- Start date: 1 January 2021
- End date: 31 August 2022
- Principal investigator: Andy Charlwood
- Co-investigators: Nicholas Wilson; Krsto Pandza; Eric Atwell (School of Computing)
To investigate how innovative machine learning methods can be used to understand the roles of innovation, digitalisation and strategic framing, and human capital, in the way that companies elucidate their strategy and how this relates to company financial performance in the population of UK listed companies.
Since 2013, UK listed companies (population circa 3,200) have been required to provide narrative strategy reports as part of their statutory annual reports to shareholders. The amount of data these reports have generated on UK listed companies (circa 3,200 companies with 6 years of reports) is beyond the capacity of traditional qualitative data analysis methods to sensibly analyse, yet such reports are likely to contain insights into aspects of company strategy held as critical for productivity, for example innovation, digitalisation and investments in human capital. Even if reports do not contain such information, this absence may itself be revealing. Recent advances in machine learning offer the possibility of using sequence to sequence mapping (seq2seq) to analyse these reports. If we can establish the viability of using machine learning to identify key features of strategic reports, we can then go on to use other forms of machine learning (e.g. random and survival forests) to investigate empirically the key features of strategy reports that predict financial performance and productivity in the population of UK listed companies.