Journal of Public Economics 15 (1981) 2955310. North-Holland Publishing Company ECONOMISTS FREE RIDE, DOES ANYONE ELSE? Experiments on the provision of public goods, IV Gerald MARWELL and Ruth E. AMES* University of Wisconsin, Mudison, WI 53706, USA Received August 1980, revised version received January 1981 Eleven closely related experiments testing the free rider hypothesis under different conditions, and sampling various subpopulations, are reported. Results question the empirical validity and generality of a strong version of the hypothesis. Some reasons for its failure are discussed. 1. Introduction The free rider hypothesis [Hardin (1968), Olson (1968)] has been one of the most widely accepted propositions in the literature on the provision of public goods by groups. This acceptance, however, has been based primarily on the strength of the theoretical argument, and the citation of commonplace example, rather than rigorous empirical test. In this paper we report on a series of experiments expressly designed to test the hypothesis and related aspects of the theory. Some of these experiments have been extensively reported in the sociological literature [Marwell and Ames (1979, 1980) contain experiments 1, 2, 5, 6; Alfano and Marwell (1980) reports on experiment 111. Others will be reported here for the first time. Our objective is to review and consider the sum of our findings. The experiments reported here add substantially to a small, but quickly growing experimental literature whose general trend has been to question the power of the free rider hypothesis for predicting behavior in collective action situations. A large portion of this work has been done by psychologists and has recently been reviewed by Dawes (1980) and Edney (1980). Almost all of this work, however, has dealt with free riding in small groups. Similarly, the few articles by economists and sociologists which have appeared in the literature also tend to deal with relatively small numbers of subjects in *This research was supported by National Science Foundation grants No. GS-3742 and No. GS-43507. A version was presented at the Public Choice Association meeting in March, 1980. We would like to thank Geraldine Alfano and John Fleishman for their advice on and participation in this work. We also thank Gene Smolensky, Michael Rothschild, Lee Hanson and Arthur Goldberger for their patience. 004772727/81/0000~0000/$02.50 G North-Holland Publishing Company296 G. Marwell and R.E. Ames, Free rider hypothesis relatively small groups [Bohm (1972) Brubaker (1975) Sweeney (1973) Smith (1978) Schneider and Pommerehne (1979)]. These latter works also tend to use very complex experimental designs, running individual subjects through multiple trials, each of which involves a somewhat different experimental condition. In the research reported below standard social psychological experimental procedures are employed. Care is taken to have subjects understand the task and the situation. Subjects are not used in more than one condition so that there is no contamination from order effects, etc. Equally important, the number of experiments and subjects involved serves to replicate and give weight to our conclusions. In our experiments ‘free riding’ refers to the absence of contribution. towards the provision of a public good by an individual, even though he or she will not be excluded from benefiting from that good. The free rider hypothesis is based on the assertion that under such conditions it is irrational for an individual to voluntarily contribute. Following Brubaker (1975) however, we will consider two versions of the group-level implication of this assertion: the ‘weak’ version of the free-rider hypothesis, which states that the voluntary provision of public goods by groups will be sub-optimal; and the ‘strong’ version, which argues that (virtually) no public goods at all will be provided through voluntary means. 2. The research paradigm To test the free rider hypothesis, and related aspects of the theory, we created a highly controlled, very abstract, experimental situation. The original experiments were designed to maximize the probability of free riding by minimizing the possible effects of normative and other pressures for contributing towards the public good. Although aspects of the paradigm changed for different experiments, it is useful to begin by describing the simplest version that we used. All experiments in the program may be seen as variants of this situation. 2.1. Dependent variable: Investment in the public good The central ingredient in the research paradigm was the operationalization of our dependent variable - investment in a public good. For this purpose, subjects were provided with a given amount of resources, in the form of tokens, which they had to invest in one of two ‘exchanges’: The ‘group exchange’ or the ‘individual exchange’. The group exchange was our operationalization of a public good, while the individual exchange was a private good. Tokens invested in the individual exchange earned a set amount, regardless of the behaviour of other group members or anythingG. Marwell and RX. Ames, Free rider hypothesis 297 else. In a typical experiment this might be one cent per token. The individual exchange was thus like a bank in assuring a specific return on investment. The return was ‘excludable’ in neither affecting, nor being affected by, returns to other group members. The group exchange, on the other hand, paid its cash earnings to all members of the group by a pre-set formula, regardless of who invested. The subject received a share of the return on his own investment in the group exchange
View Full Document
Unlocking...