Morgan (2000) and Morgan and Sefton (2000) suggest that one way of overcoming the free-rider problem in financing of public goods is to bundle contributions with purchases of tickets for a fixed-prize lottery. However, if some voluntary contributions are driven by reciprocity to expected contributions of others, introduction of a lottery can crowd out such contributions. We experimentally investigate the extent of such potential crowding out effect. We consider three treatments using a within-subject design. T1 is a regular linear voluntary contribution game. T2 introduces a lottery in which every token of contribution automatically buys a lottery ticket. One ticket is then drawn at random and wins a fixed lottery prize. In T3, three out of four group members participate in the lottery, whereas the remaining member does not, and instead receives a fixed compensation intended to neutralize the wealth effect of nonparticipation. T3 is intended to isolate the reciprocal reaction to others contributing out of desire to win the lottery prize as opposed to maximizing social efficiency from an own desire to win the lottery prize. In particular, comparison of T3 to T1 isolates the former effect, whereas the comparison of T2 to T3 isolates the latter effect. We find that, on average, contributions are lower in T2 in comparison to T1, indicating presence of the crowding out effect. We then further classify subjects by the strength of their conditional cooperation, a proxy for reciprocity, measured a’la Fischbacher et al. (2001). We find that the extent of crowding out at subject level is weakly increasing with the level of conditional cooperation. We interpret this finding as crowding-out of voluntary contributions driven by a reciprocal reaction to the perception that others contribute out of greed as opposed to out of sense of generating social welfare.
Keywords: public good, lottery, social preferences, reciprocity, crowding out.