自1973年The Bretton Woods System崩潰後，各國相繼放棄固定匯率制度，轉而採用浮動匯率制度（Floating Exchange Rate System），導致利率及匯率之變動加劇。台灣為一小型開放之島國經濟型態，對外貿易活動頻繁，故會持續性地面臨匯率與利率波動風險。加上我國加入WTO後營建產業邁向國際化，營建材料供應商由於進出口頻繁，且因舉借外幣負債或進出口貿易業務之因素，勢必面臨利率與匯率波動之風險，此外近年來營建物價更是普遍大幅上揚，對國內工程更是帶來不小之衝擊。本研究以營建業中與匯率、利率、物價息息相關之營建材料供應商為研究對象，藉由支撐向量機（Support Vector Machines；SVM）所建立出之預測模型，提供營建材料供應商日後判斷是否使用衍生性金融商品避險之參考。本研究採上市32家營建材料供應商（包含7家水泥業與25家鋼鐵業）為研究對象，取其民國91~95年度之財務報表為樣本。藉由文獻回顧法初步確認本研究之衡量變數，並探討營建材料供應商公司內部財務資訊與使用衍生性金融商品避險之關連性。此外針對材料供應商使用衍生性金融商品規避匯率、利率之實際情形，以台灣經濟新報資料庫（Taiwan Economic Journal；TEJ）中公開之財務資訊，配合統計原理獨立樣本T檢定，篩選出具有顯著影響性之衡量變數，再以VIF檢定去除變數間之共線性關係，最後以支撐向量機建構預測模型，並與傳統統計方法中學者常用之邏輯斯回歸（Logistic）預測模型分析比較結果，進行驗證且探討營建材料供應商是否須使用衍生性金融商品規避匯率與利率之風險。因此本研究之研究成果有：1.瞭解營建材料供應商目前使用衍生性金融商品之概況。2.篩選出財務報表中影響企業使用衍生性金融商品之財務變數。3.由支撐向量機建立預測模型，提供營建材料供應商日後判斷是否使用衍生性金融商品之參考。 Floating exchange rates and interest rates have enhanced financial risks for those corporations which conduct international business or contain debt capital. Risk hedging, through the use of derivatives, has provided an effective solution toward such financial risks in recent years. Most construction material suppliers usually expose to these types of risks due to a high debt capital structure and the nature of material import business. A tool that is able to predict whether such a material supplier, based on its financial status, should use derivatives to hedge financial risks is demanded. This research objective is to develop a prediction model using Support Vector Machine (SVM) to provide suggestions for hedging financial risks. The scope limits the database to all 640 financial statements published in recent 5 years from 32 listed construction material suppliers. A total of 10 input factors were identified and determined using literature review, t-test, and co linearity diagnostics. Having data trimming and normalization, 640 sets were downsized to 520 sets which contain 248 effective and 272 ineffective risk-hedging sets. The SVM prediction model, thus, based on the kernel radial basis function and normalized data, yielded the prediction accuracy rate at 80.65%. The evaluation using the cross validation method shows the practicability and validation of the model. This study concludes that (1) 10 financial ratios are proved influential to financial risk hedging using derivative, and (2) the proposed SVM prediction model is feasible and applicable for the construction material suppliers. Future studies are recommended to apply the model to construction companies.
