Do board-level controls matter? – An agency perspective on socially responsible investment (SRI) company boards in South Africa

Abstract

If board-level controls matter, the introduction of the 2008 Companies Act with its enhanced legislative requirements, should have a positive impact on firm performance. To assess board-level controls this study developed two unique control indexes to assess the boards of 84 companies over three years. The study focuses on companies on the SRI index as they have a greater focus on sustainability and transparent disclosure of board-level controls including separation of duties, monitoring, goal-aligned remuneration and oversight. The first index uses 23 board-level control indicators (CI) and the second 19 board-level direction indicators (DI). The two indexes were assessed using fixed effects estimation methods against current and negatively lagged firm performance proxies. Results show that board-level controls matter as both indexes were positively related to return on assets (RoA), however, only DI was highly significant. Both indexes changed to a negative relationship to next year’s return on assets (NYRoA), again highly significant for DI. The change to a negative relationship suggests a timing and information asymmetry problem. CI was positively related to the natural log of enterprise value per share (LEV) with a low level of significance while the positive relation continues to the next year’s natural log of enterprise value per share (LNYEV) that was highly significant. The latter suggests that the controlling role of the board are continued to be valued by the market.

Keywords:

• board-level controls
• agency theory
• directors
• directing
• internal control

Disclosure statement

This article is derived from the following PhD’s thesis: Steyn, B. (2018). The board of directors as a governance mechanism in South Africa: an agency theory perspective (PhD’s thesis, University of KwaZulu-Natal) South African link:

https://researchspace.ukzn.ac.za/handle/10413/16478

Appendix 2 LIST OF SRI COMPANIES

Table

Notes

1 A negative lag is a difference in time to the future, for example where return on assets for the current year represent t=0, negatively lagged return on assets represent t+1, or next year’s return on assets. Return on assets (RoA) is the ratio of profit before interest and tax less extraordinary profit on total assets.

2 This study views short-term as the same year and the medium-term as the next year.

3 The definitions of the 42 indicators used in the CI and DI indexes are discussed in more detail in the literature review sections under the headings Control and Direction and summarised in Appendix 1.

4 A more detailed explanation of agency problems stemming from goal divergence is discussed in the literature review under the heading Agency relationship.

5 The various King reports on corporate governance were issued by the Institute of Directors Southern Africa (IoDSA) in 1994, 2002, 2009 and 2016 and are commonly referred to as King I, King II, King III and King IV.

6 Agency cost is the cost incurred to align the goals of the board and shareholders and limit the maximisation of self-interest.

7 Bonding or bond refers to the limitations and obligations placed on the board and shareholders by the Companies Act that glue or bond them to the requirements of the Act.

8 CFO is the Chief Financial Officer

9 The study references the recommendations of King III as the document applied to the study’s period, these principles remain valid as King IV does not “represent a significant departure from the philosophical underpinnings of King III” (IoDSA, 2016).

10 The audit and S&E committees are statutory committees required by the Companies Act.

11 Negative attributes could increase goal divergence and are coded as 1 for the absence of the negative attribute.

12 Ethical leadership was calculated by assessing ethical terms using Leximancer, an automated content analysis tool, to analyse the integrated reports of all the companies for all periods. More information on the terms and process used can be obtained from the corresponding author.

13 Ubuntu comes from a Zulu phrase, “Umuntu ngumuntu ngabantu”, which means “a person is a person through other people” (Ifejika, 2006).

14 King III recommended annual performance evaluations whereas King IV recommend evaluations every second year to allow the board more time to respond to the previous results (IoDSA, 2016).

15 One company changed controlling shareholders between 2012 and 2013, and the controlling shareholders remained constant between 2013 and 2014 (85 companies with controlling shareholders), resulting in the exclusion of CONSHA from the panel analysis as the variable is a near constant over time. However, a t-test on the differences between groups show the average DI for companies without controlling shareholders at 70 in contrast to 63 for companies with a controlling shareholder (p<0.000), suggesting that DI can be viewed as an alternative to shareholder monitoring.

16 This study uses the average of the prior two year’s return on assets as a proxy for prior performance whereas other studies used a one-year lagged dependent variable.

17 This study focuses on controls and aims to better understand the relationship between board-level controls and firm performance requiring the use of control measures that are known influencers of firm performance, such as prior performance. The study added the average prior two year’s return on assets to control for the effect of prior year’s firm performance on board-level controls in line with the method used by Fosu (2013).

18 MVBV were transformed using to a logarithmic form shown as LMVBV, to improve the normality of the distribution (Wooldridge, 2015).

19 The natural log transformation is used to improve the linearity of amount variables such as total assets and audit fees. Log transformations are commonly used on positive amounts to reduce variation as the transformed distribution is closer to normal (Wooldridge, 2015)

20 It is possible to use the central limit theorem to argue asymptotic normality for a dependent variable (Wooldridge, 2015), however, the normal and logged versions were run to determine if a log would result in a better overall model. LEV improved R² from 0.991 to 0.9989 and LNYEV improved R² from 0.994 to 0.9976, as a result the logged transformation was used in the analysis.

21 As this study used a re-extraction process to ensure the data extracted was correct, a statistical reliability test was not used. Cronbach alpha is frequently used as a reliability measure for survey instruments that tests the perceptions of people to assess the level of consistency in their answers.

22 A pairwise Granger causality test was run between all the dependent and independent variables using both directions. The results suggest a causal influence between CI and RoA (F=3.2148, p<0.045) and LEV (F=4.095, p<0.02). No causal influence was found for DI or any of the dependent variables.

23 Total remuneration consists of base pay that includes benefit contributions, bonuses received, and share options exercised.

24 Excluded as the company was delisted in July 2012.

25 Excluded as the company was sold in 2013 to a private consortium and changed to a private company.

26 Excluded as the company was placed under curatorship 10 August 2014.

27 Changed the companies name from Pretoria Portland Cement Company Ltd to PPC Ltd in 2012.

28 Name changed from Rainbow Chicken Limited to RCL foods in 2013.

29 Excluded as the company became a wholly owned subsidiary of Allied Electronics Corporation Limited in 2013.

30 Name changed from Capital Shopping Centres Group Plc to Intu Properties Plc in 2013.

31 Name changed from ABSA Group Limited to Barclays Africa Group Limited in 2013.

32 Excluded as Sibanye Gold Ltd originated from a split from Gold Fields Ltd in 2013.

33 Excluded as Capevin is a passive holding company that only listed in August 2012 and was previously known as KWV Ltd.

34 Mondi and Investec have a UK based Plc and a Local Ltd company, as the boards and integrated reports are the same this study only used the Ltd.

35 Telkom SA SOC is a State-Owned Company as the South African government is the controlling shareholder.