Economic Impact Study from 2001 to 2010

Download the PDF version

About this publication

Publication author : Canada Economic Development for Quebec Regions

Collaborator : Statistics Canada

Publish date : March 27, 2015

Summary :

In 2012, Canada Economic Development for Quebec Regions commissioned Statistics Canada a study on the impact of its programs, to understand whether its assistance to firms in Quebec over the period 2001 to 2010 had a significant impact on firm performances.

Table of Contents

  1. Summary of key findings
  2. 1. Introduction
  3. 2. A brief literature review
  4. 3. CED-Q clients list and matching process
  5. 4. Selection of comparison group
  6. 5. Hypotheses
  7. 6. The model
  8. 7. Empirical results
  9. 8. Conclusions
  10. References

Summary of key findings

The Canadian Economic Development Agency for the Regions of Quebec (CED-Q) promotes long-term economic development in Quebec, particularly in those regions where slow economic growth is prevalent or opportunities for productive employment are inadequate. CED-Q provides information, advice, guidance as well as financing primarily to small and medium-sized enterprises (SME’s).

In 2012, CED-Q commissioned Statistics Canada a study on the impact of its programs, to understand whether its assistance to firms in Quebec over the period 2001 to 2010 had a significant impact on firm performances. The business performance indicators used in this analysis are revenues, labour productivity and employment growth, as well as the survival rates of businesses.

The Centre for Special Business Projects (CSBP) of Statistics Canada created a database, which included CED-Q clients and a comparison group of non-(CED-Q)-clients. The comparison group was selected on the basis of comparable employment, revenues, assets, debt ratio and profit margin. Then, a regression analysis was implemented for each cohort of CED-Q clients, to investigate the impact of the programs.

Overall results:

1. Introduction

The Canadian Economic Development Agency for the Regions of Quebec (CED-Q) is a federal agency whose mandate is regional development in Quebec. CED-Q promotes local economic development and innovation by providing financial assistance to small and medium-sized enterprises (SMEs) in Quebec.

In 2008, CED-Q and Statistics Canada (STC) began a joint initiative to identify and measure the economic impact of the enterprises that use CED-Q financing services from 2002 to 2008. In 2012, STC and CED-Q agreed to revisit the work by extending the analysis from 2001 to 2010 and by performing the study for a number of different programs. The programs being studied are: Regional Strategic Initiatives Program—Community Diversification (RSI-CD) and Innovation, Development, Export and Assistance for entrepreneurship—Business and Regional Growth (IDEA-BRG).

This document describes the methodology used to perform the analysis and provides the findings. Following this introduction, which also highlights the strengths and limitations of this analysis, Section 2 briefly reviews the early contributions concerned with relations between economic assistance and firms’ performances. Section 3 presents the matching process. Section 4 explains the method employed to select the comparison group. Section 5 presents the hypotheses that were tested. Section 6 presents the adopted methodology. The empirical results are presented and discussed in Section 7. Improvements compared to previous study are presented in section 8. Section 9 presents the limitations of the study. The last section concludes the report.

1.1. Improvements compared to the previous study

A significant contribution in this study, compared to the previous, was to perform a robustness check on the similarity between the study group and comparison group. This robustness is intended to explore the characteristic of CED-Q clients and non-clients prior to CED-Q contributions using the Kolmogorov–Smirnov test. This contribution is relevant because it ensures that the selected comparison group is closely comparable to the study group and, secondly, it provides further ground to attribute the change observed in the analysis to the CED-Q program. It is expected that firms receiving financial contribution will show higher performance indicators than a comparison group of firms with similar characteristics. But without full knowledge of the two groups, the comparison could yield biased estimates and thereby misleading interpretation. In addition, this study also presents the survival rates of CED-Q clients by type of program, which has not been done in the previous study (Annex 5).

1.2. Limitations and relevant implications

An additional caveat that should be kept in mind is about the selection of the comparison group. It is acknowledged that the selection of the comparison group is of crucial importance. In an ideal scenario, this group should be composed of business that applied to CED-Q funding but were not ultimately selected for funding. Also, the current study does not account for the fact that some business received multiple funding or benefited from other sources or types of funding. These limitations do suggest areas for future research. For instance, it would be useful, in the future, to have a list of enterprises that applied to CED-Q programs and did not succeed, to see if the pattern of enterprises that did not get funding had different characteristics vis à vis to those in the comparison group.

Some data issues should be kept in mind when interpreting these results. The R&D data showed some significant limitations when used in regression analysis. This limitation was due to the high frequency of missing values, which reduced substantially the number of usable observations. Therefore, they were omitted from the modeling.

Another data concern is related to the size of Regional Strategic Initiatives Program—Community Diversification (RSI-CD), which reported a relatively small number of beneficiaries. The last data limitation concerns the export data which was not available during the completion of the analysis.

2. A brief literature review

Assistance of firms has become a common practice in most industrialized countries. The reason for governments to intervene in the market is to foster economic growth by overcoming some type of market failure. For instance, financial assistance to SMEs is often viewed as essential due to their limited internal resources and perceived vulnerability to external competition (Holm-Pedersen et al., 2009). In most OECD countries, SMEs are engines for innovation and jobs creation.

A commonly used tool to measure the financial assistance on firm performance is the counterfactual analysis. The approach of this impact study is that of a counterfactual analysis. This methodology is well established in the literature and widely applied in economic impact studies conducted by major organizations (Lopez Acevedo, 2010; Abrell et al., 2011).

In an impact study, a counterfactual is what would have happened without a given intervention. In the literature there are two ways in which the counterfactual can be approximated: (1) using the outcomes observed for non-beneficiaries over the same period of time; or (2) using the outcomes observed for beneficiaries before they are exposed to the intervention. In most cases, however, the counterfactual is generated using the first method, that is observing the outcome of non-beneficiaries and comparing it with beneficiaries of a given intervention.

The key methodological challenge of this type of impact study is to identify the counterfactual, that is, a group which is as similar as possible to the group of beneficiaries of the given intervention or program.

The underlying idea of this analysis is that once a proper counterfactual is identified a comparison of outcomes across the two groups allows for the establishment of causality in attributing the observed changes to the intervention, while controlling for other confounding factors.

For instance, Bradshaw (2002) studied the effectiveness of the California State Loan Guarantee Program, which guaranteed small business bank loans to firms that could not otherwise obtain credit. The study tracked the change in employment of firms that received loan guarantees from 1990 to 1996 during the depths of the California recession. The study found that employment increased in firms receiving loan guarantees by 40% among all firms and 27% among non agricultural firms.

Using survey data on the Financing of Small and Medium Enterprises, Chandler (2010) investigated the economic impact of the Canada Small Business Financing Program. The analysis found that the program is responsible for the creation of 0.63 jobs per businesses participating in the program and for an increase of $78 352 in revenues. The results indicate there was no statistically significant impact on profits and salaries.

Exploiting Irish Economy Expenditure Survey (IEE) data, Girma, et al. (2007) examined the impact of government subsidies on productivity growth at the plant level during 1992-1998. The results reveals that only grants that support productivity enhancing activities increase productivity.

Ankarhem et al. (2010) looked at the effect of Swedish regional investment grants during 1990-1999 on firm performance, in terms of returns on equity and number of employees. They found that firms that received grants did not perform better in terms of returns on equity when compared to firms in the control group. They also found that recipient firms did not hire more employees.

3. CED-Q clients list and matching process

An Excel file containing 3,088 businesses that benefited from CED-Q financing services from 2001 to 2012 was submitted to the CSBP, Statistics Canada, by CED-Q. This file contained one row for each funded project, with some businesses (identified on the file by clientsID) having multiple rows. The rows after the initial year of assistance were removed. For instance, if an enterprise received CED-Q support in 2001 and 2002, only the initial year of assistance, 2001 was kept.

Out of the original 3,088 records, 690 were found to be duplicates and 5 enterprises had an initial funding year of 2011 or 2012 (outside the period under study, not included in the figure 1). Figure 1 below shows the numbers of records after duplicates have been removed.

Figure 1 Removal of duplicates from the CED-Q client list

Figure 1 Removal of duplicates from the CED-Q client list

Figure 1 – Long Description

Figure 1 shows the removal of duplicates from the CED client list in a three-bar chart. The first bar indicates the number of clients on the initial list, a total of 3,088 clients. The second bar indicates the duplications that were eliminated from the initial list, a total of 690 clients. The third bar indicates the number of clients remaining in the clean list, a total of 2,393 clients.

3.1 Matching the clean file to the Business Register

After cleaning the CED-Q file, a total of 2,393 CED-Q records remained. Of these, 2,216 were successfully matched to Business Register to obtain their EnterpriseID (a Statistics Canada unique identifier). No match was found for 177 clients. There are a number of reasons why a CED-Q client may not be identified in Statistics Canada’s Business Register. One of the main reasons is the administrative changes related to ownership, mergers and acquisitions, or partnership arrangements that can result in the issuing of new business numbers or the alteration of other key identifiers associated with an enterprise.

Table 1 provides the match rate by year and type of program. The match rate is calculated as the total number of successful matches to the Business Register divided by the total number of the client’s records in the clean file.

As can be seen from Table 1, a high match rate was found for all programs and for all years considered in the analysis. Depending on the year considered, the match rate ranges from 79% to 98% for Regional Strategic Initiatives Program—Community Diversification (RSI-CD), while it ranges from 81% to 100% for Innovation, Development, Export and Assistance for entrepreneurship—Business and Regional Growth (IDEA-BRG) program. The average match rate for all programs combined is about 93%.

Table 1 Matching rate to Business Register, by year and type of program
RSI-CD IDEA-BRG All programs
First year of funding Records Matched Match rate Records Matched Match rate Records Matched Match rate
2001 15 13 86.7% 68 63 92.6% 83 76 91.6%
2002 23 18 78.3% 205 191 93.2% 228 209 91.7%
2003 42 41 97.6% 300 279 93.0% 342 320 93.6%
2004 51 44 86.3% 320 308 96.3% 371 352 94.9%
2005 47 43 91.5% 236 220 93.2% 283 263 92.9%
2006 106 96 90.6% 157 153 97.5% 263 249 94.7%
2007 87 77 88.5% 145 145 100.0% 232 222 95.7%
2008 10 9 90.0% 172 154 89.5% 182 163 89.6%
2009 17 16 94.1% 212 201 94.8% 229 217 94.8%
2010 34 27 79.4% 146 118 80.8% 180 145 80.6%
Total 432 384 88.9% 1,961 1,832 93.4% 2,393 2,216 92.6%

3.2 Match rate of key variables

For the modeling to be robust, a high percentage of records with useable values is required. A useable value is a value that is not missing.

The key variables of interest for the CED-Q impact study are revenue, employment, productivity (revenue over employment, or sales over employment) and research and development.

Table 2 presents the number and percentage of usable records for each key variable of interest. For the variables revenue, sales, and employment, the matched clients have on average over 77% of useable values; i.e., they have a value or a value equal to zero. Revenue has the higher percentage of useable values (88%) followed by sales (81%) and employment (77%). After 2003, revenue, sales, and employment have shown a rate of useable values exceeding 71%. Research and Development in Canadian Industry (RDCI) has a rate of useable values of less than 50%Footnote 1. Data for 2010 is not available for RDCI.

The results show the match rate and the availability of the key variables of interest were sufficiently high to complete the modeling analysis of the impact study.

Table 2 Usable values in the variables of interest, all programs combined
Total revenue Total sales Average employment RDCI expenditure
First year of funding Matched Useable values % Useable values % Useable values % Useable values %
2001 76 50 65.8% 43 56.6% 38 50.0% 24 31.6%
2002 209 149 71.3% 129 61.7% 118 56.5% 71 34.0%
2003 320 264 82.5% 246 76.9% 234 73.1% 136 42.5%
2004 352 300 85.2% 270 76.7% 250 71.0% 152 43.2%
2005 263 240 91.3% 214 81.4% 201 76.4% 111 42.2%
2006 249 226 90.8% 214 85.9% 203 81.5% 112 45.0%
2007 222 213 95.9% 193 86.9% 189 85.1% 106 47.7%
2008 163 149 91.4% 140 85.9% 139 85.3% 86 52.8%
2009 217 211 97.2% 211 97.2% 206 94.9% 127 58.5%
2010 145 141 97.2% 135 93.1% 137 94.5%
Total 2,216 1,943 87.7% 1,795 81.0% 1,715 77.4% 925 41.7%

4. Selection of comparison group

Comparison group selection was done via the Statistics Canada’s Generalized Edit and Imputation software-BANFF, using the nearest-neighbour approach. CED-Q client records were flagged, with a variable indicating the year in which they first received funding. For each year, a comparison was selected using the following the criteria:

To identify similar records, the following matching variables were used:

Comparison records selection was restricted to the same NAICS classification of the CED-Q clients and within Quebec. From the 2,216 study group records matched to Business Register, any study group records or potential comparison group records missing any of the above variables were excluded from the process, resulting in counts shown in Table 3. The ratio of the number of study group to potential comparison group is also indicated in Table 3.

Table 3 Selection of comparison records
First year funding Study records Potential comparison records Ratio
2001 66 162,988 0.04%
2002 193 159,733 0.12%
2003 280 158,270 0.18%
2004 327 160,550 0.20%
2005 233 166,004 0.14%
2006 216 170,768 0.13%
2007 206 173,422 0.12%
2008 145 172,651 0.08%
2009 204 174,046 0.12%
2010 134 172,081 0.08%
Total 2,004 1,670,513 0.12%

5. Hypotheses

A series of hypotheses were tested to measure the performance outcomes of the CED-Q clients and similar enterprises that did not. This is regarded as the performance effect. The hypotheses examined were:

Hypotheses: CED-Q financing generates the following outcomes:

  1. Higher revenue growth
  2. Higher employment growth
  3. Higher labour productivity growth
  4. Higher survival rate

The CED-Q mission is to help regional economic development through businesses financing of small and medium-sized enterprises. This study was conducted to measure and assess the performance of enterprises that used CED-Q’s services to similar enterprises that did not, during the period 2001 to 2010. The empirical literature dealing with enterprise performance often cites factors such as revenue growth, employment growth and labour productivity as relevant indicators of performance. As such, the four above mentioned hypotheses were tested to assess the effectiveness of the CED-Q in influencing the economic performance of its clientele.

6. The model

The differences in revenue, productivity and employment growth between CED-Q-supported enterprises and non-supported enterprises are investigated based on the following general empirical model:

lnyit+1 - lnyit = a +β CEDQit + d Controlitit

where i is the index of the enterprise, t is the index of the year, y is the hypothesized variable (e.g. revenue growth, employment growth and labour productivity growth), lnyit+1 - lnyit is the growth of y , CEDQ is a dummy variable for CED-Q clients (takes 1 if the enterprise benefited from CED-Q funding in year t, 0 else), Control is a vector of control variables and εit is an error term.

The variable of interest, the coefficient β captures the effect for enterprises that received CED-Q support. For instance, if the estimated β in the model (1) is equal to 0.06 and is statistically significant at a usual error level, that means, supported enterprises experience on average 6% higher revenue growth compared to non–clients for the given year.

It should be noted that the estimations are based on the data for all CED-Q programs pooled together.

To test for the impact of the program on revenue growth, the following regression is run:

lnRevit+1 - lnRevit = a + ß CEDQit + d1TAit + d2LTLit + d3LPit + d4GPit + εit

where the specific variables included in the control vector are: total asset (TA) and long term liabilities (LTL), which control for the financial status of the firm; labour productivity (LP) computed as total sales over employment, and gross profit/loss (GP), which control for changes in firm performance. All the explanatory variables are expressed in growth terms and not in level terms.

To test for the impact of the program on employment growth (Empl), the following model is used:

lnEmplit+1 - lnEmplit = a + β CEDQit + d1TSGSit + d2TAit + d3GPit + d4TEit + εit

where total asset (TA) controls for the financial status of the firm, total sales of goods and services (TSGS) and gross profit/loss control for the performance indicators; and TE is the total expenses.

Lastly, the effect on labour productivity (LP) is investigated using the following specification:

lnLPit+1 - lnLPit = a + β CEDQit + d1TAit + d2GPit + d3TEit + d4AEMPLit + εit

where AEMP is average employment, which controls for firm characteristics such as size, and TA, GP and TE are as specified above.

The regression equations (2), (3) and (4) are intended to be an empirical model to explain revenue growth, employment growth and labour productivity growth between the CED-Q clients and non-clients.

The impact of a financing program is unlikely to be observable within the same reference year, for both operational as well as statistical reporting reasons. To investigate the impact of CED-Q financing on the short, medium and long run, the following lead structure of response variables was used to study the impact of CED-Q services for the reference years:

t = 2001: L → [t +1, t+2, t+3, t+4, t+5, t+6, t+7, t+8, t+9]

t = 2002: L → [t +1, t+2, t+3, t+4, t+5, t+6, t+7, t+8]

t = 2003: L → [t +1, t+2, t+3, t+4, t+5, t+6, t+7]

t = 2004: L → [t +1, t+2, t+3, t+4, t+5, t+6]

t = 2005: L → [t +1, t+2, t+3, t+4, t+5]

t = 2006: L → [t +1, t+2, t+3, t+4]

t = 2007: L → [t +1, t+2, t+3]

t = 2008: L → [t +1, t+2]

t = 2009: L → [t +1]

For instance, for a firm that received funding for the first time in 2001, its performance is compared in terms of revenue, employment and labour productivity growth to the performance of the potential comparison group for the first year after the funding (t+1) and all subsequent years until 2010 (that is, t+9).

For each cohort, a unique identifier (operating entity number) is used to track the business revenue, employment, assets, liabilities, gross profit/loss, total expenses etc, over the years; this permits following the two groups of businesses for several years after the initial funding. Control variables were allowed to vary with the lead structure, being taken from one year behind the lead period for which the impact was assessed. For example, when the lead period was 2005, control variables were calculated based on the change between 2004 and 2005. This allowed the impact of ongoing changes in the control variables on the response variable to be isolated.

The most common regression model used to investigate the effect of the independent variables on the dependent variable is the Ordinary Least Square (OLS) estimator. However, when outliers are present in the data such as vertical outliers, bad leverage points, and good leverage points, the OLS estimation may be biased. The OLS estimator is known to be vulnerable to the presence of outliers. One way to deal with outliers is to exclude the 1st and the 99th percentiles distribution of the variable of interest from all computations. An alternative way is to use a robust regression.

To overcome the outliers’ problem, Huber (1981) proposes the M-estimators which is robust with respect to vertical outliers but is not in the presence of bad leverage points. To correct for outliers’ issue, Yohai (1987) proposes a general measure that minimizes the dispersion of the residuals that is less sensitive to extreme values. Therefore, Yohai (1987) introduces an estimator that combines a high breakdown point and a high efficiency, the MM-estimator. This estimator is known to yield unbiased estimators because it corrects for all types of outliers that influence the OLS estimator. For robustness purpose, the OLS, the M-estimator and the MM-estimator have been applied; it was found that the MM-estimator has the desirable properties. Therefore, the modeling results presented in this report are based on the MM-estimator.

7. Empirical results

7.1 Profiling

In this section, we compare the distribution of CED-Q-supported enterprises with non-supported enterprises that represent the comparison group. The comparison is made with respect to revenue and employment, as recorded several years before the supported enterprises began receiving financing. To put it differently, we assess whether today’s assisted enterprises are different from non-assisted enterprises based on past values of selected indicators, when all of them did not benefit from the program. The backward tracking time-frame that we consider is three years. The empirical strategy used here to test for the difference between the two groups is the nonparametric test for first order stochastic dominance of one distribution over another that was introduced by Kolmogorov–Smirnov.

An attractive feature of the Kolmogorov–Smirnov test is that it considers all the moments rather than focusing on testing the differences in the mean value for both groups. Hence, the Kolmogorov–Smirnov test is the most appropriate test that compares the empirical distribution differences between two groups.

This nonparametric technique has been applied, for the first time in the productivity literature, to compare productivity distributions of exporting and non-exporting plants (Delgado et al. 2002). Following this seminal contribution, a growing body of empirical literature has used the technique to discuss the issue of productivity, profitability of exporting and non-exporting as well as the pre-entry differences in productivity between import starters and non-importers (see Helmut and Wagner 2010; and Bellone et al. 2010).

More formally, this method tests for stochastic dominance of a variable (revenue or employment, in the case of this study) distribution for one group over the same variable distribution for another group. For instance, let F and G denote the cumulative distribution functions of revenue for supported enterprises and non-supported enterprises, respectively. If F(x) - G(x) = 0 means that the two distributions do not differ, and the first order stochastic dominance of F relative to G means that F(z) - G(z) must be less than or equal to zero for all values of z, with strict inequality for some z.

The null hypothesis for the Kolmogorov–Smirnov test is that the distribution of employment for the two groups considered are identical against the alternative hypothesis that the distribution for the second group first-stochastically dominates the distribution of the first group. If the p-value falls below the specified alpha level, then there is evidence to suggest that the two populations are not identical.

The data set used in this empirical investigation covers the years 2001-2010. The analysis examined each cohort of enterprises that started benefiting from CED-Q program for the first time in 2002. For 2002, one year back was selected, for 2003, two years and for 2004 onwards it was possible to go back three years. For instance, if an enterprises starts to receive funding for the first time in 2004, we extracted data on revenue and employment for the CED-Q-supported enterprises as well as the comparison group three years back (i.e. for 2001, 2002 and 2003). The analysis was limited to three years time intervals to preserve the sample size.

In sum, the nonparametric method described above was applied to determine if there exists a significant difference in revenue and employment among two groups of enterprises.

7.1.1 Graphical representations

This subsection presents graphical descriptions of distributions of CED-Q-supported enterprises and non-supported enterprises for the enterprises that started receiving support in 2008. A graphical illustration permits visual comparisons between the distributions functions of the two groups; and is reported in Figure 2 to Figure 4 for the reference period of t-3, t-2, and t-1 for employment distribution. This is a graphic example for 2008. Graphs were also produced for all years (2002-2010).

Figure 2 Employment distribution of CED-Q-clients versus non-supported enterprises in t-3

Figure 2 Employment distribution of CED-Q-clients versus non-supported enterprises in t-3

Figure 2 – Long Description

Figure 2 uses two graph curves to show the employment distribution of CED-Q clients as compared to non-supported enterprises in t-3. The curves are superimposed in the chart.

 

Figure 3 Employment distribution of CED-Q-clients versus non-supported enterprises in t-2

Figure 3 Employment distribution of CED-Q-clients versus non-supported enterprises in t-2

Figure 3 – Long Description

Figure 3 uses two graph curves to show the employment distribution of CED-Q clients as compared to non-supported enterprises in t-2. The curves are superimposed in the chart.

 

Figure 4 Employment distribution of CED-Q clients as compared to non-supported enterprises in t-1

Figure 4 Employment distribution of CED-Q clients as compared to non-supported enterprises in t-1

Figure 4 – Long Description

Figure 4 uses two graph curves to show employment distribution of CED-Q clients as compared to non-supported enterprises in t-1. The curves are superimposed in the chart.

 

To give an intuitive explanation of these graphs, it is important to note that the test allows for the determination of whether both distributions are identical or not. Therefore, if the distribution of non-supported enterprises lies to the right (below) of the distribution of CED-Q-clients, it implies that non-supported enterprises stochastically dominate supported enterprises. In other words, the two groups are not the same (heterogeneous pattern). However, as can be seen from the graphs, there is no clear evidence to suggest that supported enterprises are different from non-supported and vice versa.

The same procedure has been used to depict distributions of revenue between CED-Q-supported enterprises and non-supported enterprises. Figure 5 through 7 show the empirical distributions of the two groups. Visually, once again there is no evidence to suggest that one group dominates the other group. In the next section, the statistical findings the Kolmogorov–Smirnov test results are provided.

Figure 5 Revenue distribution of CED-Q-clients versus non-supported enterprises in t-3

Figure 5 Revenue distribution of CED-Q-clients versus non-supported enterprises in t-3

Figure 5 – Long Description

Figure 5 uses two graph curves to show revenue distribution of CED-Q clients as compared to non-supported enterprises in t-3. The curves are superimposed in the chart.

 

Figure 6 Revenue distribution of CED-Q-clients versus non-supported enterprises in t-2

Figure 6 Revenue distribution of CED-Q-clients versus non-supported enterprises in t-2

Figure 6 – Long Description

Figure 6 uses two graph curves to show revenue distribution of CED-Q clients as compared to non-supported enterprises in t-2. The curves are superimposed in the chart.

 

Figure 7 Revenue distribution of CED-Q-clients versus non-supported enterprises in t-1

Figure 7 Revenue distribution of CED-Q-clients versus non-supported enterprises in t-1

Figure 7 – Long Description

Figure 7 uses two graph curves to show revenue distribution of CED-Q clients as compared to non-supported enterprises in t-1. The curves are superimposed in the chart.

 

7.1.2 Kolmogorov–Smirnov tests results

Table 4 presents the results for the Kolmogorov–Smirnov tests for employment differential between the CED-Q-supported enterprises and non-supported enterprises. In this table, the first column indicates the reference year in which the funding was first received. The second and third columns identify the Kolmogorov–Smirnov test statistics (KS test statistics) and the probability value (P-values) for year t-1 prior to first funding. As can be observed from Table 4, the null hypothesis of equality between the distributions of the two groups cannot be rejected at any reasonable level for all years considered. The Kolmogorov–Smirnov tests in Table 4 confirm the graphical presentation.

Following Table 4, Table 5 presents the hypotheses test statistics of revenue differential between the CED-Q-supported enterprises and non-supported enterprises. First, in the test applied to group of enterprises that benefited from CED-Q and non-supported enterprises from 2002 to 2009, we are not able to reject the null hypothesis of equality of distributions at any significance level and any year before the first financing. For the cohort of 2010, we are able to reject the null hypothesis of equality of revenue distributions at the 0.05 significance level for year t-2 and t-3 before the first funding. A possible explanation for rejecting the equality of distributions may be that two groups of enterprises are heterogeneous with regard to their revenue in year t-2 and t-3.

Table 4 Employment distribution differences between study and comparison groups, hypothesis test statistics
  KS test statistic P-values KS test statistic P-values KS test statistic P-values
First year funding Year t-1 Year t-2 Year t-3
2002 0.063 [0.858] ... ... ... ...
2003 0.046 [0.938] 0.056 [0.838] ... ...
2004 0.038 [0.979] 0.043 [0.953] 0.077 [0.379]
2005 0.039 [0.997] 0.063 [0.810] 0.061 [0.871]
2006 0.041 [0.995] 0.069 [0.763] 0.072 [0.765]
2007 0.065 [0.803] 0.065 [0.840] 0.084 [0.606]
2008 0.048 [0.996] 0.061 [0.972] 0.065 [0.960]
2009 0.071 [0.713] 0.058 [0.916] 0.079 [0.668]
2010 0.096 [0.590] 0.091 [0.688] 0.088 [0.765]

Note: Compare the p-values in the tables with [0.01], [0.05] significance levels. If the p-value in the tables is less than the chosen level of significance, there is evidence to suggest that the two populations are not the same.

Table 5 Revenue distribution differences between study and comparison groups, hypothesis test statistics
  KS test statistic P-values KS test statistic P-values KS test statistic P-values
First year funding Year t-1 Year t-2 Year t-3
2002 0.071 [0.744] ... ... ... ...
2003 0.059 [0.735] 0.065 [0.666] ... ...
2004 0.082 [0.240] 0.053 [0.795] 0.073 [0.429]
2005 0.038 [0.997] 0.057 [0.887] 0.073 [0.685]
2006 0.077 [0.567] 0.052 [0.960] 0.067 [0.818]
2007 0.095 [0.331] 0.074 [0.703] 0.081 [0.643]
2008 0.145 [0.108] 0.117 [0.328] 0.054 [0.994]
2009 0.097 [0.314] 0.114 [0.171] 0.094 [0.413]
2010 0.159 [0.076] 0.176 [0.043] 0.179 [0.041]

Note: Compare the p-values in the tables with 1%, 5% significance levels. If the p-value in the tables is less than the chosen level of significance, there is evidence to suggest that the two populations are not the same.

Overall, our empirical results can be interpreted as evidence supporting the similarity of the two groups under study.

7.1.3 Remarks

This section examined differences between CED-Q-supported enterprises and non-supported enterprises prior to funding. These differences were examined using enterprises over the period 2002-2010. The empirical strategy is to compare differences in employment and revenue distributions of groups of enterprises. Results can be summarized as follows.

The data suggest clearly that there is no difference of employment between CED-Q-supported enterprises and non-supported enterprises in year t-3, t-2 and t-1, where t is the reference period of the first funding received by the enterprise.

For the revenue variable, the analysis found that the only difference between the two groups is in year t-3 and t-2 prior to first funding for the CED-Q client of 2010 at 5% significance level. This finding has minor impact since it is observed only for the clients of 2010 which is the end of the period under study.

Hence, it can be concluded that overall, the empirical analysis presented in this section confirms that CED-Q-supported enterprises don’t differ from non-supported enterprises prior to first funding in terms of employment and revenue.

7.2. Regression results

This section plots the estimated coefficient β which captures the performance of CED-Q clients relative to comparison group with respect to revenue, employment and labour productivity. It is important to keep in mind that the two groups were compared in terms of revenue and employment up to three years before the CED-Q clients began receiving financing (see section 7.1). Results showed no significant difference between the two groups prior to CED-Q funding. Therefore differences in outcomes between the two groups can be more clearly attributable to CED-Q funding. The performance for each cohort between 2001 and 2010 is presented in terms of revenue, employment and labour productivity.

7.2.1 Impact on revenue, employment and labour productivity

Figure 8 shows the performance in revenue, employment and labour productivity growth for the 2001 cohort.

As can be seen, the empirical results indicate that there is no statistical difference in revenue growth between the two groups in the first year following funding. After this, CED-Q clients outperformed non-clients for three consecutive years in terms of revenue growth. Analysis showed no difference in revenue growth between the two groups in the seventh, eighth and ninth year following funding.

With respect to employment growth, the results showed that the growth of employment for CED-Q clients outperformed that of the comparison group in the year after the initial funding; then, no significantly difference between the CED-clients and the comparison group was observed for the three following years. In the fifth and sixth year, the CED-Q clients experienced negative employment growth compared to the comparison group. The rate increased to 13.7% in the ninth year.

On the labour productivity front, looking at the first year after initial funding, the impact of the program yielded no significantly difference between the study group and the comparison groupFootnote2. However, in the second year after financing, CED-Q clients experienced a 19.2% labour productivity growth compared to the non-clients. CED-Q clients’ labour productivity remained higher than that of the comparison group until the sixth year. The data showed no significantly difference between the two groups during the 2007-2009 periods. At the end of the study period, CED-Q clients recovered by showed 6.8% labour productivity growth compared to similar enterprises.

Figure 8 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2001 cohort

Figure 8 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2001 cohort

Figure 8 – Long Description

Figure 8 shows the statistical difference in the growth of revenue, employment and labour productivity between the results obtained by enterprises assisted by CED and comparable enterprises which did not receive financial assistance from CED, for 2001 cohorts. The statistical differences in percentage between these two cohorts are shown by year, from the first year following funding, that is, 2001-2002, to the last year for which data could be obtained, that is, 2009-2010.

The following are the results for the 2001 cohorts:

  • In terms of revenue growth, there was no statistical difference for the first year following funding. For the second, third and fourth years after funding, enterprises receiving assistance from CED posted growth that was 7.0%, 7.8% and 5.4% higher respectively than that of enterprises which did not receive funding. There was no statistical difference for the fifth, seventh, eighth or ninth years, but CED clients posted growth 7.5% higher than the cohort of non-assisted enterprises for the sixth year after funding.
  • In terms of employment, enterprises receiving assistance from CED posted growth that was 6.4% higher for the first year following funding. There was no statistical difference for the second, third and fourth years. For the fifth and sixth years, CED clients posted growth that was lower than that of the reference cohort by 11.0% and 7.7% respectively. There was no statistical difference for the seventh and eighth years, but CED clients experienced 13.7% higher employment growth in the ninth year.
  • In terms of labour productivity growth, there was no statistical difference during the first year following funding. For five consecutive years, from the second to the sixth year following funding, CED clients experienced labour productivity higher than that of the reference cohort by 19.2%, 12.1%, 9.5%, 5.0% and 9.0% respectively. There was no statistical difference in the seventh and eighth years following funding but CED clients once again experienced higher growth, of 6 .8%, during the ninth year.
 

Note: “N.S” stands for no significance between the two groups

Figure 9 shows results for the 2002 cohort. The 2002 cohort experienced higher revenue growth than the comparison group for two consecutive years after clients received funding. There was no difference between the groups recorded in the third year. After the third year, CED-Q clients’ started showing higher growth than their counterparts for two years. The difference in revenue growth was not significant for the remaining years.

The 2002 cohort exhibited 4.4% higher employment growth than similar enterprises in the year after funding (figure 9). From the second to the fourth year, the difference between the CED-Q funded enterprises and their counterparts was not statistically significant. During the seventh year the coefficient that captures CED-Q clients’ performance was found to be negative.

It can be seen in Figure 9 that financing enhanced the performance of the firm. CED-Q beneficiaries showed higher rate of labour productivity growth than the comparison group for four consecutive years. Then non-clients outperformed CED-Q clients during the 2006-2009. The program showed a positive impact in the eighth year.

Figure 9 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2002 cohort

Figure 9 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2002 cohort

Figure 9 – Long Description

Figure 9 shows the statistical difference in the growth of revenue, employment and labour productivity between the results obtained by enterprises assisted by CED and comparable enterprises which did not receive financial assistance from CED for the 2002 cohorts. The statistical difference in percentage between the two cohorts is shown by year, from the first year following funding, that is, 2002-2003, to the last year for which it was possible to obtain data, that is, 2009-2010.

The following are the results for the 2002 cohorts:

  • In terms of revenue, CED clients experienced growth that was 5.5% and 4.1% higher for the first and second years respectively. There was no statistical difference for the third year and revenue growth was once again higher for CED clients in the fourth and fifth years, by 4.1% and 3.8% respectively. There was no statistical difference for the sixth, seventh and eighth years.
  • In terms of employment, enterprises assisted by CED posted growth that was 4.4% higher for the first year following funding, but there was no statistical difference for the second, third and fourth years. During the fifth year, enterprises funded by CED experienced 3.5% higher growth, followed by no statistical difference during the sixth year, 2.9% lower growth during the seventh year and another lack of statistical difference during the eighth year.
  • In terms of labour productivity, CED clients experienced growth that was higher than the reference cohort during the first four years, with differences of 3.8%, 4.0%, 5.1% and 3.0% respectively. Enterprises funded by CED posted growth that was lower by 5.5% and 7.2% for the fifth and seventh years following funding, and there was no statistical difference for the sixth and eighth years.
 

Figure 10 reports the performance in revenue, employment and labour productivity growth for 2003 cohort. Enterprises that received funding for the first time in 2003 had a higher revenue growth than non CED-Q-clients for four consecutive years. There was no difference in revenue growth between CED clients and non-clients in the fifth and sixth years but were higher in the seventh year.

With respect to employment growth, CED-Q-clients showed positive sign only in two years. The coefficient had a value of 2% one year after initial funding. A negative impact observed in the sixth year. An increase of 1.8% was observed in the last period of the study.

The 2003 cohort had a higher growth in labour productivity than those in the comparison group in the first (6.2%) and second (2.6%) year following funding. Then the estimated coefficient was not significant form the third year to the sixth year. That means, there was no significant difference between the two groups for three periods. In the seventh year CED-Q client outperformed non-clients on all three growth performance measures.

Figure 10 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2003 cohort

Figure 10 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2003 cohort

Figure 10 – Long Description

Figure 10 shows the statistical difference in revenue, employment and labour productivity growth between the results of enterprises receiving assistance from CED and those of comparable enterprises not receiving financial assistance from CED, for the 2003 cohorts. The statistical difference in percentage between the two cohorts is shown by year, from the first year following the funding, that is, 2003-2004, to the last year for which it was possible to obtain data, that is, 2009-2010.

The following are the results for the 2003 cohorts:

  • In terms of revenue, CED clients experienced growth that was 4.6%, 2.8%, 3.2% and 2.6% higher from the first to the fourth year following the funding. There was no statistical difference for the fifth and sixth years, and enterprises funded by CED once again experienced higher growth, of 2.0%, for the seventh year.
  • In terms of employment growth, there was no statistical difference during the first, third, fourth and fifth years following funding. CED clients experienced 2.0% higher growth during the second year and their 1.8% higher growth during the seventh year. During the sixth year, CED clients posted 2.2% lower growth.
  • In terms of labour productivity, CED clients experienced 6.2% and 2.6% higher growth than the reference cohort for the first two years respectively. There was no statistical difference for the third, fourth, fifth and sixth years. CED client once again posted higher growth—by 3.7%—during the seventh year.
 

For the 2004 cohort, the CED-Q funding program started showing impact on revenue growth in the second year following the initial year of funding in Figure 11. CED-Q clients recorded a revenue increase of 2.4% and 3.6% in the second and third year respectively. No statistical significance difference was detected for the remaining periods.

Enterprises that used the CED-Q financing for the first time in 2004 displayed higher employment growth than similar enterprises that did not receive CED-Q funding in the third year following funding (Figure 11). During the economic downturn of 2008, non-clients performed better than CED-Q clients in terms of employment growth.

On the other hand, labour productivity of the CED-Q clients outperformed that of the comparison group in the first and second year following the initial year of funding. However, there was no difference between the two groups for the subsequent years in terms of labour productivity growth as shown in Figure 11.

Figure 11 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2004 cohort

Figure 11 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2004 cohort

Figure 11 – Long Description

Figure 11 shows the statistical difference in revenue, employment and labour productivity growth between the results of enterprises assisted by CED and those of comparable enterprises which did not receive financial assistance from CED, for the 2004 cohorts. The statistical differences in percentage between the two cohorts are shown by year, from the first year following funding, that is, 2004-2005, to the last year for which it was possible to obtain data, that is, 2009-2010.

The following are the results for the 2004 cohorts:

  • There was no statistical difference in terms of revenue growth during the first year. CED clients posted 2.4% and 3.6% higher growth than the reference cohort for the second and third years respectively. After that, there was no statistical difference for the fourth, fifth and sixth years.
  • In terms of employment growth, there was no statistical difference between the two cohorts for the first two years. CED clients posted 2.3% higher growth for the third year and 2.6% lower growth for the four and fifth years following funding. There was no statistical difference for the sixth year.
  • In terms of labour productivity, CED clients experienced 4.4% higher growth than the reference cohort during the first year and 2.8% higher growth for the second year. There was no statistical difference for the next four years.
 

For the 2005 cohort (Figure 12), the analysis shows that CED-Q enterprises did better than similar enterprises in the comparison group in the first year for all indicators of performances. Figure 12 reveals that a revenue growth of 2.6% and 3.1% was recorded in the first and second years after the year of funding. There was no significant difference between the two groups for the period 2007-2008 and 2008-2009. In the last period, CED-Q funded enterprises outperformed non-clients.

Enterprises that used the CED-Q financing for the first time in 2005 had on average 2.5% higher employment growth than the comparison group in the first year following the initial funding. CED-Q clients did worse in the fourth year, 2.5% less employment growth than the comparison group (Figure 12).

The 2005 cohort of CED-Q clients were 2.6% more productive than their counterparts in the first year after receiving funding. During the economic crisis of 2008, CED-Q clients showed a 3.2% less labour productivity growth than similar enterprises in the comparison group. A 3.5% higher labour productivity growth for CED-Q client was estimated for the fifth year (Figure 12).

Figure 12 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2005 cohort

Figure 12 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2005 cohort

Figure 12 – Long Description

Figure 12 presents the statistical difference in revenue, employment and labour productivity growth between the results obtained by enterprises assisted by CED and those of comparable enterprises which did not receive funding from CED for the 2005 cohorts. The statistical differences in percentage between the two cohorts are presented by year, from the first year following funding, that is, 2005-2006, to the last year for which it was possible to obtain data, that is, 2009-2010

The results for the 2005 cohorts are as follows:

  • In terms of revenue, CED clients experienced higher growth of 2.6%, 3.1% and 2.9% for the first, second and fifth years following funding respectively. There was no statistical difference for the third and fourth years.
  • In terms of employment growth, there were only two years in which there was a statistical difference between the two cohorts: the first and fourth years following CED funding. For the first year, enterprises which had received funding from CED experienced 2.5% higher growth, and for the fourth year, they posted 2.5% lower growth.
  • In terms of labour productivity growth, there was no statistical difference in the second and fourth years following the funding. CED clients experienced 2.6% and 3.5% higher growth than the reference cohort during the first and second years respectively, while they posted 3.2% lower growth during the third year.
 

For the 2006 cohort, the funding had a positive impact on revenue in the first year following the initial year of funding. There was no difference in revenue growth between CED clients and non-clients in the second and third years. In the fourth year, CED-Q clients had 3.3% revenue growth higher than the comparison group.

With regard to employment, the analysis shows a 2.8% higher employment growth in favour of the study group in the second year following the initial year of funding. CED-Q enterprises were outperformed by the comparison group in 2008 and then increased to 3.5% in the last year.

There was no significance difference in labour productivity growth between the two groups after the two years following the funding. In the 2008-2009 period, CED-Q clients registered a negative labour productivity growth of 2.6% (Figure 13). In 2009-2010, CED-Q clients had a labour productivity growth of 6.2% higher than the comparison group.

Figure 13 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2006 cohort

Figure 13 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2006 cohort

Figure 13 – Long Description

Figure 13 shows the statistical difference in revenue, employment and labour productivity growth between the results of enterprises assisted by CED and those of comparable enterprises which did not receive financial assistance from CED, for the 2006 cohorts. The statistical differences in percentage between the two cohorts are presented by year, from the first year following funding, that is, 2006-2007, to the last year for which it was possible to obtain data, that is, 2009-2010.

The following are the results for the 2006 cohorts:

  • In terms of revenue, CED clients experienced growth that was 3.4% and 3.3% higher for the first and fourth years following funding respectively. There was no statistical difference for years two and three.
  • In terms of employment growth, there was no statistical difference during the first year. CED clients experienced 2.8% and 3.5% higher growth during the second and four years and 0.5% lower growth during the third year.
  • In terms of labour productivity growth, there was no statistical difference for the first two years. Enterprises funded by CED experienced 2.6% lower growth for the third year and 6.2% higher growth for the fourth year.
 

For the 2007 cohort, the funding had a positive impact on revenue growth in the first year following the year of funding. CED-Q clients had revenue growth 4.2% higher than the comparison group. There was no statistically significant difference in revenue growth between the two groups in the 2008-2009 period. A 2.7% higher revenue growth was estimated for CED-Q clients in the third year, as shown in Figure 14.

The analysis shows a 3% higher employment growth in favour of the CED-Q clients for the year following the funding. There was no statistically significant difference in employment growth between the two groups for the remaining years.

In terms of labour productivity, the 2007 cohort registered a negative performance (6.3%) in the first year following the year of funding. CED-Q clients, however, outperformed non-clients in labour productivity growth by 3.8% during the 2008-2009.

Figure 14 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2007 cohort

Figure 14 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2007 cohort

Figure 14 – Long Description

Figure 14 shows the statistical difference in revenue, employment and labour productivity growth between the results obtained by enterprises assisted by CED and those of comparable enterprises which did not receive financial assistance from CED, for the 2007 cohorts. The statistical differences in percentage between the two cohorts are presented by year, from the first year following funding, that is, 2007-2008, to the last year for which it was possible to obtain data, that is, 2009-2010.

The following are the results for the 2007 cohorts:

  • In terms of revenue, CED clients experienced 4.2% and 2.7% higher growth for the first and third years following funding respectively. There was no statistical difference for the second year.
  • In terms of employment, enterprises funded by CED experienced 3.0% higher growth during the first year and there was no statistical difference for the second and third years.
  • In terms of labour productivity, CED clients experienced 6.3% lower growth in the first year, followed by 3.8% higher growth for the second year. There was no statistical difference for the third year.
 

For the 2008 cohort (Figure 15), there was no significant difference in revenue, employment and labour productivity growth between the two groups after the first year of funding. Figure 15, however, reveal that in the second year following funding a 2.3% higher revenue growth and a 4.4% higher employment growth was recorded. There was no significant difference in labour productivity growth between the two groups in the first two years.

For the 2009 cohort, the analysis found that CED-Q clients outperformed non-clients in terms of both employment and labour productivity growth on the third year (Figure 15).

Figure 15 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2008, 2009 cohorts

Figure 15 Difference in revenue, employment and labour productivity growth between CED-Q-clients and non-clients, 2008, 2009 cohorts

Figure 15 – Long Description

Figure 15 shows the statistical difference in revenue, employment and labour productivity growth between the results obtained by enterprises assisted by CED and those of comparable enterprises which did not receive financial assistance from CED, for the 2008 and 2009 cohorts. The statistical differences in percentage between the two 2008 cohorts are presented by year, from the first year following funding, that is, 2008-2009, to the last year for which it was possible to obtain data, that is, 2009-2010. For the 2009 cohorts, only the 2009-2010 data are presented.

The results for the 2008 cohorts are as follows:

  • In terms of revenue growth, there was no statistical difference for the first year and CED clients experienced 2.3% higher growth for the second year.
  • In terms of employment growth, there was no statistical difference for the first year and CED clients experienced 4.4% higher growth for the second year.
  • In terms of labour productivity, there was no statistical difference for the two years presented.

The results for the 2009 cohorts are as follows:

  • In terms of revenue growth, there was no statistical difference.
  • In terms of employment, CED clients experienced 2.9% higher growth than the reference cohort.
  • In terms of labour productivity, CED clients experienced 4.5% higher growth.
 

7.3. Survival rates

This section compares the survival rates of CED-Q clients and non-clients over the subsequent years following the reference year of funding. The survival rate was calculated by following enterprises overtime. For each cohort considered in the period of the study, if the firm did not report any employment and any revenue and did not change status (mergers/acquisitions), then this firm was considered having exited the market.

In order to generate an aggregate measure of business survival, a cumulative survival rate was calculated by aggregating cohorts for equivalent time frames, regardless of the calendar year in which the business entered the market. For instance, the business survival rate after the n-th year of entering the market was calculated by dividing the number of firms still alive after the n-th year following their start-up, by the number of start-ups at the beginning of that period, for all cohorts with a possible life span equal or greater than n.

Figure 16 shows the cumulative survival rates for all cohorts (see also Annex 4, Table 1). The estimated survival rates for CED-Q clients lies above the survival rates of non-clients throughout the entire time-period analyzed, with the exception of the first year for which the survival rate of clients (96.7%) is virtually identical to that of non client (96.9%). For instance, the percentage of CED-Q clients that were still in business five years after receiving funding (regardless of the first year in which received funding) was 85.3%; conversely, the same percentage for the comparable group of non-clients was 77.5%.

Figure 16 Cumulative survival rates, all cohorts

Figure 16 Cumulative survival rates, all cohorts

Figure 16 – Long Description

Figure 16 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for all cohorts from the first year (2001-2002) to the ninth year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 96.7%
    • Reference cohort: 96.9%
  • Year 2
    • CED clients: 94.8%
    • Reference cohort: 92.5%
  • Year 3
    • CED clients: 92.9%
    • Reference cohort: 88.1%
  • Year 4
    • CED clients: 89.3%
    • Reference cohort: 83.3%
  • Year 5
    • CED clients: 85.3%
    • Reference cohort: 77.5%
  • Year 6
    • CED clients: 79.5%
    • Reference cohort: 71.3%
  • Year 7
    • CED clients: 76.0%
    • Reference cohort: 64.0%
  • Year 8
    • CED clients: 71.3%
    • Reference cohort: 58.9%
  • Year 9
    • CED clients: 67.2%
    • Reference cohort: 50.8%
 

Figure 17 shows the survival rates for the 2001 cohort. The estimated survival rates for CED-Q clients lies above the survival rates of non-clients throughout the entire time-period analyzed. Results show that the proportion of the 2001 cohorts that were still in the business nine years after the initial year of funding was 16.4% higher than the survival rates of non-clients. Annex 5 shows the survival rates for CED-Q clients by type of program.

Figure 17 Survival rates, 2001 cohort

Figure 17 Survival rates, 2001 cohort

Figure 17 – Long Description

Figure 17 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2001 cohort from the first year (2001-2002) to the ninth year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 98.4%
    • Reference cohort: 96.9%
  • Year 2
    • CED clients: 93.8%
    • Reference cohort: 89.2%
  • Year 3
    • CED clients: 92.2%
    • Reference cohort: 81.5%
  • Year 4
    • CED clients: 89.1%
    • Reference cohort: 69.2%
  • Year 5
    • CED clients: 81.3%
    • Reference cohort: 63.1%
  • Year 6
    • CED clients: 70.3%
    • Reference cohort: 56.9%
  • Year 7
    • CED clients: 67.2%
    • Reference cohort: 53.8%
  • Year 8
    • CED clients: 67.2%
    • Reference cohort: 50.8%
  • Year 9
    • CED clients: 67.2%
    • Reference cohort: 50.8%
 

Figure 18 shows the same rates for the 2002 cohort. Overall, CED-Q clients compares favourably with that of the non-clients, despite of the length of time examined. CED-Q had higher survival rates, with nearly 73% of the CED-Q enterprises still in the business eight years after the initial year of funding (compared to almost 62% for the non-clients group).

Figure 18 Survival rates, 2002 cohort

Figure 18 Survival rates, 2002 cohort

Figure 18 – Long Description

Figure 18 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2002, from the first year (2002-2003) to the eighth year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 95.4%
    • Reference cohort: 89.1%
  • Year 2
    • CED clients: 95.4%
    • Reference cohort: 87.0%
  • Year 3
    • CED clients: 94.2%
    • Reference cohort: 86.0%
  • Year 4
    • CED clients: 91.3%
    • Reference cohort: 78.8%
  • Year 5
    • CED clients: 85.0%
    • Reference cohort: 73.6%
  • Year 6
    • CED clients: 77.5%
    • Reference cohort: 69.4%
  • Year 7
    • CED clients: 75.1%
    • Reference cohort: 64.2%
  • Year 8
    • CED clients: 72.8%
    • Reference cohort: 61.7%
 

Survival rates for the 2003 cohort are shown in Figure 19, almost 66% of non-clients survived after operating seven years in the marketplace compared close to 79% of CED-Q clients. The survival rates of the two groups are almost the same in the first and second year following the initial year of funding. After the second year enterprises that benefited from CED-Q program had higher survival rates than the comparison group.

Figure 19 Survival rates, 2003 cohort

Figure 19 Survival rates, 2003 cohort

Figure 19 – Long Description

Figure 19 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2003 cohort, from the first year (2003-2004) to the seventh year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 100%
    • Reference cohort: 100%
  • Year 2
    • CED clients: 92.9%
    • Reference cohort: 90.6%
  • Year 3
    • CED clients: 90.4%
    • Reference cohort: 82.0%
  • Year 4
    • CED clients: 85.0%
    • Reference cohort: 77.7%
  • Year 5
    • CED clients: 84.3%
    • Reference cohort: 73.0%
  • Year 6
    • CED clients: 81.8%
    • Reference cohort: 69.8%
  • Year 7
    • CED clients: 78.6%
    • Reference cohort: 66.2%
 

Figure 20 shows that the survival rates of the 2004 CED-Q clients in the first two year following the funding are slightly lower than those of the comparison group. Survival rates were similar among the CED-Q clients and non-clients in the third year. After the fifth year, non-clients start exiting the marketplace at a faster rate than CED-Q clients. At the end of the study period CED-Q clients experience a 80.4% survival rate compare to 76.5% for the comparison group.

Figure 20 Survival rates, 2004 cohort

Figure 20 Survival rates, 2004 cohort

Figure 20 – Long Description

Figure 20 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2004 cohort, from the first year (2004-2005) to the sixth year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 93.3%
    • Reference cohort: 95.4%
  • Year 2
    • CED clients: 91.4%
    • Reference cohort: 93.0%
  • Year 3
    • CED clients: 89.9%
    • Reference cohort: 89.9%
  • Year 4
    • CED clients: 85.6%
    • Reference cohort: 86.2%
  • Year 5
    • CED clients: 83.2%
    • Reference cohort: 82.6%
  • Year 6
    • CED clients: 80.4%
    • Reference cohort: 76.5%
 

As seen in Figure 21, CED-Q clients outperformed non-clients in each year subsequent to funding. The results show more CED-Q supported enterprises survived compared to those in the comparison groups after six year in business.

Figure 21 Survival rates, 2005 cohort

Figure 21 Survival rates, 2005 cohort

Figure 21 – Long Description

Figure 21 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2005 cohort, from the first year (2005-2006) to the fifth year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 99.6%
    • Reference cohort: 96.6%
  • Year 2
    • CED clients: 98.2%
    • Reference cohort: 94.0%
  • Year 3
    • CED clients: 96.0%
    • Reference cohort: 90.9%
  • Year 4
    • CED clients: 96.0%
    • Reference cohort: 87.5%
  • Year 5
    • CED clients: 91.1%
    • Reference cohort: 83.2%
 

Figure 22 shows how CED-Q clients of 2006 compares with enterprises that are in the comparison group. Approximately 95% of CED-Q clients survived through their first year compared 98% in the non-clients group. In terms of survival rate after the four year, 91.7% of the CED-Q clients were in the market the fourth year compared with 89.7% of the comparison group.

Figure 22 Survival rates, 2006 cohort

Figure 22 Survival rates, 2006 cohort

Figure 22 – Long Description

Figure 22 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2006 cohort, from the first year (2006-2007) to the fourth year (2009-2010).

Results by year are as follows:

  • Year 1
    • CED clients: 94.9%
    • Reference cohort: 97.7%
  • Year 2
    • CED clients: 94.4%
    • Reference cohort: 93.5%
  • Year 3
    • CED clients: 94.4%
    • Reference cohort: 91.1%
  • Year 4
    • CED clients: 91.7%
    • Reference cohort: 89.7%
 

Figure 23 displays the survival rates for the 2007, 2008 and 2009 cohorts. For the 2007 cohort, the percentage was high in terms of its one year survival rate. It can be seen that CED-Q clients have higher survival rate than non-clients at the end of the study period but the margin is within approximately three percentage points.

For the 2008 cohort, both the CED-Q clients and the comparison groups posted high survival rates for the year examined. However, the spread between the groups are found to be rather narrow.

For the 2009 cohort, the rate of survival for the CED-Q group was 93% after one year following funding while the survival rates was higher for the comparison group, at 99%.

Figure 23 survival rates, 2007, 2008 and 2009 cohorts

Figure 23 survival rates, 2007, 2008 and 2009 cohorts

Figure 23 – Long Description

Figure 23 presents a bar chart showing the cumulative survival rates for enterprises assisted by CED and enterprises not receiving CED assistance for the 2007, 2008 and 2009 cohorts, from the first year (2007-2008) to the third year (2009-2010).

The results by year for the 2007 cohort are as follows:

  • Year 1
    • CED clients: 100%
    • Reference cohort: 99.5%
  • Year 2
    • CED clients: 97.6%
    • Reference cohort: 95.1%
  • Year 3
    • CED clients: 94.2%
    • Reference cohort: 91.3%

The results by year for the 2008 cohort are as follows:

  • Year 1
    • CED clients: 97.9%
    • Reference cohort: 97.2%
  • Year 2
    • CED clients: 97.2%
    • Reference cohort: 96.5%

The results by year for the 2009 cohort are as follows:

  • Year 1
    • CED clients: 93.1%
    • Reference cohort: 99.0%
 

8. Conclusions

This report presents a counterfactual analysis of the economic impact of the CED-Q programs. It is aimed at studying the impact of CED-Q assistance on firm performances. The performance measures of revenue, employment and labour productivity growth were used as dependent variables in regression analysis. Using robust regression technique, coefficients were estimated for clients that benefited from programs delivered between 2001 and 2009.

Overall for the cohorts studied, CED-Q clients’ revenue, employment and labour productivity growth tended to be higher than the non-clients’ in the first two years following the funding than in the subsequent years, except for the 2007 cohort.

The difference in performance that was associated with the status of CED-Q client varied from year to year and from cohort to cohort. For instance, after controlling for other business characteristics, the 2001 cohort of CED-Q clients reported 7.8% higher revenue growth than non-clients in the third year following funding; in dollar terms this translated into an average increment of $26,000 in revenue that was associated with the CED-Q client status. This was the largest annual increment in revenue that was associated with client status across all cohorts. The smallest increment associated with client status was recorded by the 2002 cohort in their first year after funding, with 5.5% higher revenue corresponding to approximately $200 higher revenues on average.

The labour productivity growth differential also varied from year to year and from cohort to cohort; in most cases, when the difference was statistically significant, the differential was in favour of CED-Q clients. The largest differential was recorded for the 2001 cohort, two years after funding, when CED-Q clients had a 19.2% higher labour productivity, translating to approximately $2,100 higher sales per employee, on average; the smallest differential in favour of CED-Q client was recorded by the 2005 cohort five years after funding, with a higher growth of 3.5% translating into approximately $10 higher sales per worker, on average.

For the employment growth differentials, the pattern of association for client and non-client status was less clear than that seen for revenue and labour productivity. The largest differences in favour of the CED-Q client group translated into an increment of 0.2 jobs per client, on average, which was reported by three different cohorts in three different reference years. The largest difference in favour of non-clients was reported for the 2001 cohort five years after funding, translating into 0.4 more jobs for non-clients.

Overall, CED-Q clients experienced higher survival rates than non-clients at the end of the study periods for each cohort, with the exception being the 2009 cohort, which had a lower survival rate than the non-clients. This can be also observed for cumulate survival rates, which were obtained by combining all cohorts. CED-Q clients had higher survival rates than non-clients throughout the entire time period that was analyzed, with the exception of the first year for which the survival rate of clients and non-client were virtually identical. For instance, five years after funding the cumulate survival rate of CED-Q clients was 85.3% compared to 77.5% for the non-clients group.

References

Abrell, J. Anta NDoye Faye and Georg Zachmann (2011). Assessing the impact of the EU ETS using level data. Bruegel working paper.

Ankarhem, M. et al. (2010). Do Regional Investment Grants Improve Firm Performance? Evidence from Sweden. Technology and Investment, 2010, 1:221-227.

Bellone, Flora., Sarah Guillou and Lionel Nesta (2010). To what extent innovation accounts for firm export premia? University of Nice – Sophia Antipolis, mimeo, February 2010.

Bradshaw, T. (2002). The Contribution of Small Business Loan Guarantees to Economic Development, Economic Development Quarterly, vol 16 No 4:360-369.

Chandler, V. (2010). The economic impact of the Canada Small Business Financing Program. Industry Canada working paper series.

Delgado, Miguel A., Jose C. Farinas and Sonia Ruano (2002). Firm productivity and export markets: a non-parametric approach. Journal of International Economics 57:397-422.

Girma, S. et al. (2007). The effect of government grants on plant level productivity. Economics Letters 94: 439-444.

Helmut, Fryges and Joachim Wagner (2010). Exports and Profitability: First Evidence for German Manufacturing Firms, The World Economy, 33:399-423.

Holm-Pedersen, M., Millard, J., Pedersen, K. (2009). Ex Post Evaluation of Cohesion Policy Programmes 2000-2006 co-financed by the ERDF (Objective 1 and 2), Danish Technological Institute.

Huber, P. J. (1981). Robust Statistics. New York: Wiley.

Lopez Acevedo, Gladys and Hong W. Tan (2010). Impact evaluation of SME program in latin America and Caribbean’ World Bank working paper.

Mouqué, Daniel (2012). What are counterfactual impact evaluations teaching us about enterprise and innovation support? European Commission.

Yohai, V. J. (1987). High breakdown-point and high efficiency robust estimates for regression. Annals of Statistics 15:642–656.

Footnotes

Footnote 1

It is important to note that RDCI match rates are different from the others -not all businesses perform R&D.

Return to footnote 1 referrer

Footnote 2

As previously noted labour productivity was proxied by sales/employment

Return to footnote 2 referrer

Date modified: