Experimental Economics: Methods, Individual Decision-Making and Strategic Interaction

1

• Voluntary Contributions with Imperfect
  Information
• Annamaria Fiore
• University of Bari
• M. Vittoria Levati
  Andrea Morone
• MPI of Economics, Jena
  University of Bari
• University of Bari

2
Motivation

• to examine the effect of incomplete information
  on contribution levels
• two-person linear voluntary contribution
  mechanism with stochastic marginal benefits from
  the public good
• most previous experimental studies performed in
  an extremely rich informational environment
• Yet, in real life, one hardly knows
  in advance the marginal benefit she can derive
  from a public good she is asked to finance

3
Hipothesis

• A severe lack of information may lower
  individuals willingness to contribute (Bagnoli
  and Lipman, 1989, complete information is a
  very strong assumption . Incomplete
  information may lead to underprovision, p. 585)
• to establish a relationship between attitude to
  risk and willingness to contribute (Ledyard,
  1995, p. 143)

4
Experiment

• Two treatments
• PERFECT INFO
• IMPERFECT INFO
• Between design
• 32 participants in each treatment
• Computerized (Ztree, Fischbacher, 1999)
• Subject pool students in Economics
  recruitment ORSEE software (Greiner, 2004)
• Lab at MPI, Jena, Germany, on February 2005
• 2 stages
• 1 stage Vickrey auction
• 2 stage public goods game belief elicitation
• Currency ECU (10 ECU 1)

5
Results (1/4)

• The difference in contribution decisions between
  treatments is remarkable and significant
  (two-sided Wilcoxon rank-sum tests - based on
  averages over players for each pair as
  independent unit of observation plt0.001 in each
  of the first nine periods p.92 in period 10 N
  16 p.001 N 16 averaging over all periods).

6
Results (2/4)

• 1 period independent obs.
• individual contributions and individual beliefs
  highly significantly correlated (Spearman rank
  correlation coefficients are 0.83 in PERFECT
  INFO, and 0.80 in IMPERFECT INFO in each
  treatment)
• a statistically significant difference in
  first-period beliefs between treatments
  (one-tailed Wilcoxon rank-sum test p-value
  .004, N 32). No participant is expected to
  free-ride under perfect information, whereas 6
  subjects hold free-riding expectations under
  imperfect information.

7
Results (3/4)

• beliefs in all periods following the first
  endogenous
• correlation analysis between beliefs in period t
  about the partners contribution in t and the
  actual contribution of the partner in period
  (Spearman rank correlation coefficients PERFECT
  INFO .92 IMPERFECT INFO .98, p-value lt .001,
  N 16)

8
Relationship risk attitude/willingness to
contribute (4/4)

9
Conclusion

• Evidence for the impact of imperfect information
  on voluntary contribution behavior in linear
  public goods games, and for the relationship
  between risk attitudes and willingness to
  cooperate
• Linear voluntary contribution mechanism
  efficient?
• Political economy perspective individuals
  provided with a good knowledge about their
  marginal benefits if requested to contribute

10

Thanks for your attention! Annamaria
Fiore University of Bari afiore_at_dse.uniba.it