Regional Food Hubs: the key to improved farm profitability and rural economic development?

To conduct an economic impact analysis, one must have information about the level of inter-industry transactions, or purchase and sales linkages. This information involves accounting relationships detailing the extent of purchases and sales of goods and services both within and among sectors of an economy. As a business buys from and sells goods and services to businesses in other sectors of the economy and to final users, the firm stimulates additional economic activity by the other businesses and within other economic sectors. Social Accounting Matrix (SAM) models are widely used by economists to measure and understand the distributional impacts or inter-industry linkages across an economy.

The SAM methodology’s analytical capacity lies in its ability to estimate the indirect and induced economic effects stemming from the direct expenditures associated with a change in final demand for the goods and services produced by an economy. An initial (direct) expenditure driven by a change in final demand sets in motion a cascading series of (indirect) impacts in the form of additional expenditures in other sectors by each business whose sales have increased. These direct and indirect effects are also associated with increased income to labor as a result of increased economic activity. To the extent that this additional income is spent within the local economy, there are additional multiplier effects that are commonly referred to as induced impacts. The cumulative impact across all of these affected industries determines the size of this initial type of multiplier, computed as the direct plus indirect plus induced effects divided by the direct effect.

The most commonly used SAM data and software is offered by a company called the IMPLAN Group LLC (henceforth ‘IMPLAN’). IMPLAN’s database represents the entire economy in 440 sectors. Each IMPLAN sector is represented by a single, static production function – a mathematical expression that relates the quantity of inputs required to produce that industry’s resulting output (Lazarus et al., 2002b; Liu and Warner, 2009). Initiated from a national table of accounts, IMPLAN provides a comprehensive set of balanced SAMs for every county and state in the United States. These SAMs illustrate a complete picture of the economy, accounting for all inter-industry transactions, as well as transfers to and from institutional sectors.

Throughout most previous research estimating the impact of local food systems, there are two main challenges that reflect the difficulty in meeting the significant data requirements to conduct rigorous economic impact assessments. The first is what O’Hara and Pirog (2013) refer to as an ‘interpretation’ challenge. Specifically, “a critical issue for measuring net economic impacts entails stipulating how the ‘opportunity cost,’ which is what would have occurred in the absence of local food sales, is defined” (p.4). As they rightly point out, measuring opportunity cost is not straight-forward, and requires information about the extent to which increased consumer purchases of locally-grown food offsets other types of purchases, changes market prices and/or supply chain characteristics, or impacts land use. There are only a handful of local food economic impact assessments that explicitly acknowledge the need to consider opportunity cost (Conner et al. 2008; Hughes et al. 2008; Gunter & Thilmany 2012; Tuck et al. 2010; Swenson 2010). However, each of these studies makes assumptions about the sectors in which there are decreased purchases (or changes in land use) as a result of increases in local food consumption–in other words, none collects the data necessary to more fully understand the opportunity costs of increased local purchases.

The second challenge is that almost all of these studies reflect the implicit assumption that local food system participants have the same patterns of expenditure as the aggregate agricultural sector data available in IMPLAN. The production functions for each sector reflect average purchase and sales patterns across all firms in the sector, without the requisite information to be able to disaggregate them by any specific characteristic (i.e., scale of operation, or marketing channel). As IMPLAN sector data represents all inter-industry linkages, the expenditure and sales patterns are more reflective of those firms that contribute a higher proportion of total output in the sector (typically, the larger firms). Given that local food system participants tend to be smaller in scale, and represent a small overall portion of agricultural sector transactions (Low & Vogel 2011), the estimates of the impacts from increased local food sales based on existing IMPLAN data may be misleading if local food system participants have different patterns of input expenditures (e.g., different production functions) and/or they purchase a different proportion of their inputs from local sources.

IMPLAN does not include a ‘food hub’ sector within its 440 sectors. Rather, the expenditures and sales of food hubs are included within more aggregated sectors (i.e., wholesale trade). As a food hub sector is unlikely to ever exist either within IMPLAN, or in any Bureau of Economic Analysis data set, we do not construct a ‘food hub sector’ within IMPLAN, but instead model the sector based on the expenditure profile of food hubs. IMPLAN refers to this type of modeling as ‘Analysis-By-Parts’ (ABP).

We build two alternative ABP SAM impact assessment models using IMPLAN—one that incorporates additional data collected from farms selling to the food hub (Model 2) and one that does not (Model 1). Thus, the implicit assumption in Model 1 is that food hub farms have the same or similar production function to the default IMPLAN agricultural sector data. Whereas, in Model 2 we utilize additional data collected from farms selling to the food hub to create a distinct ‘food hub farm’ sector.

Given the heterogeneous structure of food hub operations and the detailed data needs required for an impact assessment, we utilize a case study approach. Regional Access, LLC (RA) was chosen for our case study due to their commitment to working directly with local farmers, their length of time in operation, the diversity of their customer base, and the size of their operation. RA was established in 1989. In 2011, it had over $6 million in sales, and employed 32 full-time equivalents. Utilizing 9 vehicles and a 25,000 square foot warehouse, RA aggregates and delivers products primarily throughout NYS. RA has over 3,400 product listings, including beverages, breads, cereals, flour, meats, produce, prepared foods, grains, and fruits and vegetables. RA purchases products directly from 96 farm vendors (86 in NYS) and 65 specialty processors (non-farm vendors), as well as from conventional sources. RA has over 600 customers, including: individual households, restaurants, institutions, other distributors, fraternities and sororities, buying clubs, retailers, manufacturers, and bakeries. RA also provides freight services to a range of businesses.

RA plays an important role connecting farmers, customers, and the community-at-large around food and agricultural issues. RA fits within USDA’s definition of a food hub given its commitment to building relationships with local farmers, managing the aggregation, distribution, and marketing of their products, and maintaining the farm’s identity.

RA provided a detailed 2011 profit and loss statement, along with estimates of the percentages expenditures in each category that were local (i.e., in NYS). Based on the data they provided and follow up discussions with the RA personnel, the hub’s expenditure categories were mapped to IMPLAN sector, value added, and other components.

Understanding how ‘food hub farms’ interact with other sectors of the economy is important in improving the precision of an impact assessment. Accordingly, we conducted interviews with 30 food hub farms. The farms represent 35% of all farms selling to RA that were located in NYS and sold more than $100 of products to RA in 2011. Farms interviewed were located in every region in NYS except New York City and Long Island. In terms of farm sales, 37% were classified as ‘small’ ($1,000-$249,999 in gross sales), 43% were classified as ‘large’ ($250,000-$999,999 in gross sales), and 20% were classified as ‘very large’ ($1 million or more in gross sales). The distribution of farms by primary commodity category was 37% meat and livestock, 30% fruit and vegetable, and 33% value added products (including cheese, butter, yogurt, honey, maple syrup, wine and juice).

Once we complete mapping of the food hub farms expenditures and sales to the relevant IMPLAN sectors, we see that our case study farms have very different patterns of expenditure than the default food sold-farm sector within IMPLAN (see Table 4). Most importantly in terms of local economic impact, per unit of output, food hub farms spend $0.77 in the local economy versus the $0.54 in the food sold-farm sector. Food hub farms spend double as much on employee compensation as the default food sold-farm sector per unit of output ($0.24 compared to $0.12), although they spend substantially less on proprietor income ($0.06 compared to $0.16). Another way of comparing these expenditure patterns is that per unit of output, the income impact of food hub farms is $0.30 compared to $0.28 in the default IMPLAN data. Food hub farms spend $0.08 per unit of output on support activities for agriculture and forestry, compared to $0.02 in the default agriculture sector per unit of output. And, food hub farms spend $0.14 per unit of output on purchases from other local farms (both food hub farms and other farms) compared to $0.06 per unit of output in the default farm sector.

RA’s customers were surveyed using an online survey to better understand the extent to which purchases from RA increase the demand for locally-grown farm products and offset purchases from other sectors. At the time of the survey, RA customers numbered 110 households and 547 businesses. Of these respondents, 57 households and 186 businesses responded to the online survey. To improve the response rate for business customers, follow up phone interviews were attempted with those customers who did not respond online. An additional 62 surveys were completed, increasing the total number of responses received to 305 (46% response rate), with 80% from business customers and 20% from individual households.

RA’s business customers are very diverse. They reported average annual gross sales of $5.7 million (median = $515,000, n=101), with a range from $3,000 to $414 million. On average, they have been in business 13 years (median = 8 years), although this ranged from new to over 130 years in operation (n=151). The average number of fulltime employee equivalents was 15 (median = 4, n=145). Business customers were also asked to identify the function their business most often performs; accordingly, 2% identified themselves as distributors, 3% as grocery/meal delivery service providers, 9% as processors/manufacturers, 11% as wholesalers, 25% as restaurants, 34% as retailers, and 17% as other—including bakery, fraternity/sorority house, caterer, coffee shop, farmers’ market vendor, and institutional cafeteria (n=245).

To understand the impact of an increase in final demand for RA food hub products and the extent of differential economy-wide impacts from the two models, we consider a scenario in which an exogenous shock increases final demand for food hub products and services by $1,000,000. The only difference between the allocation of the shock in Models 1 and 2 is that in Model 1 all local farm purchases by RA are allocated to the default agricultural sector IMPLAN data, whereas in Model 2, RA farm purchases are allocated to the food hub farm sector.

In addition to the positive shock, we consider a simultaneous negative shock to the wholesale trade sector in order to account for the opportunity cost. The customer survey results reveal that, on average, 49.39% of businesses decreased their purchases from other distributors due to their purchases from RA. Of those who reported decreasing purchases from other distributors, the average decrease was 23.09%. Accordingly, a negative shock of $114,042 was applied to the wholesale trade sector.

The above represents an abbreviated version of our methodological approach. Complete information will be made available via the USDA AMS report as soon as it is released. When the report is made public, a link will be provided here.