The Nature Services Exchange of Jamestown is a trade-mark or brand established by a cooperative effort of the University of Rhode Island (URI) and EcoAsset Markets, Inc. (EAM), as an experimental business to bring farmers together with residents of Jamestown in support of hayfield management for grassland-nesting birds, particularly the bobolink. The project was funded by a U.S. Department of Agriculture Conservation Innovation Grant through the Natural Resource Conservation Service (USDA/NRCS). Primary objectives of the project emphasize the application of methods from experimental economics and economics of “mechanism design” for delivery of public goods. Whether delivered by private action, such as farmers or non-profit organizations, or by government, public goods benefit many people simultaneously and the provider is often unable to exclude beneficiaries of the good. Without exclusion, providers are normally unable to obtain the full, economic return to producing public goods. In contrast, private goods, such as the familiar food and fiber produced by farmers, are “excludable” and the producer of private goods typically is able to withhold benefits of the goods from individuals who have not paid some price in compensation for provision. Farmers’ production of wildlife habitat services, particularly for non-game species, is a public good because typically many people can benefit from knowing that a local ecosystem is well-functioning (an existence value) or from viewing wildlife, such as from public roads, without having to compensate farmers for its provision. Many of the associated issues are elaborated in Swallow et al. (2008).

The experimental market established under this project was designed to test the performance of three basic mechanisms. The purpose of this report is to summarize the results of this experiment. The basic market was developed to enable farmers to obtain revenues from local residents who are interested in seeing grassland-nesting birds, particularly bobolinks, have a substantially higher success after a given nesting season. Normal hay harvesting practices generally lead to a harvest during the middle of the nesting season for grassland birds, with very high rates of nest and fledgling mortality. Bollinger et al. (1990), for example, show a strong tie between success of nesting bobolinks and the timing of hay harvests. While farmers are generally conservation-minded, their need to feed livestock and run their business often necessitates this direct conflict, so that if the public is to have higher levels of grassland bird populations, the public would need to compensate farmers to forego the early harvest (generally around June) of hay and to manage their farm around protecting nesting habitat.

For the experimental market reported here, it should be noted that the experimental approach placed the residents of Jamestown into a format for marketing that some may interpret as less-straightforward than what a small business might do if there was no experimental purpose. However, an ordinary small business would have been risking a lot on choosing one approach, perhaps risking complete failure or at least failure to identify the performance of any given mechanism relative to the potential value of the public good. Moreover, the project focused on methods other than the common “donations” approach, because this method is already well-used by non-profits and yet these non-profits are not typically applying the approach to encouraging farmers to managing for grassland bird nesting habitat. The project did include an extensive marketing effort that was designed to reach all residents of Jamestown with a common approach, through full-page advertisements in the Jamestown Press newspaper (already mailed to all homes in Jamestown ) and through stories in the news or posters placed in places typically visited by residents.

The Mechanisms Tested
The experimental mechanisms were introduced to each household through a mailing directed at that household. Each household was randomly assigned to one of three experimental mechanisms and to one of six farm-fields in 2007 and one of four farm-fields in 2008. Each experimental mechanism was designed to elicit an offer from an individual (household) that was willing to pay to help support bird-friendly management under a farm-wildlife contract. The mechanisms focus on reducing the incentives that individuals have to withhold payment, even though the person may value improved wildlife habitat services, because if other individuals pay for the contract the benefits still accrue to all individuals who value the services. This incentive to withhold payment is related to “free-riding” whereby individuals can “ride for free” on the payments made by others.

In each year of the market, there were three mechanisms tested. Two of these mechanisms were intended to raise revenues sufficient to cover the cost of a farm-wildlife contract. The third mechanism was designed to measure the potential value of the contracts to individuals, thereby establishing a baseline against which to measure the performance of revenue-raising mechanisms. This third mechanism was not intended to raise sufficient revenues to cover the cost of a contract, although in rare circumstances that is a possible outcome. All three mechanisms, in both years, gave residents a “money back guarantee” that if offers (revenues) collected fell short of the pre-established amount needed to pay for a farm-wildlife contract, then any contributors would have 100% of their payment refunded. The marketing materials emphasized that all monies collected would either be used to pay the costs of a farm-wildlife contract or would be refunded.
The mechanisms then differed with regard to the disposition of any money raised in excess of the amount needed to pay for a contract. The rules for such rebates (or refunds) were designed to reduce the cost to an individual of making a decision to provide an extra dollar when, in retrospect, that dollar was not absolutely necessary to cover the cost of the farm-wildlife contract. Identifying rules that work more effectively is often an objective of experimental economics and mechanism design. These rules were explained on “payment cards” or “offers” made to individuals within the direct mail marketing materials sent to each household.

One revenue-raising mechanism is dubbed the Proportional Rebate (PR) mechanism. When the total of offers made for a field exceeds the cost of the farm-wildlife contract, then excess money is returned in proportion to each individuals share of the total of all contributions. For example, if the funding-target (i.e., the cost of a farm-wildlife contract) is $100 and the PR mechanism collects $200, then half of all contributions are “extra” and the individuals who contributed are given a rebate of 50% of their contribution. (Note that the numbers here are chosen for illustration. Actual numbers would be scaled to reflect the costs of a farm-wildlife contract, which may be tailored to the unique situation of each farmer and his or her field(s).) This PR mechanism has been extensively studied by experimental economists. However, in experiments done by these economists, generally the public good is simulated through the use of carefully designed incentives and pay-offs, whereas the project here uses management of hayfields for grassland birds as a real public good and enables testing of revenue-raising mechanisms in an actual community. Work in experimental economics laboratories shows that the PR mechanism raises significantly more money (revenue) than simple donations or than donations with a money-back-guarantee without any rebate rules. An advantage to the PR mechanism is that each contributor ends up paying the same proportion of their offer, which some individuals might see as fair to all contributors.

The second revenue-raising mechanism is denoted the Uniform Price Auction (UPA). Under this mechanism, if the total of all offers for a farm-wildlife contract exceeds the cost of the contract, then the rebate rules require the administrator (i.e., the market-maker) to attempt to identify a single, uniform price that would be the same for all people who end up paying. As implemented here, the UPA would make a full refund to anyone who offered to pay less money than the calculated uniform price, and would rebate all money in excess of the uniform price for those individuals who offered more than the uniform price. This mechanism has the advantage that if an individual ends up paying into the farm-wildlife contract, that individual pays the same dollar amount as everyone else and the value the person has offered for the contract is higher than the amount the individual pays. Some individuals might view this uniform price approach as more similar to their experience in ordinary markets, and therefore more fair. A key disadvantage is that because individuals who offered something less than the uniform price will pay nothing, it is possible that even though the total of offers exceeded the cost of a farm-wildlife contract, it may be that no uniform price can be found; in that event, the farm-wildlife contract is not implemented and everyone receives a full refund. The UPA was only used in the 2007 market.

In 2008, the UPA mechanism was replaced by a mechanism denoted as the Uniform Price Cap (UPC). This mechanism was similar to the UPA but was designed to eliminate the key disadvantage of the UPA. Under UPC, if the total of all offers is sufficient to cover the cost of a contract, the market-maker then calculates a uniform price cap that sets the maximum amount that any individual will pay while still raising enough revenues to pay for the costs of implementing a farm-wildlife contract. Anyone who made an offer that is less than the price cap will only pay the amount of their offer; anyone who made an offer of more than the price cap will receive a rebate of any money in excess of the price cap. The UPC has the advantage that, for any collection of offers, if the PR mechanism ends up funding a farm-wildlife contract, the UPC could also raise enough revenues to fund the contract. UPC has the disadvantage of increasing the link between the offer that individuals make and the amount the individual ends up paying. (That is, the UPC increases the link between what an individual says a good is worth to him or her and what the individual actually pays.) This link encourages individuals to down-grade their offers relative to their true values for a farm-wildlife contract because by doing so they create an opportunity to let other contributors pay more of the costs.
In each year, the third mechanism tested was the Pivotal Mechanism (PM), which was not designed to raise revenues to actually fund a farm-wildlife contract. However, on economic theory grounds, the PM establishes an important baseline against which to measure the value of a farm-wildlife contract to the community. The PM has the property of “incentive compatibility,” meaning, in this context, that the PM gives individuals an incentive to make an offer to pay for a farm-wildlife contract that is fully consistent with that individual’s true value for the contract. Incentive compatibility is established by rules for payment (and refunds) that make each person pivotal (critical) to the final outcome for any particular farm-wildlife contract. Under PM, if the total of all offers exceeds the cost of the contract, each individual person’s payment depends only upon whether they were actually critical to the final outcome. In particular, each individual only pays their offer if, after accounting for offers from all others, the market-maker still needs this individual’s offer to cover the cost of implementing the farm-wildlife contract. The person then has an incentive to offer their full value for the contract because, if the person is not pivotal, he (she) pays nothing, but if the individual offers less than his (her) full value and the contract could have been implemented had the individual offered his full value, then the individual is no worse off but getting the contract implemented and having to pay his offer.

In the experiment then, we expect the highest offers to come from the PM, since the PR, UPA and UPC mechanisms are not incentive compatible. There is no solid basis to rank these latter three mechanisms, but for PR there is some experimental research suggesting that the mechanism is very effective and in a separately funded project, Chhandita Das and Chris Anderson have produced results that PR produces offers in experimental economics laboratories that are statistically equivalent to results for the PM.

Overview of Implementation
The entire population of residents of Jamestown was used in the experimental market, and the URI and EAM are quite grateful for their cooperation. Households were randomly assigned to the PM, PR or UPA mechanisms in 2007 and to the PM, PR or UPC mechanisms in 2008. In both years, marketing materials were sent twice during March and April, with the exception that about 50% of the households who responded in 2007 being assigned the same mechanism in 2008 (recipients of the UPA in 2007 were assigned to the UPC in 2008), with the other 50% assigned to the other two mechanisms.

Because this is a field experiment, outside the more abstract but more carefully-controlled economics laboratory, the investigators could not control each individuals choice of whether to participate in the experiment. In this context, participation means that an individual chose to mail a reply card back to the Nature Services Exchange (at a URI address), whether or not they decided to make a monetary offer. Monetary offers were received either by personal check or by a credit-card authorization.

About 10-12% of residents decided to participate in the Exchange, netting about 8% of individuals making offers. This means around 87-90% or so chose not to respond. This participation rate is somewhat higher than in many marketing efforts for familiar businesses. For our analysis, this participation rate means that our data needs to be analyzed in two steps. One is a statistical evaluation of the factors that increased the chance that an individual would decide to participate. The other is a statistical model of the factors that increased the magnitude of the offers made by those who participated, while accounting for those who participated but chose to make a zero offer. Therefore, our statistical evaluation of the offers made is conditional on an individual having made a choice to participate.

Within the marketing materials, some individuals were given a discrete-choice (DC) opportunity, while other individuals were given an open-ended opportunity (OE), again using a random assignment process. The DC form presents individuals with one amount of money and the choice to agree to offer that amount, or to decline to offer that amount; individuals were not allowed to offer any other amount. The OE form presented individuals with a list of possible offer-amounts to choose among and also allowed the individual to name their own amount (fill-in the blank). In 2007, the OE form was either “high” with listed amounts between $35 and $120, while a “low” form listed values between $20 and $80 (we designated these lists as OEhigh and OElow, respectively). The DC form is expected by economists to generate higher values because it reduces the opportunity for an individual to strategize by choosing an amount that is lower than they would actually be willing to pay rather than forego seeing a farm-wildlife contract implemented.

In addition to data on the mechanisms and the DC or OE form of presentation, we also obtained information from a commercial marketing firm indicating each resident’s attributes for age, household income (or purchasing power), and information on whether the household had a record of donating to various activities or making mail-order purchases. Finally, each direct-mail solicitation to individuals described the farm-wildlife contract available for which that person was being asked to make a decision, including information on the number of acres under contract (7-18 acres; but always 10 acres in 2008), number of bobolink territories recently observed on the field (1-4), whether the parcel was visible from a main road (whether it had a “view”) and, for 2008 only, an explicit statement of the number of fledglings that might be expected to be raised in the field given the number of bobolink territories observed in a recent year.
In what follows, we summarize what we believe our data reasonably supports about the functioning of the Exchange in Jamestown . We first discuss factors affecting participation by residents, and then discuss factors affecting a resident’s willingness to pay a higher amount in support of farm-wildlife contracts focused on the bobolink. For both pieces of analysis, we used some statistical methods (“random effects”) to account for responses from many of the same people over two years (rather than from two distinct groups of people) were used. In the “payment equation,” this two-year accounting created some more substantial challenges, so that in that case our conclusions are supported in part through more “sensitivity analysis.”

Participation Results
As in any market, Jamestown residents may choose whether or not to participate. In this test market, we designated any individual who chose to return a response to the direct-mail marketing as a participant, whether or not the individual made a monetary offer to support a farm-wildlife contract. Therefore, our primary analysis of the data separates the participation decision from the individual’s decision to make or not make an offer or payment in their response. Because participants may differ systematically from non-participants, the analysis of offers (e.g., a “payment equation” discussed in the next section) must be conditioned on adjusting for the systematic, non-random factors that influence an individual to choose to participate or not. Thus, the participation equation discussed here has two implications: it sheds light on factors affecting successful delivery of the marketing message relative to data on differences across individuals, and it supports statistical methods used to adjust non-random (or unbiased) sampling of Jamestown residents due to individual’s ability to control their participation decision (meaning their participation decision is outside the control of analysts).

The Nature Services Exchange project placed full-page newspaper advertisements in the Jamestown Press. These advertisements were available to the entire community, because the Press is mailed weekly to all households in Jamestown and it also available free at many community locations from quick-shops to town hall. Therefore, there is no way to identify statistically the effect of the details of these advertisements on the decision of individuals to participate. Between 2007 and 2008, the style of advertisements changed, and the content in 2008 was streamlined and focused on a theme to link grassland bird habitat in hayfields to the community character. Statistical estimation of a difference in participation between years is possible, however, there is no way to statistically connect any differences to the marketing as separate from other factors affecting everyone in Jamestown, including the fact that participation decisions in 2008 automatically represent the second opportunity for individuals to make the decision. In addition, the analysis relied on here allowed for an individual’s choice in 2008 to be correlated with his or her choice in 2007.

Factors considered in the statistical analysis of the participation decision include the differences in the farm-wildlife contracts offered to individuals, the lowest amount of money (above zero) that was suggested to individuals by the marketing materials, and the factors available from a commercial marketing firm with respect to the individual’s age, household purchasing power (income) and evidence that a household member had previously chosen to participate in other efforts, such as donations to causes, including other environmental causes, and mail-ordering. Because it is common for households to discard marketing materials prior to careful (or any) reading, it would not be surprising if the main factors affecting the participation decision were independent of the content of the marketing envelops.

In the analysis, the characteristics of the farm-wildlife contracts were not statistically significant factors influencing the probability that an individual participated in the market. This test allowed for the influence of contract characteristics to differ for individuals who were known to previously donate to environmental causes as compared to other individuals. However, there is strong statistical evidence that the minimum amount suggested for payment influenced the probability of participation. The higher that minimum ($20 or $35 in the OE presentations and various amounts in the DC presentations) the less likely an individual is to participate. Below, in discussion of the payment equation, it is noted that the DC presentation (and the OE presentation starting from $35) elicit statistically significant increases in the offers from individuals. The negative effect of higher requests on the participation decision is a partial off-set to the positive effect that may occur with higher amounts in the payment equation. The influence of suggested payment levels on participation is also evidence that recipients of the direct mail did open and evaluate at least some of the marketing materials.

The payment mechanism presented to individuals (the UPA, UPC, PR or PM) was not statistically significant in increasing or decreasing the probability that an individual participated. The evaluation of payment mechanisms included consideration of interaction effects between presentation (DC or OE) and the mechanism, and it considered whether the mechanism an individual received in 2007 affected their participation decision in 2008. These results are consistent both with the idea that recipients did not consider the mechanism (payment rules) in making their participation decision and, on the other hand, the idea that individuals did consider the payment rules but that the mechanisms created no differential effect on the participation decision. The data do not allow us to determine whether the payment mechanisms had no effect or just had the same effect on participation.

The presentation of the payment decision, whether DC or OE, was also not statistically significant in affecting the participation decision. This evaluation included interactions between the presentation and the economic mechanisms and interactions with the minimum suggested amount presented.
The probability that an individual chose to participate in 2008 was lower, by a statistically significant margin, as compared to 2007. However, individuals who made an offer in 2007 were more likely to participate in 2008 than were other individuals.

Characteristics of the individuals household also made a statistically significant difference to their participation decision. Individuals were more likely to participate from households with higher purchasing power (income), for older individuals, and individuals who had previously donated to any activity, particularly if they had donated to environmental causes. Individuals from households that had used mail-orders to obtain children’s items were statistically less likely to participate. Individuals for which household income, past donating choices for environmental or other causes, and past mail-ordering choices for children’s items were unknown (due to missing information) were, respectively, more likely, more likely and less likely to participate. Many of these factors affected the probability of participation in the same direction as the same factor’s effect on the magnitude of an individuals payment.

Payment Equation Results
Several factors could influence the amount that an individual is willing to offer for payment toward a farm-wildlife contract. The factors, on which data are available, include the characteristics of the field (acres covered, number of bobolink territories observed in a prior year, visibility from main roads, and projected or potential number of fledglings that nests might produce); the individual’s age and purchasing power of his or her household; the household’s previous use of mail-order or willingness to donate in other settings; and, of course, the details of the payment rules and the presentation of the opportunity to buy-into the contract. In addition, participants in the first year (2007) may make choices in a manner that differs in the second year (2008) and differs from those who did not participate in the first year.

Our analysis considered the primary effects (or “main effects”) of these factors as well as a set of interactions among these factors, whereby the effect of one factor is, in turn, affected by another factor. For example, we investigated the primary effects of the year and the economic payment mechanisms on the average willingness to pay of a typical participant, and we also investigated whether the willingness to pay under each of the economic mechanisms, or payment rules, was different in the two years. We were also able to examine whether a participant’s choice in 2008 was affected by the mechanism that the participant was given in 2007.

The available data come from offers that individuals made in response to both the discrete-choice (DC) or “yes-no” response to a pre-determined, specific amount of money, as well as the open-ended (OE) choice presented wherein individuals could choose a payment amount from a short list or could name their own amount. Except in cases where an individual actually named his or her own amount (and backed it up with actual payment), these data only tell us whether the person’s actual willingness to pay for the contract is above or below the pre-determined amount in the DC format or whether it falls between two listed amounts in the OE format. (In the OE format, a person agreeing to pay $35 when the next higher amount listed was $50 is providing data showing that their actual willingness to make an offer is at least $35 and is less than $50.) The statistical analysis (an “interval regression”) accounts for this nature of the data.

The main findings follow.

First, unlike the hypothetical or stated-preference results summarized in Uchida et al. (2007), the details of the farm-wildlife contracts appear to have little effect on the willingness to pay for that contact. This result, however, may be due to the practical limitations of market choices enabled by the Jamestown experimental market. First, in a hypothetical survey, it is feasible for marketing analysts to present several choices to individual where each choice involves an explicit comparison between two farm-wildlife contracts, but only hypothetical statement of a willingness to pay. The explicit comparison creates a framework that encourages respondents to look at the details as Uchida et al. (2007) describes and as can be seen from an examination of the survey questions. In contrast, the direct-mail marketing of farm-wildlife contracts could only present each person with a single contract to consider, so the choice-situation did not explicitly encourage participants to consider the details (i.e., acres, territories, fledglings or road-views) of the contract. Second, with the scale of farming in Jamestown (one of 39 cities and towns incorporated within the small State of Rhode Island ), most contracts were very similar. For example, while in 2007 we were able to vary the acreage from about 6 acres to 18, in 2007 most contracts involved fields of 10 acres and in 2008 the market only presented contracts for fields of about 10 acres. These types of restrictions on the range of data limit the ability to identify, statistically, the effects of some details of the contract.

While the details of the contracts were generally not significant in influencing individual’s willingness to pay, there is some evidence that there are underlying effects. In 2008, the marketing materials included a more explicit statement of the number of bobolink fledglings that ecologists might expect to be possible from each bobolink territory that is defended and stated a range for the total that might be expected to result if the number of territories present on a field in 2008 was similar to that observed in prior years. In some of our statistical models, including the simplified model introduced below, this “fledglings” characteristic of a contract was statistically significant and increasing the expectation of fledglings protected under a contract influenced monetary offers positively. Moreover, in the 2008 data taken alone, the presence of “views” from roads as a contract-characteristic increased willingness to pay as well.
Next, results showed that the effect of the change in years was not a statistically significant influence on offers of payment for a farm-wildlife contract, including interactions between the year and other factors. However, this result carries one caveat. That is, we used an indicator variable to identify those individual participants in 2008 (i.e., those observations in the 2008 data) who had already been participants in 2007. Depending on which specific variable we included in the analysis, second-time participants, on average, were willing to pay about $10 to $30 less for a contract, on average. This result is consistent with concerns about the possibility that revenues would erode over time, but we are not currently able to link that effect to any particular economic mechanism.

Marketing firms provided some data on characteristics of individual households in Jamestown . Our analysis showed that most of the available characteristics seem to have little, if any, effect on an individual’s willingness to make a payment for a farm-wildlife contract. However, it appears that participants are willing to pay a lower amount if they come from a household where a member has recently completed mail-orders of some type of items for children; this result may be an indication of overall demands on the household budget. In addition, older individuals were, on average, willing to pay less for a farm-wildlife contract than were younger participants. Marketing data also includes a measure of a household’s “purchasing power,” as a measure of household income. Participants from households with higher purchasing power were more likely to make higher offers of monetary payments; however, this effect is somewhat weak, and would not be considered statistically significant by many analysts (although a statistician’s one-tailed test for a positive effect would be considered significant in many models at a 10% level or stronger).

We now discuss the effects of the economic mechanisms and presentations. The primary effects (main effects) of the mechanisms result in a statistically significant impact on the amount of money an individual offers. In addition, the DC presentation generates a statistically significant increase in the amount of money offered, on the order of $35 relative to an OE presentation with a minimum amount of $20 listed. The OE presentation with the minimum amount listed at $35 results in a statistically significant increase in the average offer of about $13 to $15 (almost exactly the difference between the $35 and $20 in the two OE presentations tested). However, interaction effects are generally not significant; in particular, the DC presentation did not result in a statistically significant impact on the payment-amount that individuals are willing to offer, on average, under each of the different economic mechanisms (UPA, UPC, PM) relative to the amount offered under the PR mechanism. That is, the DC presentation appears to add a statistically equivalent increase in offer-amounts across the four mechanisms tested. (However, there is some weak evidence that the increase under the UPC mechanism may be a bit larger, but this evidence is not supported by models that best account for the overlap in participant-groups between the two years.)

The lack of a statistically-significant difference in the performance of the PM mechanism with and without the DC presentation has an implication for the use of PM for a baseline on performance of the other mechanisms. If individuals’ choices under the PM mechanism are consistent with the theoretical property of incentive-compatibility, then differences in the offer-amounts generated under the PM mechanism should not be statistically significant between the DC and the OE presentations. Therefore, under theoretical conditions we would expect the interaction of the DC presentation with the PM mechanism to offset the main effect of the DC presentation relative to the average; this result does not appear, so individuals may not be responding under the PM mechanism in a manner that is completely consistent with incentive compatibility. Taking the Jamestown market as a “field experiment,” our result imply there may be additional factors affecting the PM mechanism; we suspect that the presentation and rules for payment under the PM are sufficiently novel that individuals are, at least, still exploiting the strategic opportunity to make a lower offer under the OE presentations.

As noted earlier, the average offer-amounts differ across the mechanisms. Further analysis provides evidence that the main difference is that the UPA mechanism generates lower offers, on the order of $15 lower relative to the other mechanisms. Although a statistical model that allows for differences across the mechanism yields mean estimates that rank offers under PR highest, followed by PM, followed by UPC, the difference across these three mechanisms is not statistically significant. The lack of a statistically significant difference between the UPC, PR and PM mechanisms is consistent with the concept that the revenue-raising mechanisms are eliciting offers that are consistent with individuals’ full value for a farm-wildlife contract; however, the caveat is that there is evidence mentioned above, particularly with regard to the DC presentation, that the PM mechanism may not actually be operating as one expects if the PM removes incentives to understate values. In addition, these results indicate that modifications to the UPA implemented for the UPC may have offset the tendency under of individuals to make lower offers under the UPA as compared to PR.

While the PR and UPC mechanisms perform statistically equivalently to the PM, it remains to examine how these mechanisms will perform over time. There is evidence in the experimental economics literature that groups of people may begin making decisions under PR which move toward the free-riding outcomes, meaning that revenue generation may fall off significantly over time. It remains to be seen whether the UPC might maintain a more consistent level of revenues, given that the cap-structure enables individuals with higher values for a public good to retain a substantial share of their net benefits. Alternatively, over time individuals may strategize with the UPC and attempt more offers below the cap despite having personal values that exceed the expected cap price. Such strategic behavior would again reduce revenue generation. However, the experience in Jamestown provides evidence that these mechanisms can work, at least at a level that may assist in the non-profit development of a direct exchange between ex-urban communities and their farmers.

References
Bollinger, E.K., P.B. Bollinger, and T.A. Gavin. 1990. Effects of hay-croppoing on eastern populations of the bobolink. The Wildlife Society Bulletin 18:142-50.
Swallow, Stephen K., Elizabeth C. Smith, Emi Uchida, and Christopher M. Anderson. 2008. Ecosystem Services beyond Valuation, Regulation, and Philanthropy: Integrating Consumer Values into the Economy. Choices 23(2):47-52.
Uchida, E., C. Anderson, et al. 2007. Establishing a market for ecosystem services from agricultural land: Stated preferences over elicitation mechanisms. Working paper, Kingston , RI , University of Rhode Island . 34pp.