CAPÍTULO III: MARCO METODOLÓGICO
3.7 DIAGNÓSTICO DEL CENTRO EDUCATIVO COMUNITARIO
3.7.1 FODA
This section evaluates the impact o f stock-based and accounting-based performance on the likelihood of a top management change. Table 4.6 presents the basic probit estimation results on the relation between top management turnover and performance measures. Note that A stands for all Most Senior Executive departures, F for forced MSE departures and NF for non-forced MSE departures.
Table 4.6: Estimates of Probit Models Relating MSE Turnover to Stock-Based and Accounting-Based Performance, Time-Period: 1990-1998, Sample: Top 460
London Stock Exchange Firms
Independent M odel 1 M odel 2
V ariables Dependent V ariables D ependent Variables
A F NF A F NF SHR,.] -0.081 -0.053 -0.005 -0.082 -0.045 -0.018 (0.000) (0.000) (0.607) (0.000) (0.000) (0.104) sh r,.2 -0.016 -0.008 -0.004 -0.007 -0.000 -0.007 (0.249) (0.284) (0.706) (0.632) (0.984) (0.548) EBIT,., - - - -0.173 -0.122 0.003 (0.003) (0.001) (0.936) EBIT,_2 - - - 0.093 0.008 0.050 (0.079) (0.808) (0.126) SIZE -0.003 -0.001 -0.000 -0.000 0.000 0.000 (0.356) (0.471) (0.936) (0.835) (0.808) (0.847) AGE 0.003 -0.000 0.003 0.004 0.000 0.003 (0.000) (0.865) (0.000) (0.000) (0.778) (0.000)
Time Effects Yes Yes Yes Yes Yes Yes
Industry Effects Yes Yes Yes Yes Yes Yes
Observations 3039 3037 3037 2828 2826 2826
Pseudo R2 0.066 0.101 0.071 0.075 0.115 0.082
Log Lik. -848.9 -439.4 -560.1 -785.2 -408.4 -514.2
NOTES:
1. A: All Most Senior Executive changes; F: Forced changes; NF: Non-Forced changes 2. p-values in parentheses
Consistent with prior research both in the US and the UK (e.g. Parrino 1997; Conyon 1998), poor firm performance increases the probability of executive turnover. Model (1) presents estimates where two lags o f own stock return as the independent variables were included. The marginal increase in the probability of executive turnover and forced turnover, when there is a marginal decrease in stock performance, is 0.081 and 0.053 respectively; both estimates are significant at less than the 1% level. Lag two of stock return is not significant for both all changes and forced changes.
In Model (2) two lags of accounting returns as additional performance variables were included. The negative marginal effect of the first lag of EBIT ( 0.173 for all changes
and -0.122 for forced changes) reinforces the previous finding that top managers are dismissed for poor performance. Contrary to stock returns, the second lag of accounting returns has a positive marginal effect, but is not significant at conventional levels other than in the all changes model. In particular, under all changes the second lag of accounting returns enters with a positive sign of 0.093 and is significant at the 10% level. An interpretation of this result is given below. A general model of executive turnover is:
Pr(MSE Turnover,,) = a + p {Y\u_x + /? 2fl*., + /?,Size,, + PAAge,, + £„ (a)
where n is a measure of level of profit, and in this case of accounting profit. Model (a) could be re-written as:
Pt(MSE Turnover,,) = a + (/?, + )n ( n - n ,,_2) + /?, Size,, + [), Age,, + e„ (b)
According to Model (b), the sum of Pi and P2 estimates the effect o f a change in prior
year’s level of accounting earnings whilst P2 estimates the effect o f a change in prior
year’s difference in accounting earnings. An alternative interpretation of the estimates
is, therefore, that the turnover probability will increase by 0.080 (-0.173+0.093) if there
is a unit decrease in the level o f prior year’s accounting earnings or by 0.093 if there is a
one unit decrease in the growth o f prior year’s accounting earnings. Finally, the
inclusion of accounting earnings in Model (2) does not alter the effect of prior year’s shareholder returns which remains negative and significant under both all changes and forced changes.
As mentioned in Chapter 2, the effectiveness of the internal control mechanisms can be assessed by comparing forced departures with non-forced departures. The estimates of all performance measures in the non-forced turnover in Model (2) reveal that, in contrast with forced top executive changes, there is no evidence of significant relation between the likelihood of non-forced departure and performance. Instead non-forced departures are driven mainly by the age variable which enters with a positive sign (0.003) and significant at less than the 1% level. Bearing in mind that the majority of non-forced departures were due to retirement (49.4%-see Table 4.2) the result suggests that the older the executive the closer he is to retirement age and hence the higher the probability of voluntary turnover.
Overall, results in Table 4.6 suggest that internal governance institutions, such as the directors' board or large shareholders, are effective in their monitoring and disciplining tasks. This thesis attempted to expand the above analysis by identifying some o f the characteristics of an effective board. In particular, two board level variables - often considered in the literature - were incorporated in the turnover equations of Model (2) - Table 4.6. These include: a) board size (242) and b) board independence measured by the proportion of non-executive (outside) directors (243)21. CEO turnover is presumed to be positively associated with the fraction of outside directors (Fama and Jensen 1983; Williamson 1983) - although as discussed in Section 4.2.1 several outside director characteristics may inhibit their efficiency - and negatively associated with board size (Paton and Backer 1987; Lipton and Lorsch 1992; Jensen 1993).
The primary results o f this analysis can be summarised as follows. Firstly, it was found that on average UK boards consist of 10 directors of which 45% are non-executives. Secondly, analysis indicated that the fraction of non-executives is not a significant predictor of CEO turnover under all three types of changes. Thirdly, contrary to theoretical expectations, board size appears to be positive and significant under all and forced MSE departures at the 10% and 5% level respectively. Nevertheless, the economic significance of this variable is trivial; a marginal increase in board size is associated with a 0.2 percentage points increase in the forced MSE turnover likelihood. Finally, the inclusion of the two board level measures does not change the effect of the study's primary explanatory variables; the marginal effect of prior year's stock returns is -0.084 and -0.047 for all and forced changes respectively whilst the equivalent marginal effects of prior year's accounting returns are -0.184 and -0.124. Once again, performance is not associated with non-forced MSE departures.
To close this section, it is important to emphasise that the above findings should be treated with caution as they are subject to two main caveats. Firstly, the quality o f the data may bias the results. More specifically, in his pioneering study on the efficiency of board monitoring, Weisbach (1988) classifies directors into three categories: a) outsiders: those who neither work for the firm nor have extensive dealings with it, b) insiders: full-time employees of the firm and, c) grey: those who are not employees, but who may not be entirely independent of existing management due to their extensive business dealings with the company or family affiliations with the management. Since the “proportion o f outside directors” variable - employed in the current study - is not able to isolate the effect of “grey” directors, the efficiency of outsiders may be under estimated.
Finally, the endogeneity of board monitoring may limit our ability to directly model the efficacy of board in replacing poorly-performing CEOs. The vast majority of empirical studies (including the current one) treat board size and board composition as exogenous variables (e.g. Weisbach 1988; Hadlock and Lumer 1997; Dedman 2000; Dahya et al. 2001). However, a recent stream o f theoretical papers model CEO monitoring where boards are treated as endogenously-determined institutions (Hermalin and Weisbach 1998, 2000; Warther 1998). The striking insight of these studies is that the intensity with which CEO monitoring is carried out decreases under the assumption that board composition or (more generally, the behaviour of the board) is itself determined by various CEO characteristics (e.g. tenure). Therefore, until board endogeneity is explicitly addressed the power of the tests to detect the elements o f an effective board is compromised.