![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
ABSTRACT
Many public-sector programs are focused on economic development in rural areas, but few are formally evaluated. One often-overlooked component of effective rural development is the generation of local sales tax revenue, which helps fund city amenities and services. This paper evaluates whether the United States Department of Agriculture’s (USDA) Business and Industry (B&I) Loan Guarantee program impacted sales tax revenue for recipient Oklahoma communities. Sales tax revenue and demographic/business data for all Oklahoma communities that charged a sales tax between 2005 and 2015 are meshed with information on B&I loan recipients during that time. Coarsened exact matching (CEM) and multivariate regression techniques are then used to assess impacts on local sales tax revenues. The results demonstrate that the B&I program had a positive impact on total retail sales per capita over the full period, and that the effects were evident in recessionary but not expansionary periods. The evidence suggests that the B&I program can improve the financial health of recipient communities, and that their impacts are likely larger during economic downturns when tax revenue is even more vital for rural communities.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. The goal of the B&I program is to promote the creation of rural businesses to secure start-up capital, finance business expansion, and create jobs. B&I also improves the quality of life for residents, and it helps economic development in rural areas (Office of Budget and Program Analysis, Citation2019).
2. The amount of guarantee on the loan depends on the size of the loan. Loans of $5 million or less are guaranteed up to 80%, a 70% loan guarantee is offered for loans between $5 million and $10 million, and loans exceeding $10 million receive a maximum of a 60% guarantee (USDA Rural Development, Citation2017). The limit for most loan guarantees is $10 million but the administrator of RBS can grant loan guarantees up to $25 million at his or her discretion (Office of the Comptroller of the Currency, Citation2018).
3. These examples of industries come from the dataset provided by the Oklahoma RD state office, and do not necessarily represent all industries funded by the B&I program.
4. Rupasingha and Wang (Citation2017) used county level data instead of individual business level data because they did not have access to individual borrower data. Also, they did not use employment numbers as their dependent variable because the employment data are withheld to avoid disclosing data of individual companies; instead, they used establishment data as the dependent variable.
5. Both the 7(a) and 504 are loan guarantee programs, but their maximum amount to guarantee is $1,000,000 and $1,300,000 respectively. Both programs can fund businesses in rural and metro areas, unlike the B&I, which only lends to businesses in rural areas.
6. The five factors are: (1) amenities; (2) local shopping options; (3) Internet shopping; (4) potential shopper demographics; and (5) policy interventions. Two of these (policy interventions and local shopping options) are directly related to the analysis in this paper.
7. We allow for the loans to be in place for at least one year before assessing them – i.e. our cutoff for loan awards for analysis of the 2015 tax data is 2014.
8. The US Census Bureau explains that a place is any area representing officially incorporated governments such as cities, towns, villages, municipality, township, community, populated place, neighborhood, postal place/zip code, populated place, or boroughs (U.S. Census Bureau, CitationN.D.; Ratcliffe, CitationN.D.)
9. The three main assumptions of EPBR are: “X (a k-dimensional data set) is drawn randomly from a specified population X, the population distribution of X is an ellipsoidally symmetric density or a discriminant mixture of proportionally ellipsoidally symmetric densities, and the matching algorithm applied is invariant to affine transformations of X” (Iacus et al., Citation2012).
10. In alternative specifications, we replace the base-year controls (i.e. 2005) with controls from the more recent time period (i.e. 2015). The results are qualitatively and quantitatively similar, with significant impacts still found for the B&I loan guarantee amount. We also change the cutoff point to include only loans made through 2012 or 2013 (instead of 2014 in the main specification), and again the results are similar.
11. The simple OLS model includes all observations with data for all control variables, which is 301 out of the total 335 observations shown in table 2 (57 treated and 278 control).
12. These figures are calculated at the mean log of TRS per capita ($8,810 for non-recipients; ln(8,810) = 9.08). We first calculate the percentage change in the B&I grant awarded compared to the mean award: $5.26M for awardees compared to a mean of $0.89M for all communities is a ratio of 5.92. The increase in TRS per capita can be estimated by multiplying this percentage change in the B&I grant awarded by the coefficient in table 3, adding the result to the mean natural log of TRS per capita, exponentiating, and then subtracting the exponentiated mean of log TRS per capita. Exp(9.08 + (5.92 x 0.023)) – Exp(9.08) = $1,284.
13. We also exclude any communities receiving B&I loans in 2008 or 2009 (and none afterward) in the 2010-2014 comparison group.
14. In particular, we did not have access to B&I loan data prior to 2005. Earlier awardees may have continued to see benefits from these loans during our period of analysis, and we cannot control for this aspect due to the data limitation.