4.3.1 Studies from 1994 and earlier

Baysinger and Butler collected a dataset of 266 large United States companies, which are all listed in the 1970 edition of Forbes.1 They found a significant positive relationship between board independence2 in 1970 and the industry-adjusted ROE3 in 1980 (1985: 115-116). The significant relationship is measured by Pearson’s correlation. They also tested the relationship between board independence in 1970 and industry-adjusted ROE in 1970; they did the same for 1980. However, they found no contemporaneous relationship between board independence and this accounting-based performance measure in 1970 or in 1980. The lag period of ten years implies that the board composition of today would affect the financial performance over ten years. Further research shows that companies with relatively high levels of board independence at the start of the period (1970) have a higher ROE in 1980 than companies with lower levels of board independence in 1970. This advantage remains, even if companies with relatively low levels of board independence in 1970 try to make up arrears (Baysinger and Butler 1985: 116).

Schellenger et al. (1989) did find these contemporaneous relationships between board independence4 and performance. They conducted an empirical analysis of 526 companies from Compustat Industrial5 and Center for Research in Security Prices (CRSP) and used information from 1986. Besides two accounting-based performance measures – ROA and ROE – they applied a correlation analysis to market-based performance measures: shareholders’ total market return on investment and the risk-adjusted variant of this variable.6 They found a significant positive relationship between board independence and ROA and the risk-adjusted shareholders’ return, which is in line with their hypothesis which is based on the agency theory (p [amp]lt; 0.01) (Schellenger et al. 1989: 462-465). The signs of the relationships with ROE and the unadjusted shareholder returns are positive, but not significant.

Judge and Zeithaml (1992) conducted qualitative as well as quantitative research on board involvement in strategic decisions. They used a dataset of 42 United States companies, of which twelve are general hospitals, ten biotechnology companies, ten textile companies and ten diversified companies from the Fortune 500.7 Besides in-depth interviews with people involved, they constructed a database over the period 1985-1987. Board composition in this study is measured as the percentage of inside directors on the board, with insiders defined as employees and NEDs affiliated with the company (i.e. nonindependent directors). The results show that higher percentages of insiders on the board – i.e. the opposite of board independence – have a negative impact on board involvement in the strategic decision-making process (Judge and Zeithaml 1992: 781-783). This implies that more independent boards are more involved in this process. Besides this result, the relationship between the proportion of inside directors on the board and performance, measured as the five-year average ROA divided by the industry average ROA, is reported. The result of this part is negative, which means that they have found a positive relationship between board independence and performance.

In their research Pearce and Zahra (1992) also reported positive findings for the relationship between board independence and performance. They utilised four different accounting-based measures of financial performance: ROA, ROE, earnings per share and net profit margin. In the study of Pearce and Zahra thee companies supplied all the information through a survey, which explains why only accounting-based performance measures are available. The authors used a three-year average over the period 1987 to 1989. Pearce and Zahra calculated the Pearson’s correlations and found positive numbers for the correlation between the three-year average measures of performance and the 1986 board composition numbers (Pearce II and Zahra 1992: 431). The relationship between board independence and these future measures of performance is significant for ROA, ROE and earnings per share. The relationship with net profit margin is positive, but not significant. Two board composition variables are applied in this study: percentage of outsiders with no affiliation (i.e. percentage of independent NEDs) and percentage of outsiders with an affiliation (i.e. percentage of nonindependent NEDs). The sign and significance of the correlation between both board composition variables and performance measures are comparable.

Barnhart et al. (1994) used the market-to-book ratio as performance measure. They found evidence for their hypothesis that the relationship between board independence8 and performance is not linear, but curvilinear. To investigate their beliefs, they used a sample of 369 Standard & Poors 500 (S&P 500)9 companies from 1990. An ordinary least squares regression analysis and ordinary least squares with instrumental variables techniques confirmed their observations by presenting significant relationships between the variables (Barnhart et al. 1994: 335-338). The curvilinear relationship between board independence and market-to-book ratio means that the increase of board independence has a positive impact until a certain level of board independence is reached. If board independence rises further, the market-to-book ratio declines. After a certain threshold – the authors estimate that this threshold exists at a level of two thirds of the board – board independence causes the market-to-book ratio to increase rapidly. Apart from these regression analyses, Barnhart et al. found a negative correlation between board independence and performance measured by the market-to-book ratio (1994: 333).

Li (1994) investigated 390 multinational companies from ten different industrial countries10 on the determinants of outsiders in the board room. With data from 1987, he shows that performance – measured as the average of the 1986 and 1987 profit margin – has a positive influence on the ratio of outsiders11 on the board (Li 1994: 364-366). The relationship measured by ordinary least squares is significant only in one of the two estimated models. Although the causality is the other way around, this result might be evidence for a positive linear relationship between independence and the average profit margin.