Transcript Document

Self Image and Valuation of Public Goods:
Stated versus Actual Willingness to Pay
Olof Johansson-Stenman
Department of Economics, Göteborg University
Henrik Svedsäter
Organisation Behaviour, London Business School
[email protected]
Background
Hypothetical surveys to assess Willingness To Pay (WTP) for
public goods, typically environmental resources
Contingent Valuation (CVM)
Choice Experiments (CE)
Main question; do hypothetical values properly reflect real economic
commitments?
HypWTP = RealWTP (Carlsson & Martinsson, 2001; Cameron et al., 2002)
HypWTP > Real WTP, particularly for public goods and in between-subject
tests (List & Gallet, 2001; Murphy et al., 2004)
Background (ctd)
Why do the results differ?
What motivates people to contribute toward public goods?
Values here assumed to reflect two dimensions;
Instrumental value / utility of an environmental good or project;
actual change in G - conventional model
Self-image of contributing towards a ’good cause’
Moral motivation (Brekke et al., 2002)
’Warm glow’ of giving (Andreoni, 1989; 1990)
Symbolic value (Katz, 1960; Kahneman & Knetsch, 1992)
Self-signalling (Bodner & Prelec, 2003)
Background (ctd)
Hence, for issues with moral implications, contributions made partly to
improve or maintain an image as a good person / honourable citizen
Present also under conditions of anonymity (internally motivated) hence, not identical to self-presentation effects
Carries different weights in hypothetical versus real scenarios
Model and Predictions
Public goods: U = u (x, G, s)
where x = monetary costs, G = provision of good, and s = self-image
Self-image depends on (i) the degree to which the individual acts according to
her ethical views, and (ii) the estent to which she is honest toward herself
Hypothetical scenarios: no trade-offs between x, G and s;
WTP reflects maximisation of self-image, s
Real scenarios; WTP a trade-off between x, G and s (s costly)
H1: WTPHyp. > WTPReal
Model and Predictions
Private goods: U = u (x, G)
H2: WTPHyp. = WTPReal
Experimental design
Two sample groups:
A
Hypothetical
Choices
B
Real choices
Each subject made 16 pair-wise choices
8 reflecting a trade-off between (i) money to yourself, and (ii) a
restaurant voucher valid at a local restaurant - private good
8 reflecting a trade-off between (i) money to yourself, and (ii) a
donation made to WWF - public good
Altogether 32 unique choice set constructed, reflecting five levels of own
monetary payoff (0, 40, 80, 120, 160kr), and five levels of the private and
public good, respectively (0, 60, 120, 180, 240kr)
Experimental design
Il Gambero voucher
Alt. A
Alt. B
You receive
80
40
Value of voucher
0
120
Alt. A
Alt. B
160
40
0
240
}
Alt. A +40kr in own payoff,
but –120kr in voucher value
}
Alt. A +120kr in own payoff,
but –240kr in donation
Donation to WWF
You receive
Donation
One choice set randomly chosen to determine actual payoffs and voucher
value / donation
Double-blind procedure to ensure anonymity of participants
Results (n=160)
Amoral good (Restaurant voucher)
Moral good (WWF donation)
80
80
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
0
60
120
Hypothetical
180
Real
240
60
120
Hypothetical
180
Real
240
Parametric analysis
Logit, clustered Random effects
robust std
logit
errors
Fixed effects
logit
Money to subjects, β
0.027**
0.035**
0.035**
Restaurant voucher, 
0.014**
0.019**
0.019**
WWF donation, 
0.013**
0.017**
0.017**
Hyp treat * restaurant, 
0.0014
0.002
0.002
Hyp treat * WWF, 
0.005**
0.007**
0.007**
-1367.69
-1203.71
-781.59
Statistical obs
2560
2560
2560
n
160
160
160
Log-likelihood fcn
* and ** denote significance at the 0.05 and 0.01 level, respectively.
Estimation of MWTP – random utility framework
ui D E Vi D E  E
MW TPi 

 x
ui xi
Vi xi

E
where DE is the difference in donations across alternatives A and B,
and ,and xi is the difference in own payoff between alternatives
MWTP: how much can be taken away from the indvidual for each $1 given to G,
while leaving her at the same utility level?
Clustered robust std errors
Restaurant
Fixed effects logit
WWF
Restaurant
WWF
MWTPhyp
0.59 (0.024)
0.66 (0.025)
0.61 (0.048)
0.67 (0.051)
MWTPreal
0.54 (0.023)
0.47 (0.022)
0.55 (0.045)
0.47 (0.043)
Conclusions
H1 and H2 supported: WTPHyp. > WTPReal for public but not private good
Provision toward a good with moral implications tend to induce
self-image effects, which bias hypothetical statement upwards
Problematic for public-policy analysis
Results correspond to theorizing by Andreoni (1989), Kahneman
& Knetsch (1992), Brekke et al. (2002), Murninghan et al. (1996)
Conventional model not complete description of behaviour
Model also explains results of meta-analyses
Are small-stake experiments representative of behavior involving
larger financial trade-offs? Is a larger hypothetical bias likely
outside the experimental context?