Persistent earnings, also called sustainability earnings, is earnings that can repeat in the future and grow (Penman, 2013, p.96). Earnings persistence is important for the stock valuation based on discounted cash flow model, especially residual earnings model. The importance of earnings persistence is reflected in the earnings-returns relationship, for example, Nichols and Wahlen (2004) find that abnormal returns for firms with high earnings persistence is higher than that for firms with low earnings persistence, especially for firms with positive earnings changes, Báez-díaz and Alam (2013) find that the stock returns fail to reflect correctly the earnings persistence attributable to the accrual components of earnings. Zeidi et al. (2014) suggest that earnings persistence should be considered in the evaluation of earnings response coefficients. Wang (2014) extends the study on the relationship between earnings persistence by comparing investors’ assessment of “real” earnings persistence with the persistence of smoothed earnings. Wang (2014) finds that investors correctly assess the “real” earnings persistence and smoothed earnings persistence. There was a research that provided new insight into detecting earnings management which is related to the investment decision of companies listed on the Indonesian Stock Exchange (Rahmawati et al., 2015). Earnings persistence and earnings management can be used as proxies of earnings quality.

This study examines two factors that are predicted to have impacts on earnings persistence, namely accruals and deferred tax. High accruals, either positive or negative, contain more transitory factors compared to low accruals. Previous studies suggest that absolute value of accruals have the negative effect on the persistence of earnings (e.g. Sloan 1996; Dechow and Gee 2006, Blaylock et al. 2012). The tax information in financial statements is related to earnings persistence (Lev and Nissim, 2004). Thus, deferred tax is predicted to affect the persistence of earnings. In Indonesia regulation of deferred tax is contained in the Indonesian Statement of Financial Accounting Standard No. 46 (ISFAS 46) on Income Taxes.

The value relevance of deferred tax or related variables such as book-tax differences has been documented in previous studies. Phillips et al. (2003), for example, find that deferred tax can be used to detect earnings management, specifically in detecting earnings management to avoid an earnings decline and to avoid a loss, because managers may use decretion to manage earnings, and the earnings management affect the book-tax differences and deferred tax. Managed earnings associated with high deferred tax is predicted to be low persistent because it is based on transitory factors.

This study contributes to the literature in the following ways. First, this study provides empirical evidence on the effects of the deferred tax on the persistence of earnings from an emerging market, Indonesia Stock Exchange (IDX). Second, this study also provides empirical evidence on the effects of accruals on the persistence of earnings.

Deferred tax reported in the income statement is caused by temporary differences between income tax expense and income tax payable. Temporary differences are differences between the book value of an asset or liability on the balance sheet and its tax base. The temporary differences result in the unsustainability of earnings because the earnings contain transitory items which do not repeat in the future. Phillips et al. (2003) find that deferred tax expense is related with earnings management, specifically they find that deferred tax is useful in detecting earnings management to avoid an earnings decline and to avoid a loss. Earnings management indicates lower quality of earnings. Results of studies on book-tax differences (e.g. Hanlon 2005 and Blaylock et al. 2012) show that large positive or negative book-tax differences are negatively associated with earnings persistence. Thus, the absolute deferred tax is predicted to have a negative effect on the persistence of earnings.

H1.Thelarger(smaller)theabsolutevalueofdeferredtax,thelower(higher)thepersistenceofearnings.

Accruals is a noncash value flow recorded in the financial statements (Penman, 2013:130). Accruals result in the difference between net income and cash from operations in the cash flow statement. Accruals contain temporarily factors or transitory component of earnings transitory that is not sustainable. Previous studies suggest that high absolute value of accruals have a negative impact on the persistence of earnings (e.g. Sloan 1996; Dechow and Gee 2006; Blaylock et al. 2012). The relationship between accruals and earnings persistence can be explained using signaling theory developed by Spence (1973) in the context of job market signaling. Following Spence, signalling theory is used in accounting and or finance studies using various signals such as taxable stock dividends (Kuo, 2013), CSR reports (Thorne et al., 2014), IPO underpricing (Wu, 2014), ownership structure (Yeh, 2014), dividend payout ratios (Liljeblom et al., 2015). In the financial reporting context, accruals can be used as a signal of earnings quality which can be proxied by earnings persistence. Accruals are disclosed in the financial statements to reduce information asymmetry and thus to increase the usefulness of financial statements. The relationship of accruals with earnings persistence is hypothesized as follows:

H2. The larger (smaller) the absolute value of accruals, the lower (higher) the persistence of earnings.

Earnings t+1 = α + β1 Earnings t0 + β2 AbsDeferredTax t0 + β3 Earnings t0 X AbsDeferredTax t0 + β4 AbsAccruals t0 + β5 Earnings t0 X AbsAccruals t0 + ε (1)

Earnings t+1 = α + β1 Earnings t0 + β2 AbsDeferredTax t0 + β3 Earnings t0 X AbsDeferredTax t0 + β4 AbsAccruals t0 + β5 Earnings t0 X AbsAccruals t0 + β6 ∑Industry Dummy Variables t0 + β7 ∑Year Dummy Variables t0 + ε (2)

where Earnings t+1 is one-year-ahead of net income scaled by average total assets; Earnings t0 is current net income scaled by average total assets; AbsDeferredTax is the absolute value of deferred tax reported in the income statement scaled by average total assets; Accruals t0 is current net income minus current net operating cash flow scaled by average total assets; Industry Dummy Variables are the nonfinancial industry sectors or industry sectors 1-7 and sector 9 (all sectors except sector 8 which is the financial industry sector) and Year Dummy Variables is the dummy variables for the years 2008-2013.

This study uses a measure of earnings persistence following previous studies such as Sloan (1996), Hanlon (2005), and Blaylock et al. (2012). The measure of absolute accruals follows Blaylock et al. (2012), and the use of absolute deferred tax is similar to Ayers et al. (2010) that use the absolute value of negative changes in book-tax differences and Comprix et al. (2011) that use absolute values of book-tax differences. This study includes year dummy variables similar to previous studies, e.g. Baginski et al. (1999), Doukakis (2010), Habbash,

M. (2011), Habib et al. (2013), Kaczmarek et al. (2014), Lin and Wang (2014), and Lanis and Richardson (2015). The year dummy variables to control macro variables that may affect one-year-ahead earnings. Industry dummy variables are used to control the differences in earnings resulted from the different characteristics of industries. Previous studies use specific industry because of its unique characteristics (e.g. Eckels et al. 2011; Liu and Yu 2013; He and Yang 2014; Kanagaretnam et al. 2015; Nijam, H. M. 2016).

To test hypothesis 1 (H1), the coefficient β3 in the equations (1) and (2) are expected to be negative and significant. To test the hypotheses 2 (H2), the coefficient β5 in the equations (1) and (2) are also expected to be negative and significant.

The sample frame used in this study are tables of Financial Data and Ratios in the Fact Books 2008-2014 (The fact books were accessed from www.idx.co.id/en-us/home/publication/factbook.aspx). Financial data from 2007 to 2014 are used to measure the variables, and financial data from 2008 to 2013 are used to test the hypothesis (Financial statements were accessed from http://www.idx.co.id/en-us/home/listedcompanies/financialannualreport.aspx). The initial sample composed of 2,403 firm-year observations, and using a number of criteria for sample selection the final sample is composed of 1,609 observations. The sample selection procedure is presented in Table 1.

Table 2 presents the descriptive statistics. The table shows that one-year-ahead earnings (Earnings t+1) tend to be larger than current earnings (Earnings t0). This may be an early indicator of the existence of earnings persistence, that is, the positive association between current earnings and one-year-ahead earnings. Accruals tend to be negative ranging from - 0.930 to 0.740 with the mean value of -0.0148, whereas deferred tax has the mean value 0.0004 that is close to zero ranging from -0.229 to 0.222.

Regression results of the effect of the deferred tax on earnings persistence are also presented in Table 3. The coefficient of the interaction between current earnings (Earnings t0) and the absolute value of deferred tax (AbsDeferredTax t0) is negative and significant at p<0.01. This result supports the hypothesis 1 (H1) that the larger (smaller) the deferred tax, the lower (higher) the persistence of earnings, after controlling for the interaction between current earnings and the absolute value of current accruals (Table 3 Model 1). Table 3 Model 2 shows consistent regression results after controlling for Industry Sector Dummy Variables and Year Dummy Variables. Thus, if a firm has a large deferred tax, then its persistence of earnings is low, and if a firm has a small deferred tax then its persistence of earnings is high.

The regression results presented in Tabel 3 Model 2 also show that the coefficients of the Year 2012 and Year 2013 are negative and significant at p<0.05 indicating that one-year- ahead earnings (Earnings t+1) of the year 2012-2013 are lower than the one-year-ahead earnings 2008 (the excluded year dummy variable). The coefficient of Sector 2 and Sector 5 are positive and significant at p<0.05 indicating that one-year-ahead earnings (Earnings t+1) of Sector 2 and Sector 5 are higher than the one-year-ahead earnings Sector 1 (the excluded sector dummy variable). The inclusion of these dummy variables make the adjusted R-Square increases from 0.45 to 0.463, while the interaction variables remain negative significant.

Table 3 presents the regression results. The coefficient of the interaction between current earnings (Earnings t0) and the absolute value of current accruals (AbsAccruals t0) is negative, significant at p<0.01. This result supports the hypothesis 1 (H1) that the larger (smaller) the accruals, the lower (higher) the persistence of earnings after controlling for the interaction between current earnings and the absolute value of deferred tax (Table 3 Model 1). After controlling Year Dummy Variables and Industry Sector Dummy Variables (Table 3 Model 2), the results remain unchanged. Thus, if a firm has a large accrual then its persistence of earnings is low, and if a firm has a small accrual, then its persistence of earnings is high.

Table 4 Panel A presents regression results separately for positive and negative accruals and Panel B for positive and negative deferred tax. Báez-díaz and Alam (2013) use discretionary and nondiscretionary accrual ranking, whereas Jeter et al. (2008) analyze positive and negative deferred tax separately. Table 4 Panel A shows that the effects of accruals on earnings persistence are consistent for both positive and negative accrual subsamples. The coefficients of the interaction between current earnings and the absolute value of accruals (Earning t0 X AbsAccruals t0) are negative and significant at p<0.01, indicating that larger accruals, either positive or negative, have a negative effect on the persistence of earnings. However, the impact of the deferred tax on earnings persistence for positive accruals subsample and for negative accruals subsample is inconsistent. The coefficient of the interaction between current earnings and absolute value deferred tax (Earnings t0 X AbsDeferredTax t0) is negative and significant at p<0.05 for the positive accrual subsample. The coefficient of the interaction is also negative but insignificant for the negative accrual subsample. Thus, when the deferred tax is positive, the larger (smaller) the deferred tax the lower (higher) the persistence of earnings.

Table 4 Panel B shows regression results for positive and negative deferred tax subsamples. The coefficient of the interaction between current earnings and the absolute value of deferred tax (Earnings t0 X AbsDeferredTax t0) is negative, significant at p<0.01 for the positive deferred tax subsample. However, the coefficient of the interaction (Earning t0 X AbsDeferredTax t0) for the negative deferred tax subsample is also negative but insignificant. Thus, the deferred tax has the negative effect on the persistence of earnings only for positive deferred tax subsample. When the deferred tax is positive, the larger (smaller) the deferred tax the lower (higher) the earnings persistence, but when the deferred tax is negative, the deferred tax has no effect on the earnings persistence. Table 4 Panel B also shows that the effects of accruals on earnings persistence are also consistent for both positive and negative deferred tax subsamples, the coefficients of the interaction between current earnings (Earnings t0 X AbsAccruals t0) are both negative and significant at p<0.01.

The findings of this study show that deferred taxes have a negative impact on earnings persistence. Deferred tax is the difference between income before tax ("accounting income") and taxable income ("tax income"). If "accounting income" is greater than "tax income", then "deferred tax expense" presented in the income statement and "deferred tax liabilities" presented in the statement of financial position” will be generated. Conversely, if "accounting income" is less than "tax income", then "deferred tax benefit" presented in the income statement and "deferred tax assets" presented in the statement of financial position will result. Deferred tax in this study is the deferred tax expense (benefit) presented in the statement of income. ISFAS46 shows that this tax difference occurs because the tax expense for tax purposes is determined based on the tax regulations i.e. Law Number 36 Year 2008 regarding Income Tax and its implementation rules, while the tax expense under accounting rules is reported using the basis of accounting standards. The different approaches in determining these tax expense result in the differences in the amount of corporate tax expense and should be reported by referring to accounting standards relating to the reporting of tax expense (SFAS 46 concerning Income Tax Accounting). The difference in tax expense can be fixed or temporary. This study focuses on temporary tax differences since these differences are more likely to be utilized for the benefit of profit management in the income statement. As an example of this temporary difference in tax expense is the difference in the recognition of the residual value or the determination of the useful life to calculate depreciation expense. Such differences do not cause a permanent difference but are only temporary because they are due to differences in the determination of the time period and residual value in the determination of depreciation, and at a certain point, the cost of taxes will be generated in the same amount. Differences in transactions that result in deferred taxes are generally due to the recognition of "accrual-based" income versus "cash-based". Examples of the use of "accrual basis" for accounting and "cash basis" for taxation include installment sales, warranty expense, unearned income, research and development costs, while examples of the use of "cash basis" for accounting and "accrual basis "for taxation, among others, the cost of establishment and profit of long-term contracts. From the differences in accounting income and fiscal income can be identified the main nature of the difference, namely that the determination of accounting income that tends to contains discretion (principle-based) while the determination of fiscal income tends to be more rule-based. Discretion has both a positive and a negative side. Discretion has a positive side because discretion is required by managers in performing the task of managing the company for efficient contracting purposes. On the other hand, discretion may result in the use of such discretion resulting in information reduction information for users, especially investors. For example, the discretion can be used to manage the earnings that in turn result in less persistent earnings. In the process of analyzing financial statements for securities valuation, earnings persistence is useful for investors in forecasting earnings in stock valuation process for investment decision making. The empirical results of this study indicate that earnings persistence tends to be lower when the deferred tax is high.

The results of this study also indicate that the greater the accrual, either positive or negative, the lower the earnings persistence, and vice versa. Accruals in a period tend to reverse in the next period. Hence, the greater accruals in a period lead to large changes in the opposite direction in the next period. This affects the persistence of earnings to decline. In contrast, low accruals in a period tend to be followed by low accrual changes in the next period, and the impact is that the earnings persistence tends to be high. For example, a significant increase in trade receivables over a period compared to the previous period could be due to relatively large non-cash sales during the period, compared to the previous period, which has not been collected at the end of the period. This increase in accruals has a positive impact on earnings in the period. Further, normal non-cash sales in subsequent periods, especially at the end of the period, tend to decrease trade receivables compared to current accounts receivable which may be due to the collectibility of current accounts receivable at the beginning of the current period. The decrease in trade receivables has a negative impact on earnings. The accrual process in these two periods can explain that the relatively large accruals in a period have an impact on the decrease in earnings persistence. In contrast, low accruals in a period tend not to increase or decrease accruals in the next period, ceteris paribus, due to the nature of accruals that have reversed changes in subsequent periods. In turn, low accruals changes have an impact on "normal" earnings, i.e. earnings that are not affected by the accruals. Therefore, when accruals in the current period are low then current earnings tend to continue, i.e. earnings are more persistent.

This study aims to examine whether accruals and deferred taxes have an effect on earnings persistence. Results of this study showed the following findings. First, results of this study supported the hypothesis that the deferred tax had a negative impact on the earnings persistence. Companies with large deferred taxes tended to have low earnings persistence, and vice versa, although these results apply only to positive deferred taxes. Deferred tax is the temporary difference between tax expense based on accounting income before tax and tax expense based on taxable income. Taxable income is based on rules of taxation (rule- based), while pre-tax accounting income is determined by accounting standards (tend to be principle-based) that gives accounting method choices to firms in the preparation of financial statements. The choice of methods provided by accounting standards is required by firms because it is useful in carrying out efficient contracting. However, such flexibility in the selection of accounting methods also has weaknesses in the form of reducing the quality of earnings information, either whether the decline in the quality of profit information due to earnings management or whether the decline in the quality of profit information due to business conditions that impact on the company's activities. Earnings persistence can be used as a proxy of earnings quality. The findings of this study showed that there is a negative association between deferred taxes and earnings persistence.

Second, accruals negatively affected earnings persistence. Companies with high accruals tended to have low earnings persistence. On the other hand, firms with low accruals tended to have high earnings persistence. These results applied both to positive and negative accruals. That is, the greater the positive accruals or the greater the absolute negative accruals of a firm, the lower the firm's earnings persistence. Accruals in this study were total accruals that do not separate discretionary accruals and non-discretionary accruals. This means that accruals either due to the management’s discretion or accruals occurring due to business conditions (non-discretionary accruals) had an impact on the persistence of earnings. The difference between the two types of accruals was that the discretionary accruals performed by management, in general, are for managing earnings, whereas non- discretionary accruals occurred due to business conditions e.g. in the period in which accruals occurred there was a very significant increase in sales caused by a significant increase in market demand. Accruals were generally reversed because accruals for earnings management were usually only done at a certain period for a particular purpose, and an unusual increase of market demand was not occurring at any period but occurs in a certain period only. Therefore, current earnings that were affected by high accruals did not continue into the next period when the next period became a normal condition. This means that current period earnings had low persistence.

The implications of this study are as follows. This study contributes the literature on the relationship of accruals and deferred tax with earnings persistence using data from an emerging market, Indonesia Stock Exchange (IDX). Firms with large (small) accruals have low (high) persistence of earnings, and there is a negative relationship between deferred tax and the persistence of earnings but only when the deferred tax is positive. The findings of this study also provide empirical evidence that supports the usefulness of accounting standards regarding the reporting of accrual-based accounting income and deferred tax reporting in the financial statements. The practical implication of this study is that the magnitude of accruals and/or deferred taxes can be used as consideration in making investment decisions based on fundamental analysis. Its use is in the forecasting process when the accrual and/or deferred tax is large the forecasting of future earnings needs to be adjusted because the earnings that contain the large accruals and/or the large deferred taxes have an impact on the low earnings persistence.

This study has some limitations such that sample selection uses the non-probabilistic sampling method commonly used in market-based financial accounting research. The criteria used in the sampling are that the companies selected for the sample are companies in the non-financial industry sector, having the same financial reporting date as of December 31, and using the same currency (Indonesian Rupiah) in financial statements, and this study does not include companies with outlier data. These criteria are needed to control the impact of factors in the criteria on the results of the study but with the consequence of limiting the generalization of the results of the study. Future research can be done to test the external validity of the results of this study by using sample firms with characteristics different from the characteristics of this study such as different characteristics in the industrial sector, date of financial statements, observation year, and financial reporting currency. Future research can also be a further study by empirically testing on the pragmatic side of accounting theory, i.e. testing whether information about accruals and deferred taxes and its impact on earnings persistence is actually used by investors, which is an issue beyond the scope of this study.