4.5 Description of variables used in analyses of spinout formation
4.5.3 Control variables
This group of variables defines the university (university size, disclosures, and science orientation), and broader network environment (Δ regional GVA per capita and 2002 regional GVA per capita).
University size
To control for the size effect a sum of total university income over 2002/3-2013/14 is used. The data is produced by HESA annually. It is an imperfect measure among the variables utilised, as employment is more commonly used; however, employment data was not available for the year 2002/3, in effect reducing the sample by 101 observations. The variable is expressed in billions of pound sterling.
109 Disclosures
This variable captures university-generated knowledge of commercial potential as published in HEBCIS. It is constructed as a sum of annual disclosures between 2002/3 and 2013/14 and is used in two forms: untransformed in bivariate statistical tests, and expressed in natural logarithms to correct positive skew for regression analysis. In case of multiple university parents, the disclosures of all institutions are expressed as a sum. A disclosure is knowledge identified by a university TTO to have commercial potential (or recognised by faculty as having commercial potential and brought to TTO's attention, as in Bercovitz and Feldman 2008) and has been used in similar studies of spinout company formation (Di Gregorio and Shane 2003;
Hayter 2013), or academics’ entrepreneurial intentions (Fini et al. 2009; Goktepe-Hulten and Mahagaonkar 2010). It is important to note that not all commercialisable knowledge is captured by disclosures, as academics might not be inclined to disclose any invention at all (Chrisman et al. 1995). Another limitation to note is the imperfect nature of this measure of commercialisable knowledge, as some universities might require disclosures to actually be patentable inventions (Di Gregorio and Shane 2003) disregarding other inventions where monopoly market power could not be obtained.
Science orientation
University high-end research capability, which expresses university standing (Di Gregorio and Shane 2003; O’Shea et al. 2005; Powers and McDougall 2005; Lawton Smith et al. 2008), is used here to control for a bias towards science-oriented disciplines being more prone to knowledge commercialisation through spinout companies (Shane, 2004b). Shane (2004b) devoted a whole chapter in his book on spinout companies to discuss the industry or discipline bias predominantly towards spinouts from biomedical sciences, software, electronics, and industrial machinery. The key reason for the bias is in the numbers of spinout companies originating from these university departments. Shane (2004b) explains that the importance of such academic fields corresponds to a number of characteristics of such industries related to
110 greater levels of IP protection, lower levels of complementarity of assets, an earlier stage of industry development, or smaller average firm size in the industry.
Furthermore, such bias is present in the literature itself (never acknowledged or discussed), with numerous studies depicting either traditional science-based universities (128 companies from MIT were studied by Nerkar and Shane (2003); Delft University of Technology depicted in van Geenhuizen and Soetanto (2009)), science departments (researchers from Natural Sciences and Engineering Research Council of Canada studied in Landry et al. (2006); nanotechnology researchers studied at Chalmers University of Technology in Fogelberg and Lundqvist (2013)), or science spinouts' sectors (Lawton Smith and Ho (2006) profiled Oxfordshire's spinout companies to be primarily in biotechnology, IT, pharmaceuticals, measuring instruments, and optoelectronics sectors; out of 149 spinouts studied in Walter et al. (2006) 69% were in technical services and technical manufacturing sectors). There are exceptions: for example, medicine and science and engineering schools were found to be more supportive of entrepreneurial university paradigm compared to arts and humanities, and business and law schools (Philpott et al. 2011), but the bias is clearly evident in the results of such non-discriminate studies.
Based on the Research Assessment Exercise (RAE) 2008 and the Research Excellence Framework (REF) 2014, a simple measure is composed that reflects university’s focus on high-quality scientific research. RAE 2008 measured outputs from 2001 to 2007 (www.rae.ac.uk), whilst REF 2014 covered outputs produced from 2008 to 2013 (www.ref.ac.uk), and so both assessments are well aligned with the sample’s timeframe of 2002-2013. The variable first measures research quality and quantity derived either from RAE 2008 or REF 2014, where research outputs receive quality ratings from 1*-4* (unrated output is considered as well): 4*
outputs are deemed world-leading. The REF (the successor to the RAE) was used to measure research quality in D’Este et al.’s (2012) study; however, they focused on academic journal
111 submissions and used citations data as a measure of research quality (citations are also used in O’Shea et al. 2005; Powers and McDougall 2005).
The metric used in this study focuses on the traditional science fields that typically generate more spinout companies, due to their more tangible IP protection method of patenting (Shane 2004b), and therefore the focus is on REF’s (and respective fields in RAE 2008) Panels A (clinical medicine; public health, health services and primary care; allied health professions, dentistry, nursing and pharmacy; psychology, psychiatry and neuroscience; biological sciences;
agriculture, veterinary and food science) and B (earth systems and environmental sciences;
chemistry; physics; mathematical sciences; computer science and informatics; aeronautical, mechanical, chemical and manufacturing engineering; electrical and electronic engineering, metallurgy and materials; civil and construction engineering, general engineering). The approach taken here is to focus on the overall quality profile, which incorporates output, impact and environmental assessments of submissions. The end result simply transforms proportion-expressed RAE and REF scores (in percentage terms) into total submission numbers for every ith university (separately for RAE and REF):
𝐻𝐸𝑅𝐶𝑀𝑖 = ∑ 𝑅𝑂𝑖𝑗× 𝑁𝑂𝑆𝑖𝑗
𝑗
where:
HERCM – high-end research critical mass (i.e. submission numbers of 4* quality) RO – sum of percentage of research outputs with overall quality score of 4*
NOS – number of submissions i – notation for university
j – notation for research field, Panels A and B only.
In order to transform this measure into a more comparable variable it is simply divided by the total 4* research submission numbers in all panels (A-D):
112 𝑈𝑁𝐼%𝑆𝐶𝐼𝐸 = 𝐻𝐸𝑅𝐶𝑀𝐴𝐵
𝐻𝐸𝑅𝐶𝑀𝐴𝐵𝐶𝐷 where:
UNI%SCIE – university science orientation (expressed as a proportion of all research fields) HERCMAB – high-end research critical mass in Panels A and B (i.e. submission numbers of 4*
quality)
HERCMABCD – high-end research critical mass in Panels A-D (i.e. submission numbers of 4*
quality)
The variable composed here uses an average of UNI%SCIE measure for RAE 2008 and REF 2014. As a result, the variable covers a timeline from 2001 to 2013.
Change in regional GVA per capita and 2002 regional GVA per capita
In order to control for regional economic development, two simple weighted measures of GVA (Gross Value Added) per capita are used: 1) capturing change from 2002 to 2014, and 2) controlling for the value of economic activities in 2002. Clearly, a region with greater economic output translates into a larger market, an industrial structure of the region with high-value (typically high technology) activities, whilst level of growth of the region's output indicates spinouts’ performance potential in that location. Thus far, only exemplary regions were studied in terms of their spinout activity performance, such as Oxfordshire in Lawton Smith and Ho (2006); or were only used as dummy controls for such performance, such as California and Massachusetts in Toole and Czarnitzki (2007). The data for the variable was obtained from Office for National Statistics and is based on income-based regional GVA per capita (at current prices) measure.
Next, the variables used in survival analyses of university spinout companies are described.
113 4.6 Description of variables used in analyses of spinout survival
This section outlines the constructs of variables used in the analyses of spinout company survival. These variables undergo both bivariate statistical and regression examinations. The unit of analysis employed to explain spinout survival is the spinout company.
4.6.1 Dependent variable