Documente Academic
Documente Profesional
Documente Cultură
DOI 10.1007/s11187-010-9302-7
123
254 V. Chandler
into two categories: incrementality and impact. First, impact if it is designed properly. Second, it offers a
the government only wants to offer guarantees to road-map to conduct a thorough economic impact
SMEs which would not have received a loan without analysis for guarantee programs and justifies con-
the guarantee (Riding et al. 2007). Guaranteeing ducting general surveys on financing issues for
loans that do not need the program simply shifts risk SMEs.
from the private sector to the public sector without The rest of the paper is structured in the following
creating any new loans. Second, guarantees should go way: Sect. 2 introduces the CSBFP and presents
towards SMEs that generate economic impact previous research on guarantee programs. Sect. 3
because guarantees should spur economic growth describes the data used; Sect. 4 presents the method-
and create wealth. ology; Sect. 5 reviews the most important results and
The incrementality of the Canadian Small Busi- presents the conclusions.
ness Financing Program (CSBFP) has already been
assessed at 75%1 (Riding et al. 2007), meaning that
only 25% of CSBFP clients could have received a 2 Theory and previous evidence
loan without the program. The objective of this paper
is to determine the second benefit to government: the To receive guarantees from the CSBFP in 2004, a
economic impact of the CSBFP. To measure the SME would first need to seek financing at a financial
impact of the program, we use the growth of four institution. The financial institution can then choose
variables: employment, total salary, revenues and to (1) refuse the demand, (2) finance the SME with its
profit. The Survey on Financing of Small and Medium own conventional products, or (3) make a loan and
Entreprises, 2004 contains this information for register it under the CSBFP. A loan can be registered
13,042 SMEs and identifies CSBFP participants. This under the CSBFP if the SME has revenues below $5
survey also allows the issue of self-selection to be million and the loan is smaller than $250,000.2 Under
addressed. All programs attract certain kinds of the CSBFP, Industry Canada and commercial lenders
businesses. Therefore, when evaluating the impact share the risk of providing small businesses with term
of the program, it is difficult to say whether, for loans for acquiring real property and equipment and
example, high-growth businesses were attracted to making leasehold improvements. The Minister is
the program or whether the program increased the liable to pay up to 85% of eligible losses on defaulted
growth of the participants. To address this issue of loans registered under the program. To help offset the
self-selection, I compare CSBFP participants with cost of claims for losses under the program, the
three kinds of SMEs: all SMEs, denied borrowers and CSBFP charges the lender an upfront fee of 2% and a
accepted borrowers. yearly fee of 1.25% of the value of the loan that is
Using these comparison groups, I find that partic- remitted by the lender through the interest rate.
ipation to the CSBFP increases growth in employ- Additionally, the program caps the variable interest
ment, salaries and revenues by 12, 12, and 7 rate that lenders can charge SMEs at no more than
percentage points, respectively, when compared to 3% more than their prime rate and the fixed interest at
denied borrowers. Furthermore, the analysis shows the single-family mortgage rate plus 3%. There is
that the suspected self-selection problem does not also a cap on claims for loss that limits the loss rates
seem to be present: the coefficients accounting for for CSBFP loans. These measures ensure that CSBFP
CSBFP participation are stable across different loans are provided to those businesses that have
specifications. viable projects but which are riskier, often due to a
As much as this publication is specific to a lack of collateral, while riskier loans are simply
Canadian program, it is also relevant for all similar refused any financing. Riding et al. (2007) calculated
programs. First, contrary to previous evidence that 75% of program participants would not have
(Levitsky 1997), it shows that a guarantee program received a loan without the program, thus showing
can be incremental and have a positive economic
2
In 2009, this amount was increased to $350,000 for
1
A newer unpublished report estimates the incrementality at equipment or leasehold equipment and up to $500,000 for real
85% in 2007. property.
123
The economic impact of the Canada small business financing program 255
that the requirements of the CSBFP lead to oriented towards growth. Approximately 70% of
incrementality. SMEs are so-called lifestyle businesses that do not
A program must be incremental, but it must also seek growth (McMahon 2001). In that sense, if a
have a significant economic impact. Two methods program attracts high-growth SMEs, an economic
have been used in the literature to assess the impact impact study might be biased by comparing these
of guarantee programs: a macro approach and a micro high-growth SMEs to lifestyle businesses. Coleman
approach. Craig et al. (2007a, b) used a macro (1999) shows that not considering this selection bias
approach to look at the regional impact of participa- in impact studies of microfinance projects strongly
tion to a guarantee program. In their first paper, they over-estimates the impacts of the program.
analyze the impact of the number of Small Business In this study, I use the micro approach, but explicitly
Administration (SBA) loans on the growth of address the issue of self-selection. First, I include a lag
regional per capita income, while the second one variable for each growth indicator. Second, I compare
they use the same method to study the impact of the program participants with other SMEs that seek debt
same variable on the average annual level of financing. Finally, I include a variable that accounts for
employment. In both cases, the authors conclude that growth intentions. The significance of these variables
the number of SBA loans is positively correlated to and the changes in the coefficients accounting for
higher per capita income and higher employment. It CSBFP participation will give a good indication of the
is, however, impossible to say whether the SBA loans extent of the self-selection bias.
spur economic development or whether the economic
development increases the demand for SBA loans
(Craig et al. 2009). This problem of reverse causality 3 Data and descriptive statistics
is typical of the macro approach.
The micro approach, which consists of following This paper uses the Survey on Financing of Small and
individual program participants, avoids the problem of Medium Enterprises, 2004, a part of the Small and
conflicting causality. For example, Bradshaw (2002) Medium Enterprise Financial Data Initiative (SME
compared the employment and economic activity of FDI), which targets Canadian SMEs and investigates
759 firms having used the California State Loan their financing behavior.3 The main objective of this
Guarantee Program before and after the loan. Over survey conducted by Industry Canada, the Depart-
the period of the loan, the number of employees of ment of Finance and Statistics Canada was to
participants grew on average by 40%, while the determine which kind of SMEs request financing in
average Californian SME shrank by 11%. In this the form of debt, leasing, equity and trade credit, and
comparison, therefore, the program was responsible which are approved.4 The results of the survey were
for an economic growth of 51%. Kang and Heshmati then linked to the tax files of the respondents to
(2008), using a pseudo-panel of the 20,702 applica- provide financial information for the 5 years preced-
tions of the Korea Technology Credit Guarantee Fund ing and following the survey. Finally, the CSBFP
(KOTEC), report that guaranteed firms were able to randomly selected participants to be surveyed and
achieve a better performance (growth of sales and over-sampled (Table 1).
productivity). The impact of the KOTEC varied greatly A first analysis of difference in growth rates
across industries. Finally, an assessment carried out by (Table 2) shows that CSBFP participants in the
Industry Canada (2009a), using self-administered sample grew significantly faster than the three other
questionnaires asking participants to determine the categories of SMEs (CSBFP non-participants). For
number of jobs created by the loan, reports that CSBFP
created on average 2.2 new full-time jobs.
3
The challenge of the micro approach, one which Statistics Canada chose 34,363 SMEs from the Canadian
has not been addressed in earlier publications, is to Business Registry. Of these 21,710 could be contacted, and
13,042 agreed to complete the phone interview and allowed
overcome the selection bias or, in other words, to
Statistic Canada to share the information with Industry Canada.
reach the sixth step of program evaluation (Storey 4
For more information on this survey, see the following
2000). Businesses that apply for guarantees are website: http://www.sme-fdi.gc.ca/eic/site/sme_fdi-prf_pme.
different from other businesses because they are nsf/eng/01561.html.
123
256 V. Chandler
Number of employees Average number of employees in a given year as reported to the Canadian Revenue Agency
Salaries Total salary reported by the company to the Canadian Revenue Agency
Revenues Total revenue reported by the company to the Canadian Revenue Agency
Profit Total net income reported by the company to the Canadian Revenue Agency
Growth intentions (binary) Answer to the question: ‘‘Do you intend to significantly grow employment or revenues over the
next 2 years’’
Age of small and medium The number 2004 minus the answer to the question: ‘‘During what year did the business first start
enterprise (SME) selling goods and services’’
Size (full-time equivalent) Number of full-time employees in 2004 ? 0.5 (number of part-time employees in 2004)
Return on asset Net income in 2004/total assets in 2004
Leverage Total liabilities in 2004/total asset in 2004
example, CSBFP borrowers grew 9.2% more than the and 2 years would provide enough time for the firm to
average SME with respect to the number of employ- invest, while not exacerbating the survivorship bias.
ees. The difference in growth intention between The key variable to explain growth is CSBFP
CSBFP participants and all SMEs is significant and participation. The CSBFP randomly chose SMEs
could account for this difference between the two having received a guarantee in 2004 to be over-
groups. However, denied borrowers, which grew sampled in the survey. These businesses are the ones
more slowly than CSBFP borrowers, intended to considered to be CSBFP participants. It is possible
grow more than CSBFP borrowers. With respect to that some of the surveyed SMEs participated in the
their age and size, CSBFP borrowers are very similar program but are not identified as such because they
to denied borrowers in that they are younger and were randomly chosen from the business registry.
smaller than the average SME. Moreover, CSBFP This eventuality is, however, highly unlikely because
participants, just like denied borrowers, are more the program usually services 10,000 SMEs per year
leveraged than the average SME. Finally, CSBFP from a population of 1.3 million SMEs in Canada. To
borrowers are not present in the same industries as all ensure that the CSBFP dummy only captures the
SMEs, being over-represented in the retail and impact of the program—and not the self-selection
transportation sector and under-represented in man- into the program—I also use a number of control
ufacturing and mining. These samples are not repre- variables. First, the lag of the dependent variable is
sentative of all CSBFP participants because only included because businesses that have already expe-
participants that were in business from 2002 to 2006 rienced high growth will probably continue growing.
are considered in this sample. Since this survivorship To ensure that the lag is not impacted by the CSBFP
bias also affects the comparison groups, the growth participation, the lag growth of the dependent vari-
comparison is still valid. able from 2002 to 2003 is included. Second, owners
answered whether they intended on expanding the
scope of their business in the coming years. These
4 Methodology businesses should grow more than lifestyle busi-
nesses. Finally, even though Storey (1994, Table 5.7)
In line with the existing literature, economic impact shows that the literature on SME growth cannot
is measured through the growth of the following identify key determinants, I take into consideration
variables: revenues, number of employees, total other variables that could have an impact on growth.
salary and profit. The growth is calculated over a The age and size of business should have a negative
2-year period (2004–2006). This time span was impact on growth based on Gibrat’s law (see
chosen because the guarantee was given in 2004, Santarelli et al. 2006). Financial indicators, such as
123
The economic impact of the Canada small business financing program 257
return on asset (net income/total asset) and financial The first sample consists of all SMEs. This sample is
leverage (total liability/total asset), could also affect the one usually taken by impact studies. The impact
growth. The kind of industries in which an SME is of the program should be larger for this sample
active should also have an impact on its growth rate. because the self-selection bias is not accounted for, so
Therefore, the equation to be estimated is: CSBFP participation captures some of that bias. The
second sample consists of program participants and
Economic indicator ¼ b1 CSBFP þ b2 self selection
denied borrowers. According to Riding et al. (2007),
þ b3 control þ e
75% of CSBFP participants would have been rejected
Three samples are used to evaluate the coefficients had there not been the program. Consequently, had
in order to control for unobservable characteristics. the program not existed, 75% of CSBFP borrowers
123
258 V. Chandler
123
The economic impact of the Canada small business financing program 259
Table 3 Impact of the guarantee program on salary growth between 2004 and 2006
Explanatory variable All SMEs Denied borrowers Approved borrowers
I II III IV V VI
not participate. In the case of a continuous variable, impact studies, but at least it does not invalidate the
the coefficient represents the percentage point impact results of previous studies that did not deal explicitly
of an increase of one unit on the growth rate. The with the issue of self-selection.
value of the intercept represents the average growth Instead of looking at three distinct regression
for all SMEs and can be attributed to the economic models, the MANCOVA analysis combines the four
cycle. indicators in one analysis (Table 7). It shows that
The most interesting result is the lack of changes CSBFP participation always plays a positive role in
across specifications. One of the assumptions is that growth when CSBFP participants are compared to all
the self-selection bias played a major role in the borrowers. If the comparison is done only with
evaluation of collateral guarantee programs. In other denied borrowers, CSBFP participation only has a
words, participants would be more growth oriented significant impact on the number of employees and
than the usual SME, so the coefficient accounting for salaries. CSBFP participation does not significantly
participation would be over-estimated when partici- explain the simultaneous growth rates when a com-
pants are compared with usual SMEs. However, the parison with approved borrowers is performed.
impact of the model is the same whether participants
are compared to all SMEs, to denied borrowers or to 5.2 Absolute differences
accepted borrowers. In that sense, there does not
seem to be an important self-selection problem. It is To be able to put actual numbers on the impact of the
difficult to say whether this finding applies to other program, I also conducted OLS robust regressions to
123
260 V. Chandler
Table 4 Impact of the guarantee program on employment growth between 2004 and 2006
Explanatory variable All SMEs Denied borrowers Approved borrowers
I II III IV V VI
explain the differences between levels in 2004 and created approximately 5,000 jobs. From the 131,471
2006 in a very similar framework as the extensive jobs created in private small businesses from 2004 to
one used for Tables 3, 4, 5, 6 comparing CSBF 2006 (Industry Canada 2007), CSBFP would be
borrowers with denied borrowers. The results show responsible for 3.8% of these positions.
that the program would be responsible for no Assuming that the 2003–2004 cohort will cost
significant absolute increase in salaries, for the approximately $30 million5 (Industry Canada 2009b,
creation of 0.63 jobs, for an increase of $78,352 in p. 24), each job created costs of about $6,000. In
revenues and for no significant increase in profit over comparison, the HRSDC (1998) offers some indication
the 2-year period per loan guaranteed (Table 8). as to the cost per job of other public programs: the
The number of jobs estimated contrasts with the Local Economic Development Assistance (1980–
2.2 mentioned by Industry Canada (2009a). It must be 1981), for example, spent approximately $5,000 per
stressed that this number was calculated based on job, while Young Canada Works (1977–1980) spent
SMEs self-reported number of jobs created by the
loan and is therefore not very reliable, as already
noticed by the KPMG (2009, Sect. VI).
Of the approximately 11,000 loans guaranteed in
2004 (Industry Canada 2009a, Table 2), 75% were 5
Up to 2010, the cohort has cost $28.1 millions. The program
incremental (Riding et al. 2007), and they would have costs in this equation exclude the salary and operating expenses
created 0.63 jobs each, meaning that the CSBFP for the program.
123
The economic impact of the Canada small business financing program 261
Table 5 Impact of the guarantee program on revenues growth between 2004 and 2006
Explanatory variable All SMEs Denied borrowers Approved borrowers
I II III IV V VI
$7,260 per job and the New Employment Expansion I included variables that capture growth intention
and Development (1982–1983) generated jobs at a cost and estimate the models in three different settings:
of $11,900 per job. When taking into account inflation, with all SMEs, with denied borrowers and with
the cost per job created by the CSBFP is very small, approved borrowers.
especially when taking into consideration that job When compared to denied borrowers, I have
creation is not the main objective of the CSBFP. estimated that the CSBFP has an impact on growth
of 12 percentage points in the cases of salaries and
employment and 7 percentage points in the case of
6 Conclusion revenues. This impact is very similar across differ-
ent specifications, meaning that the self-selection
This paper attempts to quantify the economic impact problem is probably not as important as initially
of the CSBFP. To do so, I have considered four key assumed. In the case of absolute differences, CSBFP
dependent variables: salary, employment, revenues borrowers would have hired 0.63 employees more
and profits. Using a simple OLS robust regression, than non-borrowers and would have had earnings
I have estimated the coefficient, taking into account $78,352 superior to non-borrowers between 2004
CSBFP participation, and determined the contribu- and 2006.
tion of the program to the growth of the participant. This research offers only a preliminary analysis of
To address the potential issue of self-selection, the economic impact. Due to the variety of methods
123
262 V. Chandler
Table 6 Impact of the guarantee program on profit growth between 2004 and 2006
Explanatory variable All SMEs Denied borrowers Approved borrowers
I II III IV V VI
Table 7 MANCOVA results (p values of CSBFP) explaining growth rates between 2004 and 2006
Dependent variables Comparison groups
All SMEs Denied borrowers Approved borrowers
used in the literature on firm growth, it is difficult to on testing the present results using different econo-
determine which econometric method should be used metric methods. Furthermore, the surprising absence
when assessing the impact of a program on SME of self-selection in guarantee programs should be
growth. Consequently, future research should focus confirmed using other similar programs.
123
The economic impact of the Canada small business financing program 263
Table 8 Impact of the guarantee program on economic impact differences between 2004 and 2006
Explanatory variable Salary Employment Revenues Profits
Acknowledgments The author wishes to thank the two Craig, B. R., Jackson, W. E., III, & Thomson, J. B. (2009). The
anonymous reviewers, Allan Riding, Adele Deschamps, economic impact of the small business administration’s
Derek Gowan, Katherine Clubine, Owen Jung and Richard intervention in the small firm credit market: A review of
Archambault for helpful comments. Any remaining mistakes the literature. Journal of Small Business Management,
are the sole responsibility of the author. 47(2), 221–231.
Cressy, R. (2002). Funding gaps: A symposium. The Economic
Journal 112, F1–F16.
References Honohan, P. (2008). Partial credit guarantees: Principles and
practice. Institute for International Integration Studies
Bradshaw, T. K. (2002). The contribution of small business (IIIS) Discussion Paper no. 244. Dublin: IIIS.
loan guarantees to economic development. Economic HRSDC (Human Resources and Social Development Canada)
Development Quarterly, 16, 360–369. (1998). Direct job creation programs: Evaluation lessons.
Coleman, B. E. (1999). The impact of group lending in Available at: http://www.hrsdc.gc.ca/eng/cs/sp/hrsdc/edd/
northeast Thailand. Journal of Development Economics, reports/1998-000392/page01.shtml. Accessed 25 Nov 2009.
60, 105–141. Industry Canada (2007). Key small business statistics. Avail-
Craig, B. R., Jackson, W. E., III, & Thomson, J. B. (2007a). able at: http://www.ic.gc.ca/eic/site/sbrp-rppe.nsf/eng/
SBA-loan guarantees and local business growth. Journal h_rd02188.html. Accessed 25 Nov 2009
of Small Business Management, 45, 116–132. Industry Canada (2008). Key small business statistics. Avail-
Craig, B. R., Jackson, W. E., III, & Thomson, J. B. (2007b). able at: http://www.ic.gc.ca/eic/site/sbrp-rppe.nsf/eng/
Credit market failure intervention: Do government spon- h_rd02296.html. Accessed 25 Nov 2009
sored small business credit programs enrich poorer areas? Industry Canada (2009a). Canada Small Business Financing
Small Business Economics, 30, 345–360. Act, Annual Report 2008–2009. http://www.ic.gc.ca/eic/
123
264 V. Chandler
site/csbfp-pfpec.nsf/eng/h_la02973.html. Accessed 25 Oct Parker, S. C. (2002). Do banks ration credit to new enterprises?
2010. And should governments intervene. Scottish Journal of
Industry Canada (2009b). Comprehensive Review Report Political Economy, 49(2), 162–195.
2004–2009. Available at: http://www.ic.gc.ca/eic/site/ Riding, A. L., Madill, J., & Haines, G. (2007). Incrementality
csbfp-pfpec.nsf/eng/h_la02973.html. Accessed 25 Nov 2009 of SME loan guarantees. Small Business Economics, 29,
Kang, J. W., & Heshmati, A. (2008). Effects of credit guarantee 47–61.
policy on survival and performance of SMEs in Republic of Santarelli, E., Klomp, L., & Thurik, A. R. (2006). Gibrat’s
Korea. Small Business Economics, 31, 445–462. Law: An overview of the empirical literature. In E. San-
KPMG (2009). Study of the economic costs and benefits of the tarelli (Ed.), Entrepreneurship, growth, and innovation:
Canada Small Business Financing Program. Report pre- The dynamics of firms (pp. 41–73). New York: Springer.
sented to the CSBFP. Available at: http://www.ic.gc.ca/ Statistics Canada (2004). Survey on financing of small busi-
eic/site/csbfp-pfpec.nsf/eng/la02951.html. Accessed 25 nesses. Available at: http://www.sme-fdi.gc.ca/eic/site/
Oct 2010. sme_fdi-prf_pme.nsf/eng/Home. Accessed 25 Nov 2009.
Levitsky, J. (1997). SME guarantee schemes: A summary. The Storey, D. J. (1994). Understanding the small business sector.
Financier 4(1&2), 5–11. London: Routledge.
McMahon, R. G. P. (2001). Deriving an empirical development Storey, D. J. (2000). Six steps to heaven: Evaluating the impact
taxonomy for manufacturing SMEs using data from of public policies to support small businesses in developed
Australia’s business longitudinal survey. Small Business economies. In D. L. Sexton & H. Landstrom (Eds.),
Economics, 17, 197–212. Handbook of entrepreneurship. Oxford: Blackwells.
123