Presentation by Neha Prashar to Scottish Enterprise Roundtable events 28.10.2020

1. Fast-Growth SMEs in the Scottish
Economy:
Developing an Econometric Model of Growth Episodes
Prof Mark Hart, Dr Neha Prashar, Aston University,
Dr Karen Bonner, Ulster University Economic Policy Centre

2. Background
• Presentation builds on a report to Scottish Enterprise (SE) in January 2020 – “Analysis and Benchmarking of
Business High-Growth Performance in Scotland”
• SE commissioned analysis to explore the high-growth dynamics of Scotland’s business base, benchmarking this
against other regions of the UK.
• The objective was to focus on the level to which Scottish businesses achieve and then sustain levels of high-
growth, ultimately reaching the definition of becoming a High-Growth Firm (HGF) as defined by the OECD, and
how this compares with elsewhere.
• The analysis also presents a critique of the definition of a HGF and sets out some alternatives definitions and
methodologies to allow policymakers to more accurately focus on business growth over time.
• Data – ONS Business Structure Database (BSD)

3. Key Takeaways
• Step away from growth rates as the central concern towards ‘growth trajectories’ -
our shorthand term for the dynamics of job creation over a firm’s life.
• This better captures the interplay between growth and survival.
• It provides a different approach to measuring the contribution of rapidly growing
firms to job creation and economic growth.

4. High-Growth Episodes – a cohort
perspective
• The OECD HGF definitions and its variants are less than optimum and we urge SE not to base a scale-
up strategy upon this sub-optimal metric.
• A HGE is defined as firms with 10+ employees experiencing at least 20% or 10% growth in
employment the following year. We compute the annual growth rate for each year (2010-11 to 2017-
18) and place each year into three categories:
– ‘High-growth’: ten or more employees and growth of 20% or 10% in that year.
– Alive but not high-growth: firms that don’t meet the ‘high-growth’ threshold, which are growing more slowly, not growing
at all or declining in that year and also firms that did grow by 20% or 10% but have fewer than 10 employees.
– Not active or no employment: when a firm has no employment or disappears from the database. This could be for a
number of reasons that don’t necessarily relate to the death or closure of a firm, such as, the firm could have been
acquired and still be operating under another legal entity.

5. High-Growth Episodes - Scotland
• Analysis of a cohort
of start-ups in 2010
placed the concept
of ‘high-growth’
within the life cycle
of the business -
‘high-growth’ can
occur at any stage.
• Weakness of the
arbitrary OECD
definition by
rendering invisible
many firms as ‘high-
growth’ because
their rapid growth
was not consistent
year on year and
took place in
discrete one or two-
year episodes over
the decade.
Scottish Start-ups in 2010: timing of ‘high-growth’
events (20% emp. def) 2010-18

6. A tale of two ‘high-growth’ metrics
• SE needs to adopt a more nuanced view about business growth and high-growth in particular.
• Reliance upon a single definition (i.e., the OCED HGF) is less than optimal and as we have shown
renders invisible much of the growth and indeed high-growth we observe in businesses across the
Scottish economy.
• OECD HGF definition? – arbitrary and leads to misdirection of policy - it is well past its sell-by date
• High-growth episodes? – reflecting the reality of growth for those firms that do grow.
• Research into the triggers of these episodes and to examine the role of the various interventions
associated with the Account Managed system in Scotland would be an invaluable next steps project.

7. Modelling High-Growth Episodes in Scotland
2010-18
• Compelling evidence from the descriptive analysis of Scotland having a lower
proportion of HGFs as defined by the OECDs.
• But, our econometric analysis indicates that when we control for the nature of the
business population (size, age, sector, prior growth), together with environmental
variables such as education, ethnicity and other macro variables such as growth and
new venture formation.
• …….. then the case for a ‘high-growth deficit’ in Scotland is severely weakened.

8. Source: Business Structure Database 2010-2018 (BSD)
6
8
10
12
14
16
2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18
HGErate(%)(20%over1year)
North East North West Wales Scotland UK UK (Excl London)
High Growth (20%pa) (1 year periods)

9. What have we done so far?
𝑮𝑮𝑮𝑮𝑮𝑮𝑮𝑮𝑮𝑮𝑮𝑮 𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝑴𝒄𝒄𝒊𝒊𝒊𝒊 = α𝒊𝒊𝒊𝒊 + β𝒊𝒊𝒊𝒊 𝑿𝑿𝒊𝒊𝒊𝒊 + γ𝒊𝒊 𝒕𝒕−𝟏𝟏 𝒁𝒁𝒊𝒊 𝒕𝒕−𝟏𝟏 + δ𝒊𝒊 + θ𝒕𝒕 + ε𝒊𝒊𝒊𝒊
• HGE – High growth episode (10% growth in employment over 1 year for firms with more than or equal to 10 employees)
• HGE_20 – High growth episode (20% growth in employment over 1 year for firms with more than or equal to 10 employees)
• sHGE – small high growth episode (growth in employment by at least 8 employees for firms with less than 10 employees)
• allHGE - High growth episode (10% growth in employment over 1 year for firms with more than or equal to 10 employees)
and sHGE (growth in employment by at least 8 employees for firms with less than 10 employees) together.
• allHGE_20 - High growth episode (20% growth in employment over 1 year for firms with more than or equal to 10
employees) and sHGE (growth in employment by at least 8 employees for firms with less than 10 employees) together.
Data is combined from the Business Structure Database (BSD), Annual Population Survey (APS) and the Global Entrepreneurship
Monitor (GEM) for 2010-2018.
Logistic regression analysis with random effects is estimated using the panel data. The data was restricted to those born before
2007 and still alive in 2018. This was done due to memory restrictions in the Secure Lab – something we are working on to
modify during the course of this research.

10. 𝑿𝑿𝒊𝒊𝒊𝒊 includes
• Age = Year – Birth of a firm
• Sector = 1 digit SIC code classifications (not including Agriculture, mining and public sectors)
• Average TEA = Total early stage entrepreneurial activity (GOR level for each year). This is taken from the Global
Entrepreneurship Monitor UK.
• UKborn = percentage of working aged population (16-64) who are non –UK born (NUTS2 level for each year)
• Ethnic = percentage of working aged population (16-64) who are ethnic minority (NUTS2 level for each year)
• NVQ4+ = percentage of working aged population (16-64) who have qualification of NVQ4 or over
• Netemployment = calculated from job creation and destruction estimates as the number of jobs resulting from firm births +
firm expansions – (firm deaths + firm contractions) at NUTS3 level for each year
• GOR = GOR level dummies
• Year = Year level dummies
𝒁𝒁𝒊𝒊 𝒕𝒕−𝟏𝟏 includes
• Size = size in terms of number of employees
• Empgr = previous employment growth over one year (ie, for 2010, it would be the employment growth between 2008 and
2009)
• Turngr = previous turnover growth (similar to empgr)
δ𝒊𝒊 represents the regional dummies at GOR level and θ𝒕𝒕 are time dummies. Results presented are the odds ratios.

11. Model 1 Model 2 Model 3 Model 4 Model 5
Employment growth (t-1) 1.204*** 1.203*** 1.203*** 1.203***
(0.0105) (0.0105) (0.0106) (0.0106)
Turnover growth (t-1) 1.274*** 1.274*** 1.273*** 1.273***
(0.00845) (0.00845) (0.00849) (0.00849)
Average TEA 1.657** 1.650** 1.690**
(0.335) (0.337) (0.346)
% of working pop with NVQ4+ 1.002** 1.002**
(0.00114) (0.00114)
% of working pop that are ethnic minorities 0.999 0.999
(0.00166) (0.00166)
% of working pop that are Non-UK Born 1.005* 1.004*
(0.00258) (0.00258)
netemployment 1.004***
(0.000768)
Base = North East
North West 1.080*** 1.077*** 1.071*** 1.052** 1.050**
(0.0239) (0.0235) (0.0235) (0.0237) (0.0236)
Yorkshire and the Humber 1.042* 1.041* 1.025 1.006 1.005
(0.0239) (0.0235) (0.0239) (0.0239) (0.0238)
East Midlands 1.069*** 1.066*** 1.064*** 1.034 1.034
(0.0248) (0.0243) (0.0243) (0.0251) (0.0251)
West Midlands 1.040* 1.037 1.031 1.007 1.005
(0.0236) (0.0231) (0.0231) (0.0239) (0.0238)
East of England 1.084*** 1.078*** 1.062*** 1.020 1.019
(0.0242) (0.0236) (0.0241) (0.0255) (0.0255)
London 1.190*** 1.174*** 1.152*** 0.954 0.951
(0.0255) (0.0248) (0.0258) (0.0429) (0.0428)
South East 1.084*** 1.079*** 1.064*** 1.001 1.001
(0.0232) (0.0227) (0.0232) (0.0245) (0.0246)
South West 1.083*** 1.078*** 1.063*** 1.026 1.023
(0.0244) (0.0239) (0.0244) (0.0246) (0.0245)
Wales 0.883*** 0.882*** 0.879*** 0.871*** 0.870***
(0.0233) (0.0229) (0.0228) (0.0229) (0.0229)
Scotland 1.049** 1.049** 1.047** 1.004 1.004
(0.0243) (0.0238) (0.0238) (0.0248) (0.0248)
Northern Ireland 1.321*** 1.318*** 1.316*** 1.299*** 1.306***
(0.0353) (0.0347) (0.0346) (0.0360) (0.0362)
Constant 0.623*** 0.553*** 0.541*** 0.501*** 0.506***
(0.0157) (0.0138) (0.0144) (0.0179) (0.0181)
Observations 1,108,837 1,108,837 1,108,837 1,095,858 1,095,858
Number of ID 161,318 161,318 161,318 160,657 160,657
SE form in parentheses
*** p<0.01, ** p<0.05, * p<0.1
• HGE – High growth episode
(10% growth in employment
over 1 year for firms with
more than or equal to 10
employees)
• Age, size, sector and average
TEA play an important role
in determining HGE
• Previous growth is
important in determining
future high growth.
• When adding
regional/national controls
for skills and dynamism, this
changes the importance of
regional/national
differences – no ‘Scottish’
effect

12. Next Steps? – missing variables….
Access to Finance
• Link current data (FAME)
Business Support
• Link to SE Account Managed Firms and other business support programmes if data is
available
HMRC Data
• Trade data - Import/Export data
• This one is more tricky as access to the HMRC lab is difficult and takes time to set up.
Growth Ambition or Mindset
• TEA and Established Businesses ambition to grow (expects growth of 10 or more
employees and 50% employee growth) (GEM) - % of businesses in a local economic
area
• This definition can be tweaked.
Any others…data permitted….

13. Other things…
Look at different time periods
• Focused on 1-year time periods.
• Can look at 2, 3 and 5 year time periods.
Look at different base year
• Focused on all firms born before 2007.
• Can look at different years and even a specific cohort (i.e., firms born in one particular year).
Omit London from the analysis
• Can re-run the model with London omitted for a better comparison of Scotland with the other regions.
Use a different measure of “high-growth”
• Employment growth
• Turnover growth
• Productivity growth
• Productivity Heroes (those that experience positive employment growth and positive turnover growth
where turnover>employment)
Is the model the same for the UK and Scotland? Should our analysis focus on comparison of
Scotland with the North of England and/or other home nations?

14. Thank you
For further details please visit :
www.enterpriseresearch.ac.uk
@ERC_Uk
n.prashar14@aston.ac.uk
@npra31 ERC Funded by