Education have on voluntary contributions to education

EDUCATION’S GAMBLING PROBLEM: EARMARKED LOTTERY
REVENUES AND CHARITABLE DONATIONS TO EDUCATION
DANIEL B. JONES
I examine the impact that lotteries introduced to support education have on voluntary
contributions to education. State lotteries, and the causes they are introduced to support,
are highly publicized. This provides the opportunity to assess whether donors are
crowded-out by government spending of which they are almost certainly aware. Using
donor-level survey data and nonprofits’ tax returns, I find that donations to educationrelated organizations fall with the introduction of a lottery. This result is driven by
donors’ response to the new (highly publicized) government revenue source (rather than
a decrease in nonprofit fundraising efforts). (JEL D64, H3, H75)
I. INTRODUCTION
Over the past several decades, state governments in the United States have come to embrace
lotteries as an alternative source of revenue. Lotteries have proven to be successful in raising revenue; on average, lotteries add nearly 500 million
dollars to each state’s budgets yearly.1 While a
handful of states add lottery revenue to their general funds, states typically earmark the revenue
to support particular public goods. States adopt
lotteries with the intention of funding causes as
diverse as environmental protection, the arts, and
support for their elderly, but most commonly lottery funds are earmarked for education. Twenty
of the 43 states that currently sponsor lotteries direct all of their revenues toward education, while several more dedicate at least some
fraction to education. However, existing research
suggests that, at best, education earmarking fails
to increase education funding by the promised
amount (Evans and Zhang 2007; Novarro 2005);
at worst, total education funding either remains
constant (Garrett 2001; Spindler 1995) or falls
with the introduction of a lottery (Borg, Mason,
and Shapiro 1991; Erekson et al. 2002).
Even if earmarked lottery revenues do not
increase government’s contribution to the
intended public good, government of course
Jones: Department of Economics, Darla Moore School of
Business, 1705 College Street, Columbia, SC 29208.
Phone (803) 777–4940, Fax (803) 777–6876, E-mail
[email protected]
1. On the basis of the 2008 Survey of Government
Finances.
is not the only source of funding for many public
goods. In most cases, the causes supported by
state lotteries also benefit from and rely on charitable contributions. This is especially true of
education. In aggregate, education-related organizations consistently receive more donations
than any other secular cause in the United States.
Americans donated a total of 38.87 billion dollars
toward education in 2011, which is roughly twice
the amount of money that was raised through
state lotteries in the same year.2 An examination
of government expenditures alone therefore does
not capture the full impact that a lottery has on
public good provision, as the lottery may also
affect charitable contributions.
With this in mind, I examine the impact of
the introduction of education lotteries on private donations. Standard models of public good
provision suggest that if an individual’s utility depends at least in part on the overall level
of the public good, then government spending
serves as a substitute for charitable contributions
(Andreoni 1989; Bergstrom, Blume, and Varian
1986). Thus, we should expect charitable contributions to decrease with an increase in government spending. If, on the other hand, donors
are motivated entirely by “warm glow” or the
2. Giving USA, 2012.
ABBREVIATIONS
CES: Consumer Expenditure Survey
COPPS: Center on Philanthropy Panel Study
DID: Difference-in-Differences
GVS: Giving and Volunteering Survey
906
Economic Inquiry
(ISSN 0095-2583)
Vol. 53, No. 2, April 2015, 906–921
doi:10.1111/ecin.12194
Online Early publication January 11, 2015
© 2015 Western Economic Association International
JONES: EDUCATION LOTTERIES AND DONATIONS 907
joy of giving, then government spending is not
a good substitute for one’s own donation and
should therefore have no impact on donations.
Numerous empirical tests have generally found
some drop in giving, though the magnitude of the
crowd-out is typically small.3 If donors respond
to the announced increase in government funding associated with the introduction of a lottery,
then this—combined with the fungibility of lottery revenue—may imply that lotteries lead to a
decrease in total provision.
I assess the degree to which lottery revenue
impacts donors’ contributions using several
individual-level surveys: Center on Philanthropy
Panel Study (COPPS), Giving and Volunteering
Survey (GVS).4 Collectively, these surveys span
from 1989 to 2008, so all of the analysis in the
paper focuses on this time period. All of these
surveys ask respondents to indicate how much
money they have donated recently to a variety
of causes, including education. In a differencein-differences (DID) framework I compare the
level of education-related donations before and
after a state has introduced an education-funding
lottery. I find a significant decrease in education
giving when an education lottery is introduced.
I then address why contributions fall in this
context. In doing so, I speak to more general
questions in the literature on donors’ motivations
and response to government activity. Andreoni
and Payne (2003, 2011) show that the negative relationship between charitable contributions
and government grants to nonprofits can in some
cases almost entirely be explained by a decrease
in fundraising. Their results might suggest that
donation decisions are in fact relatively unresponsive to the overall level of the public good. In this
paper, I empirically examine a different explanation for their result and for the small degree of
crowd-out that is often observed in the literature:
donors may be largely unaware of government
activity in most settings. While this theoretical
possibility has been discussed in the literature,5
(to my knowledge) this is the first paper to empirically assess the importance of the salience of
government spending.
3. See Vesterlund (2006) for a review of the empirical
crowd-out literature.
4. Some additional analysis in an online appendix makes
use of a third survey: the Consumer Expenditure Survey.
5. In particular, both Garrett and Rhine (2010) and Monti
(2010) point to higher awareness of government activity as a
potentially important difference between direct government
spending and spending through government grants. Monti
presents a model demonstrating the impact that increased
awareness may have on donations.
Unlike government spending in the form of
grants to nonprofits, the intended increase in government spending associated with the introduction of a lottery is highly publicized and the beneficiary is well known. States are eager to advertise that revenues go toward a “good cause,” perhaps to overcome moral opposition to the lottery
and draw in customers who might not otherwise
gamble (Clotfelter and Cook 1990); advertisements therefore typically include some reminder
of the cause supported by lottery revenues (Clotfelter and Cook 1991). Thus, state lotteries provide the opportunity to test whether donors (and
not just fundraisers) respond to government activity in a setting where government spending is
highly salient.
To determine whether donors or nonprofits
drive crowd-out, I analyze the tax returns of
a random sample of nonprofits in the same
DID framework. I find that an education lottery
decreases donations received by educationrelated organizations by roughly 8%. This is
not driven by a change in fundraising behavior. Moreover, there is a negative relationship
between donations received and a proxy for a
state’s lottery advertising expenditures. This suggests that donors’ response to (the perception of)
increased government spending on education is
dependent on the salience of government activity.
While researchers have examined a variety of
issues related to state lotteries, the general impact
of state lotteries as a means to finance public
goods is not well understood. This paper fills this
gap by examining the side of education funding
that has been neglected in this literature: charitable donations. In doing so, the results also
contribute to the more general literature on the
interaction of government activity and charitable giving. Recent work in this area has generally found that crowd-out is largely explained by
fundraiser behavior. The results presented here
point to the importance of salience of government activity; when donors are more aware of
government activity, their behavior is more in line
with the crowd-out predicted by classic models of
voluntary contributions to public goods.
II. ADDITIONAL BACKGROUND: LOTTERY AND
CHARITABLE SUPPORT FOR EDUCATION
Before proceeding to the analysis, some
additional detail on state lottery and charitable
support for education will help fix ideas. In
particular, the degree to which we might expect
donors to reduce their contributions depends in
908 ECONOMIC INQUIRY
part on their perception of the overlap between
the causes they support and the specific causes
supported by the lottery. Thus, despite the fact
that most of the analysis will center on the impact
of lotteries on education spending and giving in
general, here I discuss which particular causes
within education tend to benefit from each source
of funding.
As noted, education is typically the most
popular secular category of giving in the United
States, second only to religious giving. According to a recent Giving USA report, 38.87 billion
dollars and 13% of all charitable donations
went to education-related causes in 2011.
This figure—and the “education giving” discussed throughout—includes donations to a
wide array of education-related organizations:
“giving to the education subsector includes
giving to support nonprofit, public, and charter pre-K through grade 12 schools; nonprofit
and public colleges and universities; vocational and technical schools; nonprofit and
public libraries; education research and policy;
adult education programs; tutoring programs;
and student services organizations.”6 However, a majority of donations to education
(roughly 78% in 2011) support public and private higher education (including scholarship
and financial aid programs). There is a fairly
even split between support for public and private institutions: in 2011, private institutions
received 55% of donations to higher education, while public institutions received 45%
of donations.7
So while a broad array of causes fall under
the umbrella of “support for education,” the main
beneficiaries are public and private institutions of
higher education. We will see that this is generally true of education lotteries introduced during
the sample period as well. The main donor-level
data I use spans from 1989 to 2008 so I focus
on this time period throughout. The states that
introduced education lotteries during these years
and the specific causes that they currently support are listed in Table 1.8 Like private charitable
support for education, a majority of these lotteries
are currently designed—at least in part—to fund
6. Source: Giving USA 2012 report.
7. Source: Council for Aid to Education Annual Survey
(2012).
8. A comprehensive historical listing of specific beneficiaries is not available. All lotteries listed have supported
some education-related cause(s) since the date indicated.
Some states (like South Carolina) adjust the specific composition of their beneficiaries on a year-to-year basis.
higher education. Many of these lotteries were
accompanied by the introduction of large-scale,
state-run, lottery-funded scholarship programs.9
Many of the lottery programs also support programs outside of higher education that often fall
within the private nonprofit sector, such as literacy programs and pre-kindergarten programs for
low-income children.
III. GENERAL EMPIRICAL APPROACH
Throughout the paper, I employ a DID
approach to identify the impact of an education
lottery on donors’ contributions (Section IV), and
donations received by nonprofits (Section V).
The generic empirical specification employed
throughout is:
yist =α+βDIDEdulotist +βxXist +[
state FE′
s
]
s
+ [
year FE′
s
]
t
where yist is the outcome variable of interest
and Xist is a vector of individual-level covariates.
“Edulotist” is an indicator variable equal to one if
observation i is in a state (s) that, at that point in
time (t), sponsors an education-funding lottery.10
Throughout, standard errors are adjusted to allow
for clustering at the state level.
One potential concern with the DID approach
is that states may introduce education lotteries
in response to a decrease in the availability of
education funding from either public or private
sources. This would violate the assumption of
parallel pre-treatment trends across treatment
and control states. However, factors that are
unrelated to education financing (e.g., withinstate religiosity, the adoption of a lottery in
a neighboring state) have been shown to be
more important predictors of lottery adoption than fiscal crises (Coughlin, Garrett, and
Hernández-Murillo 2006), particularly in lotteries introduced after the 1970s (Alm, McKee,
and Skidmore 1993). In online appendix Table
A1, I report results suggesting that donations are
not different in treatment states just prior to the
9. Georgia’s HOPE Scholarship is a prominent example
and seems to have served as a model for several states that
followed.
10. An “education-funding lottery” is defined here as a
lottery that is introduced solely for the purpose of funding
education. Some states defined here as education lotteries use
a small fraction of their revenues for other causes, but only
after achieving a certain threshold of funding for education.
Thus, more precisely, an education lottery is defined here as a
lottery for which the entire first dollar of revenue is earmarked
for education.
JONES: EDUCATION LOTTERIES AND DONATIONS 909
TABLE 1
Education Lotteries Introduced during Sample Period
State
Education
Lottery
Established Specific Beneficiaryb
Georgia 1993 Higher ed. scholarships (public & private schools), funding for pre-K programs
Missouri 1993a Programs at all levels of public education
New Mexico 1996 Higher ed. scholarships (public universities/community colleges)
Texas 1997a Public K-12
Vermont 1998a “Education fund”
Virginia 2000a Public K-12
Washington 2001a Higher ed. scholarships/fin. aid (public and private schools), low-income pre-K programs
South Carolina 2002 All levels of education, scholarships/fin. aid for public and private universities
Tennessee 2004 Higher ed. scholarships (public and private), pre-K and after-school programs
Kentucky 2005a Higher ed. scholarships (public and private), early childhood literacy programs
Oklahoma 2005 Public K-12, Higher ed. grants/loans/scholarships, Other higher ed. programs
North Carolina 2006 Public K-12, Higher ed. scholarships/financial aid, and pre-K programs
aThese states already had a lottery (with revenues going toward a different cause or a general fund) but switched to earmarking
funds only for education in the year indicated. bInformation on specific beneficiaries is obtained from state lottery websites and was current as of early 2013.
application of treatment.11 In Section V, I use an
event-study methodology and find no difference
between treatment and control states with regards
to donations received by education organizations
prior to treatment.
IV. DONOR RESPONSE TO EDUCATION LOTTERY
REVENUE
A. Data and Empirical Approach
How do donors respond to the introduction
of an education lottery? To begin to answer this
question, I primarily draw from two individuallevel surveys: the Giving and Volunteering in
the United States Survey (GVS) and the Center on Philanthropy Panel Study (COPPS).12
These surveys ask respondents to indicate how
much they have donated to a variety of causes,
including education.
GVS and COPPS were designed to gather
information about individuals’ charitable activities and are two of the most widely used sources
of data on the topic. Both surveys ask detailed
questions about the amount donated to various
charitable causes such as education, health, public services, etc., in addition to more basic demographic information. COPPS follows a panel of
11. Additionally, in analyses not reported here, I find
that lottery states’ governments do not experience drops in
revenue or increases in education expenditures in the years
preceding to the adoption of a lottery.
12. Additional results drawing from the Consumer
Expenditure Survey (CES) are in an online appendix.
individuals between 2001 and 2009 (with surveys
every 2 years). GVS is not a panel, but I have
constructed a repeated cross-section of surveys
between 1990 and 1999 (again, with waves every
2 years—until 1996, when the next wave was not
administered until 1999). In both surveys, participants are asked about their charitable giving in
the preceding calendar year, so collectively GVS
and COPPS provide results for the years 1989
through 2008.
The COPPS data are preferable as it is a
panel and allows for individual fixed effects,
thereby controlling for unobserved differences
in altruism. However, given that identification in
the DID framework stems from a state establishing an education lottery within 2001–2009,
one might be concerned that the results are
driven by something specific about this handful of states. Thus, the GVS data is included to
further support the robustness of the results by
providing additional observations during a different decade with different states introducing
education lotteries.
The primary outcome variable of interest in
all three datasets is total giving to education. In
both COPPS and GVS, respondents report education giving for the preceding calendar year. To
provide some sense of the magnitude of giving
in these samples, the mean of education giving
is roughly $40 in both datasets. In GVS, only
16% of respondents report making any donation to an education-related cause. In COPPS,
13% of respondents report positive donations
to education. To provide some sense of the
910 ECONOMIC INQUIRY
FIGURE 1
Distribution of Education-Related Donations in
the Survey of Giving and Volunteering
(Conditional on Making any Education
Donation)
distribution of giving, Figure 1 is a histogram
of donations in the COPPS data conditional on
making any donation. The distribution is similar
in GVS. As documented in the figure, even when
restricting attention to positive donations, the
distribution of donations is still rather skewed
toward low donations.
Included covariates vary by the survey and
empirical approach being used. In the COPPS
data, all specifications control for family income.
When the COPPS data is estimated without individual fixed-effects, I also include a variety of
additional controls: number of children, employment (respondent and spouse), marital status,
urban-rural residence status, age, sex, and race.
GVS specifications include controls for race,
gender, employment (respondent and spouse),
church attendance, age, education level, income,
marital status, children in household, and confidence in education (as indicated in the survey). Regardless of the survey being used, all
specifications include year fixed effects and state
fixed effects (unless the specification includes
individual-level fixed effects.)
B. Results
Results from the baseline specifications in
both datasets are presented in Table 2. Column
1 reports the results of a fixed-effects regression
in the COPPS data; Column 2 reports the results
of the repeated cross-section analysis in the GVS
data. In either case, we find that education giving
significantly decreases when an education lottery
is introduced.
I also estimate logit models to assess how an
education lottery impacts giving on the extensive
margin. Results are presented in columns 3 and 4
for COPPS and GVS, respectively. In both models, the dependent variable is equal to one if the
respondent reports any education giving. There
is little response to the introduction of an education lottery. Thus, the baseline results are driven
by changes on the intensive margin. This provides an initial indication that these results may
not be entirely driven “fundraiser crowd-out.” If
the only reason that contributions decrease is a
decline in the number of donors being solicited,
then we might expect to find that the drop in
giving is driven by the extensive margin.
There is reason to be concerned that the simple baseline results might be biased due to the
large number of individuals who contribute nothing to education. This concern is addressed in two
ways, with results reported in Table 3. In analyzing the COPPS data, I can restrict the sample to
“education-givers”—individuals who donate to
education at any point in the panel. This substantially reduces the number of zero-contribution
observations. This specification is also interesting in its own right as it estimates the impact of
the treatment on the individuals who would be
giving. As the GVS is not a panel, the GVS parallel to this is to restrict the sample to observations
with positive education contributions.13 Results
from these estimations are reported in Columns
1 and 3, respectively. Again, we see a significant
decrease in giving in both datasets/decades but,
as we would expect, the magnitude is much larger
than the baseline result.
In Models 2 and 4, I estimate Tobit models to
address “censoring” of contributions at $0. There
is not a straightforward and unbiased implementation of fixed effects in Tobit models for panel
data, so in the COPPS data I instead estimate a
standard Tobit model, adjusting standard errors
for clustering at the individual-level (Column
3).14 Similarly, in Column 4 I report the results
of estimating a Tobit model in the GVS data. For
each of these specifications, I report the marginal
effect of “Edulot” on the unconditional expected
value of observed giving. In both cases, we continue to observe a significant decrease in giving
13. Restricting our attention to education givers would be
problematic if the treatment changed the set of donors and not
just the size of their contribution.
14. Estimating a random-effects Tobit model yields
similar results.
JONES: EDUCATION LOTTERIES AND DONATIONS 911
TABLE 2
Baseline Results— Impact of Education Lottery on Education Giving
(1) (2) (3) (4)
Variables Educ. giving Educ. giving Any educ. giving Any educ. giving
Edulot −9.372* −33.49*** −.033 −.003
(5.153) (11.94) (.032) (.019)
Observations 29,715 11,017 7,985 11,012
Dataset COPPS (2000–2008) GVS (1989–1998) COPPS (2000–2008) GVS (1989–1998)
Model FE Reg. OLS FE Logit Logit
Robust standard errors (clustered at state level) in parentheses.
Columns 3 and 4 report marginal effects.
***p < .01, **p < .05, *p < .1.
TABLE 3
Alternative Specifications— Impact of Education Lottery on Education Giving
(1) (3)
Educ. giving (2) Educ. giving (4)
Variables (Educ. givers only) Educ. giving (Educ. givers only) Educ. giving
Edulot −27.77* −8.44* −191.9*** −11.96*
(16.27) (5.12) (53.09) (7.01)
Observations 9,279 28,426 1,801 11,017
Dataset COPPS COPPS GVS GVS
Model FE Tobit OLS Tobit
Robust standard errors in parentheses (clustered at state level in Models 1 and 3, individual level in Models 2 and 4). Columns
2 and 4 report marginal effect on the unconditional expected value of observed giving.
***p<.01, **p<.05, *p<.1.
after accounting for the large number of censored observations. These estimates suggest that
the introduction of an education lottery decreases
average giving by between $8 and $12; from an
average of $40, this represents a drop in giving of
between 20% and 30%.
What is driving this drop in giving? The
decrease is consistent with classic models of
crowd-out; the expected introduction of a new
source of funding for a public good serves as
a substitute for individual contributions and as
such donors reduce their level of giving. There
are of course alternative explanations. Kearney
(2005) finds that, for the average lottery player,
lottery spending is entirely financed by a reduction in non-gambling expenditures; thus, it is reasonable to expect that lottery spending may come
at the expense of a particular category of nongambling expenditures: charitable giving. If this
were the case, we would expect charitable giving to decrease generally instead of finding a drop
only in education-related giving.
It is not the case that giving to other causes
substantially decreases with the introduction of
an education lottery. To show this, I again estimate the baseline specifications (Columns 1 and
2 of Table 2) and the Tobit models (as in Table 3)
but take “non-education giving” as the dependent variable.15 Results are reported in Table 4.
With the exception of Column 3 (where there is
a small and insignificant drop in non-education
giving), we see that giving to other causes actually slightly increases, but this increase is not significant. (The magnitudes of these coefficients are
larger than those of the education-only estimations as the mean of giving to the sum of other
causes is naturally much higher than giving to just
education. For instance, average non-education
giving in GVS is roughly $224.)
Another alternative to the crowd-out explanation is that donors view the lottery as a new
way to contribute to education. The idea that
donors view donations and lottery expenditures
as equally good ways to contributions drives
15. In GVS, “non-education giving” is defined as total
reported giving minus education giving. In COPPS, respondents do not report “total giving” and the way that they are
asked to report giving to several causes changed between
the 2002 wave and the remaining waves. However, questions
regarding education giving, religious giving, “combined purpose” giving (e.g., United Way), health giving, and “help
for the needy” are consistent across waves. Thus, in COPPS
“non-education giving” is the sum of these consistently
measured categories (religious, health, combined purpose,
and needy).
912 ECONOMIC INQUIRY
TABLE 4
Giving to Other Non-Education Related Causes
(1) (2) (3) (4)
Variables Non-educ. giving Non-educ. giving Non-educ. giving Non-educ. giving
Edulot 68.98 35.61 −25.38 22.69
(73.38) (47.70) (47.16) (32.15)
Observations 29,715 28,426 11017 11,017
Dataset COPPS COPPS GVS GVS
Model FE-Reg Tobit OLS Tobit
Robust standard errors in parentheses (clustered at state level in Models 1 and 3, individual level in Models 2 and 4). Columns
2 and 4 report marginal effect on the unconditional expected value of observed giving.
***p<.01, **p<.05, *p<.1.
Morgan’s (2000) theoretical result (and following experimental work) that lotteries are better at
providing public goods than voluntary contributions. If donors are (in their mind) simply shifting
to another way of contributing, this would not
truly be considered crowd-out. However, using
the Consumer Expenditure Survey, which allows
me to observe both lottery and charitable expenditures, I find that it is the individuals who do
not play the lottery who are reducing their donations. These results are reported and discussed in
greater detail in an online appendix.
To summarize, the introduction of an education lottery reduces donors’ contributions to education by 20–30%. This reduction appears to be
driven by changes on the intensive margin; the
lottery does not impact the probability that an
individual will make a contribution. Moreover,
there is evidence that the drop in giving might
be explained by (expected) government spending
crowding out private contributions, as opposed to
individuals sacrificing charitable contributions to
play the lottery. However, while the decrease in
giving seems to be a response to new government
funds, empirical results from recent literature on
crowd-out suggests that the crowd-out may be
a response by nonprofit organizations and not
donors. This issue is explored in the next section.
V. NONPROFIT ORGANIZATION RESPONSE TO
EDUCATION LOTTERY REVENUE
How does the introduction of an educationfunding lottery impact donations received by
education-related nonprofits? We have already
seen that an education lottery crowds out donations to education organizations, but it is possible that the result is driven by a reduction in
the effort of fundraisers—either because they
expect the marginal benefit of fundraising to be
lower or because they have benefitted directly
from lottery revenues and their level of need has
reduced. Andreoni and Payne (2003) document
that, in a more general setting, the crowd-out that
results from government grants to nonprofits can
be almost entirely explained by this “fundraising
crowd-out.” In some of their results that account
for fundraising, donors’ contributions are either
unaffected by or slightly increase with grants.
One explanation for the observed lack of “traditional crowd-out” in Andreoni and Payne’s
work (and, more generally, for the relatively
small degree of crowd-out typically observed in
response to government grants) is that individuals are largely unaware of government grants to
nonprofits. In the United States, the introduction
of a lottery to fund education tends to be highly
publicized and as such individuals are more likely
to be aware of this change in government funding. Thus, it may be reasonable to expect that
the crowd-out observed in the previous section
is in fact driven by donor preferences, consistent with classic models of charitable giving
and crowd-out.
A. Data and Empirical Approach
To examine whether this is the case, I next
turn to data on nonprofit organizations’ revenue
and expenses from federal tax returns spanning
from 1989 to 2007. The data are collected and
constructed by the IRS Statistics of Income
division, and then compiled and provided for
research purposes by the National Center for
Charitable Statistics (NCCS). Each year a subset
of tax returns from nonprofit organizations that
hold 501(c)(3) status are randomly sampled for
inclusion in the dataset, which reports a variety
of financial variables from their tax return (from
the year sampled) such as operation expenses,
charitable contributions received, fundraising expenses, etc. The dataset also includes
JONES: EDUCATION LOTTERIES AND DONATIONS 913
groupings of nonprofit organizations by function, categorizing organizations as Arts, Education, Health, Human Services, or Other.
A broad array of education-related organizations are represented in the data, including
colleges, universities, preschools, libraries, remedial reading organizations, etc. However, as noted
in the introduction, a vast majority of charitable
activity in the education subsector is directed
toward higher education. These data of course
include private nonprofit colleges and universities, but many public universities and colleges
are also represented: either because (1) they
officially hold 501(c)(3) status or (2) their
fundraising activities are accomplished
through an affiliated but independent nonprofit
foundation, both of which are common.
While the dataset is not constructed as a true
panel of nonprofit organizations, nonprofit organizations reappear in the data often enough that
it can be treated as panel (as Andreoni and
Payne do, for instance).16 Thus, I construct an
unbalanced panel where each observation is a
particular nonprofit organization in a particular year; there are typically (but not always)
gaps between a nonprofit organization’s appearances in the panel but these appearances are
randomly determined.
The goal of this section is to examine the donations received and the fundraising behavior of
nonprofit organizations in response to the introduction of a lottery. Thus, I restrict my sample to
organizations that receive donations at any point
in the panel. The resulting dataset consists of a
total of 192,478 observations and 19,505 unique
nonprofit organizations; 39,410 of these observations are education-related organizations.
Throughout this section, I use a fixed-effects
approach (with fixed-effects at the organization level) within the same DID framework
employed in previous sections. Two questions
are of primary interest: First, how does the
introduction of an education lottery impact the
amount of donations received by educationrelated organizations? This essentially tests
the robustness of the results from the previous
section, but with much richer data. Here, for
instance, we do not suffer from the censoring
at $0 that plagued the assessment of the donorlevel data. Second, are changes in donations
received by nonprofits driven by changes in
fundraising efforts?
16. The median organization in the dataset I use appears
seven times.
With these questions in mind, the primary outcome variables of interest is log of contributions received.
17 I regress contributions on the
“Edulot” indicator variable and, in all specifications, I include controls for the log of total revenue (excluding public support), the log of total
expenditures (excluding fundraising), and year
fixed effects. I additionally control for state-level
covariates which may impact donations: log of
income per capita, log of state population, log of
education expenditure per capita, and log of other
expenditures per capita. To address the impact
of fundraising, I then control for fundraising
expenditures. In doing so, I use an instrumental
variables approach to account for the endogeneity between fundraising and donations received,
using liabilities at the beginning of the fiscal year
as an instrument for fundraising.18
B. Results
Table 5 provides an initial assessment of the
impact that a lottery has on contributions received
by nonprofits. Columns 1 and 2 report the results
of fixed-effects estimations, with fixed-effects at
the organization level, for education organizations and non-education organizations, respectively. Consistent with the findings from the
previous section, the introduction of an education lottery reduces the contributions received
by education organizations—in this case, by an
estimated 8%—but has no significant impact on
contributions to other causes.19
17. Taking the log of contributions was not feasible in the
previous section given the large number of $0 donations. In
this section, where the organizations in the sample mostly all
receive some positive amount of donations, it is both feasible
and preferable; the log specification better handles outliers.
18. Andreoni and Payne (2011) use this instrument for
fundraising as well arguing that higher debt impacts the need
for fundraising in a way that is unrelated to the amount of
donations one expects to receive.
19. To link this result more closely to the existing literature on crowding-out of charitable giving, we would ideally
like to know the extent to which charitable giving decreases
as a function of the amount that government spends. Answering this question is difficult because there is very little actual
increase in spending, but we do know how much government
claims it will spend. That is, in the state government finance
data I observe “lottery proceeds,” which is the amount of
money remaining for the beneficiaries after accounting for
prizes awarded and administrative costs. Thus, we can estimate the continuous impact of treatment by adopting the same
specifications as before but replacing the “Edulot” dummy
with log of lottery proceeds in education-lottery states. Online
Appendix Table A2 reports the results of these estimations
for both education and non-education organizations. Based
on the instrumental variable specification which controls for
fundraising (Table A2, Panel B), a 10% increase in lottery
proceeds is associated with a 5.25% decrease in contributions
received by education-related organizations.
914 ECONOMIC INQUIRY
TABLE 5
The Impact of an Education Lottery on Contributions Received
(1) (2) (3) (4)
Variables
Contributions
received:
Educ. orgs.
Contributions
received:
Non-educ. orgs.
Contributions
received:
Educ. orgs.
Contributions
received:
Educ. orgs.
FE-Reg. FE-Reg. FE-Reg. FE-Reg.
Edulot −.0817*** .0119 −.0734*
(.0275) (.0282) (.0369)
Failed Edulot −.0247
(.0817)
Non-educ. lottery .0138
(.0363)
Other expenditures .339*** .282*** .340*** .339***
(.0374) (.0142) (.0373) (.0374)
Other revenues −.0747*** −.0659*** −.0745*** −.0747***
(.0216) (.00876) (.0217) (.0217)
State: income .850*** 1.053*** .783*** .851***
(.258) (.294) (.258) (.258)
State: population .433 .666*** .264 .425
(.270) (.198) (.241) (.280)
State: educ. exp. .0443 .0149 .0565 .0435
(.105) (.0889) (.108) (.104)
State: non-educ. exp. −.131 −.217* −.149 −.135
(.140) (.115) (.145) (.139)
Observations 38,585 129,267 38,585 38,585
R2 .263 .066 .227 .331
Robust standard errors (clustered at state level) in parentheses.
***p<.01, **p<.05, *p<.1.
Columns 3 and 4 offer two robustness tests.
Between 1989 and 2008, three states20 attempted
to introduce a lottery through referenda or ballot initiatives, but failed to achieve enough votes.
In Column 3, I replace the “Edulot” dummy
with a “Failed Edulot” dummy. If the treatment
effects here are merely picking up trends in giving that cause a state to introduce a lottery, the
coefficient on “Failed Edulot” should be negative and significant. While negative, the coefficient is substantially smaller in magnitude than
the result from Column 1 and is not significantly
different than zero. Column 4 adds a dummy to
indicate a non-education lottery to the main specification. No drop in giving to education organizations is observed when the lottery is not
intended to benefit education.
Is the decrease in contributions to education
organizations driven by a change in fundraising efforts? To answer this, I add a control
for fundraising to the preceding specification.
However, to account for potential endogeneity
between fundraising and donations, I do so in
an instrumental variables framework, taking
liabilities as an instrument for fundraising.
20. Oklahoma in 1995, Alabama in 2000, Arkansas
in 2001.
Results for education (Columns 1 and 2) and
non-education organizations (Columns 3 and
4) are presented in Table 6. Columns 1 and 3
report the first stage of the instrumental variables regression. Notably, the introduction of
an education lottery has very little impact on
education organizations’ fundraising expenditures (Column 1). Thus, in turning to the impact
of the lottery after accounting for fundraising
(Column 2), it is unsurprising to find that the
estimated decrease in giving is very close to the
estimate from Table 5.
Next, I report the results of an event study
approach to analyzing the data. This is done for
two reasons. First, the DID approach I rely on
throughout the paper requires that there is not
a difference in trends across treated and control states prior to treatment. The event study
approach explicitly tests this. Second, the dynamics of the crowd-out detected thus far is interesting in its own right. Does the lottery cause just a
brief drop in giving, or is there a more persistent
treatment effect?
To assess this, I adapt the two-stage least
squares approach reported in Table 6 and
employ an event study approach similar to that
of Jacobson, LaLonde, and Sullivan (1993)
JONES: EDUCATION LOTTERIES AND DONATIONS 915
TABLE 6
The Impact of an Education Lottery—Accounting for Fundraising
(1) Educ. orgs:
Fundraising
FE-Reg.
(first-stage)
(2) Educ. orgs:
Contributions
received
IV-FE-Reg.
(3) Educ. orgs:
Contributions
received
FE-Reg.
(4) Non-educ.
orgs:
Fundraising
FE-Reg.
(first-stage)
(5) Non-educ.
orgs:
Contributions
received
IV-FE-Reg.
(6) Non-educ.
orgs:
Contributions
received
FE-Reg.
Edulot −.00296 −.0689** −.0650** .0685 .00314 .0247
(.0324) (.0279) (.0296) (.0422) (.0412) (.0372)
Liabilities .0346** .0432***
(.0130) (.00644)
Fundraising .0614 .130*** .4851*** .176***
(.2656) (.0116) (.1273) (.00924)
Obs. 27,905 27,905 28,828 58,374 58,374 61,996
R2 .452 .155 .512 .245 .469 .363
Robust standard errors (clustered at state level) in parentheses.
Additional controls included as noted in text.
***p<.01, **p<.05, *p<.1.
and Kline (2012). Specifically, I now estimate
the following:
yist =α+ ∑
t
δtDst + βffundraising ̂ ist + 𝛃Xist
+ [
state FE′
s
]
s + [
year FE′
s
]
t
where, instead of the simple “treatment” dummy,
I now include a vector of dummies Dst that are
equal to one only when treatment is exactly t
periods away in state s. Given the small amount
of noise in defining when a state is “treated,”
one “period” in this specification is 2 years. So
that the results can be interpreted as the treatment effect of the lottery, one period prior to
the introduction of the lottery—that is, 0–2
years before the lottery—is the omitted category.
(More precisely, δ−1 =0.) If the parallel trends
assumption is satisfied δt =0 as long as t <0.
The remainder of the specification is no different
than the one used to generate results in Table 6;
“fundraisingist” is predicted in a first-stage using
the assets instrument.
The results are documented graphically in
Figure 2, which plots the estimated event study
coefficients, δt. The horizontal axis indicates
the period in question, while the vertical axis
reports the estimated percentage change in
giving. First, note that the parallel trends
assumption is satisfied; for all periods prior
to the introduction of a lottery, there is no statistical difference between treated and untreated
nonprofit organizations. After the lottery has
been introduced (starting with “0–2 years after
lottery”), donations fall. Donations decline by
4.5% in the first 2 years (p<.01), 10.4% in
the third and fourth years (p<.01), and 9% in
the fifth and sixth years (p<.10). In the long
run, the effect is dampened. Six-to-eight years
after treatment there is an estimated drop in
giving of 6%; there is an even smaller (and
insignificant) difference between treated and
untreated organizations more than 8 years
after treatment.
C. Variation in Crowd-Out by Type of Lottery
and Type of Organization
As noted in Section II, different states designate funds for different levels of education. Some
states that introduce an education lottery during the sample period aim to support both K-12
and higher education. Some aim to support only
higher education, usually creating a scholarship
fund. Two states (namely, Texas and Virginia)
support only K-12 education. In this subsection,
I allow each of these categories of treatment to
have their own effect. Similarly, some states generate funds for both public and private institutions
(usually in the form of scholarship funds) while
others are designated only for public institutions.
I also allow these two groups of states to have
different impacts. (I do not decompose the treatment effects in the main analysis of the paper as,
in some cases, there are very few states that fit
into these categories.)
Table 7, Column 1 reports the results of
extending the basic specification to allow for
three different types of treatments: lotteries that
are intended to support both K-12 and higher
education, lotteries that are intended to support K-12 education only, and lotteries that are
916 ECONOMIC INQUIRY
FIGURE 2
The Impact of an Education Lottery—Event Study Approach
-0.15
-0.1
-0.05
0
0.05
0.1
> 6 years
before
6-4 years
before
4-2 years
before
2-0 years
before
0-2 years
after
2-4 years
after
4-6 years
after
6-8 years
after
> 8 years
after
Pct. change in donations received
intended to support higher education only. These
results suggest that crowd-out is most prominent
in states that designate the lottery either for K-12
only or higher education only.
Of course, we may expect the impact of these
three separate treatments to vary for different
types of nonprofit organizations represented in
the data. For instance, a higher education organization should be impacted in states where lottery
funds are used for higher education, but not in
states where lottery funds are used for K-12 education only. This notion is explored in the remaining columns of Table 7.
Specifically, in each remaining column,
I restrict the sample to a particular type of
organization: K-12 organizations in Column 2,
Higher education organizations in Column 3, and
“Fundraising/support organizations” in Columns
4–6. This final category requires some additional
explanation: in many cases, the organization that
raises funds for a university (public or private)
registers as its own separate nonprofit organization. These organizations appear in the data
as “Fundraising/support organizations.”21 Also
21. For example, the University of South Carolina’s (a
public university) “Development Foundation” and “Educational Foundation” both appear in the data and are categorized
as “fundraising/support organizations.”
captured in this category are any other organizations “that raise and distribute funds for multiple
organizations within the Education major group
area” or “Organizations existing as a support and
fund-raising entity for a single institution.” This
therefore also includes organizations such as:
Library Friends Groups, School Athletic Booster
Clubs, School Booster, United Negro College
Fund, and Funding for School Districts. Column
4 includes all such organizations. A separate
variable in the data further categorizes these
organizations as being related to K-12 or higher
education, but this coding is only available for a
fraction of the data and therefore cannot be used
on its own to make sample restrictions. Thus, in
attempting to split the sample further, I simply
drop organizations that are clearly designated
as being related to K-12 in Column 5 (therefore
leaving behind mostly higher education related
organizations) and drop higher education-related
organizations in Column 6 (leaving behind
mostly K-12 organizations).
Columns 3 and 5 document declining contributions to higher education organizations
and higher education fundraisers (respectively)
in states where higher education is the only
beneficiary. This lends credence to the argument
that declining giving is indeed crowd-out rather
than a more general drop in giving to education.
JONES: EDUCATION LOTTERIES AND DONATIONS 917
TABLE 7
The Treatment Effects of Different Types of Lotteries on Different Types of Nonprofit
Organizations—K-12 Lotteries Versus Higher Education Lotteries
(1) (2) (3)
Variables ln(Donations received) ln(Donations received) ln(Donations received)
Sample Full K-12 education organizations Higher education organizations
Treated—K-12 and higher ed. −.0313 .742 −.0112
(.0405) (10.26) (.0340)
Treated—K-12 only −.0907*** −.602 −.0740
(.0330) (6.457) (.0488)
Treated—Higher ed. only −.0939* –2.177 −.109**
(.0545) (27.12) (.0440)
Observations 27,905 7,696 14,335
(4) (5) (6)
Variables ln(Donations received) ln(Donations received) ln(Donations received)
Sample Fundraising/support
organizations
Fundraising/support
organizations (excluding
K-12 support)
Fundraising/support
organizations (excluding
higher ed. support)
Treated—K-12 and higher ed. .0152 .0801 .203
(.0924) (.0665) (.175)
Treated—K-12 only −.122 −.0782 −.0785
(.121) (.120) (.106)
Treated—Higher ed. only −.291 −.425** .0755
(.221) (.193) (.230)
Observations 4,129 3,972 2,393
Robust standard errors in parentheses.
***p<.01, **p<.05, *p<.1.
It may also be unsurprising to find that higher
education organizations suffer more when a
lottery is introduced as higher education organizations do receive a larger share of donations to
start with.
Interestingly, there is little or no evidence that
higher education organizations are impacted in
states where all levels of education benefit from
the lottery. This could arguably be consistent
with the salience argument: it is only in states
where advertisements can focus on the fact that
the money goes toward higher education that
higher education organizations suffer. In states
where advertising only makes broad claims about
“benefitting education,” there is less crowd-out.
We might also expect differences across states
that do and do not generate funds for private
institutions. In particular, as noted in Section
II, roughly 55% of higher education donations
are directed at private institutions. Thus, we
might expect more crowd-out in states that generate funding for both public and private institutions. Table 8 shows that this is the case.
Crowd-out of donations, especially to higher
education organizations, is most prominent in
states that generate funds for both public and
private institutions.
D. Salience of Government Activity as an
Explanation for Crowd-Out?
Consistent with the findings from the previous section, contributions to education-related
organizations fall after the introduction of an education lottery, which is not true of contributions
to other organizations. However, we can now say
that this result appears to be driven by donors’
decisions to reduce their contributions as opposed
to reduced fundraising efforts. This result differs
from a recent literature that demonstrates that
crowd-out is often largely explained by a change
in nonprofits’ fundraising behavior (Andreoni
and Payne 2003, 2011; Heutel 2009; Hughes,
Luksetich, and Rooney 2014; Monti 2010). I have
suggested that an important difference between
state lotteries and other forms of government
spending is the high level of publicity that
lotteries receive. Relative to government
grants to nonprofits, donors are likely to be
more aware of government spending resulting from lotteries—and therefore more likely to
respond—in large part because states themselves
heavily advertise the recipient of lottery revenues.
Is there more direct evidence to support
this suggestion? I take two approaches to
answer this question. First, if the crowd-out
918 ECONOMIC INQUIRY
TABLE 8
The Treatment Effects of Different Types of Lotteries on Different Types of Nonprofit
Organizations—Public Education Lotteries Versus Public and Private Education Lotteries
(1) (2) (3)
Variables ln(Donations received) ln(Donations received) ln(Donations received)
Sample Full K-12 education organizations Higher education organizations
Treated—Public and private −.101*** –1.888 −.0869***
(.0319) (24.18) (.0291)
Treated—Public only −.0477 .162 −.0358
(.0377) (3.971) (.0376)
Observations 27,905 7,696 14,335
(4) (5) (6)
Variables ln(Donations received) ln(Donations received) ln(Donations received)
Sample Fundraising/support
organizations
Fundraising/support
organizations (excluding
K-12 support)
Fundraising/support
organizations (excluding
higher ed. support)
Treated—Public and private −.160 −.177 .0927
(.164) (.168) (.140)
Treated—Public only −.0503 −.0206 .0604
(.124) (.124) (.135)
Observations 3,914 3,834 2,210
Robust standard errors in parentheses.
***p<.01, **p<.05, *p<.1.
observed in this paper is indeed driven by
donors’ awareness of government activity and if
this awareness is (at least in part) the result of
government advertising, then we would expect
the magnitude of the crowd-out to increase with
governments’ advertising activities. The Census
Bureau’s Survey of Government Finances reports
states’ yearly lottery administrative costs, which
includes advertising expenditures.22 Advertising expenditures are not reported, so I use the
ratio of administrative costs to ticket sales as
a proxy for advertising. In addition to advertising, administrative costs include the cost of
printing and distributing tickets which obviously
varies with the number of tickets sold, so most
of the variation in administrative costs after
accounting for tickets sales presumably comes
from advertising.23
I extend the previous empirical specifications (FE and FE-IV-Regressions controlling
for fundraising) to include controls for the ratio
22. According to the Census Bureau, administrative costs
“includes salaries of officials as well as advertising, supplies,
and the like.”
23. In an online appendix, I assess the strength of this
ratio as a proxy for advertising cost. I obtain data on actual
advertising budgets from one year from LaFleur’s World
Lottery Almanac; I find a strong positive relationship between
my constructed proxy and actual advertising budgets from
that year.
of administrative costs to ticket sales (“Advertising”) and the interaction of “Advertising”
with “Edulot.” In doing so, I re-center “Advertising” around its mean so that the main effect
of “Edulot” can be interpreted as the impact
of an education lottery evaluated at the mean
level of advertising. If crowd-out is increasing in
advertising we would expect the coefficient on
“Edulot X Advertising” to be negative.
This is indeed the case, as can be seen in
Columns 1 and 3 of Table 9 which report the
results of these estimations for education organizations. On the basis of Column 3, an education
lottery is associated with a 6% decrease in
contributions received by education organizations. For each additional cent of ticket sales
that a state devotes to administrative costs, contributions decrease by an additional 1%. The
same significant relationship does not hold for
non-education organizations (Columns 2 and 4).
A second approach allows for the possibility
that the political method of introducing the lottery impacts crowd-out. In particular, seven of
the 12 states that introduced an education lottery
between 1989 and 2008 did so through referenda
or ballot initiatives.24 The remaining states introduced their lottery through legislative action. One
might expect that citizens are more aware of the
24. These states are Georgia, Missouri, Virginia, Washington, South Carolina, Tennessee, and Oklahoma.
JONES: EDUCATION LOTTERIES AND DONATIONS 919
TABLE 9
Crowd-Out and Awareness of Government Spending—Proxy for Advertising Expenditures
(1) (2) (3) (4)
Variables
Contributions
received:
Educ. orgs.
(FE-Reg.)
Contributions
received:
Non-educ. orgs.
(FE-Reg.)
Contributions
received:
Educ. orgs.
(FE-IV-Reg.)
Contributions
received:
Non-educ. orgs.
(FE-IV-Reg.)
Edulot −.0686** .0263 −.0594** .0154
(.0286) (.0280) (.0256) (.0393)
Edulot X Advertising –1.274*** −.589 −.988** −.247
(.387) (.485) (.466) (.488)
Advertising .655*** .00285 .556*** −.243
(.200) (.270) (.197) (.221)
Observations 38,585 129,267 27,905 58,374
R2 .294 .068 .156 .000
Robust standard errors (clustered at state level) in parentheses.
***p<.01, **p<.05, *p<.1.
TABLE 10
Crowd-Out and Awareness of Government Spending—Political Method of Lottery Introduction
(1) (2) (3) (4)
Variables
Contributions
received:
Educ. orgs.
(FE-Reg.)
Contributions
received:
Non-educ. orgs.
(FE-Reg.)
Contributions
received:
Educ. orgs.
(FE-IV-Reg.)
Contributions
received:
Non-educ. orgs.
(FE-IV-Reg.)
Edulot (Legislative) −.0627* −.00842 −.0413 −.00215
(.0329) (.0296) (.0473) (.0468)
Edulot (Direct vote) −.0921** .0261 −.0842*** .00658
(.0355) (.0374) (.0275) (.0562)
Observations 38,585 129,267 27,905 58,374
R2 .266 .066 .155 .000
Robust standard errors (clustered at state level) in parentheses.
***p<.01, **p<.05, *p<.1.
lottery and its beneficiary when they vote directly
on the issue. Thus, if salience is important to
crowd-out, there should be more crowd-out in
states that introduced their lotteries through direct
voting (referenda/ballot initiatives).
I test whether this is the case in Table 10,
which includes a separate treatment dummy
for legislative action and direct vote states.
Columns 1 and 3 report the results of these estimations for education organizations. Crowd-out
is indeed higher in Direct vote states. The same
relationship is not observed for non-education
organizations (Columns 2 and 4).
Of course, these results should be taken as
merely suggestive: we cannot directly observe
advertising expenditures, nor do we know that
donors are more aware of the lottery beneficiary in “Direct voting” states. However, the
results are consistent with the suggestion that
a higher level of awareness of government
activity leads to more crowd-out. This may help
explain why crowd-out is driven by donors when
the source of funding is a state lottery, while
crowd-out is driven mostly by nonprofits when
the source of funding is much-less-publicized
government grants.
The reaction to government activity only when
it is salient is not without precedent in the literature. These results relate and contribute to
the recent literature on salience and taxation.
In both experimental and observational data,
Chetty, Looney, and Kroft (2009) find that consumers adjust their buying decisions in the face
of higher taxes only when taxes are clearly highlighted; for instance, posting “after-tax” price
next to items in a grocery store dramatically
decreased demand despite evidence that consumers knew and could calculate after-tax prices.
Similar effects have been documented in other
settings (Cabral and Hoxby 2012; Finkelstein
2009; Goldin and Homonoff 2013). The results
920 ECONOMIC INQUIRY
reported in this paper highlight a different context where decision-makers react to government
activity in the way that theory might predict, but
only when the activity is highly salient.
VI. CONCLUSION
In this paper, I assess the impact that
education-funding state lotteries have on
donations to education. I find that charitable
contributions to education significantly decrease
after the introduction of an education lottery;
contributions received by education-related nonprofit organizations drop by 8% with a lottery.
There is evidence to suggest that this drop is
driven by a crowding-out of donations, consistent with classic models of voluntary public
good provision. In particular, I am able to rule
out alternative explanations that might suggest
that individuals are merely shifting charitable
expenditures to lottery expenditures.
Additionally, unlike recent work that finds that
crowd-out stemming from grants to nonprofits
is often mostly explained by nonprofit fundraising behavior, here the effect is almost entirely
driven by donors. I argue that this is because of
the high level of publicity that lotteries and their
intended beneficiaries receive. Consistent with
this suggestion, I show that crowd-out is increasing in a measure of state advertising activity.
Also, crowd-out is higher for states that introduce
a lottery through referenda instead of legislative
action, which is presumably less salient to citizens. Though the potential importance of salience
as a determinant of charitable crowd-out has been
discussed in recent work by Monti (2010), to my
knowledge this is the first paper to provide empirical evidence that crowd-out is indeed increasing
in awareness of government activity.
There are of course a variety of policyoriented reasons why some oppose statesponsored lotteries; for instance, it has been
repeatedly shown that, as a tax, lotteries are
highly regressive. This paper highlights an additional trade-off that states face in implementing a
lottery as a way to fund public goods. While some
existing work shows that earmarking for a “good
cause” increases a lottery’s revenue (Landry and
Price 2007), I find that this comes at a price:
private, voluntary support for the cause falls.
However, the fact that state governments are
vocal about the particular cause being supported
(education) seems to be critical to this result.
This suggests that a government that is vocal
about supporting “good causes,” but does not
support or highlight any one cause in particular,
may enjoy the benefits of higher revenue without
disrupting charitable activity. The UK National
Lottery operates in this manner, advertising that
the Lottery supports “380,000 … good causes
… across the UK.”25 Indeed, in an analysis of
UK charities that have received lottery grants,
Andreoni, Payne, and Smith (2014) find no
evidence of charitable crowd-out.26
REFERENCES
Alm, J., M. McKee, and M. Skidmore. “Fiscal Pressure, Tax
Competition, and the Introduction of State Lotteries.”
National Tax Journal, 46, 1993, 463–76.
Andreoni, J. “Giving with Impure Altruism: Applications to
Charity and Ricardian Equivalence.” Journal of Political Economy, 97, 1989, 1447–58.
Andreoni, J. “Philanthropy,” in Handbook on the Economics
of Giving, Reciprocity and Altruism, Vol. 2, edited by
S.-C. Kolm and J. M. Ythier. Amersterdam, The Netherlands: North Holland, 2006, 1201–69.
Andreoni, J., and A. A. Payne. “Do Government Grants to
Private Charities Crowd Out Giving or Fund-Raising?”
American Economic Review, 93(3), 2003, 792–812.
Andreoni, J., and A. A. Payne. “Is Crowding Out Due Entirely
to Fundraising? Evidence from a Panel of Charities.”
Journal of Public Economics, 95, 2011, 334–43.
Andreoni, J., A. A. Payne, and S. Smith. “Do Grants to
Charities Crowd Out Other Income? Evidence from the
UK.” Journal of Public Economics, 114, 2014, 75–86.
Apinunmahakul, A., and R. A. Devlin. “Charitable Giving
and Charitable Gambling: An Empirical Investigation.”
National Tax Journal, 57, 2004, 67–88.
Bergstrom, T., L. Blume, and H. Varian. “On the Private Provision of Public Goods.” Journal of Public Economics,
29(1), 1986, 25–49.
Borg, M. O., P. M. Mason, and S. L. Shapiro. The Economic
Consequences of State Lotteries. Westport, CT: Greenwood Publishing Group, 1991.
Cabral, M., and C. Hoxby. “The Hated Property Tax:
Salience, Tax Rates, and Tax Revolts.” No. w18514.
National Bureau of Economic Research, 2012.
Chetty, R., A. Looney, and K. Kroft. “Salience and Taxation:
Theory and Evidence.” American Economic Review,
99(4), 2009, 1145–77.
Clotfelter, C. T., and P. J. Cook. “On the Economics of
State Lotteries.” Journal of Economic Perspectives,
4(4), 1990, 105–19.
Clotfelter, C. T., and P. J. Cook, Selling Hope: State Lotteries
in America. Cambridge, MA: Harvard University Press,
1991.
Coughlin, C. C., T. A. Garrett, and R. Hernández-Murillo.
“The Geography, Economics, and Politics of Lottery
Adoption.” Review: Federal Reserve Bank of Saint
Louis, 88(3), 2006, 165.
Erekson, O. H., K. M. DeShano, G. Platt, and A. L. Ziegert.
“Fungibility of Lottery Revenues and Support of Public
Education.” Journal of Education Finance, 28, 2002,
301–11.
Evans, W. N., and P. Zhang. “The Impact of Earmarked
Lottery Revenue on K-12 Educational Expenditures.”
Education Finance and Policy, 2(1), 2007, 40–73.
25. http://www.national-lottery.co.uk/player/p/good
causesandwinners.ftl
26. In fact, for small organizations, they find evidence
of crowd-in.
JONES: EDUCATION LOTTERIES AND DONATIONS 921
Finkelstein, A. “E-ztax: Tax Salience and Tax Rates.” Quarterly Journal of Economics, 124(3), 2009, 969–1010.
Garrett, T. A. “Earmarked Lottery Revenues for Education: A
New Test of Fungibility.” Journal of Education Finance,
26, 2001, 219–38.
Garrett, T., and R. Rhine. “Government Growth and Private
Contributions to Charity.” Public Choice, 143(1), 2010,
103–20.
Goldin, J., and T. Homonoff. “Smoke Gets in Your Eyes:
Cigarette Tax Salience and Regressivity.” American
Economic Journal: Economic Policy, 5(1), 2013,
302–36.
Heutel, G. “Crowding Out and Crowding In of Private
Donations and Government Grants.” Technical Report,
National Bureau of Economic Research, 2009.
Hughes, P., Luksetich, W., and Rooney, P. “Crowding Out
and Fundraising Efforts.” Nonprofit Management and
Leadership, 2014.
Jacobson, L. S., R. J. LaLonde, and D. G. Sullivan. “Earnings Losses of Displaced Workers.” American Economic Review, 83, 1993, 685–709.
Kearney, M. S. “State Lotteries and Consumer Behavior.”
Journal of Public Economics, 89(11), 2005, 2269–99.
Kline, P. “The Impact of Juvenile Curfew Laws on Arrests
of Youth and Adults.” American Law and Economics
Review, 14(1), 2012, 44–67.
Landry, C. E., and M. K. Price. “Earmarking Lottery Proceeds for Public Goods: Empirical Evidence from US
Lotto Expenditures.” Economics Letters, 95(3), 2007,
451–5.
Lange, A., J. A. List, and M. K. Price. “Using Lotteries
to Finance Public Goods: Theory and Experimental
Evidence.” International Economic Review, 48(3),
2007, 901–27.
Monti, H. “Environmental Policy and Giving: Does Government Spending Affect Charitable Donations?” Working
Paper, 2010.
Morgan, J. “Financing Public Goods by Means of Lotteries.”
Review of Economic Studies, 67(4), 2000, 761–84.
Morgan, J., and M. Sefton. “Funding Public Goods with
Lotteries: Experimental Evidence.” Review of Economic Studies, 67(4), 2000, 785–810.
Novarro, N. K. “Earmarked Lottery Profits: A Good Bet for
Education Finance?” Journal of Education Finance, 31,
2005, 23–44.
Spindler, C. J. “The Lottery and Education: Robbing Peter to
Pay Paul?” Public Budgeting & Finance, 15(3), 1995,
54–62.
Vesterlund, L. “Why Do People Give?,” in The Nonprofit
Sector: A Research Handbook, edited by W. W. Powell
and R. Steinberg. New Haven, CT: Yale University
Press, 2006, 568–90.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article:
APPENDIX S1. Online Appendices: Data Notes and
Additional Results.
Copyright of Economic Inquiry is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder’s
express written permission. However, users may print, download, or email articles for
individual use.

---

Get Professional Assignment Help Cheaply

Are you busy and do not have time to handle your assignment? Are you scared that your paper will not make the grade? Do you have responsibilities that may hinder you from turning in your assignment on time? Are you tired and can barely handle your assignment? Are your grades inconsistent?

Whichever your reason is, it is valid! You can get professional academic help from our service at affordable rates. We have a team of professional academic writers who can handle all your assignments.

Why Choose Our Academic Writing Service?

• Plagiarism free papers
• Timely delivery
• Any deadline
• Skilled, Experienced Native English Writers
• Subject-relevant academic writer
• Adherence to paper instructions
• Ability to tackle bulk assignments
• Reasonable prices
• 24/7 Customer Support
• Get superb grades consistently

Online Academic Help With Different Subjects

Literature

Students barely have time to read. We got you! Have your literature essay or book review written without having the hassle of reading the book. You can get your literature paper custom-written for you by our literature specialists.

Finance

Do you struggle with finance? No need to torture yourself if finance is not your cup of tea. You can order your finance paper from our academic writing service and get 100% original work from competent finance experts.

Computer science

Computer science is a tough subject. Fortunately, our computer science experts are up to the match. No need to stress and have sleepless nights. Our academic writers will tackle all your computer science assignments and deliver them on time. Let us handle all your python, java, ruby, JavaScript, php , C+ assignments!

Psychology

While psychology may be an interesting subject, you may lack sufficient time to handle your assignments. Don’t despair; by using our academic writing service, you can be assured of perfect grades. Moreover, your grades will be consistent.

Engineering

Engineering is quite a demanding subject. Students face a lot of pressure and barely have enough time to do what they love to do. Our academic writing service got you covered! Our engineering specialists follow the paper instructions and ensure timely delivery of the paper.

Nursing

In the nursing course, you may have difficulties with literature reviews, annotated bibliographies, critical essays, and other assignments. Our nursing assignment writers will offer you professional nursing paper help at low prices.

Sociology

Truth be told, sociology papers can be quite exhausting. Our academic writing service relieves you of fatigue, pressure, and stress. You can relax and have peace of mind as our academic writers handle your sociology assignment.

Business

We take pride in having some of the best business writers in the industry. Our business writers have a lot of experience in the field. They are reliable, and you can be assured of a high-grade paper. They are able to handle business papers of any subject, length, deadline, and difficulty!

Statistics

We boast of having some of the most experienced statistics experts in the industry. Our statistics experts have diverse skills, expertise, and knowledge to handle any kind of assignment. They have access to all kinds of software to get your assignment done.

Law

Writing a law essay may prove to be an insurmountable obstacle, especially when you need to know the peculiarities of the legislative framework. Take advantage of our top-notch law specialists and get superb grades and 100% satisfaction.

What discipline/subjects do you deal in?

We have highlighted some of the most popular subjects we handle above. Those are just a tip of the iceberg. We deal in all academic disciplines since our writers are as diverse. They have been drawn from across all disciplines, and orders are assigned to those writers believed to be the best in the field. In a nutshell, there is no task we cannot handle; all you need to do is place your order with us. As long as your instructions are clear, just trust we shall deliver irrespective of the discipline.

Are your writers competent enough to handle my paper?

Our essay writers are graduates with bachelor's, masters, Ph.D., and doctorate degrees in various subjects. The minimum requirement to be an essay writer with our essay writing service is to have a college degree. All our academic writers have a minimum of two years of academic writing. We have a stringent recruitment process to ensure that we get only the most competent essay writers in the industry. We also ensure that the writers are handsomely compensated for their value. The majority of our writers are native English speakers. As such, the fluency of language and grammar is impeccable.

What if I don’t like the paper?

There is a very low likelihood that you won’t like the paper.

Reasons being:

• When assigning your order, we match the paper’s discipline with the writer’s field/specialization. Since all our writers are graduates, we match the paper’s subject with the field the writer studied. For instance, if it’s a nursing paper, only a nursing graduate and writer will handle it. Furthermore, all our writers have academic writing experience and top-notch research skills.
• We have a quality assurance that reviews the paper before it gets to you. As such, we ensure that you get a paper that meets the required standard and will most definitely make the grade.

In the event that you don’t like your paper:

• The writer will revise the paper up to your pleasing. You have unlimited revisions. You simply need to highlight what specifically you don’t like about the paper, and the writer will make the amendments. The paper will be revised until you are satisfied. Revisions are free of charge
• We will have a different writer write the paper from scratch.
• Last resort, if the above does not work, we will refund your money.

Will the professor find out I didn’t write the paper myself?

Not at all. All papers are written from scratch. There is no way your tutor or instructor will realize that you did not write the paper yourself. In fact, we recommend using our assignment help services for consistent results.

What if the paper is plagiarized?

We check all papers for plagiarism before we submit them. We use powerful plagiarism checking software such as SafeAssign, LopesWrite, and Turnitin. We also upload the plagiarism report so that you can review it. We understand that plagiarism is academic suicide. We would not take the risk of submitting plagiarized work and jeopardize your academic journey. Furthermore, we do not sell or use prewritten papers, and each paper is written from scratch.

When will I get my paper?

You determine when you get the paper by setting the deadline when placing the order. All papers are delivered within the deadline. We are well aware that we operate in a time-sensitive industry. As such, we have laid out strategies to ensure that the client receives the paper on time and they never miss the deadline. We understand that papers that are submitted late have some points deducted. We do not want you to miss any points due to late submission. We work on beating deadlines by huge margins in order to ensure that you have ample time to review the paper before you submit it.

Will anyone find out that I used your services?

We have a privacy and confidentiality policy that guides our work. We NEVER share any customer information with third parties. Noone will ever know that you used our assignment help services. It’s only between you and us. We are bound by our policies to protect the customer’s identity and information. All your information, such as your names, phone number, email, order information, and so on, are protected. We have robust security systems that ensure that your data is protected. Hacking our systems is close to impossible, and it has never happened.

How our Assignment Help Service Works

1. Place an order

You fill all the paper instructions in the order form. Make sure you include all the helpful materials so that our academic writers can deliver the perfect paper. It will also help to eliminate unnecessary revisions.

2. Pay for the order

Proceed to pay for the paper so that it can be assigned to one of our expert academic writers. The paper subject is matched with the writer’s area of specialization.

3. Track the progress

You communicate with the writer and know about the progress of the paper. The client can ask the writer for drafts of the paper. The client can upload extra material and include additional instructions from the lecturer. Receive a paper.

4. Download the paper

The paper is sent to your email and uploaded to your personal account. You also get a plagiarism report attached to your paper.

GET A PERFECT SCORE!!!

---