Corporate Social Responsibility, Corporate Governance, And Managerial Risk-Taking

This dissertation investigates the association between corporate social responsibility (CSR) and managerial risk-taking, as well as the differences in governance structure that affect this association. Using a sample of US public firms from 1995 to 2009, we find that firms with strong CSR records engage in higher risk-taking. Furthermore, we find that this relationship is robust when accounting for differences in governance structure and correcting for endogeneity via simultaneous equations modeling. Additional testing indicates that performance in the employee relations dimension of CSR in particular increases with risk-taking, while high firm visibility dampens the association between CSR and the accounting-based measures of risk-taking. Prior literature establishes that high managerial risk-tolerance is necessary for the undertaking of risky yet value-enhancing investment decisions. Thus, the main findings suggest that CSR, rather than being a waste of scarce corporate resources, is instead an important aspect of shareholder value creation. They contribute to the debate on CSR by documenting that corporate risk-taking is one mechanism among others through which CSR maps into higher firm value.

Vulnerability to adverse consequences of drinking and problem drinker status as predicted by risky drinking behaviours, drug use, sex differences and affect : a test of multiple models /

Although alcohol problems and alcohol consumption are related, consumption does not fully account for differences in vulnerability to alcohol problems. Therefore, other factors should account for these differences. Based on previous research, it was hypothesized that risky drinking behaviours, illicit and prescription drug use, affect and sex differences would account for differences in vulnerability to alcohol problems while statistically controlling for overall alcohol consumption. Four models were developed that were intended to test the predictive ability of these factors, three of which tested the predictor sets separately and a fourth which tested them in a combined model. In addition, two distinct criterion variables were regressed on the predictors. One was a measure of the frequency that participants experienced negative consequences that they attributed to their drinking and the other was a measure of the extent to which participants perceived themselves to be problem drinkers. Each of the models was tested on four samples from different populations, including fIrst year university students, university students in their graduating year, a clinical sample of people in treatment for addiction, and a community sample of young adults randomly selected from the general population. Overall, support was found for each of the models and each of the predictors in accounting for differences in vulnerability to alcohol problems. In particular, the frequency with which people become intoxicated, frequency of illicit drug use and high levels of negative affect were strong and consistent predictors of vulnerability to alcohol problems across samples and criterion variables. With the exception of the clinical sample, the combined models predicted vulnerability to negative consequences better than vulnerability to problem drinker status. Among the clinical and community samples the combined model predicted problem drinker status better than in the student samples.

Evolutionary synthesis of stochastic gene network models using feature-based search spaces

A feature-based fitness function is applied in a genetic programming system to synthesize stochastic gene regulatory network models whose behaviour is defined by a time course of protein expression levels. Typically, when targeting time series data, the fitness function is based on a sum-of-errors involving the values of the fluctuating signal. While this approach is successful in many instances, its performance can deteriorate in the presence of noise. This thesis explores a fitness measure determined from a set of statistical features characterizing the time series' sequence of values, rather than the actual values themselves. Through a series of experiments involving symbolic regression with added noise and gene regulatory network models based on the stochastic 'if-calculus, it is shown to successfully target oscillating and non-oscillating signals. This practical and versatile fitness function offers an alternate approach, worthy of consideration for use in algorithms that evaluate noisy or stochastic behaviour.