We would like to thank Bill Brown, Ravi Dharwadkar, Randy Elder, and David Harris. Additionally, we thank the workshop participants at Syracuse University, State University of New York at Binghamton, Texas A & M University, University of South Florida, Villanova University, and attendees of the 2008 Financial Accounting Reporting Section Conference for helpful comments.

# Evidence on the role of accounting conservatism in monitoring managers’ investment decisions

Article first published online: 14 SEP 2010

DOI: 10.1111/j.1467-629X.2010.00369.x

© 2010 The Authors. Accounting and Finance © 2010 AFAANZ

Additional Information

#### How to Cite

Ahmed, A. S. and Duellman, S. (2011), Evidence on the role of accounting conservatism in monitoring managers’ investment decisions. Accounting & Finance, 51: 609–633. doi: 10.1111/j.1467-629X.2010.00369.x

#### Publication History

- Issue published online: 1 AUG 2011
- Article first published online: 14 SEP 2010
- Received 26 September 2009; accepted 28 May 2010 by Robert Faff (Editor).

- Abstract
- Article
- References
- Cited By

### Keywords:

- Accounting conservatism;
- Corporate governance;
- Agency costs

- G3;
- M41

### Abstract

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

Watts (2003), among others, argues that conservatism helps in corporate governance by mitigating agency problems associated with managers’ investment decisions. We hypothesize that if conservatism reduces managers’*ex ante* incentives to take on negative net present value projects and improves the *ex post* monitoring of investments, firms with more conservative accounting ought to have higher future profitability and lower likelihood (and magnitude) of future special items charges. Consistent with this expectation, we find that firms with more conservative accounting have (i) higher future cash flows and gross margins and (ii) lower likelihood and magnitude of special items charges than firms with less conservative accounting.

### 1. Introduction

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

Prior work suggests that conservatism is potentially useful in mitigating agency problems associated with managers’ investment decisions. In particular, Ball and Shivakumar (2005) argue that if managers know *ex ante* that losses will be recognized on a timely basis, they are (i) less likely to make negative net present value (NPV) investments and (ii) more likely to act quickly to limit economic losses from poorly performing investments. In addition to this *ex ante* role, Watts (2003) argues that conservatism provides directors and shareholders with timely signals for investigating the existence of negative NPV projects and taking corrective actions.

Providing evidence on whether conservatism plays a role in corporate governance is important for several reasons. First, while prior research documents a positive relationship between conservatism and the strength of board governance (Beekes *et al.*, 2004; Ahmed and Duellman, 2007; Garcia Lara *et al.*, 2009), it does not provide evidence on the specific benefits of conservatism in monitoring firms’ investment policies predicted by Watts (2003) and Ball and Shivakumar (2005). Furthermore, because of the endogeneity problems inherent in such association studies, one cannot make strong inferences about the direction of causality from these studies.

Second, our evidence on the benefits of conservatism in monitoring management investment decisions provides additional insight into the net benefits of conservatism which has been questioned by standard setters. For example, the FASB (2008, p. 28) in its Exposure Draft on the Conceptual Framework for Financial Reporting states that conservatism will ‘*lead to a bias in the reported financial position and financial performance*’ and ‘*this framework does not include prudence or conservatism as desirable qualities of financial reporting information*’. Commenting on such perceptions Watts (2003, p. 208) states: ‘*The long survival of conservatism and its apparent resilience to criticism strongly suggests that conservatism’s critics overlook its significant benefits. If regulator and standard-setter critics try to eliminate conservatism without understanding its benefits, the resultant standards are likely to be seriously detrimental to financial reporting*.’

If conservatism provides the monitoring/stewardship role predicted in the accounting literature, then the removal of conservatism from the conceptual framework is likely to increase the costs of monitoring managers. Thus, evidence on the benefits of conservatism in monitoring firms’ investments has important implications for standard setting.

Because investment decisions are not directly observable, we rely on two observable *ex post* manifestations of investment decisions, future profitability and future special items charges, to draw inferences about the benefits of conservatism in mitigating agency problems associated with managers’ investment decisions. We hypothesize that if conservatism facilitates governance in the manner predicted by Watts (2003) and Ball and Shivakumar (2005), then firms that use more conservative accounting should (i) have higher future profitability and (ii) take fewer and smaller special items charges in the future relative to firms that use less conservative accounting.

Utilizing a sample of US firms, we test our predictions on 23 681 firm-year observations from 1989–2001. In our primary tests, we use two conservatism measures: an asymmetric timeliness measure, following Roychowdhury and Watts (2007), and a market-value-based measure. We find that firms that use more conservative accounting have higher industry-adjusted future profitability measured by (i) operating cash flows (deflated by sales revenue or cash sales) and (ii) gross profit margins, up to 3 years in the future. Our profitability tests control for average performance over the current and past 2 years, standard deviation of performance, leverage, growth opportunities, and industry.

Consistent with conservatism reducing *ex ante* incentives to take on negative NPV projects and improving *ex post* monitoring of investments, we also find that firms with more conservative accounting have a significantly lower likelihood of taking a special items charge in the future and take special items charges of lower magnitudes than firms that use less conservative accounting. Our tests control for the determinants of special items charges identified in prior work including size, leverage, growth opportunities, variability of profits, and industry.

Taken together, the evidence from our profitability tests and the special items tests is consistent with conservatism playing an important role in providing managers with *ex ante* incentives to avoid negative NPV projects and in *ex post* monitoring of managers’ investment decisions as predicted in Watts (2003) and Ball and Shivakumar (2005).

We contribute to the literature in three ways. First, we not only provide evidence on the specific benefits of conservatism in mitigating agency problems associated with managers’ investment decisions but also are able to draw stronger inferences relative to prior studies by examining the relation between conservatism and *future* manifestations of investment policy. Second, our evidence on the benefits of conservatism in corporate governance challenges standard setters’ current views that conservatism is not useful. The elimination of conservatism is consistent with the FASB and the IASB removing stewardship as one of the objectives of the financial statements.^{1}

Third, our findings suggest that accounting choices affect managers’ investment/divestment decisions and their monitoring by governance mechanisms. In other words, accounting choices have ‘real’ effects on firms’ investment policies. This is similar to the notion in Biddle and Hilary (2006) that high quality accounting leads to greater efficiencies in the investment process. Consistent with conservatism being a characteristic of high quality accounting Garcia Lara *et al.* (2010) find evidence that conditional conservatism constrains over- and under-investment.^{2} Similarly, Bushman *et al.* (2007) find that conservatism, measured at the country level, is associated with investment efficiency.

Our study has a number of limitations. First, we do not examine the impact of conservatism on other manifestations of investment policy such as mergers and acquisitions, large capital expenditures, and divestitures.^{3} Examining all of these manifestations together with profitability and special items would be beyond the scope of a single study. Thus, our study should be viewed as an exploratory analysis of the predicted link conservatism and monitoring of investment decisions as captured by future profitability and special items.

A second limitation of our study is that because we require 3 years of prior and 3 years of future data for inclusion in our sample, we may have a survivorship bias. However, to assess the seriousness of this bias, we repeat our tests requiring only 1 year of future and 1 year of past data and obtain similar results. Therefore, our results are unlikely to be driven solely by survivorship bias. A third limitation of our study is that we utilize a sample of US firms, and the results may not be generalizable to other accounting regimes.

The remainder of this paper proceeds as follows. Section 2 presents the development of the hypotheses. Section 3 presents a discussion of the proxies for accounting conservatism and the research design. Section 4 presents the evidence. Section 5 presents a discussion of additional robustness tests. Section 6 presents the conclusion.

### 2. Hypotheses development

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

#### 2.1. Background

The separation of ownership and control in the modern corporation results in conflicts of interest arising between managers and other parties to the firm (Berle and Means, 1932; Jensen and Meckling, 1976). These conflicts cannot be resolved completely through contracts because it is costly, if not impossible, to write and enforce complete contracts (Fama and Jensen, 1983; Hart, 1995). Thus, in a world with incomplete contracts, corporate governance mechanisms arise to mitigate these conflicts (Shleifer and Vishny, 1997; Zingales, 1998; Hermalin and Weisbach, 2003). The optimal combination of governance mechanisms is chosen to maximize firm value and is likely to vary systematically across firms because these costs and benefits likely vary with firm characteristics such as the investment opportunity set, leverage, and the relative importance of external financing (Agrawal and Knoeber, 1996; Watts, 2006).

One important governance mechanism that is directly involved in monitoring managers is the board of directors (Fama and Jensen, 1983). Boards ratify and monitor top managers’ decisions because it is efficient to separate decision initiation and implementation from decision ratification and monitoring. Directors are given the power to hire and fire managers, determine managers’ compensation, and approve key decisions such as acceptance of major investment projects (Grinstein and Tolkowsky, 2004). Directors also advise managers on proposed strategies, provide outside expertise, and monitor the progress of major projects (Boone *et al.*, 2007; Coles *et al.*, 2008; Linck *et al.*, 2008).^{4}

To effectively monitor and advise managers, directors (particularly outside directors) need verifiable information. The accounting and financial reporting system is a critical source of verifiable information that is useful in monitoring and evaluating managers as well as their decisions and strategies (Watts and Zimmerman, 1986; Bushman and Smith, 2001). Furthermore, conservatism is an important characteristic of a firm’s accounting system that can help directors in reducing deadweight losses and disciplining other sources of information thereby increasing firm and equity values (Watts, 2003, 2006).

In the next two sub-sections, we discuss the specific benefits of conservatism in monitoring firms’ investments and the testable implications of these benefits.

#### 2.2. Benefits of conservative accounting in investment monitoring

Prior studies suggest that conservatism can improve investment efficiency in at least two ways. First, Ball and Shivakumar (2005) argue that if managers know *ex ante* that losses from poorly performing projects will be recognized during their tenure because of conservative accounting, their *ex ante* incentives to take on negative NPV projects are greatly reduced (Ball, 2001). In contrast, if managers can defer losses by using aggressive accounting, they will have incentives to accept negative NPV projects.

Second, Watts (2003) argues that managers have *ex post* incentives to hide losses from poorly performing projects to avoid being fired before their tenure is over. He suggests that conservatism provides directors and shareholders with a signal to investigate the reasons for the losses. Such investigations can lead to corrective actions such as abandonment of a negative NPV project or even discharging of managers responsible for such projects. Similarly, Ball and Shivakumar (2005) also argue that with conservative accounting, managers have incentives to act quickly to limit economic losses from poorly performing investments because without such action these losses would grow and be recognized during their tenure.

Taken together, the above arguments suggest that conservatism (i) provides incentives for *ex ante* efficient investment decisions and (ii) facilitates *ex post* monitoring of managers’ investment decisions by the board of directors.

One question that arises is how the level of conservatism is selected and why managers do not deviate from conservatism *ex post*. Efficient contracting theory suggests that managers benefit from reducing contracting and agency costs and thereby maximizing firm value. Furthermore, the board of directors can also direct managers to use conservative accounting as well as monitor accounting choices. Beekes *et al.* (2004) and Ahmed and Duellman (2007), among other studies, document that strong boards are associated with the use of conservative accounting. This is consistent with boards having some influence on accounting choices. Furthermore, managers may find deviating from conservatism costly in terms of the loss of reputation and/or credibility. Ultimately, whether these mechanisms are effective in constraining managers’ accounting choices and whether conservatism plays an important role in monitoring investment policy is an empirical question. Thus, assuming that conservatism plays the hypothesized role in governance discussed earlier, we develop testable implications and see if these implications are supported empirically.

### 2.3. Testable predictions

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

While the predicted benefits of conservatism discussed earlier are intuitively appealing, they cannot be directly tested because researchers cannot observe managers’ investment decisions or directors’ monitoring decisions. To draw inferences about these potential benefits of conservatism in corporate governance, we rely on two observable *ex post* manifestations of investment decisions: future profitability and future special items charges.

We choose these manifestations of investment decisions for two reasons. First, if conservatism results in *ex ante* avoidance of negative NPV projects and/or timely identification of poorly performing projects, this should result in higher future profitability, holding other things equal. Furthermore, under conservative accounting managers’ incentives to act quickly to limit economic losses from poorly performing projects will also lead to higher future profitability relative to firms that use aggressive accounting and as a result continue operating negative NPV projects.

Second, we expect that firms using conservative accounting will be less likely to take special items charges (such as an asset write-down or restructuring charge) if under conservative accounting negative NPV projects are avoided and/or corrective actions on projects that *ex post* become unprofitable are taken on a more timely basis.^{5} Thus, under conservative accounting, losses are unlikely to accumulate to a level that requires taking a special items charge or a one-time asset write-down. Furthermore, this also implies that any special items charge taken by a firm using conservative accounting will likely be for a smaller amount.

However, it should be noted that the use of conservative accounting may also cause management to forgo small positive NPV projects or prematurely terminate projects with negative cash-flow realizations in early periods. This effect offsets the benefits of conservative accounting in monitoring investment policy and reduces the likelihood of observing evidence consistent with the predicted benefits. To summarize, we test the following hypotheses:

*H1: Firms that use more conservative accounting will have higher future profitability than firms that use less conservative accounting.**H2: Firms that use more conservative accounting will take fewer special items charges, and charges of smaller magnitude, than firms that use less conservative accounting.*

### 3. Research design

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

#### 3.1. Proxies for accounting conservatism

We use two alternative proxies for conservatism in our main tests. The first proxy is an asymmetric timeliness proxy along the lines of Basu (1997) estimated using the backward cumulation procedure in Roychowdhury and Watts (2007).^{6} The rationale for the backward cumulation procedure is that although the Basu model captures the asymmetric verification standards of gains and losses it does not appropriately capture the conservatism prior to the estimation period (Pae *et al.*, 2005; Roychowdhury and Watts, 2007). Roychowdhury and Watts (2007) suggest that cumulating earnings and returns over the three prior years mitigates this limitation.^{7} Although a firm may gradually become more or less conservative over time, Watts and Zimmerman (1986) describe conservatism as a firm characteristic that is *relatively* stable as it is a function of static firm components such as size, age, industry, and debt structure.

We utilize the asymmetric timeliness measure of accounting conservatism as it is among the most prevalent measures of conservatism in the accounting literature. However, the asymmetric timeliness measure of conservatism may be biased when there is good news that is not related to earnings (Givoly *et al.*, 2007). Additionally, Dietrich *et al.* (2007) claim that the observed differences between good and bad news in the Basu regression is because of the sample truncation bias and sample variance ratio.

Our second proxy for conservatism is the industry-adjusted book-to-market ratio (multiplied by negative one). Intuitively, conservative accounting results in reducing book values relative to market values. Furthermore, book-to-market ratios capture the cumulative effects of conservatism since the inception of the firm. However, book-to-market ratios are also likely to capture economic rents and growth opportunities (Lindenberg and Ross, 1981). To mitigate the effects of these other determinants of book-to-market ratios, we use (i) an industry adjustment to control for industry-level growth opportunities and (ii) firm-specific controls for growth opportunities discussed in the following sections.

#### 3.2. Empirical tests using the asymmetric timeliness measure of conservatism

Consistent with previous research, such as Beekes *et al.* (2004), Ahmed and Duellman (2007), and Roychowdhury and Watts (2007), we interact our variables of interest with the measure of asymmetric timeliness to test our hypotheses. The full model to test for the relation between special items charges and future profitability and asymmetric timeliness is provided in equation (1). Consistent with the recommendations of Ryan (2006), we estimate several different specifications of the model.

- (1)

where *E*_{i,t,t−3} is income before extraordinary items cumulative from year *t* − 3 to year *t*, *P*_{i,t,t−4} is the market value of equity at the end of the year *t*, *D*_{i,t,t−3} is an indicator variable set equal to one if *R*_{i,t,t−3} is less than one, zero otherwise, *R*_{i,t,t−3} is the buy and hold return starting 4 months after the end of the fiscal year *t* − 3 and ending 4 months after the end of year *t*, *Profitability*_{i,t+3} is either 3-year ahead industry-adjusted *PM* or *CFO*, *PM*_{i,t+3} is one less cost of goods sold divided by total revenues, *CFO*_{i,t+3} is cash flows from operations divided by total revenue, *Special Items*_{i,t+1} is amount of special items taken in year *t* + 1 divided by sales and multiplied by negative one.

We expect future profitability (*Profitability*_{t+3}) to be positively related to the current level of asymmetric timeliness. Further, we expect *Special Items*_{t+1} to be negatively related to asymmetric timeliness. Consistent with the majority of asymmetric timeliness research we report the results without controls to highlight the variables of interest.

#### 3.3. Empirical tests using firm-specific measures of conservatism

##### 3.3.1. Conservatism and future profitability

To test for the effects of conservatism on future profitability, we employ the following regression:

- (2)

where *Profitability*_{i,t+3} is either *PM* or *CFO*, *PM*_{i,t+3} is one less cost of goods sold divided by total revenues, *CFO*_{i,t+3} is cash flows from operations divided by total revenue, *Conservatism*_{i,t} is the book-to-market ratio multiplied by −1, *Avg. Profitability*_{i,t} is the profitability measure averaged over years *t* to *t* − 2, *SD Profitability*_{i,t} is the standard deviation of the profitability measure over years *t* to *t* − 2, *Leverage*_{i,t} is total long-term liabilities divided by total sales, *Growth*_{i,t} is the percentage change in revenues from the previous year, *Size*_{i,t} is the natural log of total assets, *R&D + ADV*_{i,t} is research and development plus advertising expense divided by total sales, and *Past*_{i,t} is one if the firm takes a special items charge >1 per cent of total revenues during the time period *t* through *t* − 2, zero otherwise.

We control for industry in our tests, following Ahmed *et al.* (2002), by subtracting the industry median value of the variables from both the dependent and independent variables. We define industry as in Barth *et al.* (1999). The rationale for industry differencing is that the market-based conservatism measure reflects economic rents or growth opportunities as well as conservatism. Because the ability to earn economic rents and growth opportunities likely vary across industries, controlling for industry reduces the likelihood that the results are driven by these economic factors.

We utilize two different measures of future profitability: cash flows from operations divided by total revenues (*CFO*) and gross profit margin (*PM*). We deflate cash flows and profit margins by total revenue of the firm to avoid the problems associated with deflating by total assets. However, one drawback of deflating by total revenues is that firms using less conservative accounting may have less restrictive revenue recognition policies.

We use the average *ROE* (*PM*) over years *t* to *t* − 2, to control for prior profitability because firms that are profitable in the current period are likely to retain their operating edge in the short term. We expect the sign on prior profitability to be positive. Additionally, we utilize *Avg. ROE* as a predictor of future *CFO* as Dechow *et al.* (1998) find that current earnings more accurately predict future operating cash flows than current operating cash flows. We deflate our earnings measure by total market equity as deflation by average total assets may cause inflated profitability ratios for firms using conservative accounting.

We control for firm risk by including the standard deviation of *ROE* (*PM*); this is consistent with the proxy of operating performance risk utilized in Core *et al.* (1999). Firms with greater variance in their operating performance may be more likely to report either high or low earnings in any particular year. Therefore, we do not predict a sign on the coefficient of *SD ROE* (*SD PM*).

We control for capital structure by including *Leverage* as a control variable. Although interest expense is not included in both of the profitability measures utilized, the capital structure of the firm may affect the riskiness of investments undertaken. Firms that utilize more debt-based financing may be constrained by debt-holders from taking risky projects and therefore have lower future profitability. Furthermore, *Leverage* also controls for the investment opportunity set (Smith and Watts, 1992). We expect the coefficient on *Leverage* to be negative.

We control for firm growth, as firms in the growth stage of the business life cycle frequently have lower profitability than mature firms. Anthony and Ramesh (1992) document evidence on the effect of the firm life cycle on profitability. Therefore, we expect the coefficient on *Growth* to be negative. We control for firm size (*Size*) as Fama and French (1995) find that firm size is positively related to future accounting performance. Thus, we predict a positive coefficient on *Size*.

We control for research and development and advertising expense (*R&D* + *ADV*) because Lev and Sougiannis (1996) find a relation between research and development costs and future profitability. Furthermore, this variable also controls for growth opportunities (Ahmed, 1994). However, we do not predict a sign on this coefficient because firms that are R&D-intensive at time *t* are likely R&D-intensive at time *t* + 3 which could reduce future profitability. We also include a control for prior large special items charges (*Past*) as firms that take special items charges may have greater future performance because of the divestment of poorly performing operating segments (Atiase *et al.*, 2004). However, because a firm taking special items charges may be taking these charges to manage earnings we do not make a prediction about the sign of the coefficient.

##### 3.3.2. Conservatism and the likelihood of future special items charges

To test for the effects of conservatism on the likelihood of future special items charges, we employ the following logistic regression:

- (3)

where *Future*_{i,t} is equal to one if the firm takes a special items charge in the time period *t* + 1, zero otherwise,^{8}*Conservatism*_{i,t} is the book-to-market ratio multiplied by −1, *Past*_{i,t} is set equal to one if the firm has taken a special items charge in the time period *t* through *t* − 2 and is zero otherwise, *Avg. ROE*_{i,t} is income from operations plus special items divided by the fiscal year end market value of equity and averaged over years *t* to *t* − 2, *SD ROE*_{i,t} is the standard deviation of ROE over the time period *t* to *t* − 2, *Return*_{i,t} is the buy and hold market return for fiscal year *t* beginning 4 months after the fiscal year end, *Leverage*_{i,t} is total long-term liabilities divided by total sales, *Growth*_{i,t} is the percentage change in revenues from the previous year, and *Size*_{i,t} is the natural log of total assets. We additionally include industry and year dummy variables to control for industry and year effects. Industry is defined consistent with Barth *et al.* (1999).^{9}

We control for prior special items charges in the time period *t* to *t* − 2 (*Past*), because some firms take special items charges frequently (Elliott and Hanna, 1996).^{10} Thus, we expect firms that have taken special items charges in the past to be more likely to take a special items charge in the future.

We control for past profitability (*Avg. ROE*) because firms that restructure have lower earnings than peer firms in the period before the special items charge (Brickley and Van Drunen, 1990). Therefore, we predict a negative coefficient on prior profitability. We control for the standard deviation of profitability (*SD ROE*) because firms with more volatile profitability are more likely to have special items charges than other firms. Conversely the operating uncertainty may cause the firms to restructure and take special items charges. We control for the stock market return (*Return*) because Francis *et al.* (1996) find that the market return is negatively related to the size of the special items charge.

We control for leverage (*Leverage*), because it is a proxy for risk as well as the investment opportunity set. Furthermore, holders of debt may monitor the firm to ensure that assets are written off in a timely manner. We control for firm growth (*Growth*) because firms in the growth stage should be less likely to take a special items charge. However, rapidly growing firms may need to restructure their operations because of their changing business dynamics. Therefore, we do not predict a sign on the coefficient for *Growth*. Finally, we control for firm size (*Size*) because large firms have a greater scope of operation and thus are more likely to have a special items charge in any given year.

### 4. Results

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

#### 4.1. Sample selection and descriptive statistics

To test our predictions, we utilize a sample of US firms listed in Compustat with available data from 1989–2001.^{11} To be included in our sample, we require that a firm has all data items available for 3 years before (after) year *t* to measure past (future) profitability and special items charges. All variables are obtained from Compustat except for *Return*, which is obtained from CRSP. We exclude financial service and insurance firms from the analysis as the classification between operating, investing, and financing activities is ambiguous for these firms (Richardson, 2006). Furthermore, we eliminate observations where income from operations plus special items divided by average total assets is less than −0.50 to control for outliers.^{12}

Cohen *et al.* (2003) discuss the justification for restricting the data range and recommend eliminating the observation if the observation is ±3 standard deviations from the mean. Cohen *et al.* (2003) contend that the researcher can justify these truncations if (i) the initial sample including outliers skews the results and interpretation, or (ii) the data truncated is outside the area of interest for the study. The justification for our truncation is that the firms with very large losses are typically young growth/IPO firms with small asset bases investing their resources in research and development (Darrough and Ye, 2007). Because these firms typically have higher managerial ownership than other firms (McGuire, 2000), the benefits of conservative accounting in alleviating information asymmetry or in monitoring managers’ investment decisions would be reduced for such firms.^{13}

After imposing the data restrictions we have a sample of 56 980 firm-years over the time period 1989–2001. We lose 8721 firm-years because of the profitability restriction and 24 578 firm-years because of the requirement of having seven consecutive years of data. Thus, we have a final sample of 23 681 firm-years.

Table 1 presents the industry-adjusted descriptive statistics for the sample. We Winsorize all variables at the top and bottom 1 per cent to mitigate the effects of extreme observations. To ensure that our sample is consistent with previous research we compare the means and medians of our sample prior to the industry differencing to relevant papers in the accounting literature.

Mean | SD | Min | 25% | Median | 75% | Max | |
---|---|---|---|---|---|---|---|

^{}We utilize a sample of Compustat and CRSP firms over 1989–2001. We require that the firm has all data items available for 3 years before (after) year *t*to measure past (future) profitability and special items charges. We exclude financial service firms because of the difference in reporting requirements and profitability. We also exclude observations where income from operations plus special items divided by average total assets is less than −0.50. All variables are obtained from Compustat except for*Return*, which is obtained from CRSP.*Conservatism*is the book-to-market ratio multiplied by −1,*Avg. ROE*is income from operations plus special items divided by the fiscal year end market value of equity and averaged over years*t*to*t*− 2,*Avg. CFO*is cash flows from operations divided by sales averaged over years*t*to*t*− 2,*Avg. PM*is one less cost of goods sold divided by sales averaged over years*t*to*t*− 2,*SD ROE*is the standard deviation of ROE over years*t*to*t*− 2,*SD PM*is the standard deviation of PM over years*t*to*t*− 2,*Return*is the buy and hold market return for fiscal year*t*beginning 4 months after the fiscal year end,*Leverage*is total long-term liabilities divided by total sales,*Growth*is the percentage change in revenues from the previous year,*Size*is the natural log of total assets,*R&D + ADV*is research and development plus advertising divided by total sales,*Past*is one if the firm takes a special items >1 per cent of total revenues during the time period*t*through*t*− 2, zero otherwise, and*Special Items*is special items multiplied by −1 divided by total revenues.
| |||||||

Conservatism | −0.124 | 0.514 | −2.651 | −0.297 | 0.000 | −0.195 | 1.051 |

Avg. ROE | −0.031 | 0.124 | −0.714 | −0.039 | 0.000 | 0.024 | 0.212 |

Avg. CFO | 0.006 | 0.092 | −0.553 | −0.040 | 0.000 | 0.045 | 0.496 |

Avg. PM | −0.007 | 0.105 | −0.523 | −0.036 | 0.000 | 0.036 | 0.415 |

SD ROE | 0.055 | 0.155 | −0.147 | −0.018 | 0.000 | 0.055 | 0.966 |

SD PM | 0.034 | 0.054 | −0.098 | −0.002 | 0.000 | 0.060 | 0.425 |

Return | 0.091 | 0.527 | −1.276 | −0.221 | 0.000 | 0.266 | 2.737 |

Leverage | 0.033 | 0.166 | −0.347 | −0.090 | 0.000 | 0.126 | 0.749 |

Growth | 0.035 | 0.240 | −1.051 | −0.082 | 0.000 | 0.107 | 1.339 |

Size | 0.097 | 1.937 | −5.646 | −1.279 | 0.000 | 1.355 | 6.464 |

R&D + ADV | 0.018 | 0.056 | −0.094 | −0.002 | 0.000 | 0.027 | 0.323 |

Past | 0.399 | 0.490 | 0.000 | 0.000 | 0.000 | 1.000 | 1.000 |

Special Items_{t+1} | 0.007 | 0.030 | −0.105 | 0.000 | 0.000 | 0.004 | 0.154 |

The means and medians of future profitability, *Avg. PM* and *Avg. CFO*, prior to industry differencing (unreported), are consistent with the levels reported in Richardson *et al.* (2005) despite slightly different time periods.^{14} Furthermore, the percentage of firms taking special items charges (*Past*), and the magnitudes of these charges (*Special Items*), is consistent with the levels reported in Elliott and Hanna (1996) and McVay (2006). Finally, the means and medians of *Conservatism* and the control variables are consistent with previous research.

Table 2 presents the correlations between the market-based conservatism measure, profitability and the control variables. *Conservatism* is positively correlated with the size of a future special items charges using Spearman correlation and negatively (albeit insignificant) correlated using Pearson correlation. Consistent with expectations, *Conservatism* is positively correlated with *Avg. ROE* and *Avg. PM* consistent with Fama and French (1995). The univariate results should be interpreted with caution as they likely suffer from an omitted variables bias.

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | ||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

^{}We utilize a sample of Compustat and CRSP firms from 1989–2001. *Conservatism*is the book-to-market ratio multiplied by −1,*Avg. ROE*is income from operations plus special items divided by the fiscal year end market value of equity and averaged over years*t*to*t*− 2,*Avg. CFO*is cash flows from operations divided by sales averaged over years*t*to*t*− 2,*Avg. PM*is one less cost of goods sold divided by sales averaged over years*t*to*t*− 2,*SD ROE*is the standard deviation of ROE over years*t*to*t*− 2,*SD PM*is the standard deviation of PM over years*t*to*t*− 2,*Return*is the buy and hold market return for fiscal year*t*beginning 4 months after the fiscal year end,*Leverage*is total long-term liabilities divided by total sales,*Growth*is the percentage change in revenues from the previous year,*Size*is the natural log of total assets,*R&D + ADV*is research and development plus advertising divided by total sales,*Past*is one if the firm takes a special items >1 per cent of total revenues during the time period*t*through*t*− 2, zero otherwise, and*Special Items*is special items multiplied by −1 divided by total revenues.
| ||||||||||||||

1 | Conservatism | 1 | 0.04 | 0.23 | 0.21 | −0.34 | −0.02 | 0.34 | 0.01 | 0.27 | 0.24 | 0.16 | −0.01 | 0.01 |

2 | Avg. CFO | 0.15 | 1 | 0.25 | 0.01 | −0.44 | −0.11 | 0.11 | −0.05 | 0.09 | 0.08 | −0.12 | −0.47 | −0.06 |

3 | Avg. ROE | 0.15 | 0.22 | 1 | 0.35 | −0.32 | −0.03 | 0.10 | −0.03 | 0.03 | 0.33 | 0.02 | −0.02 | 0.02 |

4 | Avg. PM | 0.17 | 0.07 | 0.36 | 1 | −0.18 | 0.18 | 0.03 | −0.07 | 0.06 | 0.01 | 0.30 | 0.11 | 0.08 |

5 | SD ROE | −0.31 | −0.73 | −0.19 | −0.12 | 1 | 0.09 | −0.13 | 0.12 | −0.23 | −0.19 | −0.01 | 0.40 | 0.00 |

6 | SD PM | −0.06 | −0.12 | −0.02 | 0.16 | 0.14 | 1 | −0.02 | 0.01 | −0.01 | −0.10 | 0.00 | 0.13 | 0.09 |

7 | Return | 0.28 | 0.06 | 0.04 | 0.05 | −0.04 | 0.04 | 1 | −0.02 | 0.24 | 0.06 | 0.01 | −0.07 | −0.04 |

8 | Leverage | 0.05 | −0.10 | −0.01 | −0.05 | 0.11 | 0.02 | −0.01 | 1 | 0.05 | 0.28 | −0.15 | 0.10 | 0.05 |

9 | Growth | 0.21 | 0.12 | 0.00 | 0.06 | −0.15 | 0.06 | 0.20 | 0.09 | 1 | 0.03 | 0.04 | −0.11 | 0.00 |

10 | Size | 0.22 | 0.11 | 0.27 | 0.01 | −0.14 | −0.09 | −0.01 | 0.23 | 0.01 | 1 | −0.02 | 0.11 | 0.10 |

11 | R&D + ADV | 0.12 | −0.05 | −0.03 | 0.33 | −0.01 | 0.03 | 0.03 | −0.15 | 0.01 | −0.07 | 1 | 0.06 | 0.04 |

12 | Past | −0.02 | −0.38 | −0.02 | 0.11 | 0.31 | 0.12 | −0.04 | 0.10 | −0.05 | 0.12 | 0.05 | 1 | 0.12 |

13 | SpecialItems_{t+1} | −0.02 | −0.03 | 0.04 | 0.10 | 0.01 | 0.08 | −0.03 | 0.02 | 0.01 | 0.07 | 0.03 | 0.12 | 1 |

#### 4.2. Tests based on the asymmetric timeliness measure of conservatism

In this section, we test our hypotheses using the asymmetric timeliness measure of conservatism. Consistent with Roychowdhury and Watts (2007), we estimate asymmetric timeliness cumulatively over multiple periods. Table 3 reports the results of our tests using the asymmetric timeliness of earnings as our measure of conservatism (equation (1)).

Profitability measure | (i) | (ii) | (iii) | (iv) | (v) | (vi) | |
---|---|---|---|---|---|---|---|

– | CFO_{t+3} | PM_{t+3} | – | CFO_{t+3} | PM_{t+3} | ||

^{}Regression coefficients are presented in the same row as the variable name and Newey–West corrected *t*-statistics are reported in parentheses. Significance is based on two-tailed*t*-tests. */**/*** represents significance at the 10/5/1 per cent level.*ROE*is income from operations plus special items divided by the fiscal year end market value of equity,*CFO*is cash flows from operations divided by sales,*PM*is one less cost of goods sold divided by sales,*Dummy*is a dichotomous variable equal to one (zero) if the cumulative buy and hold return is less (greater) than zero,*Return*is the cumulative buy and hold return during the years*t*− 3 to*t*beginning 4 months after the fiscal year end,*Profitability*is the profitability measure listed at the top of the column, and*Special Items*is special items from year*t*+ 1 multiplied by −1 divided by total revenues.
| |||||||

Intercept | ? | 0.227 (46.09)*** | 0.203 (31.03)*** | 0.259 (23.98)*** | 0.228 (44.91)*** | 0.203 (30.72)*** | 0.260 (24.02)*** |

Dummy | ? | 0.058 (4.63)*** | 0.062 (3.44)*** | 0.048 (1.51) | 0.059 (4.58)*** | 0.063 (3.48)*** | 0.050 (1.56) |

Return | + | 0.122 (46.93)*** | 0.139 (40.55)*** | 0.185 (31.71)*** | 0.124 (45.54)*** | 0.141 (40.12)*** | 0.185 (31.71)*** |

Dummy*Return | + | 0.434 (31.51)*** | 0.487 (23.81)*** | 0.444 (14.91)*** | 0.442 (31.14)*** | 0.495 (23.83)*** | 0.452 (15.11)*** |

Profitability | ? | 0.229 (5.70)*** | −0.104 (−3.88)*** | 0.229 (5.69)*** | −0.103 (−3.81)*** | ||

Profitability*Dummy | ? | −0.028 (−0.18) | 0.011 (0.12) | −0.035 (−0.22) | 0.010 (0.11) | ||

Profitability*Return | ? | −0.159 (−7.90)*** | −0.150 (−11.45)*** | −0.158 (−7.83)*** | −0.149 (−11.26)*** | ||

Profitability*Dummy*Return | + | 0.348 (2.39)** | −0.087 (−0.93) | 0.354 (2.43)** | −0.082 (−0.88) | ||

Special Items | ? | −0.147 (−0.92) | −0.170 (−1.07) | −0.142 (−0.89) | |||

Special Items*Dummy | ? | −0.226 (−0.49) | −0.221 (−0.48) | −0.296 (−0.65) | |||

Special Items*Return | ? | −0.155 (−2.08)** | −0.136 (−1.82)* | −0.023 (−0.30) | |||

Special Items*Dummy*Return | − | −1.304 (−2.54)** | −1.355 (−2.65)*** | −1.441 (−2.82)*** | |||

Sample period | 1989–2001 | 1989–2001 | 1989–2001 | 1989–2001 | 1989–2001 | 1989–2001 | |

N | 23 681 | 23 681 | 23 681 | 23 681 | 23 681 | 23 681 | |

AdjustedR^{2} | 0.2000 | 0.210 | 0. 202 | 0.201 | 0.211 | 0.203 |

Column (i) provides the basic regression as seen in Table 3 of Roychowdhury and Watts (2007). The coefficient on *Dummy * Return* is 0.434 compared to the coefficient of 0.332 found in Roychowdhury and Watts (2007).

Columns (ii) and (iii) present our tests of H1, which predicts that firms that use more conservative accounting will have higher future profitability. Consistent with H1, the coefficient on *Profitability * Dummy * Return*, 0.348, is positive and significant in column (ii) when future profitability is measured by cash flows from operations. However, the coefficient on *Profitability * Dummy * Return*, −0.087, is not significantly different from zero in column (iii) when future profitability is measured by future profit margins. In column (iv), we present our test of H2 that predicts that firms with more conservative accounting take smaller special items charges. Consistent with our prediction, the coefficient on *Special Items * Dummy * Return*, −1.304, is negative and significant.^{15} Columns (v) and (vi) present the simultaneous tests of H1 and H2. The inferences from the full model are consistent with those previously discussed.^{16}

Overall, using the asymmetric timeliness measure of conservatism our results are consistent with the benefits of conservative accounting in corporate governance as predicted by Ball (2001), Watts (2003), and Ball and Shivakumar (2005).

#### 4.3. Evidence on the relation between conservatism and future profitability

In the previous tests, we used the asymmetric timeliness measure of conservatism. Table 4 reports the results of regressions of future profitability on conservatism and the control variables (equation (2)) on the 13 year sample period. Columns (i) and (ii) present the Fama–MacBeth regressions of industry-adjusted 3-year ahead *CFO* and *PM* on industry-adjusted conservatism and controls. Columns (iii) and (iv) present the pooled regression of industry-adjusted 3-year ahead *PM* on industry-adjusted conservatism and controls with Newey–West *t*-statistics.

Dependent variable | Fama–MacBeth estimation | Pooled estimation | |||
---|---|---|---|---|---|

(i) | (ii) | (iii) | (iv) | ||

CFO_{t+3} | PM_{t+3} | CFO_{t+3} | PM_{t+3} | ||

^{}The Fama–MacBeth regressions show the mean of coefficient estimates obtained from estimating equation (2) on each yearly sample of the 23 681 firm-years; and the adjusted *R*^{2}is the average*R*^{2}of the yearly regressions. Regression coefficients are presented in the same row as the variable name and Newey–West corrected*t*-statistics are reported in parentheses. Significance is based on two-tailed*t*-tests. */**/*** represents significance at the 10/5/1 per cent level.*ROE*is income from operations plus special items divided by the fiscal year end market value of equity,*CFO*is cash flows from operations divided by sales,*PM*is one less cost of goods sold divided by sales,*Conservatism*is the book-to-market ratio multiplied by −1,*Avg. ROE*is income from operations plus special items divided by the fiscal year end market value of equity and averaged over years*t*to*t*− 2,*SD ROE*is the standard deviation of ROE over years*t*to*t*− 2,*Avg. PM*is one less cost of goods sold divided by sales averaged over years*t*to*t*− 2,*SD ROE*is the standard deviation of ROE over years*t*to*t*− 2,*SD PM*is the standard deviation of PM over years*t*to*t*− 2,*Leverage*is total long-term liabilities divided by total sales,*Growth*is the percentage change in revenues from the previous year,*Size*is the natural log of total assets,*R&D + ADV*is research and development plus advertising divided by total sales, and*Past*is one if the firm takes a special items >1 per cent of total revenues during the time period*t*through*t*− 2, zero otherwise.
| |||||

Intercept | ? | 0.026 (9.17)*** | −0.028 (−9.30)*** | 0.024 (−33.78)*** | −0.013 (−3.59)*** |

Conservatism_{t} | + | 0.022 (8.00)*** | 0.044 (4.61)*** | 0.019 (12.51)*** | 0.042 (19.40)*** |

Avg. ROE_{t} | + | 0.044 (3.99)** | – | 0.049 (5.05)*** | – |

SD ROE_{t} | ? | −0.019 (−2.20)** | – | −0.018 (−2.24)** | – |

Avg. PM_{t} | + | – | 0.809 (21.49)*** | – | 0.732 (26.39)*** |

SD PM_{t} | ? | – | 0.330 (7.85)*** | – | 0.330 (14.43)*** |

Leverage_{t} | − | −0.021 (−4.21)*** | −0.010 (−1.78)* | −0.017 (−3.78)*** | 0.005 (0.73) |

Growth_{t} | − | −0.001 (−0.23) | −0.017 (−5.28)*** | 0.008 (1.08) | −0.008 (−1.61) |

Size_{t} | + | 0.010 (19.10)*** | 0.000 (0.08) | 0.009 (25.64)*** | −0.004 (−6.80)*** |

R&D + ADV_{t} | ? | −0.036 (−0.98) | 0.185 (7.41)*** | −0.032 (2.07)** | 0.176 (10.41)*** |

Past_{t} | ? | 0.000 (0.58) | 0.001 (0.36) | 0.001 (0.36) | 0.028 (5.71)*** |

Included dummies | Industry | Industry | Industry and year | Industry and year | |

AdjustedR^{2} | 0.086 | 0.227 | 0.082 | 0.216 |

Consistent with H1, *Conservatism* is positively related to 3-year ahead profitability at the 1 per cent level of significance for both cash flows from operations and profit margins at the 1 per cent level of significance in all four columns. Thus, using both the asymmetric timeliness measure of conservatism and unconditional measure of conservatism, we obtain support for H1.^{17} This result is consistent with the findings of Fama and French (1995) who find that the market correctly projects future growth rates based on portfolios of book-to-market. However, our study is different in its research design as we use regression analysis and control for industry and other factors such as growth, leverage, and prior profitability. Thus, our tests and findings are fundamentally different from those reported in Fama and French (1995).

The coefficients on the control variables are generally consistent with expectations. However, the results for the control variables are sensitive to the estimation method used (Fama–MacBeth vs. Pooled regression). For example, in column (iv), the coefficient on *Size* is negative and significant whereas it is positive in the other three columns. Similarly, the coefficient on *Leverage* is positive but not significant in column (iv) and negative and significant in the other three columns. Furthermore, *Growth* is negative and significant in column (ii) but unrelated to future profitability in the other three columns. Finally, the coefficient on *Past* is positive and significant in column (iv) but not significantly different from zero in the other columns.

#### 4.4. Evidence on the relation between conservatism and future special items charges

The results of the logistic regression equation (3) on the likelihood of future special items charges on conservatism and controls are presented in Table 5. Column (i) presents the logistic regression of future special items > 0 per cent of total revenues on conservatism and controls. Column (ii) presents the logistic regression of future special items >1 per cent of total revenues on conservatism and controls. Columns (iii) and (iv) present the logistic regression of future special items on conservatism for a sub-sample controlling for executive turnover. When the logistic regression coefficients are positive, it indicates a greater likelihood of future special items charges.

Future_{t+1} | (i) | (ii) | (iii) | (iv) | |
---|---|---|---|---|---|

0% of total revenues | 1% of total revenues | 0% of total revenues | 1% of total revenues | ||

^{}Regression coefficients are presented in the same row as the variable name and chi-squares are reported in parentheses. Significance is based on two-tailed Chi-squares. */**/*** represents significance at the 10/5/1 per cent level. *Future*is one if the firm takes a special items charge greater than the amount specified in the column during the time period*t*+ 1,*Conservatism*is the book-to-market ratio multiplied by −1,*ROE*is ROE before the special items charge,*CEO Turnover*is a dichotomous variable equal to one if the CEO was replaced in the last two fiscal years, zero otherwise,*Past*is one if the firm takes a special items greater than the amount specified in the column during the time period*t*through*t*− 2, zero otherwise,*Avg. ROE*is income from operations plus special items divided by the fiscal year end market value of equity and averaged over years*t*to*t*− 2,*SD ROE*is the standard deviation of ROE over years*t*to*t*− 2,*Return*is the buy and hold market return for fiscal year*t*beginning 4 months after the fiscal year end,*Leverage*is total long-term liabilities divided by total sales,*Growth*is the percentage change in revenues from the previous year, and*Size*is the natural log of total assets.
| |||||

Intercept | ? | −2.135 (709.72)*** | −2.639 (817.24)*** | −1.548 (38.62)*** | −2.503 (78.27)*** |

Conservatism_{t} | − | −0.071 (4.89)** | −0.124 (11.60)*** | −0.272 (5.07)** | −0.493 (11.42)*** |

ROE_{t+1} | − | − | − | −5.827 (262.58)** | −5.298 (307.36)*** |

CEO Turnover_{t} | + | − | − | 0.179 (3.97)** | 0.214 (4.06)*** |

Past_{t} | + | 0.750 (532.73)*** | 0.658 (315.09)*** | 0.950 (120.92)*** | 0.858 (86.40)*** |

Avg. ROE_{t} | − | −0.512 (8.20)*** | −0.632 (9.26)*** | – | – |

SD ROE_{t} | ? | −0.078 (0.29) | −0.300 (3.05)* | −0.479 (8.04)*** | −0.498 (4.52)*** |

Return_{t} | − | −0.141 (22.47)*** | −0.202 (33.30)*** | −0.054 (0.53) | −0.005 (0.01) |

Leverage_{t} | ? | 0.259 (8.01)*** | 0.073 (0.48) | 0.737 (9.27)*** | 0.149 (0.29) |

Growth_{t} | ? | 0.152 (5.87)** | 0.161 (5.20)** | 0.242 (2.26) | 0.139 (0.64) |

Size_{t} | + | 0.150 (491.99)*** | 0.156 (236.03)*** | 0.154 (31.27)*** | 0.159 (27.48)*** |

Industry and yeardummies | Yes | Yes | Yes | Yes | |

Sample period | 1989–2001 | 1989–2001 | 1997–2001 | 1997–2001 | |

# of Firms with Special Items > the specified % | 7609 | 4803 | 1799 | 1096 | |

N | 23 681 | 23 681 | 3933 | 3933 | |

Pseudo R^{2} | 0.130 | 0.100 | 0.252 | 0.258 |

Consistent with H2, *Conservatism* is negatively related to the likelihood of a future special items charge at the 5 per cent level of significance in column (i) and at the 1 per cent level in column (ii). As in Elliott and Hanna (1996), the sign on the coefficient of previous special items charges (*Past*) is consistent with firms making frequent use of special items charges. The coefficient on *Avg. ROE* in columns (i) and (ii) indicates that more profitable firms are less likely to take a future special items charge. The coefficient on *Return* is negative and significant at the 1 per cent level in both columns (i) and (ii) consistent with the market anticipating bad news. *Leverage* is unrelated to future special items in column (ii) and positively related to future special items charges in column (i). *Growth* is positively related to future special items charges in column (i) at the 5 per cent level of significance. The coefficient on *Size* is positive and significant consistent with larger firms taking more special items charges. Overall, these results are consistent with the notion that conservatism helps prevent special items through timelier recognition of losses.

In columns (iii) and (iv), we include two controls for management’s incentive to take special items charges consistent with Francis *et al.* (1996). These controls are profitability before the special items charge in year *t* + 1 and management turnover. Firms that are underperforming expectations may take special items charges and shift future charges into the current period (Francis *et al.*, 1996; McVay, 2006). To control for future profitability, we use the return on equity before the special items charge in year *t* + 1 consistent with Francis *et al.* (1996).^{18} Additionally, management turnover may cause increased asset write-downs as the new management team may want to change the strategic focus of the firm and rid themselves of negative NPV projects (Francis *et al.*, 1996). To control for management turnover, we include a dummy variable that takes the value of one if the CEO has been replaced in the previous two fiscal years.

We obtain the CEO turnover data from Execucomp which provides executive compensation data on S&P 1500 firms. The inclusion of the Executive turnover variable reduces our sample size to 3933 firm-years from 1997–2001. Because of the differences in samples the coefficients differ significantly across comparable columns. However, we continue to find a negative and significant coefficient on *Conservatism* at the 5 per cent level of significance. Furthermore, we find that management turnover (*CEO Turnover*) is positively related to the likelihood of a special items charge while future profitability (*ROE*_{t+1}) is negatively related to the likelihood of a special items charge.

### 5. Additional robustness checks

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

In this section, we report the results of additional robustness checks. First, we repeat our tests with an accrual-based measure of conservatism. This measure of conservatism, *CON-ACC*, is defined as income before extraordinary items less cash flows from operations plus depreciation expense and special items deflated by average total assets, and averaged over years *t* to *t* − 2, multiplied by negative one. Positive values of *CON-ACC* indicate greater conservatism. This measure has been previously used in Givoly and Hayn (2000) and Ahmed *et al.* (2002). A predominance of negative accruals over a period of time is evidence of conservatism because over the long run earnings and cash flow are equal (Givoly and Hayn, 2000). By averaging accruals over time, we are more likely to capture the effects of conservatism rather than the effects of any potential earnings management because accruals tend to reverse within one to two years (Richardson *et al.*, 2005). However, one drawback of this measure is that it does not reflect the level of conservatism in prior periods.

Another drawback is a potential mechanical relation between *CON-ACC* and future special items charges. The mechanical relation would occur if the use of conservative accounting obviated the need for special items charges. However, the difference in *CON-ACC* (unreported) between firms that take a special items charge and firms that do not take a special items charge is 0.32 per cent of total revenue. This value is well below that of the average special items charge (unreported), 3.23 per cent of total revenue, for the subset of firms taking a special items charge. In other words, negative operating accruals are unlikely to substitute for special items charges. Thus, our results using the *CON*-*ACC* measure of conservatism are unlikely to be caused by a mechanical link between conservatism and future special items charges. Our tests based on this alternative firm-specific measure of conservatism (not reported) yield similar inferences to those reported in Tables 4 and 5. Thus, we obtain consistent results using three distinct proxies for accounting conservatism.

Second, a potential limitation of our study is that we require 7 years of consecutive data, and therefore our results may be affected by a survivorship bias. To alleviate these concerns, we repeat our tests using the firm-specific measures of conservatism on a less restrictive sample requiring only 3 years of consecutive data (*t* − 1 thru *t* + 1).^{19} Under these specifications, we have a sample of 45 765 firm-years. Overall, the results based on this less restrictive are qualitatively similar to those reported in the paper.

Third, we test for the effects of conservatism on the *size* of the special items charges (rather than just a dichotomous variable indicating the presence of special items). We employ the following regression:

- (4)

where *Special Items*_{i,t+1} is special items multiplied by −1 divided by total revenue and all other variables are as previously defined. Consistent with McVay (2006), we deflate special items charges by total revenues. We find that *Conservatism* is negatively related to the size of the special items charge at the 5 per cent level of significance. This finding is consistent with accounting conservatism assisting managers in abrogating negative NPV projects before losses accumulate. When we estimate equation (4) including *CON-ACC* as an explanatory variable, we also find a negative and significant relation between *CON-ACC* and the size of the special items charge.

Fourth, our reported results focus on the relation between conservatism and 1-year ahead special items charges. As an additional robustness check, we test if conservatism is related to 2-year (*t* + 2) and 3-year (*t* + 3) ahead special items charges. Our book-to-market measure of conservatism (*Conservatism*) is unrelated to the likelihood of future special items charges in years *t* + 2 and *t* + 3. However, *CON-ACC* is negatively related to the likelihood of special items charges in years *t* + 2 and *t* + 3. Similarly, *CON-ACC* is negatively related to the size of special items charges in years *t* + 2 and *t* + 3 while our book-to-market measure of conservatism is insignificant in these regressions. Finally, we find that the asymmetric timeliness of earnings is negatively related to future special items charges in year *t* + 2. However, the asymmetric timeliness of earnings is unrelated to future special items charges in year *t* + 3. Overall, our results are consistent with conservatism assisting directors in terminating negative NPV projects in a timelier manner.

### 6. Conclusion

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

Relying on two observable *ex post* manifestations of investment decisions, future profitability and future special items charges, we provide evidence on the potential benefits of conservatism in mitigating agency problems associated with managers’ investment decisions predicted by Watts (2003) and Ball and Shivakumar (2005). We find that relative to firms with less conservative accounting, firms with more conservative accounting have significantly higher profitability up to 3 years in the future. Furthermore, we find that firms with more conservative accounting have a significantly lower likelihood of taking a special items charge in the future and take special items charges of significantly lower magnitudes than firms that use less conservative accounting. Our tests control for the determinants of future profitability and special items charges identified in prior work including size, leverage, growth opportunities, variability of profits, and industry. Taken together, the evidence is consistent with the role of conservatism in corporate governance predicted in Watts (2003) and Ball and Shivakumar (2005).

- 1
The removal of stewardship as a principal objective of accounting caused two IASB members to issue alternative views to the released preliminary viewpoint.

- 2
Additionally, Garcia Lara

*et al.*(2010) find evidence of a positive relation between conditional conservatism and future profitability as measured by returns and profit margin. We distinguish ourselves from this concurrent research by focusing on investment outcomes and the use of conservative accounting to quickly uncover poorly performing projects allowing management to act in a timely basis; while Garcia Lara*et al.*(2010) focus on the relation between conservatism and over/under-investment. - 3
Francis and Martin (2010) find that firms that use more timely incorporation of economic losses make acquisition decisions that result in higher post-merger operating performance and are less likely to divest the acquired firm.

- 4
Recent theoretical work on boards includes Gillette

*et al.*(2003), Harris and Raviv (2008), Raheja (2005), and Adams and Ferreira (2007). - 5
Managers may take these negative NPV projects as their perquisites often increase with investment causing management to over-invest (Jensen, 1986; Stulz, 1990).

- 6
Ryan (2006, p. 511) claims that ‘despite its limitations documented in the literature, asymmetric timeliness is the most direct implication of conditional conservatism’.

- 7
We do not estimate firm-specific Basu coefficients because Givoly

*et al.*(2007) find that the asymmetric timeliness coefficients estimated from successive non-overlapping periods are not correlated, while other measures of conservatism (such as market-to-book) are highly correlated over the same period. - 8
We focus on time period

*t*+ 1 as our conservatism measures reflect conservatism over the life of the firm. Therefore, focusing on a 1-year ahead*Special Items*reduces the likelihood of changes in other factors that could bias our results. However, we report the effects of conservative accounting on the size of future special items charges in years*t*+ 2 and*t*+ 3 in the robustness tests. - 9
In additional unreported testing, we utilize industry-year dummy variables to control for the macroeconomic conditions for each year in a particular industry and we obtain qualitatively similar results to those reported.

- 10
Firms may frequently take special items charges because analysts and investors tend to treat these items as transitory earnings charges (Lipe, 1986; Fairfield

*et al.*, 1996). - 11
As we require 3 years of profitability data following year

*t*, we collect data from Compustat from 1986–2004. We start our sample in 1989 because of cash flows from operations not appearing in the Compustat database until 1986 for a small sample of firms. Cash flows from operations become widely available in Compustat in 1987. - 12
Darrough and Ye (2007) report that the frequency of loss firms reported on Compustat has grown from 3 per cent in 1963 to 40 per cent in 2001.

- 13
A discussion on how inclusion of these firms can skew tests involving profitability may be found in Hawawini

*et al.*(2003) and McNamara*et al.*(2005). - 14
Richardson

*et al.*(2005) use a sample of Compustat firms from 1962–2001 while we use a sample of Compustat firms from 1989–2001. - 15
In untabulated tests we also estimate column (iv) replacing the variable

*Special Items*with an indicator variable equal to one if the special items charge in year*t*+ 1 is >1 per cent of sales and zero otherwise. The coefficient on the interaction of the*Special Items Dummy * Return * Dummy*is −0.037 and is significant at the 5 per cent level. - 16
In untabulated testing, we find qualitatively similar results after controlling for firm leverage, growth, size, advertising expense, and research and development as well as their interaction with market return, negative return dummy, and asymmetric timeliness.

- 17
The significant difference in

*R*^{2}between columns (i) and (ii) is because of the use of*Avg. ROE*as a predictor of future*CFO*rather than*Avg. CFO*(*ROA*). The variable of interest (*Conservatism*) remains positive and significant at the 1 per cent level if*Avg. CFO*(*ROA*) is used to predict future*CFO*. - 18
We remove

*Avg. ROE*from the logistic regression in this specification because of the high correlation with*ROE*_{t+1}. - 19
We cannot run our measure of asymmetric timeliness because of the Roychowdhury and Watts (2007) methodology requiring 3 years of cumulative data prior to year

*t*. We also adjust our conservatism and control variables to adjust for the shorter lead/lag time.

### References

- Top of page
- Abstract
- 1. Introduction
- 2. Hypotheses development
- 2.3. Testable predictions
- 3. Research design
- 4. Results
- 5. Additional robustness checks
- 6. Conclusion
- References

- 2007, A theory of friendly boards, Journal of Finance 62, 217–250. , and ,
- 1996, Firm performance and mechanisms to control agency problems between managers and shareholders, Journal of Financial and Quantitative Analysis 31, 377–396. , and ,
- 1994, Accounting earnings and future economic rents: an empirical analysis, Journal of Accounting and Economics 17, 377–400. ,
- 2007, Accounting conservatism and board of director characteristics: an empirical analysis, Journal of Accounting and Economics 43, 411–437. , and ,
- 2002, The role of accounting conservatism in mitigating bondholder-shareholder conflicts over dividend policy and in reducing debt costs, The Accounting Review 77, 867–890. , , , and ,
- 1992, Association between accounting performance measures and stock prices: a test of the life cycle hypothesis, Journal of Accounting and Economics 15, 203–227. , and ,
- 2004, Operational restructuring charges and post-restructuring performance, Contemporary Accounting Research 21, 493–522. , , and ,
- 2001, Infrastructure requirements for an economically efficient system of public financial reporting and disclosure, in: R. E.Litan and R.Herrig, eds, Brookings–Wharton Papers on Financial Services (Brookings Institution Press, Washington, DC), 127–169. ,
- 2005, Earnings quality in UK private firms: comparative loss recognition timeliness, Journal of Accounting and Economics 39, 83–128. , and ,
- 1999, Accruals, cash flows, and equity values, Review of Accounting Studies 4, 205–229. , , , and ,
- 1997, The conservatism principle and the asymmetric timeliness of earnings, Journal of Accounting and Economics 24, 3–37. ,
- 2004, The link between earnings timeliness, earnings conservatism and board composition: evidence from the UK, Corporate Governance: An International Review 12, 47–59. , , and ,
- 1932, The Modern Corporation and Private Property (MacMillan, New York). , and ,
- 2006, Accounting quality and firm-level capital investment, The Accounting Review 81, 963–982. , and ,
- 2007, The determinants of corporate board size and composition: an empirical analysis, Journal of Financial Economics 85, 66–101. , , , and ,
- 1990, Internal corporate restructurings: an empirical analysis, Journal of Accounting and Economics 12, 251–280. , and ,
- 2001, Financial accounting information and corporate governance, Journal of Accounting and Economics 32, 237–333. , and ,
- 2007, Capital allocation and timely accounting recognition of economic losses, Working paper (University of Chicago, Chicago, IL). , , and ,
- 2003, Applied Multiple Regression and Correlation Analysis for the Behavioral Sciences (Erlbaum, Mahwah, NJ). , , , and ,
- 2008, Boards: does one size fit all? Journal of Financial Economics 87, 329–356. , , and ,
- 1999, Corporate governance, chief executive compensation, and firm performance, Journal of Financial Economics 51, 371–406. , , and ,
- 2007, Valuation of loss firms in a knowledge-based economy, Review of Accounting Studies 12, 61–93. , and ,
- 1998, The relation between earnings and cash flows, Journal of Accounting and Economics 25, 131–214. , , and ,
- 2007, Asymmetric timeliness tests of accounting conservatism, Review of Accounting Studies 12, 95–124. , , and ,
- 1996, Repeated write-offs and the information content of earnings, Journal of Accounting Research 34, 135–155. , and ,
- 1996, Accounting classification and the predictive content of earnings, The Accounting Review 71, 337–355. , , and ,
- 1995, Size and book-to-market factors in earnings and returns, The Journal of Finance 50, 131–155. , and ,
- 1983, Separation of ownership and control, Journal of Law and Economics 26, 301–325. , and ,
- Financial Accounting Standards Board (FASB), 2008, Conceptual framework for financial reporting: the objective of financial reporting and qualitative characteristics and constraints of decision-useful financial reporting information, 29 September 2008 (FASB, Norwalk, CT).
- 2010, Acquisition profitability and timely loss recognition, Journal of Accounting and Economics 49, 161–178. , and ,
- 1996, Causes and effects of discretionary asset write-offs, Journal of Accounting Research 34, 117–134. , , and ,
- 2009, Accounting conservatism and corporate governance, Review of Accounting Studies 16, 161–201. , , and ,
- 2010, Conditional conservatism and firm investment efficiency, Working paper (Universidad Carlos III de Madrid, Madrid). , , and ,
- 2003, Corporate board composition, protocols, and voting behavior: experimental evidence, Journal of Finance 58, 1997–2031. , , and ,
- 2000, The changing time-series properties of earnings, cash flows and accruals: has financial reporting become more conservative? Journal of Accounting and Economics 29, 287–320. , and ,
- 2007, Measuring reporting conservatism, The Accounting Review 82, 65–106. , , and ,
- 2004, The role of the board of directors in the capital budgeting process – evidence from S&P 500 firms, Working paper (Cornell University, Ithaca, NY). , and ,
- 2008, A theory of board control and size, Review of Financial Studies 21, 1797–1832. , and ,
- 1995, Corporate governance: some theory and implications, The Economic Journal 105, 678–689. ,
- 2003, Is performance driven by industry- or firm-specific factors? A new look at the evidence, Strategic Management Journal 24, 1–16. , , and ,
- 2003, Boards of directors as an endogenously determined institution: a survey of the economic literature, Economic Policy Review 9, 2–26. , and ,
- 1986, Agency costs of free cash flow, corporate finance, and takeovers, American Economic Review 76, 323–329. ,
- 1976, Theory of the firm: managerial behavior, agency costs, and ownership structure, Journal of Financial Economics 3, 305–360. , and ,
- 1996, The capitalization, amortization, and value relevance of R&D, Journal of Accounting and Economics 21, 107–138. , and ,
- 2008, The determinants of board structure, Journal of Financial Economics 87, 308–328. , , and ,
- 1981, Tobin’s
*q*ratio and industrial organization, Journal of Business 54, 1–32. , and , - 1986, The information contained in the components of earnings, Journal of Accounting Research 24, 37–64. ,
- 2000, Corporate governance and growth potential: an empirical analysis, Corporate Governance: An International Review 8, 32–42. Direct Link: ,
- 2005, Is performance driven by industry- or firm-specific factors? A response to Hawawini, Subramanian, and Verdin, Strategic Management Journal 26, 1075–1081. , , and ,
- 2006, Earnings management using classification shifting: an examination of core earnings and special items, The Accounting Review 81, 501–531. ,
- 2005, The link between earnings conservatism and the price-to-book ratio, Contemporary Accounting Research 22, 693–717. , , and ,
- 2005, Determinants of board size and competition: a theory of corporate boards, Journal of Financial and Qualitative Analysis 40, 283–306. ,
- 2006, The overinvestment of free cash flows, Review of Accounting Studies 11, 1159–1189. ,
- 2005, Accrual reliability, earnings persistence and stock prices, Journal of Accounting and Economics 39, 437–485. , , , and ,
- 2007, Asymmetric timeliness of earnings, market-to-book and conservatism in financial reporting, Journal of Accounting and Economics 44, 2–31. , and ,
- 2006, Identifying conditional conservatism, European Accounting Review 15, 511–525. ,
- 1997, A survey of corporate governance, The Journal of Finance 52, 737–783. , and ,
- 1992, The investment opportunity set and corporate financing, dividend and compensation policies, Journal of Financial Economics 32, 263–292. , and ,
- 1990, Managerial discretion and optimal financing policies, Journal of Financial Economics 26, 3–27. ,
- 2003, Conservatism in accounting part I: explanations and implications, Accounting Horizons 17, 207–221. ,
- 2006, What has the invisible hand achieved, Working paper (Massachusetts Institute of Technology, Cambridge, MA). ,
- 1986, Positive Accounting Theory (Prentice Hall, NJ). , and ,
- 1998, Corporate Governance, in: P.Newman, ed., The New Palgrave Dictionary of Economics and the Law (MacMillan, NY), 497–503. ,