4.4.2 Studies from 1995 until 2004

Yermack found a significant negative relationship between board independence1 and financial performance. Financial performance was measured in this article by Tobin’s Q (1996: 195). Yermack used ordinary least squares models: with and without firm fixed effects. In both models the relationship is negative in a sample of 452 American companies and 3,438 company year observations over a period of eight years between 1984 and 1991.

Coles et al. (2001) used the changes in the market-based performance measure market value added,2 and the accounting-based performance measure economic value added3 over the period 1984 and 1988 as performance measures in their study. They conducted research on a sample of 144 large companies from the United States, which relied partly on the sample of Jensen and Murphy (1990). They conjectured that companies with higher levels of board independence4 would have higher performance than companies with high proportions of inside directors on their board. A correlation analysis showed the opposite result: the level of board independence is negatively correlated with market value added as well as economic value added (Coles et al. 2001: 38-39). The ordinary least square regression analysis showed similar results. Board independence has a negative impact on market value added and economic value added (Coles et al. 2001: 39-42). These results are consistent for all models. When the models were controlled for CEO tenure effects, the relationship between board independence and market value added appeared to be significant (negative) at a 5 per cent level. In the end, the authors found no evidence for their hypothesis; their results even point in the opposite direction.

Bhagat and Black (2001) researched boards in the United States and found negative relationships between board independence5 and performance, which was measured by Tobin’s Q, ROA and the ratio between sales and total assets. Their analysis was based on a data sample of 934 large American companies over the period 1985 until 1995, but results were only published for the aggregated periods 1988-1990 and 1991-1993. These results, derived from ordinary least squares and three-stage least squares regressions, show – without exception – a negative impact of board independence on the different performance measures (Bhagat and Black 2001: 247-253).

Mishra et al. focused their study on the Norwegian market, which is considerably influenced by family ownership (2001: 237-238). A quarter of the companies in Norway is owned by founding families, who possess at least twenty per cent of the shares in these companies. Due to high family ownership percentages in Norway the stock market was not very well developed. The annual market turnover, calculated as the percentage of active stocks, was only 3 per cent in 1981 and rose to 40 per cent in 1996. This is still rather small. Therefore, Mishra et al. aimed to investigate the impact of family ownership on the financial results of a company measured by Tobin’s Q. Board independence6 is a control variable in these regressions. First of all, the descriptive statistics showed that the level of board independence is significantly lower for family companies, which are defined as companies in which the founding family has a shareholding of at least twenty per cent (Mishra et al. 2001: 243-246). They found that family ownership has a significant positive effect on company performance, but board independence has a significant negative impact on that performance (Mishra et al. 2001: 249-251). Mishra et al. argue that board independence does not improve corporate governance in a family company and is less desired, because the board is already aligned with the founding family and propagates values like trust and altruism (2001: 255). Boards in these family companies are consequently aimed more at strategy matters than monitoring.

Beekes et al. (2004) investigated whether board composition influences accounting conservatism. They conjectured that the time that elapses before bad news is reflected in the earnings statement is positively related and the time that elapses before good news is reflected is negatively related to the percentage of NEDs on the board (Beekes et al. 2004: 49-50). In these hypotheses higher percentages of NEDs on the board are associated with accounting conservatism. The underlying assumption is that NEDs are more concerned with their reputation and possible liability, which makes them announce bad news earlier. Since executive directors are more focused on their performance-based pay, they are more inclined to accelerate the inclusion of good news in the earnings statement. In a sample of 205 United Kingdom listed companies, 501 company year observations over the period 1993 to 1995, the authors found greater time lines for bad news than for good news, but could not attribute the difference to the composition of the board (Beekes et al. 2004: 57-58). The performance measure chosen is earnings per share divided based on the share price at the start of the period. The performance measure shows, in an ordinary least squares regression, a negative relationship with the proportion of NEDs on the board (Beekes et al. 2004: 54-57). In a number of estimated models, the coefficients of the percentage of NEDs are negative and some are significant, but overall the results for the relationship between performance and the percentage of NEDs are not robust. In addition, the total sample is split into two groups, based on the median of the percentage of NEDs on the board. This shows that the group of companies with boards that have relatively low percentages of NEDs has a significant, higher average performance. This only applies to the average value of the performance: the median values do not differ significantly (Beekes et al. 2004: 51-53).