The problem of free riding in voluntary generic advertising: parallelism and possible solutions from the lab.

To avoid potential impacts from subjects considering this to be a finite game, subjects were not told the total number of rounds of the experiment. The second treatment of the experiment was conducted in much the same way as the first. The primary difference was that producers in the second treatment were assessed at $0.25 for each unit sold and the assessments funded an advertising program that increased demand in the subsequent round. Subjects were informed not only that the advertising program "in previous experiments" had increased demand but also that the higher demand would result in higher prices and greater profits for producers. In fact, generic advertising programs traditionally inform producers of all the benefits of generic advertising prior to implementation of the program. Specifically, the advertising program increased the demand in the next round by [Q.sub.D-Increase] = 2/3 * [[summation].sup.n.sub.i=1] [A.sub.i] where [A.sub.i] is the assessment collected for each subject, i = 1, ..., 20. The advertising program increased demand above the level determined by the draw of the bingo ball. (4) Subjects were informed of the expected price that would result from different amounts of assessments collected. The assessment rate and corresponding increase in expected price were set to parallel the high return on investment, roughly four to one, frequently found with generic commodity advertising. Thus the average MPCR in the experiment was 0.25. As previously discussed, in VCM settings where the MPCR is less than one, the Nash equilibrium for a subject is to request a refund of her assessment. Subjects always have the financial incentive to request a refund (free ride), as it would provide them with the highest possible earnings in any particular round.

To simulate the discussions among producers that typically occur in the referendum process related to generic advertising programs, subjects were given up to five minutes prior to the start of VCM to discuss the advertising program as a group. In these discussions (commonly referred to as "cheap talk") subjects were permitted to discuss only strategy regarding the advertising program and not price collusion. After the rounds began, subjects were not allowed to communicate with each other.

As with some historic generic advertising programs, subjects could request a refund of some or all of their assessments. To request a refund, subjects submitted, via instant messaging, a confidential one-sentence request stating the amount of the refund desired with a maximum refund of $0.25 per unit of quantity sold. An example is "Subject #5 requests a refund of $0.25 for Round Eight. Sincerely, Jane Doe." All refund requests were granted and refunds were added back to the subject's profits. In each round, the administrator announced the total assessments possible, the total assessments actually collected, and the corresponding increase in demand.

After participating in a voluntary program for eleven years, U.S. egg producers held a referendum in 1988 on whether to create a mandatory program or to have no program at all. Egg producers voted overwhelmingly in favor of a mandatory program (84% in support). To mimic this historical transition from a voluntary to a mandatory program, subjects were asked to vote on whether they wanted a mandatory program with no option of a refund or no advertising program and thus no assessments. Three of the experimental designs included the mandatory treatment and the total sample size was 160 (table 1).

If the subjects elected the mandatory program, then producers were assessed $0.25 for every unit sold and the assessments were used to fund the advertising program. If the subjects elected the no-advertising-program option, then producers operated identically to the first treatment, which had no advertising program. Subjects were given up to five minutes to discuss the referendum with the entire group. A majority vote determined the outcome and producers went through five rounds. Subjects were informed that whatever assessments were collected in the last round of the third treatment affected demand in the first round of the fourth treatment.

After tabulating the confidential votes, the administrator announced the election results and directed the subjects to the treatment of their spreadsheet that corresponded to the election outcome. If the mandatory program was elected, then in each round the administrator announced the total assessments collected and the increase in demand, the market price, and the number of units sold. If no advertising program was elected, the administrator announced the market price and the number of units sold.

To simulate the potential transition that could result if a mandatory generic advertising program were replaced by a voluntary PPM funding mechanism, the fourth and final treatment was identical to the second treatment except that a PPM with a refund-by-request feature (referred to as PPM-Refund) was employed. In the PPM-Refund, subjects were assessed for each unit sold and could submit confidential requests for refunds of their assessments. However, in this treatment the advertising program was implemented only if at least 70% of the subjects did not request refunds. (5) If seven or more of the twenty subjects in each experimental session requested refunds, the advertising program was not implemented and all contributors received a complete refund of their assessments.

Note that the provision point of 70% was based on the number of subjects not requesting refunds instead of applying the provision point to the total possible contributions. The advantage of tying the provision point to the number of subjects was its transparency since the number of subjects in the experiment remained constant while the total possible contributions could potentially change in each round. Additionally, for practical policy purposes, a PPM based on the percentage of producers participating would likely be preferred because it would be perceived as more democratic.

If the 70% provision point was met, the advertising program operated as described in the second treatment--the amount of money actually collected and the corresponding increase in demand were announced. In addition, the number of subjects in each round who did not request a refund was announced to the subjects. If the provision point was not met, the round operated identically to the first treatment without an advertising program and subjects were given the opportunity to reach the provision point in the subsequent round.

Experimental Design III--VCM versus PPM

To answer the second research question of whether a PPM can yield higher levels of voluntary contributions, 40 subjects participated in an experimental design that was identical to the Baseline design except that it repeated the VCM-Refund as the fourth treatment of the session. Consequently, the results from the VCM versus PPM design can be compared to the results of the 80 subjects who participated in the PPM-Refund as the fourth treatment in the Baseline design (table 1).

Experimental Design III--Status Quo Bias and the VCM

To determine whether status quo bias is a reliable source of increased contributions for the VCM, two sessions (n = 40) were conducted in which the second treatment involved a reversal of the status quo for the contributions. These results can then be compared to those from the 160 subjects who participated in the VCM-Refund as the second treatment of their session (table 1).

The Status Quo Bias and the VCM design exactly duplicated the first two treatments of the Baseline design except that the status quo of the donation was changed to not contributing rather than contributing. The design consisted of two treatments. The first treatment was five rounds without the advertising program. The second treatment was eleven rounds in which the funds for the advertising campaign were raised by contributions given by subjects (referred to as VCM-Contribution). (6) The subjects were not aware of the actual number of rounds in the second treatment of the experiment and none of the experiment parameters were changed. The written instructions were identical to those for the first two treatments of the Baseline design except that subjects were no longer automatically assessed for every unit sold and subsequently given the opportunity to request a refund of these assessments. Instead, subjects were given an opportunity to contribute up to twenty-five cents for each unit they sold. To make a contribution, subjects entered the amount into their spreadsheets and completed a one-sentence instant message stating their intents. A sample message is "Subject #5 contributes $0.25 for Round Eight. Sincerely, Jane Doe." All contributions were accepted and the amounts were deducted from the producer's profits.

Experimental Design IV--Status-Quo Bias and the PPM

The first three treatments of the Status Quo Bias and the PPM design were identical to the Baseline design. However, in the fourth treatment subjects were informed that the advertising campaign would be implemented only if 70% or more gave "complete contributions," where their contributions were the maximum possible of $0.25 for each unit sold (referred to as PPM-Contribution). This design mirrors the fourth treatment of the Baseline design (PPM-Refund), in which the advertising campaign was implemented if 70% or more of the subjects "did not request a refund" of any amount.

Results