SME Financing Data Initiative
sme-fdi.gc.ca
Common menu bar links
First
Previous
Financing Innovative Small and Medium-Sized Enterprises in Canada
V. Multivariate Analysis
The descriptive analysis presented so far has shown some differences between innovative and non-innovative SMEs in terms of financing activities, experiences, and terms and conditions. To investigate the differences in more depth, several research approaches were used: binary logistic regression model, multinomial logistic regression model and linear regression model.
It has been shown that innovative SMEs were more likely to request external financing than non-innovative SMEs. Moreover, innovative SMEs were more likely to request debt and equity financing than non-innovative ones. However, innovative SMEs were less successful in obtaining the loans. Four binary logistic models were employed in order to investigate if the differences were statistically significant after controlling for some firm-specific characteristics such as firm size and sector. The reason why the binary logistic models were used was because the dependent variables were dummy variables. For example, in the first binary logistic model, the dependent variable was whether (=1) or not (=0) a respondent had requested external financing. In the second binary logistic model, the dependent variable was whether (=1) or not (=0) a respondent has applied for debt financing. In the third binary logistic model, the dependent variable was whether (=1) or not (=0) a respondent had applied for equity financing. In the last binary logistic model, the dependent variable was whether (=1) or not (=0) a respondent got credit approved. The firm-specific characteristics used were: firm status (a dummy variable 1=start-up and 0=non-start-up); firm size (a dummy variable 1=number of full-time employees greater than or equal to 0 and 0=number of full-time employees fewer than 5); firm's growth intention (1=firm has growth intention and 0=firm has no growth intention); industrial sector; number of years with the financial institution; and the type of firm (1=innovative and 0=non-innovative).
In general, the results were reflective of those already presented in the descriptive analysis, as shown in Table 5. The results from Model 1 indicates that innovative SMEs were 1.68 times more likely to request external financing than non-innovative ones, after controlling for the firm-specific characteristics. Moreover, Model 2 and Model 3 show that innovative SMEs were 1.18 times more likely to request debt financing and 4.48 times more likely to request equity financing than non-innovative SMEs.
Table 6 shows the results from the last binary logistic model. The results from Model 4 confirm that innovative SMEs were less likely to get the credit approvals than non-innovative SMEs. The coefficient associated with innovative SMEs is negative and statistically significant at the p<0.05 level. The odds of getting credit authorized among innovative SMEs were estimated to be 26 percent lower than among non-innovative SMEs (1-0.74=0.26), after controlling for the firm-specific characteristics.
Multinomial logistic regression is the extension for the binary logistic regression when the categorical dependent variable outcome has more than two levels. In the following analysis, the dependent variable "collateral" has four levels: personal collateral required; business collateral required; both personal and business collaterals required; and neither required. "Neither was required" was chosen as a reference group for comparison.
Table 7 shows the results from the multinomial logistic regression. The second column of Table 7 has the outcome of "personal collateral required" compared to "neither was required." Comparing with innovative SMEs, non-innovative SMEs were about 3.5 times more likely to be asked for personal collateral. Conversely, we can say that innovative SMEs were less likely to be asked for personal collateral, odds ratio is 0.29 (given by the reciprocal of 3.5). The third column has the outcome of "business collateral required" compared to "neither was required." Non-innovative SMEs were 6.2 times more likely to be asked for business collateral compared to innovative SMEs. The last column has the outcome of "both personal and business collateral required" compared to "neither was required." Non-innovative SMEs were 1.3 times more likely to be asked to provide both types of collaterals.
Table 8 shows the linear regression results on loan terms for three types of loans: short-term loan, term loan and mortgage loan. For term loan, comparing the length of terms offered to innovative and non-innovative SMEs, we would expect the length of term for innovative SMEs be 29 months shorter, on average, holding all the other independent variables constant. It is worth noting that, for short-term loan and mortgage loan, the difference was not statistically significant.
Table 9a shows the linear regression results on fixed interest rates charged by the financial institution for six different financing instruments. The seventh financing instrument on fixed rates for increasing in the credit limit of current lines of credit was excluded from the analysis due to low response rates. For short-term loans, compared to the fixed rates paid by non-innovative SMEs, we would expect the fixed rates paid by innovative SMEs be 0.9 percentage points higher, on average, holding all the other independent variables constant. However, for mortgage loans and new credit cards, we would expect the fixed rates paid by innovative SMEs be 2.3 percent and 0.7 percent points lower. For term loans, it is not statistically significant that innovative SMEs paid higher fixed rates than non-innovative ones.
Table 9b shows the linear regression results on variable interest rates charged by the financial institution for six different financing instruments. The seventh financing instrument on variable rates for new credit card and increase in the credit limit of current credit cards were excluded from the analysis due to low response rates. For new line of credit and increase in credit limit of current lines of credits, compared to the variable rates paid by non-innovative SMEs, we would expect the variable rates paid by innovative SMEs be 0.1 percent and 0.9 percent points higher, holding all the other independent variables constant. However, for mortgage loans, we would expect the variable rates paid by innovative SMEs be 0.7 percent points lower. For term loans, it is not statistically significant that innovative SMEs paid higher variable rates than non-innovative SMEs.
