The Effect of Book-Tax Differences, and Executive Compensation on Earnings Persistence with Real Earnings Management as Moderating Variable

This study analysed the effect of book-tax differences, and executive compensation for earnings persistence with real earnings management as moderating variables. The panel data is 192 samples from the Banking sector which were listed on the Indonesia Stock Exchange in 2014-2017. Use multi regression moderation analysis. The results show that current tax, permanent differences, and temporary differences have a significant positive effect on earnings persistence, executive compensation has a significant negative effect with earnings persistence, and cash flows operations have a positive and insignificant effect on earnings persistence. The results of the moderating variable test, real earnings management weaken the influence of current tax and permanent differences on earnings persistence. The effect of cash flows operations, temporary differences and compensation on earnings persistence is strengthened by real earnings management.


2.2.3
The effect of permanent differences on earnings persistence Martinez and de Souza (2012), permanent differences occur because of differences in recognition between accounting rules and tax regulations related to certain income and expenditure. Permanent differences arise when certain income and expenses recognized by accounting principles, but are not based on tax regulations. Permanent differences result in a permanent difference between accounting earnings and taxable income. Permanent differences usually arise because tax laws must that some transactions not be included calculation of taxable income. According to the Income Tax Act, some items which constitute a permanent difference are final income tax (Article 4 paragraph 2), non-taxable objects (Article 4 paragraph 3), and costs that are not allowed gross income (Article 9). paragraph 1) (Waluyo, 2016). Hypothesis 3: Permanent differences have a positive effect on earnings persistence 2.2.4 The effect of temporary differences on the persistence of earnings Temporary differences occur when accounting rules and tax laws recognize the same number of transactions, however, differing with respect to the time of recognition. A transaction recognized by accounting principles, but not by tax regulations; or vice versa (Sonnier et al., 2012). This difference is temporary because it identified in the next accounting period. In other words, all transactions are actually recognized by the accounting system and taxes but they differ with regard to the time of their allocation. Temporary differences are usually caused by differences in the methods used by accounting rules and tax rules, in terms of accrual and realization, depreciation and amortization, inventory valuation, and calculation of compensation for losses (Noor et al., 2009). Hypothesis 4: Temporary differences negatively affect on earnings persistence 2.2.5 The effect of executive compensation on earnings persistence The managerial view of power states that CEO compensation and other corporate governance practices reflect the exercise of managerial power and behavior more than providing efficient incentives. Bonus payments all BOD regularly halt as a punishment when the company's execution break down or when a outrage including company administration uncovered. Compensation is also useful for maintaining a competent workforce in managing the company (Anthony and Govindarajan 2007). Balsam (1998) said that company supervisors utilize profit administration as it were when benefits surpass costs. Hypothesis 5: Executive compensation has a negative effect on earnings persistence 2.2.6 The effect REM as a moderating variable between of current earnings on earnings persistence. The way an agent can do to influence accounting numbers profit engineering or earnings management in financial statements. Current tax burden and large deferred tax burden will reduce the level of profits obtained by a company, and vice versa, the current tax burden and small deferred tax burden will increase the level of profits obtained by a company. Income consists of accrual and cash flows. Juliati and Tjaraka's research (2014) proves that the tax burden now has ability to detect REM also finds a weak result that there is an earnings management action on the company consideration of the income tax burden. Hypothesis 6: REM weakens the effect of current tax profits on earnings persistence 2.2.7 The effect REM as a moderating variable between cash flows operating on earnings persistence Dechow and Ge (2005), mentions operating cash flows related to company processes in generating operating profits from aspects of cash transactions. Cash income and expenditure will affect the company's profit figures such as when cash sales or payment of expenses that go directly out of cash. Evaluation of earnings and cash flow is sensitive to the economic characteristics of the company (Watson and Wells, 2005). Profits compiled on accrual basis contain elements of manager's interest in the reporting so that information on operating cash flows needed as one of the considerations in predicting the company's performance in the future. Fairfield et al. (2003), using operating cash flow as one of the predictive components of the company's future operating profit. Hypothesis 7: REM strengthens the effect of operating cash flows on earnings persistence 2.2.8 The effects REM as a moderating variable between permanent differences on earnings persistence The difference between book and taxable income may not be informative about earnings management or crosssectional variation in generating earnings persistence. On the other hand, with tax conformation the second book of financial accounting and tax systems is separated (Goncharov and Zimmermann, 2006). Hypothesis 8: REM weakens the effect of permanent differences on earnings persistence 2.2.9 The effect REM as a moderating variable between temporary differences on earnings persistence Book-tax differences generated by tax planning strategies. Book-tax conformity is consistent with the general system for accounting and taxation purposes, where financial accounting is directly used to calculate taxable profits. The difference in book-tax classified as temporary differences (Hanlon and Shevlin, 2008). Temporary differences defined as items included in book income or tax income but which included in both according to the time of use. Hypothesis 9: REM strengthens the effect of temporary differences on earnings persistence 2.2.10 The effect REM as a moderating variable between of executive compensation on earnings persistence Balsam (1998), that shareholders tend to remunerate supervisors who utilize optional discretionary accrual  Vol.10, No.5, 2019 57 smoothing. Smoothing anticipated to diminish detailed instability in net pay and offer assistance company directors to defeat income targets in a push. Hence, the relationship between official rewards and optional collection is moderately higher when company supervisors utilize optional collection for salary smoothing. This appears a shifted relationship between optional gathering and official rewards. Hypothesis 10: REM strengthens the effect of executive compensation with earnings persistence

Research Methodology 3.1 Population dan Sample
Population uses secondary financial statement data from public banking companies listed on the Indonesia Stock Exchange. This study uses non-random sampling technique, namely purposive sampling. The total banking 48 companies, the observation period from 2014-2017 was 4 years so that 192 observations obtained.

Variable Measurement
Pre-tax book income (PTBI t ) is pre-tax income as measured by TA (Sundvik, 2017;Dridi dan Adel, 2016;Hanlon, 2005). Operating cash flow (CFO) referred to as the amount of cash arising from a company's operations on income, expenses, income and expenses. CFO describes how companies earn income and convert it into cash (Conyon and He, 2012;Carter and Hillegeist, 2010;Penman, 2001). Permanent differences (PERM) obtained from accounting earnings and taxable income. Permanent differences arise because tax laws must that some transactions not be included calculation of taxable income (Sundvik, 2017;Waluyo, 2016;Martinez et. Al, 2016;Martinez and de Souza, 2012). Temporary differences (TEMP) obtained due to differences in the rules of both accounting rules and tax laws recognizing the same number of transactions, however, they differ with respect to time of recognition. A transaction recognized by accounting rules, but not by tax regulations; or vice versa. For example, accruals and realization, depreciation and amortization, inventory valuation, and compensation calculations (Sundvik, 2017;Waluyo, 2016;Martinez et.al, 2016;Jackson, 2009). Executive compensation (COMP) is Natural log compensation (basic salary, cash bonus, benefits, etc.) Carter and Hillegeist (2010), Conyon and He (2012).
Dependent variable, earnings persistence (PTBIt + 1) is taxable income divided by average TA. Previous research Blaylock et al., 2012;Hanlon, 2005). Moderating variables, REM are measured by the number of abnormal provisions for loans / or loss of assets, abnormal cash flows and abnormal discretionary expenditure equations. REM banks are measured by the size of Robb (1998) combined with measurements of Roychowdhury (2006), as used by Chou & Chan (2018) ,t and the last one DISEXPi.t/ TAi. t-1 = 1/ TAi. t-1 + REVi. t-1 / TAi. t-1 , LLPi, t and LLPi, t-1 are bank provisions for loans / or asset losses against TA in year t and t-1, respectively; WOi, t and WOi, t +1 is Net Charge-Off bank i to TA in the year t and t + 1, respectively; , is the TA of the bank in the year t. Estimated error term i, t are unexpected provisions for loans / or asset losses, i.e. abnormal provisions for loans / or loss of assets for i bank year t.
, is cash flow from bank operations i year t; , − 1 is TA bank i year t-1; , is total income from the bank i during the year t; ∆ , is the change in income from banks in the year t; estimate yi, t is an estimation error term, used as a measure of abnormal cash flow for i bank.
the bank is discretionary spending i year t, which is defined as the amount of advertising costs, and sales, general and administrative costs; estimate i, t is a term estimate of error, that is abnormal discretionary costs for i bank.
The control variable consists of: company size (SIZE) which is measured by the natural logarithm of TA (Hung et.al, 2018;Liu and Schneible, 2017;Koubaa and Jarboui, 2017), LEVERAGE measured by Total liabilities / TA (Hung et.al, 2018;Koubaa and Jarboui, 2017), Return On Assets (ROA) obtained from profit before tax / average TA (Hung et.al, 2018;Koubaa and Jarboui, 2017), SALES obtained from current year's sales / average TA (Hung et.al, 2018), effective tax rate at this time (CETR) obtained from current tax income / 58 pre-tax profit (Zhou, 2016, Xian et.al, 2016Hanlon, 2005). Table 1 show that the Pretax Book Income in the next period t + 1 (PTBIt + 1) has an average of 0.005. With positive earnings persistence, it important role in assessing company performance. The highest PTBIt + 1 (maximum) is 0.109 and the lowest (minimum) is 0,000 and the std.dev is 0.010. The minimum value of Pretax Book Income in the current period (PTBIt) is 0,000, while the average value is 0.019 or 1.9%. The std.dev value is 0.018, which indicates variations in the homogeneous PTBIt. The average PTBIt value is greater than the std.dev value which indicates that PTBIt + 1 gives a large response from the profit information generated at PTBIt.

Results and Discussion 4.1 Descriptive Statistics
The average operating cash flow is smaller than the std.dev, which indicates that the company uses operating cash flows with an average value of 0.012 to improve the following year's pre-tax profit. Executive compensation (COMP) shows the minimum value of 6.486 stated by the company. The std.dev value of 1.574 which indicates the type of compensation executive is homogeneous. The average value of executive compensation is 10.283 with a std.dev value lower than the average value so that the average value used as a benchmark for presentation in the entire data.
The average REM is higher than the std.dev value, indicating that the company uses REM with an average value of 7.2% of the following year's profit before tax, while the average permanent difference value (PERM) and temporary difference (TEMP) is smaller than the std.dev value, indicating that the company uses permanent and temporary differences with an average value of 1% and less than 1% of the following year's pre-tax profit.
The minimum return on assets (ROA) value of -11.50 recorded by the company, the average return on assets value is 1.011 with a std.dev value of 2.407 which indicates that the type of return on assets is homogeneous. The average value of return on assets is greater than the value of the std.dev of the company, which indicates that the company responds to tax returns the next year responds more to information on earnings return on assets. Sales (SALES) sets a minimum value of 0.022 while the average value is 0.097. The average value of earnings for the following year is 0.005, which means that the company has the following year's pre-tax profit of 0.5% on the TA. The average value of current tax effectiveness (CETR) is 0.250 so that the company uses that effectiveness to control the company's activities and is also implied by the average bank company to see the following year's pre-tax profit. The average value of company size (SIZE) is 17,031, the next year's average tax profit value is 0.01, which means that the company's operating activities range from 1% of the TA. The average leverage value (LEV) is greater than the std.dev, so the implication is that the leverage level is around 85% to provide information to stakeholders in their business strategy management.

Correlation Matrix
The correlation matrix shown in table 2 reveals current pre-tax profit (PTBIt), operating cash flow (CFO), fixed difference (PERM), temporary difference (TEMP) and next year's pre-tax profit (PTBIt + 1) significant correlation . This correlation shows that the higher component of the difference in tax and accounting income the higher the taxable profit the following year or vice versa. In conclusion, a significant correlation between variables is complementary.

Result
From table 3, the results of the classic assumption of residual normality test are: fallow finger value of 0.313 with p value of 0.054 where> 0.05 so that means the residual is normally distributed. The coefficient of determination on Adjusted R-squared is 0.717, which means 71.7% can predict the effect of independent variables on the dependent variable and 28.3% is explained by other variables outside the tested variable. Based on the partial regression test presented in table 3, tax income now has a significance value of 0,000 <0,005 and a regression coefficient of 0,996, which means that tax returns are now positive and significant at levels below 1%. So that hypothesis 1 is accepted, which indicates that tax profits now have an influence on earnings persistence. Operating cash flow has a significance value of 0.406> 0.005 and a regression coefficient of 0.001, which means that operating cash flows are positive and not significant. So that hypothesis 2 is rejected, which indicates that the operating cash flow has no effect on earnings persistence. Permanent differences have a significance value of 0,000 <0,005 and a regression coefficient of 1,020, which means that positive permanent differences are significant at levels below 1%. So that hypothesis 3 is accepted, which indicates that permanent differences have an influence on earnings persistence. Temporary differences have a significance value of 0,000 <0,005 and a regression coefficient of 0,983, which means that positive and significant temporary differences. So that hypothesis 4 is rejected, which indicates that operating cash flows have the opposite effect on earnings persistence. Executive compensation has a significance value of 0.029 <0.005 and a regression coefficient of -0.001, which means that executive compensation is negative and significant at levels below 5%. So that hypothesis 5 is accepted, which indicates that executive compensation has a negative influence on earnings persistence. Table 3. The Effect of Book-Tax Dif'ferences, and Executive Compensation on Earnings Persistences with REM as Moderating Variable PTBIt+1 = α + β1PTBIt + β2CFO + β3PERM + β4TEMP + β5COMP + β6REM β7PTBIt*REM+ β8CFO*REM + β9PERM*REM+ β10TEMP*REM + β11COMP*REM + β12ROA + β13SALES + β14CETR + β15SIZE + β16LEV + ε Variable Predictions Coefficient Prob. The test results of the REM moderation variable between current tax income and earnings persistence obtained a value of 0.996 and after being moderated to 0.803. This shows that REM weakens the influence of current taxes with the persistence of earnings. The significant level produced is 0.183> 0.05 after being moderated so the results are not significant and hypothesis 6 is rejected, the conclusion is that the moderating value weakens the effect of current tax on earnings persistence. The effect of operating cash flow on earnings persistence which is moderated by REM the influence of operating cash flows with earnings persistence. The significant level produced is 0.481> 0.05 after being moderated so the results are not significant and hypothesis 7 is rejected, the conclusion is that the moderating value strengthens the influence of current tax on earnings persistence. The effect of permanent difference on the persistence of earnings which is moderated by REM obtained a value of 1.020 and after being moderated to -7.782. This shows that REM weakens the influence of permanent differences with persistence of earnings. The significant level produced is 0,000 <0,05 after being moderated so that the results are significant and hypothesis 8 is rejected, the conclusion is that the moderating value weakens the effect of current tax on earnings persistence.
The effect of temporary differences on earnings persistence which is moderated by REM is -0.001 and after being moderated to -2,951. This shows that REM strengthens the influence of temporary differences with persistence of earnings. The significant level produced is 0,000 <0,05 after being moderated so that the results are significant and hypothesis 9 is accepted, the conclusion is that the moderating value strengthens the effect of current tax on earnings persistence. The effect of executive compensation on earnings persistence which is moderated by REM is the value of -0.081 and after being moderated to -0.008. This shows that REM weakens the influence of executive compensation with earnings persistence. The significant level produced is 0.114> 0.05 after being moderated so the results are not significant and hypothesis 10 is rejected, the conclusion is that the moderating value strengthens the influence of current tax on earnings persistence.

Discussion
The results of the first hypothesis are consistent with previous researchers by Dridi and Adel (2016) which state that tax profits now have an influence on next tax profits (persistence of earnings) because the market still considers the profits obtained can now be used as predictions for earnings in the next period considered financial information is most easily obtained without looking at all components of future earnings predictions. The results of the second hypothesis test results are not consistent with previous researchers by Jackson (2015) where the results of this test obtained have a positive but not significant effect, perhaps because the company is not consistent in classifying operating cash flows that have an inconsistency in the data presented, operating cash flows also useful for users of financial statements so that if part of forming the following year's profit has high validity then the future earnings prediction results will be directly proportional.
The results of the fourth hypothesis test are inconsistent with the research by Waluyo (2016) finding the results of the temporary difference test are negative so that it is different from the results of this study which has a positive and significant influence, this might occur because of the many perceptions of how to imply temporary difference components to carry out earnings management so that the difference between the differences obtained from accounting and tax differences varies. The test results of the REM moderation variable of five hypothesis testing, only the ninth hypothesis where temporary differences in earnings persistence reinforced by REM. This is consistent with Jackson's (2015) study, where temporary difference components are still an area that is often used as a way to make earnings management so that application of real-based earnings management will reduce the practice of earnings management a little. Of the five control variables tested, only two showed an influence on earnings persistence, namely Return On Assets (ROA) and firm size (SIZE) while sales (SALES), Current Effective Tax Rate (CETR) and Leverage (LEV) did not produce the effect significant to the next tax return.

Conclusions, Limitations and Suggestions 5.1 Conclusions
The results of the first hypothesis test show that tax profits now influence on earnings persistence. The results of this study are consistent with Dridi and Adel (2016). The results of the second hypothesis test show that operating cash flow has a positive impact that is not significant to earnings persistences. The results of this study are inconsistent with Jackson (2015). The third hypothesis test results show that permanent differences have a significant positive influence on earnings persistence consistent with Waluyo (2016) and Marinez and Souza's (2012) research. The results of testing the fourth hypothesis show that temporary differences have a positive and significant effect, these results are not consistent with the research by Waluyo (2016). The results of the fifth hypothesis test show that executive compensation has a significant negative effect. This result is consistent with research by Neokleous (2015) and Auwor (2012).
The results of the REM hypothesis test as a moderating variable weaken the effect of current tax profits, permanent differences in earnings persistence, while strengthening operating cash flows, temporary differences and executive compensation against earnings persistence. But from the results of existence of moderating variables, the results that are after the hypothesis are the ninth hypothesis where temporary differences in earnings persistence strengthened by REM. This is consistent with Jackson's (2015) study. Control variables that show significant effects on earnings persistence are ROA and SIZE, while SALES, CETR and LEV are not significant.

Limitations and Suggestions
This research only uses banking data, so that it more trusted and generalized, for the next researcher the researcher recommends taking data not only in the banking sector so that the two can compare the results of the