Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Calamity, Aid and Indirect Reciprocity:

The Long Run Impact of Tsunami on Altruism

Leonardo Becchetti1 Stefano Castriota1 Pierluigi Conzo2

1Univ. of Rome “Tor Vergata”; 2CSEF

June 18, 2012

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

1 Introduction

2 Research Design

3 Descriptive Evidence

4 Hyp. Testing

5 Econometric Analysis

6 Robustness

7 Conclusions

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Introduction

Natural Disasters:

destruction of material wealth and capital stock + job losses

affect expectations, preferences and choices →consumption/saving and human capital investment decisions.

What we do:

test the impact of 2004 tsunami on a sample of Sri LankanMF borrowers’ altruistic preferences

...7 year later, dictator game, exploiting variation invictimization and damage type at individual level.

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Related experimental literature (1)

Conflicting results!

Cassar et al. (2011): Thai tsunami victims become slightlymore impatient, trusting, trustworthy, risk-averse (uncertaintyabout the future, unrealistically higher probabilities of natural disasters).

Callen (2010): Sri Lankan victims of the same Tsunamicalamity are more patient.

Whitt and Wilson (2007): missing family members reducecontributions among evacuated from New Orleans to Houstonshelters after Hurricane Katrina

Castillo and Carter (2011): non-linear impact of 1998hurricane Mitch on trust and reciprocity of Honduran victims(intermediate shocks help coordination around a higher equilibrium, while

extreme shocks undercut it)

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Related experimental literature (2)

Fleming et al. (2011): victims of Chilean 2010 earthquakereveal significantly lower trustworthiness (“aftermath moralhazard”, migration, displacement)

Eckel et al. (2009): survivors of Hurricane Katrina in NewOrleans tended to act more risk lovingly in the short term(“heaven owes me for my past bad luck”)

Voors et al. (2012): people exposed to violence in Burundihave higher discount rates.

Becchetti et al. (2011): violence after 2007 political outbreaksnegatively affects trustworthiness in Kenya.

Heterogeneous results may be explained by type of damages

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Our contribution/identification strategy (1)

Our experiment set in Dec. 2011 = longer time horizon(Callen: mid 2007; Cassar et al.: mid 2009)

Info on individuals’ victimization within each village → reducesheterogeneity between “treatment” & “control”

Discriminate among damages to individuals, to their homesand to their economic activity.

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Our contribution/identification strategy (2)

Causality tsunami → preferences supported by:

limited differences between damaged and non damaged in thesame villages;

differences not significant when we compare types of shocks

both damaged and non damaged belong to same MFI → samescreening, same “ethos” → less heterogeneity.

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Main Findings

The shock affects participants’ altruistic preferences even in thelong run:

who report at least one damage give and expect less thanthose who did not

who reports only losses to the ec. activities give and expectsignificantly less than those who suffer also injuries or housedamages

the latter receive significantly more help than the former

Possible interpretation of such generosity: indirect reciprocity.

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Background

Over 1,000 Km of coast (2/3) affected by the wave;

Dramatic human impact: over 35,000 dead;

Economic losses: 24,000 boats, 11,000 businesses and 88,500houses damaged or destroyed;

Several international organizations and NGOs provided helpand support.

Affecting almost randomly individuals living at short distancefrom each other...

...based on their location at the moment of the tsunami withrespect to the waterline

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Sampling

Randomly selected 380 borrowers from list of AMF → identifydamaged and non damaged in each village.

3 villages (Galle, Matara and Hambantota) only partiallyaffected by the calamity → within-village heterogeneity

Tsunami Impact Figure

Why AMF sampling:

initial screening by AMF (and/or potential self-selection intoit) is likely to reduce heterogeneity;

loan officers informed us about the damaged/non-damagedstatus → avoided potential framing.

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Experiment

Dictator Game: 2 players, a Sender (S) and a Receiver (R).Anonymity.

S is endowed with 900 LKR (5.74 eur) and has to decide howmuch of it to send to R;

R takes no actions in this game and receives the amount ofmoney S has sent.

S’s maximum utility is reached by sending 0 LKR and keepingthe whole endowment (900 LKR).

S’s deviation from 0 can be interpreted as a measure ofaltruism.

Additional games to control for risk and time preferences+ socio-demographic survey

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Protocol

Participants are told about the sequence of the interviewprocess i.e. an experimental session + survey + a final lottery.

They are paid just for one game extracted before they play.

If the participant is chosen to be R in DG we elicit her/hisFirst Order Beliefs (50 LKR for a correct guess)

Max payoff 900 LKR = about 51% of the median per capitamonthly food expenditure in our sample

Damage Classification:

1 injuries to family members2 damages to house3 damages to economic activity (raw materials, economic

activity, working tools)

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

...working hard and/or hardly working...

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Descriptive Statistics

Age = 47; n. of household members = 4.5; n. of educationyears = 10.5; about 63% = relatively impatient and 60% ofthe amount in the risky option.Declare at least 1 type of damage (“damaged”)= 53%.Damage distribution:

Variable Obs Mean Std. Dev. Min Max

Injury 201 0.239 0.427 0 1Economicloss 201 0.841 0.367 0 1Housedamage 201 0.259 0.439 0 1

Injuryonly 201 0.065 0.247 0 1Eclossonly 201 0.473 0.5 0 1

Housedamageonly 201 0.005 0.071 0 1

Injuryhousedamage 201 0.06 0.238 0 1Injuryeconomic 201 0.174 0.38 0 1

Eclosshousedamage 201 0.254 0.436 0 1

Alldamages 201 0.06 0.238 0 1

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Balancing Properties (Damaged vs. Non-Damaged)Variable Group Obs Mean Non-par test (z, p)

age Rest of sample 179 44.53 -3.41Damaged 201 48.93 0.001

male Rest of sample 179 0.05 -1.485Damaged 201 0.09 0.137

married Rest of sample 179 0.9 3.001Damaged 201 0.79 0.003

separated Rest of sample 179 0.01 -1.754Damaged 201 0.03 0.08

widowed Rest of sample 179 0.07 -1.88Damaged 201 0.12 0.06

single Rest of sample 179 0.03 -1.493Damaged 201 0.06 0.135

n house members Rest of sample 179 4.37 -2.598Damaged 201 4.69 0.009

years schooling Rest of sample 178 10.81 1.981Damaged 196 10.29 0.048

agricolture Rest of sample 179 0.31 3.95Damaged 201 0.14 0.000

manufacturing Rest of sample 179 0.32 -0.103Damaged 201 0.32 0.918

fishery Rest of sample 179 0.01 -2.503Damaged 201 0.06 0.012

trading Rest of sample 179 0.32 -2.098Damaged 201 0.42 0.036

(cont.) Balancing Properties (Damaged vs. Non-Damaged)Variable Group Obs Mean Non-par test (z, p)foodexp Rest of sample 178 8.27 -0.646

Damaged 201 9.16 0.518switch Rest of sample 179 5.7 -1.292

Damaged 201 6.07 0.197impatient Rest of sample 179 0.6 -1.186

Damaged 201 0.66 0.236riskloving Rest of sample 179 176.31 -0.146

Damaged 201 179.1 0.884trustindex Rest of sample 179 1.21 0.538

Damaged 199 1.2 0.591most can be trusted Rest of sample 179 1.97 0.088

Damaged 199 1.96 0.93cant rely Rest of sample 179 1.54 0.802

Damaged 201 1.47 0.423people take advantage Rest of sample 179 1.06 -1.202

Damaged 199 1.11 0.229BMI Rest of sample 177 22.97 -1.671

Damaged 200 24.11 0.095distance housecoast Rest of sample 173 10.73 9.988

Damaged 197 3.54 0.000distant Rest of sample 179 0.74 9.026

Damaged 201 0.28 0.000

Helpindexi =∑

8k=1 aidsk

8 , where aids concern money, credit, food,medicines, raw materials, tools, consumption, other.

Damaged Obs 201Mean 0.211

[95% CI] .1726659 .2489759Non Damaged Obs 179

Mean 0.071[95% CI] .0482759 .0941822

Injury Obs 48Mean 0.24

[95% CI] .1532893 .3258774Economic Losses Obs 169

Mean 0.231[95% CI] .1877707 .2737678

House damage Obs 52Mean 0.361

[95% CI] .2689816 .4521723Injury Or Housedamaged Obs 88

Mean 0.29[95% CI] .2249546 .3545909

Ec. Loss only Obs 95Mean 0.172

[95% CI] .1227889 .221948

Balancing Properties (InjuryOrHouseDamaged vs. Only Ec. Loss)Variable Group Obs Mean Non-par test (z, p)

age InjuryOrHouseDamaged 88 49.67 1.618Only Ec. Loss 95 46.93 0.106

male InjuryOrHouseDamaged 88 0.08 -0.363Only Ec. Loss 95 0.09 0.717

married InjuryOrHouseDamaged 88 0.78 -0.627Only Ec. Loss 95 0.82 0.531

separated InjuryOrHouseDamaged 88 0.02 -0.366Only Ec. Loss 95 0.03 0.714

widowed InjuryOrHouseDamaged 88 0.15 1.869Only Ec. Loss 95 0.06 0.062

single InjuryOrHouseDamaged 88 0.05 -1.055Only Ec. Loss 95 0.08 0.291

n house members InjuryOrHouseDamaged 88 4.66 -0.422Only Ec. Loss 95 4.83 0.673

years schooling InjuryOrHouseDamaged 84 10.05 -1.144Only Ec. Loss 94 10.54 0.252

agricolture InjuryOrHouseDamaged 88 0.14 0.201Only Ec. Loss 95 0.13 0.841

manufacturing InjuryOrHouseDamaged 88 0.31 -0.433Only Ec. Loss 95 0.34 0.665

fishery InjuryOrHouseDamaged 88 0.1 1.925Only Ec. Loss 95 0.03 0.054

trading InjuryOrHouseDamaged 88 0.43 0.291Only Ec. Loss 95 0.41 0.771

(cont.) Balancing Properties (InjuryOrHouseDamaged vs. Only Ec. Loss)Variable Group Obs Mean Non-par test (z, p)foodexp InjuryOrHouseDamaged 88 8.4 -0.494

Only Ec. Loss 95 8.75 0.621switch InjuryOrHouseDamaged 88 5.77 -1.13

Only Ec. Loss 95 6.28 0.259impatient InjuryOrHouseDamaged 88 0.6 -1.307

Only Ec. Loss 95 0.69 0.191riskloving InjuryOrHouseDamaged 88 188.86 1.606

Only Ec. Loss 95 169.89 0.108trustindex InjuryOrHouseDamaged 87 1.18 -1.836

Only Ec. Loss 94 1.25 0.066most can be trusted InjuryOrHouseDamaged 87 2 2.575

Only Ec. Loss 95 1.93 0.01cant rely InjuryOrHouseDamaged 88 1.42 -1.649

Only Ec. Loss 95 1.57 0.099people take advantage InjuryOrHouseDamaged 87 1.14 -0.274

Only Ec. Loss 94 1.11 0.784BMI InjuryOrHouseDamaged 88 24.13 0.704

Only Ec. Loss 94 24.04 0.481distance housecoast InjuryOrHouseDamaged 86 3.39 -2.406

Only Ec. Loss 95 3.35 0.016distant InjuryOrHouseDamaged 88 0.25 -0.363

Only Ec. Loss 95 0.27 0.717

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Testing Giving By Damage

Giving by: Obs Mean Std dev Non-par test (z, p)Rest of sample 89 0.37 0.205 2.051

Damaged 101 0.31 0.168 0.04Rest of sample 169 0.34 0.191 0.518

Injury 21 0.32 0.16 0.604Rest of sample 103 0.37 0.2 2.069Economicloss 87 0.31 0.169 0.039Rest of sample 165 0.34 0.194 -1.056Housedamage 25 0.37 0.139 0.291Rest of sample 96 0.37 0.202 0.41Only injured 6 0.32 0.17 0.682

Rest of sample 96 0.37 0.202 2.329Only economicloss 54 0.29 0.181 0.02

Rest of sample 96 0.37 0.202 -0.977Only housedamage 1 0.5 . 0.329

Injury 21 0.32 0.16 0.837Only Ec. Losses 54 0.29 0.181 0.403

InjuryOrHouseDamage 40 0.34 0.148 1.667Only Ec. Losses 54 0.29 0.181 0.096

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Testing Expected Receiving By Damage

Expected Receiving by: Obs Mean Std dev Non-par test (z, p)Rest of sample 90 0.44 0.189 2.396

Damaged 100 0.39 0.209 0.017Rest of sample 163 0.41 0.194 0.067

Injury 27 0.43 0.242 0.947Rest of sample 108 0.44 0.191 2.473Economicloss 82 0.38 0.209 0.013Rest of sample 163 0.42 0.204 0.198Housedamage 27 0.42 0.185 0.843Rest of sample 101 0.44 0.183 0.151Only injured 7 0.46 0.307 0.88

Rest of sample 101 0.44 0.183 3.221Only economicloss 41 0.33 0.203 0.001

Injury 27 0.43 0.242 1.647Only Ec. Losses 41 0.33 0.203 0.100

InjuryOrHouseDamage 48 0.44 0.22 2.235Only Ec. Losses 41 0.33 0.203 0.025

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Testing the Solidarity Norm by Damage

Solidarity Norm by: Obs Mean Std dev Non-par test (z, p)Rest of sample 179 0.41 0.2 3.164

Damaged 201 0.35 0.193 0.002Rest of sample 332 0.38 0.195 0.271

Injury 48 0.38 0.216 0.786Rest of sample 211 0.4 0.198 3.298Economicloss 169 0.34 0.193 0.001Rest of sample 328 0.38 0.203 -0.588housedamage 52 0.39 0.165 0.557Rest of sample 197 0.4 0.195 0.436Only injured 13 0.39 0.255 0.663

Rest of sample 197 0.4 0.195 4.106Only economicloss 95 0.31 0.191 0.000

Rest of sample 379 0.38 0.198 -0.869Only housedamage 1 0.5 . 0.385

Injury 48 0.38 0.216 1.87Only Ec. Losses 95 0.31 0.191 0.062

InjuryOrHouseDamage 88 0.39 0.196 2.863Only Ec. Losses 95 0.31 0.191 0.004

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Determinants Of Giving (OLS)

Dep. Var: Giving (1) (2) (3) (4)OLS OLS OLS OLS

damaged -0.0593** -0.0582**(0.0274) (0.0291)

injury -0.0111 -0.018(0.0381) (0.0419)

economicloss -0.0556** -0.0505*(0.0272) (0.0299)

impatient -0.0452 -0.046(0.0298) (0.0299)

riskloving ratio 0.108** 0.107**(0.0486) (0.0504)

SocioDem. Controls NO YES NO YES

Observations 190 185 190 185R-squared 0.025 0.127 0.023 0.123

Robust se in parentheses; Controls: age, marital status, gender, foodexp, village, yearsschool., house members, employment sector, trustindex, loancycle

Robustness: WLS and TOBIT

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Determinants Of Expected Receiving (OLS)

Dep Var (1) (2) (3) (4)Expected Receiving OLS OLS OLS OLS

damaged -0.0491* -0.0754**(0.0289) (0.0304)

injury 0.0460 0.0242(0.0511) (0.0546)

economicloss -0.0651** -0.0715**(0.0314) (0.0320)

impatient -0.0689** -0.0677**(0.0302) (0.0299)

riskloving ratio -0.0353 -0.0360(0.0602) (0.0590)

SocioDem. Controls NO YES NO YES

Observations 190 187 190 187R-squared 0.015 0.185 0.026 0.181

Robust se in parentheses; Controls: age, marital status, gender, foodexp, village, yearsschool., house members, employment sector, trustindex, loancycle

Robustness: WLS and TOBIT

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Determinants Of Solidarity Norm (1)

Dep. Var.: Solidarity Norm OLS OLS TOBIT TOBIT

receiver 0.0750*** 0.0780*** 0.0761*** 0.0794***(0.0198) (0.0204) (0.0205) (0.0207)

damage -0.0542*** -0.0619*** -0.0539*** -0.0618***(0.0199) (0.0213) (0.0206) (0.0215)

impatient -0.0626*** -0.0653***(0.0214) (0.0217)

riskloving ratio 0.0407 0.0394(0.0388) (0.0401)

SocioDem. Controls NO YES NO YES

Observations 380 372 380 372R-squared 0.055 0.132

Robust se in parentheses; Controls: age, marital status, gender, foodexp, village, years school., housemembers, employment sector, trustindex, loancycle

Robustness: WLS

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

Determinants Of Solidarity Norm (2)

Dep. Var: Solidarity Norm OLS OLS OLS OLS

receiver 0.0731*** 0.0695*** 0.0762*** 0.0728***(0.0198) (0.0197) (0.0205) (0.0205)

injury 0.0200 0.0149(0.0328) (0.0365)

economicloss -0.0598*** -0.0612***(0.0206) (0.0226)

eclossonly -0.0845*** -0.0711***(0.0227) (0.0233)

impatient -0.0624*** -0.0605***(0.0212) (0.0212)

riskloving ratio 0.0374 0.0354(0.0389) (0.0385)

SocioDem. Controls NO YES NO YES

Observations 380 380 372 372R-squared 0.058 0.070 0.130 0.133

Robust se in parentheses; Controls: see above. Robustness: WLS and TOBIT

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

WLS and IV

All OLS and TOBIT results robust to WLS, where the weightsare the inverse p-score [Blattman and Annan (2010) and Hirano et al. (2003)]:

damagedpscore(damaged)

+1−damaged

1−pscore(damaged)

Tables: Giving ; Expected Giving ; Solidarity (1) ; Solidarity (2)

All results on Solidarity Norm are robust to IV-estimates usingDistance from Coast and BMI to instrument the Damageddummy. IV

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

What we do

We test the long run impact of 2004 Tsunami in Sri Lanka onaltruistic preferences;

We exploit within-village variability betweendamaged/non-damaged and across damage types;

Among victims, we compare those who report only losses tothe economic activity vs. those who report also damages tothe house and/or injuries to relatives;

Differences in observables (including the distance from thecoast) between the two groups disappear;

Sample composed by MF borrowers → common unobservables→ reduced heterogeneity + IV estimate

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Introduction Research Design Descriptive Evidence Hyp. Testing Econometric Analysis Robustness Conclusions

What we find

Those who report at least one damage give and expect lessthan those who do not;

Among damaged, those suffering not only losses to economicactivities (but also damages to house or injuries to relatives)give and expect significantly more than other damaged;

since those who report only losses to economic activityexperience on average less aid than those who also reporthouse damages or injuries...

...we interpret the superior pro-sociality of the latter in termsof indirect reciprocity.

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

THANK YOU!

We thank:

CSEF and EIEF fellows for all useful comments and suggestions.

C. Angelico, C. Pagano, E. Agostino and N. Kurera for the precioussupport on field data collection.

Etimos Foundation, Etimos Lanka and AMF for financial and logistic aid.

Back

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Risk Preferences

Simplified Charness and Genicot (2009)’s approach.

Participants endowed with 300 LKR; decide if KEEP orINVEST any portion x in a risky asset;

risky asset pays 3x with 1/2 prob. and 0 otherwise;

x = rough proxy of risk aversion (the higher x , the less riskaverse).

Back to the main experiment

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Time Preferences

If you win the lottery, would you like to receive:Option n. A B

1 10,000 after 2 month 10,100 after 8 months2 10,000 after 2 month 10,198 after 8 months3 10,000 after 2 month 10,368 after 8 months4 10,000 after 2 month 10,607 after 8 months5 10,000 after 2 month 10,840 after 8 months6 10,000 after 2 month 11,180 after 8 months7 10,000 after 2 month 12,247 after 8 months8 10,000 after 2 month 14,142 after 8 months

Back to the main experiment

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Socio-Demographic Survey

socio-demographic information

social preferences: standard GSS questions regarding trust,negative reciprocity, sociability

damage received in the 2004 on six dimensions: familymembers, house, economic activity, buildings/assets, workingtools, raw materials.

Classified as:

1 injuries to family members2 damages to house3 damages to economic activity

Back to the main experiment

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Determinants Of Giving (WLS)

Dep. Var: Giving WLS WLS WLS WLS

damaged -0.0913*** -0.0802**(0.0338) (0.0346)

injury -0.0579 -0.0662(0.0501) (0.0509)

economicloss -0.0848** -0.0667**(0.0327) (0.0336)

impatient -0.0757** -0.0760**(0.0354) (0.0353)

riskloving ratio 0.136*** 0.139***(0.0520) (0.0524)

SocioDem. Controls NO YES NO YES

Observations 184 183 184 183R-squared 0.047 0.184 0.050 0.186

Notes: Robust standard errors in parentheses; Weights: damagedpscore(damaged )

+ 1−damaged1−pscore

(damaged );

pscore estimated using age, years school., village, employm. sector, BMI, distant, loancycle

Back to OLS or back to Robustness

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Determinants Of Giving (TOBIT)

Dep. Var: Giving (5) (7) (9) (11)TOBIT TOBIT TOBIT TOBIT

damaged -0.0590** -0.0566**(0.0282) (0.0284)

injury -0.00903 -0.016(0.038)1 (0.0402)

economicloss -0.0560** -0.0494*(0.0278) (0.0293)

impatient -0.0478 -0.0486*(0.0291) (0.0291)

riskloving ratio 0.111** 0.110**(0.0478) (0.0494)

SocioDem. Controls NO YES NO YES

Observations 190 185 190 185

Notes: Robust standard errors in parentheses; *** p

Determinants Of Expected Receiving (WLS)

Dep Var: Expected Receiving WLS WLS WLS WLS

damaged -0.0313 -0.0507(0.0314) (0.0326)

injury -0.0579 -0.0662(0.0501) (0.0509)

economicloss -0.0848** -0.0667**(0.0327) (0.0336)

impatient -0.0999*** -0.0760**(0.0326) (0.0353)

riskloving ratio -0.0381 0.139***(0.0554) (0.0524)

SocioDem. Controls NO YES NO YES

Observations 187 186 184 183R-squared 0.007 0.190 0.050 0.186

Notes: Robust standard errors in parentheses; Weights: damagedpscore(damaged )

+ 1−damaged1−pscore

(damaged );

pscore estimated using age, years school., village, employm. sector, BMI, distant, loancycle

Back to OLS or back to Robustness

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Determinants Of Expected Receiving (TOBIT)

Dep Var (5) (7) (9) (11)Expected Receiving TOBIT TOBIT TOBIT TOBIT

damaged -0.0488 -0.0759**(0.0302) (0.0301)

injury 0.0534 0.0317(0.0540) (0.0554)

economicloss -0.0678** -0.0751**(0.0329) (0.0319)

impatient -0.0723** -0.0712**(0.0300) (0.0295)

riskloving ratio -0.0413 -0.0423(0.0620) (0.0607)

SocioDem. Controls NO YES NO YES

Observations 190 187 190 187

Notes: Robust standard errors in parentheses; *** p

Determinants Of Solidarity Norm (WLS)

Dep. Var.: Solidarity Norm WLS WLS I. W. TOBIT I. W. TOBIT

receiver 0.0582** 0.0646*** 0.0588** 0.0655***(0.0243) (0.0234) (0.0251) (0.0238)

damaged -0.0611*** -0.0648*** -0.0602** -0.0637***(0.0232) (0.0238) (0.0239) (0.0238)

impatient -0.0957*** -0.0998***(0.0243) (0.0247)

riskloving ratio 0.0529 0.0524(0.0397) (0.0404)

SocioDem. Controls NO YES NO YES

Observations 371 369 371 369R-squared 0.045 0.149

Notes: Robust standard errors in parentheses; Weights: damagedpscore(damaged )

+ 1−damaged1−pscore

(damaged );

pscore estimated using age, years school., village, employm. sector, BMI, distant, loancycle

Back to OLS or back to Robustness

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Determinants Of Solidarity Norm (TOBIT)

Dep. Var: Solidarity Norm TOBIT TOBIT TOBIT TOBIT

receiver 0.0740*** 0.0703*** 0.0774*** 0.0740***(0.0205) (0.0205) (0.0208) (0.0209)

injury 0.0246 0.0201(0.0340) (0.0372)

economicloss -0.0612*** -0.0632***(0.0214) (0.0230)

eclossonly -0.0878*** -0.0744***(0.0239) (0.0241)

impatient -0.0651*** -0.0631***(0.0215) (0.0215)

riskloving ratio 0.0357 0.0337(0.0402) (0.0398)

SocioDem. Controls NO YES NO YES

Observations 380 380 372 372

Notes: Robust standard errors in parentheses; *** p

Determinants Of Solidarity Norm (WLS)

Dep. Var: Solidarity Norm WLS WLS WLS WLS

receiver 0.0580** 0.0560** 0.0638*** 0.0618***(0.0241) (0.0246) (0.0232) (0.0237)

injury -0.0683*** -0.0627***(0.0231) (0.0230)

economicloss -0.0143 -0.0284(0.0434) (0.0426)

eclossonly -0.0725*** -0.0608***(0.0244) (0.0231)

impatient -0.0942*** -0.0942***(0.0241) (0.0244)

riskloving ratio 0.0526 0.0529(0.0398) (0.0401)

SocioDem. Controls NO YES NO YES

Observations 371 371 369 369R-squared 0.049 0.042 0.149 0.140

Notes: Robust standard errors in parentheses; Weights and pscore: see above.

Back to OLS or back to Robustness

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”

Determinants Of Solidarity Norm (IV)

Dep. Var.: Solidarity Norm IV IV IV IVdamaged -0.0947** -0.0998** -0.0961** -0.0959**

(0.0431) (0.0491) (0.0417) (0.0477)receiver 0.0748*** 0.0783*** 0.0775*** 0.0807***

(0.0199) (0.0199) (0.0200) (0.0200)impatient -0.0613*** -0.0627***

riskloving ratio 0.0403 0.0347(0.0379) (0.0381)

SocioDem. Controls NO YES NO YESObservations 380 372 377 369

R-squared 0.045 0.124 0.044 0.123Instruments distant distant distant, BMI distant, BMI

First Stage: F 51.64 13.28 37.49 13.40Exogeneity test: χ2 1.062 0.699 1.415 0.642Exogeneity test: p 0.303 0.403 0.234 0.423

Weak Id.Test: F 103.3 76.32 56.23 40.24Overid.test: χ2 0.0687 0.454Overid.test: p 0.793 0.500

Notes: Robust standard errors in parentheses.

back to Robustness

Becchetti, Castriota, Conzo Univ. of Rome “Tor Vergata”, CSEF

“Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism”