Small Business and Special Surveys Division Statistics Canada

1
Small Business and Special Surveys
DivisionStatistics Canada

• Entrepreneurship Indicators Project
• Steering Group Meeting
• Paris Nov 19-20, 2007

2
EIP Sensitivity Analysis Objectives

• Have countries test various strategies for
  determining High Growth (HG) rates
• Specifically look at how the HG rate is affected
  by
• Period of observation 3 year vs. 5 year
• Varying growth rate per year ()
• Business Size Class
• Business Sector
• Age of the business (Gazelles)

3
Canadian Sensitivity Testing

• Reference data 2001 to 2006
• Notes
• Employment 12 month average head count as
  reported to CRA via payroll remittance data
• Agriculture NAICS 11 and Government NAICS 91 are
  NOT included in any outputs
• Acquisition, mergers take-overs not considered
• Only businesses active prior to 2001 with 5
  -249 employees in 2001 were in scope

4
Evaluation Variables

• Business Size Classes 5-9 10-19 20-49 50-99
  100-249 employees
• Industrial Sectors
• Construction Wholesale Retail Prof. Scientific
  Tech Services Other All
• Window of observation
• 3 yr (2001 - 2004) 5 yr (2001 2006)
• Age of business (Gazelles lt 6 yrs old)
• Growth rate / yr 10 12 14 16 18 20

5
Disclaimer
Data is not for publication and should not be
cited
6
Findings-Observation Period-

• The 5 year measurement resulted in 50 fewer HG
  firms overall using 20 growth/yr
• Distribution by size class varied little for 3
  5 yr HG rates
• Size 3yr 5yr
• 5-9 8,948 6 4,006 3
• 10-19 4,354 5 2,167 3
• 20-49 2,488 5 1,176 2
• 50-99 619 4 309 2
• 100-249 227 3 119 2
• 5-249 16,636 5 7,777 2

7
Findings-Observation Period-

• Hypothesis
• A HG rate is NOT sustainable for firms (They go
  through a HG period but then plateau until the
  next HG period)
• Recommendation
• HG should be measured within a 3 yr window
• Given that HG rate varies greatly by size class
  (6 HG rate for size class 5-9 and 3 for size
  class 100-249) it is essential to produce HG
  rates by size class
• Using an overall HG rate will distort
  international comparisons as the proportion of
  large firms in countries varies greatly

8
Findings-High Growth (HG) Definition-

• 20 growth per year is high
• 19 of businesses 5-249 Emp did not survive
• Only 5 (16,636) businesses were HG
• If we want to examine HG businesses by size and
  industry or size and geography, data will be very
  sparse and confidentiality may become an issue
• Especially large sized firms (100-249 Emp) in
  Const. 11 HG, Whole. 11 HG, Retail 13 HG

9
Findings-High Growth (HG) Definition-

• Data suggests that there is no natural breakpoint
  in HG definition the number of businesses who
  qualify decreases progressively as the growth /
  yr raises
• Growth/yr HG
• 10 15
• 12 11
• 14 9
• 16 8
• 18 6
• 20 5

10
Findings-High Growth (HG) Definition-

• Hypothesis
• There is no right or wrong measure of HG, only a
  measure that we can all agree to that yields
  useful and meaningful results for international
  comparison
• Therefore we want a measure that clearly
  separates businesses that are growing
  significantly more than the average but that
  yields sufficient observations to be useful in
  demographic analysis examining business size,
  industry, geography etc.
• Recommendation
• Pick a HG rate somewhere between 15 20 per
  year over a 3 year period

11
Findings-(HG) Revenue vs. Employment-

• Results
• HG measured by Revenue results in twice as many
  HG enterprises as Employment (33,653 vs. 16,636)
  using 20 growth/yr
• The difference in the 2 measures becomes more
  pronounced as the growth / yr increases
• 10 growth/yr (firms 5-249) Emp 15 Rev 21
• 16 growth/yr (firms 5-249) Emp 8 Rev 14
• 20 growth/yr (firms 5-249) Emp 5 Rev 11

12
Findings-(HG) Revenue vs. Employment-

• Summary
• Data suggests that using Emp or Rev changes the
  of HG firms but not demographic trends
• We need to decide whether having 2 measures
  of HG will enlighten users or confuse users

13
Findings-(HG) Sector Analysis-

• The proportion of HG by Sector is similar whether
  Emp or Rev is used to calculate HG
• Rev 33,653 HG firms
• Const 14 Manuf 10 Whole 8
• Ret 12 Prof 8 Oth 47
• Emp 16,636 HG firms
• Const 11 Manuf 9 Whole 7 Ret 17
  Prof 7 Oth 48
• Differences are confined to Const and Retail

14
Findings-(HG) Sector Analysis-

• Based on Emp HG
• A lower concentration of HG small firms 5-49 is
  found in Manuf 91 vs. norm of 95-96
• A marginally higher concentration of HG small
  firms 5-49 is found in Const 97 vs. norm of
  95-96
• Results were very similar when using Rev rather
  than Emp

15
Findings-(HG) Sector Analysis-

• Summary
• Sector analysis is important
• For many sectors it will not be possible to
  drill down too deeply due to confidentiality but
  probably not essential for international
  comparisons
• Detail sector analysis for large firms 50-249 is
  impossible due to too few observations

16
Findings-(HG) Gazelles-

• Tracking firms lt 6 yrs old
• They represent 1/3 (105,141) firms 5-249 Emp
• 72 of these firms survived vs. 81 overall
• Although they represent 1/3 of firms they
  represent 41 of HG firms (6,838)
• Using 20 growth/yr over 3 yrs
• The proportion of HG Gazelles decreases as bus
  size class increases
• Size class HG Gazelles Prop. Gazelles
• 5-9 8948 4124 46
• 10-19 4354 1676 38
• 20-49 2488 817 33
• 50-99 619 171 28
• 100-249 227 50 22

17
Findings(HG) Gazelles

• A higher concentration of small Gazelles 5-49 is
  found in Const 99 vs. norm of 96
• Severe disclosure issues arise when looking at
  Gazelles by size and industry even if one were to
  collapse medium sized firms together 50-249