Darryl Laws

 Regression method used. Malmendier and Tate use a binary logistic regression methodology to predict categorical outcomes from categorical and continuous predictors when predicting which of the two categories a CEO was likely to belong to; overconfident CEOs (independent variable) verses rational CEOs. The logistic regression or logit model method is appropriate as it is often used to model dichotomous outcome variables such as the dependent variable: CEOs holding their stock options to maturity and CEO’s overly acquisitiveness. In the logit model the log odds of the outcome are modeled as a linear combination of the predictor variables. When OLS regression is used with a binary response variable it becomes known as a linear probability model and can be used to describe conditional probabilities. However, the errors (residuals) from the linear probability model violate the homoskedasticity and normality of errors assumptions for OLS regression, resulting in invalid standard errors.

Malmendier and Tate’s theoretical framework, distinguishes overconfidence manifests itself in two forms, 1) an overconfident CEO overestimates the value of the potential merger, either due to the belief that his leadership skills are “better than average” (and better than the target’s current CEO) or due to an underestimation of the downside to the merger and 2) he believes that his company’s equity is undervalued by the capital markets due to the overestimation of his leadership skills and his ability to select profitable acquisition opportunities.

They commence their analysis by introducing the basic notation of their model … there are two companies, Acquiror A and Target T , which have market values of VA and VT respectively. The manager of A chooses whether or not to acquire T. They denote by c the total internal resources (cash and riskless debt) available to the manager of A and by c the amount of cash he pays to the target shareholders as part of the merger financing. V (c) is the market value of the combination of A and T , bV (c) the A manager’s valuation of the combination of A and T, and bVA his perception of his own company’s value if he does not pursue the merger. they call a CEO overconfident when bVA > VA and bV (c) − V (c) > bVA −VA for some c. The first condition is that the CEO overvalues his own company. The second condition is that the CEO overvalues the merger. To test the effect of managerial overconfidence on acquisitiveness, Malmendier and Tate use a general regression model of:

Pr{Yit = 1|Oit, Xit} = G(β1 + β2Oit + X0itB)


O is the long holder overconfidence measure. The set of controls X includes Tobin’s Q, cash flow, size, a measure of corporate governance, ownership, unexercised vested options (normalized by total number of shares outstanding) and year fixed effects. Y is a binary variable that takes the value 1 if the CEO made at least one successful merger bid in a particular year. Throughout the paper, they assume that G is the logistic distribution. The null hypothesis is that β2, the coefficient on overconfidence, is equal to zero.

Comments

Post a Comment

Popular posts from this blog

Darryl Laws

  Their model is binary because there are two kinds of variation that they use to identify the effect of overconfidence on acquisitiveness, cross-sectional and within-company variation. Malmendier and Tate (2008) solve their regression equation using three estimation procedures: 1) a logit regression, makes use of both types of variation, 2) a logit regression with random effects, also makes use of both types of variation but, it explicitly models the effect of the firm, rather than the CEO, on acquisitiveness. I noted that if the estimated effects of overconfidence in the logit equation are due to company effects, they should expect to see a decline in their estimates if they include random effects. The remainder of their solution uses a logit regression with fixed effects. This regression equation makes use only of the second type of variation, the effect of overconfidence on acquisitiveness using only variation between overconfident and rational CEOs within a particular firm. Fo...
Read more

Darryl Laws

  Log likelihood. Had I conducted the research I would have chosen to use log likelihood statistic to assess and to depict the  deviance, or -2 log-likelihood (-2LL) statistic . The deviance is basically a measure of how much unexplained variation there is in our logistic regression model the higher the value the less accurate the model. It compares the difference in probability between the predicted outcome and the actual outcome for each case and sums these differences together to provide a measure of the total error in the model. This is similar in purpose to looking at the total of the  residuals  (the  sum of squares ) in linear regression analysis in that it provides us with an indication of how good our model is at predicting the outcome. The -2LL statistic (often called the deviance) is an indicator of how much unexplained information there is after the model has been fitted, with large values of -2LL indicating poorly fitting models. The  deviance...
Read more

Darryl Laws

  The measurement they use is the coefficient odds ratio (log likelihood statistic). The odds ratio is crucial to the interpretation of logistics regression. It is an indicator of the change in odds resulting from a unit change in the predictor. The resulting statistic is based upon comparing observed frequencies with the predicted model. When the predictor variable is categorical the odds ratio is easier to explain. This is referred to as the likelihood ratio Field, 2018, pg. 614)  There are two very different approaches to answering the goodness of fit question. One is to get a statistic that measures how well you can predict the dependent variable based on the independent variables. These kinds of statistics are referred to as measures of predictive power . Typically, they vary between 0 and 1, with 0 meaning no predictive power whatsoever and 1 meaning perfect predictions. The other approach to evaluating model fit is to compute a goodness-of-fit statistic. Ordinarily the ...
Read more