Source: UNIV OF CONNECTICUT submitted to NRP
ECONOMIC RURALITY: USING INFOGROUP TO CONSTRUCT NEW MEASURES OF RESILIENCY, DEPENDENCE, AND CONNECTEDNESS
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
Reporting Frequency
Annual
Accession No.
1025384
Grant No.
2021-67024-34435
Cumulative Award Amt.
$500,000.00
Proposal No.
2020-04604
Multistate No.
(N/A)
Project Start Date
Mar 1, 2021
Project End Date
Feb 28, 2025
Grant Year
2021
Program Code
[A1661]- Innovation for Rural Entrepreneurs and Communities
Recipient Organization
UNIV OF CONNECTICUT
438 WHITNEY RD EXTENSION UNIT 1133
STORRS,CT 06269
Performing Department
Agricultural Resource Economic
Non Technical Summary
The economies of rural America continue to lag those in metropolitan areas and some rural regions are experiencing significant economic hardship. It is increasingly recognized that identifying the attributes of rural economies that explain why they lag their urban counterparts and the characteristics that cause some rural areas to flourish while others flounder are critically important lines of inquiry in public policy research. While social science scholars have begun to examine these issues, it is widely recognized that existing urban-rural place typologies do not adequately capture the profuse economic, sociological and demographic heterogeneity among non-metropolitan areas of the country. This project seeks to use InfoGroup data to generate a new place typology that will complement existing place typologies currently published by USDA ERS by incorporating supply-chain connectedness in a spatial gravity model. These will generate a hierarchy of Economic Catchment Areas (ECAs). Once defined, we will construct standard measures of economic activity and economic dynamism by aggregating InfoGroup establishment data to the ECA-level. We will then use these measures for two purposes. First, we will estimate the relationship between economic growth and supply-chain connectedness with particularly emphasis on the role of establishments in food-and-agricultural industries. We will also develop a web-based platform that will allow users to select the gravity model parameters that define the ECA hierarchy, visualize the spatial distribution of ECAs through a mapping tool, and download data via an extract function for their own research or policy evaluation needs.
Animal Health Component
100%
Research Effort Categories
Basic
(N/A)
Applied
100%
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
60360503010100%
Knowledge Area
603 - Market Economics;

Subject Of Investigation
6050 - Communities, areas, and regions;

Field Of Science
3010 - Economics;
Goals / Objectives
The economies of rural America continue to lag those in metropolitan areas, and some rural regions are experiencing significant economic hardship:In 2018, total employment in rural America was two percent lower compared to 2008; in contrast, employment grew by eight percent in metropolitan areas over the same period.Between 2015 and 2016, nominal wages in rural America grew by 3.8 percent, compared to 5.5 percent in mid-sized metropolitan areas (250,000 to 1 million residents) and 4.9 percent in small metropolitan areas (less than 250,000 residents).Population growth in rural America was negative every year between 2010 and 2016. 2017 witnessed the first population increase of the decade. In contrast, the annual population growth rate in metropolitan areas during this period has averaged 0.8 percent.The labor force in rural America decreased 2.5 percent between 2008 and 2016, compared to a 4.5 percent increase in the labor force in metropolitan areas. It is increasingly recognized that identifying the attributes of rural economies that explain why they lag their urban counterparts and the characteristics that cause some rural areas to flourish while others flounder are critically important lines of inquiry in public policy research. While social science scholars have begun to examine these issues, including shifting rural-urban boundaries and urban-to-rural migration (e.g., Lichter and Brown 2011; Nelson, Oberg, and Nelson 2010), rural development policy (Kandilov and Renkow 2010; Rasker et al. 2009), and the features of rural entrepreneurship (Acs and Malecki 2003; Ring, Peredo, and Chrisman 2010), three deficiencies in the federal statistical system hamper research efforts: 1) a lack of systematic data collection and reporting on economic activity within the food-and-agriculture supply chain; 2)inadequate reporting of economic activity in rural areas; and 3) place typologies that obscure profuse economic, sociological, and demographic heterogeneity among non-metropolitan areas of the country. Historically, the connection between farming (economic activity) and rurality (demographic concept) underlies the rationale for USDA administering rural development programs. Increasingly, however, rural areas have become less dependent on agriculture as a source of economic activity. Further,researchers recognize that available urban-rural taxonomic systems increasingly strain to capture the complex interaction of economic, sociological, demographic, and geographic attributes that define place in America. In the past, the delineations between urban and rural were stark along these dimensions. Demographically, urban was synonymous with densely populated, while rural areas were sparsely populated. Economically, urban was synonymous with manufacturing and trade, while rural was associated with agriculture, forestry, and resource extraction. Moreover, urban-rural distinctions based on demographics or economics mostly largely aligned. As a result, both the definitional (sparsity, farming) and consequential attributes (e.g., higher fertility, higher out-migration, limited economic opportunity) summarized by the category rural were widely recognized. Today, the distinctions between rural and urban places are increasingly blurred. Rural communities support a more diverse set of economic activities, while urban agriculture has gained momentum. Further, the problems of population loss and economic stagnation associated with rural America are increasingly experienced by urban areas of all sizes. Consequently, researchers generally agree that the previously clear contrast between place type is now a complex set of interdependent layers, rendering traditional measures of "rural" increasingly less accurate and less useful for social scientists and policymakers (Champion and Hugo 2004 ; Kulcsár and Brown 2011 ; Lichter and Ziliak 2017). Although there is general agreement that binary approaches (urban vs rural) are no longer applicable, there is no consensus in the research community on alternative, improved measures of the rural-urban gradient. Until relatively recently, most analytic attention in the construction of place typologies focused on the urban end of the rural-urban continuum. As a result, available measures fails to account for the multidimensional quality of rural places or are organized at geographic aggregations that masks important sub-unit heterogeneity or both (Cromartie and Bucholtz 2008), e.g. MSAs and Rural-Urban Continuum (RUC) Codes. Despite their widespread use, both these place typologies exhibit recognized shortcomings. First, they mask sub-county heterogeneity, which is problematic in contemporary rural America since many "metropolitan" counties contain rural territories and populations. Second, in both taxonomic systems rural is simply the residual category, i.e., the populations and places that are not included in an urban designation, rather than a class of place defined on its own terms. Acknowledging the complexity of social and economic trends that define place requires researchers to develop taxonomic systems that better organize the economic, sociological, and demographic attributes of where Americans live and work, thereby generating the research structure to better assess the forces underlying the disparate fortunes among rural places (Curtis and Kulcsár 2019).Recognizing these issues, researchers at ERS have taken a leading role in developing more nuanced classification schema: RUCA codes, Typology Codes, andFrontier and Remote (FAR) Area Codes. Although each of these taxonomic systems are valuable improvements as we move away from the overly simplistic rural-as-residual and rural-as-farming frameworks, it is also clear that they do not exhaust all salient environmental-socioeconomic relationships characterizing place in America. Therefore, we propose to create a taxonomy of place based on economic interdependence that complements the systems described above. Specifically, we will offer a hierarchical classification scheme that links geographic units through the supply of intermediate production inputs. This projectscontinues work toward the long-term goals of better documenting the sources of economic growth in rural America by using InfoGroup data to generate new place typologies defined by economic connectedness. Therefore, we propose the following specific project objectives:Revise the InfoGroup dataset to act as a synthetic LBD by redefining key variables of analysis (establishment and firm identifiers, industry codes, location of operation, and employment);Generate Economic Catchment Area (ECA) codes that complement existing place typologies currently published by USDA ERS by incorporating supply-chain connectedness as calculated in InfoGroup using establishment industry codes and BEA input-output tables;Construct measures of economic activity (sales, employment, number of establishments, productivity) for place categories by aggregating InfoGroup establishment data. Further, we will harness the longitudinal nature of InfoGroup to construct measures of economic dynamism (growth rates of employment, sales, and productivity; entry and exit rates) for place categories;Identify ECAs that are strongly associated with the food-and-agricultural supply chain and then estimate the relationship between economic growth in FAI and economic connectedness;Develop a web-based platform that allows users to explore project data and download outputs. An interactive mapping component will allow users to examine the spatial properties of different place typologies and their relationship to one another. Users will be able to adjust the parameters that define ECA codes to see changes to their spatial extent in real time. A data extract function will enable users to download project datasets at a variety of geographic scales for their own use.
Project Methods
1. Revise the InfoGroup dataset to act as a synthetic LBDAlthough the InfoGroup data is collected, organized, and published for very different reasons using very different approaches than federal administrative and survey data, in many important respects it is similarly structured to the Longitudinal Business Database (LBD). We thus propose to build a crosswalk between the LBD and InfoGroup that effectively makes the latter a synthetic version of the former:2. Generate Economic Catchment Area (ECA) codesWe propose to use core units in the RUCA codes system as the basis for building Rural-Urban Economic Catchment Areas codes from InfoGroup and the Census of Agriculture. Specifically, we will define a primary economic catchment area (ECA) centered on each small UC core (RUCA 7) that includes associated commuting Census tracts (RUCA 8 and 9) and nearby rural tracts that include large employers in InfoGroup. We will use the Census of Agriculture to identify the primary crop and livestock production activities and the County Business Patterns to identify the presence of forestry and mining activity in rural Zip codes surrounding the primary ECA (the smallest published unit for both programs is the Zip code). We anticipate that for the overwhelming majority of the country, using the county as the basic unit will suffice for identifying agricultural, forestry, and extraction, but western counties are sufficiently large to benefit from a more disaggregated reporting unit. We will then use the most recent Use Tables published by BEA as part of its Input-Output Accounts program (2018 for 71 two- and three-digit NAICS industries and 2012 for 415 four-digit NAICS industries) to calculate absolute and relative rural use coefficients for each primary ECA with respect to the agricultural, forestry, and extraction (AFE) activity in its vicinity. We then will then assign remaining rural tracts to primary ECAs using a gravity model based on the use coefficients defined above. Gravity models have been applied as far back as the 1850s (Carey, 1858) to describe social phenomena and continue to be used across disciplines (Erlander, 1980; Haynes and Fotheringham, 1984; Sen and Smith, 1995; Wilson 2000), including contemporary analyses of international trade (USITC, 2020). We will then move to assigning the remaining Micropolitan and Metropolitan commuting tracts. We will then move to define secondary ECAs. To do so, we will again treat the United States as a plane with nodes defined by Micropolitan cores and primary ECAs attracted to those nodes by total use coefficients. Primary ECAs are assigned to secondary ECAs centered on Micropolitan cores when gravity is sufficiently strong. Next, contiguous groups of yet-unassigned primary ECAs with sufficiently strong gravity will be aggregated into secondary ECAs independent of a Micropolitan core; those without are classified as isolated. This is repeated with Metropolitan cores to define tertiary ECAs. The approach described above is based on the three categories of urbanized places from the RUCA system, but it generalizes broadly to an N-level hierarchy of nodes where the nth ECA is defined by a parameter pair: the upper limit on the population of the core and the gravity threshold that defines economic islands, respectively. Refinements based on the former would permit the definition of small and large metropolitan cores, or even Megapolitan cores. The latter parameter can be varied according to moments of the use coefficient distribution (the lowest quintile or decile) or absolute values with a priori relevance, e.g., a parameter from a specified decay function in a formal model.3. Construct measures of economic activity and economic dynamism by place type The smallest unit in the typology dataset described above is the Census tract. Every establishment observation in the InfoGroup dataset includes a Census tract identifier. Further, InfoGroup is linked longitudinally at the establishment level. It is therefore possible to construct measures of economic activity, the net change in economic activity, and the gross flows of economic activity at the Census tract level. For example, we can calculate the number of establishments (level), the change in the number of establishments (net), and the number of establishments that started and ceased operation (gross). We can do likewise for employment and sales. Further, because every ECA at every level of the proposed hierarchical typology is constructed by agglomerating Census tracts, these measures can be calculated for every ECA in the hierarchy by summing over their constituent tracts.4. Examine dynamism and FAI activity across place typesWith measures of economic growth at the ECA-level, we propose the following descriptive analyses:-For every level of the ECA hierarchy, we will identify the ECAs in the top and bottom quintile of the growth distribution.-For every level of the ECA hierarchy, we will construct the CDF of the use coefficient (i.e., the extent to which a unit at one level of the hierarchy is economically connected to a unit at a higher level of the hierarchy). We will then calculate the mean net growth rate for ECAs at each quintile of the use coefficient distribution. This will allow us to test whether economic connectedness is related to economic growth.-We will further explore this relationship by estimating regressions of growth rates in ECA i in year t on the relative use coefficient between ECA i and the core of ECA j. To study the specific role of FAI industries on growth, we will also estimate regression models that interact use coefficients with the share of economic activity in FAI industries. Further, we will decompose use coefficients into FAI and non-FAI components, allowing these to enter the regression separately.-Finally, we are interested in how growth in FAI translates through the supply chain to influence regional economic development. We will therefore estimate linear regressions of the growth rate of FAI in ECA i in year t with the growth rate of FAI in ECA j.5. Develop a web platform that stakeholders and researchers can use to download, analyze, and visualize economic activity and dynamism for different place types. We believe the creation of a web-based platform with interactive mapping that allows users to examine the spatial properties of different place typologies is a particularly innovative component of the proposed work that will provide great benefit to both researchers and policymakers. Users will be able to adjust the threshold parameters that generate ECA codes and see changes to their spatial extent in real time. Measures of economic activity and economic dynamism will be generated at the ECA-level and a data extract function will enable users to download shape files, Census tract and ECA identifiers, and these measures of economic activity and dynamism for specified geographies for use in their own analyses. For this project, the web-based interface will draw from a library of datasets created by the project team. Recall, that each typology is defined by model parameters. While the number of levels in the hierarchy, N, is discreet, the population thresholds that define the core of each level and the gravity thresholds are continuous. Beginning with N=3, we will discretize population to provide users options for each threshold with the grid size increasing at each level in the hierarchy. We will similarly discretize the gravity thresholds ten options, but unlike population, these will be based on the quantile of the use parameter distribution.

Progress 03/01/23 to 02/29/24

Outputs
Target Audience:We reached academic researchers and research staff at government statistical agencies through presentation of results and dissemination of research findings. We also lead several workshops for graduate students in economics to share our methodology. Early results were presented at theannual meeeting of the Rural Sociological Society at the University of Vermont August 2-6 2023. Changes/Problems:The project director, Anton Babkin, has left the University in August 2023. The role has been transfered to Austin Sandler. The project director, Austin Sandler, is leaving the University in June 2024. All project objectives will be completed at that time. Please note no cost extension with end date of 02/28/2025. What opportunities for training and professional development has the project provided?Post-doctoral researcher Sandler has performed the work listed under project objectives and has agreed to take over the role of Project Director after departure of Project Director Babkin in August 2023. A graduate student continuedworking on the project through the entire reporting period developing new methods and skills in spatial and tidy programming and has presented at the RSS conference in August 2023. Former Project Director Babkin, conducted a workshopat the University of Wisconsin for a broad audience of students, researchers, and other data practionerson tools and methods of collabrative research ininteractive Python and R notebooks. How have the results been disseminated to communities of interest?Results on demographic and health outcomes in rural areaswere presented at the RSS conference in Burlington,VT. What do you plan to do during the next reporting period to accomplish the goals?The team will finalize the Web dashboard and make the ECA typology data availible online(Objective 5). The team will also identify areas with strong FAI economic contribution and separately identify the relationship between connectedness and economic growth in this subset of areas (Objective 4). Please note no cost extension: new end date of 02/28/2025.

Impacts
What was accomplished under these goals? With respect to objectives 2 and 3, we have completed these tasks. Preliminary calculations performed using County Business Patterns have been replaced with finalized methodology using InfoGroup data and augmended with Census of Agriculture data. The methodology hasbeen significantly refined to account for international trade,government spending, commodity and industry dichotomy, as well as general trade balance constraints. With respect to objective 5, we have worked with APL to develop preliminary version of public facing interactive Web dashboard.

Publications


    Progress 03/01/22 to 02/28/23

    Outputs
    Target Audience:We reached academic researchers and research staff at government statistical agencies through presentation of results and dissemination of research findings. We also lead several workshops for graduate students in economics to share our methodology. Changes/Problems:The project director, Richard Dunn, is leaving the University in February. A project direct change is being requested. What opportunities for training and professional development has the project provided?A post-doctoral researcher started at the beginning of the project period and has been trained in our distributed version control collaborative research environment and fully integrated into our economic research program. They have agreed to continue working on the project for this coming year. A graduate student started during the second half of the project period and is currently undergoing similar training in methods and tools. Project researcher and Co-Investigator Babkin, conducted two workshops at the University of Wisconsin for graduate students in economics and agricultural economics on tools and methods in interactive reporting with Jupyter notebooks using a subset of project activities as the example application. How have the results been disseminated to communities of interest?Results on economic activity at the county and census tract level for alternative urban-rural typologies were presented at the 2022 Annual RDC Research Conference in Kansas City, Missouri. What do you plan to do during the next reporting period to accomplish the goals?Objectives 2 and 3 have been completed using publicly available aggregate data. In the forthcoming reporting period, we will repeat the analysis using establishment-level microdata as proposed in the project. The team will also identify areas with strong FAI economic contribution and separately identify the relationship between connectedness and economic growth in this subset of areas (Objective 4).

    Impacts
    What was accomplished under these goals? With respect to objective 2, the project team constructed economic catchment areas using data from the County Business Patterns, Census of Agriculture, and BEA Input-Output tables. With respect to objective 3, the project team constructed measures of economic activity and dynamism using the County Business Patterns, Census of Agriculture, and Business Dynamics Statistics. With respect to objective 5, the project team has developed interactive Jupyter and RStudio notebooks that allow users to analyze, map, and explore patterns of connectedness.

    Publications


      Progress 03/01/21 to 02/28/22

      Outputs
      Target Audience: Nothing Reported Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?We successfully hired a post-doctoral researcher who started employment January 30, 2021. During the reporting period, introducing the researcher to our computing environment and research tasks was initiated. How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?With respect to Objective 2, we anticipate developing an algorithm in Python and/or R that links counties by economic connectedness using the BEA Input/Output tables and four datasets: theCounty Business Patterns (CBP), the Quarterly Census of Employment and Wages (QCEW), InfoGroup, and NETS. Three of these were not originally included in the proposal but we are already using CBP as a smaller version of InfoGroup to develop our algorithms. The QCEW is a similar, publicly available dataset that uses a different sample frame that may be of interest and relatively low-cost to evaluate. We have decided to also consider NETS as this is the proprietary dataset that ERS has opted to use in the construction of an Agriculture and Food Business Database (AFBD). With respect to Objective 3, we will use these new economic activity based geographic aggregations to construct measures of economic activity and dynamism.

      Impacts
      What was accomplished under these goals? With respect to objective 1, we have completed this task and have processed InfoGroup to act as a synthetic LBD. With respect to objectives 2-4, we have completed the process of hiring a post-doctoral researcher who will undertake these tasks in the next two years of the project. With respect to objective 5, we have constructed an interactive Jupyter notebook that allows users to map rural typologies published by the Census Bureau and USDA ERS at different levels of geographic aggregation, along with rollover information boxes that provide summary statistics of demographics and economic activity. While built using existing place typologies, this product will act as the framework for the web-based platform described in the project once objectives 2-4 are completed.

      Publications