This research project is based on three individual essays which are inter-linked. Our research starts from the development of a firm’s profitability forecasting model. Given the much shorter history of quantile regression and its infrequent use in finance and accounting compared to the popular ordinary-least-squares (OLS) regression, it is not clear whether stock prices have fully reflected the incremental information from quantile-regression profitability forecasts over and above the information contained in their OLS counterparts. This thesis examines the issue using a forecasting and a hedge portfolio analysis. We construct quantile-regression forecasts on an economy-wide and an industry-specific basis and compare them to their OLS counterparts out-of-sample. We verify that the quantile-regression forecasts are more accurate than their OLS counterparts on either basis. We further show that a hedge portfolio formed by contrasting the quantile-regression forecasts to their OLS counterparts, whether on an economy-wide or industry-specific basis, can earn an abnormal return. Our results hold for a number of new and traditional profitability measures and are not sensitive to various methodological and sample choices. Following similar interests in the profitability forecasting model, we focus on loss-making firms in the second essay. We examine the effects of diversification on loss persistence. Diversified firms have higher probability of loss reversal than focused firms. Using various measures of the abandonment option, we find that diversified firms can liquidate their loss-making assets or segments more efficiently than focused firms so as to achieve profits in the following year. Additional tests suggest that the efficiency of the abandonment option for diversified firms can be dampened by the agency problem of over-investment. Our findings are robust to various agency problem proxies and our analyses are controlled for endogeneity.In our third essay, we investigate the effect of firm structure on dividend payout ratio. Consistent with the substitute theory, we find that diversified firms have significantly higher payout ratio than focused firms. We develop our analysis based on two hypotheses which are agency problem hypothesis and efficient internal capital market hypothesis. Under the agency problem hypothesis, we find that diversified firms have much higher dividend payout ratio than focused firms when they are under the agency problem but no difference in payout ratio when they are out of agency problem. Under the efficient internal capital market hypothesis, we find that diversification increases the dividend payout ratio for firms that are financially constrained but not for financially unconstrained firms. These findings further confirm the substitute theory that dividend payment is used as a signal for firms with agency problems and financial constraints. Our results are robust under various additional tests of endogeneity problems.
|Date of Award||22 Dec 2018|
|Supervisor||David Newton (Supervisor)|