4.5.1 Studies from 1994 and earlier

Kesner (1987) focused on 250 Fortune 500 companies from the United States of 1983. The relationship between the percentage of insiders on the board and six financial performance indicators was measured by correlations. Kesner included both accounting-based performance measures – profit margin, ROA, ROE and earnings per share – and market-based performance measures – stock price and total return to investors – in his database. The overall results are mixed (Kesner 1987: 503-505). The relationships between the percentage of insiders and profit margin and ROA were significantly positive, which implies that the relationships between board independence and these two measures are negative. ROE and earnings per share are positively correlated with percentage of insiders (i.e. negatively to board independence), but these results are not significant. The market-based performance measures do show negative results: stock prices are significantly negatively related to insiders and the relationship with total return to investors is insignificantly negative. This means that the market has a positive view of board independence (i.e. lower percentages of insiders) and rewards this with higher returns. Kesner also investigated whether the percentage of insiders of 1983 influenced the future performance in 1984 and 1985. These results show positive relationships, but only the relationship with total return to investors is significantly positive. This entails that board independence has an insignificant negative impact on future performance, according to Kesner.

Hill and Snell (1988) constructed a dataset with information from 94 Fortune 500 companies from five research intensive industries – chemicals, electrical and electronics, computers, industrial and farm equipment, and pharmaceuticals – in 1980. As a measure for board composition, Hill and Snell used the ratio between the number of inside directors and the number of outside directors.1 A correlation analysis shows a positive correlation of 0.88 with ROA (Hill and Snell 1988: 583-584). Although the significance of the relationship was not reported in the article, the magnitude of the relationship can be considered to be quite high. This means that lower levels of independence are positively related to (accounting) performance; higher levels of board independence lead to lower performance. However, these results are challenged by an ordinary least squares regression analysis in the same study. The insider to outsider ratio has a reported negative relationship with ROA. When the authors also included variables in the regression that correct for the strategy of the companies, the relationship became significantly negative (Hill and Snell 1988: 584-586). This can be interpreted as a positive impact of board independence on overall performance, which supports the view that outside directors act as guardians for the welfare of shareholders.

Molz (1988) applied a confirmatory factor analysis on nine measures of board composition and corporate governance, of which the percentage of outside directors2 on the board is one. Diversity, CEO duality, the number of shareholdings of insiders and the number of meetings are examples of other variables. The factor analysis was used to combine the nine corporate governance measures into one single factor (Molz 1988: 238-239). It appears that the percentage of outside directors has the strongest power in the constitution of the factor. The factor was used to divide the total sample of 45 randomly selected food companies from the 1983 Fortune 500 into two distinct groups: 21 companies with management-dominated boards and 24 companies with pluralistic boards. Molz tested whether ROA, ROE and total return to shareholders differ between these two groups. The results do not show significant differences (Molz 1988: 245-246). ROA and ROE are slightly higher in companies with pluralistic boards, whereas total return to shareholders is higher in companies with management-dominated boards. However, none of the results are significant and board independence is only part of the factor that splits the total group into two subgroups.

Fizel and Louie (1990) conducted research on the determinants of CEO replacement. In order to investigate this relationship, they selected all the 72 companies from the Forbes annual report on executive compensation that had replaced their CEO between 1984 and 1985. Each of these 72 companies was matched with another company from the same Forbes lists that had retained their CEO in that time period. The matching process was based on the industry in which the company was active. The probability of a CEO replacement does not appear to be influenced by board independence.3 Although the sign of the relationship is negative, the t-test does not point to a significant relationship. However, the squared variant of board independence shows a consistent and significant positive relationship with the probability of replacement (Fizel and Louie 1990: 171-174). The explanation for this can be that really independent boards, with high levels of independence, are more likely to replace a CEO than boards with lower percentages of board independence. In addition, the article published correlations between board independence and three performance measures: the deviation of the four-year average of ROA, earnings per share and the yearly change in share price. ROA and the share price change are negatively correlated with board independence, whereas earnings per share shows a positive correlation with board independence (Fizel and Louie 1990: 175).

Hermalin and Weisbach did not find any relationship between board composition and financial performance using the data of 142 NYSE companies from 1971, 1974, 1977, 1980, 1983. The dependent variable used is Tobin’s Q and the independent variable to capture board composition and independence in particular is a dummy variable. Hermalin and Weisbach constructed three dummy variables: from companies with a percentage of independent directors on their board of less than forty per cent, companies with board independence ranging from forty to sixty per cent, and a dummy if board independence exceeded sixty per cent. Whether the observations over the different years were pooled or not, a relationship – measured by an ordinary least squares regression – was not found between board independence and Tobin’s Q (1991: 106-111). Hermalin and Weisbach admitted that the relationship between these two variables might exist and that their methods failed to detect it, but in that case the relationship is at least very small or economically insignificant (1991: 111).