Printer Friendly

Firm Investment Efficiency and Auditor Perception of Dividend Initiation.

Introduction

Auditors are tasked with assessing the business risk of the firm. Any information that helps clarify the risk of the firm should influence the audit risk and consequently the audit fees that are imposed. Dividends may be initiated to balance over-investment by the firm with the degree of agency conflict by reducing available cash. Furthermore, dividends may reflect future performance expectations. Managers have previously stated that dividend initiation signifies that the firm is confident their future earnings will be strong enough to support future dividend payments.' Although dividend initiation generally produces positive abnormal market returns, all market participants may not share the same perspective. The goal of this study is to examine how dividend initiation is interpreted by auditors as evidenced by a premium or discount in audit fees.

The decision to initiate a dividend payment has serious consequences for managers and shareholders, in theory and practice. The commitment to a recurring dividend payout reduces managers' available discretionary income. Dividend initiation is often considered a signal of confidence and a long-term commitment to support a payout to shareholders, regardless of short-term free cash flow fluctuations. Further, dividend initiations predict subsequent positive earnings and a decline in risk (Healy and Palepu 1988).

This paper examines the auditors' perception of the reduction in the agency costs of the over-investment problem by examining firms that initiate dividends, considering that initiation may reduce audit risk in two ways. If the dividend initiation represents a signal of management's confidence in future performance, auditors perceive a lower risk of financial distress. From an agency perspective, dividends serve as a monitoring mechanism, restricting management's tendency to over-invest.

Our investigation contributes two new findings for the auditing and corporate financing policy literatures. First, we provide evidence that auditor perception of dividend initiations is conditioned on the firm's investment opportunities, suggesting that the audit fee reduction for dividend initiation is ascribed to the agency problem of over- or underinvestment. Combining the firm's investment efficiency with the dividend initiation event, we find that over-invested firms receive the greatest fee reduction upon initiation. Secondly, we find that over-invested, dividend initiating firms pay lower fees even when compared to other dividend-paying firms, suggesting that dividend initiation is also priced by auditors because of management's signal of confidence in sustainable ongoing financial performance. This second finding of auditors responding to management's signal for high future performance is reinforced with evidence that under-invested firms with high growth also have lower audit fees on dividend initiation. In addition, our findings shed light on how auditors value dividend initiation, separate from market valuation of the resulting cash flow.

In section 2, we review the literature regarding dividend initiations and auditor pricing of risk and then develop and present our hypotheses. In section 3, we describe our data and empirical procedures used to test our hypotheses. In section 4, we present our results, and Section 5 concludes the paper.

Theoretical Framework and Hypothesis Development

AUDITOR PERCEPTIONS OF DIVIDEND INITIATIONS

Although several papers examined dividend initiations from the shareholder perspective, auditor perception of risk related to dividend initiations has been largely neglected. Griffin, Lont and Sun (2010) find that dividend payout does not moderate audit fees, but they do not differentiate between payers and initiators. Therefore, empirical research lacks a large body of evidence regarding how auditors perceive dividend initiations and instead rest heavily on stock market reactions, which have a different loss function and set of considerations than an audit team.

When assessing audit fees, auditors consider the risk the client poses for them from reputational capital and legal perspectives (Simunic 1980). Auditors take into account business risk, which is the risk that the business will not achieve its objectives. Furthermore, auditors assess engagement risk, the potential legal and reputational effects to the CPA firm resulting from taking on the client. Business risk creates liability for auditors because the risk of the firm not meeting objectives increases the rationality component of the fraud triangle, and because, if it leads to investor loss, the plaintiffs may assert that the auditors allowed the indicators of decline to be masked in the financial statements. In addition, the auditor must assess the time investment, provide assurance for investor losses and insure against unassociated risks.

According to AICPA, SAS 107, para. .02, audit risk comprises two components. First is "the risk that the auditor may unknowingly fail to appropriately modify his or her opinion in financial statements that are materially misstated" (Zimbelman 1997; Clover, Prawitt, Schultz and Zimbelman 2003 and Bedard and Johnstone 2004). Second, auditors compensate themselves for expected losses. Losses may arise from class action lawsuits resulting from, for instance, negative performance outcomes for the client and corresponding impacts on shareholder wealth. Losses may also include foregone future audit fees resulting from the audit-client's financial distress and any reputational damages due to engaging with a risky client (see Menon and Williams 1994; Houston, Peters and Pratt 1999; Johnstone 2000; Lyon and Maher 2005 and Houston, Peters and Pratt 2005).

DIVIDEND INITIATIONS

Lawson and Wang (2016) document that auditors assess lower audit fees to clients who initiate dividends. We use the findings of Lawson and Wang (2016) as the starting point for our investigation of auditor perception of strategic corporate financing policy changes conditioned upon growth opportunities. From a theoretical standpoint, there are two perspectives on dividend initiations, which are not mutually exclusive: 1) they can reduce agency costs through discretionary cash flow reduction, and 2) they can signal future cash flows to investors (Miller and Modigliani 1961).

Signaling Theory

The audit risk related to financial failure can be alleviated by evidence of strong future earnings, and prior research supports such a relationship between dividends and future earnings (Healy and Palepu 1988). Firms that initiate a recurring dividend have higher earnings after the announcement of recurring payout, and the authors suggest that investors can understand the dividend payout as a future earnings forecast. Benartzi, Michaely and Thaler (1997) find positive and significant initial and three-year market returns for firms that increase dividends. Dividend increases are also associated with a persistence of higher future earnings (Koch and Sun 2004).

Bhattacharya (1979) and Miller and Rock (1985) argue that changes in dividend policy signal news about future cash flow prospects in that they present a credible signal to market participants. A credible signal is sufficiently but not prohibitively costly, such that other firms cannot easily mimic it. Dividends provide credible signals because, as quasi-contracts, they are rigid. Unless managers are certain they can maintain the new payout stream following initiation, they hesitate to initiate dividends, and only managers of strong firms can ensure that they will be able to maintain the dividend. Managers of weak firms cannot initiate a dividend payment because they cannot be confident of their ability to uphold the quasi-contract. Since they will suffer a stock price decline if the dividend is cut or terminated, weak firms cannot easily mimic a strong firm's signal (Brav, Graham, Harvey and Michaely 2005 and Allen, Bernardo and Welch 2000). Thus, the signaling hypothesis posits that managers initiate dividends in order to signal strong future cash flow prospects (Kale, Kini and Payne 2012).

Dividend initiation may indicate reduced risk. Skinner and Soltes (2011) suggest that dividend policies signal information regarding the quality and risk of firms' future earnings. Charitou, Lambertides and Theodoulou (2011) report that dividend initiations impact the firm's default risk, arguing that a dividend initiation signals the firm's confidence that its future earnings will persist and cover future dividends. Moreover, they propose that firms initiating dividends normally reveal that they have stable cash flows and are mature enough to reduce investment cash. Eventually, these firms' systematic and default risk declines. Dyl and Weigand (1998) find that the announcement of initial dividends significantly reduces firms' total risk (especially for larger firms) and earnings volatility (especially for smaller firms). Hones, Gu and Liu (2014) find that implied volatility declines after the dividend initiation announcement, along with a decline in traditional ex-post risk measures, such as the standard deviation and beta, in the period following the initiation announcement.

Agency Theory

From the agency perspective, dividend initiations may mitigate opportunistic behavior by managers by reducing their available discretionary funds (Jensen 1986). Firms with large pools of internally-generated funds and a propensity to over-invest may be plagued by managers who misdirect resources toward perquisite consumption. In support of this conjecture, firms with weak governance and those with a propensity to over-invest are greeted with the most favorable market reaction when they announce a dividend initiation (Officer 2011). Furthermore, dividend initiations are likely to result in more frequent trips to the equity markets to raise capital (as a result of lower discretionary cash flows). This results in better quality governance through greater scrutiny of managerial activity by equity market participants (Easterbrook 1984 and Jensen 1986).

Caskey and Hanlon (2013) find that dividend-paying firms are less likely to engage in intentional financial report misstatements than non-dividend-paying firms. In otherwords, dividend initiations appearto constrain managers' ability to intentionally misstate their firms' financial statements. Based on Caskey and Hanlon's (2013) evidence, it could be argued that clients' dividend initiations send signals to auditors about the firms' risk of material misstatements. Gleason, Greiner and Kannan (2017) find a positive relationship between excess cash and audit fees, attributed to an increase in auditor perception of business risk. They conclude that excess cash holdings signal to the auditor that the firm has greater agency costs.

Therefore, the agency problem associated with cash holdings, beyond the available positive Net Present Value (NPV) investments, creates an engagement risk for the auditor that would be priced in the audit fee. The initiation of a dividend payout is a long term commitment on the part of management rather than an attempt to lower audit fees by reducing excess cash in a less permanent manner, and it is associated with risk reduction for the firm and to the auditor. Assuming that the auditor perceives the informational signal about dividend initiations as affecting risk, the auditor should design tests to reduce risk accordingly. This will eventually affect the resources the auditor allocates, as well as the audit fees.

INVESTMENT EFFICIENCY AND DIVIDEND INITIATION

Evaluating dividend initiation in combination with the firm's investment efficiency provides an opportunity to distinguish signaling from agency theory in dividend initiating firms and to determine if auditors change their risk assessment in response to one or both theories. We identify firms as inefficiently invested by the extent to which their investment deviates from industry peers with similar sales growth.

Over- and under- investment are similar in that they both represent investment inefficiency and impair performance of the firm; however, they stem from different situations. Over-investment is a symptom of agency theory, although over-invested firms may also use dividend initiation to signal future performance. Firms that are under-invested may simply not have the funding sources necessary to take on all of the positive net present value projects. Alternatively, they may suffer from problems related to information asymmetry, where investors do not realize the growth opportunities available to the firm and therefore do not support external funding at the competitive rate expected by management. In this case, managers may pass up positive NPV projects due to capital rationing. Under-invested and over-invested firms can call attention to the high quality of their growth prospects by initiating a dividend.

From the agency perspective, a dividend initiation reduces the discretionary cash flows available to managers, who may misappropriate them for personal rents (Jensen 1986 and Officer 2006). However, misappropriation of resources may be constrained by the investment opportunity set. Bulan, Subramanian and Tanlu (2006) find that dividend initiators are large, stable firms that consistently achieve high profitability levels and generate much cash but do not find many profitable investment opportunities. However, executives profit personally and professionally from over-investing. Grinstein and Hribar (2004) and Harford (1999) find that CEOs with the power and excess cash to over-invest in value-destroying projects will do so. Therefore, we investigate the potential for inefficient investment that accompanies growth to affect auditors' perception of dividend initiations. The initiation of a dividend that will reduce management's access to excess cash flows should reduce engagement risk for auditors.

H1: Firms that invest more inefficiently are assessed lower audit fees when they initiate a recurring dividend.

Over-investment

To clarify the impact of the dividend initiation on the determination of audit fees, we control for the investment inefficiency and look at the difference between dividend initiating firms and firms with no dividend payment. If a firm has a propensity to misallocate resources through under- or over-investment and its managers have greater discretion in making investment choices benefitting personal interests, initiating firms may exhibit lower audit fees compared to non-paying firms. The quality of a firm's investments is higher when management's access to free cash flow is constrained by a dividend payout regimen (Lang, Stulz and Walkling 1991). Therefore, we directly investigate the relationship between over-investment propensity and audit fees for firms that initiate dividends versus non-paying firms (following Officer 2006).

For firms that are over-invested and have excess free cash flow, investments are the most value-depreciating (Lang, Stulz and Walkling 1991 and Nohel and Tarhan 1998). Agency costs are more pronounced when there are few investment opportunities because CEOs are rewarded for their ability to meet quarterly earnings targets but are not held accountable for over-investment that leads to lower firm value (Harford and Li 2007 and Hermalin and Weisbach 1998). Management may use over-investment to disguise firm losses, to "buy" sales increases through acquisition or to complicate the firm operations for job security (Morck, Shleifer and Vishny 1990), all of which add risk to the audit. The reduction in free cash flow due to the initiation of a dividend may help mitigate agency costs and reduce audit risk.

Over-invested firms are at a greater risk of misallocation of resources; thus, auditors may perceive a greater benefit for over-invested firms to initiate a dividend. Our second hypothesis is as follows:

H2: Over-invested firms that initiate a recurring, cash dividend have lower audit fees compared to firms that do not pay dividends.

Under-investment

In rich investment opportunity set environments, firms that are under-invested may be less inclined to commit to what are effectively fixed cash flows, such as dividends, and therefore less likely to initiate dividends. Those firms that initiate dividends and are under-invested may be perceived as more likely to exhibit distress or to have higher cost capital market access. High-growth firms tend to hold a large proportion of intangible assets, which are prone to exhibit greater information asymmetries and lead to difficulties in retaining these assets' full value should the firm end up bankrupt (John 1993). High-growth firms also experience higher costs of accessing the capital markets, relative to firms with greater assets in place (Berger and Udell 1995 and Myers 1977). Thus, auditors may perceive firms with greater investment opportunities who initiate dividends as increasing business risk and consequently increase audit fees. However, signaling can be used to reduce information asymmetry in under-invested firms, especially for under-invested firms that are not less financially constrained (Biddle, Hilary and Verdi 2009).

Firms that have less invested in capital expenses, R&D and acquisitions (less disposals) compared to similarly growing peer firms are under-invested. In other words, under-invested firms have net present value projects available, and, consequently, they do not exhibit agency problems such as over-invested firms (Biddle, Hilary and Verdi 2009). Lower levels of investment are associated with lower risk by auditors. Chy and Hope (2016) find that auditor conservatism causes under-investment. When auditors tightly control the ability of the firm to use earnings management, management reduces the optimal spending of R&D to avoid the negative short-term effects on the firm's performance. This suggests that under-investment is evidence of excess conservatism. Furthermore, since under-invested firms have available growth opportunities, they should have lower perceived audit risk and lower audit fees.

Under-investment may exist due to management's perception of scarcity of affordable external funding given existing levels of information asymmetry and may be related to management passing up risky but positive NPV projects. Low levels of liquidity and reduced access to funding have been shown to increase audit fees (Francis and Wang 2005; Whisenant, Sankaraguruswamy and Raghunandan 2003 and Raghunandan and Rama 2006). Firms that initiate a recurring dividend signal to the auditors and to the market that they are not under-invested due to low liquidity. This signal indicates that management expects sufficient cash flow for the foreseeable future.

Dividend initiation may be a signal to the market and tangible evidence to auditors of strong future growth projects. La Porta, Lopez-De-Silanes, Shleifer and Vishny (2002) suggest that firms may initiate dividends according to a substitute model, where dividends are initiated as an appeasement to shareholders to keep the firm held in high esteem so that a future seasoned equity offering will be more successful. The concept behind under-invested firms initiating dividends is to alter their cost of capital. Therefore, we anticipate that under-invested dividend initiators have lower audit fees than under-invested firms that do not pay dividends.

H3: Under-invested firms that initiate a recurring, cash dividend have lower audit fees compared to firms that do not pay dividends.

Compared to Dividend Paying Firms

We further seek to examine the role of investment inefficiency in audit fee determination by examining dividend initiating firms versus firms with existing dividend payments. In the agency paradigm, firms presently paying dividends are less likely to misallocate resources than those that do not because managers must be cognizant of how reducing dividends could impact shareholder wealth. In this way, dividend paying firms with a propensity to over-invest may exhibit lower agency conflict through dividend payouts to shareholders than nonpaying firms. However, dividend initiations are a unique signal in that dividends can only be initiated once, whereas most firms' corporate financing policies account for regular dividend increases once dividends commence. Dividend increases are made in small increments based on prior dividends, whereas initiations are large shifts in corporate financing policy that are typically unanticipated by the financial markets.

Dividend initiation may result in lower fees when management is calling the market's attention to the positive governance implications of dividend initiations. An initiation is a one time, unequivocal signal of commitment to value maximizing behavior. Further, once dividends are initiated, market participants are subject to greater oversight by the external capital markets (Easterbrook 1984). Accordingly, a favorable capital market response to dividend initiations is indicated by the significant, positive reaction to dividend initiation announcements by shareholders (Asquith and Mullins 1983 and Michaely, Thaler and Womack 1995). The capital market response to both increases and initiations is positive, but abnormal returns are larger for initiations than for increases because it is a unique, less anticipated, more visible shift in corporate financing policy than a dividend increase.

With respect to financial signaling theory, firms that initiate a dividend should present lower risk to auditors than non-paying firms, but not necessarily in comparison to dividend paying firms as they are already committed to future cash payout. In order to be a lower risk than dividend paying firms, dividend initiators must have near-term performance expectations or risk levels that are more favorable than their dividend paying peers. Prior studies suggest that dividend initiators may be poised for the near future to outperform even dividend payers in terms of earnings stability and quality. Dyl and Weigand (1998) find that dividend initiators have a significant decline in earnings volatility for twelve quarters after dividend initiation, and they show that the market prices this downward risk shift in the firm. Smith and Pennathur (2017) find that dividend initiating firms employ downward earnings management in the period just prior to dividend initiation to create reserves of cash and earnings that can be drawn on for more stable performance in the near future.

Auditors have access to the firm's complete financial information and can see if near-term future performance appears optimistic. However, when management is willing to commit to a quasi-contract of recurring, cash dividend payout, they signal confidence and credibility related to governance to the auditors, which lessens the audit risk. We anticipate finding a differential impact between audit fees and dividend paying firms versus initiating firms, conditioned on the extent of the firm's investment inefficiency. We leave the determination of the specific effects of over- or under-investment to empirical results. Therefore, our fourth hypothesis is as follows:

H4: Inefficiently invested firms that initiate a recurring, cash dividend have lower audit fees compared to firms that already pay dividends.

Data and Methods

DATA

To obtain our sample, we begin with firms that have current year fee data in Compustat and have matching dividend payment data in CRSP. Firm-year observations must also have audit fee data in Audit Analytics and segment data from S&P Research Insights. Firms with Compustat SIC codes for utilities and financial institutions (4400 - 4999 and 6000 --6999) are excluded, as are firm-years with missing data or assets less than $1 million.

All dividend information is obtained from CRSP, and dividends are only considered if they are recurring, ordinary cash dividends, identified with CRSP dividend codes 1222, 1232, 1242 or 1252. Firms without recurring dividend payments are designated as nonpayers. To identify firms that initiate a dividend, we require that the firm did not pay dividends for three prior years. All of the sample firm-years were matched to CRSP to confirm payer, nonpayer or initiator status. The final sample includes 17,520 firms-years from 2003 to 2012. This includes 12,042 non-paying firm-years, 5,196 dividend paying firm-years and 282 firm-years of dividend initiation. Fifty-seven of the dividend-initiating firms are over-invested, and 225 of the initiating firms are under-invested.

Methods

We use ordinary least squares regression with clustered errors by firm to estimate our model, which follows Raghunandan and Rama (2006). We present all analyses by using standard errors clustered by firm to address any potential cross-sectional correlation of error terms (Petersen 2009). All continuous variables used in the analyses are winsorized at the 1st and 99th percentiles. Our audit fee model includes proxies for variables that Simunic (1980) identifies as loss exposure determinants. We rely on the Raghunandan and Rama (2006) model because of its high explanatory power and its controls for auditor and engagement attributes, which are significantly associated with audit fees (Hay, Knechel and Wong 2006). We also include a benchmark sample using non-payers (following Officer 2006). The model used to test our hypotheses is as follows:

[LNAF.sub.i,t] = [[alpha].sub.0] + [[alpha].sub.1] ([INITIATOR.sub.i,t]) + [[alpha].sub.2] (Investment Inefficiency [Variable.sub.i,t]) + [[alpha].sub.31](INITIATOR x Investment Inefficiency [Variable.sub.i,t]) + [[alpha].sub.4]([LNTA.sub.i,t]) + [[alpha].sub.5][FOROPS.sub.i,t]) + [[alpha].sub.6]([XDOPS.sub.i,t]) [[alpha].sub.7]([INVREC.sub.i,t]) + [[alpha].sub.8]([ROA.sub.i,t]) + [[alpha].sub.g] ([LIQ.sub.i,t]) + [[alpha].sub.10]([DA.sub.i,t]) + [[alpha].sub.11]([BIG4.sub.i,t]) + [[alpha].sub.12]([NEWAUDITOR.sub.i,t]) + [[alpha].sub.13]([GC.sub.i,t]) + [[alpha].sub.14]([RESTATEMENT.sub.i,t]) + [[alpha].sub.15]([UNOUALIFIED.sub.i,t]) + [[alpha].sub.15]([WEAKIC.sub.i,t]) + [[alpha].sub.17]([ICFR.sub.i,t]) + [[alpha].sub.18]([BUSY.sub.i,t]) + [[alpha].sub.19]([GROWTH.sub.i,t]) + [[alpha].sub.20]([LOSS.sub.i,t]) + [[alpha].sub.21]([NYSE.sub.i,t]) + [[alpha].sub.22] ([SQSEGS.sub.i,t]) + [[alpha].sub.23](INDUSTRY [EFFECTS.sub.i,t]) + [[alpha].sub.24](FIXED YEAR [EFFECTS.sub.i,t]) + [e.sub.i,t] (1)

The dependent variable LNAF is the natural logarithm of the audit fees for firm i at fiscal year-end, and test variable INITIATOR is equal to 1 if the firm initiated dividends in the current year. The variable INITIATOR is interacted with the Investment Inefficiency Variable used in the regression. The Investment Inefficiency Variable measures the propensity of a firm to invest inefficiently. Relying on the methodology of Biddle, Hilary and Verdi (2009), we estimate the following equation:

[INVESTMENT.sub.i,t,t+1] = [beta]0 + [beta]1([GROWTH.sub.i,t-1]) + [[epsilon].sub.i,t] (2)

where, INVESTMENT is the average 2-year (t and t+1) capital investment, and GROWTH is the percentage change in sales at fiscal year-end t-1. More specifically, investment is computed by summing three Compustat variables (CAPX + XRD + ACQ) minus SPPE and then dividing the result by total assets.

We estimate equation 2 from 2003 to 2012 and extract the residual, which proxies for misallocation of resources. The residual identifies the firm's investment efficiency compared to peer firms, and we label this proxy, INEFF, for investment inefficiency. This proxy is continuous and includes negative and positive values so that further values from zero represent more inefficiency in the firm's investment. Therefore, we square this term to construct a proxy for increasing inefficiency in investment (INEFF-SQ) following Morgado and Pindado (2003). (1)

Another way to test inefficient investment is to split the variable between positive and negative values. Residuals greater than or equal to zero represent over-investment (OVERINVEST), and higher values of OVERINVEST indicate more inefficient over-investment. Residuals less than zero represent under-investment (UNDERINVEST). We multiply the under-invest residuals by negative one so that a higher value of UNDERINVEST indicates more inefficient under-investment. Over-invested (under-invested) firms are spending more (less) on investment than they should, given their rate of sales growth. As the value for OVERINVEST (UNDERINVEST) increases, the firm is more inefficiently over-invested (under-invested).

Control variables are defined as follows: LNTA is the natural log of total assets; FOROPS is equal to 1 if the firm has foreign exchange gain or loss at the end of the current fiscal year, and 0 otherwise; XDOPS is equal to 1 if the firm i reported extraordinary items or discontinued operations in the current fiscal year; INVREC is inventory and accounts receivables scaled by total assets; ROA is return on assets, defined as the ratio of operating income to total assets; LIQ is the ratio of current assets to current liabilities; DA is total debt divided by total assets; BIG4 is equal to 1 if firm i engages a Big 4 auditor in the current fiscal year-end; NEWAUDITOR is equal to 1 if the auditor is in the first or second year of an engagement in the current fiscal year; GC is equal to 1 if the firm received a going concern opinion on its financial statements in the current fiscal year; RESTATEMENT is equal to 1 if the firm restated its prior financial statements in the current fiscal year; UNQUALIFIED is equal to 1 if the firm received an unqualified opinion on its financial statements in the current fiscal year; WEAKIC is equal to 1 if the firm had an adverse opinion with respect to internal controls over financial reporting in the current fiscal year (where IC_IS_EFFECTIVE = N); ICFR is equal to 1 if the firm received an opinion on its internal control; BUSY is equal to 1 if the firm has a December year end; GROWTH is the percentage sales growth from the prior fiscal year; LOSS is equal to 1 if ROA is negative in the prior year; NYSE is equal to 1 if the firm is traded on the New York Stock Exchange; and SQSEGS is the number of business segments squared. All models include controls for years and for Fama and French (1997) industries.

Over- versus Under- invested Samples

We follow the methodology of Biddle, Hilary and Verdi (2009) to construct the over- and under- investment proxies. In their study, Biddle, Hilary and Verdi (2009) examine over-invested firms as a separate sample from under-invested firms. While we follow this methodology, we also combine the samples. This requires setting the OVERINVEST (UNDERINVEST) proxy equal to zero for firms that are under-invested (over-invested). Consequently, the regression model is estimated for all firms (with over- and under- invest proxies set to zero where they are undefined), for only over-invested firms and for only under-invested firms.

High and Low Growth

If a firm is investing inefficiently, then doing more investing is expected to magnify the inefficiency. In their study of entrenched CEOs, Grinstein and Hribar (2004) conclude that the CEOs who destroy the most value do the most investing relative to their firm size. Morck, Shleifer and Vishny (1990) find evidence that managers make value-depreciating acquisitions of high growth targets to buy growth. High growth firms that invest inefficiently should be rewarded more for initiating a dividend. To evaluate the effect of high growth, we estimate equation 1 after dividing the sample by the median of sales growth.

Details of the sample collection process are provided in Table 1, Panel A, and the sample distribution by year is displayed in Panel B.

Our sample selection, Panel A, results in 17,520 firm-years, which can be further divided into non-paying firms (12,042 firm-years), dividend paying firms (5,196 firm-years) and dividend initiators (282 firm-years). In Table 1, Panel B, we show that the lowest number of initiators, 12 firms, occurs in the year 2008. The largest number of initiators is found in 2004. Fifty-six percent of initiators are contained in the first five years of the sample, 2003-2007, and 44 percent of dividend initiators are represented in the latter half of the sample, 2008-2012.

Table 2 provides descriptive data for the sample, divided into three sections for treatment variables, continuous control variables and dichotomous proxies.

Audit fees and investment variables are shown in Panel A of Table 2. The mean natural log of audit fees is 13.61 for non-initiating firms, with a standard deviation of 1.24. This is very similar to the mean, 13.57, and standard deviation, 1.26, for dividend initiators. There are 6,232 (11,006) over-invested (under-invested) control firm-years, and 57 (225) of the dividend initiators are over-invested (under-invested). Continuous variables provided in Panel B show that the mean (median) total assets of control firms are $2,735 ($393) million, which is lower than total assets of dividend-initiating firms with a mean (median) of $3,240 ($545) million. Profitability of control firms is lower than for dividend-initiating firms. The mean ROA for control firms, 0.03, is lower than that of initiating firms, 0.13, but mean liquidity is higher for control firms, 3.05, compared to 2.99 for initiators. Leverage is slightly higher for control firms, with a mean (median) debt to asset ratio of 0.52 (0.450) versus the initiating firms' leverage, 0.49 (0.40). Discrete variables, displayed in Panel C, indicate that control firms are slightly more likely to have discontinued operations, a going concern opinion and weak internal control. Dividend initiators are more likely to have a restatement, a Big 4 auditor, a new auditor and an unqualified financial statement opinion. In the initiators and control samples, 13 percent of firms have foreign operations.

Results

We first assess the expectation that dividend initiation for inefficiently invested firms results in lower audit fees. Table 3 provides analysis for our first hypothesis, H1, examining whether the firm's investment opportunity set influences the impact of the dividend initiation on audit fees.

The results for model 1 reflect analysis of the audit fee model for dividend initiators compared to all firms whether they currently pay dividends or not. In model 1, dividend initiating firms have significantly lower audit fees with a coefficient of -0.094, and dividend payers also have significantly lower audit fees with a slightly lower coefficient of -0.088. The coefficient on OVERINVEST is not significant, but UNDERINVEST has a highly significant, negative coefficient of -0.020. Taken together, these results indicate that, overall, firms are assessed lower audit fees when there are excess available projects that have not been funded. However, we do not find evidence in the larger population of firms that over-invested firms have higher audit fees.

In model 2, an interaction term is included to quantify dividend initiation acting on over- and under-investment. Dividend initiation for over-invested firms results in lower audit fees, as the coefficient is -2.337 and significance is at the 5 percent level. However, the coefficient on dividend initiation for under-invested firms is not significant.

In model 3, we leave the over- and under- investment residuals combined in their original, continuous variable (INEFF), where negative values indicate increasingly under-invested firms and positive values indicate increasingly over-invested firms. The advantage of this proxy is that it retains the over- and under- investment effects as a single variable and does not require setting undefined values to zero. Because our test is on firms that deviate further from zero in investment inefficiency, we include a squared term, INEFF-SQ, following Morgado and Pindado (2003). In this model, we find evidence that the more inefficiently invested a firm is, the more its audit fees decline on initiating a recurring dividend. This result supports H1.

H2 and H3 examine the effect of dividend initiation compared to firms that are not committed to a recurring dividend. In Table 4, model 1, OVERINVEST and UNDERINVEST are included in the same model. Compared to non-paying firms, dividend initiators earn lower audit fees, and then they are over-invested (INITxOVERINVEST). The result for the proxy, INITxOVERINEST, supports H2 in that over-investment influences the perceived audit risk related to dividend initiation. This result is consistent with both signaling and agency explanations of auditor dividend initiations pricing. From an agency perspective, the negative coefficient suggests that dividend initiating firms with remaining positive net present value projects have lower audit fees than those that do not initiate dividends. This is consistent with firms that are less likely to be over-invested in value-depreciating projects having lower perceived audit risk.

Model 1 also shows that auditors perceive lower risk for dividend initiating firms when they are under-invested (INITxUNDERINVEST). Since under-invested firms are less likely to have the same agency problems as over-invested firms, this suggests that auditors perceive a signal of reduced risk with dividend initiation. The results for over-invested initiators has higher significance, 5 percent as opposed to 10 percent, and a larger coefficient than for under-invested initiators. The magnitude and significance of the coefficient on INITxOVERINVEST indicates that over-invested dividend initiators have lower priced risk than their under-invested counterparts.

In models 2 and 3, we follow Biddle, Hilary and Verdi (2009) in dividing the sample by the firms' status as over- or under-invested. Among over-invested firms in model 2, we find that dividend initiators have lower audit fees. The coefficient, -5.056, is highly significant and more than offsets the higher fees assessed to over-invested firms with a coefficient of 0.158. Therefore, we find support for Hypothesis 2, that over-invested firms have lower fees when they initiate dividends.

Examining under-invested firms in model 3, we find that the coefficient on under-invested dividend initiators, INITxUNDERINVEST, is not significant. While previously, in model 1, we noted that under-invested dividend initiators had lower audit fees, supporting H3 with weak significance (p-value = 0.054), the results in model 3 for INITxUNDERINVEST are not significant and do not support H3.

The propensity for firms to misallocate assets under high versus low growth is presented in Table 5. The sample of non-paying firms and dividend initiators is divided by the median of sales growth. We expect that investment inefficiency is magnified in firms that have higher growth. In model 1, for low growth firms, the interaction terms for dividend initiation and over- or under- investment are not significant. In model 2, limited to firms with sales growth above the median, dividend initiators that are the most over- or under- invested have the lowest audit fees in the year of dividend initiation, supporting Hypotheses 2 and 3. We interpret these findings as strong support for H2, that over-invested firms have lower fees on dividend initiation, and mixed support for under-invested firms having lower fees on dividend initiation. The under-invest proxy is Less indicative of agency problems (Biddle, Hilary and Verdi 2009) and, where significant, indicates that auditors respond favorably to the firm's market signal and thus assess lower audit fees.

Dividend paying firms have already engaged the quasi-contract of dividend initiation and have committed to paying a recurring dividend. In Table 6, audit fees for dividend initiating firms are compared to those for dividend paying firms. In model 1, the interaction term for dividend initiators that are over- or under- invested is not significant. In model 2, comparing only over-invested firms, dividend initiators that are over-invested have audit fees that are significantly lower, with 95 percent confidence, than dividend paying firms. In untabulated results, under-invested initiators do not have significantly lower fees compared to other under-invested, paying firms. High growth payers are examined in model 3, and we note that audit fee reduction is only significant for over-invested dividend initiators compared to dividend payers. This finding supports H4 in that inefficiently invested initiating firms have lower audit fees compared to dividend paying firms. We do not find evidence that under-invested dividend initiators have lower audit fees than dividend paying firms. Therefore, H4 is supported for over-invested, initiating firms but not for under-invested, initiating firms.

Robustness Tests

We conduct a number of robustness tests. (2) The control sample is large compared to our test sample of dividend initiators, so we repeat regressions using a propensity-matched control sample of firms. The propensity matching is done using a logit model to determine the likelihood of similarity to the dividend initiating firm, and three matches are chosen for each dividend initiator without replacement. The regression results are similar to those using the Compustat universe of dividend paying and non-paying firms.

Biddle, Hilary and Verdi (2009) provide an alternative proxy for over- and under- investment, computed with Tobin's Q rather than sales growth. The authors explain that this measure is less reliable because of endogeneity between Tobin's Q and investment. Results using this alternate proxy are less significant but are consistent with the sales growth measure.

To test for endogeneity, we conduct a Hausman test on the models with the over-invest and under-invest proxies. The residuals are obtained from the first pass regression on the dividend initiation variable, but in the second stage regression, we find that the residuals are not statistically significant. We conclude that endogeneity is not present in the models and that the test statistics of the OLS regression are reliable.

An alternate form of payout policy is share repurchase, so we perform an additional test to evaluate the effect of share repurchases by including a repurchase dummy variable in our regression models. While the repurchase dummy is statistically significant in reducing audit fees, the coefficient is almost zero, indicting a very low level of economic significance. This is consistent with our argument that reducing excess cash addresses the agency problem with over-investment. Share repurchase is weakly significant but does not have the economic effect of a firm's commitment to continued dividend payout. Therefore, we find support for our hypotheses regarding dividend initiation.

Our final test is to construct a timeline of audit fees surrounding the year of dividend initiation. This timeline requires that firms have audit fee data for the five years surrounding dividend initiation (year -2 through year +2), which reduces the sample to 69 firms. The result is consistent with our argument that audit fees are lower due to the agency and signaling effects of dividend initiation. Audit fees are on a somewhat sharp, upward trajectory prior to dividend initiation. Then, they decline in the year of dividend initiation before increasing more slowly for the two years post- dividend initiation. We interpret this to indicate that the signaling effect and/or governance commitment to dividend payout brought audit fees down in the year of dividend initiation; then, over the period from the year of dividend initiation through the next two years, audit fees remain lower either due to the realization of the signaled, strong performance or the benefits of the reduced agency problems.

Conclusion

This paper is the first to examine how auditors price dividend initiations in the context of the firm's level of investment. Specifically, we test whether the auditor's perception of risk due to dividend initiation is explained by the firm's investment efficiency. We document a negative and significant association between audit fees and the extent of a firm's investment inefficiency when dividends are initiated. Auditors perceive either a positive signal arising from managerial confidence in future cash flows and corresponding lower risk of client distress or from lower risk associated with a reduction of discretionary cash flows available to managers.

Furthermore, we find that over-invested initiating firms exhibit lower audit fees, but we find mixed evidence that under-invested initiating firms have lower fees. This finding is consistent with the argument that under-invested firms are less likely to suffer from agency problems related to over-investment and have better financial prospects. When firms are over-invested, audit fees reflect the risk of potential agency conflict.

Firms that already pay dividends have already reduced their propensity to over-invest. However, our results show that initiators have lower audit fees proportional to the extent of their over-investment when compared to dividend paying firms. This finding is not significant when the control sample is all dividend paying firms. Rather, over-invested dividend initiators are perceived as lower risk by auditors compared to over-invested dividend paying firms, which may have continued agency problems even though they pay dividends. Dividend initiating firms that are over-invested also have lower audit fees when compared to high growth dividend payers. We interpret this consistent with Michaely, Thaler and Womack (2005), who determine that dividend initiations are a more abrupt, visible shift in corporate financing policy than dividend increases and therefore a more unequivocal signal. However, the results comparing to dividend paying firms are not as conclusive as the control sample of non-paying firms.

Overall, we find that a firm's dividend initiation appears to moderate the perceived risk associated with asset misappropriation, in the sense that initiators with a greater propensity to over-invest exhibit lower audit fees than either payers or nonpayers. Further, we find that even firms already paying dividends in order to mitigate over-investment's discretionary cash flow hazards are priced higherthan dividend initiators, suggesting that signaling plays a role in the reduction of perceived audit risk. Furthermore, we find some evidence, although mixed, that under-invested firms have lower audit fees when they initiate dividends. Thus, we find support for the agency and signaling explanations of the negative association between audit fees and dividend initiations, but the strongest support is for over-invested firms with the potential for agency problems.

APPENDIX A

Variable Definitions
Panel A: Dependent and Investment Inefficiency Variables

VARIABLE     DEFINITION

LNAF         The natural log of (1+AF), where Audit fees (AUDIT FEES)
             is for the client company for fiscal year t, from Audit
             Analytics.
INITIATOR    1 if the firm initiated a cash dividend. Initiating firms
             are identified with CRSP dividend codes 1222, 1232, 1242
             or 1252 and have not paid a regular dividend in the past
             three years.
PAYER        1 if the firm pays a recurring cash dividend. Dividends are
             identified with CRSP dividend codes 1222, 1232, 1242 or
             1252.
INEFF        We estimate Equation 2 from 2003 to 2012 and extract the
             residual, which proxies for misallocation of resources,
             INEFF, or investment inefficiency. This residual is the
             firm's investment efficiency compared to industry peer
             firms.
OVERINVEST   Over-investment is the positive values of INEFF, otherwise,
             set to zero.
UNDERINVEST  Under-investment is the negative values of INEFF multiplied
             by (-1) so that higher values indicate more inefficient
             under-investment, otherwise, set to zero.

Panel B: Audit Fee Model Control Variables (all variables from
Compustat unless stated otherwise)

   VARIABLE     DEFINITION

    FOROPS      1 if the firm has foreign exchange gain or loss (FCA)
                at the end of the current fiscal year.
    XDOPS       1 if the firm reported extraordinary items or
                discontinued operations (XIDO) in the current fiscal
                year.
 UNQUALIFIED    1 if the firm received an unqualified audit opinion in
                the current fiscal year.
    ICWEAK      1 if the firm had a material weakness in internal
                control in the current fiscal year.
     ICFR       1 if firm i received an audit opinion on internal
                controls over its financial reporting in the current
                fiscal year, using (IC_IS_EFFECTIVE) from Audit
                Analytics.
GOING_CONCERN   1 if firm i received a going concern audit opinion in
                the current fiscal year, based on GOING_CONCERN from
                Audit Analytics.
 RESTATEMENT    1 if the firm filed a restatement of previously issued
                financial statements in the current fiscal year, from
                Audit Analytics.
     BIG5       1 if engagement auditor for the current fiscal year is
                a Big4/5 auditor, using Audit Analytics and Compustat
                auditor information.
  DECEMBERYE    1 if fiscal year-end month is December.
   INITIAL      1 if an auditor is in the first or second year of an
                engagement in the current fiscal year, using Audit
                Analytics and Compustat auditor information.
     LOSS       1 if ROA at fiscal year-end is less than 0.
     NYSE       1 if the firm is traded on the NYSE index.
    INVREC      Inventory and accounts receivable scaled by total
                assets ((RECT + INVT)/AT) at fiscal year-end.
     ROA        Return on assets, or the ratio of income before
                extraordinary items at fiscal year-end (IB) to lagged
                fiscal year-end total assets (AT).
      DA        The ratio of total liabilities to total assets ((DLC +
                DLTT)/AT) at fiscal year-end.
     LIQ        The ratio of current assets net of inventory (ACT-INVT)
                to current liabilities (LCT) at fiscal year-end.
     LNAT       The natural logarithm of 1 plus total assets (AT) at
                fiscal year-end for the audit fee year.
    SOSEGS      The square root of the number of firm segments.
    GROWTH      Growth in sales (SALE), found as the proportional
                change in sales for the current fiscal year over the
                prior fiscal year.


References

Allen, F., A. Bernardo, and I. Welch. "A Theory of Dividends Based on Tax Clienteles." Journal of Finance, vol. 55, no. 6, 2000, pp. 2499-2536.

Asquith, P., and D. Mullins, Dr. "The Impact of Initiating Dividend Payments on Shareholders' Wealth." Journal of Business, vol. 56, no. 1, 1983, pp. 77-95.

Bedard, J.C., and K.M. Johnstone. "Earnings Manipulation Risk, Corporate Governance Risk, and Auditors' Planning and Pricing Decisions." Accounting Review, vol. 79, no. 2, 2004, pp. 277-304.

Benartzi, S., R. Michaely, and R. Thaler. "Do Changes in Dividends Signal the Future or the Past?" Journal of Finance, vol. 52, no. 3, 1997, pp. 1007-1034.

Berger, A.N., and G.F. Udell. "Relationship Lending and Lines of Credit in Small Firm Finance." Journal of Business, vol. 68, no. 3, 1995, pp. 351-381.

Bhattacharya, S. "Imperfect Information, Dividend Policy, and The Bird in the Hand' Fallacy." Bell Journal of Economics, vol. 10, no. 1, 1979, pp. 259-270.

Biddle, G.C., G. Hilary, and R.S. Verdi. "How Does Financial Reporting Quality Relate to Investment Efficiency?" Journal of Accounting and Economics, vol. 48, no. 2, 2009, pp. 112-131.

Brav, A., D.R. Graham, C.R. Harvey, and R. Michaely. "Payout Policy in the 21st Century." Journal of Financial Economics, vol. 77, no. 3, 2005, pp. 483-527.

Bulan, L., N. Subramanian, and L. Tanlu. "On the Timing of Dividend Initiations." Financial Management, vol. 36, no. 4, 2006, pp. 31-65.

Caskey, X, and M. Hanlon. "Dividend Policy at Firms Accused of Accounting Fraud." Contemporary Accounting Research, vol. 30, no. 2, 2013, pp. 818-850.

Charitou, A., N. Lambertides, and G. Theodoulou. "Losses, Dividend Reductions, and Market Reaction Associated with Past Earnings and Dividends Patterns." Journal of Accounting, Auditing & Finance, vol. 26, no. 2, 2011, pp. 351-382.

Chy, M., and 0-K. Hope. "Does High Auditor Litigation Risk Discourage Corporate Innovation?" econbiz.de, 10 Sept 2016, https://www.econbiz.de/Record/does-high-auditor-litigation-risk-discourage-corporate-innovation-chy-mahfuz/10011547582.

Dyl, E.A., and R.A. Weigand. "The Information Content of Dividend Initiations: Additional Evidence." Financial Management, vol. 27, no. 3, 1999, pp. 27-35.

Easterbrook, F. "Two Agency Cost Explanations of Dividends." American Economic Review, vol. 74, no. 4, 1984, pp. 650-59.

Fama, E.F., and K.R. French. "Industry Costs of Equity." Journal of Financial Economics, vol. 43, no. 2,1997, pp.153-193.

Francis, 1R., and D. Wang. "Impact of the SEC's Public Fee Disclosure Requirement on Subsequent Period Fees and Implications for Market Efficiency." Auditing: A Journal of Practice & Theory, vol. 24, no. s-1, 2005, pp. 145-160.

Gleason, K.C., A J. Greiner, and Y.H. Kannan. "Auditor Pricing of Excess Cash Holdings." Journal of Accounting, Auditing & Finance, vol. 32, no. 3, 2017, pp. 423-443.

Glover, S.M., D.F. Prawitt, 11 Schultz, Jr., and M.F. Zimbelman. "A Test of Changes in Auditors' Fraud-related Planning Judgments since the Issuance of SAS No. 82." Auditing: A Journal of Practice & Theory, vol. 22, no. 2, 2003, pp. 237-251.

Griffin, PA., D.H. Lont, and Y. Sun. "Agency Problems and Audit Fees: Further Tests of the Free Cash Flow Hypothesis." Accounting & Finance, vol. 50, no. 2, 2010, pp. 321-350.

Grinstein, Y., and P. Hribar. "CEO Compensation and Incentives: Evidence from M&A Bonuses." Journal of Financial Economics, vol. 73, no. 1,2004, pp. 119-143.

Harford, 1. "Corporate Cash Reserves and Acquisitions." Journal of Finance, vol. 54, no. 6,1999, pp. 1969-1997.

Harford, 1., and K. Li. "Decoupling CEO Wealth and Firm Performance: The Case of Acquiring CEOs." Journal of Finance, vol. 62, no. 2, 2007, pp. 917-949.

Hay, D.C., W.R. Knechel, and N. Wong. "Audit Fees: A Meta-analysis of the Effect of Supply and Demand Attributes." Contemporary Accounting Research, vol. 23, no. 1, 2006, pp.141-191.

Healy, RM. "The Effect of Bonus Schemes on Accounting Decisions." Journal of Accounting and Economics, vol. 7, no. 1, 1985, pp. 85-107.

Healy, RM., and K.C. Palepu. "Earnings Information Conveyed by Dividend Initiations and Omissions." Journal of financial Economics, vol. 21, no. 2,1988, pp. 149-175.

Hermalin, B.E., and M.S. Weisbach. "Endogenously Chosen Boards of Directors and Their Monitoring of the CEO." American Economic Review, vol. 88, no. 1, 1998, pp. 96-118.

Houston, R.W., M.F. Peters, and 3.H. Pratt. "The Audit Risk Model, Business Risk and Audit-planning Decisions." Accounting Review, vol. 74, no. 3, 1999, pp. 281-298.

Houston, R.W., M.F. Peters, and J.H. Pratt. "Non-litigation Risk and Pricing Audit Services." Auditing: A Journal of Practice & Theory, vol. 24, no. 1, 2005, pp. 37-53. http://www.cbrands.com/news-media/constellation-brands-announces-initiation-quarterly-dividend. Last accessed Nov. 9, 2015. https://www.moodys.com/research/Moodys-says-Constellation-Brands-dividend-initiation-is-credit-negative-but--PR_322583. Last accessed Nov. 9, 2015.

Jensen, M.C. "Agency Cost of Free Cash Flow, Corporate Finance, and Takeovers." American Economic Review, vol. 76, no. 2, 1986, pp. 323-329.

John, K., and X Williams. "Dividends, Dilution and Taxes: A Signaling Equilibrium." Journal of Finance, vol. 40, no. 4, 1985, pp. 1053-1070.

John, T.A. "Accounting Measures of Corporate Liquidity, Leverage, and Costs of Financial Distress." Financial Management, vol. 22, no. 3,1993, pp. 91-100.

Jones, J.S., J. Gu, and P. Liu. "Do Dividend Initiations Signal a Reduction in Risk? Evidence from the Option Market." Review of Quantitative Finance and Accounting, vol. 42, no. 1, 2014, pp. 143-158.

Johnstone, K.M. "Client-acceptance Decisions: Simultaneous Effects of Client Business Risk, Audit Risk, Auditor Business Risk, and Risk Adaptation." Auditing: A Journal of Practice & Theory, vol. 19, no. 1,2000, pp. 1-25.

Kale, J.R., O. Kini, and J.D. Payne. "The Dividend Initiation Decision of Newly Public Firms: Some Evidence on Signaling with Dividends." Journal of Financial and Quantitative Analysis, vol. 47, no. 02, 2012, pp. 365-396.

Koch, A.S., and A.X. Sun. "Dividend Changes and the Persistence of Past Earnings Changes." Journal of Finance, vol. 59, no. 5, 2004, pp. 2093-2116.

La Porta, R., F. Lopez-De-Silanes, A. Shleifer, and R.W. Vishny. "Agency Problems and Dividend Policies around the World." Journal of Finance, vol. 55, no. 1, 2000, pp. 1-33.

Lang, L.H.P., R.M. Stultz, and R.A. Walkling. "A Test of the Free Cash Flow Hypothesis." Journal of Financial Economics, vol. 29, no. 2, 1991, pp. 315-335.

Lawson, B.P., and D. Wang. "The Earnings Quality Information Content of Dividend Policies and Audit Pricing." Contemporary Accounting Research, vol. 33, no. 4, 2016, pp. 1685-1719.

Lyon, ID., and M.W. Maher. "The Importance of Business Risk in Setting Audit Fees: Evidence from Cases of Client Misconduct." Journal of Accounting Research, vol. 43, no. 1, 2005, pp. 133-151.

Menon, K., and D.D. Williams. "The Insurance Hypothesis and Market Prices." Accounting Review, vol. 69, no. 2, 1994, pp. 327-342.

Michaely, R., R.H. Thaler, and K.L. Womack. "Price Reactions to Dividend Initiations and Omissions: Overreaction or Drift?" Journal of Finance, vol. 50, no. 2, 1995, pp. 573-608.

Miller, M.H., and F. Modigliani. "Dividend Policy, Growth, and the Valuation of Shares." Journal of Business, vol. 34, no. 4, 1961, pp. 411-433.

Miller, M.H.,and K. Rock. "Dividend Policy under Asymmetric Information." Journal of Finance, vol. 40, no. 4, 1985, pp. 1031-1051.

Morck, R., A. Shleifer, and R.W. Vishny. "Do Managerial Objectives Drive Bad Acquisitions?" Journal of Finance, vol. 45, no. 1, 1990, pp. 31-48.

Morgado, A., and 1 Pindado. "The Underinvestment and Overinvestment Hypotheses: An Analysis Using Panel Data." European Financial Management, vol. 9, no. 2, 2003, pp. 163-177.

Myers, S.C. "Determinants of Corporate Borrowing." Journal of Financial Economics, vol. 5, no. 2, 1977, pp. 147-175.

Nohel, T, and V. Tarhan. "Share Repurchases and Firm Performance: New Evidence on the Agency Costs of Free Cash Flow." Journal of Financial Economics, vol. 49, no. 2, 1998, pp. 198-222.

Officer, M.S. Dividend Policy, Dividend Initiations, and Governance. Dissertation, University of Southern California, 5 Oct. 2006.

Officer, M.S. "Overinvestment, Corporate Governance, and Dividend Initiations." Journal of Corporate Finance, vol. 17, no. 3, 2011, pp. 710-724.

PCAOB, Auditing Standard #12. pcaobus.org, http://pcaobus.org/Standards/Auditing/Pages/Auditing_Standard_12.aspx#riskassessment Last accessed 9 Nov. 2015.

Petersen, M.A. "Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches." Review of Financial Studies, vol. 22, no. 1,2009, pp. 435-480.

Raghunandan, K., and D.V. Rama. "SOX Section 404 Material Weakness Disclosures and Audit Fees." Auditing: A Journal of Practice & Theory, vol. 25, no. 1, 2006, pp. 99-114.

Simunic, D.A. "The Pricing of Audit Services: Theory and Evidence." Journal of Accounting Research, vol. 18, no. 1,1980, pp. 161-190.

Skinner, D.l, and E. Soltes. "What Do Dividends Tell Us about Earnings Quality?" Review of Accounting Studies, vol. 16, no. 1, 2011, pp. 1-28.

Smith, D., and A. Pennathur. "Signaling versus Free Cash Flow Theory: What Does Earnings Management Reveal about Dividend Initiation?" Journal of Accounting, Auditing and Finance, 4 Aug. 2016, http://journals.sagepub.com/doi/10.1177/0148558X17724051.

Whisenant, S., S. Sankaraguruswamy, and K. Raghunandan. "Evidence on the Joint Determination of Audit and Non-audit Fees." Journal of Accounting Research, vol. 41, no. 4, 2003, pp. 721-744.

Zimbelman, M.F. "The Effects of SAS No. 82 on Auditors' Attention to Fraud Risk Factors and Audit Planning Decisions." Journal of Accounting Research, vol. 35, 1997, pp. 75-97.

MALEK ALSHARAIRI

German-Jordanian University

DEBORAH DRUMMOND SMITH

Cleveland State University

MINA GLAMBOSKY

The City University of New York - Brooklyn College

KIMBERLEY GLEASON

American University of Sharjah

(i) On April 9, 2015, the CEO of an international alcoholic products distributor, Constellation Brands, announced the company would initiate a regular quarterly dividend to its common shareholders. Explaining his motivation to establish the dividend, the CEO stated that the decision "represents a significant milestone in the company's history and demonstrates confidence in our growth prospects, free cash flow generation and financial outlook over the long-term" (http://www.cbrands.com/news-media/constellation-brands-announces-initiation-quarterly-dividend). Further, he reinforced the argument that Constellation has stable cash-generation and the capacity to support a continued high return on investment for its sharefolders, while maintaining its target leverage ratio. The CEO's announcement affirmed finance theory's claim that dividend initiations represent a cedible signal to the market regarding the firm's ability to maintain the dividend stream. Despite the CEO's evident optimism, Moody's released an announcement that the dividend initiation was a credit negative event for Constellation Brands, even as its share price rose by 1 percent (https://www.moodys.com/research/Moodys-says-Constellation-Brands-dividend-initiation-is-credit-negative-but--PR_322583).

(1) Morgado and Pindado (2003) investigate the quadratic nature of investment inefficiency.

(2) The details of these tests are untabulated but are available upon request. We are very grateful to Associate Editor, Dr. Pawel Bilinski, for recommending each of these robustness tests and for many other suggestions that improved the paper.
TABLE 1
Sample Determination and Data Sources

             Panel A. Sample Description
                                   Firm-Year Observations

 Full Sample of firms with CRSP
 and Compustat data and dividend    52,405
      designation 2003-2012
Less: Firm-year observations with
   regulated or financial SIC      (15,373)
Less: Firm-year observations with
     missing audit fee data        (13,795)
Less: Firm-year observations with
       missing variables:           (5,717)
          Total Sample              17,520
          Payer Status
           Non-payers               12,042
         Dividend Payers             5,196
       Dividend Initiators             282
   All Firm-Year Observations       17,520

         Panel B. Sample Distribution by Year

            Non-Initiating Firms
Year  No.    Percent  Total   Total
      Firms           Firms   Percent

2003  1,715  10.0      1,715   10.0
2004  1,931  11.2      3,646   21.2
2005  1,867  10.8      5,513   32.0
2006  1,832  10.6      7,345   42.6
2007  1,836  10.7      9,181   53.3
2008  1,891  11.0     11,072   64.2
2009  1,912  11.1     12,984   75.3
2010  1,827  10.6     14,811   85.9
2011  1,794  10.4     16,605   96.3
2012    633   3.7     17,238  100.0

        Panel B. Sample Distribution by Year
             Dividene Initiating Firms
Year   Year  No.    Percent  Total  Total
             Firms           Firms  Percent

2003    2003  41     14.5     41     14.5
2004    2004  43     15.3     84     29.8
2005    2005  35     12.4     119    42.2
2006    2006  20      7.1     139    49.3
2007    2007  18      6.4     157    55.7
2008    2008  12      4.3     169    59.9
2009    2009  16      5.7     185    65.6
2010    2010  38     13.5     223    79.1
2011    2011  40     14.2     263    93.3
2012    2012  19      6.7     282   100.0

TABLE 2
Sample Description
This table provides descriptive statistics of variables used in
statistical tests. Panel A shows descriptive statistics for treatment
and interaction variables for non-initiators versus initiators, Panel B
shows descriptive statistics for control variables and Panel C shows
descriptive statistics for discrete variables.

Panel A. Descriptive Statistics of the Treatment and Interaction
Variables

Non-Initiators
                N       Mean   St. Dev.  25th %  Median  75th %

    LAF         17,238  13.61  1.24      12.75   13.61   14.41
   INEFF        17.238   0.00  0.13      -0.08   -0.04    0.05
OVERINVEST       6,232   0.13  0.13       0.05    0.09    0.16
UNDERINVEST     11,006   0.07  0.03       0.05    0.07    0.10

Initiators
                N       Mean   St. Dev.  25th %  Median  75th %

    LAF         282     13.57  1.26      12.64   13.62   14.40
   INEFF        282     -0.05  0.06      -0.09   -0.07   -0.03
OVERINVEST       57      0.07  0.04       0.03    0.06    0.08
UNDERINVEST     225      0.07  0.03       0.05    0.07    0.10

Panel B. Descriptive Statistics of Continuous Variables

Non-Initiators
                 N       Mean          St. Dev.

Total Assets     17,238  2,735         10,457
   INVREC        17,238      0.27           0.18
     ROA         17,238      0.03           0.24
     LIQ         17,238      3.05           3.10
     DA          17,238      0.52           0.28
   SQSEGS        17,238      1.41           0.49

Non-Initiators
                  25th %  Median     75th %

Total Assets      102.03  393       1,560
   INVREC           0.13    0.24        0.38
     ROA            0.01    0.07        0.12
     LIQ            1.47    2.18        3.49
     DA             0.36    0.45        0.61
   SQSEGS           1.00    1.41        1.73

Initiators
                 N       Mean   St. Dev.  25th %  Median  75th %

 Total Assets       282  3.240  13,124    148     545     2.048
    INVREC          282  0.29    0.19       0.15    0.26  0.41
     ROA            282  0.13    0.11       0.08    0.12  0.17
     LIQ            282  2.99    2.24       1.52    2.21  3.50
      DA            282  0.49    0.19       0.36    0.40  0.58
    SQSECS          282  1.39    0.47       1.00    1.41  1.73

Panel C. Descriptive Statistics of Discrete Variables

Non-Initiators
                N       Mean

  FOROPS        17,238  0.13
   XDOPS        17,238  0.09
RESTATEMENT     17,238  0.13
    GC          17,238  0.02
   BIG4         17,238  0.77
  INITIAL       17,238  0.13
UNQUALIFIED     17,238  0.57
  WEAKIC        17,238  0.05

Initiators
                N       Mean

  FOROPS           282  0.13
   XDOPS           282  0.08
RESTATEMENT        282  0.16
    GC             282  0.00
   BIG4            282  0.78
  INITIAL          282  0.15
UNQUALIFIED        282  0.65
  WEAKIC           282  0.03

TABLE 3
Audit Fees for Dividend Initiating Firms That Are Investing
Inefficiently
Table 3 provides an analysis of our first hypothesis, related to the
association between audit fees and dividend initiation (INITIATOR)
after controlling for all baseline model variables. The benchmark
sample is all firms, including dividend payers and non-payers. The
sample covers the time period 2003 to 2012 and excludes firms in the
financial and utility sectors. We include year and industry dummies in
all model analysis. The dependent variable LAF is the natural logarithm
of the audit fees in $ thousands. INITIATOR is equal to 1 if the firm
initiated dividends in the current fiscal year. Our proxy for
over-investment is OVERINVEST, and our proxy for underinvestment is
UNDERINVEST. INEFF is the extent of the firm's inefficiency in
investing. INEFF-SQ is INEFF, squared. Over-invest and Under-invest are
the positive and negative values of INEFF, but UNDERINVEST is
multiplied by-1 so that higher values represent more inefficient
under-investment. Initiator is interacted with over- and under-invest,
INEFF and INEFF-SQ to form the proxies INITxOVERINVEST,
INITxUNDERINVEST, INITxINEFF and INITxINEFF-SQ. Control variables are
defined in Appendix 1.
Note: Continuous variables are winsorized at the 1st and 99th
percentiles, t-statistics are based on clustered standard errors
(Peterson 2009). (*). (**) and (***) represent significance at the
0.10. 0.05 and 0.01 levels, respectively, based on two-tailed tests.

                                             Model 1
                                             All Firms
                    Exp. Sign   Est.         p-value

   Intercept                     9.923       0.000         (***)
   INITIATOR        -           -0.094       0.004         (***)
     PAYER          -           -0.088       0.000         (***)
   OVERINVEST       +           -0.020       0.829
  UNDERINVEST       -           -0.710       0.001         (***)
INITxOVERINVEST     -
INITxUNDERINVEST    -
     INEFF          +
    INEFF-SQ        +
   INITxINEFF       -
 INITxINEFF-SQ      -
     FOROPS                      0.146        0.000        (***)
     XDOPS                       0.169        0.000        (***)
  UNQUALIFIED                   -0.082        0.000        (***)
     ICWEAK                      0.306        0.000        (***)
      ICFR                       0.293        0.000        (***)
 GOING_CONCERN                   0.091        0.038         (**)
  RESTATEMENT                    0.036        0.008        (***)
      BIG5                       0.271        0.000        (***)
   DECEMBERYE                    0.119        0.000        (***)
    INITIAL                      0.024        0.139
      LOSS                       0.123        0.000        (***)
      NYSE                       0.093        0.000        (***)
     INVREC                      0.508        0.000        (***)
      ROA                       -0.308        0.000        (***)
       DA                        0.153        0.000        (***)
      LIQ                       -0.020        0.000        (***)
      LNAT                       0.489        0.000        (***)
     SQSEGS                      0.161        0.000        (***)
     GROWTH                     -0.001        0.036         (**)
      Obs.                               17,520
    R-Square                                  0.8256

                                  Model 2
                                  All Firms
                     Est.         p-value

   Intercept         9.909        0.000      (***)
   INITIATOR         0.054        0.489
     PAYER
   OVERINVEST       -0.016        0.864
  UNDERINVEST       -0.740        0.001      (***)
INITxOVERINVEST     -2.337        0.044      (**)
INITxUNDERINVEST    -1.454        0.140
     INEFF
    INEFF-SQ
   INITxINEFF
 INITxINEFF-SQ
     FOROPS          0.148        0.000      (***)
     XDOPS           0.168        0.000      (***)
  UNQUALIFIED       -0.084        0.000      (***)
     ICWEAK          0.313        0.000      (***)
      ICFR           0.296        0.000      (***)
 GOING_CONCERN       0.092        0.037      (**)
  RESTATEMENT        0.034        0.013      (**)
      BIG5           0.274        0.000      (***)
   DECEMBERYE        0.125        0.000      (***)
    INITIAL          0.026        0.106
      LOSS           0.131        0.000      (***)
      NYSE           0.077        0.001      (***)
     INVREC          0.501        0.000      (***)
      ROA           -0.306        0.000      (***)
       DA            0.161        0.009      (***)
      LIQ           -0.020        0.000      (***)
      LNAT           0.484        0.000      (***)
     SQSEGS          0.159        0.000      (***)
     GROWTH         -0.001        0.041      (**)
      Obs.                   17,520
    R-Square                      0.8249

                                Model 3
                                All Firms
                    Est.        p-value

   Intercept        9.877       0.0001     (***)
   INITIATOR        0.006       0.912
     PAYER
   OVERINVEST
  UNDERINVEST
INITxOVERINVEST
INITxUNDERINVEST
     INEFF          0.273       0.009      (***)
    INEFF-SQ       -0.330       0.154
   INITxINEFF      -0.250       0.571
 INITxINEFF-SQ    -12.687       0.039       (**)
     FOROPS         0.149       0.000      (***)
     XDOPS          0.167       0.000      (***)
  UNQUALIFIED      -0.084       0.000      (***)
     ICWEAK         0.313       0.000      (***)
      ICFR          0.296       0.000      (***)
 GOING_CONCERN      0.091       0.038       (**)
  RESTATEMENT       0.034       0.013       (**)
      BIG5          0.274       0.000      (***)
   DECEMBERYE       0.124       0.000      (***)
    INITIAL         0.026       0.111
      LOSS          0.129       0.000      (***)
      NYSE          0.076       0.002      (***)
     INVREC         0.494       0.000      (***)
      ROA          -0.298       0.000      (***)
       DA           0.158       0.009      (***)
      LIQ          -0.020       0.000      (***)
      LNAT          0.485       0.000      (***)
     SQSEGS         0.160       0.000      (***)
     GROWTH        -0.001       0.032       (**)
      Obs.                 17,520
    R-Square                    0.8247

TABLE 4
Auditor Pricing of Dividend Initiation and Investment Inefficiency
Compared  to Non-paying Firms
Table 4 provides an analysis of dividend initiation for over- and
under-invested firms compared to non-dividend paying firms. Models 2
and 3 compare to only over- and only under-invested firms,
respectively, following Biddle, Hilary and Verdi (2009). The sample
covers the time period 2003 to 2012. The dependent variable, LAF, is
the natural logarithm of the audit fees in $ thousands. INITIATOR is
equal to 1 if the firm initiated dividends in the current fiscal year.
Our proxy for over-investment is OVERINVEST, and our proxy for
under-investment is UNDERINVEST. Initiator is interacted with over- and
under-invest to form the proxies, INITxOVERINVEST and INITxUNDERINVEST.
Variables are defined in Appendix 1, and estimations on control
variables are excluded from the table for brevity. We include year and
industry dummies in all model analysis.
Note: Continuous variables are winsorized at the 1st and 99th
percentiles, t-statistics are based on clustered standard errors
(Peterson 2009). (*), (**) and (***) represent significance at the
0.10, 0.05 and 0.01 levels, respectively, based on two-tailed tests.

                             Model 1
                             All Non-paying Firms

                  Exp. Sign  Est.           p-value

   Intercept                 10.442         0.000    (***)
   INITIATOR      -           0.066         0.414
   OVERINVEST     +           0.017         0.854
INITxOVERINVEST   -          -2.575         0.03      (**)
  UNDERINVEST     -          -0.66          0.006    (***)
INITxUNDERINVEST  -          -1.925         0.054      (*)
[SIGMA]Controls                             Yes
  [SIGMA]Year                               Yes
[SIGMA]lndustry                             Yes
      OBS                              12,324
     ADJRSQ                                 0.7938

                   Model 2
                   Over-innvested
                   Non-paying Firms
                   Est.        p-value

   Intercept       10.407      0.000      (***)
   INITIATOR        0.311      0.022       (**)
   OVERINVEST       0.158      0.086        (*)
INITxOVERINVEST    -5.056      0.003      (***)
  UNDERINVEST
INITxUNDERINVEST
[SIGMA]Controls                Yes
  [SIGMA]Year                  Yes
[SIGMA]lndustry                Yes
      OBS                  5,303
     ADJRSQ                    0.7803

                  Model 3
                  Under-invested
                  Non paying Firms
                  Est.         p-value

   Intercept      10.131       0.000    (***)
   INITIATOR      -0.061       0.520
   OVERINVEST
INITxOVERINVEST
  UNDERINVEST     -0.941       0.003    (***)
INITxUNDERINVEST  -0.114       0.920
[SIGMA]Controls                Yes
  [SIGMA]Year                  Yes
[SIGMA]lndustry                Yes
      OBS                  7,021
     ADJRSQ                    0.8127

TABLE 5
The Propensity to Misallocate under High Growth (Control Sample Is
Non-payers)
Table 5 provides an analysis of dividend initiaiton for over- and
under-invesled firms compared to non-dividend paying firms, but in
these models the sample is divided by firms above and below the
Compustat universe median of sales growth. Observations span the period
from 2003 to 2012. The dependent variable, LAF, is the natural
logarithm of the audit fees in $ thousands. INITIATOR is equal to 1 if
the firm initiated dividends in the current fiscal year. Our proxy for
over-investment is OVERINVEST, and our proxy for under-investment is
UNDERINVEST. Initiator is interacted with over- and under-invest to
form the proxies INITxOVERINVEST and INITxUNDERINVEST. Variables are
defined in Appendix 1, and estimations on control variables are
excluded from the table for brevity. We include year and industry
dummies in all model analysis. Note: Continuous variables are
winsorized at the 1st and 99th percentiles. t-statistics are based
on clustered standard errors (Petersen 2009). (*), (**) and
(***) represent significance at the 0.10. 0.05 and 0.01 levels,
respectively, based on two-tailed tests.

                             Only Low Growth Firms
                  Exp. Sign

   Intercept                 10.410          0.000   (***)
   INITIATOR      -          -0.088          0.428
   OVERINVEST     +          -0.025          0.832
INITxOVERINVEST   -          -0.488          0.768
  UNDERINVEST     -          -1.000          0.001   (***)
INITxUNDERINVEST  -          -0.267          0.842
[SIGMA]Controls                              Yes
[SIGMA]Industry                              Yes
  [SIGMA]Year                                Yes
      Obs.                               5,822
    R-Square                                 0.8063

                  Only High Growth Firms

   Intercept      10.460      0.000  (***)
   INITIATOR       0.184      0.113
   OVERINVEST      0.047      0.663
INITxOVERINVEST   -4.189      0.008  (***)
  UNDERINVEST     -0.617      0.034  (**)
INITxUNDERINVEST  -3.348      0.034  (**)
[SIGMA]Controls               Yes
[SIGMA]Industry               Yes
  [SIGMA]Year                 Yes
      Obs.                6,502
    R-Square                  0.786

TABLE 6
Compared to Dividend Payers
Table 6 provides an analysis of dividend initiation for over- and
under-invested firms compared to firms that already pay dividends. The
sample period is 2003 through 2012. The dependent variable LA, is the
natural logarithm of the audit fees in $ thousands. Our proxy for
over-investment is OVERINVEST, and our proxy for under-investment is
UNDERINVEST. Variables are defined in Appendix 1, and estimations on
control variables are excluded from the table for brevity. We include
year and industry dummies in all model analysis. Note: Continuous
variables are winsorized at the 1st and 99th percentiles. t-statistics
are based on clustered standard errors (Petersen 2009). (*), (**) and
(***) represent significance at the 0.10, 0.05 and 0.01 levels,
respectively, based on two-tailed tests.

                       Model 1
                       All Payers

    Intercept           9.326         0.000    (***)
    INITIATOR      -    0.138         0.105
   OVERINVEST      +   -0.640         0.093    (*)
 INITxOVERINVEST   -   -1.134         0.372
   UNDERINVEST     -   -0.563         0.172
INITxUNDERINVEST   -   -1.274         0.238
 [SIGMA]Controls                      Yes
 [SIGMA]Industry                      Yes
   [SIGMA]Year                        Yes
      Obs.                        5,478
    R-Square                          0.8579

                   Model 2
                   Over-invested Payers

    Intercept       9.452                0.000     (***)
    INITIATOR       0.351                0.011     (**)
   OVERINVEST      -0.492                0.227
 INITxOVERINVEST   -4.036                0.039     (**)
   UNDERINVEST
INITxUNDERINVEST
 [SIGMA]Controls                         Yes
 [SIGMA]Industry                         Yes
   [SIGMA]Year                           Yes
      Obs.                               1,043
    R-Square                             0.8717

                   Model 3
                   High Growth Payers

    Intercept       9.278              0.000   (***)
    INITIATOR       0.247              0.044   (**)
   OVERINVEST      -0.724              0.103
 INITxOVERINVEST   -3.036              0.067   (*)
   UNDERINVEST     -0.876              0.065   (*)
INITxUNDERINVEST   -2.704              0.111
 [SIGMA]Controls                       Yes
 [SIGMA]Industry                       Yes
   [SIGMA]Year                         Yes
      Obs.                         2,391
    R-Square                           0.8541
COPYRIGHT 2018 Creighton University
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2018 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Alsharairi, Malek; Smith, Deborah Drummond; Glambosky, Mina; Gleason, Kimberley
Publication:Quarterly Journal of Finance and Accounting
Article Type:Report
Geographic Code:9CHIN
Date:Jun 22, 2018
Words:11628
Previous Article:Asset Allocation: A Recommendation for Resolving the Collision between Theory and Practice.
Next Article:Product Market Competition and Credit Ratings.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters