Abstract
Increased non-financial disclosure has made it essential for researchers to develop new ways of measuring the quality of non-financial disclosure. This thesis aimed to develop new instruments to measure non-financial information disclosed in integrated reports, incorporate the measures into the Ohlson (1995) Model, and evaluate the value relevance. The objective was to evaluate two types of non-financial information: business model disclosure quality and the scope of the fourth industrial revolution (4IR) disclosure, the latter focusing on projects, technologies, and stature disclosure. Firstly, an innovative instrument was designed and developed to measure the quality of business model disclosure in integrated reports. A multi-dimensional disclosure index to measure the quality of business model disclosure was developed and implemented into a new text analysis software named the Business Model Analysis Tool (BMAT). Herein, the content analysis process was facilitated by algorithms based on natural language processing techniques in BMAT and removed the tedium of manual content analysis. As a result, an increase in business model quantity, dispersion, coverage, and depth for 370 integrated reports over five years was noted. Next, a new instrument was designed to measure the scope of 4IR disclosure in integrated reports. A disclosure index, comprising a 4IR dictionary, was developed and incorporated into a new text analysis software named the 4IR Disclosure Analysis Tool (4IRDAT). The use of algorithms based on natural language processing techniques in 4IRDAT removed the tedium of manual content analysis. It was established that there was an increase in the disclosure of 4IR technologies for a sample of 285 integrated reports over three years. In addition, the thesis assessed whether the quality of business model disclosure influences share price. The scores obtained from BMAT were utilised to evaluate the value relevance of business model disclosure. Two scenarios were evaluated: including loss-making companies and excluding loss-making companies. The sample size for the value relevance of business model disclosure quality, including loss-making companies and excluding loss-making companies, was 350 and 260, respectively, for five years. The results for both scenarios indicated that the quality of business model disclosure had yet to be incorporated in the share price of South African listed companies. Finally, due to the emergence of 4IR and its effect on existing business models, we evaluated the effect of 4IR disclosure on the share price. The scores obtained from 4IRDAT were utilised to
v
evaluate the value relevance of the scope of 4IR disclosure. Two scenarios were evaluated: loss-making companies and the other excluding loss-making companies. The sample size for the value relevance of the scope of 4IR disclosure, including loss-making companies and excluding loss-making companies, was 270 and 216, respectively, for three years. The results indicated that 4IR disclosure had yet to be incorporated in the share price of South African listed companies for both scenarios. This thesis is indispensable to regulators, practitioners, standard setters, and academics because it highlights the design and development of multi-dimensional quality measures, which can be amplified using algorithms based on natural language processing techniques. It provides a platform for future research to explore the use of artificial intelligence tools in developing software to enable content analysis of financial and non-financial reports, enabling deeper analysis of corporate reporting. It also provides insights into the disruptive effects of 4IR technologies on business models and is an area yet to be interrogated in a South African context.
Keywords - Integrated reporting, Business model disclosure, Non-financial reporting, Multi-dimensional disclosure index, Content analysis, Value relevance, Ohlson (1995) Model, 4IR, Algorithms, Natural language processing techniques