VCM or PPM? A comparison of the performance of two voluntary public goods mechanismsIntroductionExperimental designAggregate contribution levels and demand revelationRelative efficiency of the VCM and PPMA tournamentEfficiency as successes and failuresEfficiency as welfare gainsA synthetic field experiment and final remarksAcknowledgementsReferencesVCM or PPM? A comparison of the performance oftwo voluntary public goods mechanismsDaniel Rondeaua, Gregory L. Poeb, William D. Schulzeb,*aDepartment of Economics, University of Victoria, CanadabDepartment of Applied Economics and Management, Cornell University, United StatesReceived 6 May 2003; received in revised form 6 May 2004; accepted 3 June 2004Available online 21 November 2004AbstractLittle progress has been made toward understanding the relative performance of the twomechanisms most widely used for fundraising: the Voluntary Contribution (VCM) and ProvisionPoint (PPM) mechanisms. This paper provides direct comparisons of the relative performance ofvariants of the VCM and PPM as they are most commonly implemented in the field. The researchmakes use of 1296 individual observations from 721 subjects, including 40 observations from a fieldexperiment. A meta-analysis of the determinants of contribution levels and bootstrap estimates of therelative efficiency of the two mechanisms provide novel analyses of public goods experimental data.Overall, the PPM is found to increase total contributions, to be more responsive to changes in inducedvalue, and to be generally more efficient than the VCM. For public goods with a benefit–cost ratio inthe interval [1, 1.4), however, the VCM captures a greater portion of available benefits than the PPM.D 2004 Elsevier B.V. All rights reserved.JEL classification: H41 public goods; C92 design of experiments, laboratory, group behavior; C93 design ofexperiments, field experimentsKeywords: Voluntary public goods mechanisms; Provision point; VCM; PPM; Efficiency1. IntroductionEconomists have long sought to develop public goods contribution mechanisms thatovercome the free or cheap-rider problem (e.g. Clarke, 1971; Groves, 1973; Groves and0047-2727/$ - see front matter D 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.jpubeco.2004.06.014* Corresponding author. Fax: +1 607 255 9611.E-mail address: [email protected] (W.D. Schulze).Journal of Public Economics 89 (2005) 1581 – 1592www.elsevier.com/locate/econbaseLedyard, 1977). While much insight has been gained from the development of incentive-compatible mechanisms, these instit utions are of little practical significance since they areeither too complex, require some form of coercive power, or rely on unanimity or iterativegroup procedures. This makes them unsuitable for most field applications. Such barriersled Laffont (1987) to conclude that bany real application will be made with methods whichare crude approximat ions [of incentive-compatible mechanisms]Q (p. 567).Fundraisers most often rely on the Voluntary Contributions Mechanism (VCM) or theProvision Point Mechanism (PPM), neither of which is theoretically incentive compatible.Although these mechanisms are widely used in the real world and have been extensivelystudied in the laboratory, remarkably little is known about their relative performance.This paper provides a direct large-scale cross-mechanism comparison of variants of theVCM and PPM as they are implemented in real fund raising campaigns. The large datasetand its collection method enable the use of meta-analysis and bootstrapp ing techniques.With these methods, we econometrically analyze the determinants of contributions in theVCM and PPM, and study their relative efficiency using 430,000 simulated public goodswith benefit–cost ratios ranging from 0.42 to 2.48. The paper concludes by reporting theresults of a within-subject experiment conducted in the field with both mechanisms.2. Experimental designThe core of the analysis makes use of 1176 individual observations obtained from singleshot, no-feedback laboratory experi ments. These experiments were set in an environmentdeliberately chosen to mimic field conditions. They were conducted with large groups ofsubjects who had heterogeneous induced values and incomplete information about both thedistribution of values and the exact number of individuals in their group.Design choices were generally motivated by the pragmatic need to understand andimprove the application of public goods mechanisms in the field. Specifically theexperimental design was intended to closely correspond to the fieldwork of Champ et al.(1997) in which a real VCM was implemented to fund the removal of abandoned roads thatcontinued to provide access to ecologically sensitive areas of Grand Canyon National Park.The removal of all 215 miles of unwanted roads had an estimated total cost of US$142,000.However, the project was presented to potential donors as a divisible good by stressing thateach US$1 donation would cover the cost of removing eight feet of road. No provisions weremade for spending donations in excess of removal costs and it was assumed that little or noadditional benefits would accrue to donors from contributions exceeding the total cost.11Marks and Croson (1998) note that a similar effort to raise funds for a new baseball stadium in Seattle devotedthe surplus to the State’s general funds, making it impossible for contributors to assess the existence or value ofreturns from excess contributions. We contend that there is a major class of public goods in the real world wherethere are either institutional or physical reasons for a sudden drop in marginal utility: often fund drives arecentered on constructing a building of a specific size or purchasing an existing building; many VCMs use athermometer to measure how close the total contributions are to the stated goal which allows some discrete levelof activity to be reached which the organization argues is necessary or dire consequences will result; land trustsusually try to fund specific purchases.D. Rondeau et al. / Journal of Public Economics 89 (2005) 1581–15921582Thus, if one assumes a constant level of individual benefits per unit of road removed, thisproject corresponds to a standard VCM up to the total cost of the project, and no benefitsthereafter.The VCM we implemented in the laboratory reflects these features of the Grand Canyoncampaign. Each individual received a constant positive individual return from the publicgood, up to a maximum