Finance and Growth at the Firm Level: Evidence from SBA Loans

• Author: J. DAVID BROWN,JOHN S. EARLE
• Date: 01 June 2017
• DOI: http://doi.org/10.1111/jofi.12492
• Published date: 01 June 2017

THE JOURNAL OF FINANCE •VOL. LXXII, NO. 3 •JUNE 2017

Finance and Growth at the Firm Level: Evidence

from SBA Loans

J. DAVID BROWN and JOHN S. EARLE∗

ABSTRACT

Weanalyze linked databases on all SBA loans and lenders and on all U.S. employers to

estimate the effects of financial access on employment growth. Estimation exploits the

long panels and variation in local availability of SBA-intensive lenders. The results

imply an increase of 3–3.5 jobs for each million dollars of loans, suggesting real effects

of credit constraints. Estimated impacts are stronger for younger and larger firms

and when local credit conditions are weak, but we find no clear evidence of cyclical

variation. Weestimate taxpayer costs per job created in the range of $21,000–$25,000.

“...[S]omeinstitutionalsolution to the problem of providing small busi-

ness with long-term capital must be sought. In view of the special re-

quirements of small business financial aid and the problems encountered

in financing small business, it is doubtful if any of our existing financial

institutions are equipped to perform this service.”

— Charles H. Schmidt (Journal of Finance (1951))

THE “SPECIAL REQUIREMENTS AND PROBLEMS in financing small business” have

been recognized since at least the early 1950s, when the Small Business Ad-

ministration (SBA) was founded. SBA loans have since become one of the

most significant interventions affecting firm-level access to finance, recently

∗Brown is with the Center for Economic Studies–U.S. Census Bureau. Earle is with the Schar

School of Policy and Government at George Mason University and IZA. An earlier version of this

paper circulated as “Do SBA Loans Create Jobs?” We thank the National Science Foundation for

support (Grant 1262269 to George Mason University) and participants in presentations at the

International Monetary Fund, Southern Economic Association Annual Meetings, SGE-ASSA An-

nual Meetings, Comparative Analysis of Enterprise Data conferences in Nuremberg and Istanbul,

George Mason University,Oberlin College, Central European University, the U.S. Census Bureau,

the Hungarian National Bank, the Small Business Administration, the Kauffman-Brandeis En-

trepreneurial Finance and Innovation Conference, Wesleyan University, the Consumer Financial

Protection Bureau, and the Comptroller of the Currency as well as Zoltan Acs, Emek Basker,David

Hart, Deborah Lucas, Traci Mach, Javier Miranda, E.J. Reedy, Alicia Robb, Rebecca Zarutskie, two

anonymous referees, and Editor Michael Roberts for helpful discussions and comments. We thank

Mee Jung Kim and Yana Morgulis for excellent assistance, and the SBA for providing the list of

loans we use in the analysis. We have read the Journal of Finance’s disclosure policy and have no

conflicts of interest to disclose. Any opinions and conclusions expressed herein are ours only and

do not necessarily reflect the views of the U.S. Census Bureau. All results have been reviewed to

ensure that no confidential information on individual firms is disclosed.

DOI: 10.1111/jofi.12492

1039

1040 The Journal of Finance R

reaching “all-time records in the Agency’s history, with over $30 billion in lend-

ing support to 60,000 small businesses in its top two lending programs—7(a)

and 504” in fiscal year 2011 (SBA (2011, p. 1)). Yet little is known about the

programs’ outcomes.

In this paper, we estimate the firm-level impact of access to SBA loans on

employment growth, which the SBA describes as its number one “strategic

goal.”1Theoretically, the employment effects of SBA loans are ambiguous. On

the one hand, easier access to capital may enable expansion, a scale effect.

But it may also induce capital-labor substitution, and it would reduce employ-

ment if capital and labor are gross substitutes. Moreover, even if the scale

effect dominates, so that the factors are gross complements, the employment

increase may be attenuated if the program crowds out other sources of capi-

tal, and the aggregate effect may be reduced if there are general equilibrium

displacement effects (negative spillovers onto competing firms).2Empirically

analyzing the programs is also difficult because many factors influence em-

ployment and growth, loan receipt may be subject to selection bias (positive or

negative), and appropriate firm-level microdata have usually been unavailable.

Perhaps as a result of these problems—and despite the prominence of SBA

programs and the high hopes in their power to stimulate business growth—

there have been few attempts to measure their impacts using appropriate data

and econometric methods. Unlike evaluations of job training programs, for ex-

ample, research on SBA loan effects has had to rely on small samples, short

time series, or aggregated data that do not permit the use of recent devel-

opments in econometrics (e.g., Imbens and Wooldridge (2009)). Most studies

consist of simple comparisons before and after the interventions, with little use

of comparison groups of nonrecipients. The most common unit of observation

in SBA studies is a geographic area such as county or metropolitan statistical

area, with outcomes measured as overall employment or per-capita income in

the local area; Craig, Jackson, and Thomson (2009) review these studies. Many

factors affect county-level outcomes, of course, making it difficult to isolate the

effects of a program that is small relative to the local economy. The SBA itself

reports a “performance indicator”—the number of “jobs supported,” reported

in recent years at over 0.5 million, based on summing up borrowers’ estimates

on their loan applications of the number of jobs they anticipate creating or

retaining as a result of the loan.3

1See SBA (2012, p. 7). The first goal is “growing businesses and creating jobs,” while the other

two (which would be still more difficult to investigate) are “building an SBA that meets the needs

of today’s and tomorrow’s small businesses” and “serving as the voice for small business.”

2Hurst and Pugsley (2011) have recently criticized SBA programs on the basis that typical

(median) small firms neither grow nor report wanting to grow, and thus that the emphasis on

small businesses is misplaced. Our paper does not analyze the full range of assistance for small

businesses, but we do show that recipients of SBA loans differ from the median in tending to grow

prior to loan receipt, which is an important issue for our identification strategy.

3The figure is 583,737 for fiscal year 2010 (SBA (2011, p. 4)). This number includes jobs sup-

ported not only by the 7(a) and 504 loan programs, but also by the much smaller microloan and

surety bond guarantee (SBG) programs.

Finance and Growth at the Firm Level 1041

Our research contributes to estimation of these effects by using much richer

data than were heretofore available and by applying econometric methods

developed for estimating causal effects with such data. We link administrative

data on every SBA 7(a) and 504 loan to long-panel data on all employers in

the U.S. economy, and we use the linked data to implement a longitudinal

matching estimator (e.g., Heckman, Ichimura, and Todd (1997,1998)). The

annual panels in our data from 1987 to 2012 permit us to select control firms

based on size, age, industry,year, and several years of growth history to control

for fixed effects for time- and treatment-control groups, and to measure the

evolution of employment before and after the loans were awarded. To address

potential correlation of SBA loan availability with local financial development

and growth, we also add time-varying controls for the local employment share

of banks, local volume of small business lending, and growth in the recipient’s

county-industry cell during the estimation period.

To account for time-varying unobservables in the selection into SBA loan

receipt and in the amount of loan received, we exploit institutional features of

the SBA program as well as empirical regularities about small business credit

markets. The SBA program works through partial (50% to 85%) loan guar-

antees provided to private lenders, some of which have a special status that

reduces their administrative costs (preferred lender (PLP) status in the Pre-

ferred Lenders Program).4As we show, these PLP lenders account for a large

share of all SBA loans, particularly the larger loans, implying that participation

in SBA loan provision is to some extent a function of corporate bank (firm-level)

policy,such that some banks participate a great deal in SBA lending, while oth-

ers do little or not at all. Furthermore, the PLP banks are unevenly dispersed

across the United States and have changed over time. These are key facts for

SBA loan access, as they imply variation in the ease and likelihood of obtain-

ing an SBA loan, especially when combined with the continuing importance

of the local credit market for small businesses (e.g., Brevoort and Hannan

(2006)). Indeed, we find that only 3.2% of SBA loans are in counties where

there were no branches of PLP lenders in the previous year, implying that SBA

borrowers also tend to rely on local lenders. At the same time, SBA loans are

a tiny share of the total banking business, accounting for only about 1% of all

small business loans, which themselves are only a fraction of all banking busi-

ness, suggesting that locational decisions of banks are unlikely to be based on

SBA business.

Based on this reasoning, we construct instrumental variables (IVs) that re-

flect the local presence of at least one branch of a PLP bank, defined by activity

outside the local lending market. In alternative specifications, presence is mea-

sured as a dummy variable for any PLP branch, as the number of PLP banks

with local branches, or as the local share of local bank branches affiliated with

PLP lenders. In some specifications, the presence of a branch of a PLP bank

4Strictly speaking, PLP is the term used in the 7(a) loan program. The analogous status for 504

loans is called the Preferred Certified Lenders Program (PCLP). Hereafter, we use the term PLP

to refer to both PLP and PCLP status.