Some of these conclusions can be also found inanother research papers such as Lipton and Lorsch (1992); and Jensen 1993 inwhich is highlighted that when the boards members size is larger they can showcommunication problems so that less efficiency.
Also, in these studies can befound a recommendation that the board size need to be around 8 or 9 persons (nomore than that). However, there are another studies that show that this is notnecessary true because it depends of the firm size ( Boone et al, 2007 andLinck et al 2008). There are another studies such as Adams and Mehran, 2005 andBeiner et al 2006 that using instrumental variable techniques have shown thatthe company performance is not related of the board size. However, Wintoki(2007) highlighted that these variable techniques are not a good way to explorethe performance of a company, he point out that the best model to evaluatethese variables is to apply the GMM estimator. But, due to the complexity ofthe relationship of these variables another studies show procedures andimportant findings such as that the performance of the company may differdepending of the type of company and it is not related of the board size (Lincket al, 2008; Coles et al (2008)). Objectivesof the StudyThe objectives of this study can be described asfollows:1) Analyse the relationship between the boardgovernance and company performance based on the information available on somecompanies2) Find out the relationship between ownershipstructure and corporate performance, and also the relationship between boardcomposition and corporate performance3) Find out how the CEO characteristic affect thecorporate performance2) Provide some tools that can be applied toevaluate the relationship previous mentioned3) Develop a set of recommendations that can beapplied in any type of company which can be used to evaluate the boardgovernance and company performance. Empirical issues, Model & Data In the studiespresented in the literature review, data from several firms was studied andanalysed, using hierarchical regression model, multilevel regression model,linear models and association between variables was tested by statisticallearning methods such as ttest for testing the difference between two groupsmean value, the association was checked using pearson correlation coefficient.According to the distribution of the data, parametric or non-parametricstatistical hypothesis test could be used to assess the relation between thebroad governance and company performance.
Sample descriptionFor the statistical model it was applied regressionmodels such as lineal model (LM) and generalized linear model (GLM):g(E(Y)) = ?0 + ?1 x1 + ?2 x2 +…+ ?n xnAgeneral linear model (also called GLM, hence create confusion), there is no gfunction and f functions are scalar multiplication by numbers. So, the model isof the form:Y = ?0 +?1 x1 + ?2 x2 +…+ ?n xnThemain differences between these two methods is that for general linear modelsthe distribution of residuals is assumed to be Gaussian. If it is not the case,it turns out that the relationship between Y and the model parameters is nolonger linear.
But if the distribution of residuals is one from the exponentialfamily such as binomial, Poisson, negative binomial, or gamma distributions,there exists some functions of mean of Y, which has linear relationship withmodel parameters. This function is called link function.Forthe evaluation of the hypothesis it can be employed methods such asMann-Whitney U test. This methodology is used to compare differences betweentwo independent groups when the dependent variable is either ordinal orcontinuous, but not normally distributed.
In this method is necessary to checkif the data that will be analysed fulfil four assumptions in order to get avalid result. Assumption # 1 The depend variable will be measured at the ordinal or continues level Assumption # 2 The independent variable should consist of two categorical, independent groups. Assumption # 3 The observations are independent which means that there is not a relationship between the observations in each group or between the groups themselves. Assumption # 4 A Mann-Whitney U test can be used when your two variables are not normally distributed. Afterthat it will be applied the Pearson Product-Moment Correlation or Pearsoncorrelation coefficient which is a measure of the strength of a linearassociation between two variables and is denoted by r.
In the cases that is obtained a value of0 that means that there is not a relationship between the variables. A useful technique to analyse the effect of different factors on aresponse is to perform an Analysis of Variance. It is important to know whichtype of analysis can be done in order to determine the main factors that have agreat effect in the data or how much of the variability in the responsevariable is attributable to each factor.This study was developedwith a dataset from 2004 to 2015 considering full board members, and evaluatingif these members are responsible for risk, risk committee, or not. And, isevaluated if the CEO is responsible for risk management or how the CEO isinvolved in the risk management. Table 1 shows an overview of the variablesthat was used to fulfil the objectives of this study.
More details of this datawill be shown in the Annex.