Debt maturity structure, institutional ownership and accounting conservatism: Evidence from Iranian listed companies

2.1 Accounting conservatism defined

As the old adage “anticipate no profits but anticipate all losses” goes, accountants are often inclined to require higher verifiability for recognizing good news as gains than for bad news as losses (Basu, 1997). Accordingly, conservative reporting is typically defined as the timely recognition of expected unfavorable events in income than the recognition of the effects of expected favorable events (Givoly et al., 2007). Lower cost or market accounting for inventories and the timely recognition of cost estimates leading to future expected losses on long-term contracts than those resulting in higher future profits are typical examples of financial conservatism (Basu, 1997). Nevertheless, from a broader perspective, conservatism is also defined and interpreted as accountants’ preference for the adoption of accounting methods leading to lower values of assets and revenues along with higher values of liabilities and expenses (Belkaoui, 1985; Basu, 1997).

2.2 Conditional vis-à-vis unconditional conservatism

Prior literature draws a sharp distinction between the two types of conservatism using different terminologies and/or several pairs of terms. In this respect, several papers have classified conservatism under income statement and balance sheet conservatism categories (e.g. Basu, 1997; Ball et al., 2000; Pae et al., 2004). Beaver and Ryan (2005), arguably, criticize this classification for being rather unrealistic, primarily due to the consistency of the impact of conservatism on the income statement and balance sheet when the firm employs clean surplus accounting. The authors refer to conditional and unconditional conservatism just as Ball and Shivakumar (2005) do. Other pairs of names used for classifying conservatism are ex post and ex ante as well as news-dependent and news-independent conservatism. Following Beaver and Ryan (2005) and Ball and Shivakumar (2005), we use conditional and unconditional terms to name the two types of conservatism. Based on the argument of Beaver and Ryan (2005), unconditional conservatism can be triggered when the expected unrecorded goodwill is captured through the accounting process of assets and liabilities such as immediate expensing of the costs of intangible assets, accelerated depreciation of property, plant, and equipment and the historical cost accounting for positive net present value projects. The authors also define conditional conservatism as the timely recognition of book values under sufficiently adverse circumstances than their recognition under favorable circumstances.

2.3 Active vis-à-vis passive institutional ownership

Prior studies suggest that institutional investors are not alike in terms of their monitoring functions. In other words, they have different incentives for actively monitoring corporate managers (Navissi and Naiker, 2006; Cornett et al., 2007). Accordingly, institutional ownership is generally comprised of both active and passive investors. Passive institutional investors possess a high portfolio turnover and transient trading strategy. They are interested only in short-term profits and their performance is evaluated primarily based on the short-term returns they generate (Navissi and Naiker, 2006). On the other hand, active institutional investors are more inclined to gain corporate board representation than passive ones and consequently exercise efficient monitoring (Navissi and Naiker, 2006).

2.4 Debt maturity structure and accounting conservatism

To our knowledge, to date, prior literature on the relationship between accounting conservatism and the agency costs arising from debt financing has focused on different aspects of debts, such as debt maturity structure, the interest charged on the debts and in some cases, the extent of debt used in a firm’s capital structure (Ahmed et al., 2002; Zhang, 2008; Beatty et al., 2008; Khurana and Wang, 2015). The collective evidence provided by Ahmed et al. (2002), for instance, highlights the significant role of accounting conservatism in reducing corporate debt costs. Employing both market-based and accrual-based proxies for accounting conservatism, the authors indicate that corporate debt is negatively associated with the adoption of conservatism in financial reports, ceteris paribus. In a similar vein, Zhang (2008) recognizes mutual benefits for two sides of a debt contract, i.e., the lender and the borrower. Specifically, he predicts and finds that more conservative borrowers, as compared to their counterparts, enjoy lower initial interest rates offered by lenders as a result of their covenant violation and the consequent signaling of default risk. Using 3,641 private debt agreements with net worth covenants issued during 1994 to 2004, Beatty et al. (2008) documented the existence of conservative contract modifications, but not ubiquitously. Specifically, they provided some evidence that debt contracts are modified mostly when there are higher agency costs and lower litigation, tax and equity demands for conservatism. Recent evidence by Khurana and Wang (2015) focuses on the relationship between the debt maturity structure and accounting conservatism and indicates that the adoption of conservative accounting approach is negatively influenced by the short-maturity debt, which is also more pronounced among financially distressed firms. Ball and Shivakumar (2005) took into consideration the extent of debt used in the capital structure of a sample of the privately held and publicly traded companies of the UK and demonstrated that the publicly held companies of the UK were more likely to recognize losses in a timelier manner (i.e. adopt conservative financial reporting) than their counterparts.

It has been argued that shortening debt maturity could provide the ground for mitigating both the agency costs arising from debt financing and underinvestment problems. The latter can be fulfilled through facilitating frequent debt re-pricing and making debt mature before the growth options expire (Myers, 1977; Barnea et al., 1980; Khurana and Wang, 2015). However, although shortening debt maturity has the above-cited strengths, firms are required to consider the suboptimal liquidation risks stemming from too much refinancing and the consequent bankruptcy costs. Indeed, they have to trade off the benefits and costs of using short-maturity debt (Sharpe, 1991; Diamond, 1991; Khurana and Wang, 2015). Furthermore, accounting conservatism could be a curative mechanism through which lender–borrower conflicts and the agency costs of debt can be mitigated (Watts, 2003; Ball and Shivakumar, 2005; Khurana and Wang, 2015). Prior literature also suggests that accounting conservatism can act as much of a deterrent against managers’ incentives to undertake self-serving projects that do not enhance shareholder value. To put it more simply, accounting conservatism, to some extent, prevents managers from deferring the recognition of economic losses and thus pursuing negative net present value projects by providing timelier revisions of earnings and asset book values (Watts, 2003; Ball and Shivakumar, 2005; Khurana and Wang, 2015). In this context, Bushman et al. (2011) examined the relations between corporate investment behavior and the timeliness of accounting recognition of economic losses by using firm-level investment decisions for a sample of 25 countries and found that managers’ tendency toward investment spending is positively associated with investment opportunities in countries where conservative reporting is more pronounced. Based on above-cited arguments, we hypothesize that short-maturity debt contracting is likely to be negatively associated with the demand for accounting conservatism, particularly when it facilitates the monitoring function of lenders and discourages shifting toward risky projects. As Khurana and Wang (2015) point out, it is predictable from the incomplete financial contracts perspective that short-maturity debt financing is likely to negatively affect the demand for accounting conservatism. Therefore, we present the following hypotheses to examine our prediction:

Despite the above predictions, several papers have focused on the conflicts between lenders and shareholders in the presence of default risk. In this case, several events cause the longevity of effective debt maturity such as choosing riskier investment projects, over-financing, lower equity financing and concealing problems from creditors (Myers, 2001; Khurana and Wang, 2015).

Myers (2001) asserts that where the default risk is very small or negligible, debts values are rarely influenced by news regarding the economic performance of corporations. However, as the level of default risk rises, the value of debts are further influenced, and, consequently, the creditors are provided with an incentive to receive the performance-related news of the corporations. Eisdorfer (2008) examines the relationship between investment and volatility and finds the expected negative relation between the two variables to be reversely affected by shareholders’ risk-shifting incentives. Indeed, the author offers two major results. The first result is related to the fact that financially distressed firms are faced with a weak negative relationship (and, in some cases, a positive relationship) between their investment intensity and volatility. The second finding reveals that the value-creation potential of investments in financially distressed firms is lower during times of high uncertainty. Based on the previously mentioned discussions, it can be concluded that financial distress (FD) is likely to influence the relationship between short maturity debts and accounting conservatism. Therefore, we posit the following hypotheses in the null form:

Some recent studies indicate a positive association between the level of accounting conservatism and agency problems. LaFond and Roychowdhury (2008), for instance, examined the impact of managerial ownership on accounting conservatism employed by corporations and found that manager-owned firms with an accentuated state of ownership and control separation as well as agency conflicts are faced with lower conservatism. Likewise, LaFond and Watts (2008) demonstrate that there is a positive association between conservatism and information asymmetry arising from agency problems. These studies are indicative of how equity investors demand conservatism. Nevertheless, Ramalingegowda and Yu (2012) predict and find that institutional investors that are more likely to monitor corporate managers drive the demand for conservatism more than the individual investors. Indeed, institutional investors, as more sophisticated and important price-setters in capital markets, perceive and value corporate governance benefits of conservative financial reporting and, consequently, demand more conservative accounting from managers (Bartov et al., 2000; Chakravarty, 2001; Sias et al., 2006; Ramalingegowda and Yu, 2012). Prior literature also introduces some institutional characteristics inducing higher monitoring incentives such as long investment horizons, concentration of share holdings and independence from management (Shleifer and Vishny, 1986; Hartzell and Starks, 2003; Chen et al., 2007; Ramalingegowda and Yu, 2012). This leads to our next hypotheses as follows:

Finally, based on the above-mentioned relationships, we posit the following hypotheses as well:

3.1 Data sources and sample selection procedure

We obtain our required data manually from the hardcopy financial statements held in the TSE library (Codal[1] and its supplementary software known as Rahavard Novin) for the period 2011–2016. To construct our sample for the paper’s hypotheses, we begin with all client-year observations present on the Codal database, i.e., a potential population of 2,982 firm-year observations. We then exclude delisted observations (696 firm-year observations), observations with missing or insufficient variable data (126 firm-year observations) and newly-listed observations[2] (408 firm year observations). We also exclude firms operating in banking industry as well as financial and investment institutions (894 firm-year observations) to calculate the variables used in our equations, primarily because financial institutions and banking industry have different reporting requirements that could influence the figures associated with dependent variables. This leaves us with a primary sample of 858 firm-year observations. Table I discusses the breakdown of sample attrition.

3.2 Measures of accounting conservatism

The present paper employs Basu’s (1997) earning-return model along with Beaver and Ryan’s (2000) book-to-market model to measure accounting conservatism. Basu’s (1997) model regresses earnings on positive (negative) stock returns (i.e. the sign of the return coefficient could be either positive or negative) to capture good or bad economic news as follows:

In Equation (1), the coefficient on NEG_JT (β₂) captures the timeliness of earnings concerning good news and the coefficient on the interaction variable of NEG_JT×RET_JT (β₃) captures asymmetric timeliness regarding bad news vs good news (i.e. the measure of accounting conservatism). We use Equation (1) to measure conditional conservatism. Following Ismail and Elbolok (2011), we also incorporate MB (ratio of market value to book value of stockholders’ equity) in Equation (1) in order to measure unconditional conservatism.

Beaver and Ryan (2000) designed a model for measuring accounting conservatism by capturing the difference between the book value and market value of net total assets. To put it simply, the authors measured conservatism as the ratio of book to market value of stockholders’ equity:

3.3 Regression models

Following Khurana and Wang (2015), we use Basu’s (1997) modified model to empirically examine the relationship between short-maturity debts and conditional (unconditional) conservatism reflected in H1–H1b as follows:

In Equation (2), ST represents the extent of short-maturity debt in a firm’s capital structure. Following Fan et al. (2012) and Wang et al. (2010), we use the ratio of short-term debts to total debts as a proxy for the debt maturity structure. The market-to-book (MB) ratio is computed as the market value scaled by the book value of common equity. MB is used to measure unconditional conservatism and its relation with the debt maturity structure. LEV represents the financial leverage and is calculated as the book value of debts scaled by the book value of total assets. SIZE is a proxy for firm size and computed as the natural logarithm of market value of equity. To examine the relation between debt maturity structure and accounting conservatism, we also estimate the following model (Beaver and Ryan, 2000):

where CON is the accounting conservatism; ST is the extent of short-maturity debt in a firm’s capital structure; LEV is the financial leverage, calculated as the book value of debts scaled by the book value of total assets; SIZE is the firm size, computed as the natural logarithm of the market value of equity.

To test the association between short-maturity debts and conditional and unconditional conservatism in financially distressed firms (i.e. H2–H2b), we employ Altman’s (1968) bankruptcy prediction model (Z-score). This model is dependent on five different ratios: namely, liquidity, solvency, profitability and activity:

In Equation (5) WCTA represents the ratio of working capital to total assets; ROA, the ratio of retained earnings to total assets; EBTA, the ratio of earnings before income and tax to total assets; MVTD, the ratio of the market value of equity to book value of total debts; STA, the ratio of sales income to total assets; and Z, the overall Z index. The lower a firm’s Z-score, the higher its probability of bankruptcy. Indeed, Z values lower than 1.81 are indicative of bankruptcy. Based on the previous measure of FD or bankruptcy and following Khurana and Wang (2015), we estimate the following regression model to test H2–H2b:

where FD is our variable of interest which equals 1 if the firm is faced with FD, and 0 otherwise. The rest of the variables are similar to those defined in Equation (3). To test the relationship between the debts maturity structure and accounting conservatism by our second conservatism measure, we estimate the following Beaver and Ryan’s (2000) model:

All of the variables included in Equation (7) are similar to variables used in Equations (3), (4) and (6). We examine the effect of active and passive institutional ownerships on accounting conservatism (H3–H3d) by modifying Basu’s (1997) model as follows:

In Equation (8), the positive or negative sign of β₇ indicates that greater (lesser) institutional ownership results in greater (lesser) accounting conservatism. Next, we draw a distinction between active and passive institutional ownerships and then modify the above-mentioned model:

In Equations (8) and (9), INST represents the percentage of shares held by institutional investors; ACINST, the percentage of shares owned by active institutional investors (i.e. the institutional investors with representatives on corporate board of directors); INACINST, the percentage of shares owned by passive institutional investors (i.e. the institutional investors without any board representation). The rest of variables are similar to previous equations. In addition to Equation (9), we attempt to examine the effect of active and passive institutional ownerships on the second measure of accounting conservatism proposed by Beaver and Ryan (2000) by estimating the following equation:

Finally, the following equation captures the effect of active and passive institutional ownerships on the debt maturity structure:

3.4 Specification tests (diagnostics) in panel data models

We conduct several diagnostic tests by using the R programming language to estimate the most appropriate models. What follows is a succinct explanation of these tests at 0.05 significance level: the results of F-limer specification test (P1<0.001; P2<0.001; P3<0.001; P4<0.001) confirms the preference of panel data model for all four models and for both accounting conservatism measures. The results of Hausman test indicates that all models are better fitted using the random effects model (P1: 0.497; P2: 0.962; P3: 0.961; P4: 0.204) for both conservatism measures. Next, we use Lagrange Multiplier Test (Breusch-Pagan) to choose the appropriate model between a random effects regression and a simple OLS regression. The null hypothesis in the LM test is that variances across entities are zero. That is, there is no significant difference across units (i.e. no panel effect). This test confirms the appropriateness of the pooled OLS model with time effects for all regression models (P1: 0.023; P2: 0.003; P3: 0.388; P4: 0.387) and for both conservatism measures. Finally, the results of specification tests indicate that the panel of random effects is the most appropriate model so far all models. To test the serial correlation of (the idiosyncratic component of) the errors in panel models, we also conduct the Breusch-Godfrey/Wooldridge test on the residuals of the (quasi-) demeaned model, which should be serially uncorrelated under the null of no serial correlation in idiosyncratic errors for both conservatism measures. The results suggest that our final models are fitted using the Panel Generalized Linear Model (PGLM). A brief summary of the above-cited tests is shown in Table II.

4.1 Descriptive statistics

Table III reports the descriptive statistics of the variables used in the regression models. As it is evident in panel A of this table, current debts on average account for 86 percent of the total debts of the sample firms. Comparing the average and standard deviation of debt maturity structure (ST) and Beaver and Ryan (2000) conservatism measure also reveals a negative relationship between the two, which could provide supporting evidence for H2. In panels B and C of Table III, the proxy used for FD indicates that an average of 72 percent of sample firm are faced with FD. The indicator variable with respect to Basu’s (1997) model (NEG) demonstrates an average of 46 percent negative return on common equity of sample firms, suggesting the critical condition of most companies listed on the TSE. The average net income before extraordinary items of sample firms is 10 percent of market value of stockholders’ equity. Finally, Table III indicates that 57 percent of institutional investors are active institutional investors (i.e. with board representation) and 13 percent of them are passive institutional investors (i.e. without any representative on the corporate board of directors).

4.2 Estimation results

Table IV exhibits the estimation results of model (3) using PGLM. As it is shown in panel A of this table, while LEV is negatively (C: −0.936) and significantly (two-tailed p<0.001) associated with accounting conservatism, the coefficient on RET×NEG (C: 3.336) as well as the coefficient on RET×LEV (C: 0.283) suggest a significant and positive relation with accounting conservatism. In addition, the interaction variables of RET×NEG×LEV (C: −1.120; P: 0.017) and NEG×RET×SIZE (C: −0.158; P: 0.047) indicate a negative and significant relation with accounting conservatism. As the estimation results show, the rest of variables used in model (3) are not significantly associated with accounting conservatism. Based on preceding discussions and considering the non-significant coefficient on RET×NEG×ST, our results do not provide supporting evidence for H1. However, the results confirm the negative relationship between debt maturity structure and accounting conservatism.

Panel B of Table IV also suggests that LEV (C: 2.933; p<0.001) and SIZE (C: 0.311; p<0.001) are negatively and significantly associated with accounting conservatism computed by Beaver and Ryan’s (2000) model (at 0.01 significance level). Furthermore, the figures reported in panel B are also indicative of the significance of the coefficient on ST at 0.05 of significance level. Therefore, our findings provide supporting evidence for H1. That is, accounting conservatism measured by Beaver and Ryan’s (2000) model is negatively influenced by debt maturity structure.

The figures shown in Table V are the estimation results of models (6) and (7) using PGLM. Panel A of this table indicates that LEV and the interaction variables of RET×NEG×LEV as well as RET×NEG×SIZE are negatively and significantly associated with accounting conservatism measured by modified Basu’s (1997) model. Since the coefficient on the interaction variable of RET×NEG×ST×FD (C: −0.321; P: 0.882) is non-significant, our findings do not support H2. Nevertheless, our findings confirm the negative relation between debt maturity structure and accounting conservatism. Panel (B) of Table V indicates a significant relationship between all the variables used in model (7) and accounting conservatism measured by Beaver and Ryan’s (2000) model. In this respect, the significant coefficients on ST (C: −1.122; p<0.001) and FD (−1.245; p<0.001) provide support for H2.

Table VI indicates the estimation results of model (9) and (10) using PGLM. As it is evident in panel A of Table VI, the ratio of MB value of stockholders’ equity and passive institutional investors (INACINST) is significantly associated with accounting conservatism (at 0.05 of significance level). However, since the coefficients on the interaction variables of INACINST×RET×NEG (C: 0.860; P: 0.076) and ACINST×RET×NEG (C: −1.578; P: 0.104) are not statistically significant, H3 is not supported. Panel B of Table VI indicates that LEV and SIZE are significantly and positively associated with accounting conservatism measured by Beaver and Ryan’s (2000) model. However, these figures do not show any significant relationship between active (C: −0.152; P: 0.387) and passive (C: −0.243; P: 0.329) institutional investors (ACINST and INACINST) and accounting conservatism. Accordingly, H3 is not supported by our findings.

Finally, the estimation results of model (11) reported in Table VII suggest that active (C: 0.0008; P: 0.978) and passive (C: −0.006; P: 0.895) institutional ownerships are not significantly associated with accounting conservatism and consequently H4 is not supported as well.