Financially-Distressed Firms in Taiwan: Predictions and Earnings Management Behaviors

• Main Authors: Wan-Ting Hsieh, 謝宛庭
• Other Authors: TsingZai Wu
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/17887284119705457808

博士 === 國立成功大學 === 會計學系碩博士班 === 94 === Part I There are clues before a firm faces financial disaster. To reduce costs resulting from information asymmetry and protect investors and creditors, it is desirable to develop a better model for predicting financially-distressed firms. This study aims at developing a hazard model to predict firms that face the financially distressed risk in Taiwan. Unlike prior studies, for the first time I employ Wu and Lilien (2003)’s earnings outliers method by using outlier analysis on earnings series in my prediction model for financially-distressed firms in Taiwan. In this study, I employ the discrete-time hazard model to construct my prediction model, which combines survival analysis and logit model to predict financially-distressed firms. A hazard model corrects coefficients for PERMiods at risk and allows time-varying covariates. It uses all available information to estimate the probability for financially-distressed firms at each point in time in the study. Our empirical results show the followings: significant independent variables for predicting financially-distressed firms in the discrete-time hazard model are time dummy variables (T3, T5, T7), the current ratio, the debt ratio, the net income to assets ratio, the retained earnings to total assets ratio, the cash flow to liabilities ratio, stock abnormal returns, variance of stock returns, directors and managers shareholdings, board chairman acting as CEO, temporary earnings outliers, auditor switch, receiving a going-concern GC, default in debt covenants, changes in CEO, CFO and director of internal auditor. However, the hypothesis 3 (SIZE, AGE) are not supported from my empirical study. Furthermore, Empirical results show that two earnings outliers can make model concordance reaching 75%, which implies Taiwan’s listed firms having negative permanent or temporary earnings outliers may face financial disaster. I further compare the prediction accuracy of discrete-time hazard model, logit model and auditor GC. Considering the actual beforehand proportion and cost ratio, our empirical results show the followings: the prediction accuracy of discrete-time hazard model is higher than that of logit model and auditor’s GC. However, the comparative bases of discrete-time hazard model and auditor GC are different. An auditor’s GC is based on beforehand information before firms face the financially distress risk. On the other hand, the discrete-time hazard model is based on the ex-post information after firms face the financially distress risk. Such difference may lead to the higher prediction accuracy of discrete-time hazard model than that of auditor’s GC. In addition, the prediction accuracy of discrete-time hazard model is also higher than that of logit model. Although discrete-time hazard model needs information in all survival time, it adds value in predicting accuracy comparing to single-PERMiod logit model. Part II As a firm faces financial difficulties, its management is apt to window-dress financial reports for meeting outsiders’ expectations. Earnings management behaviors may include lower-cost discretionary accounting treatments or higher-cost real activities that affect operational cash flows. This study aims at exploring earnings management behaviors of financially-distressed firms in Taiwan. The results of earnings management of financially-distressed firms will overstate income. First, I investigate whether managers of financially-distressed firms are more likely to make income-increasing earnings management. Empirical results indicate that, for concealing unfavorable financial situation and increasing firms’ income, managers of financially-distressed firms in Taiwan take both accounting accruals and real activities manipulations. Empirical results support my hypothesis I. Second, I explore whether going-concern opinion issued by an auditor will affect its client’s earnings management behaviors. When a firm gets a going-concern opinion from its CPA, the auditor will generally require its client to reverse the overstated earnings. My empirical results show that firms receiving going-concern opinions from their CPA’s will not take accounting accruals and real activities manipulations that increase firms’income. This conclusion is consistent with my hypothesis II.