研發支出之會計處理，一直以來，因著研究發展之特性，始終有著相當分歧的看法。由於研究發展支出具有長期性及未來的經濟效益，有人主張將其以「資本化」方式處理；亦由於研究發展支出具有高度的風險與不確定性，有人主張將其以「費用化」方式處理。 Shi(2003)認為研究發展支出資本化與費用化之爭論，正反映了研發支出未來效益及其風險間之抵換關係(trade-off)，亦即，若研發支出之未來效益大於其風險，則較傾向將其資本化，其會計處理同於一般的無形資產；相反地，若研發之風險大於其未來效益，則較傾向將其以費用化方式處理，於發生當期即以費用入帳。 本研究以台灣債券市場為研究對象，探討研發支出未來效益與風險間之抵換關係，文中檢視「債券風險衡量因子」(bond risk measures)與「研究發展支出」之相關性，並以「債信評等等級」與「債券風險溢酬」為債券風險衡量因子，決定平均數效果(預期未來效益)與變異數效果(風險)於債券的評價上何者較為顯著。 一般而言，以債券投資者的角度觀之，若「債券風險衡量因子」與「研究發展支出」兩者呈現負相關，亦即平均數效果較強，則代表研究發展之未來預期效益大於研究發展之風險；若此兩者呈現正相關，亦即變異數效果較強，則代表研究發展之風險大於研究發展之未來預期效益。本研究之實證結果與發現如下： 1.對全體樣本而言，研發支出與債信評等等級呈顯著之正相關(本研究採用TCRI為債信評等衡量變數，等級愈高，風險愈大)，代表研發支出之風險大於其未來效益。然研發支出與債券風險溢酬之關係未達統計顯著水準，無法再次驗證上述結果。 2.對電子業樣本而言，與上述對全體樣本之結論相同。 3.對非電子業樣本而言，研發支出與債券風險溢酬為顯著之負相關，代表研發支出之未來效益大於其風險。然研發支出與債信評等等級之關係未達統計顯著水準，無法再次驗證上述結果。 4.在全體樣本、電子業樣本、及非電子業樣本中，將研發支出以費用化或資本化方式予以衡量，兩者之實證結果並無不同，顯示兩者對研發支出未來效益與風險間之抵換關係並無顯著差異存在。 5.電子業與非電子業所獲之結論不同，再次驗證產業別對於研發支出之效果確實有其差異性。 6.針對電子業而言，本研究之實證結果較傾向以費用化之方式處理其研發支出；然針對非電子業而言，較傾向以資本化之方式處理之。 / The debate about the alternative accounting treatments of R&D expenditures reflects trade-offs between the future benefits of R&D and its risk. In general, if the uncertainty regarding future benefits is not so high that it disqualifies the measurability criterion of asset recognition, then one may argue in favor of capitalizing R&D expenditures (as is typical for intangible investment). Conversely, if future outcomes are risky and unpredictable, the expensing treatment may be warranted. This is study examines the associations among bond risk measures (bond rating and risk premium) and R&D expenditures to determine whether their mean effect (expected future benefits) or their variance effect (risk) is more significant in pricing bonds. In general, from the perspective of bondholders, a negative correlation between bond risk parameter and R&D expenditures would indicate a stronger mean effect; that is, the expected future benefits of R&D expenditures are more than enough to compensate for the added risk of R&D. Conversely, a positive correlation would imply a stronger variance effect that swamps the mean effect of future benefits from R&D expenditures. The empirical results indicate follows: (1) For all samples, R&D expenditures are significantly positively associated with bond rating. The evidence suggests that, from the perspective of bondholders, the risk and uncertainties of R&D appear to dominate its expected future benefits. However, R&D expenditures have no significant effect on risk premium. (2) For electronic industry samples, the empirical results are the same with all samples. (3) For nonelectronic industry samples, R&D expenditures are significantly negatively associated with risk premium. The evidence suggests that, from the perspective of bondholders, the expected future benefits of R&D appear to dominate its risk. However R&D expenditures have no significant effect on bond rating. (4) The interpretation of this issue are not significant different through the expensing and capitalizing of R&D expenditures. (5) The industry effect is supported by the empirical results that show different effects of R&D on the bond risk measures between electronic industry and the nonelectronic industry. (6) The results indicate that it may be in favor of expensing R&D expenditures for electronic industry and capitalizing R&D expenditures for nonelectronic industry.