[144 words]Governments around the world want to develop their ICT industries. Researchers and policymakers thus need a clear picture of digital businesses, but conventional datasets and typologies tend to lag real-world change. We use innovative 'big data' resources to perform an alternative analysis for all active companies in the UK, focusing on ICT-producing firms. Exploiting a combination of observed and modelled variables, we develop a novel 'sector-product' approach and use text mining to provide further detail on key sector-product cells. We find that the ICT production space is around 42% larger than SIC-based estimates, with around 70,000 more companies. We also find ICT employment shares over double the conventional estimates, although this result is more speculative. Our findings are robust to various scope, selection and sample construction challenges. We use our experiences to reflect on the broader pros and cons of frontier data use.
JEL
1/ IntroductionThis paper uses novel 'big data' sources to expand our understanding of digital businesses in the UK. We produce alternative counts of ICT-producing firms and set out key descriptive characteristics. We then draw on this experience to critically reflect on some of the opportunities and challenges presented by big data tools and analytics for economic research and policymaking.Information and Communications Technologies (ICTs) -and the 'digital economy' they support -are of enduring interest to researchers and policymakers. Digital sectors and firms are the subject of much analysis both at the organisational level (Bloom et al., 2012;Bresnahan et al., 2002) and in the growth field. Human capital and innovation shape long term economic development (Lucas, 1988; Romer, 1990); high value-added sectors such as ICT make direct contributions to national growth, as well as indirect contributions through spillovers and supply chains (Audretsch and Feldman, 1996;Moretti, 2012).National and local government are thus keen to exploit the growth potential of digital businesses. Given the recent resurgence of interest in industrial policy across many developed countries (Aghion et al., 2013;Aiginger, 2007;Block and Keller, 2011;Harrison and Rodríguez-Clare, 2009;Mazzucato, 2011; Rodrik, 2004), there is now substantial policy interest in developing stronger, more 'competitive' digital economies. For example, the UK's new industrial strategy agenda (Cable, 2012) Real-world features of an industry tend to evolve ahead of any given industrial typology. For researchers, these data challenges present particular barriers to understanding the extent and nature of ICT production, where the pace of change can be very rapid. Data coverage is often imperfect, industry typologies can lack detail, and product categories do not closely align with sector categories. For policymakers, these information gaps feed through into policy gaps, which can limit the ability to design effective interventions.
3To tackle these issues we use an innovative commercial dataset develope...