How Is This Research Useful to Practitioners?
In 2011, the Small Business Administration (SBA) provided over $30 billion of lending support to more than 60,000 small businesses. Measuring the impact of these loan dollars measures the success of the SBA and provides a lower-bound estimate of the marginal value of additional loan dollars.
The authors’ analysis provides robust estimates that suggest 3.0–3.5 new jobs are created within three years for every million dollars of SBA loans. The recipients of these loans are credit constrained, the SBA loans do not replace conventional loans, and the loans help borrowers grow more rapidly than they would otherwise. Larger firms and younger firms return slightly higher job growth from SBA loans, and there is greater job creation when local credit conditions are weak, but there is no clear evidence of cyclical variation. The average cost per job created during the study is $21,580–$25,450, which includes SBA administrative expenses and loan write-offs, and is considerably less than alternative fiscal stimulus programs targeting employment growth. The average wage for these newly created jobs is $30,000, but the authors note that their study is not a cost–benefit analysis and thus many factors were not considered.
The uniformly positive influence that increased capital access has on employment growth suggests that these factors of production are complementary rather than representative of capital–labor substitution. The authors’ findings will be of interest to financiers, macro-level investment analysts, government policymakers, and small businesses seeking financing.
How Did the Authors Conduct This Research?
Administrative data on SBA 7(a) and 504 loan programs are linked to the US Census Bureau’s Longitudinal Business Database employment measure by matching firm IRS Employer Identification Numbers (EINs), and loan years are set to correspond with the Census Bureau’s fiscal year. The dataset spans 25 years, from 1987 to 2012, and loan values are adjusted for inflation by using 2010 dollars. Canceled loans are excluded from the treatment group, multiple SBA loans to a single firm in a single year are aggregated, and only the first year of SBA loans is considered when firms receive loans in multiple years.
Firms with no employees prior to receiving SBA loans, such as startups, are excluded because employment history is necessary to identify similar control firms. If these excluded firms have a stronger (weaker) employment response to financing, this exclusion implies a negative (positive) bias. Control firms are matched on the basis of firm age, industry, year, pre-loan size, and kernel-based matching on propensity scores estimated as a function of employment history and other variables. A quadratic function assigns a greater weight to more closely matching controls.
The authors run several regression specifications estimating the effect of loan receipt and loan size on employment. The regression results are robust across ordinary least-squares and instrumental variable specifications controlling for local industry growth and local credit conditions. These controls have no discernible difference in the loan estimate.
SBA loans are not randomly given to recipients as in a controlled trial; rather, firms applying for SBA loans are thoroughly screened and vetted by SBA lenders so that loans are given only to firms with the greatest potential for successful employment growth and repayment. That fact alone may be an important signal to analysts looking at comparable public companies. Receipt of loans is partially a function of firm-level characteristics, and the fact that recipients demonstrate larger growth leading up to the loan may indicate some selection bias. Unobservable factors may bias the results; for example, SBA loan recipients may also receive financing from non-SBA sources. Failure to link all SBA loan recipients to the Longitudinal Business Database is another area of potential bias analysts should note.
An opportunity for additional research focusing on the positive and negative spillover effects of total lending on SBA firms exists. The authors examine some spillover effects from SBA loans, but SBA loans account for only approximately 0.25% of total outstanding loans. A very small percentage of firms receive SBA loans. A thorough cost–benefit analysis could examine producer surpluses of borrowers, lenders, and workers, as well as consumer surpluses from lower production costs due to a scale effect, spillover effects, and the external effects of increased employment on society or government budgets.