財務報表資料推估課稅所得之研究

An Estimation Model of Taxable Income

實證研究上經常須要以財務報表資料推估企業之課稅所得額，但國內外文獻上均顯示現行常用之推估課稅所得公式均存有偏誤。本文建立適合我國所得稅制之課稅所得推估公式，以上市櫃公司建檔之所得稅附註揭露資料與其申報之營利事業所得稅課稅所得額，測試本文建立課稅所得推估公式之適合度，實證結果顯示，本文所建立之課稅所得推估公式符合不偏估計式之檢定，並且在電子業及非電子業、研發支出比率高及研發支出比率低之樣本間，並無顯著之差異。本文建立之推估公式可作為未來研究以財務報表資料推估課稅所得額之參考。

The growing differences between financial income and taxable income have attracted many researchers’ attention. Recently, many studies aim to explore the implications of book-tax income differences, including addressing the relationship between book-tax income differences and earnings management, and examining the incremental information revealed by book-tax income differences. However, due to confidentiality of tax return data, most previous studies use financial statement data to estimate taxable income variable in the regression models. However, prior studies show that there exist nontrivial biases in using financial statement data to estimate taxable income. These potential biases not only place a restriction on accounting research that requires taxable income data but also cause results in prior studies inconclusive. Taiwan’s income tax system is less complicated than the U.S. system. Moreover, Taiwan’s listed companies disclose detailed tax information in financial statements, providing a research opportunity to overcome the limitation on the availability of taxable income data. Therefore, this paper constructs an estimation model for estimating taxable income by using financial statement data of Taiwan’s listed companies and their detailed disclosures on income tax expenses. This paper also uses tax return data of Taiwan’s listed companies to test whether the constructed estimation model is consistent with an unbiased estimator. Considering Taiwan’s income tax law, the estimation model of this paper, to reduce estimation errors in estimating taxable income, adjusts the potential biases in income tax expenses that reported in financial statements, including (1) a 10% of surtax on undistributed earnings, (2) investment tax credits, and (3) the amount of operating losses carried forward. The testing results show that the estimated taxable income and actual taxable income in tax return are not significantly different, and that the constructed estimation model is consistent with an unbiased estimator. Further the goodness of fit of the regression models is of no significant differences in applying the estimation model to the samples between electronic and nonelectric firms, and between firms with high- and low-R&D intensities. The results suggest that the constructed estimation model is applicable to companies within different industries that may differ in opportunities of obtaining investment tax credits (e.g., electronics vs. traditional industries). Likewise, the estimation model is also applicable to firms that differ in R&D intensities as well as R&D tax credits. Financial statements are publicly and periodically available data. This paper contributes to research on book-tax income differences and corporate tax research by providing an unbiased estimation method for researchers to estimate taxable income of Taiwan’s listed companies by using financial statement data.