The association between discretionary accruals and future financial performance

(1)

The Association between Discretionary

Accruals and Future Financial

Performance

Name: Salihah Ahmed Student number: 10750258

Thesis supervisor: dr. S.W. Bissessur Date: June 25, 2018

Word count: 8000

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by student Salihah Ahmed who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

This study examines the relation between discretionary accruals and future financial performance. Prior literature suggests that accruals adjust the recognition of cash flows over time and thereby improve the ability of earnings to reflect firm performance. This happens when credible signals are given to users of financial statements. A credible signal is given when private information is revealed about the firm’s performance. However, it is also known that managers can exercise judgment in reporting accruals and thereby mislead the users of financial statements. This occurs when managers use their discretion to opportunistically manage earnings, also known as earnings management. According to prior literature discretionary accruals are used as a proxy to measure earnings management. Therefore, in this study discretionary accruals are estimated using the modified-Jones (1991) model. It is expected that the association between discretionary accruals and future financial performance will be higher for firms in a bad state. It is also expected that the association between discretionary accruals and future financial performance will be higher for firms having a long operating cycle. The findings indicate that a bad state of a firm is negatively associated with future performance. It is also shown that firms’ having a long operating cycle is negatively associated with future performance. However, no evidence is gathered to show that the association between discretionary accruals and future financial performance will increase due to the firms being in a bad state or having a long operating cycle.

(4)

List of Figures & Tables

Table 1 – Sample Selection………...………17

Table 2 – Descriptive Statistics………...22

Table 3 – Correlation Matrix………...24

Table 4 – Future Earnings, Discretionary Accruals and Firm Characteristics…..…………...22

Table 5- Table 6 – 2-Digit Firm Frequencies….………...35

Table 7 – Variance Inflation Factors………...36

(6)

6

1 Introduction

In this study, it is examined to what extent discretionary accruals are associated with future financial performance. In addition, it is examined to what extent firm characteristics, like the change in operating cash flows and the operating cycle of a firm, are associated with future performance. In addition, it will be examined to what extent these firm characteristics affect the association between discretionary accruals and future financial performance. Therefore, the main research question of this study is the following:

What is the association between discretionary accruals and future financial performance?

Financial information should be relevant and reliable for investors to make investment decisions. In some cases, managers can exercise judgement in financial reporting. For instance, when predicting future economic events, judgement is required. Managers can use their judgement to make financial reports more informative. This happens when mangers communicate private inside information and users of financial reports perceive credible signals about a firm’s financial performance. However, when managers use their judgement to opportunistically manage earnings and mislead stakeholders, the signals about firm performance are distorted. This is also known as earnings management. According to prior literature, it is known that managers’ discretion is often revealed in accruals. Dechow, Ge and Schrand (2010) state that discretionary accruals are used as a proxy to measure earnings management.

The association between managers’ accrual choices and earnings management is still a relevant concern for academics. Recent studies of Owens, Wu and Zimmerman (2017) and Collins, Pungaliya and Vijh (2017) and Frankel and Sun (2017), show how for instance shocks to firms’ underlying economics affect managers’ accrual choices and earnings management. The studies regarding this concern began with Healy (1985), DeAngelo(1986) and Jones (1991). Since Healy (1985), some studies provided evidence of earnings management, while others argued that weaknesses of the research designs limit the reliability of these studies (Bernard and Skinner, 1996; Frankel and Sun, 2017; Chen, Hribar and Melessa, 2017). Moreover, while some stated that managers’ accrual choices are opportunistic and therefore caused noise, others argued that managers used their discretion to improve the informational value of earnings (Bernard and Skinner, 1996; Watts and Zimmerman, 1986; Healy and Palepu, 1993).

Subramanyam (1996) has shown that discretionary accruals are positively associated with stock prices, future earnings and cash flows. He found evidence that managers use accruals

(7)

7 to increase the informativeness of earnings; so, in an efficient market, managerial discretion can improve the ability of earnings to reflect the value of the firm. Nevertheless, he has also shown that discretionary accruals are opportunistic and not relevant for the value creation of the firm; but are still priced by an inefficient market (Subramanyam, 1996; Guay et al., 1996; Louis and Robinson, 2005). This study will follow Subramanyam (1996) by examining the association between discretionary accruals and future earnings. However, this study will differentiate itself from prior literature as in this study it will be examined whether firm specific characteristics influence this association.

According to Owens et al. (2017) accrual models neglect the underlying economic circumstances that lead to an increase of firm performance. These circumstances differ across firms and over time and therefore result in different levels of accruals. For example, two similar firms will experience increases in sales. However, it may be that for one firm it is due to an increase in demand, whilst for the other it is due to longer credit terms. This will result in different levels of accruals in the firms, as the first may have below-normal levels of inventory whilst the latter will have higher accounts receivable. So, although firms are in the same industry and face the same external environment they have different business models. Therefore, this study will investigate whether firm specific characteristics, like the state of a firm and the length of its operating cycle, will influence the level of accruals.

It is expected that the level of accruals will be different for firms being in a bad state. A bad state is defined as the state in which a firm’s change in cash flow from operations is negative. It is expected that during a bad state, cash flows and earnings will decrease as firms are facing problems. This decrease in cash flows might lead to an increase in accruals if managers, by using their discretion, estimate and report accruals deliberately wrong in order to present higher earnings to investors. Furthermore, in this study it will be examined how the length of the operating cycle of a firm will affect its future financial performance. Recent literature, Frankel and Sun (2017), have shown empirically that the relation between accruals and revenues varies with the length of the operating cash cycle. More specifically, a positive relation between accruals and current period revenues is shown, which increases with the length of the operating cash cycle. However, as it is not known yet what the effect of operating cycle is on future financial performance, investigating it will be relevant as it may contribute to existing literature.

This study is also relevant for several other reasons. First, financial statement users, such as stakeholders, are interested in a clarification of how discretionary accruals should be interpreted and under which circumstances these discretionary accruals can increase the

(8)

8 informativeness of earnings. Second, standard setters are trying to reduce managers’ ability to use their discretion in financial reports in order to reduce earnings management. However, if this study shows that managers use their discretion to increase the informativeness of earnings, standard setters may want to re-assess their approach.

This study makes use of panel data of USA public listed firms for the years 1987-2017. Discretionary accruals are measured with the Modified Jones model (1991) and future financial performance is measured with future earnings. Furthermore, building on existing literature of Dechow (1996), Dechow and Dichev (2002), Frankel and Sun (2017) and Chen et al. (2017), an empirical model is constructed. This model shows a linear relation between future earnings, discretionary accruals, the state of a firm and the length of the operating cycle. It is hypothesized that the association between discretionary accruals and future financial performance will be higher for firms being in a bad state. It is also hypothesized that the association between discretionary accruals and future financial performance will be higher for firms having a long operating cycle.

The results indicate that a firm being in a bad state will negatively affect future performance. In addition, the results indicate that a firm with a long operating cycle will also negatively affect future performance. However, there is not gathered enough evidence to state that the association between discretionary accruals and future financial performance will increase due to the state of the firm or the length of the operating cycle.

This study has some limitations as well. Given the findings of this study, it can be concluded that there is not enough evidence to state that the association between discretionary accruals and future financial performance will increase due to the firms being in a bad state or having a long operating cycle. This may imply that using discretionary accruals as a proxy to measure earnings management is not the most suitable proxy in this setting. Furthermore, the measurement error and potential bias associated with discretionary accruals is a concern. Therefore, it would be interesting to see of using a different proxy would lead to changes.

The remainder of this study is organized as follows: section two commences with a literature review, which gives an overview of the existing literature about accruals. Based on existing literature, the research hypotheses will be developed. Section three will describe the research methodology and it will therefore specify the methods and data which will be used. The results are presented and discussed in section 4 and subsequently in section five, a conclusion of this study will be given along with the limitations.

(9)

9

2 Literature Review and Hypothesis Development

In this part, relevant prior studies will be discussed to provide an extensive summary of the literature. In addition, I will discuss the variables and examine findings of prior studies in order to formulate hypotheses that will help to answer the main question of this research.

2.1 Information asymmetries

Information asymmetries between investors and the firm’s managers creates demand for a summary measure of firm performance. With this measure investors can evaluate managers and the ability of firms to generate cash. However, the problem is that managers are also rewarded based on the firm’s performance. As a consequence, investors and other external parties have difficulties in assessing to what extent the signals produced by management are reliable (Dechow, 1994).

One could evaluate firms and managers based on realized cash flows as these can be measured objectively. However, the problem is the timing of cash receipts and

disbursements. For instance, management could be penalized for making investments even though the net present value is positive. The reason for this is the timing of the cash receipts, these will be received at a later stage. According to Dechow (1994) an alternative could be the attempt of management to determine the firms expected future cash flows. The problem with this alternative is that the extent to which managers are flexible to report increases. As a consequence, it becomes difficult to verify any signal which is produced by the firm and therefore this alternative is also considered as an unreliable measure of firm performance. Using accruals can be seen as trading-off the above-mentioned problems. In addition, when using accruals, there are rules on the timing of cash flows that should be followed. Therefore, earnings will more closely reflect firm performance than realized cash flows Dechow (1994)

2.2 Clarifying Accruals

Earnings consist of cash flow from operations adjusted by accruals (Dechow, 1994). Accruals are transactions which must be recorded in the accounts to match the revenues and expenses, even though cash hasn’t been received or paid yet (Stolowy and Lebas, 2013). The method in which transactions should be recognized and recorded in the period to which they relate is called accrual accounting. The opposite of this is cash accounting, where transactions are recorded and reported in the period where cash is received or paid.

(10)

10 The ability of a firm to generate more cash inflows than cash outflows can indicate how successful a firm is. Therefore, net realized cash flows could be used as a performance measure. However, according to Dechow and Dichev (2002), the role of accruals is to adjust the recognition of cash flows over time. So, they state that firm performance can be measured more accurately with the adjusted numbers. As already mentioned, cash flows reduce the ability to measure firm performance due to timing and matching problems. Accruals can also mitigate the timing and mismatching problems of cash flows. These timing and mismatching problems of cash flows arise because the achievements or value creation of the firm often differs from the timing of the related cash flows. When net cash receipts and disbursements take place in periods different from the underlying economic event, there is a timing problem. Similarly, when cash inflows and outflows differ from the period in which the underlying event took place, there is a matching problem (Frankel and Sun, 2017).

However, despite the above-mentioned benefits of accruals, the use of accruals involves some problems as well. Namely, accruals are based on assumptions and estimates, and if wrong then they must be corrected in the future. For example, if the net proceeds from a receivable item are less than the estimated amount, then the subsequent entry should record both the cash collected and the correction of the estimation error. According to Dechow and Dichev (2002), estimation errors and their corrections are considered as noise which reduces the beneficial role of accruals. Another example is if cash for a transaction will be received or paid after the matching expenses and revenues are recognized in the earnings, the amount of cash that would be received or paid in the future must be estimated. Consequently, managers may estimate accruals deliberately wrong in order to present higher and persistent earnings to investors. This process is called earnings management.

Earnings management occurs when managers manipulate transactions or use their own judgment in financial reports in such a way that stakeholders are misinformed about the performance of the company (Healy and Wahlen, 1999). Earnings management is enabled through the use of accruals. Healy (1985) defined accruals as the difference between the reported earnings and cash flows from operations. The total accruals of a firm can be decomposed into discretionary and discretionary accruals. Healy (1985) states that non-discretionary accruals, also called the normal accruals, are the mandated expected level of adjustments to the firms’ cash flows. These accounting adjustments to the cash flows are mandated by the accounting standard setting bodies like the Financial Accounting Standards Board (FASB). Whereas, discretionary accruals are selected adjustments made by the manager. Here, managers get enabled to opportunistically change the cash flows due the discretion they

(11)

11 get by some generally accepted procedures which are again defined by the accounting standard setting bodies. Some examples are the flexibility to choose the depreciation method of assets or the flexibility to allocate fixed overhead costs between the cost of goods sold and inventories.

2.3 The operating cash cycle

As mentioned above, accruals can be used opportunistically to achieve earnings management goals and mislead users of financial statements. However, Dechow and Dichev (2002) argue that even though there is no intentional earnings management, the quality accruals will be associated with firm and industry characteristics such as the volatility of operations or the length of the operating cycle. The quality of accruals is negatively related to estimation errors as these errors reduce the beneficial role of accruals. According to Dechow and Dichev (2002) even with good intentions and skills, there is a likelihood that managers of volatile firms make larger estimation errors. In addition, they argue that it is important to make a distinction between firm and industry characteristics because such characteristics are likely to be both recognizable and persistent. This means that the volatility of operations is related to the propensity to make estimation errors compared to the determinants of managerial opportunism that are often unrecognizable and irregular.

Therefore, Dechow and Dichev (2002) investigate the relation between accrual quality and firm characteristics, as the operating cycle. The operating cycle measures the average time between the outflow of cash to produce a product and the receipt of cash from the sale of the product. They show that the longer the operating cycle, the lower the accrual quality. A long operating cycle implies that future cash flows will be recognized in current earnings by for example e accruing receivables. So, the intuition behind this result is that a long operating cycle leads to more uncertainty, more estimation and estimation errors, and consequently lower quality of accruals (Dechow and Dichev, 2002).

Besides these findings, Frankel and Sun (2017) have shown empirically that the relation between accruals and revenue varies with the length of the operating cash cycle. More specifically, they show that the relation between accruals and current period revenues changes is positive and this positive relation strengthens as the operating cash cycle increases. The intuition behind this result is that firms having a longer operating cycle will lead to more timing and matching problems for the firm’s cash flows. So, if the operating cycle increases than the difference between the credit terms for accounts receivable and accounts payable will also be greater. As a consequence, for a given change in revenue, a larger level of

(12)

12 accruals is needed to offset cash flows that will be realized in periods different than the creation of the profit. This is also in line with Dechow (1994) that the longer the operating cash cycle of a firm, the larger the change in working capital for a given change in sales or operations.

2.4 Other empirical results

2.4.1 Signaling

As accruals allows managers to transfer earnings between periods, management typically have some discretion over the recognition of accruals. This discretion can be used by management to signal their private information or to opportunistically manipulate earnings.

Subramanyam (1996) has examined whether managers use discretionary accruals to reveal some private information or whether they use it opportunistically to manage earnings. He finds that discretionary accruals are priced by the market. This means that the market attaches value to discretionary accruals. There are given two explanations for this finding. In the first scenario, discretionary accruals are priced by the market because it represents value-relevant information. In addition, under this scenario managers use their discretion to improve the ability of reported earnings to reflect the value of the firm. So, managers may use their discretion to reveal their knowledge about the profitability of the firm.

Subramanyam (1996) documented that discretionary accruals predict future profitability after controlling for the current levels of nondiscretionary accruals and operating cash flow. In addition, he found a positive association between discretionary accruals and dividend changes. These findings are consistent with managers who use accounting and dividend choices to give signals about the firm value and managers who communicate information about future profitability by using discretionary accruals.

Furthermore, these findings are in line with dissemination of private information by using accruals: if discretionary accruals allow investors to better predict future performance (profitability and dividends), they are more informative.

2.4.2 Earnings management

In the second scenario, mangers use their discretion to opportunistically manage their reported earnings (Watts and Zimmerman, 1986). As a consequence, the pricing of discretionary accruals under this scenario leads to market mispricing. According to Xie et al. (2003), financial information is crucial for capital markets as it is used to set security prices and

(13)

13 investors use this information to design an investment strategy. A consequence of incorrect information is that it could lead to mispricing of securities. According to the efficient market hypothesis, securities will be priced fairly, given all the information available to investors (Mohanram, 2014). Therefore, incorrect information will lead to mispricing of securities and this will reduce the probability that shareholders make correct informed decisions (Xie et al., 2003).

A reason for managers to manipulate earnings is for example to maximize their own and/or the firm’s wealth. Incentives for managers to maximize their own wealth arise when managers are compensated based on the firm’s earnings performance (Xie, Davidson and DaDalt, 2003). Examples of such situations are when managers receive bonusses, prestige and promotions if they achieve a pre-established target of earnings. If a firm for instance fails to achieve this target, managers may modify short-term expenses, such as advertising or training cost in order to achieve this target of earnings and as a result the bonusses and promotions (Stolowy Lebas and Ding, 2013).

Beyond this compensation problem for managers, it could be that managers manipulate earnings to maximize the firm’s wealth. A situation in which this is the case is for instance when contracts between the firm and other parties, such as debtholders and suppliers, are based on reported earnings (Becker, Defond, Jiambalvo and Subramanyam, 1998).

Lenders would like to know if the firm can pay back the amount of money borrowed and the interest upon it. The earnings of a firm indicate the financial position of a firm and may indicate the ability of the firm to hold to a debt contract. According to Dechow, Sloan and Sweeney (1996) high and stable earnings will encourage debtholders to lend their money to the firm. So, in order to attract external financing managers may manipulate the firm’s earnings. Another reason why managers will manipulate earnings is to avoid hostile takeovers. Easterwoord (2011) states that targets of hostile takeover attempts, will let their earnings increase in order to prevent that shareholders will support the takeover.

Another motivation of earnings management is to smooth earnings. The goal hereby is to reduce the difference between the reported earnings and the number of earnings that is normal or expected. Some reasons to opportunistically smooth earnings is to benefit from bonus plans or to signal lower risk (Healy, 1985).

While income smoothing seems opportunistic it does not have to be necessarily opportunistic. For instance, managers may smooth earnings to increase the persistence of earnings (Subramanyam (1996); Hand (1989)). If earnings are adjusted to increase the persistence of earnings or to get a more stable trend in the reported earnings and if earnings are

(14)

14 smoothed to mitigate the effects of transitory cash flows, then income smoothing may enhance the value relevance of earnings (Subramanyam, 1996).

2.5 Hypothesis Development

Following the literature discussed above, in this section hypotheses will be developed to answer the research question. As there has been a number of studies, Dechow (1994); Dechow and Dichev (2002); Frankel and Sun (2017); Chen et al. (2017), investigating the role of accruals, it is known that accruals and operating cash flows are negatively correlated. According to Frankel and Sun (2017) for a given increase in revenue, more accruals are needed to offset cash flows that will be realized in periods other than the realization of the revenue. It is expected that during a “bad state”, like a crisis for example, cash flows and earnings will decrease as firms are facing problems. This decrease in cash flows might lead to an increase in accruals if managers, by using their discretion, estimate and report accruals deliberately wrong in order to present higher earnings to investors. So, reporting accruals opportunistically wrong might lead to an increase in discretionary accruals as well. (Dechow and Dichev, 2002). For example: it could be argued that the likelihood of firms performing an impairment will increase due to a bad state. Impairments may be considered as informative as they can signal that the firm is not performing well. However, it could be that managers do not impair the full amount that should be impaired and therefore give a wrong signal. The reason for this also to present relatively higher earnings to investors.

In this study a bad state is defined as the state where the cash flow from operations is negative. Therefore, it is predicted that when a firm is in a bad state, the level of discretionary accruals will be higher. This leads to the first hypothesis:

H1: The association between discretionary accruals and future earnings is higher for firms in

a bad state.

After investigating the association between discretionary accruals and future earnings, it will be examined whether the operating cash cycle affects future earnings. From prior literature, it is known that a long cash operating cycle leads to more uncertainty, more estimation and estimation errors, and consequently lower the quality of accruals (Dechow and Dichev (2002). Therefore, it predicted that for firms’ having a longer operating cycle there is much uncertainty and the problems regarding the discretionary use of accruals will increase. This leads to the second hypothesis

(15)

15

H2: The association between discretionary accruals and future earnings is higher for firms

(16)

16

3 Research Methodology

In this part of the paper, methods that will be applied and followed in order to conduct the research will be discussed. Furthermore, the data and sources that are needed will be discussed. Lastly, the measures of the variables will be defined and an appropriate statistical model with control variables using prior studies will be designed.

3.1 Data and sample selection

For this study, data will be gathered through databases within Wharton Research Data Services (WRDS). Compustat North America - Fundamentals Annual database will be used to obtain annual company-year level data for U.S publicly listed firms. The sample period will be a thirty-year period from 1985 –2017. Consistent with prior research, all variables are scaled by average total assets. Initially, the sample consists of 26,774 firms and 290,032 firm-years, as can be observed in Table 1. However, financial firms are excluded, SIC codes 6000-6999, as prior literature is also conducted with solely non-financial firms. This leads to a drop of 1,190 firms and 36,268 firm-years. Furthermore, firms with missing data about cash flow from operations are left out. This resulted in a drop of the whole year 1985 and 1986, as insufficient data was available for them.As can be seen in Table 1, this resulted in a drop of 11,938 firms and 139,057 firm-years. In addition, firms with other missing financial statement data are also removed. Furthermore, to filter the data from extreme observations, the data is winsorized at the 1st_{and 99}th_{percentiles of their distribution. This results in a drop of 8,628 firms and 78,255}

firm-years. Lastly, firms with data of less than five consecutive years are removed. This resulted in a drop of 2,025 firms and 7,947 firm-years. Consequently, the final sample consists of 2,993 firms and 28,536 firm-years.

When examining the 2-digit SIC codes of the final sample, it is noted that 21.91 percent of the firms are within the SIC2 codes 40-49. These codes are attributable to transportation & public Utilities. More specifically, 17.27 percent of the final sample has a 2-digi SIC code of 49. Which means that the sample has a large share of firms in the electric, gas and sanitary services industry. More details of this can be seen in Table 5 (Appendix).

(17)

17 Table 1

Sample Selection

Description Firms Firm-years

Firms on Compustat North America 1985-2017 26,774 290,063

Dropped: financial firms 1,190 36,268

Dropped: firms missing cash flow from operations 11,938 139,057 Dropped: firms missing financial statement data 8,628 78,255 Dropped: firms with less than five consecutive years 2,025 7,947

Final sample 2,993 28,536

3.2 Description of variables and empirical models

The main question of this study is: what is the association between discretionary accruals and future financial performance? The predictive validity framework presented in figure 1 (see the appendix) shows how the conceptual relationship, which will be examined in this study, will be operationalized. The procedure is to first estimate discretionary accruals as the residual from an ordinary least squares (OLS) regression. Then these residuals, i.e. discretionary accruals, are used as an independent variable in the second OLS regression to test the predictions made in this study. Discretionary accruals will be estimated with the modified Jones (1991) model. Jones (1991) developed a regression model in which he controlled for factors influencing total accruals. More specifically, he made a linear relation between total accruals, change in revenues and property, plant and equipment. The factors influencing the total accruals are the nondiscretionary accruals whereas the errors terms are called discretionary accruals. This model is still the most used method to estimate discretionary accruals, but this model has been criticized by academics because of its weaknesses, such as it is subjected to large measurement errors and potential bias (Lawrence et al., 2011; Chen, Hribar and Melessa, 2017). As stated before, discretionary accruals are estimated error terms. These errors are the difference between total accruals and non-discretionary accruals and therefore represent the unexplained or discretionary part of the total accruals. However, it could be argued that in a given year a firm has high total accruals because of the underlying business activities. So, the level of the estimated errors, which are the discretionary accruals, can be due to higher total accruals. Therefore, these error terms are correlated with firm performance and this can lead to potential bias.

However, there are also some strengths of this model namely: it has a high statistical power in detecting earnings management. In addition, this proxy may signal more extreme

(18)

18 undetected misstatements and captures quality variation for a large number of firms. (Defond and Zhang, 2014). To be consistent and comparable with prior literature, the modified-Jones (1991) model, as recommended by Kothari, Leony and Wasley (2005), is used to estimate discretionary accruals. Future earnings will be used as a measure of future financial performance. As in this study it will also be examined whether the associaton between discretionary accruals and future financial performance differs with the state of the firm, the state of the firm has to be defined. Cash flows from operations are therefore used as a proxy to assess the state of a firm. A good state is defined as a positive change in the cash flows from operations whereas a bad state is defined as a negative change in the cash flow from operations. As can be seen in the model beneath, a dummy variable is created, which is equal to one if the company is in a bad state. Lastly, as I also want to investigate if the association between discretionary accruals and future financial performance differs with the length of the operating cycle, another dummy variable is created. This dummy variable is equal to one if the operating cycle is long and zero otherwise. Following Dechow and Dichev (2002) first the length of the operating cycle is calculated. Hereafter a distinction is made between short and long operating cycles. Firms with an operating cycle above the median are categorized as firms with long operating cycles and firms with an operating cycle beneath the median are categorized as firms with short operating cycles. Thus, to test our predictions and in order to answer the main question of this study, the following regression model will be used:

EARN i, t+1 = α + β1DAC i,t + β2BADSTATE i,t + β3DAC i,t * BADSTATE i,t + β4OCYCLE i,t +

β5DAC i,t *OCYCLE i,t + β6SIZE i,t + β7ROA i,t + β8LEVi,t + β9GROWTH i,t + ε i,t (1)

Where the (control) variables stand for: EARN i, t+1 = Firm i’s earnings in year t+1;

DAC i,t = Firm i’s year t abnormal accrual, estimated as the residual of the modified

Jones model;

BADSTATEi,t = 1 if the change in cash flow from operations is negative and otherwise 0;

OCYCLE i,t = 1 if the operating cash cycle is long and otherwise 0;

SIZE i,t = The natural logarithm of firm i’s year t market value of common equity;

(19)

19 LEV i,t = Total long term debt plus total debt in current liabilities scaled by total

assets;

GROWTH i,t = (Current Year Sales – Previous Year Sales)/ Previous Year Sales.

In order to test the hypotheses by using the model above, discretionary accruals and the operating cash cycle have to be calculated first. As mentioned earlier, the modified-Jones (1991) model is used for this. With this model, the abnormal accruals (discretionary accruals) are estimated for every industry group per year. Industry groups are defined by the two-digit SIC codes (Veenman, 2013). It is expected that a regression for every industry group will estimate the abnormal accruals more precisely than a regression with all the firms together. Estimating the abnormal accruals with all the firms and industries together would not take in to account the differences of the underlying business economics of the firms. However, with the two-digit grouping the differences in the business activities are taken into account. The exact model used is as equation (6) in Kothari et al. (2005, p. 173) with a modification for the change in receivables. This model, in which all the variables are scaled by lagged total assets, is the following:

TACC i,t = β0 + β1(1/Assets i,t-1) + β2(∆Sales i,t - ∆REC i,t) + β3PPE i,t + ε i,t (2)

Where the variables stand for: TACC i,t = Total accruals

= (Income before extraordinary items t – net cashflow from operating

activities t) / total assets t-1

ASSETS i,t-1 = Total assets in year t-1;

∆SALES i,t = (Sales revenue in year t – Sales in year t-1) / total assets in year t-1 ;

∆REC i,t = (Accounts receivable in year t – Accounts receivable in year t-1) / total assets

in year t-1 ;

PPE i,t = (Net property, plant and equipment in year t) / total assets in year t-1;

ε i,t = Estimated Discretionary Accrual.

As mentioned earlier, in line with Dechow (1994) and Dechow and Dichev (2002) the length of the operating cycle will be calculated. This will be done by using the following model:

(20)

20 Operating Cycle =

(

(𝐴𝑅 𝑡 + 𝐴𝑅 𝑡−1) /2 𝑆𝑎𝑙𝑒𝑠/360 + (𝐼𝑁𝑉 𝑡 + 𝐼𝑁𝑉 𝑡−1) /2 𝐶𝑜𝑠𝑡 𝑜𝑓 𝑔𝑜𝑜𝑑𝑠 𝑠𝑜𝑙𝑑/360

)

(3) where: AR = Accounts receivable; INV = Inventory; AP = Accounts payable;

The first component of the operating cycle formula measures the number of days’ sales in accounts receivable. The second component measures the number of days’ it takes to produce and sell the product. sales in accounts receivable. The sum of these two indicates the average time between the outflow of cash to produce a product and the receipt of cash from the sale of the product.

In line with Cameran et al. (2014) and Ghosh and Moon (2005), firm size is included as a control variable. According to the political cost theory (Kim et al., 2003)., managers of larger firms are more likely to exploit the latitude in accounting to decrease political costs. In other words, managers of large firms are more incentivized to keep earnings stable over time. In addition, Kim et al. (2003) state there is higher probability that large firms manage their earnings to meet or beat certain expectations. As this affects earnings quality it is necessary to control for firm size. Leverage is also included as companies with higher leverage are more likely to exploit their latitude to avoid debt covenant restrictions (Warfield et al., 1995). Return on asset (ROA) is another important variable which needs to be controlled. According to Meek et al. (2007) firms with higher return on asset, have higher discretionary accruals. It is possible that managers need to achieve a certain percentage of return on asset to get their bonus. Lastly, there will be controlled for the growth of firms. According to Desai et al. (2004), high growth firms are likely to have large accruals, while low growth firms are likely to have a smaller number of accruals.

(21)

21

4 Empirical results

In this part of the study, the results will be discussed. First, the descriptive statistics and correlations are discussed. Afterwards, the empirical results obtained from the regression model, are discussed. Lastly, these results will be compared with the predictions and existing literature.

4.1 Descriptive statistics and correlations

Table 2 provides descriptive statistics on the variables. As it can be observed in the table, the mean of discretionary accruals is 0. According to prior literature, the mean of residuals should be equal to zero (Cohen et al., 2008). As discretionary accruals are estimated residuals, this observation is in line with prior literature. Another interesting variable to discuss is the operating cycle. The operating cycle is firm i’s operating cycle in days. As it can be seen, the mean of the operating cycle is 123.306. This indicates that on average the operating cycle of the firms’ included in this sample is equal to 124 days. Dechow and Dichev (2002) report an average operating cycle of 141.08 for 1,725 firms between 1987 and 1999. The lower average operating cycle might be due to the differences in doing business over time. For example, todays’ society is way more advanced in technology than the society in the 90’s (Khan, Khan and Aftab, 2015). This advanced technology and digitization makes purchasing and paying easier and faster. Also, mobile-banking is so advanced that transactions could be made in less than a minute. So, as digitization makes receiving money easier and quickly, it might be the reason for a decrease in the operating cycle (the average time between the outflow of cash to produce a product and the receipt of cash from the sale of the product).

The descriptive statistics for the control variable SIZE are remarkable. The mean is 6.744 whereas the other control variables have a mean between the 0 and 2. Moreover, the standard deviation (2.105) is also relatively higher than the standard deviations of the other control variables. A possible reason could be that the size of firms in this sample is very divergent as the sample contains firms throughout different industries. Lastly, the control variable ROA has a mean of 0.269 and a standard deviation of 0.325. Remarkable here is that the minimum value of ROA is -0.034 whilst the maximum is 19.349. This indicates that the ROA is divergent and the reason for this also might be the sample which consist of different firms throughout different industries.

(22)

22 Table 2 Descriptive Statistics N Mean Std. P1 P25 Median P75 P99 EARNINGS 24.563 0.256 0.270 0.000 0.119 0.194 0.314 8.693 DAC 28.542 0.000 0.149 -1.357 -0.065 0.005 0.072 1.127 BADSTATE 28.542 0.500 0.500 0.000 0.000 1.000 1.000 1.000 CFO 28.542 0.469 0.491 0.000 0.189 0.318 0.579 10.751 DAC * BADSTATE 28.542 0.021 0.103 -1.067 0.000 0.000 0.036 1.127 OpCycle 28.542 123.306 78.445 2.766 72.694 108.053 155.152 1,572.120 LongOpCycle 28.542 0.484 0.500 0.000 0.000 0.000 1.000 1.000 DAC * LongOpCycle 28.542 0.003 0.077 -1.009 0.000 0.000 0.006 1.100 SIZE 25.34 6.744 2.105 1.518 5.310 6.814 8.183 11.477 LEVERAGE 28.542 1.238 1.685 0.000 0.137 0.560 1.646 8.702 ROA 28.542 0.268 0.325 -0.034 0.117 0.199 0.328 19.349 GROWTH 24.568 0.002 0.025 0.000 0.000 0.000 0.000 2.794

Notes: Table 1 presents descriptive statistics for key variables used in this study. The sample consists of all Compustat firm-years during 1987 - 2017 for which the relevant data to compute the partitioning variables reported in this table are available. All variables are scaled by average total assets and winsorized at the 1 percent and 99 percent levels. Variables measurement: EARNINGS = Earnings of firm i in year t+1; DAC = Firm i’s year t abnormal accrual, estimated as the residual of the modified Jones (1991) model as recommended by Kothari et al. (2005); CFO = Firm i's cash flow from operations in year t, scaled by average total assets; BADSTATE= an indicator variable that equals 1 if the change in CFO < 0, and equals 0 otherwise; OpCycle = Firm i's operating cycle in days; LongOpCycle = An indicator variable that equals 1 if the OpCycle > median, and equals 0 otherwise; SIZE= the natural logarithm of firm i’s year t market value of common equity (Compustat PRCC_F *CSHO); LEVERAGE = Firms i's year t leverage, measured as total long-term debt (DLTT) plus total debt in current liabilities (DLC) scaled by total assets (AT); ROA = Firm i's year t return on assets, computed as net income divided by lagged total assets; GROWTH = Firm i's difference in sales from year t to t-1 , divided by sales in year t -1.

The correlations in Table 3 illustrate the relations between the sample variables. These correlations are in line with existing literature (Dechow and Dichev, 2002; Frankel and Sun, 2017). As it can be seen in the table, the (Spearman) correlation between cash flow from operations and future earnings is positive (0.817). The relation between discretionary accruals and future earnings is negative (-0.026). Moreover, the relation between a firm in a bad state and its future earnings is negative (-0.065). In addition, it can be seen that the operating cycle is negatively correlated with future earnings (-0.253) and cash flow from operations (-0.362). However, the relation between operating cycle and discretionary accruals is positive (0.043). Furthermore, discretionary accruals are negatively correlated with cash flow from operations (-0.233), which based on previous findings indicate that accruals offset components in operating cash flows (Frankel and Sun, 2017). Since cash flows suffer from temporary mismatching of cash receipts and disbursements, it is likely that cash flows contain components which are reversed over time. In addition, as accruals are used to offset the

(23)

23 mismatching between cash receipts and cash disbursements, accruals and cash flows are negatively correlated.

Remarkably, both indicator variables: BADSTATE and LongOpCycle are negatively correlated with the control variables. For instance, SIZE has a negative relation with the operating cycle of -0.047 and it is also negatively correlated with BADSTATE. Another noticeable variable is LEVERAGE. LEVERAGE is positively related with future earnings and cash flow from operations (0.453 and 0.620, respectively) but it is negatively correlated with discretionary accruals (-0.012). Lastly, the control variable GROWTH is positively related with Discretionary accruals. McNichols (2000) has shown that firms with greater expected growth are likely to have more income increasing accruals. So, this might explain the positive relation between the control variable GROWTH and discretionary accruals.

(24)

24 Table 3 Correlation Matrix 1) 2) 3) 4) 5) 6) 7) 8) 9) 10) 1) EARNINGS -0.042 -0.084 0.705 -0.338 -0.300 0.249 0.290 0.780 0.102 2) DAC -0.026 0.321 -0.332 0.044 0.038 -0.052 -0.030 0.060 -0.016 3) BADSTATE -0.065 0.285 -0.217 0.018 0.014 -0.015 -0.016 -0.042 -0.153 4) CFO 0.817 -0.233 -0.132 -0.516 -0.463 0.239 0.494 0.721 0.103 5) OpCycle -0.253 0.043 0.020 -0.362 0.866 -0.039 -0.374 -0.323 -0.121 6) LongOpCycle -0.260 0.041 0.014 -0.371 0.692 -0.046 -0.348 -0.288 -0.096 7) SIZE 0.132 -0.036 -0.015 0.118 -0.014 -0.047 0.246 0.246 -0.187 8) LEVERAGE 0.453 -0.012 -0.027 0.620 -0.329 -0.386 0.114 0.262 0.026 9) ROA 0.797 0.059 -0.019 0.776 -0.217 -0.224 0.117 0.394 0.101 10) GROWTH 0.102 0.008 -0.033 0.075 -0.028 -0.031 -0.117 0.026 0.117

Notes: Table 2 presents Spearman (Pearson) correlations above (below) the diagnal among key variables that are used in this study. The sample consists of all Compustat firm-years during 1987 - 2017 for which the relevant data to compute the partitioning variables reported in this table are available. All variables are scaled by average total assets and winsorized at the 1 percent and 99 percent levels. Variables measurement: EARNINGS = Earnings of firm i in year t+1; DAC = Firm i’s year t abnormal accrual, estimated as the residual of the modified Jones (1991) model as recommended by Kothari et al. (2005); CFO = Firm i's cash flow from operations in year t, scaled by average total assets; BADSTATE= an indicator variable that equals 1 if the change in CFO < 0, and equals 0 otherwise;

OpCycle = Firm i's operating cycle in days; LongOpCycle = An indicator variable that equals 1 if the OpCycle > median, and equals 0 otherwise; SIZE= the natural logarithm of firm i’s year t market value of common equity (Compustat PRCC_F *CSHO); LEVERAGE = Firms i's year t leverage, measured as total long-term debt (DLTT) plus total debt in current liabilities (DLC) scaled by total assets (AT); ROA = Firm i's year t return on assets, computed as net income divided by lagged total assets; GROWTH = Firm i's difference in sales from year t to t-1 , divided by sales in year t -1.

(25)

25

4.2 Tests of predictions

The variance inflation factors (VIFs) are examined to assess the seriousness of multicollinearity. As it can be seen in Table 7 (appendix), the highest VIF is 2.52. According to Chatterjee and Price (1997), VIFs’ higher than 10 indicates serious multicollinearity problems. So, this means that multicollinearity is not a serious problem in this study.

4.2.1 Test of H1

The first hypothesis examines whether the association between discretionary accruals and future performance is higher for firms in a bad state. Table 4 reports results for the entire sample. The results show that the coefficient of discretionary accruals is significantly negative 0.077 with a t-stat of -3.07). The coefficient of the variable BADSTATE is also negative (-0.016 with a t-stat of -4.94). The variable BADSTATE is an indicator variable that takes the value of 1 if the change in cash flow from operations is negative and simultaneously it implies that the firm is being in a bad state. This significantly negative result indicates that firms being in a bad state this year, will have a lower performance in the following year. Furthermore, the coefficient of the interaction term, (DAC* BADSTATE), is significantly positive (0.178, with a t-stat of 4.01) in column (1). So, this would mean that the association between discretionary accruals and future financial performance becomes lower for firms in a bad state. It was expected that the association between discretionary accruals and future earnings would be higher for firm being in a bad state. The intuition behind this was that due to the bad state of firm i’s in year t, managers of these firms would opportunistically report accruals wrong to present a better performance of the firm in year t+1. However, according to the results in column 1, the opposite is shown of what was predicted. The reason for this opposite result might be that a firm being in a bad state in year t may impact several years of operations as managers react by modifying their strategies. This would imply that it is possible that the interaction term likely fails to capture how the shock in year t affects earnings in year t+1. However, as it can be seen in column 4, the coefficient of this interaction term becomes insignificant when controlling for the firm size, growth, leverage and profitability (ROA). Elimination of these control variables would provide a biased view.

In addition, as the table shows, the adjusted R-squared increases significantly from 0.6% in column (1) to 7.2% in column (3) when the operating cash cycle and its interaction with discretionary accruals (DAC*OCYCLE) are included. Lastly, when the control variables are included, the adjusted R-squared increases significantly to 66.8%.

(26)

26 Table 4

Future Earnings, Discretionary Accruals and Firm Characteristics EARN i, t+1 = α + β1DAC i,t + β2BADSTATE i,t + β3DAC i,t * BADSTATE i,t +

β4LongOpCycle i,t + β5DAC i,t *LongOpCyclei,t + β6SIZE i,t + β7ROA i,t + β8LEVERAGEi,t

+ β9GROWTH i,t + ε i,t

Column nr. 1 2 3 4 DAC -0.097** -0.043 -0.071*** -0.077*** (-2.11) (-0.81) (-2.63) (-3.07) BADSTATE -0.037*** -0.035*** -0.016*** -0.016*** (-7.70) (-8.07) (-5.43) (-4.94 ) DAC * BADSTATE 0.178*** 0.119*** -0.071 -0.055 (4.01) (2.67) (-1.26) (-1.03) LongOpCycle -0.141*** -0.019** -0.016** (-14.19) (-2.31) (-2.46) DAC * LongOpCycle -0.023 -0.007 0.001 (-0.44) (-0.17) (0.02) SIZE 0.338*** 0.003* 0.005** (0.011) (1.89) (2.01) LEVERAGE 0.025*** 0.030*** (3.42) (3.62) ROA 0.597*** 0.561*** (5.05) (4.54) GROWTH 0.135 0.100 (0.67) (0.64) Constant 0.271*** 0.338*** 0.056*** 0.116** (37.50) (30.70) (3.06) (2.49) Observations 24,563 24,563 19,538 19,538 Adjusted R-squared 0.006 0.072 0.663 0.668

Fixed effects No No No Yes

Notes: This table presents the coefficients estimated from model (1). T-statistics are in parentheses. Furthermore, the sample consists of firm-years observations from 1987 to 2017. The sample consists of all Compustat firm-year during 1987 - 2017 for which the relevant data to compute the partitioning variables reported in this table are available. All variables are scaled by average total assets and winsorized at the 1 percent and 99 percent levels. Variables measurement: EARNINGS = Earnings of firm i in year t+1; DAC = Firm i’s year t abnormal accrual, estimated as the residual of the modified Jones (1991) model as

recommended by Kothari et al. (2005); CFO = Firm i's cash flow from operations in year t, scaled by average total assets; BADSTATE= an indicator variable that equals 1 if the change in CFO < 0, and equals 0

otherwise; OpCycle = Firm i's operating cycle in days; LongOpCycle = An indicator variable that equals 1 if the OpCycle > median, and equals 0 otherwise; SIZE= the natural logarithm of firm i’s year t market value of common equity (Compustat PRCC_F *CSHO); LEVERAGE = Firms i's year t leverage, measured as total long-term debt (DLTT) plus total debt in current liabilities (DLC) scaled by total assets (AT); ROA = Firm i's year t return on assets, computed as net income divided by lagged total assets; GROWTH = Firm i's

difference in sales from year t to t-1 , divided by sales in year t -1. *** , ** and * indicate statistical significance at the 1%, 5% and 10% confidence levels, respectively.

(27)

27

4.2.2 Test of H2

The second hypothesis concerns the association between discretionary accruals and future performance for firms having a long operating cycle. As mentioned earlier, the operating cycle indicates the average time between the outflow of cash to produce a product and the receipt of cash from the sale of the product. According to prior literature, longer operating cycles indicate more uncertainty, more estimation and errors of estimation. Therefore, it was expected that due to the increased uncertainty, the association between discretionary accruals and future performance for firms having long operating cycles will increase. As it can be seen in Table 3, the coefficient of the variable, LongOpCycle, is significantly negative (-0.016, with a t-stat of -2.46). The variable, LongOpCycle, is an indicator variable that takes the value of 1 if firm i has a long operating cycle. This significantly negative result indicates that a long operating cycle will negatively affect future performance. This finding can partially be explained by the findings of prior literature.

As mentioned earlier, Dechow and Dichev (2002) state that long operating cycles indicate more uncertainty, more estimations and more errors of estimation and therefore lowers the quality of accruals. So, the negative association between a long operating cycle of and future performance is possibly caused by the increased uncertainty. It could also be that this uncertainty is perceived by third parties, like investors or customers, and their negative reaction causes a further decline of future earnings. Moreover, Wang (2017) has examined whether the operating cycle predicts future stocks returns after controlling for profitability. He has shown a strong negative relation between a firm’s operating cycle and its subsequent returns. So, the finding of the negative association between the length of the operating cycle and future earnings is in line with prior findings.

In the table it can also be seen that the coefficient of the interaction term, DAC*OCCYLE, is insignificant (0.001 with a t-stat of 0.02). It was predicted that the association between discretionary accruals and future performance would be higher for firms having a long operating cycle. The intuition behind this prediction was the following: as longer operating cycles goes along with higher uncertainty, more estimation and more estimation errors, it might be that due to longer operating cycles and uncertainty, the discretion of managers to deliberately report accruals wrong, in order to present better firm performance, would increase. However, the results do not support this hypothesis. A possible reason might be the misspecification in the modified Jones model to estimate discretionary accruals. As this model ignores the fact that accruals are affected by future expected growth. For example,

(28)

28 increases and decreases in inventory often lead to expected future changes in sales. Therefore, when estimating discretionary accruals these growth effects should be controlled for (Collins et al., 2017).

Summarizing the results, it is shown that a firm being in a bad state will negatively affect future performance. In addition, the results indicate that firms’ having a long operating cycle will also negatively affect future performance. Lastly, there is not gathered enough evidence to state that the association between discretionary accruals and future financial performance will increase due to the state of the firm or the length of the operating cycle.

4.3 Robustness test

4.3.1 Fama-Macbeth statistics

Table 5 reports the time-series averages of the coefficients of the independent variables. One advantage of the Fama-Macbeth (1973) regression is that it corrects for cross-sectional correlation in the pooled sample.

Table 5

Future Earnings, Discretionary Accruals and Firm Characteristics EARN i, t+1 = α + β1DAC i,t + β2BADSTATE i,t + β3DAC i,t * BADSTATE i,t +

β4LongOpCycle i,t + β5DAC i,t *LongOpCyclei,t + β6SIZE i,t + β7ROA i,t + β8LEVERAGEi,t

+ β9GROWTH i,t + ε i,t

Column nr. 1 2 3 DAC -0.097** -0.043 -0.071*** (-2.11) (-0.81) (-2.63) BADSTATE -0.037*** -0.035*** -0.016*** (-7.70) (-8.07) (-5.43) DAC * BADSTATE 0.178*** 0.119*** -0.071 (4.01) (2.67) (-1.26) LongOpCycle -0.141*** -0.019** (-14.19) (-2.31) DAC * LongOpCycle -0.023 -0.007 (-0.44) (-0.17) SIZE 0.338*** 0.003* (0.011) (1.89) LEVERAGE 0.025*** (3.42) ROA 0.597*** (5.05) GROWTH 0.135

(29)

29 (0.67) Constant 0.271*** 0.338*** 0.056*** (37.50) (30.70) (3.06) Observations 24,563 24,563 19,538 Adjusted R-squared 0.006 0.072 0.663

Notes: This table presents the coefficients estimated from model (1). T-statistics are in parentheses. Furthermore, the sample consists of firm-years observations from 1987 to 2017. The sample consists of all Compustat firm-year during 1987 - 2017 for which the relevant data to compute the partitioning variables reported in this table are available. All variables are scaled by average total assets and winsorized at the 1 percent and 99 percent levels. Variables measurement: EARNINGS = Earnings of firm i in year t+1; DAC = Firm i’s year t abnormal accrual, estimated as the residual of the modified Jones (1991) model as

recommended by Kothari et al. (2005); CFO = Firm i's cash flow from operations in year t, scaled by average total assets; BADSTATE= an indicator variable that equals 1 if the change in CFO < 0, and equals 0

otherwise; OpCycle = Firm i's operating cycle in days; LongOpCycle = An indicator variable that equals 1 if the OpCycle > median, and equals 0 otherwise; SIZE= the natural logarithm of firm i’s year t market value of common equity (Compustat PRCC_F *CSHO); LEVERAGE = Firms i's year t leverage, measured as total long-term debt (DLTT) plus total debt in current liabilities (DLC) scaled by total assets (AT); ROA = Firm i's year t return on assets, computed as net income divided by lagged total assets; GROWTH = Firm i's

difference in sales from year t to t-1 , divided by sales in year t -1. *** , ** and * indicate statistical significance at the 1%, 5% and 10% confidence levels, respectively.

(30)

30

5 Conclusion

This study investigated whether there is an association between discretionary accruals and future financial performance. More specifically, it is examined if the association between discretionary accruals and future performance increases for firms being in a bad state. Furthermore, it is examined what the association is between firms having a long operating cycle and future performance. Lastly, it is examined whether the association between discretionary accruals and future performance increases for firms having a long operating cycle.

As discussed in the theoretical part of this study, the purpose of financial reporting is to provide information that is useful for making decisions. Financial information should be relevant and reliable for investors to make investment decisions. However, managers can exercise judgement in financial reporting, which makes financial statements less reliable. In some cases, judgement is required in predicting future economic events. Managers can use their judgement to make financial reports more informative. This happens when mangers communicate private inside information and users of financial reports perceive credible signals about a firm’s financial performance. However, when managers use their judgement to opportunistically manage earnings and mislead stakeholders, the signals about firm performance are distorted. This is also known as earnings management. Discretionary accruals are used as a proxy to measure earnings management. Therefore, in this study discretionary accruals are estimated using the modified-Jones (1991) model.

Furthermore, building on existing literature of Dechow (1996), Dechow and Dichev (2002), Frankel and Sun (2017) and Chen et al. (2017), an empirical model is constructed. This model shows a linear relation between future earnings, discretionary accruals, the state of a firm and the length of the operating cycle. It was hypothesized that the association between discretionary accruals and future financial performance would be higher for firms in a bad state. It was also hypothesized that the association between discretionary accruals and future financial performance would be higher for firms having a long operating cycle.

The sample consists of 28,536 firm-year observations over the period 1987-2017. Conducting a regression with the model constructed from prior literature, it is shown that a bad state of a firm is negatively associated with future earnings. More specifically, a firm being a bad state this year will cause a decrease in the future financial performance. The findings also indicate that long operating cycles are negatively associated with future earnings. The intuition behind is that a long operating cycle increases uncertainty and estimation errors. Therefore, users of financial reports could perceive this as a negative signal and as a consequence future

(31)

31 performance will decline. No significant evidence is gathered to show that the association between discretionary accruals and future financial performance will increase due to the fact that a firm is in a bad state. In addition, there is also no significant evidence gathered to show that the association between discretionary accruals and future financial performance increase due to firms’ having a long operating cycle.

This study has important implications for investors, firms and accounting researchers. First, this study contributes to the existing literature by providing evidence that a firm being a bad state this year will cause a decrease in the future financial performance. In addition, the finding that a long operating cash cycle decrease future financial performance also extend existing literature. Second, the findings might help investors for making investment-decisions in certain firms. As this study shows, the future financial performance of firms having a long operating cycle will decline. This information might be helpful for investors to (re)consider their investment decisions.

However, this study has also some limitations. Given the results, it can be concluded that there is not enough evidence to state that the association between discretionary accruals and future financial performance increases for firms’ being in a bad state or having a long operating cycle. This may imply that using discretionary accruals as a proxy to measure earnings management is not the most suitable proxy is in this setting. Furthermore, the measurement error and potential bias associated with discretionary accruals is a concern. In this study, discretionary accruals were estimated by the modified-Jones (1991) model and afterwards used as an independent variable to explain future performance. According to Chen et al. (2017), a two-step regression procedure leads to biased coefficients and t-statistics. Consequently, they state that the two-step procedure frequently yields incorrect inferences. Another limitation is concerned the sample selection. The sample consist of 28,536 firm-year observations over the period 1987-2017. It would be interesting to increase the sample and thereby increase firm-year observations. As the sample also incorporates the recent financial crisis, it would be interesting to exclude the financial crisis or control for it.

(32)

32

References

Becker, C. L., DeFond, M. L., Jiambalvo, J., & Subramanyam, K. R. (1998). The effect of audit quality on earnings management. Contemporary accounting research, 15(1), 1-24.

Berk, J., & DeMarzo, P. (2013). Corporate finance. Edinburgh Gate: Pearson Education Limited.

Bernard, V. L., & Skinner, D. J. (1996). What motivates managers' choice of discretionary accruals?. Journal of Accounting and Economics, 22(1), 313-325.

Chatterjee, S., & Mächler, M. (1997). Robust regression: a weighted least squares approach. Communications in Statistics-Theory and Methods, 26(6), 1381-1394.

Chen, W., Hribar, P., & Melessa, S. (2017). Incorrect Inferences When Using Residuals as Dependent Variables.

Cohen, D. A., Dey, A., & Lys, T. Z. (2008). Real and accrual-based earnings management in the pre-and post-Sarbanes-Oxley periods. The accounting review, 83(3), 757-787. Collins, D. W., Pungaliya, R. S., & Vijh, A. M. (2016). The effects of firm growth and model

specification choices on tests of earnings management in quarterly settings. The Accounting Review, 92(2), 69-100.

DeAngelo, L. (1981). Auditor size and audit quality. Journal Of Accounting And Economics, 3(3), 183-199.

Dechow, P. M. (1994). Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals. Journal of accounting and economics, 18(1), 3-42.

Dechow, P., Ge, W., & Schrand, C. (2010). Understanding earnings quality: A review of the proxies, their determinants and their consequences. Journal of Accounting and

Economics 50,344-401.

Dechow, P.M. and I.D. Dichev, 2002, The Quality of Accruals and Earnings: The Role of Accrual Estimation Errors. The Accounting Review 77, 35-59.

DeFond, M., & Zhang, J. (2014). A Review of Archival Auditing Research. Journal of Accounting and Economics, 58(2–3), 275-326.

(33)

33 Desai, M. A., Foley, C. F., & Hines, J. R. (2004). A multinational perspective on capital structure choice and internal capital markets. The Journal of Finance, 59(6), 2451-2487.

Dopuch, N., & Simunic, D. (1980). The nature of competition in the auditing profession: a descriptive and normative view. Regulation and the accounting profession, 34(2), 283-289.

Easterwood, C. M. (2011). Takeovers and incentives for earnings management: an empirical analysis. Journal of Applied Business Research (JABR), 14(1), 29-48.

FASB ,1978, The Statement of financial Accounting Concepts, No.1, par.44

Fama, E. and J. MacBeth (1973). "Risk, return, and equilibrium: Empirical tests." Journal of Political Economy. 81: 607-636.

Francis, J. R., E. L. Maydew, and H. C. Sparks. 1999. The Role of Big 6 Auditors in The Credible Reporting of Accruals. Auditing: A Journal of Practice and Theory 18(2), 17-34.

Francis, J. R. (2004). What do we know about audit quality?. The British accounting review, 36(4), 345-368.

Frankel, R., & Sun, Y. (2018). Predicting accruals based on cash-flow properties. The Accounting Review.

Guay, W. R., Kothari, S. P., & Watts, R. L. (1996). A market-based evaluation of discretionary accrual models. Journal of accounting research, 83-105.

Harigaya, T. (2016). Effects of Digitization on Financial Behaviors: Experimental Evidence from the Philippines. Working Paper). Cambridge, MA: Harvard University.

Healy, P. M. (1985). The effect of bonus schemes on accounting decisions. Journal of accounting and economics, 7(1-3), 85-107.

Healy, P.M. and J.M. Wahlen, 1999, A Review of the Earnings Management Literature and Its Implications for Standard Setting. Accounting Horizons 13, 365-383.

Jere R. Francis and Michael D. Yu (2009) Big 4 Office Size and Audit Quality. The Accounting Review, 84(5), 1521-1552.

Johnstone, K. M., & Bedard, J. C. (2004). Audit firm portfolio management decisions. Journal of Accounting Research, 42(4), 659-690.

The association between discretionary accruals and future financial performance