| dc.description.abstract | In this thesis we have investigated if firms engage in earnings management prior to issue of debt, and if asset volatility increases after. We used three models to look for earnings management, a cross-sectional model, a decile divided model and a panel data model. The cross-sectional model struggled to provide prediction of “normal” accruals, and thus proving to not be reliable to conclude on significance. The decile divided model proved to be a better fit than the cross-sectional model. However there seemed to be a systematical error which we couldn’t control for, resulting in negative Z-values. This meant that although we got promising average prediction errors, we couldn’t conclude if there was a significant difference. The panel data approach was implemented to control for covariance between error terms. Both the cross-sectional and decile divided model assumed covariance to be zero. We used the cross-sectional model as the basis when building the model. Using the cross-sectional model as basis proved to be detrimental for the panel model, this is because we transferred the inability to predict “normal” accruals to the panel model. The result was insignificant results from the panel model as well. We found an increase in average volatility of 11,5 percentage points after debt issue. We also found an increase in the number of firms which had significantly higher volatility one year after issue, compared to issue year. The risk-shift analysis gave support to our hypothesis of increased volatility after debt issue. We were also unable to confirm a relation between firms engaging in earnings management prior to debt issue and subsequently increasing volatility after. | en_US |
