You are on page 1of 48

Who is more Generous with the Most Needy?

Experimental Evidence from Bogotá Stratification

Mariana Blanco Patricio S. Dalton


Universidad del Rosario Tilburg University, CentER, CAGE

FCE - UNC
Abril 2019

Blanco & Dalton (2019) Generosity and Socioeconomic Status 1 / 38


Motivation

Charitable giving is an increasing important economic activity around


the world.
In 2014, the Americans gave US$ 358.38 billion (2.1% of the US GDP).
In 2015, a third of the world’s population have given money to charity.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 2 / 38


Motivation

Charitable giving is an increasing important economic activity around


the world.
In 2014, the Americans gave US$ 358.38 billion (2.1% of the US GDP).
In 2015, a third of the world’s population have given money to charity.

Large amount of literature about charitable giving.


No clear understanding of the relationship of income and wealth with
giving.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 2 / 38


Motivation

Charitable giving is an increasing important economic activity around


the world.
In 2014, the Americans gave US$ 358.38 billion (2.1% of the US GDP).
In 2015, a third of the world’s population have given money to charity.

Large amount of literature about charitable giving.


No clear understanding of the relationship of income and wealth with
giving.
Little is known about charitable giving in developing countries.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 2 / 38


Motivation
Mixed evidence relating income and wealth with giving

Administrative or survey data for the US:


Auten et al (1997) - Administrative data: Positive correlation
Schervish & Havens (1998) - Survey data: Flat relationship
Andreoni (2006): U-shape
Data from the lab:
Eckel et al (2007): The experimental endowment does not affect the
probability to give. Positive correlation between experimental
endowments and the amount of giving to charity
Erkal et al (2011): Inverted U-shape between experimental earnings
and the probability to giving to other participant
Andreoni & Vesterlund (2001): No relationship
Tonin & Vlassopoulos (2013): No correlation between experimental
earnings and giving to charity
Field Experiment:
Andreoni et al (2017): No correlation between income and prosocial
behavior.
Blanco & Dalton (2019) Generosity and Socioeconomic Status 3 / 38
Motivation
Concerns about administrative/survey data

Difficult to identify the relationship between wealth and giving.

Measurement errors in self-reported income and self-reported giving


(may differ across wealth).

Different tax incentives: Charitable giving is deduced from taxable


income. The price of giving is lower the richer you are.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 4 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship

Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship
Different absolute income? The rich have more income to donate

Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship
Different absolute income? The rich have more income to donate
Different information? The rich may be better informed about the
charity and how it works

Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship
Different absolute income? The rich have more income to donate
Different information? The rich may be better informed about the
charity and how it works
Different weight to self-image? The rich may be more prone to
donate to improve self-image (particularly if they know they are
observed)

Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship
Different absolute income? The rich have more income to donate
Different information? The rich may be better informed about the
charity and how it works
Different weight to self-image? The rich may be more prone to
donate to improve self-image (particularly if they know they are
observed)
Different transaction costs? The rich may have better access to
the donation points (e.g. internet)

Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship
Different absolute income? The rich have more income to donate
Different information? The rich may be better informed about the
charity and how it works
Different weight to self-image? The rich may be more prone to
donate to improve self-image (particularly if they know they are
observed)
Different transaction costs? The rich may have better access to
the donation points (e.g. internet)
Different beliefs? The rich may have different beliefs about
deservedness (effort-luck), success alone or with help, etc.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38


Motivation
Concerns about administrative/survey data

Difficult to identify the determinants of the wealth/income−giving


relationship
Different absolute income? The rich have more income to donate
Different information? The rich may be better informed about the
charity and how it works
Different weight to self-image? The rich may be more prone to
donate to improve self-image (particularly if they know they are
observed)
Different transaction costs? The rich may have better access to
the donation points (e.g. internet)
Different beliefs? The rich may have different beliefs about
deservedness (effort-luck), success alone or with help, etc.
Different preferences? Are the rich more altruistic?
Blanco & Dalton (2019) Generosity and Socioeconomic Status 5 / 38
Motivation
Concerns about income distributions generated in the lab

External validity of the artificially generated income distribution in


the lab? Suitable to study income elasticity of giving, but not enough
to study giving behaviour across people from different socio-economic
stratum.

Wealth and socio-economic stratum include dimensions that are not


captured by the income distribution created in the lab (i.e.
discrimination, beliefs, opportunities, preferences, etc).

People’s real (rather than artificial) position in the income distribution


may shape their preferences, beliefs and information which in turn
may determine charitable giving.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 6 / 38


Our contribution

Take advantage of a unique feature of Colombia

observe giving behavior in a controlled environment, rather than


self-reported.

to a real charity helping the most needy, rather to other subjects


in the lab

observing real wealth, rather than self-reported or wealth created in


the lab.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 7 / 38


How we account for (or rule out) different possible
explanations?

Taxes incentives by not giving proof of donations (so it cannot be


used for tax refunds)
Information asymmetries by providing same information to people
about the Charity
Self-image by a double-blind design (donations are truly anonymous).
Transaction costs as the donations are requested in the University
where the subjects study.
Measurement errors by using an objective proxy of income/wealth.
Beliefs by elicitation with a post-experimental questionnaire.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 8 / 38


Colombia - Social Stratification by Law

All cities in Colombia are divided in 6 strata defined by law.

The stratification policy was made into law in 1994 to grant subsidies
to the poorest residents.

People living in upper layers (strata 5 and 6) pay higher electricity,


water and sewage rates to subsidise lower strata.

Unique identification: No need to rely on self-reported income or to


induce a fictitious income distribution.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 9 / 38


Strata distribution across Bogotá

Blanco & Dalton (2019) Generosity and Socioeconomic Status 10 / 38


The Strata in images - Bogotá

Stratum 2 Stratum 6

Blanco & Dalton (2019) Generosity and Socioeconomic Status 11 / 38


Strata, income and expenditure
Survey data for Bogotá (2011)

3000

2500

2000

1500

1000

500

0
2 3 4 5 6
-500

Income per capita Expenditure per capita

Source: Gallego et al. (2015)

Blanco & Dalton (2019) Generosity and Socioeconomic Status 12 / 38


The Charity

Non-profit that builds houses for the homeless.


Mainly present in Latin America & the Caribbean.
Run by young people (age limits for participation).
To date over 4,856 houses has been built across Colombia.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 13 / 38


Experiment 1: Design

Double-blinded dictator game with TECHO as recipient.


Recruiting e-mail.
Fill in a survey: identification of stratum and other individual
characteristics
When entering the lab participants receive:
an envelope marked inside according to their strata (Envelope D).
a blank envelope (envelope M),
the equivalent to US$10 in bills of US$1,
as many blank papers of the size of the bills as bills received and
a letter with information about TECHO.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 14 / 38


Experiment 1: Decision

Distribute bills of US$1 and blank papers between envelope D and M.


Each envelope had to have same number of bills + blank papers.
Keep the envelope M.
Complete a questionnaire and put it in a new blank envelope.
Put together both envelope D + Envelope with questionnaire into an
urn.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 15 / 38


Experiment 1: Sample

210 students
2 universities
11 sessions, ≈ 30-40 minutes
Average earning: US$ 6.83

Blanco & Dalton (2019) Generosity and Socioeconomic Status 16 / 38


Experiment 1: Sample

Stratum 1 2 3 4 5 6

Experiment 1 1 18 50 47 58 36

Blanco & Dalton (2019) Generosity and Socioeconomic Status 17 / 38


Do strata really identify income/wealth?
Financing tuition

0,80

0,60

0,40

0,20

0,00
2 3 4 5 6

Outside family (loan/scholarship) Family funds

Blanco & Dalton (2019) Generosity and Socioeconomic Status 18 / 38


Do strata really identify income/wealth?
Property ownership by stratum

1.00#

0.80#

0.60#

0.40#

0.20#

0.00#
2# 3# 4# 5# 6#

0+1# 2+3#

Blanco & Dalton (2019) Generosity and Socioeconomic Status 19 / 38


Do strata really identify income/wealth?

Table: Individual characteristics and strata ranges


Strata range: Low Middle Top Fisher’s exact test

Father’s education 2.444 3.063 3.723 (p=0.00)***


(0.511) (0.971) (1.239)
Mother’s education 2.556 2.907 3.426 (p=0.00)***
(0.784) (0.990) (1.112)
Father’s employment 0.833 0.794 0.798 (p=1.00)
(0.383) (0.407) (0.404)
Mother’s employment 0.778 0.784 0.606 (p=0.02)**
(0.428) (0.414) (0.491)
Student’s employment 0.111 0.206 0.266 (p=0.32)
(0.323) (0.407) (0.444)
Origin of funds 1.444 2.134 2.426 (p=0.00)***
(0.616) (0.671) 0.664)
Property owners 0.667 1.423 2.053 (p=0.00)***
(0.840) (0.966) (1.009)
Car owners 0.778 1.072 1.840 (p=0.00)***
(0.943) (0.904) (0.942)

Blanco & Dalton (2019) Generosity and Socioeconomic Status 20 / 38


Experiment 1: Results
Average Donation by Stratum

1
.8
.6
.4
.2
0

2 3 4 5 6
Strata

Experiment 1

Blanco & Dalton (2019) Generosity and Socioeconomic Status 21 / 38


Experiment 1: Results

We ran three regressions:

Donationi = α + βStratai + X1i 0 γ1 + X2i 0 γ2 + X3i 0 γ3 + ε i (1)

Donationi =α + β 1 Stratum 2i + β 2 Stratum 3i + β 3 Stratum 4i + β 4 Stratum 5i + (2)


+ X1i 0 γ1 + X2i 0 γ2 + X3i 0 γ3 + ε i

Donationi = α + β 1 Low Classi + β 2 Middle Classi + X1i 0 γ1 + X2i 0 γ2 + X3i 0 γ3 + ε i (3)

Blanco & Dalton (2019) Generosity and Socioeconomic Status 22 / 38


Experiment 1: Results

Dependent variable: Proportion of Endowment donated to Techo


(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Strata 0.039*** 0.032** 0.035** 0.039**


(0.015) (0.016) (0.016) (0.018)
Stratum 2 -0.175** -0.154* -0.199*** -0.193**
(0.078) (0.081) (0.072) (0.079)
Stratum 3 -0.141** -0.119* -0.096 -0.104
(0.057) (0.061) (0.062) (0.070)
Stratum 4 -0.113* -0.092 -0.107* -0.072
(0.061) (0.064) (0.062) (0.071)
Stratum 5 -0.097 -0.089 -0.081 -0.053
(0.060) (0.061) (0.060) (0.066)
Poor class -0.115* -0.098 -0.148** -0.160**
(0.067) (0.070) (0.062) (0.068)
Middle class -0.068* -0.049 -0.051 -0.054
(0.037) (0.041) (0.041) (0.046)
Constant 0.155** 0.224 -0.044 0.100 0.419*** 0.458** 0.182 0.346 0.360*** 0.372* 0.100 0.256
(0.061) (0.200) (0.226) (0.244) (0.049) (0.206) (0.255) (0.274) (0.029) (0.197) (0.233) (0.247)

Individual characteristics X X X X X X X X X
Participation in Techo X X X X X X
Beliefs X X X
Observations 209 209 195 180 209 209 195 180 209 209 195 180
R-squared 0.034 0.048 0.119 0.164 0.041 0.053 0.128 0.167 0.024 0.040 0.117 0.161
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Blanco & Dalton (2019) Generosity and Socioeconomic Status 23 / 38


Conclusion from Experiment 1

Amount donated increases with wealth

But... Are the rich intrinsically more generous?


or
Is it that the cost of giving is lower for them?

Blanco & Dalton (2019) Generosity and Socioeconomic Status 24 / 38


Experiment 2: Design

IDEM Experiment 1 but...


3 different endowments in random order (≈ 10, 18 and 25 USD).
Endowments = per capita daily expenditure of households in stratum
2, 4 and 6.
Strategy method: after all decision are made and envelopes sealed,
each participant rolls a dice to determine the decision that will be
paid out.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 25 / 38


Experiment 2: Sample

Stratum 1 2 3 4 5 6

Experiment 1 1 18 50 47 58 36

Experiment 2 1 24 40 42 43 16

Blanco & Dalton (2019) Generosity and Socioeconomic Status 26 / 38


Experiment 2 replicates Experiment 1
Average donation by stratum

Donations with endowment of USD 10 in Experiment 2 = donations in


Experiment 1
1
.8
.6
.4
.2
0

2 3 4 5 6
Strata

Experiment 1 Experiment 2

Replication test
Blanco & Dalton (2019) Generosity and Socioeconomic Status 27 / 38
Experiment 2: Results
OLS regressions using strata-equivalent endowment

Dependent variable: Proportion of relevant endowment donated to Techo


(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Strata 0.019 0.015 0.014 0.014


(0.017) (0.016) (0.018) (0.019)
Stratum 2 -0.035 -0.030 -0.032 -0.033
(0.088) (0.081) (0.091) (0.090)
Stratum 3 -0.034 -0.020 -0.044 -0.058
(0.076) (0.070) (0.071) (0.072)
Stratum 4 0.049 0.060 0.048 0.024
(0.078) (0.075) (0.074) (0.073)
Stratum 5 0.034 0.028 0.002 -0.003
(0.082) (0.077) (0.080) (0.078)
Poor class -0.060 -0.050 -0.033 -0.030
(0.065) (0.061) (0.076) (0.076)
Middle class -0.016 0.000 0.007 -0.007
(0.043) (0.043) (0.045) (0.046)
Constant 0.231*** -0.029 -0.308 -0.297 0.298*** 0.022 -0.242 -0.213 0.322*** 0.042 -0.261 -0.242
(0.066) (0.178) (0.246) (0.271) (0.070) (0.201) (0.251) (0.278) (0.036) (0.194) (0.253) (0.279)

Individual characteristics X X X X X X X X X
Participation in Techo X X X X X X
Beliefs X X X
Observations 165 165 142 137 165 165 142 137 165 165 142 137
R-squared 0.009 0.085 0.111 0.122 0.022 0.099 0.126 0.132 0.006 0.085 0.109 0.119
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

Blanco & Dalton (2019) Generosity and Socioeconomic Status 28 / 38


Conclusion from Experiment 2

Generosity is the same across strata

Blanco & Dalton (2019) Generosity and Socioeconomic Status 29 / 38


Conclusion from Experiment 2

Generosity is the same across strata

Do people from different strata differ in their motivations to give?

Blanco & Dalton (2019) Generosity and Socioeconomic Status 29 / 38


Experiment 3: Design

IDEM Experiment 2 but...


We match donations at a 1:1 ratio.
This experiment was ran only at one university.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 30 / 38


Experiment 3: Results
OLS regressions comparing matching with no matching

Dependent variable: Amount donated to Techo


(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Strata 0.026 0.026 -0.002 -0.006


(0.030) (0.029) (0.033) (0.035)
Stratum 2 -0.041 -0.030 0.068 0.061
(0.114) (0.112) (0.145) (0.149)
Stratum 3 -0.026 -0.038 0.008 0.014
(0.062) (0.061) (0.062) (0.067)
Stratum 4 0.086 0.082 0.116* 0.089
(0.069) (0.069) (0.069) (0.071)
Poor class 0.002 0.010 0.098 0.092
(0.104) (0.100) (0.125) (0.128)
Middle class 0.048 0.039 0.070 0.058
(0.054) (0.053) (0.053) (0.056)
Matching 0.013 -0.001 -0.127 -0.163 0.074 0.058 0.076 0.069 0.058 0.044 0.046 0.037
(0.149) (0.145) (0.160) (0.164) (0.071) (0.070) (0.071) (0.072) (0.062) (0.060) (0.062) (0.065)
Strata×Matching -0.001 0.000 0.030 0.039
(0.039) (0.037) (0.040) (0.041)
Stratum 2×Matching -0.013 -0.024 -0.110 -0.110
(0.140) (0.134) (0.162) (0.162)
Stratum 3×Matching -0.089 -0.079 -0.160* -0.174*
(0.094) (0.092) (0.087) (0.090)
Stratum 4×Matching -0.130 -0.116 -0.116 -0.094
(0.095) (0.097) (0.101) (0.102)
Poor class×Matching -0.033 -0.042 -0.110 -0.116
(0.127) (0.120) (0.141) (0.141)
Middle class×Matching -0.094 -0.084 -0.107 -0.103
(0.077) (0.076) (0.079) (0.081)
Constant 0.184 0.274* 0.069 0.166 0.264*** 0.344** 0.001 0.088 0.248*** 0.305* -0.030 0.067
(0.112) (0.142) (0.166) (0.177) (0.054) (0.169) (0.190) (0.222) (0.046) (0.162) (0.187) (0.214)

Individual characteristics X X X X X X X X X
Participation in Techo X X X X X X
Beliefs X X X
Observations 159 159 141 134 159 159 141 134 170 170 152 145
R-squared 0.014 0.049 0.128 0.149 0.043 0.079 0.181 0.191 0.009 0.046 0.125 0.140
Robust standard errors in parentheses. Using only observations from Universidad del Rosario
*** p<0.01, ** p<0.05, * p<0.1

Blanco & Dalton (2019) Generosity and Socioeconomic Status 31 / 38


Conclusion from Experiment 3

Weak evidence of warm-glow ONLY for stratum 3.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 32 / 38


Conclusion from Experiment 3

Weak evidence of warm-glow ONLY for stratum 3.

We can claim that all other strata mainly give motivated by pure
altruism.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 32 / 38


Within-subject Analysis - Experiment 2
Proportion of endowment donated by stratum

Proportion of endowment donated by strata


Same Increasing Decreasing Others
proportion proportion proportion
Low class 0.02 0.04 0.01 0.08
Middle class 0.02 0.10 0.07 0.31
High class 0.02 0.07 0.09 0.17
Cumulative 0.06 0.21 0.17 0.56

Blanco & Dalton (2019) Generosity and Socioeconomic Status 33 / 38


Within-subject Analysis - Experiment 2
Proportion of endowment donated by stratum

1
.8
.6
.4
.2
0
10 USD 18 USD 25 USD
Endowment

Strata 2
1

1
.8

.8
.6

.6
.4

.4
.2

.2
0

0
10 USD 18 USD 25 USD 10 USD 18 USD 25 USD
Endowment Endowment

Strata 3 Strata 4
1

1
.8

.8
.6

.6
.4

.4
.2

.2
0

10 USD 18 USD 25 USD 10 USD 18 USD 25 USD


Endowment Endowment

Strata 5 Strata 6

Blanco & Dalton (2019) Generosity and Socioeconomic Status 34 / 38


Within-subject Analysis - Experiment 3
Proportion of endowment donated by class

Proportion of endowment donated by class


Same Increasing Decreasing Others
proportion proportion proportion
Low class - 0.05 0.05 0.14
Middle class 0.06 0.10 0.11 0.18
High class 0.02 0.05 0.07 0.17
Cumulative 0.08 0.20 0.23 0.49

Blanco & Dalton (2019) Generosity and Socioeconomic Status 35 / 38


Within-subject Analysis - Experiment 3
Proportion of endowment donated by class
1

1
.8

.8

.8
.6

.6

.6
.4

.4

.4
.2

.2

.2
0

0
10 USD 18 USD 25 USD 10 USD 18 USD 25 USD 10 USD 18 USD 25 USD
Endowment Endowment Endowment

Low class Middle class High class

Blanco & Dalton (2019) Generosity and Socioeconomic Status 36 / 38


Conclusion

We identify socio-economic status without relying on artificially


induced distribution or self-reported data.
No difference across strata in the amount donated, once we account
for differences in income.
We control for information about the charity, personal characteristics,
beliefs about why people is poor, so we are only left out with the
preference channel.
We rule out that donations are motivated by warm-glow for all strata
but for stratum 3. Warm-glow motivations do not differ sharply
across strata.
The rich are not more generous than the poor, they just have more
money and that is why they donate more.

Blanco & Dalton (2019) Generosity and Socioeconomic Status 37 / 38


Experiment 1 Experiment 2 p-value
Full sample 0.318 0.307 0.5952
(0.257) (0.262)
Stratum 2 0.244 0.262 0.8261
(0.262) (0.268)
Stratum 3 0.278 0.255 0.5339
(0.207) (0.2025)
Stratum 4 0.306 0.317 0.7113
(0.246) (0.239)
Stratum 5 0.322 0.358 0.7656
(0.266) (0.310)
Stratum 6 0.419 0.331 0.3076
(0.296) (0.298)

Experiment 2

Blanco & Dalton (2019) Generosity and Socioeconomic Status 38 / 38

You might also like