財稅所得差異決定因素及課稅所得推估之研究

Determinants of Book-Tax Income Differences and Estimation of Taxable Income

本文探討我國上市櫃公司財稅所得差異之決定因素，並測試文獻上使用的推估課稅所得額計算方法，是否可以無偏誤地估計我國企業實際申報之課稅所得額。實證結果顯示，我國企業在適用稅法上會因為其企業特性造成不同之財稅所得差異情形而影響其租稅負擔，且財稅所得差異在次一期具有顯著的迴轉效果。再者，本文的迴歸測試結果也顯示，文獻上使用財務報表資料推估課稅所得額的六種方法，皆無法無偏誤地估計我國企業實際申報課稅所得額，其原因可能與我國租稅制度上的特性，如未分配盈餘加徵10％所得稅、投資抵減等因素有關，而影響推估結果之精確性。因此，我國相關文獻研究以財務報表資料推估企業的課稅所得額，可能會存有潛在偏誤之情形。

This study examines the determinants of book-tax income differences and tests the accuracy of six estimation methods of using financial data to estimate taxable income. The increasing trend in book-tax income differences has caused growing concerns in that the book-tax differences may be associated with firms’ deliberate tax planning and earnings management behaviors. Thus, the increasing book-tax differences may have a negative impact on tax equity and financial reporting quality. Yet, few studies have examined the causes of book-tax differences. Taking into consideration the features of Taiwan’s income tax system, we construct regression models to empirically examine the determinants of book-tax differences using listed and OTC companies as our sample. We use panel data methods to control for both firm individual effects and time effects, and also adjust the regression models to account for potential heteroskedasticity and autoregressive problems. Further, due to the confidentiality of tax return data, most previous studies used financial statement data to estimate the taxable income variable in the regression models. However, the potential biases of using financial data to estimate taxable income have not been addressed. The nontrivial differences between accounting standards and tax laws suggest that financial income may not be an unbiased estimate of taxable income. However, the accuracy of using financial data to estimate taxable income has not been examined in previous studies. Therefore, we use tax return data and construct regression models based on the six estimation methods used in previous literature to examine the potential bias in using financial statement data to estimate taxable income. Our analyses show that during our sample period of 2000 through 2003, the average of book-tax differences was positive and grew more divergent, suggesting there was a growing trend that firms tended to report greater financial income than taxable income. We also find several important factors associated with the book-tax differences, such as depreciation expenses, bad debt expenses, losses on declines in market value of inventories, investment income (losses), gains (losses) on disposal of investments, and gains (losses) on disposal of assets. The relations between those income/loss items and book-tax differences are consistent with the differences between Taiwan’s tax laws and financial accounting standards. Therefore, companies reporting greater amounts of those items on their financial statements are more likely to have greater book-tax differences. Further, firms with greater profitability tend to have greater book-tax differences, possibly due to their economies of scales in tax planning, and having greater propensities to tax-savings. Finally, ceteris paribus, the electronics industry, having enjoyed greater preferential tax treatments, also appears to have higher book-tax differences than other industries. Our analyses also indicate that the benefits of tax incentives are an important factor causing the book-tax differences and are unevenly distributed, to a great extent, among the electronics and traditional industries. The electronics industry alone captures about 95% of the total tax benefits. Our estimates show that during our sample period, the electronics industry had exempted from the tax payable of about NT$64.8 billion, while the traditional industries in aggregate had only about NT$3.6 billion. The results suggest that the current tax incentive system has been detrimental to tax equity. The results show that differences in firm characteristics may lead to different extents of book-tax differences, causing firms to bear tax payables that may not be parallel to their economic income. Finally, we find that book-tax income differences exhibit a significant reversing effect in the next year following their occurrence. For a dollar of current book-tax income difference, the reversing effect is about 0.1497 dollars in the following year. The magnitude of the reversal is considerable, suggesting it needs to take into account the reversing effect in evaluating the benefits of the tax planning on the book-tax differences. Further, the accuracy test results of using six estimation methods for taxable income indicate that all six methods based on financial statement data fail to qualify as a consistent and unbiased estimator for taxable income, possibly due to the features in Taiwan’s tax system. For example, the 10% surtax on undistributed earnings and investment tax credits may cause the overstatement and understatement of actual tax payables on current taxable income, respectively, and thus reduce the accuracy in estimation of taxable income using income tax expenses in financial reporting. We, however, find that the estimation models have greater explanatory power for firms with lower R&D intensity than those with higher R&D intensity, suggesting that tax incentive effect associated with R&D expenditures may be a factor causing noises in using financial statement data to estimate taxable income. Nevertheless, our results suggest caution on the potential bias in the findings of previous studies using financial statement data to estimate taxable income. Our paper is the first research exploring the effects of tax system and firm characteristics on causing book-tax differences and testing the accuracy of estimating taxable income based on financial income and other financial statement variables. The growing divergence in book-tax differences has undermined tax equity and earnings quality. Hence, our findings have implications for both tax reform and financial statement disclosure policies. The results suggest that a more neutral tax system is needed to prevent firms from taking advantage of current tax system and accounting standards to increase book-tax differences. Further, the limitations in using financial statement data to measure taxable income suggest that more detailed disclosure is needed to reduce the costs of information asymmetry in book-tax differences.