本文利用統計科學方法建立企業（特別是科技產業）Logistic信用風險模型(Credit Risk Models)，考慮代理問題(Agency Problems)，找出影響公司績效之顯著變數，儘可能在公司尚未發生財務風險(Financial Risk)以前，預測出可能發生之機率，達事前預警與風險管理之效。實證分析上，利用財務與代理問題變數，區分經營權與所有權是否同一人，探討比較其差異。實證結果發現，經營權與所有權未分開，可降低代理問題其預測準確率亦較高，提供投資人選股策略參考；董事長與總經理為同一人，影響公司績效有償債能力、經營能力、獲利能力與現金流量指標；經營權與所有權未分開，企業儘早提高現金流量比率，以降低信用風險與發生財務危機之機率。董事長與總經理為不同人，影響公司績效有財務結構、償債能力、經營能力、獲利能力與現金流量指標；專業經理人(Chief Executive officer, CEO)宜降低現金再投資比率，增加業外收支率，多角化經營降低經營風險；專業經理人宜降低負債比率，提高總資產週轉率、每股現金流量與營業毛利率，增強獲利能力，提高公司績效。因此本文之實證價值與管理涵義為代理問題、財務資訊與公司績效。

This paper primarily used statistical methods to establish logistic credit risk models in electronic companies, discussed the agency problem on whether management and ownership belong to the same person. That made it be possible to predict in advance the probability of electronic companies experiencing financial risk. In the empirical analysis, the author built up a credit risk model using financial ratios and agency problems as the explanatory variables and the logistic regression model as the statistical method. Finding, In the case where management and ownership are combined, there had no agency problems and increased the degrees of predictive accuracy, empirical results suggested that the key performance indicators are the solvency, the operating performances, the profitability, and the cash flow, it had also been found that increasing cash flow ratios that reducing minimize agency problem and credit risk of the company. As for the case where management and ownership are separated, empirical results suggested that the key performance indicators are the financial structure, the solvency, the operating performances, the profitability, and the cash flow. It had also been found that chief executive officer (CEO) reducing the cash flow reinvestment ratios, and enhance ratio of non-operating revenue and expenses could reducing credit risk. It had also been found that reducing the debt ratio can enhance the total assets turnover, the cash flow per share, and the gross margin of the company. Therefore, this paper may be useful for researchers and practitioners who are focusing on agency problems, financial information, and corporate performance implementation.