charity ratings 1 camp resource june 2010 laura ellyn grant voluntary contributions and watchdog...
Post on 21-Dec-2015
216 views
TRANSCRIPT
Charity Ratings1
Camp ResourceJune 2010 Laura Ellyn Grant
Voluntary Contributions and Watchdog Ratings: Introduction and Signaling Effects
Laura Ellyn Grant
University of California, Santa Barbara
Charity Ratings2
Camp ResourceJune 2010 Laura Ellyn Grant
Question & Scope
• What are the effects of ratings, ranging from 0- to 4-stars in this study, in changing donations to charities? What do the responses indicate about donor behavior?
• Motivation:– Identifying the extent of and response to missing information – Providing the ratings (financial metrics) publicly likely changes donations– Allowing donors to contribute strategically to orgs with better outcomes
• Nearly $400 BILLION in donations in 2008; $3 TRILLION in revenue• Recession implies a tight spot for philanthropy • Need to know charities are reputable, putting $$ to the best use.
– Outcomes will be relative to expectations and elasticities– Differentiate INTRODUCTION of ratings from SIGNALING by ratings
Charity Ratings3
Camp ResourceJune 2010 Laura Ellyn Grant
• Voluntary– Health Marketing: low-fat, natural– Eco-labeling: organic, marine stewardship– Social Responsibility: fair-trade
• Government Policies– Education/School Performance– Nutrition and Content Labels– Hospital Performance– Restaurant Hygiene Report Cards
• Third party– Media Campaigns: Fox News and Republican membership– Ratings Organizations: Morningstar, Moody’s, Standard and Poor’s
Broader Literature: Information Disclosure
Charity Ratings4
Camp ResourceJune 2010 Laura Ellyn Grant
Approach• Conceptual Framework:
– Expected demand for charity– Effects of information in the form of ratings
• Data: – Charity Navigator (CN), complete data from the largest third-party
evaluator; Ratings from 0- to 4-stars– 8-years of longitudinal ratings data on more than 5400 large charities– 8 additional years of previous tax data from IRS source– Total observations: 60,000+
• Econometrics: – Introduction Effect: Before-after, with-in charity effects– Signaling Effect: With-in charity effects of published levels/change in ratings– Heterogeneity by sector & size
Charity Ratings5
Camp ResourceJune 2010 Laura Ellyn Grant
Demand for Public Goods
Standard charitable giving model
max Ui(xi, G, gi) subject to xi + p*gi = i
potential donor with utility over a private good, a public good, and private benefits of giving to the public good
Also called impure public goods [Andreoni (1990), Cornes and Sandler (1996), Kotchen (2005, 2006)]
Can likely omit public good aspect: U(x, g)–Anonymous gifts to large charities likely independent of others donations–Effect of information likely acts on private benefits with no immediate consequence to public supply–Solve for demand/marginal benefit of giving: g(p, )
xi , gi
Charity Ratings6
Camp ResourceJune 2010 Laura Ellyn Grant
MC0
Introduction Effect: Expected Demand, Missing InformationElastic Inelastic
Higher Expectations
Lower Expectations
Q
$
0
MB
MC0MB
MChighRating MC0
MB
MChighRating
Q
MC0
MClowRating
$
MBMClowRating
Charity Ratings7
Camp ResourceJune 2010 Laura Ellyn Grant
Response to Introduction• Cannot observe original ‘expected equilibrium’, g0
*
• Define 1
• is non-zero & defined given information has value & g0 0. • How consumers react to information will depend on BOTH expectations and
elasticities. A priori, sign of the effect is ambiguous:
gR*
g0*
Elastic Inelastic
Higher Expectations
Lower Expectations
DECREASE
( – )
DECREASE
( – )
INCREASE
( + )
INCREASE
( + )
Charity Ratings8
Camp ResourceJune 2010 Laura Ellyn Grant
Signaling Effect: Expected Demand, Changing InformationElastic Inelastic
Q
$
0
MB
MB
MChighRating
Q
MClowRating
Suppose that the empirically found sign of is negative, these two cases remain
The intuitive outcome is that higher star-rating yields more donations, but effect is also unknown, a priori
Can now measure response to changes in rating, from low to high, to deduce which case is correct
MClowRating
MChighRating
0
Charity Ratings9
Camp ResourceJune 2010 Laura Ellyn Grant
Tax Data• Public Charity designated by US law 501(c)(3)• Tax exempt but must file IRS form 990 if receipts
exceed $25000• 280,000 filed in 2008• Hundreds of fields on the tax form• Publically available• Estimated 1 million charities rated• Tax forms are complex, confusing, and incomparable
Charity Ratings10
Camp ResourceJune 2010 Laura Ellyn Grant
Charity Ratings11
Camp ResourceJune 2010 Laura Ellyn Grant
CN Website
• Launched in 2002• Online only, over 5500 charities with $10bil/yr contributions• Can search for charities by name, location, attributes• “Guide to intelligent giving,” evaluating the financial health
of each of the charities. • Third-party: not paid by charities, charities cannot opt-in or
out• 0- to 4-Stars rank from ‘exceptionally poor’ to ‘exceptional’
Charity Ratings12
Camp ResourceJune 2010 Laura Ellyn Grant
CN Website
Charity Ratings13
Camp ResourceJune 2010 Laura Ellyn Grant
Seeking Information
0
0.5
1
1.5
2
2.5
3
3.5
4
01/0
4/04
03/1
4/04
05/2
3/04
08/0
1/04
10/1
0/04
12/1
9/04
02/2
7/05
05/0
8/05
07/1
7/05
09/2
5/05
12/0
4/05
02/1
2/06
04/2
3/06
07/0
2/06
09/1
0/06
11/1
9/06
01/2
8/07
04/0
8/07
06/1
7/07
08/2
6/07
11/0
4/07
01/1
3/08
03/2
3/08
06/0
1/08
08/1
0/08
10/1
9/08
12/2
8/08
03/0
8/09
05/1
7/09
07/2
6/09
10/0
4/09
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
South Asia
Tsunami Dec 2004
Hurricane Katrina
Aug 2005Christmas
2006
Thanksgiving 2007
Christmas 2008
Christmas 2007
Christmas 2005
Christmas 2004
Sichuan EarthquakeMay 2008
Kashmir Earthquake Oct 2005
CN President appears on
O'Reilly Factor
May 2005
Charity Ratings14
Camp ResourceJune 2010 Laura Ellyn Grant
Ratings CalculationsExpense data
Trend data
program expenses = programmatic costs/total func exps
administrative expenses = admin costs/total func exps
fundraising expenses = fund costs/total func exps
fundraising efficiency = fund costs/contributions
program exp growth = (prog_t2/prog_t1 - 1)
revenue growth
= (rev_t2/rev_t1 - 1)
capital ratio = net assets/total exps
Charity Ratings15
Camp ResourceJune 2010 Laura Ellyn Grant
Ratings CalculationsExpense data
Trend data
program expenses = programmatic costs/total func exps
administrative expenses = admin costs/total func exps
fundraising expenses = fund costs/total func exps
fundraising efficiency = fund costs/contributions
program exp growth = (prog_t2/prog_t1 - 1)
revenue growth
= (rev_t2/rev_t1 - 1)
capital ratio = net assets/total exps
Convert all raw scores to a scale of 0 to 10
Continuous re-scaling or thresholds
Charity Ratings16
Camp ResourceJune 2010 Laura Ellyn Grant
Ratings CalculationsExpense data
Trend data
program expenses = programmatic costs/total func exps
administrative expenses = admin costs/total func exps
fundraising expenses = fund costs/total func exps
fundraising efficiency = fund costs/contributions
program exp growth = (prog_t2/prog_t1 - 1)
revenue growth
= (rev_t2/rev_t1 - 1)
capital ratio = net assets/total exps
efficiency rating
= 0 - 40, scaled to 0- to 4-stars
capacity rating
= 0 - 30, scaled to 0- to 4-stars
overall rating
= efficiency rating + capacity rating
= 0 - 70, scaled to 0- to 4-stars
Charity Ratings17
Camp ResourceJune 2010 Laura Ellyn Grant
Introduction Effect: Preliminary Specification
• Publication Signal, before and after, with-in Charity (i), flexible time trend (t)– Treatment: Observerd publication, charities added over time– Control: Same charities, unpublished scores
• Append historical data and calculate ratings using aforementioned process
• Provides a with-in charity counterfactual/falsification
ln_contit = *Star0 + *Star1 + *Star2 + *Star3 + *Star4 +
+ K*StarK*Observed + *scoreit + f(Fundit, Prog_Serveit, Assetit, Liabsit) + t +i + it
K*StarK
Charity Ratings18
Camp ResourceJune 2010 Laura Ellyn Grant
Comparing calculated to true scores
0.0
1.0
2.0
3.0
4
0 20 40 60 80 0 20 40 60 80
0 1D
ensi
ty
final_overall_sGraphs by ratedUnpublished/Calculated Published/True
Thresholds of ratings: Star0 = 0-24.9, Star1 = 25-39.9, Star2 = 40-49.9, Star3 = 50-59.9, Star4 = 60-70
Charity Ratings19
Camp ResourceJune 2010 Laura Ellyn Grant
(1) (2) (3) (4) ln_contributions ln_contributions ln_contributions rating premium
0-Star, obs -0.233 -0.247 (0.046)** (0.048)** 0.068**
1-Star, obs -0.194 -0.179 (0.016)** (0.018)** 0.033**
2-Star, obs -0.152 -0.146 (0.012)** (0.013)** 0.055**
3-Star, obs -0.095 -0.091 (0.010)** (0.011)** 0.063**
4-Star, obs -0.019 -0.029 (0.011) (0.012)*
1-Star -0.014 -0.012 (0.026) (0.027)
2-Star 0.016 0.007 (0.031) (0.032)
3-Star 0.045 0.015 (0.035) (0.036)
4-Star 0.085 0.035 (0.039)* (0.039)
ln (Score) 0.018 0.052 0.026 (0.025) (0.012)** (0.025)
Observations 61561 61561 61561 Charities 5468 5468 5468 R-squared 0.36 0.37 0.37
Robust standard errors in parentheses + significant at 10%; * significant at 5%; ** significant at 1%
ln_contit = K*StarK*Obs + K*StarK + f(Score) + ln(covars) + t +i + it
Introduction Effect
Signaling Effect
Charity Ratings20
Camp ResourceJune 2010 Laura Ellyn Grant
Overall Ratings by Category
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Religion
Arts & CultureHealth
Public Benefit
Environment
Human Services
Animals
International
EducationTotal
Cu
mu
lati
ve P
erce
nt
0
0.4
0.8
1.2
1.6
2
2.4
2.8
3.2
3.6
4
Ove
rall
Sta
rs
Stars4
Stars3
Stars2
Stars1
Stars0
Mean
Charity Ratings21
Camp ResourceJune 2010 Laura Ellyn Grant
Animals
Art, Culture,
Humanities Education Envrnmt Health
Human Services
Internat’l Public Benefit
Religion
0-Star, obs -0.116 -0.129 -0.126 -0.251 -0.289 -0.388 -0.111 -0.169 -0.314 (0.135) (0.117) (0.145) (0.253) (0.107)** (0.099)** (0.199) (0.123) (0.125)*
1-Star, obs -0.202 -0.116 -0.169 -0.272 -0.152 -0.109 -0.255 -0.198 -0.175 (0.075)** (0.046)* (0.056)** (0.067)** (0.047)** (0.041)** (0.063)** (0.041)** (0.047)**
2-Star, obs -0.227 -0.155 -0.174 -0.156 -0.107 -0.113 -0.178 -0.115 -0.037 (0.059)** (0.032)** (0.040)** (0.056)** (0.036)** (0.030)** (0.047)** (0.030)** (0.032)
3-Star, obs -0.119 -0.072 -0.126 -0.094 -0.022 -0.106 -0.075 -0.094 0.039 (0.044)** (0.029)* (0.033)** (0.041)* (0.030) (0.024)** (0.045)+ (0.027)** (0.031)
4-Star, obs -0.061 -0.082 -0.034 -0.064 0.058 -0.055 0.006 -0.070 0.139 (0.045) (0.037)* (0.033) (0.040) (0.030)+ (0.025)* (0.047) (0.030)* (0.045)**
Score, obs -0.077 -0.091 0.232 0.038 0.115 0.034 -0.103 0.065 0.105 (0.078) (0.091) (0.082)** (0.090) (0.047)* (0.064) (0.065) (0.055) (0.052)*
ln_fundraising 0.044 0.028 0.033 0.029 0.047 0.035 0.041 0.026 0.017 (0.008)** (0.007)** (0.007)** (0.010)** (0.010)** (0.005)** (0.010)** (0.006)** (0.005)**
ln_prog_exp 0.034 -0.003 0.036 0.030 0.043 0.073 0.064 0.040 0.054 (0.008)** (0.006) (0.010)** (0.008)** (0.008)** (0.008)** (0.010)** (0.008)** (0.011)**
ln_assets 0.322 0.408 0.312 0.471 0.359 0.374 0.328 0.390 0.229 (0.123)** (0.069)** (0.086)** (0.027)** (0.061)** (0.021)** (0.041)** (0.046)** (0.053)**
ln_liabilities -0.010 0.005 0.022 0.019 0.023 0.005 0.013 0.011 0.005 (0.006) (0.010) (0.009)* (0.008)* (0.007)** (0.006) (0.005)** (0.011) (0.006)
Observations 4539 8072 6519 3745 6656 12858 5300 9877 3995 Charities 393 676 563 337 583 1150 514 893 359 R-squared 0.34 0.27 0.30 0.41 0.44 0.45 0.48 0.37 0.40
Results by Sector
Charity Ratings22
Camp ResourceJune 2010 Laura Ellyn Grant
Economic Impact
• Calculate the median annual contributions by sector• Weight by average proportions in each star rating• Multiply respectively by estimated percent changes in
contributions in each sector and rating
$1 Billion/year loss, 2007 dollars
Charity Ratings23
Camp ResourceJune 2010 Laura Ellyn Grant
Discussion
• Introduction: Unambiguously reduces donations, on average– Findings vary by sector and size
• Signaling: Higher stars, greater contributions. Together the effects imply demand for charity is overly auspicious &
price elastic, on average• Is the money disappearing?
– Some is lost in transactions costs– Transfer to other unrated charities is likely
• If aggregate donations do not decrease, as if donors do not want to know the information.
• May be particularly a problem if ratings cause distortions and/or are uncorrelated with social impact
Charity Ratings24
Camp ResourceJune 2010 Laura Ellyn Grant
Further Work
• Can we predict the sensitivity to changes in the rating distribution or metrics used?
• Macro-economic trend in contributions affected by ratings?
• Trade-off between rating and reference charities?– Effect of other published charities gives cross-price of
ratings– Effect of unpublished charities gives transfer
• Does event analysis demonstrate trends of effects?– Learning versus salience– Growing popularity of ratings– Cohort effects and number of times rated
Charity Ratings25
Camp ResourceJune 2010 Laura Ellyn Grant
Thanks
• Camp Resources Organizers & Funders
• Charity Navigator
• NCCS of The Urban Institute
• Matt Kotchen, Paulina Oliva
• Funding from NSF IGERT, UC Regents,
& Bren School Toyota Fellowships, and
UCSB Economics Dept Data Grants.
Charity Ratings26
Camp ResourceJune 2010 Laura Ellyn Grant
Cumulative Star-Rating Probabilities by Nth Rating
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 Total
Star4
Star3
Star2
Star1
Star0
Charity Ratings27
Camp ResourceJune 2010 Laura Ellyn Grant
Analysis of residualsResidit = ln_Contit – (ln_Covarsit + i + t )