060)
Fraction Urban Land
5.222***
(0.952)
Population Density
0.001**
(4.89E-04)
Per Capita Income
5.21E-05***
(8.47E-06)
High School Education
-4.220***
(0.335)
Ln Mean Precipitation
0.264***
(0.045)
Year Fixed Effects
Yes
State Fixed Effects
Yes
Trans-boundary Fixed
Effects
Yes
Constant
-1.870***
(0.523)
Observations
R2
F – statistic
Prob > F
Stock-Yogo F – statistic
Prob > F
12,560
0.57
70.55
0.000
47.19
0.000
* p < 0.1; ** p < 0.05; *** p < 0.01. Std. errors in parentheses.
22 Table 4. Effects of Presence of Water Groups on Water Quality
Dependent Var.: Number of
Impaired Water Body Listings
Explanatory Variables
OLS
Number of Water Groups t - 1
3.058
-10.606***
(0.409)
(3.984)
Violations t – 1
Fraction Rural Land
Fraction Urban Land
Population Density
Per Capita Income
High School Education
2SLS
***
6.489
5.622***
(1.595)
(2.116)
0.474
-3.283**
(0.925)
(1.545)
***
61.379
128.695***
(9.102)
(26.428)
***
-0.013
0.004
(0.004)
(0.009)
3.79E-04***
0.001***
(8.11E-05)
(2.67E-04)
***
-13.201
**
(5.716)
Ln Mean Precipitation
-72.823***
(19.124)
4.341
7.905***
(0.525)
(1.304)
Year Fixed Effects
Yes
Yes
State Fixed Effects
Yes
Yes
Trans-boundary Fixed Effects
Yes
Yes
-55.082***
-73.171***
Constant
Observations
R2
F – statistic
Prob > F
Wald χ2 – statistic
Prob > χ2
***
(6.826)
(14.987)
12,560
0.46
35.03
0.000
12,560
0.19
5944.11
0.000
* p < 0.1; ** p < 0.05; *** p < 0.01. Std. errors in parentheses.
23 Table 5. First – Stage Regressions: Determinants of Number of Water Groups, Total
Revenue, and Total Expenditures
Explanatory Variables
Dependent Var.:
Number of Water
Groups
Dependent Var.:
Ln Total Revenue
Dependent Var.: Ln
Program
Expenditures
Home Ownership t – 1
-1.450***
-4.232***
-4.282***
(0.213)
(0.609)
(0.598)
Church Membership t – 1
-0.184
-2.184
-2.037***
(0.120)
(0.373)
(0.369)
Violations t – 1
-0.053
0.227
0.098
(0.092)
(0.217)
(0.213)
Fraction Rural Land
-0.161***
-0.415***
-0.513***
***
(0.061)
Fraction Urban Land
(0.158)
5.198***
(0.956)
Population Density
Per Capita Income
High School Education
0.001
(1.387)
12.852***
(1.342)
0.001
0.001
(4.91E-04)
(0.001)
(0.001)
5.24E-05***
7.67E-05***
7.99E-05***
(8.47E-06)
(1.22E-05)
(1.20E-05)
-4.154
**
***
-16.047
***
(0.329)
Ln Mean Precipitation
(0.156)
12.084***
-16.113
(0.819)***
0.717
0.875
(0.045)
(0.099)
(0.100)***
Year Fixed Effects
Yes
Yes
Yes
State Fixed Effects
Yes
Yes
Yes
Trans-boundary Fixed
Effects
Yes
Yes
Yes
-1.720***
-0.640
-2.361
(0.533)
(1.157)
(1.161)**
Constant
Observations
R2
F – statistic
Prob > F
Stock-Yogo F – statistic
Prob > F
0.257
(0.825)
***
12,557
0.57
70.92
0.000
26.93
0.000
***
12,557
0.36
28.97
0.000
45.61
0.000
12,560
0.39
39.74
0.000
44.69
0.000
* p < 0.1; ** p < 0.05; *** p < 0.01. Std. errors in parentheses.
24 Table 6. Effects of Water Group Activity on Water Quality
Dependent Var.: Number of Impaired Water Body Listings
Explanatory Variables
OLS
Ln Total Revenue t – 1
1.307
2SLS
***
OLS
13.989
(0.123)
(4.361)
Ln Program Expenditures t – 1
Number of Water Groups t – 1
1.876
***
(0.395)
Violations t – 1
Fraction Rural Land
Fraction Urban Land
6.107
Per Capita Income
High School Education
Ln Mean Precipitation
(17.293)
-0.231
(1.598)
(5.478)
1.372
(0.911)
52.188
(9.076)
Population Density
-52.296
***
***
***
2SLS
***
1.052***
15.543***
(0.125)
(5.235)
***
2.060
-57.880***
(0.412)
(20.412)
***
1.150
(1.605)
(5.749)
-1.605
1.280
-0.337
(3.494)
(0.909)
(3.782)
173.281
***
(60.212)
***
-0.013
0.040
(0.004)
6.326
***
53.419
171.470***
(9.057)
(64.623)
***
0.045
(0.031)
(0.004)
(0.034)
3.46E-04***
0.002***
3.51E-04***
0.002***
(8.04E-05)
(0.001)
(8.06E-05)
(0.001)
4.076
-13.904
0.577
-11.331
(5.426)
(38.737)
(5.437)
(42.035)
3.632
***
7.714
**
-0.013
***
3.603
5.501
(3.471)
(0.524)
(3.242)
(0.526)
Year Fixed Effects
Yes
Yes
Yes
Yes
State Fixed Effects
Yes
Yes
Yes
Yes
Trans-boundary Fixed
Effects
Yes
Yes
Yes
Yes
Constant
Observations
R2
F – statistic
Prob > F
Wald χ2 – statistic
Prob > χ2
-75.297***
-117.166***
-71.099***
-98.726**
(8.290)
(39.452)
(8.064)
(40.276)
12,557
0.47
35.57
0.000
12,557
12,560
0.46
35.09
0.000
12,560
2789.53
0.000
2525.17
0.000
* p < 0.1; ** p < 0.05; *** p < 0.01. Std. errors in parentheses.
25 Table 7. Sensitivity – Identification and Model Specification
Dependent Var.: Number of Impaired Water Body Listings
Explanatory Variables
Home Ownership t – 1
OLS Reduced
***
Form
16.773
Poisson IV
(4.892)
Church Membership t – 1
-20.922***
(2.728)
0.183*
Ln Total Revenue t – 1
(0.100)
0.162**
Ln Program Expenditures t – 1
(0.076)
***
Number of Water Groups t – 1
-0.215
(0.024)
Violations t – 1
Fraction Rural Land
***
5.736
***
0.840
(0.213)
1.027
-0.026
-0.026
(0.963)
(0.100)
(0.097)
***
***
(9.965)
(0.000)
Population Density
-0.006
0.006***
(0.005)
(0.002)
***
0.001
(8.74E-05)
Ln Mean Precipitation
0.813***
(0.204)
70.447
High School Education
(0.025)
(1.646)
Fraction Urban Land
Per Capita Income
-0.210***
***
0.456
***
2.60E-05
(9.45E-06)
1.028***
(0.000)
0.005***
(0.002)
2.69E-05***
(1.00E-05)
-21.109
0.069
0.018
(5.361)
(0.716)
(0.690)
4.265***
0.471***
0.417***
(0.536)
(0.059)
Year Fixed Effects
Yes
Yes
Yes
State Fixed Effects
Yes
Yes
Yes
Trans-boundary Fixed
Effects
Yes
Yes
Yes
-2.435 ***
-0.934
(0.895)
(1.187)
Constant
-56.617***
(7.424)
Observations
R2
F – statistic
Prob > F
12560
0.45
33.10
0.000
12557
(0.064)
12560
* p < 0.1; ** p < 0.05; *** p < 0.01. Std. errors in parentheses.
26 Table 8. Sensitivity – Trans-boundary Watersheds.
Dependent Var.: Number of Impaired Water Body Listings
Explanatory Variables
Ln Total Revenue t – 1
Use only trans-boundary
watersheds for states in
sample
15.040***
10.767***
(5.245)
(3.212)
***
Ln Program Expenditures t – 1
Number of Water Groups t – 1
Incorporate transboundary FE into state
FE
11.690***
15.819
(5.716)
***
-58.831
***
-62.192
(3.669)
***
-32.942***
(8.523)
(9.592)
-30.377
(21.408)
(23.332)
Violations t – 1
-1.077
-0.721
5.435
7.118
(6.350)
(6.580)
(5.674)
(6.108)
Fraction Rural Land
-2.256
-1.023
1.657
2.642
(3.914)
(4.065)
(2.596)
(2.893)
189.640***
202.683***
65.270
68.313
(69.007)
(73.434)
(49.660)
(50.099)
0.044
0.041
0.074**
0.074**
(0.035)
(0.036)
(0.034)
(0.036)
***
***
***
0.002***
(4.13E-04)
(4.46E-04)
Fraction Urban Land
Population Density
Per Capita Income
0.003
0.003
0.001
(0.001)
(0.001)
-34.151
-34.993
6.511
14.708
(44.336)
(46.274)
(31.146)
(34.156)
8.853**
6.542*
4.814**
2.788
(3.806)
(3.833)
(2.327)
(2.766)
Year Fixed Effects
Yes
Yes
Yes
Yes
State Fixed Effects
Yes
Yes
Yes
Yes
No
No
High School Education
Ln Mean Precipitation
Trans-boundary State Fixed
Effects
Constant
Observations
Wald χ2 – statistic
Prob > χ2
For states in
sample only
For states in
sample only
-125.823**
-103.873**
-77.570***
-62.045**
(50.756)
(48.813)
(23.553)
(24.769)
11557
1921.64
0.000
11560
1921.64
0.000
12557
1537.12
0.000
12560
1406.47
0.000
* p < 0.1; ** p < 0.05; *** p < 0.01. Std. errors in parentheses.
27 Figure 1. 303(d)-Listed Water Bodies
Figure 2. Water Groups by Watershed
28 References
Albers, H.J., A.W. Ando, and X.Chen. 2008. “A Spatial-Econometric Analysis of Attraction and
Repulsion of Private Conservation by Public Reserves.” Journal of Environmental Economics
and Management 56(1): 33 - 49.
Aldashev, G. and T. Verdier. 2010. “Goodwill Bazaar: NGO Competition and Giving to
Development.” Journal of Development Economics 91(1): 48 - 63.
Anderson, L.R., J.M. Mellor, and J. Milyo. 2004. “Social Capital and Contributions in a PublicGoods Experiment.” American Economic Review 94(2), Papers and Proceedings: 373-376.
Andreoni, J. 1989. “Giving with Impure Altruism: Applications to Charity and Ricardian
Equivalence.” Journal of Political Economy 97(6): 1447 – 1458.
Andreoni, J. 1990. “Impure Altruism and Donations to Public Goods: A Theory of Warm-Glow
Giving?” Economic Journal 100(401): 464 – 477.
Andreoni, J. and A. Payne. 2011. "Is Crowding Out Due Entirely to Fundraising? Evidence from
a Panel of Charities." Journal of Public Economics 95: 334 - 343.
Angrist, J., and J.Pischke. 2009. Mostly Harmless Econometrics: An Empiricist’s Companion.
Princeton: Princeton University Press.
Association of Religion Data Archives. US Congregational Membership Reports. Available at
www.TheARDA.com. Accessed July 2013.
Baik, K.H. and J.F. Shogren. 1994. “Environmental Conflicts with Reimbursement for Citizen
Suits.” Journal of Environmental Economics and Management 27: 1 – 20.
Baker, L.A. 1992. “Introduction to Nonpoint Source Pollution in the U.S. and Prospects for
Wetland Use.” Ecological Engineering 1: 1-26.
Bergstrom, T.C., L. Blume, and H. Varian. 1986. “On the Private Provision of Public Goods.”
Journal of Public Economics 33: 25 – 49.
Bouma, J., E. Bulte, and D. van Soest. 2008. “Trust and Cooperation: Social Capital and
Community Resource Management.” Journal of Environmental Economics and Management 56:
155 – 166.
Castaneda, M.A., J. Garen, and J. Thornton. 2007. “Competition, Contractibility, and the Market
for Donors to Nonprofits.” Journal of Law, Economics, and Organization 24(1): 215 - 246.
Chernozhukov, V., and C. Hansen. 2008. “The Reduced Form: A Simple Approach to Inference
With Weak Instruments.” Economics Letters100: 68–71.
29 DiPasquale, D., and E.L. Glaeser. 1999. “Incentives and Social Capital: Are Homeowners Better
Citizens?” Journal of Urban Economics 45: 354 – 384.
The Center on Philanthropy at Indiana University. 2010. Overview of Overall Giving.
http://www.philanthropy.iupui.edu/files/research/2007_copps_key_findings.pdf
Accessed December 2012.
The Center on Philanthropy at Indiana University. 2012. Giving USA 2012. The Annual Report
on Philanthropy for the Year 2011.
Cronin, J., and R.F. Kennedy, Jr. 1999. The Riverkeepers. New York: Touchstone.
Desai, M.A. and R.J. Yetman. 2006. “Constraining Managers Without Owners: Governance of
the Not-for-Profit Enterprise.” NBER Working Paper.
Duncan, B. 1999. “Modeling Charitable Contributions of Time and Money.” Journal of Public
Economics 72: 213 - 242.
Gerber, A., J. Gruber, and D.M Hungerman. 2008. “Does Church Attendance Cause People to
Vote? Using Blue Laws’ Repeal to Estimate the Effect of Religiosity on Voter Turnout.” NBER
Working Paper 14303.
Glaeser, E.L., D. Laibson, and B. Sacerdote. 2002. “An Economic Approach to Social Capital.”
The Economic Journal 112: F437 – F458.
Glaeser, E.L. 2003. “Introduction.” In E.L. Glaeser, Ed. The Governance of Not-for-Profit
Organizations. Chicago: The University of Chicago Press.
Glaeser, E.L. and A. Shleifer. 2001. “Not-For-Profit Entrepreneurs.” Journal of Public
Economics 81: 99 – 115.
Glazer, A., and Konrad, K.A. 1996. “A Signaling Explanation for Charity.” American Economic
Review 86: 1019 – 1028.
Grant, L. E., and Grooms, K. 2013. “Do nonprofits encourage compliance? Watershed groups
and the U.S. Clean Water Act.” Working Paper.
Gruber, J. 2005. “Religious Market Structure, Religious Participation and Outcomes: Is Religion
Good for You?” Advances in Economic Analysis and Policy 5(1): Article 5.
Harbaugh, W.T. 1998. “What Do Donations Buy? A Model of Philanthropy Based on Prestige
and Warm Glow.” Journal of Public Economics 67: 269 – 284.
Henderson, J.V. and Y.S. Lee. 2011. “Organization of Disaster Aid Delivery: Spending Your
Donations.” NBER Working Paper.
30 Heyes, A.G. 1997. “Environmental Regulation by Private Contest.” Journal of Public Economics
63: 407 – 428.
Holländer, H. 1990. “A Social Exchange Approach to Voluntary Cooperation.” American
Economic Review 80: 1157 – 1167.
Khanna, J. and T. Sandler. 2000. "Partners in Giving: The Crowding-In Effects of UK
Government Grants." European Economic Review 44(8): 1543 - 1556.
Kingma, R.B. 1989. “An Accurate Measurement of the Crowd-Out Effect, Income Effect, and
Price Effect for Charitable Contributions.” Journal of Political Economy 97: 1197 – 1207.
Kotchen, M.J. and M.R. Moore. 2007. “Private Provision of Environmental Public Goods:
Household Participation in Green-Electricity Programs.” Journal of Environmental Economics
and Management 53(1): 1 - 16.
Langpap, C. 2007. “Pollution Abatement with Limited Enforcement Power and Citizen Suits.”
Journal of Regulatory Economics 31(1): 57-81.
Liu, B., S.D. Wright Austin, and B.D. Orey. 2009. “Church Attendance, Social Capital, and
Black Voting Participation.” Social Science Quarterly 90(3): 576 – 592.
Nunnenkamp, P. and H. Öhler. 2012. “Funding, Competition, and the Efficiency of NGOs: An
Empirical Analysis of Non-charitable Expenditure of US NGOs Engaged in Foreign Aid.”
Kyklos 65(1): 81 – 110.
Okten, C., and B.A. Weisbrod. 2000. “Determinants of Donations in Private Nonprofit Markets.”
Journal of Public Economics 75: 255–272.
Owen, A.L, and J.R. Videras. 2007. “Culture and Public Goods: The Case of Religion and the
Voluntary Provision of Environmental Quality.” Journal of Environmental Economics and
Management 54: 162 – 180.
Payne, A. 1998. "Does the Government Crowd-Out Private Donations? New Evidence from a
Sample of Non-profit Firms." Journal of Public Economics 69: 323 – 345.
Posnett, J., and Sandler, T. 1989. “Demand For Charity Donations in Private Non-Profit Markets
The Case of the U.K.” Journal of Public Economics 40: 187 – 200.
Pretty, J. 2003. “Social Capital and the Collective Management of Resources.” Science 302:
1912 – 1914.
PRISM Climate Group, Oregon State University, http://prism.oregonstate.edu, accessed July 8,
2013.
Putnam, R.D. 2000. Bowling Alone: The Collapse and Revival of the American Community. New
31 York: Simon and Schuster.
Ribar, D.C., and M.O. Wilhelm. 2002. “Altruistic and Joy-of-Giving Motivations in Charitable
Behavior.” Journal of Political Economy 110(2): 425 – 457.
Rose-Ackerman, S. 1982. “Charitable Giving and Excessive Fundraising.” Quarterly Journal of
Economics 97(2): 193 - 212.
Rose-Ackerman, S. 1996. “Altruism, Nonprofits, and Economic Theory.” Journal of Economic
Literature 34(2): 701 – 728.
Seaber, P.R., F.P Kapinos, and G.L. Knapp. 1987, Hydrologic Unit Maps: U.S. Geological
Survey Water Supply Paper 2294. Available at http://pubs.usgs.gov/wsp/wsp2294/
Smith, V.H., M.R. Kehoe, and M.E. Cremer. 1995. “The Private Provision of Public Goods:
Altruism and Voluntary Giving.” Journal of Public Economics 58:107 – 126.
Stock, J. and M.Yogo. 2005. “Testing for Weak Instruments in Linear IV Regression.” In Stock,
Andrews, Eds., Identification and Inference for Econometric Models: Essays in Honor of
Thomas J. Rothenberg,CambridgeUniversityPress.
Sundberg, J.O. 2006. “Private Provision of a Public Good: Land Trust Membership.” Land
Economics 82 (3): 353 – 366.
Tang, Z., B. Engel, B. Pijanowski, and K. Lim. 2005. “Forecasting Land Use Change and its
Impact at a Watershed Scale.” Journal of Environmental Management 30: 391 – 405.
U.S. Environmental Protection Agency. 2009. “Fact Sheet: Introduction to Clean Water Act
Section 303(d) Impaired Waters Lists.” Available at http://www.epa.gov/owow/tmdl/results/pdf/
aug_7_introduction_to_clean.pdf. Accessed July 16, 2013.
Videras, J., A.L. Owen, E. Conover, and S. Wu. 2012. “The Influence of Social Relationships on
Pro-Environment Behaviors.” Journal of Environmental Economics and Management 63: 35 –
50.
Wane, W. 2004. “The Quality of Foreign Aid: Country Selectivity or Donors Incentives?”
World Bank Policy Research Working Paper 3325.
Werker, E. and F.Z. Ahmed. 2008. “What Do Nongovernmental Organizations Do?” Journal of
Economics Perspectives 22(2): 73 - 92.
32 
Download