Hosting a Market Is Just the First Step: Exploring the Relationship Between Community Characteristics and Farmers Market Size

INTRODUCTION

Environmental justice scholarship has long noted disparities that exist in food systems across space. ^{1 , 2 , 3 , 4} For example, Gripper et al. (2022) describe a system of food apartheid that negatively affects certain neighborhoods. ⁵ Current research, often using geographic information systems analysis, backs up this claim and has consistently found that neighborhoods with high proportions of Black and low-income residents have disproportionately high numbers of fast food restaurants, lower quality foods available to purchase, and lower numbers of grocery stores.^6–8

The modern local food movement emerged in the late 1990s as a way to increase access to healthy and nutritious foods in communities historically underserved by other food outlets. ⁹ Within the local food movement, the farmers market (FM)—public temporary assemblages where farmers market their own goods directly to consumers, and which are often hailed as the flagship of the movement—has been motivated by goals related to environmental justice, sustainability, and economic development. ^{10 , 11}

Access to FMs has been found to increase fruit and vegetable consumption among patients at family planning clinics, pregnant women and women, infant, and children recipients, and residents of low-income communities. ^{12 , 13 , 14} Rural food producers benefit from the expansion of local markets for farm products, and FMs inject dollars into regional economies through income substitution. ^{15 , 16}

Numerous case studies have also detailed how vibrant FMs can bring foot traffic to downtown areas, jumpstart local food economies, and provide jobs and volunteer opportunities to residents. ^{17 , 18} Driven by consumer demand, hard work from farmers, and strong support from health and economic development advocates, the number of FMs in the United States increased from 1755 in 1994 to 8771 in 2019, and sales totaled $711 million in 2015. ^{19 , 20}

The success and popularity of FMs, however, have led researchers to ask whether the benefits that FMs provide for host communities are skewed toward members of already privileged social groups. ^{21 , 22} Existing studies that use large-scale census data to examine where FMs are located give good reason to think that communities with relatively high percentages of college-educated and White residents are significantly more likely to host an FM. ^{23 , 24 , 25}

But this conclusion, although sobering, only partially addresses the question of whether and to what extent inequalities exist in access to the social and environmental goods that FMs provide. This is because there are many salient dimensions to FMs, beyond the basic question of whether or not they exist at a certain location. FMs are not all the same, and indeed, one of the most significant dimensions on which FMs differ is also one that is directly relevant to their ability to make available key goods, and deliver benefits, to host communities and their residents.

Specifically, FMs can differ dramatically in their size, as measured by such characteristics as the number of farmers and other vendors, and the number of types of goods—including fresh fruits and vegetables, meat and dairy products, and prepared foods and value-added products—that are sold. ^{26 , 27} More farmers and food vendors, in most cases, would mean a greater potential supply of fresh and healthy food.

Larger more visible markets would tend to constitute a more enticing destination for relatively casual shoppers, thus potentially leading to more patronage—and greater consumption of healthy food—by people who might otherwise shop only at stores where processed foods are more prevalent. More vendors may also be associated with a greater variety of healthy food products, and increased competition among vendors may drive down prices for consumers.

The size of an FM would also multiply in straightforward ways its economic impact on the host community. Larger markets generate more foot traffic, create reasons for people to spend time in traditional downtown business districts, and potentially lead to the circulation of more dollars among small businesses. ^{28 , 29} It is no exaggeration to say that a large and thriving FM, by attracting hundreds of visitors over the course of an entire day, can have a significant impact on the trajectory of entire neighborhoods. ³⁰

Where the benefits of FMs for consumers and communities are concerned, FM size may be nearly as important as whether a market exists in a particular location at all. Therefore, it is perhaps surprising that no previous study has used nationwide or multistate data to look at the relationship between market size and community social, economic, and demographic characteristics. Indeed, to the best of our knowledge, only one study has specifically focused on environmental justice questions related to FM size.

Lowery et al., using data from an audit of FMs in Los Angeles, conclude that FMs in low-income and non-White communities are smaller, with fewer vendors and less space devoted specifically to fresh fruits and vegetables, than those in high-income majority-White communities. ³¹ However, the relatively small sample size and restricted geographic reach of the data for the Lowery et al. study both makes it difficult to draw more general conclusions about regions other than Los Angeles, and restricts the analysis to examinations of bivariate relationships.

Further research is needed to explore whether market size is related to community social privilege while controlling for other factors, such as market history and organizational characteristics, that might affect how many vendors FMs can attract and retain.

Our contention in this article is that FM size, in terms of number of vendors and number of kinds of goods for sale, is likely to be largely a function of three kinds of factors. First, FMs are better able to attract and retain vendors if they are well-organized and professionally run. For a farmer with a small or medium-sized operation who is interested in direct-to-consumer markets, selecting the right FM (or multiple markets) is a decision that can mean the difference between succeeding or failing as a farm.

FMs that can clearly communicate to vendors what is needed to succeed at a particular location, and which themselves give the impression of being on firm financial and organizational footing, seem likely to have the best chance of impressing potential vendors as a promising place to do business. A coalition of state agencies, private foundations, and farmers groups in Vermont, for instance, recommends written bylaws, rules, and a board of directors as key steps toward recruiting vendors and creating a thriving FM. ³²

These same resources and dimensions of human capital may also give FMs a relatively good chance of actually succeeding, thereby creating a virtuous circle where existing vendors do well enough to stay on, and new vendors see this success as reason to try to associate themselves with the FM in question.

Second, straightforward access to a large pool of potential customers seems likely to constitute a draw for potential vendors. FMs can be located, and do well, in a wide range of geographic areas, from dense urban cores to remote and small rural towns. But just as the number of grocery stores in an area correlates with the number of residents, FMs that can potentially serve as a shopping destination for relatively large numbers of people would likely attract and be able to accommodate more vendors, including farmers. ^{33 , 34}

Third, much of the literature examining FMs in the context of environmental justice questions has focused on whether the racial composition and economic resources of communities are associated with where FMs are located. FMs are in many respects an emblem of the local food movement, and the social justice aspirations of local food advocates come through in claims that FMs can and do bring fresh fruits and vegetables and other farm products to consumers in communities that are underserved by conventional stores. ^{35 , 36}

Food deserts—areas where convenient access to healthy food at stores and restaurants is severely limited—are associated with relatively high percentages of non-White residents and high levels of racial and economic segregation.^37–39 So if farmers who prioritize selling to local consumers are also strongly committed to making healthy food more available to underserved consumers, then proximity to majority non-White low-income communities may be positively correlated with FM size.

In contrast, it has also sometimes been argued that FMs are “white spaces” that serve mainly White highly educated high-income populations. While noting that White farmers and customers “dominate” FMs and community supported agriculture in California, Alkon and McCullun argue that the “affluent, liberal habitus of whiteness” at FMs involves more than “just the presence of pale-skinned bodies.” ⁴⁰

Rather, the whiteness of local food spaces emerges from practices and symbols that signify social privilege, and which have the effect of excluding, if informally and unintentionally, people who do not share in this privilege. Theory on the whiteness of alternative food institutions gives reason to hypothesize that FMs are likely to be larger in White communities, compared with communities that differ in racial composition but are otherwise similar in many ways.

In sum, a number of different kinds of factors—internal FM organization; proximity to potential customers; and community demographics, among others—might reasonably be expected to affect FM size. Little if any research, however, has explored this question systematically and using a large geographically diverse sample of markets and places where they are located. That is the task that we undertake in the rest of this article.

METHODS

Three data sets were used to address questions of interest for this study. First, a survey was conducted in 2018 of key agents—managers and board members—at FMs in Massachusetts (N = 231), Rhode Island (N = 38), Connecticut (N = 125), New Hampshire (N = 64), Vermont (N = 59), Iowa (N = 155), Kentucky (N = 135), Colorado (N = 86), and Oregon (N = 133). A list of FMs in these states (N = 1026) was generated using United States Department of Agriculture (USDA) and state databases of FMs and local food markets.

Key agents for FMs were identified using contact information from these databases, as well as FM websites and social media. Potential respondents were invited by e-mail to take a survey about their FM. Following Dillman's Tailored Design Method, ⁴¹ respondents received follow-up invitations to take the survey up to four times by e-mail, and, finally, a mailed paper copy of the survey with a $2 incentive for participation. Once data had been cleaned for duplicate entries, the response rate for FMs was 54.7% (n = 561).

All FMs identified in the study area were geocoded and linked to census tracts using two publicly available data sets: the American Community Survey (ACS) 2014–2018 5-year estimates, and the USDA Rural–Urban Commuting Areas (RUCA) Codes from 2010. The nine-state study area region contained 7046 census tracts, of which 468 hosted an FM, and 93 hosted two or more FMs.

Given the diversity of states included in the study, statistical tests, including t-tests and chi-square, were used to compare census tracts with an FM that responded to the survey with tracts with an FM that did not respond to the survey; no significant differences were detected, suggesting that the final survey sample was representative of the population of FMs and their neighborhoods in the states that were selected.

Variables for statistical analyses were generated from both the original survey and census data sets. Data from the original survey were used to create variables for each FM that measured (1) number of vendors present; (2) number of kinds of goods available, including fruit, vegetables, cheese, meat, eggs, prepared goods, canned goods, mushrooms, flowers, plants, and beverages, and an option for any other good not included previously; (3) years the FM had been in existence; (4) organizational strategies employed at an FM (dummy variables), including using a vendor handbook, vendor application, tracking of vendor sales, written bylaws, board of directors, professional account, strategic/business plan, and a producer-only rule.

The “number of vendors” variable initially included 10 FMs with more than 100 vendors that were 5 standard deviations (SDs) above the mean; these outlier FMs were found to significantly skew results of statistical tests, and were, therefore, not included in regression analyses. Census data from the ACS were used to create variables for the percentage of White residents and the overall socioeconomic status (SES) for each census tract. The SES latent variable was created by standardizing (mean = 0; SD = 1) and averaging five manifest variables from the ACS: median household income, average education level, median home value, median rent, and proportion of individuals “doing okay economically” (Cronbach's α = 0.91).

Other social and demographic variables from census data, including measures of social inequality and neighborhood quality, were tested but ultimately not included in regression models due either to high multicollinearity with SES, or to results from goodness-of-fit tests that indicated that their inclusion did not improve the predictive power of models. Finally, RUCA data provided a measure of the rural-urban character of census tracts.

Two statistical analyses were conducted to examine possible relationships between FM size and community and FM characteristics. Ordinary least squares (OLS) regression models were generated using the number of vendors as the dependent variable. Poisson regression models were generated using the number of kinds of goods available as the dependent variable.

Based on the literature review and theoretical framework discussed earlier, we hypothesized that both the number of vendors and the number of kinds of goods available would (1) increase as the percentage of White residents and neighborhood SES increase; (2) increase for FMs that used organizational best practices to recruit vendors, manage finances, and plan for the future; (3) increase as the number of years an FM has been in existence increases; (4) decrease in areas of high-commuting or noncore areas of cities and towns.

RESULTS

Table 1 reports descriptive statistics for variables used as outcomes in regression models. Fruits (95%) and vegetables (97%) were almost universally available at FMs. Eggs (85%), prepared goods (89%), plants (81%), meat (68%), canned goods (61%), and flowers (74%) were available at most FMs, whereas beverages (55%), cheese (34%), mushrooms (31%), and other goods (36%) were more selectively available.

The average FM had 17.93 vendors. Table 2 reports descriptive statistics for variables used as explanators in regression models, as well as for component measures of the index variable for SES. Table 1 shows that neighborhoods with an FM were slightly whiter and had a lower SES than the average census tract in the entire nine-state study area. RUCA data show that FMs are located mainly in core areas of large cities, small cities, and towns.

Table 3 reports the results of the OLS and Poisson regression models. Model 1 shows the results of the OLS model where the dependent variable was the number of vendors present. Findings were generally in line with hypothesized relationships: percentage of White residents in a census tract, tract SES, and FM years in existence, all had a significant positive relationship with the number of vendors at an FM.

For every 1% increase of White residents in a neighborhood, the number of vendors at the FM in that neighborhood increases by 0.127. For every 1 unit increase in tract SES, the number of FM vendors increases by 1.46. Every year of existence for an FM is associated with an additional 0.24 vendors. FM location was also statistically significant in several cases. Compared with FMs located in census tracts at the core of metropolitan areas, FMs had fewer vendors when located in tracts in high-commuting metropolitan areas, small town cores, and rural areas.

A number of FM organizational strategies were also positively associated with the number of vendors. The model predicts a statistically significant increase of vendors for FMs that had a vendor handbook (+4.46), a vendor application (+6.20), a board of directors (+4.77), and used professional accounting (+5.28) compared with those that did not use each organizational strategy.

Table 3 shows the results of the Poisson regression model where the dependent variable was the number of goods available (Model 2). Similar to Model 1, Model 2 finds a statistically significant positive relationship between the percentage of White residents in a neighborhood and the number of types of goods offered. For every 1% increase of White residents, the number of types of goods at an FM increases by 0.004.

Among FM organizational strategies, however, only having a vendor application had a statistically significant positive relationship with the dependent variable; older FMS also were larger in terms of number of goods available. FMs in both small town core and small town high-commuting areas had significantly fewer kinds of goods, compared with FMs in the metropolitan core. SES status (p = 0.06) was positively related to the number of goods at an FM, but the effect was not statistically significant at the p < 0.05 level.

DISCUSSION

The purpose of this research was to explore how neighborhood characteristics and market organizational strategies influence the size and scope of FMs. Study findings largely support our hypotheses. In line with Hypothesis 1, both models presented in Table 3 predict positive relationships between number of vendors present and the number of types of goods offered with the percentage of White residents and SES level.

Each was found to be statistically significant in Model 1 and percentage White was also significant in Model 2. Results also offer support for Hypotheses 2 and 3. The models in Table 3 show that years of existence and a variety of organizational practices were associated with significant increases in the number of vendors. Impactful practices included having a vendor handbook, vendor application, or a board of directors, and using professional accounting or a possessing a strategic/business plan.

Years of existence and the presence of a vendor application were also associated with significant increases in the number of types of goods. Finally, the results in Table 3 provide some support to Hypothesis 4, in that being located in metro area, high-commuting or rural areas were found to significantly decrease the number of vendors at FMs, as was small town high commuting for the number of types of goods offered. However, the fact that location in small town core areas was found to significantly reduce the outcome variables in both models did not conform to our expectations.

As a whole, study findings highlight two trends in the data that can be taken to represent positive news for FM proponents specifically and local food proponents generally. First, there are several concrete organizational best practices at FMs that are associated with relatively large market size in terms of number of vendors. In theory, the direction of causality in this relationship could run either way; it could be that markets do not adopt practices such as having a board of directors or a vendor application system until after they have already grown quite large.

Also possible, however, that FMs that are well-organized and professionally run have greater success in attracting a relatively large number and, in some cases, broad diversity of vendors. To the extent that the cause and effect in this relationship start with practices and points toward size, our study strongly suggests that the resources and time required to implement these best practices are well worth the investment.

Crucially, our study also gives reason to think that these organizational best practices can be effective in increasing the size of FMs, regardless of the racial or ethnic composition or SES level of a particular community. Whether a neighborhood is mostly Black or mostly White, high-income or low-income, our data are consistent with the idea that running an FM in a professionalized business-oriented way is likely to increase the chances of the FM attracting a sizeable number of vendors—and, therefore, customers.

Second, the results of Model 2 indicate that the impact of neighborhood characteristics on the variety of types of goods offered at FMs is, practically speaking, relatively small. For instance, Table 4 shows that although the percentage of White residents variable was statistically significant, our modeling predicts only an increase of 0.37 types of goods for a neighborhood composed of 100% White people when compared with a neighborhood with no White people. We would contend, therefore, that Model 2 suggests that markets, regardless of location, are offering a similar breadth of good type offerings.

Although there are these positive trends in our data, we note that overall, the results of this study do conform to the environmental justice literature's continued theme of significant disparities in FM depth (i.e., number of vendors present) by neighborhood type. Unlike the types of goods model, Tables 3 and 4 show the predicted impact of neighborhood characteristics substantially affects the predicted number of vendors present during a market session, even after taking into consideration the organizational best practices that FMs have at their disposal.

To show this, the right side of Table 4 shows the predicted decrease or increase in the number of vendors based on a neighborhood's percentage of White residents and SES percentile. Consistent with the broader literature on FMs, there is a positive effect of affluence on the vendor size. Most striking is the magnitude of the model's predictions of the increased number of vendors as the proportion of White residents in a neighborhood increases.

For example, a neighborhood with 70% White residents would be predicted to have an increase of almost twice the number of vendors when compared with a neighborhood made up of 30% White residents. We contend that Tables 3 and 4 show a penalty for low SES neighborhoods, and a reward for whiteness, as to what we have termed the depth of an FM. This relationship is the inverse of that which many would argue is called for by principles of environmental justice.

To the best of our knowledge, this study is one of the first to highlight the importance of FM size and scope as variables that merit the attention of environmental justice researchers. Given the affirmative findings of inequities in access to larger FMs, future research could expand on this study in a number of ways. In particular, although key agents at FMs in nine states were surveyed, this group does not capture every U.S. region where FMs are located.

Greater geographical diversity in the data would strengthen the generalizability of findings. Data that captured the evolution of FMs over time would also help to resolve the question of whether FM organizational best practices contribute to success and growth—or whether success and growth provide FMs with the resources needed to adopt practices that can secure their stability.

CONCLUSION

The results of this study support the idea that access to FMs ought to be treated as a multidimensional concept encompassing more than just whether an FM is located in a particular place. FM size, measured by number of vendors and diversity of goods in this study, was found to vary in our data, when considering racial and socioeconomic composition of neighborhoods and by the organizational characteristics of markets themselves.

With these central findings in mind, it is worth returning to why the issue of FM size and good availability deserve more sustained attention from environmental justice researchers. We contend that larger and more robust FMs benefit consumers in that there are more vendors to shop from and more goods to select. Equally important is that large, vibrant, and prosperous FMs seem more likely to offer the noneconomic benefits seen in the literature, such as fostering rich social network connections between farmers and residents, and helping to build community institutions that extend beyond the market itself.

An FM is a place for people to gather, socialize, and get to know local farmers. ^{42 , 43} More broadly, a successful FM acts as a magnet for bringing people to main streets and downtown areas—people who might go on to spend time and money at other local businesses. In short, if FMs are an engine for generating social, economic, and community capital while addressing issues of environmental justice—then the larger the FM, the more the capital that can be generated and reinvested back into communities.

For environmental justice, it matters not just whether an FM exists, but what kinds of FM exist. We encourage policy makers and local food advocates to think beyond the location of FMs and to consider mechanisms that can contribute to increasing robust vendor participation from the onset, particularly in communities with relatively large proportions of low-income and non-White residents. For instance, a percentage of public and private funds could be set aside for expanding FMs where they already exist, in addition to starting new FMs in communities where they are currently missing.

In addition, this study shows that implementing effective organizational practices, such as using a vendor application and professional service providers, could benefit the breadth and depth of FMs. Encouraging practices such as these would be a useful focus moving forward, especially if the goal is not just to create new FMs, but also to increase access to healthy food by adding vendors and goods to markets that already have a foothold in food-insecure communities.