Canada Small Business Financing Program: Economic impact analysis – July 2019

---

Executive summary

This analysis quantifies the economic impact of the Canada Small Business Financing Program (CSBFP) on Canadian small and medium-sized enterprises using data from the 2014 Survey on Financing and Growth of Small and Medium Enterprises linked with administrative tax data from 2012 to 2016. Results indicate that the program continued to positively affect the economic growth of Canadian small and medium-sized enterprises from 2014 to 2016. In particular, CSBFP loans increased firms' growth in revenues, profits and employment by 6, 7 and 3 percentage points respectively compared with non-CSBFP borrowers. Furthermore, CSBFP borrowers' growth in revenues and profits was $52,000 and $22,000 higher, respectively, than that of non-CSBFP borrowers. In addition, CSBFP borrowers were 3 percentage points more likely to survive to 2016 than non-CSBFP borrowers.

---

1. Introduction

The Canada Small Business Financing Program (CSBFP) is a loan loss-sharing program designed to support access to financing for Canadian small and medium-sized enterprises (SMEs). The program partners with private sector financial institutions to share the risk of lending to facilitate access to affordable financing for SMEs to start up, expand and modernize their small businesses.

To assess and improve the administration and economic impact of the CSBFP, Innovation, Science and Economic Development Canada (ISED) conducts a comprehensive review of the program every five years in accordance with the Canada Small Business Financing Act. Each review includes an evaluation of whether the program continues to achieve its objective of improving access to financing for SMEs, thereby encouraging growth and job creation.

This study feeds into the upcoming CSBFP evaluation for the period 20142019, quantifying the economic impact of the program, and complementing recent evaluations of whether the program improves access to financing (Rivard 2018) and whether it generates economic and social benefits for Canadians (Huang and Rivard 2019). Specifically, this study follows previous economic impact analyses (Chandler 2010; Song 2014) in using data from the Survey on Financing and Growth of Small and Medium Enterprises to estimate robust regressions and matching estimators that quantify the impact on firm performance of participation in the CSBFP. Firm performance is measured using a wide array of metrics: revenues, salaries, assets, employment, profits, profit margin, labour productivity and two-year survival.

Results suggest that CSBFP loans increased firms' growth in revenues, profits and employment by 6, 7 and 3 percentage points, respectively, from 2014 to 2016, compared with non-CSBFP borrowers. In addition, growth in revenues and profits of CSBFP borrowers was $52,000 and $22,000 higher, respectively, between 2014 and 2016, than that of non-CSBFP borrowers. Finally, CSBFP borrowers were 3 percentage points more likely to survive to 2016 than non-CSBFP borrowers.

The remainder of this report is organized as follows:

• Section 2: Overview of the program
• Section 3: 2010 and 2014 economic impact analyses, and related research
• Section 4: Data
• Section 5: Methodology
• Section 6: Results
• Section 7: Conclusions

---

2. Canada Small Business Financing Program

The primary objective of the CSBFP is to increase the availability of financing for the establishment, expansion, modernization and improvement of small businesses. In doing so, the program encourages economic growth and job creation by SMEs. The CSBFP came into effect on April 1, 1999, and it is governed by the Canada Small Business Financing Act.^{Footnote 1}

Small businesses play a critical role in the Canadian economy, accounting for 98 percent of all businesses in 2017 and 68 percent of job creation between 2013 and 2017 (ISED 2019). Yet, small businesses face unique challenges in accessing financing. If credit markets are characterized by credit rationing, for example, riskier small businesses may be unable to borrow even when they are willing to pay higher interest rates.

The CSBFP, by providing lenders with a guarantee on loans, is designed to facilitate access to financing for higher risk SMEs. Specifically, the program encourages lending to SMEs by financial institutions by reimbursing up to 85 percent of eligible losses on loans made through the program in the event of default.

Under the program, a small business applies for a loan from a financial institution (e.g., chartered bank, credit union or Caisse populaire) of its choice. Financing under the program is available to all small businesses with annual sales of $10 million or less, except for agricultural businesses, not-for-profit organizations, and charitable or religious organizations.

When the financial institution receives the loan application, it makes its lending decision based upon the expected returns from lending. The lender's expected return from a loan is a function of the interest rate it charges and the loan applicant's default risk (i.e., creditworthiness). Creditworthiness is based upon the applicant's credit score (i.e., propensity to repay debts), security (e.g., collateral or personal guarantee) and serviceability (i.e., income and debt levels, which impact the ability to repay the loan).

The lender has three options to maximize returns: approve a non-CSBFP loan, approve a CSBFP loan or deny financing. Offering non-CSBFP loans to the most creditworthy small businesses, which have low default risk, maximizes expected returns as the gain from charging higher interest rates and avoiding administration fees outweighs the risk from carrying the full cost of default. Likewise, lenders will choose to offer CSBFP loans to less creditworthy small businesses as the gain from the loan guarantee (i.e., reduced exposure to risk) outweighs the reduced profits from capped interest rates.^{Footnote 2} The least creditworthy small businesses will be denied financing altogether as lenders do not expect returns from loans made to such businesses, even if the loan is partially guaranteed.

Loans extended to small businesses under the CSBFP are registered with the program. The lender is responsible for all aspects of loan administration, including disbursement of loan proceeds and, in the event of default, realization on security and guarantees.

The total loan amount, which cannot exceed $1 million, can be used towards the purchase or improvement of real property, while a maximum of $350,000 can be used for other allowable expenses, such as leasehold improvements, equipment and registration fees. The maximum period of government coverage on a CSBFP loan is 15 years for real property and 10 years for all other eligible expenses.

To access CSBFP financing, borrowers pay registration and administration fees, as well as a capped interest rate. The registration fee is 2 percent of the total amount of the CSBFP loan and can be financed as part of the loan. In addition, borrowers pay an annual administration fee of 1.25 percent based upon loan balances. This fee can be included in the interest rate charged to the borrower. The lender submits the registration and administration fees to ISED, which are used to help offset the costs it incurs in claims paid on defaulted loans.

The second cost is the interest rate, which depends upon the financial institution and general credit market conditions. The maximum interest rate for a variable rate loan is the lender's prime rate plus 3 percent. The maximum interest rate for a fixed rate loan is the lender's single family residential mortgage rate plus 3 percent.

3. Previous economic impact analyses and related research

As noted in the 2010 economic impact analysis (Chandler 2010), the economic impact of loan guarantee programs, such as the CSBFP, may be evaluated using a macroeconomic or microeconomic approach. One shortcoming of a macroeconomic approach, which uses regional data on economic indicators, is that it is difficult to identify the effect of the program as loans to small businesses may increase economic growth but, at the same time, growth may increase demand for small business loans.

The microeconomic approach, which uses firm-level data, suffers from a different identification issue: firms may self-select into CSBFP loans. If, for example, CSBFP borrowers are inherently more growth oriented than other SMEs, the estimate of the impact of the CSBFP will capture both the effect of the program and the growth effect of these inherent characteristics. This overstates the impact of the program. Early studies, such as Bradshaw (2002) and Kang and Heshmati (2008), may not have adequately accounted for self-selection.

At the same time, the advantage of firm-level data is that such data typically offer a richer set of variables within which controls for self-selection may be found. Indeed, the 2010 economic impact analysis of the CSBFP (Chandler 2010) offers an example of this. The analysis exploits the richness of the 2004 Survey on Financing of Small and Medium Enterprises, including lagged growth and growth intentions as controls for selection, to estimate the effect of CSBFP participation on salaries, employment, revenues and profits. The other novelty of the methodology is the use of robust regression, in place of ordinary least squares regression, to limit the influence of outliers. Results suggest that CSBFP loans help SMEs grow: participation in the program is estimated to increase salary, employment and revenue growth by 10 percentage points. Using alternative metrics, CSBFP borrowers are found to have hired 0.63 employees more and to generate revenues $78,000 higher than non-CSBFP borrowers.

The 2014 economic impact analysis of the CSBFP (Song 2014), using data from the 2007 Survey on Financing of Small and Medium Enterprises, extends the analysis to include a wider variety of indicators. In addition to estimating the impacts, using robust regression, of program participation on salaries, employment, revenues and profits, the analysis also estimates the impacts on capital, value added and labour productivity. The analysis also uses logit models to estimate the impact of CSBFP participation on two-year survival and investment in research and development. Results indicate that CSBFP loans increase both the value and growth of revenues, salaries, profits and value added. Some evidence that the program increases labour productivity growth, propensity to spend on research and development, and two-year survival is also found. Estimating impacts using non-parametric estimators yields qualitatively similar results.

To generate results comparable with those of Chandler (2010) and Song (2014), this analysis uses the 2014 Survey on Financing and Growth of Small and Medium Enterprises to estimate robust regressions of the impact of participation in the CSBFP on the wide array of economic indicators evaluated in Song (2014).^{Footnote 3} The methodology used here is largely unchanged from Song (2014), with one notable exception: this analysis estimates robust regressions using more recently developed algorithms in Stata that better identify leverage points, are more robust to certain types of leverage points and offer heteroskedasticity-robust standard errors.^{Footnote 4} As in Song (2014), parametric estimates are qualitatively compared with their non-parametric analogues.

4. Data

Statistics Canada microdata used in this analysis link the 2014 Survey on Financing and Growth of Small and Medium Enterprises with Canada Revenue Agency's General Index of Financial Information^{Footnote 5} and payroll deduction data for the years 2012 to 2016.

The Survey on Financing and Growth of Small and Medium Enterprises is a cross-sectional survey, conducted every three years, that gathers data on small and medium-sized businesses and their financing activities, and provides detailed information on firm and owner characteristics.^{Footnote 6} The survey oversamples for special populations, including CSBFP borrowers, which gives a sufficiently large sample of CSBFP borrowers for this analysis. The linkage with the General Index of Financial Information and payroll deduction data is critical as it provides the data necessary to calculate growth from 2014 to 2016, as well as lagged growth from 2012 to 2013.

The microdata file contains 11,111 SMEs, of which 743 were CSBFP borrowers and 10,368 were non-CSBFP borrowers (Table 1). The analysis is limited to 9,627 incorporated firms as only these firms file T2 income tax data containing the General Index of Financial Information variables needed to calculate firms' economic indicators. The analysis further excludes 765 firms that entered the market after 2012 (i.e., start-ups) as these firms did not have lagged growth from 2012 to 2013, an important control for selection into CSBFP borrowing.^{Footnote 7}

Of the remaining 8,862 firms, a further 1,724 firms with incomplete records are excluded: 763 firms missing covariates (i.e., return on assets, leverage or lagged growth) and 961 firms that had exited the market by 2016.^{Footnote 8} This leaves a sample of 7,138 firms used for the analysis.

Table 1: Microdata and Sample Exclusions

	CSBFP Borrowers	Non-CSBFP Borrowers	Total
Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; and payroll deduction data.
Total	743	10,368	11,111
Unincorporated	33	1,451	1,484
Entered the market after 2012	206	559	765
Incomplete records
Exited the market by 2016	35	926	961
Missing covariates	88	675	763
Total	123	1,601	1,724
Total usable records	381	6,757	7,138

For this report, we use the following economic indicators: revenues, salaries, profits, assets, employment, profit margins, labour productivity and two-year survival. Financial variables are based upon tax filings by firms to the Canada Revenue Agency, deflated using Statistics Canada's consumer price index^{Footnote 9} and expressed in 2014 dollars. Employment is the average number of employees during a year; profit margins are calculated as profits divided by total sales; labour productivity is calculated as total sales divided by employment and two-year survival indicates that the firm survives until 2016.

Economic growth (of all economic indicators except two-year survival) is calculated in percentage and level terms. Specifically, for economic indicator ${}_{2014}$, percentage growth between 2014 and 2016 is calculated as $\left(\raisebox{1ex}{${}_{2016}$}\!\left/ \!\raisebox{-1ex}{${}_{2014}-1$}\right.\right)\times 100$, while (level) growth between 2014 and 2016 is calculated as ${}_{2016}-{}_{2014}.$ The indicator for two-year survival is equal to one if the firm employed at least one worker in 2016 and equal to zero otherwise.^{Footnote 10}

Table 2: Variables and Definitions

Variable	Definition
Firm Characteristics
CSBFP	Dummy variable indicating CSBFP borrower.
Lagged growth	Growth of economic indicators (except for two-year survival) between 2012 and 2013.
Firm age	Difference between 2014 and year firm was established.
Firm size	Average number of employees in 2014.
Return on assets (ROA) (Reference: Quartile 1)	Dummy variables for each quartile of ROA, where ROA = profits/assets. Quartiles are calculated by industry sector.
Leverage (Reference: Quartile 4)	Dummy variables for each quartile of leverage, where leverage = liabilities/assets. Quartiles are calculated by industry sector.
Innovation	Dummy variable indicating firm developed or introduced an innovation between 2012 and 2014.Footnote 1
Export	Dummy variable indicating firm exported in 2014.
Urban	Dummy variable indicating firm operated in a census metropolitan area (CMA)Footnote 2 in 2014.
Franchise	Dummy variable indicating firm was a franchise in 2014.
Growth intentions	Dummy variable indicating expected average yearly growth in sales or revenues, between 2015 and 2017, of more than 10 percent.
Industry sector (Reference: Retail trade)	Dummy variables for the following industry sectors: agriculture, forestry, fishing and hunting; mining and oil and gas extraction; manufacturing; construction; wholesale trade; transportation and warehousing; information and communication technology (ICT); professional, scientific and technical services; accommodation and food services; and other services (except public administration).Footnote 3
Region (Reference: Ontario)	Dummy variables indicating firm operated in the following regions: Atlantic (New Brunswick, Prince Edward Island, Nova Scotia, and Newfoundland and Labrador), Quebec, Prairies (Manitoba, Saskatchewan), Alberta, British Columbia and Territories (Yukon, Northwest Territories and Nunavut).
Owner Characteristics
Age	Age of primary decision maker of firm.
Level of education (Reference: Less than high school diploma)	Dummy variables indicating highest level of education attained by primary decision maker of firm: high school diploma, college/Cégep/trade school diploma, bachelor's degree, and master's degree or above.
Gender	Dummy variable indicating percentage of business owned by women is greater than 50 percent.
Immigration status	Dummy variable indicating firm owner was born outside of Canada.
Visible minority status	Dummy variable indicating percentage of business owned by a person from a visible minority group is greater than 50 percent.
Aboriginal status	Dummy variable indicating percentage of business owned by an Aboriginal person is greater than 50 percent.

The analysis uses a wide array of covariates that are important to economic growth. These variables are defined in Table 2. Table 3 presents summary statistics for all variables used in the analysis. Columns I, II, III and IV of Table 3 contain means or propensities for all non-CSBFP borrowers, approved borrowers, denied borrowers and CSBFP borrowers respectively. The first group is representative of all SMEs except for CSBFP borrowers.^{Footnote 11} It includes firms that sought financing in 2014 (approved borrowers and denied borrowers) and firms that did not seek financing in 2014. Approved borrowers are firms that were approved for non-CSBFP financing in 2014. Denied borrowers are firms that were denied any financing in 2014. The final group, CSBFP borrowers, are firms that were approved for CSBFP financing in 2014. The last column indicates whether the difference between columns I and IV is statistically significant at the 10 percent level.

A discussion of these statistics follows. It should be noted that the analysis in this section does not control for factors related to economic growth. In other words, the economic performance of CSBFP borrowers might not be attributed solely to the program, but also to other factors related to economic growth.

Table 3 suggests that CSBFP borrowers (column IV) outperform all non-CSBFP borrowers (column I) in terms of revenue, profit, employment, salary, assets and labour productivity growth rates. Revenues, employment and salaries of CSBFP borrowers grew significantly more between 2014 and 2016 than those of all non-CSBFP borrowers, while differences in profits, assets, labour productivity and profit margins are not significant.

More generally, seeking financing correlates with higher growth, with approved, denied and CSBFP borrowers (columns II, III and IV, respectively) growing more than all non-CSBFP borrowers. Interestingly, growth is highest for denied borrowers and CSBFP borrowers, two groups of SMEs that are — given CSBFP lending is incremental (e.g., Rivard 2018) — less creditworthy. These growth outcomes appear to be well predicted by respondents' growth intentions, with approved borrowers reporting higher expected revenue growth than all non-CSBFP borrowers, and denied and CSBFP borrowers reporting higher expected revenue growth than approved borrowers.^{Footnote 12}

With respect to firm characteristics, CSBFP borrowers, which have been in operation for an average of 14 years, are significantly younger than non-CSBFP borrowers, which have been in operation for an average of 23 years. In part due to their younger age — having had less time to grow and look beyond local markets — CSBFP borrowers are also significantly smaller in firm size and have a significantly lower propensity to export. Notably, CSBFP borrowers are significantly more likely to be franchises, with franchises comprising 17 percent of CSBFP borrowers, compared with 9 percent of all non-CSBFP borrowers.

Owners of firms extended CSBFP loans are significantly younger, with an average age of 46 years, than those owners of all non-CSBFP borrowers, whose average age is 52 years. Canada Small Business Financing Program borrowers have significantly fewer owners who were born outside of Canada (16 percent) than all non-CSBFP borrowers (20 percent).

There are significant differences in the education profiles of CSBFP borrower owners and non-CSBFP borrower owners. In particular, 11 percent of CSBFP borrowers do not hold a high school diploma, compared with 7 percent of all non-CSBFP borrowers. Among CSBFP borrowers, 35 percent hold a college, Cégep or trade school diploma, compared with 31 percent of all non-CSBFP borrowers, and 6 percent of CSBFP borrowers hold a master's degree or above, compared with 13 percent of all non-CSBFP borrowers.

Broadly speaking, Table 3 points to greater growth for CSBFP borrowers than for all non-CSBFP borrowers. Furthermore, there is a positive relationship between having sought financing in 2014 and growth over the subsequent two years. This relationship is particularly evident for denied borrowers and CSBFP borrowers. However, Table 3 also points to significant differences between all groups, along firm and owner characteristics dimensions. These differences could partially explain differences in economic performance. Thus, the remainder of this analysis aims to disentangle these relationships, controlling for firm and owner characteristics in comparing growth of CSBFP borrowers with that of other groups.

Table 3: Summary Statistics, Non-CSBFP Borrowers and CSBFP Borrowers

Parameters		Non-CSBFP Borrowers			(IV) CSBFP borrowers (N = 381)	Significant difference at the 10 percent level between (I) and (IV)
		(I) All (N = 6,757)	(II) Approved (N = 1,935)	(III) Denied (N = 206)
Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; and payroll deduction data. and authors' calculations.
Growth rates 20142016	Revenue (%)	3.8	5.4	7.6	16.2	Yes
	Profit (%)	22.9	26.3	12.1	26.7	No
	Employment (%)	5.9	6.5	8.6	15.5	Yes
	Salary (%)	10.0	9.1	20.7	17.7	Yes
	Assets (%)	14.7	15.8	22.4	15.0	No
	Profit margin (%)	24.8	62.2	7.5	8.8	No
	Labour productivity (%)	9.1	9.5	16.0	11.4	No
Business characteristics	Firm age (years)	22.5	22.6	16.7	14.4	Yes
	Firm size (number of employees)	27.4	35.0	15.6	12.6	Yes
	Innovation (%)	43.5	52.7	56.8	45.4	No
	Exporter (%)	16.9	20.0	17.5	8.1	Yes
	Return on assets	2.9	1.4	2.2	1.3	No
	Leverage	0.9	0.8	1.4	0.9	No
	Urban (%)	61.8	57.9	56.8	44.1	Yes
	Franchise (%)	8.6	8.1	7.8	16.8	Yes
	Growth intention (%)	18.1	23.7	34.0	30.7	Yes
Owner characteristics	Age (years)	52.1	50.3	48.6	46.3	Yes
	Female (%)	11.3	9.7	13.1	13.1	No
	Immigrant (%)	19.6	16.0	27.7	15.5	Yes
	Visible minority (%)	8.2	6.4	12.1	6.0	No
	Less than high school diploma (%)	6.8	8.8	6.9	11.3	Yes
	High school diploma (%)	23.0	22.4	20.9	23.4	No
	College/Cégep/trade school diploma (%)	30.5	29.5	32.5	35.2	Yes
	Bachelor's degree (%)	26.6	26.0	28.6	24.1	No
	Master's degree or above (%)	13.2	13.3	11.1	6.0	Yes

---

5. Methodology

Following the methodology of previous economic impact analyses, this report estimates the impact of CSBFP lending on revenues, salaries, profits, employment, profit margins, labour productivity, assets and two-year survival.

In particular, to assess the impact of CSBFP lending on the growth of SMEs' revenues, salaries, profits, employment, profit margin, labour productivity and assets, between 2014 and 2016, the following equation is estimated by robust regression:^{Footnote 13}

where ${\mathrm{economicindicator}}_i$ measures the growth of firm $i$ between 2014 and 2016 for a given economic indicator listed above;^{Footnote 14} ${\mathrm{CSBFP}}_i$ is a binary variable indicating CSBFP participation; and vector ${\chi }_i$ includes other firm and owner characteristics important to economic growth, such as the lagged growth of the dependent variable between 2012 and 2013 and firm size, as well as owner education and experience.

To assess the impact of CSBFP lending on firms' viability, two-year firm survival is modelled as

where the latent variable ${\mathrm{firmsurvival}}_i^{*}$ indicates survival of firm $i$ until 2016 and the other variables are the same as listed in Appendix A (Table A9).

A firm's survival or exit is observed based upon the value of the latent variable ${\mathrm{firmsurvival}}_i^{*}$ as

Assuming the error follows an extreme value distribution, this model is estimated as a probit.

Finally, to check the robustness of our estimates qualitatively, we use propensity score matching to non-parametrically estimate the average treatment effect of CSBFP participation on a given economic indicator of the treated (i.e. the group comprised of CSBFP borrowers), conditional on the variables used as explanatory variables (See Appendix B, Table B1). The samples here are chosen to correspond with those of the robust regressions (i.e., observations given a weight of zero in the robust regressions are excluded).

As in previous economic impact analyses, the above regressions are estimated for three samples. The first sample, all SMEs, is comprised of all non-CSBFP borrowers and CSBFP borrowers.^{Footnote 15} The estimated coefficient for CSBFP participation captures the economic impact of receiving a CSBFP loan. These estimates are the focus of the analysis.

The second sample includes approved borrowers (of non-CSBFP financing) and CSBFP borrowers. This sample compares creditworthy (i.e., approved for non-CSBFP financing) borrowers with incremental borrowers. The estimated coefficient for CSBFP participation gives the economic impact of receiving a CSBFP loan for SMEs that accessed financing. However, caution should be taken in interpreting these estimated CSBFP coefficients as incremental CSBFP borrowers are less creditworthy than approved (non-CSBFP) borrowers. The CSBFP coefficient, then, may be biased if creditworthiness is related to increased economic performance (in a manner not captured by the other covariates).^{Footnote 16}

The third sample includes denied borrowers and CSBFP borrowers. As CSBFP loans are incremental, and CSBFP borrowers would be denied financing in the absence of the program, this sample compares less creditworthy SMEs. In particular, the estimated coefficient for CSBFP participation gives the economic impact of receiving a CSBFP loan for less creditworthy SMEs that sought financing. Following the above logic, if creditworthiness correlates with economic performance, the CSBFP coefficient may be biased as incremental borrowers are more creditworthy than borrowers denied even partially guaranteed financing.

---

6. Results

As noted above, regression results for the first sample, all non-CSBFP borrowers and CSBFP borrowers, are the focus of the analysis. Table 4 summarizes the regression results, with detailed results presented in Appendix A.

Table 4: Impact of the CSBFP on Growth of Economic Indicators between 2014 and 2016

Economic Indicator	All SMEs	Approved Borrowers	Denied Borrowers
Note 1: Robust standard errors are in parentheses. Note 2: The impact of CSBFP borrowing on two-year survival refers to the marginal effects estimated from the probit regression as outlined in Section 5. Sources Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; and payroll deduction data. and authors' calculations.
Revenue	0.06Note *** (0.02)	0.05Note *** (0.02)	0.09Note *** (0.04)
Salary	0.03 (0.02)	0.03 (0.02)	0.06 (0.05)
Profit	0.07Note *** (0.05)	0.05Note ** (0.02)	0.07Note * (0.04)
Assets	−0.004 (0.01)	−0.03 (0.02)	0.01 (0.03)
Employment	0.03Note ** (0.02)	0.04Note ** (0.02)	0.06Note * (0.04)
Profit margin	−0.004 (0.005)	−0.003 (0.01)	−0.004 (0.01)
Labour productivity	0.02 (0.01)	0.02 (0.02)	0.05 (0.04)
Two-year survival	0.03Note *** (0.01)	0.01 (0.01)	0.05Note ** (0.02)

Table 4 and corresponding tables in Appendix A highlight three key results. First, a CSBFP loan, controlling for firm, owner and economic performance, significantly increased firms' growth in revenues, profits and employment by 6, 7 and 3 percentage points, respectively, from 2014 to 2016. CSBFP borrowers were also 3 percentage points more likely to survive until 2016 than non-CSBFP borrowers. Finally, in levels (see Appendix A, Table A8), CSBFP loans significantly increased two-year growth in revenues and profits by $52,000 and $22,000 respectively.^{Footnote 17}

Interestingly, indicators for CSBFP and growth intentions, particularly the latter variable, offer the best predictors for growth. Other covariates, such as firm size and owner age, tend to be statistically significant but not economically meaningful. Controls for innovation and exporting, activities with established links to economic performance, are notable for their lack of significance.^{Footnote 18} One explanation for the lack of significance among owner covariates may be that growth intentions best capture the effect of owner ability and motivation on a firm's success.

Note that some caution should be taken in interpreting these results. Given the nature of the sampling scheme, survey weights are not usable for this analysis.^{Footnote 19} In addition, recall that 763 firms were dropped due to their records missing covariates, while 961 firms were dropped because they did not survive until 2016. This latter group may create bias in the results if, for example, CSBFP borrowers that exited the market in 2015 differ significantly from non-CSBFP borrowers that exited the market in 2015. These factors may limit the representativeness of the estimates.

However, these results may also be viewed as robust for the following reasons. First, non-parametric specification of the impact of CSBFP loans on economic growth yields qualitatively similar results (Table 5).^{Footnote 20} Furthermore, these results largely hold across samples, which offer different points of comparison for CSBFP borrowers. Finally, estimated coefficients for lagged growth and growth intentions both offer intuitively credible controls for selection into CSBFP borrowing and significantly control for variation related to growth in revenues, profits and employment and to the likelihood of survival.^{Footnote 21}

Table 5: Summary of Results

Variables	Significant Impact from CSBFP: Growth 2014–2016		Significant Impact from CSBFP: Level Difference 2014–2016
	Robust Regression	Non-Parametric Matching	Robust Regression	Non-Parametric Matching
Note 1: Statistical significance is calculated at the 10 percent level. Note 2: The significant impact of CSBFP borrowing on two-year survival refers to estimates of probit regression as outlined in Section 5 (and not of robust regression). Sources Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; and payroll deduction data. and authors' calculations.
Revenue	Yes	Yes	Yes	Yes
Salary	No	No	No	No
Profit	Yes	Yes	Yes	Yes
Assets	No	No	No	No
Employment	Yes	Yes	No	No
Profit margin	No	No	No	No
Labour productivity	No	No	No	No
Two-year survival	Yes	N/A	N/A	N/A

---

7. Conclusions

This report quantifies the economic impact of receiving a CSBFP loan on the growth of Canadian SMEs. This study follows closely the methodology of previous economic impact analyses by estimating the impact using robust regression and propensity score matching. Specifically, results indicate that receiving a CSBFP loan significantly increased firms' growth in revenues, profits and employment by 6, 7 and 3 percentage points, respectively, from 2014 to 2016. CSBFP loans also significantly increased the probability of survival to 2016 by 3 percentage points. Finally, in absolute terms, CSBFP loans significantly increased growth in revenues and profits by $52,000 and $22,000, respectively, from 2014 to 2016. Corresponding non-parametric matching estimators give qualitatively similar results. Overall, this analysis indicates that the CSBFP positively impacted the economic growth of its borrowers.

---

References

Austin, P. C. (2009) "Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples", Statistics in Medicine, 28(25), 3083–3107.

Becker, S. O. and M. Caliendo. (2007). "Sensitivity analysis for average treatment effects", The Stata Journal, 7(1), 71–83.

Bradshaw, T. K. (2002). "The contribution of small business loan guarantees to economic development", Economic Development Quarterly, 16(4), 360–369.

Caliendo, M. and S. Kopeinig. (2008). "Some practical guidance for the implementation of propensity score matching", Journal of Economic Surveys, 22(1), 31–72.

Cameron, A. C. and P. K. Trivedi. (2005). Microeconometrics: Methods and Applications, New York, Cambridge University Press.

Canada Revenue Agency. General Index of Financial Information (GIFI). Ottawa, Canada Revenue Agency.

Chandler, V. (2010). "The Economic Impact of the Canada Small Business Financing Program." Ottawa: Industry Canada.

Chen, C. (2002). "Robust Regression and Outlier Detection with the ROBUSTREG Procedure", Proceedings of the Twenty-Seventh Annual SAS Users Group International Conference, 1–13.

Farcomeni, A. and L. Ventura. (2012). "An overview of robust methods in medical research", Statistical Methods in Medical Research, 21(2), 111–133.

Gangl, M. (2004). "RBOUNDS: Stata Module to Perform Rosenbaum Sensitivity Analysis for Average Treatment Effects on the Treated.", Statistical Software Components S43830: Boston College Department of Economics.

Garrido, M.M., A.S. Kelley, J. Paris, K. Roza, D.E. Meier, R.S. Morrison and M.D. Aldridge. (2014). "Methods for constructing and assessing propensity scores." Health Services Research, 49(5), 17011720.

Heritier, S., E. Cantoni, S. Copt and M.-P. Victoria-Feser. (2009). "Robust Methods in Biostatistics." Chichester: John Wiley & Sons Ltd.

Ho, D.E., K. Imai, G. King and E.A. Stuart. (2007). "Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference." Political Analysis, 15(3), 199236.

Huang, L. and P. Rivard. (2019). "Canada Small Business Financing Program: Cost-Benefit Analysis." Ottawa: Innovation, Science and Economic Development Canada.

Imai, K., G. King and E.A. Stuart. (2008). "Misunderstandings between experimentalists and observationalists about causal inference." Journal of the Royal Statistical Society, 171(2), 481502.

Imbens, G.W. (2004). "Nonparametric estimation of average treatment effects under exogeneity: A review." The Review of Economics and Statistics, 86(1), 429.

Innovation, Science and Economic Development Canada. (2019). "Key Small Business Statistics." Ottawa: Innovation, Science and Economic Development Canada.

Kang, J. and A. Heshmati. (2008). "Effect of credit guarantee policy on survival and performance of SMEs in Republic of Korea." Small Business Economics, 31(4), 445462.

Renaud, O. and M.-P. Victoria-Feser. (2010). "A robust coefficient of determination for regression." Journal of Statistical Planning and Inference, 140(7), 18521862.

Rivard, P. (2018). "Incrementality Study of the Canada Small Business Financing Program." Ottawa: Innovation, Science and Economic Development Canada.

Rosenbaum, P.R. (2002). "Observational Studies (2nd ed.)." New York: Springer.

Song, M. (2014). "Canada Small Business Financing Program: Updated and Extended Economic Impact Analysis." Ottawa: Industry Canada.

Statistics Canada. Survey on Financing and Growth of Small and Medium Enterprises, 2014. Ottawa: Statistics Canada.

Stuart, E.A. and D.B. Rubin. (2008). "Best Practices in Quasi-Experimental Designs: Matching Methods for Causal Inference." Best Practices in Quantitative Methods, 155176. Edited by J. Osborne. New York: Sage Publications.

Unterlass, F. (2013). "The Nexus of Innovation, Exports and Economic Performance of Firms — Revisiting Self-Selection and Learning-by-Exporting." Prague: EcoMod Conference.

Verardi, V. and C. Croux. (2009). "Robust regression in Stata." The Stata Journal, 9(3), 439453.

---

Appendix A: Regression output tables

Table A1: Impact of the CSBFP on Revenue Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; and payroll deduction data. and authors' calculations.
Firm Characteristics
CSBFP borrower	0.06Footnote *** (0.02)	0.08Footnote *** (0.02)	0.05Footnote *** (0.02)	0.06Footnote *** (0.02)	0.09Footnote *** (0.04)	0.09Footnote *** (0.03)
Lag growth (20122013)	0.001Footnote *** (0.0003)		0.0001Footnote *** (0.00001)		0.0001Footnote *** (0.00002)
Firm size	0.0002Footnote *** (0.0001)		0.00009 (0.0001)		0.00004 (0.001)
Firm age	0.0004Footnote ** (0.0002)		0.0006Footnote * (0.0003)		0.001 (0.001)
Return on assets — Quartile 2	0.01 (0.01)		0.04Footnote ** (0.02)		0.001 (0.05)
Return on assets — Quartile 3	0.01 (0.01)		0.04Footnote ** (0.02)		0.01 (0.05)
Return on assets — Quartile 4	0.02 (0.01)		0.04Footnote ** (0.02)		0.04 (0.05)
Leverage — Quartile 1	0.001 (0.01)		0.03 (0.02)		0.001 (0.08)
Leverage — Quartile 2	0.00002 (0.01)		0.001 (0.01)		0.01 (0.03)
Leverage — Quartile 3	0.02Footnote * (0.01)		0.02 (0.01)		0.01 (0.03)
Exporter	0.02 (0.01)		0.02 (0.02)		0.01 (0.05)
Innovation	0.01 (0.01)		0.02 (0.01)		0.002 (0.03)
Urban	0.01 (0.01)		0.03Footnote ** (0.01)		0.03 (0.04)
Franchise	0.01 (0.01)		0.003 (0.02)		0.004 (0.05)
Growth intentions	0.07Footnote *** (0.01)		0.08Footnote *** (0.02)		0.08Footnote ** (0.04)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.001Footnote *** (0.0003)		0.001Footnote * (0.001)		0.001 (0.002)
Female	0.003 (0.01)		0.02 (0.02)		0.05 (0.04)
Visible minority	0.01 (0.01)		0.02 (0.03)		0.02 (0.07)
Immigrant	0.01 (0.01)		0.01 (0.02)		0.02 (0.05)
High school diploma	0.01 (0.01)		0.004 (0.03)		0.04 (0.06)
College/Cégep/trade school diploma	0.003 (0.01)		0.01 (0.03)		0.03 (0.06)
Bachelor's degree	0.01 (0.01)		0.01 (0.03)		0.06 (0.07)
Master's degree or above	0.02 (0.02)		0.03 (0.03)		0.11 (0.08)
Constant	0.10Footnote *** (0.03)	0.002 (0.003)	0.11Footnote ** (0.05)	0.01Footnote ** (0.01)	0.11 (0.12)	0.01 (0.02)
Percentage of outliers	4.2	X	4.1	X	3.4	X
Robust R2 (w)Footnote 22	0.19	0.006	0.21	0.01	0.23	0.03
Robust R2 (rho)Footnote 23	0.06	0.002	0.08	0.004	0.09	0.01
N	7,138	7,138	2,316	2,316	587	587

Table A2: Impact of the CSBFP on Salary Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	0.03 (0.02)	0.05Footnote *** (0.02)	0.03 (0.02)	0.03Footnote * (0.02)	0.06 (0.05)	0.07Footnote * (0.04)
Lag growth (20122013)	0.0003Footnote *** (0.0001)		0.003 (0.01)		0.01Footnote ** (0.005)
Firm size	0.0001 (0.0001)		0.00003 (0.0001)		0.0001 (0.001)
Firm age	0.0003 (0.0002)		0.0001 (0.0004)		0.0002 (0.001)
Return on assets — Quartile 2	0.01 (0.01)		0.01 (0.02)		0.03 (0.06)
Return on assets — Quartile 3	0.02 (0.01)		0.01 (0.02)		0.01 (0.06)
Return on assets — Quartile 4	0.01 (0.01)		0.002 (0.02)		0.002 (0.06)
Leverage — Quartile 1	0.03Footnote ** (0.01)		0.04Footnote * (0.02)		0.13Footnote * (0.07)
Leverage — Quartile 2	0.02Footnote ** (0.01)		0.02 (0.02)		0.02 (0.04)
Leverage — Quartile 3	0.04Footnote *** (0.01)		0.02 (0.02)		0.05 (0.04)
Exporter	0.02Footnote * (0.01)		0.02 (0.02)		0.04 (0.05)
Innovation	0.01Footnote * (0.01)		0.02 (0.02)		0.03 (0.04)
Urban	0.01 (0.01)		0.03Footnote * (0.02)		0.0003 (0.04)
Franchise	0.01 (0.01)		0.01 (0.03)		0.05 (0.06)
Growth intentions	0.07Footnote *** (0.01)		0.06Footnote *** (0.02)		0.13Footnote *** (0.05)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.001Footnote *** (0.0004)		0.00004 (0.0008)		0.001 (0.002)
Female	0.02 (0.01)		0.02 (0.02)		0.04 (0.05)
Visible minority	0.02 (0.02)		0.05 (0.03)		0.18Footnote ** (0.08)
Immigrant	0.01 (0.01)		0.03 (0.02)		0.07 (0.05)
High school diploma	0.01 (0.02)		0.05 (0.03)		0.04 (0.07)
College/Cégep/trade school diploma	0.01 (0.02)		0.04 (0.03)		0.01 (0.07)
Bachelor's degree	0.02 (0.02)		0.06Footnote * (0.03)		0.002 (0.08)
Master's degree or above	0.01 (0.02)		0.04 (0.04)		0.08 (0.09)
Constant	0.01 (0.03)	0.004 (0.004)	0.04 (0.06)	0.01 (0.01)	0.05 (0.16)	0.03 (0.04)
Percentage of outliers	4.6	X	4.1	X	4.6	X
Robust R2 (w)	0.10	0.002	0.13	0.002	0.18	0.01
Robust R2 (rho)	0.03	0.0007	0.04	0.0009	0.07	0.004
N	7,138	7,138	2,316	2,316	587	587

Table A3: Impact of the CSBFP on Profit Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	0.07Footnote *** (0.05)	0.08Footnote *** (0.02)	0.05Footnote ** (0.02)	0.06Footnote *** (0.02)	0.07Footnote * (0.04)	0.08Footnote ** (0.03)
Lag growth (20122013)	0.0003Footnote *** (0.00005)		0.01Footnote *** (0.003)		0.0002Footnote *** (0.00002)
Firm size	0.0002Footnote *** (0.0001)		0.00003 (0.0001)		0.0004 (0.001)
Firm age	0.0002 (0.0002)		0.0005 (0.0004)		0.001 (0.001)
Return on assets — Quartile 2	0.02 (0.01)		0.05Footnote ** (0.02)		0.02 (0.06)
Return on assets — Quartile 3	0.03Footnote ** (0.01)		0.07Footnote *** (0.02)		0.01 (0.07)
Return on assets — Quartile 4	0.04Footnote *** (0.01)		0.07Footnote *** (0.02)		0.04 (0.07)
Leverage — Quartile 1	0.03Footnote *** (0.01)		0.01 (0.02)		0.02 (0.08)
Leverage — Quartile 2	0.02Footnote * (0.01)		0.02 (0.02)		0.08Footnote * (0.04)
Leverage — Quartile 3	0.01 (0.01)		0.02 (0.02)		0.01 (0.04)
Exporter	0.02 (0.01)		0.01 (0.02)		0.03 (0.06)
Innovation	0.01Footnote * (0.01)		0.03Footnote * (0.02)		0.05 (0.04)
Urban	0.01 (0.01)		0.03Footnote * (0.02)		0.005 (0.04)
Franchise	0.02 (0.01)		0.003 (0.02)		0.01 (0.06)
Growth intentions	0.07Footnote *** (0.01)		0.07Footnote *** (0.02)		0.05 (0.04)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.002Footnote *** (0.0004)		0.001Footnote * (0.001)		0.002 (0.002)
Female	0.01 (0.01)		0.01 (0.02)		0.04 (0.05)
Visible minority	0.001 (0.01)		0.01 (0.03)		0.03 (0.08)
Immigrant	0.01 (0.01)		0.001 (0.02)		0.03 (0.06)
High school diploma	0.02 (0.02)		0.02 (0.03)		0.10 (0.06)
College/Cégep/trade school diploma	0.01 (0.02)		0.01 (0.03)		0.10Footnote * (0.06)
Bachelor's degree	0.004 (0.02)		0.04 (0.03)		0.11 (0.06)
Master's degree or above	0.03Footnote * (0.02)		0.06 (0.04)		0.10 (0.09)
Constant	0.14Footnote *** (0.03)	0.007 (0.004)	0.16Footnote ** (0.06)	0.01Footnote * (0.01)	0.24 (0.13)	0.0002 (0.03)
Percentage of outliers	5.7	X	5.3	X	5.1	X
Robust R2 (w)	0.18	0.005	0.20	0.008	0.23	0.02
Robust R2 (rho)	0.06	0.002	0.07	0.003	0.09	0.006
N	7,138	7,138	2,316	2,316	587	587

Table A4: Impact of the CSBFP on Asset Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	0.004 (0.01)	0.03Footnote * (0.01)	0.03 (0.02)	0.04Footnote *** (0.02)	0.01 (0.03)	0.002 (0.02)
Lag growth (2012–2013)	0.003Footnote *** (0.001)		0.01Footnote ** (0.003)		0.01 (0.01)
Firm size	0.0002Footnote *** (0.0001)		0.0003Footnote *** (0.0001)		0.001 (0.001)
Firm age	0.00001 (0.0002)		0.0001 (0.0003)		0.0002 (0.001)
Return on assets — Quartile 2	0.01 (0.01)		0.01 (0.01)		0.01 (0.03)
Return on assets — Quartile 3	0.01 (0.01)		0.01 (0.02)		0.06 (0.03)
Return on assets — Quartile 4	0.01 (0.01)		0.03 (0.02)		0.03 (0.04)
Leverage — Quartile 1	0.07Footnote *** (0.01)		0.08Footnote *** (0.02)		0.10 (0.06)
Leverage — Quartile 2	0.03Footnote *** (0.01)		0.05Footnote *** (0.01)		0.04 (0.03)
Leverage — Quartile 3	0.01 (0.01)		0.003 (0.01)		0.01 (0.02)
Exporter	0.02Footnote ** (0.01)		0.01 (0.02)		0.004 (0.04)
Innovation	0.004 (0.01)		0.01 (0.01)		0.02 (0.03)
Urban	0.01 (0.01)		0.003 (0.01)		0.02 (0.03)
Franchise	0.002 (0.01)		0.03 (0.02)		0.05 (0.04)
Growth intentions	0.05Footnote *** (0.01)		0.08Footnote *** (0.02)		0.07Footnote ** (0.03)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.001Footnote *** (0.0003)		0.001Footnote ** (0.001)		0.001 (0.001)
Female	0.02Footnote ** (0.01)		0.01 (0.02)		0.01 (0.03)
Visible minority	0.02Footnote * (0.01)		0.01 (0.03)		0.08 (0.05)
Immigrant	0.01 (0.01)		0.02 (0.02)		0.004 (0.04)
High school diploma	0.02Footnote * (0.01)		0.01 (0.03)		0.05 (0.05)
College/Cégep/trade school diploma	0.03Footnote * (0.01)		0.01 (0.03)		0.02 (0.05)
Bachelor's degree	0.03Footnote * (0.01)		0.01 (0.03)		0.02 (0.05)
Master's degree or above	0.04Footnote ** (0.02)		0.02 (0.03)		0.02 (0.08)
Constant	0.06Footnote *** (0.03)	0.02Footnote *** (0.003)	0.07 (0.05)	0.03Footnote *** (0.01)	0.07 (0.09)	0.01 (0.02)
Percentage of outliers	5.8	X	5.7	X	7.8	X
Robust R2 (w)	0.25	0.0007	0.14	0.005	0.17	0.00002
Robust R2 (rho)	0.04	0.0003	0.05	0.002	0.06	0.000008
N	7,138	7,138	2,316	2,316	587	587

Table A5: Impact of the CSBFP on Employment Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	0.03Footnote ** (0.02)	0.04Footnote *** (0.01)	0.04Footnote ** (0.02)	0.03Footnote ** (0.02)	0.06Footnote * (0.04)	0.08Footnote *** (0.03)
Lag growth (2012–2013)	0.02Footnote *** (0.005)		0.04Footnote *** (0.01)		0.02 (0.02)
Firm size	0.0001 (0.0001)		0.0001 (0.0001)		0.0001 (0.001)
Firm age	0.0002 (0.0002)		0.0001 (0.0003)		0.0003 (0.001)
Return on assets — Quartile 2	0.01 (0.01)		0.01 (0.02)		0.01 (0.05)
Return on assets — Quartile 3	0.01 (0.01)		0.01 (0.02)		0.04 (0.05)
Return on assets — Quartile 4	0.004 (0.01)		0.02 (0.02)		0.07 (0.05)
Leverage — 1er quartile	0.01 (0.01)		0.03 (0.02)		0.09 (0.08)
Leverage — Quartile 2	0.01 (0.01)		0.03Footnote ** (0.01)		0.001 (0.04)
Leverage — Quartile 3	0.02Footnote ** (0.01)		0.02 (0.02)		0.05 (0.03)
Exporter	0.002 (0.01)		0.01 (0.02)		0.003 (0.04)
Innovation	0.01 (0.01)		0.004 (0.01)		0.03 (0.03)
Urban	0.0001 (0.01)		0.03Footnote * (0.01)		0.02 (0.04)
Franchise	0.01 (0.01)		0.005 (0.02)		0.02 (0.04)
Growth intentions	0.05Footnote *** (0.01)		0.06Footnote *** (0.02)		0.06Footnote * (0.04)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.001Footnote * (0.0003)		0.0002 (0.001)		0.0001 (0.002)
Female	0.003 (0.01)		0.02 (0.02)		0.05 (0.04)
Visible minority	0.01 (0.01)		0.03 (0.03)		0.02 (0.08)
Immigrant	0.01 (0.01)		0.02 (0.02)		0.09 (0.06)
High school diploma	0.01 (0.01)		0.001 (0.02)		0.05 (0.04)
College/Cégep/trade school diploma	0.004 (0.01)		0.02 (0.03)		0.05 (0.04)
Bachelor's degree	0.002 (0.01)		0.03 (0.03)		0.06 (0.05)
Master's degree or above	0.01 (0.01)		0.02 (0.03)		0.004 (0.07)
Constant	0.01 (0.02)	0.01Footnote *** (0.003)	0.004 (0.05)	0.006 (0.006)	0.11 (0.11)	0.04Footnote * (0.02)
Percentage of outliers	5.9	X	4.6	X	5.1	X
Robust R2 (w)	0.06	0.003	0.09	0.003	0.15	0.03
Robust R2 (rho)	0.02	0.0008	0.03	0.001	0.05	0.01
N	7,138	7,138	2,316	2,316	587	587

Table A6: Impact of the CSBFP on Profit Margin Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	0.004 (0.005)	0.003 (0.004)	0.003 (0.01)	0.004 (0.005)	0.004 (0.01)	0.008 (0.007)
Lag growth (2012–2013)	0.001Footnote *** (0.00001)		0.01 (0.005)		0.13Footnote *** (0.02)
Firm size	0.00004Footnote * (0.00002)		0.00001 (0.0001)		0.0004 (0.0004)
Firm age	0.00003 (0.0001)		0.0002 (0.0002)		0.0005 (0.001)
Return on assets — Quartile 2	0.002 (0.01)		0.01 (0.01)		0.02 (0.03)
Return on assets — Quartile 3	0.001 (0.005)		0.01 (0.01)		0.01 (0.02)
Return on assets — Quartile 4	0.0002 (0.005)		0.01 (0.01)		0.03 (0.02)
Leverage — Quartile 1	0.01Footnote *** (0.003)		0.01Footnote ** (0.01)		0.02 (0.02)
Leverage — Quartile 2	0.01Footnote *** (0.002)		0.01Footnote ** (0.01)		0.01 (0.01)
Leverage — Quartile 3	0.01Footnote * (0.003)		0.01 (0.01)		0.01 (0.01)
Exporter	0.0002 (0.003)		0.01 (0.01)		0.01 (0.02)
Innovation	0.004 (0.002)		0.01 (0.01)		0.004 (0.01)
Urban	0.001 (0.003)		0.01 (0.01)		0.02 (0.01)
Franchise	0.001 (0.004)		0.001 (0.01)		0.004 (0.01)
Growth intentions	0.004 (0.003)		0.01 (0.01)		0.01 (0.01)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.00003 (0.0001)		0.0003 (0.0002)		0.0002 (0.001)
Female	0.003 (0.003)		0.001 (0.01)		0.01 (0.01)
Visible minority	0.004 (0.004)		0.002 (0.01)		0.02 (0.02)
Immigrant	0.001 (0.003)		0.0003 (0.01)		0.01 (0.01)
High school diploma	0.01* (0.01)		0.03Footnote *** (0.01)		0.01 (0.02)
College/Cégep/trade school diploma	0.01Footnote * (0.01)		0.02Footnote ** (0.01)		0.002 (0.02)
Bachelor's degree	0.01 (0.01)		0.02 (0.01)		0.005 (0.02)
Master's degree or above	0.01 (0.01)		0.02Footnote * (0.01)		0.002 (0.02)
Constant	0.01 (0.01)	0.0003 (0.001)	0.04Footnote * (0.02)	0.001 (0.003)	0.01 (0.04)	0.005 (0.006)
Percentage of outliers	16.0	X	14.1	X	16.5	X
Robust R2 (w)	0.02	0.0001	0.05	0.0003	0.28	0.004
Robust R2 (rho)	0.004	0.00002	0.01	0.00008	0.02	0.0008
N	7,138	7,138	2,316	2,316	587	587

Table A7: Impact of the CSBFP on Labour Productivity Growth between 2014 and 2016

Variable	All SMEs		Approved Borrowers		Denied Borrowers
	I	II	III	IV	V	VI
Note 1: Robust standard errors are in parentheses. Note 2: "X" indicates data were not released to meet confidentiality requirements of the Statistics Act. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	0.02 (0.01)	0.03Footnote * (0.01)	0.02 (0.02)	0.03Footnote * (0.02)	0.05 (0.04)	0.02 (0.03)
Lag growth (2012–2013)	0.0003Footnote *** (0.00001)		0.01Footnote *** (0.001)		0.0003Footnote *** (0.00002)
Firm size	0.0003Footnote *** (0.0001)		0.0003Footnote *** (0.0001)		0.0005 (0.001)
Firm age	0.00005 (0.0002)		0.0003 (0.0003)		0.001 (0.001)
Return on assets — Quartile 2	0.003 (0.01)		0.01 (0.02)		0.03 (0.04)
Return on assets — Quartile 3	0.002 (0.01)		0.01 (0.02)		0.03 (0.04)
Return on assets — Quartile 4	0.003 (0.01)		0.004 (0.02)		0.04 (0.05)
Leverage — Quartile 1	0.01 (0.01)		0.01 (0.02)		0.06 (0.05)
Leverage — Quartile 2	0.002 (0.01)		0.01 (0.01)		0.01 (0.03)
Leverage — Quartile 3	0.001 (0.01)		0.02 (0.01)		0.05 (0.03)
Exporter	0.002 (0.01)		0.02 (0.02)		0.06 (0.06)
Innovation	0.004 (0.01)		0.03Footnote ** (0.01)		0.01 (0.03)
Urban	0.002 (0.01)		0.004 (0.01)		0.04 (0.03)
Franchise	0.01 (0.01)		0.03 (0.02)		0.01 (0.04)
Growth intentions	0.003 (0.01)		0.01 (0.02)		0.01 (0.03)
Industry	Yes	No	Yes	No	Yes	No
Region	Yes	No	Yes	No	Yes	No
Owner Characteristics
Age	0.001Footnote ** (0.0004)		0.001 (0.001)		0.0002 (0.001)
Female	0.001 (0.01)		0.03 (0.02)		0.0004 (0.04)
Visible minority	0.01 (0.02)		0.02 (0.04)		0.001 (0.09)
Immigrant	0.01 (0.01)		0.01 (0.02)		0.05 (0.06)
High school diploma	0.002 (0.02)		0.003 (0.03)		0.08 (0.05)
College/Cégep/trade school diploma	0.02 (0.02)		0.005 (0.03)		0.08 (0.05)
Bachelor's degree	0.01 (0.02)		0.01 (0.03)		0.10Footnote ** (0.05)
Master's degree or above	0.01 (0.02)		0.03 (0.03)		0.15Footnote ** (0.07)
Constant	0.08 (0.03)	0.02Footnote *** (0.004)	0.08 (0.05)	0.02Footnote *** (0.01)	0.09 (0.10)	0.01 (0.03)
Percentage of outliers	4.6	X	1.7	X	4.8	X
Robust R2 (w)	0.10	0.0007	0.12	0.002	0.18	0.002
Robust R2 (rho)	0.03	0.0003	0.04	0.0009	0.06	0.0006
N	7,138	7,138	2,316	2,316	587	587

Table A8: Impact of the CSBFP on the Level of Growth in Economic Indicators between 2014 and 2016

Variable	Revenue ($)	Salary ($)	Profits ($)	Assets ($)	Employment	Profit Margin	Labour Productivity ($)
Note: Robust standard errors are in parentheses. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010—2016; payroll deduction data; and authors' calculations.
Firm Characteristics
CSBFP borrower	51,899.53Footnote *** (15,678.75)	4,287.83 (4,737.85)	21,883.10Footnote ** (8,814.00)	4,332.76 (7,736.64)	0.15 (0.13)	0.002Footnote * (0.002)	1,223.38 (1,654.86)
Lag growth (2012–2013)	0.04Footnote *** (0.01)	0.02Footnote *** (0.001)	0.04Footnote *** (0.001)	0.03Footnote *** (0.004)	0.05Footnote *** (0.01)	0.02 (0.01)	0.01 (0.01)
Firm size	1,283.93Footnote * (722.20)	451.29Footnote *** (117.82)	903.43Footnote *** (270.72)	394.33 (292.38)	0.01 (0.01)	0.00002Footnote * (0.00001)	28.33Footnote *** (9.88)
Firm age	251.55 (241.46)	1.23 (75.41)	157.75 (130.03)	6.53 (137.79)	0.001 (0.002)	0.000002 (0.00003)	23.98 (27.00)
Return on assets — Quartile 2	2,475.45 (13,141.30)	2,333.32 (3,273.41)	23,373.65Footnote *** (6,145.42)	5,157.56 (7,353.08)	0.02 (0.09)	0.003Footnote * (0.002)	1,590.37 (1,596.00)
Return on assets — Quartile 3	6,703.20 (12,420.88)	5,460.25Footnote * (3,148.73)	19,891.83Footnote *** (6,045.91)	549.31 (6,990.75)	0.05 (0.08)	0.01Footnote *** (0.002)	1,674.89 (1,516.74)
Return on assets — Quartile 4	15,397.98 (12,148.10)	1,045.01 (3,265.54)	30,229.05Footnote *** (6,007.53)	1,722.70 (6,795.19)	0.21 (0.09)	0.01Footnote *** (0.001)	1,434.11 (1,440.67)
Leverage — Quartile 1	10,907.10 (9,440.32)	2,076.52 (2,752.90)	15,484.58Footnote *** (5,088.60)	33,436.22Footnote *** (4,874.29)	0.07 (0.07)	0.01Footnote *** (0.001)	1,934.33 (1,163.84)
Leverage — Quartile 2	2,037.13 (9,282.84)	919.58 (2,652.63)	7,353.87 (5,075.95)	14,110.47Footnote *** (4,450.83)	0.03 (0.07)	0.003Footnote *** (0.001)	125.39 (1,028.32)
Leverage — Quartile 3	1,960.83 (10,109.76)	2,873.01 (2,755.60)	4,866.75 (5,148.47)	2,474.84 (4,515.84)	0.04 (0.08)	0.003Footnote *** (0.001)	916.34 (1,106.87)
Exporter	13,207.12 (15,855.66)	4,624.15 (3,850.54)	5,257.99 (7,945.15)	8,439.44 (7,300.62)	0.03 (0.09)	0.00002 (0.001)	1,705.60 (1,583.76)
Innovation	10,032.27 (8,155.50)	4,119.49Footnote * (2,273.72)	8,785.56Footnote ** (4,330.72)	2,987.42 (3,900.52)	0.003 (0.06)	0.002 (0.001)	552.41 (1,007.26)
Urban	4,869.99 (8,930.21)	386.92 (2,468.38)	443.09 (4,645.34)	871.46 (4,124.51)	0.09 (0.06)	0.0001 (0.001)	496.98 (1,020.81)
Franchise	7,972.88 (15,826.97)	1,697.40 (4,282.37)	19,394.71Footnote ** (7,609.36)	8,658.48 (7,002.91)	0.05 (0.14)	0.002 (0.002)	885.00 (1,393.40)
Growth intentions	60,230.11Footnote *** (11,511.87)	15,681.55Footnote *** (3,120.79)	24,093.35Footnote *** (6,003.47)	20,103.42Footnote *** (5,182.19)	0.26 (0.08)	0.002 (0.001)	527.44 (1,331.89)
Industry	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Owner Characteristics
Age	1,290.26Footnote *** (381.41)	286.37Footnote *** (108.57)	280.74 (199.92)	622.78Footnote *** (180.25)	0.005 (0.003)	0.00001 (0.00004)	85.50Footnote * (45.36)
Female	10,413.69 (9,717.76)	1,034.24 (2,792.53)	5,123.46 (5,009.72)	9,447.71Footnote ** (4,545.32)	0.06 (0.08)	0.002 (0.001)	395.70 (1,162.71)
Visible minority	15,122.84 (13,443.19)	5,207.64 (4,056.81)	9,457.66 (7,213.38)	6,552.68 (5,830.46)	0.10 (0.11)	0.002 (0.002)	295.29 (1,785.08)
Immigrant	8,112.21 (10,499.36)	652.55 (2,936.38)	1,415.02 (5,538.62)	5,656.11 (4,836.66)	0.04 (0.08)	0.002 (0.001)	304.82 (1,337.34)
High school diploma	231.11 (14,194.08)	6,465.74 (4,042.04)	1,604.32 (7,748.35)	4,181.55 (6,970.52)	0.01 (0.10)	0.004Footnote ** (0.002)	526.91 (2,158.19)
College/Cégep/trade school diploma	11,695.68 (13,699.51)	5,314.58 (3,986.08)	12,733.89Footnote * (7,494.83)	7,053.74 (6,831.62)	0.0003 (0.10)	0.004Footnote ** (0.002)	983.29 (2,133.34)
Bachelor's degree	6,386.70 (15,621.08)	1,969.52 (4,361.31)	10,802.90 (8,354.02)	4,843.36 (7,545.44)	0.16 (0.12)	0.004Footnote * (0.002)	1,815.28 (2,183.54)
Master's degree or above	22,960.90 (16,859.61)	2,976.18 (4,918.87)	9,338.73 (9,596.00)	13,214.66 (9,171.91)	0.17 (0.12)	0.003 (0.002)	1,096.75 (2,415.82)
Constant	80,210.58 (29,762.51)	13,465.54 (8,440.71)	56,068.51Footnote *** (15,312.25)	37,425.30Footnote *** (14,453.10)	0.30 (0.23)	0.001 (0.004)	10,370.01Footnote ** (4,103.16)
Percentage of outliers	20.5	17.9	18.5	20.3	15.0	12.0	12.1
Robust R2 (w)	0.09	0.11	0.30	0.09	0.08	0.03	0.05
Robust R2 (rho)	0.01	0.01	0.01	0.01	0.008	0.006	0.01
N	7,138	7,138	7,138	7,138	7,138	7,138	7,138

Unselect

Table A9: Marginal Effects on the Probability of Firms Surviving until 2016

Variable	Estimated Marginal Effects
	All SMEs	Approved Borrowers	Denied Borrowers
Note: Standard errors are in parentheses and derived using the delta method. Sources: Statistics Canada, Survey on Financing and Growth of Small and Medium Enterprises, 2014; Canada Revenue Agency, General Index of Financial Information 2010–2016; payroll deduction data; and authors' calculations.
CSBFP borrower	0.03Footnote *** (0.01)	0.01 (0.01)	0.05Footnote ** (0.02)
Firm size	0.02Footnote *** (0.003)	0.01Footnote ** (0.004)	0.03Footnote ** (0.01)
Firm age	0.02Footnote *** (0.003)	0.01Footnote *** (0.005)	0.004 (0.01)
Croissance des revenus en 2013-2014 — Quartile 2	0.03Footnote *** (0.01)	0.02Footnote ** (0.01)	0.05Footnote *** (0.02)
Growth revenue 20132014 — Quartile 3	0.04Footnote *** (0.01)	0.03Footnote *** (0.01)	0.04Footnote ** (0.02)
Growth revenue 20132014 — Quartile 4	0.04Footnote *** (0.01)	0.03Footnote *** (0.01)	0.08Footnote *** (0.02)
Return on assets — Quartile 2	0.04Footnote *** (0.01)	0.03Footnote *** (0.01)	0.03 (0.03)
Return on assets — Quartile 3	0.03Footnote *** (0.01)	0.02Footnote ** (0.01)	0.001 (0.03)
Return on assets — Quartile 4	0.02Footnote ** (0.01)	0.01 (0.01)	0.05 (0.04)
Leverage — 1er quartile	0.03Footnote *** (0.01)	0.02Footnote * (0.01)	0.01 (0.03)
Leverage — Quartile 2	0.05Footnote *** (0.01)	0.04Footnote *** (0.01)	0.04Footnote ** (0.02)
Leverage — Quartile 3	0.03Footnote *** (0.01)	0.02Footnote *** (0.01)	0.03 (0.02)
Exporter	0.01 (0.01)	0.004 (0.01)	0.03 (0.03)
Innovation	0.01 (0.01)	0.01 (0.01)	0.003 (0.02)
Urban	0.0001 (0.01)	0.01 (0.01)	0.03Footnote * (0.02)
Franchise	0.003 (0.01)	0.02 (0.01)	0.02 (0.03)
Growth intentions	0.03Footnote *** (0.01)	0.01 (0.01)	0.02 (0.02)
Owner age — Quartile 1	0.003 (0.01)	0.01 (0.01)	0.04Footnote * (0.02)
Owner age — Quartile 2	0.01Footnote * (0.01)	0.02 (0.01)	0.06Footnote ** (0.02)
Owner age — Quartile 3	0.02Footnote ** (0.01)	0.001 (0.01)	0.04Footnote * (0.02)
Female	0.02Footnote * (0.01)	0.02 (0.01)	0.02 (0.03)
Visible minority	0.003 (0.01)	0.0003 (0.01)	0.02 (0.03)
N	8,658	2,760	742

---

Appendix B: Variables Used in Propensity Score Matching

Table B1: Variables Used in Propensity Score Matching

Variable	Definition
Firm Characteristics
Firm size	Natural logarithm of firm size in 2013.
Firm age (Reference: Quartile 4)	Dummy variables for each quartile of firm age.
Revenue growth 2013–14 (Reference: Quartile 1)	Dummy variables for each quartile of revenue growth between 2013 and 2014.
Return on assets (ROA) (Reference: Quartile 1)	Dummy variables for each quartile of ROA, where ROA = profits/assets. Quartiles are calculated by industry sector.
Leverage (Reference: Quartile 4)	Dummy variables for each quartile of leverage, where leverage = liabilities/assets. Quartiles are calculated by industry sector.
Export	Dummy variable indicating firm exported in 2014.
Innovation	Dummy variable indicating firm developed or introduced an innovation between 2012 and 2014.Footnote 1
Urban	Dummy variable indicating firm operated in a census metropolitan area (CMA).Footnote 2 in 2014.
Franchise	Dummy variable indicating firm was a franchise in 2014.
Growth intentions	Dummy variable indicating expected average yearly growth in sales or revenues, between 2015 and 2017, of more than 10 percent.
Industry sector (Reference: Retail trade)	Dummy variables for the following industry sectors: agriculture, forestry, fishing and hunting; mining and oil and gas extraction; manufacturing; construction; wholesale trade; transportation and warehousing; information and communication technology (ICT); professional, scientific and technical services; accommodation and food services; and other services (except public administration).Footnote 3
Region (Reference: Ontario)	Dummy variables indicating firm operated in the following regions: Atlantic (New Brunswick, Prince Edward Island, Nova Scotia, and Newfoundland and Labrador), Quebec, Prairies (Manitoba, Saskatchewan), Alberta, British Columbia and Territories (Yukon, Northwest Territories and Nunavut).
Owner Characteristics
Age (Reference: Quartile 4)	Dummy variables for each quartile of owner age.
Gender	Dummy variable indicating percentage of business owned by women is greater than 50 percent.
Visible minority status	Dummy variable indicating percentage of business owned by a person from a visible minority group is greater than 50 percent.

---

Appendix C: Propensity Score Matching Quality Tests and Robustness Checks

Best practices and diagnostics outlined in Stuart and Rubin (2008), Caliendo and Kopeinig (2008) and Garrido et al. (2014) were followed to assess the validity of the assumptions underlying the matching procedure. Generally speaking, the matched observations in the control group should be comparable with those in the treatment group; differences in outcomes, then, can be attributed to the treatment.^{Footnote 24}

Propensity score matching yields a well-identified estimate of the average treatment effect on the treated only when key assumptions hold.^{Footnote 25} The first assumption is the common support, or overlap, condition (Caliendo and Kopeinig 2008). In this context, common support means that firms with the same values of observable covariates have a positive probability of being a CSBFP borrower or a non-CSBFP borrower.

Common support is typically assessed using the estimated propensity scores. In particular, the common support assumption is satisfied if propensity scores overlap for both the treatment group (CSBFP borrowers) and the control group (non-CSBFP borrowers). Visual inspection of the distribution of propensity scores, for each group, is one method of evaluating common support. The (visual) similarity between graphs of propensity score distributions for the treatment and control groups suggests that this assumption is satisfied.

The second key assumption is that the treatment and control groups are balanced with respect to observable variables, that is, the distributions (e.g., means of continuous variables or proportions of categorical variables) of the covariates in both groups should be similar.

To assess balance, standardized differences (or standardized bias) were calculated.^{Footnote 26} A rule of thumb is that the standardized difference should not be greater than 10 percent (Ho et al. 2007). This is the case for nearly all covariates, for each economic indicator, suggesting that the covariates are balanced for each estimated treatment effect.

Another diagnostic for balance is visual inspection of propensity score distributions of the control and treatment groups. As noted above, these distributions are similar, which points to covariate balance. In addition, as suggested by Caliendo and Kopeinig (2008), propensity scores on the matched sample — CSBFP borrowers and matched non-CSBFP borrowers — were re-estimated and the joint significance of all covariates was estimated. For all economic indicators, likelihood ratio test statistics do not reject the null hypothesis that the covariates are jointly significant from zero, implying that the matched sample does not explain variation in the probability of being a CSBFP borrower.

The third key assumption is conditional independence, or unconfoundedness. This assumption, that unobservables affect neither the probability of treatment (being a CSBFP borrower) nor the outcome of treatment (economic growth), must hold for an unbiased estimate of the average treatment effect (CSBFP participation). In other words, if propensity scores control for all factors that impact the outcomes and probability of treatment, this assumption is met, and estimates will be unbiased (Becker and Caliendo 2007). By contrast, if unobservables do impact the outcomes or probability of treatment, the estimated treatment effect is biased.

Rosenbaum (2002) suggests a sensitivity test to determine if the estimator is robust to the possible presence of such unobservables.^{Footnote 27} This test suggests that propensity score matching estimates for the impact of CSBFP borrowing on percentage growth in revenues, profits and size are relatively robust to the possibility of unobserved heterogeneity, which points to the robustness of the estimated impacts on percentage growth in these economic indicators.

However, estimated impacts on the level of growth in revenues, profits and size may be sensitive to unobservables. Therefore, caution should be taken in interpreting the matching estimates of program impacts on the level of growth.

Finally, nearest-neighbour matching (i.e., matching on Mahalanobis distance^{Footnote 28}) was used to compare propensity score matching estimates. Overall, results are similar, which suggests that the latter estimates are robust.