Financial Constraints, Auditing, and External Financing

Abstract

This paper examines the role of costly audit signals in future external financing activity of financially constrained firms. We document that when facing higher financial constraints, firms pay higher audit fees and have their audit reports completed sooner. Equally important, we find that costlier and timelier audits are associated with a greater amount of future financing raised by equity-seeking, but not debt-seeking, constrained firms. Our results are robust to controlling for various audit characteristics and risk factors. Additional analyses show that equity-seeking constrained firms that underwent costlier audits exhibit more favorable outcomes with respect to long-run stock performance and investment efficiency following seasoned equity offerings. Our findings suggest that while financially constrained firms are pressured to make cuts across various expenditure categories, negotiating lower audit fees in the face of higher financial constraints may not be a wise strategy.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental data

Supplemental data for this article can be accessed at https://doi.org/10.1080/09638180.2022.2159471.

Notes

1 We borrow the financial constraints definition of Lamont et al. (2001, p. 529), who write: “By ‘financial constraints’, we mean frictions that prevent the firm from funding all desired investments. This inability to fund investment might be due to credit constraints or inability to borrow, inability to issue equity, dependence on bank loans, or illiquidity of assets. We do not use ‘financial constraints’ to mean financial distress, economic distress, or bankruptcy risk, although these things are undoubtedly correlated with financial constraints.” In our empirical analyses, we control for bankruptcy risk and financial risk.

2 There is evidence that higher audit fees are an indication of greater commitment to financial statement reliability on the part of both firms and auditors (e.g., Carcello et al., 2002; Srinidhi & Gul, 2007; Dhaliwal et al., 2008; Blankley et al., 2012; Asthana & Boone, 2012), whereas longer audit lags signal potential issues regarding financial statement reliability such as future restatements (e.g., Blankley et al., 2014, 2015). We follow these studies in choosing audit fees and audit lag as relevant information signals for investors.

3 We acknowledge that compared with financially unconstrained firms, financially constrained firms face greater operational and financial uncertainty, which typically affects audit outcomes. Hence, we control for multiple measures of risk in our main and robustness tests to strengthen our inferences. We do not claim that financially unconstrained firms are less risky. We also note that no study, including this one, can fully control for audit or litigation risk in our setting. This represents a limitation of our study.

4 The last fiscal year covered in Hoberg and Maksimovic’s financial constraints database is 2015.

5 Hoberg and Maksimovic (2015) report that their measure is reasonably correlated with other commonly used measures of financial constraints but it is distinct enough (e.g., ρ(FCS, firm age) = -0.156). They write (p. 1326): “our variables also contain a significant amount of information that goes beyond the conclusions of [Hadlock and Pierce] as many smaller and younger firms appear to be unconstrained and some medium sized firms appear to be constrained.” They also note: “our financial constraint variables are somewhat persistent, but not overwhelmingly so […] a firm that is constrained in a given year is quite likely to be constrained the next year, but it is unlikely to be still constrained two to three years later.”

6 If higher audit fees paid by financially constrained firms were simply an indication of higher audit risk, one would expect to see a negative or no moderation effect in this case.

7 Financially constrained versus unconstrained firms may be more likely to switch to Big 4 auditors from non-Big 4 auditors in hopes of signaling better audit quality. There is some evidence of this behavior. The results show that in the face of higher financial constraints, firms tend to switch to Big 4 auditors (β = 0.0006, t = 1.86, p = 0.064). We control for the Big 4 dummy and auditor change dummy in all the models presented in the paper.

8 The results are similar when we use continuous variables instead of decile-ranked variables.

9 Prior to December 15, 2016, the annual report filing deadline for large accelerated filers was 75 days. Also, prior to December 15, 2003, accelerated filers were under a 90-day filing deadline.

10 Table OA3 reports the sample averages separately for firms with above and below median FCS scores.

11 To examine whether our sample is representative, we compared the size characteristics of our sample with those of all industrial U.S. public firms covered by Compustat. For the same sample period, on average, our sample has slightly larger total assets (2.36 billion vs. 2.24 billion; p < 0.05), comparable market value ($2.58 billion vs. $2.44 billion; p = 0.238) and higher firm age (19.54 vs. 18.01; p < 0.01).

12 In the audit fee regression model, we also control for audit lag (and vice versa).

13 As an additional test, we replaced FCS with ESCS and DSCS. The results show that ESCS interacts positively with Total Q to predict audit fees (β = 0.0027, p < 0.05). That is, equity-seeking constrained firms with more investment opportunities are likely to value the costly signal of high audit assurance more than those with fewer investment opportunities (b = 0.0049, t = 2.34, p < 0.05). In contrast, there is no significant interaction between DSCS and Total Q (β = 0.0003, p = 0.819), though the main effect for DSCS is positive (β = 0.0022, p < 0.05).

14 To facilitate the interpretation of estimated coefficients, we standardized Res_ln(Audit fees) before creating the interaction term by subtracting its mean and dividing by its standard deviation.

15 We also tested whether the documented negative interaction between financial constraints and residual audit fees in predicting audit lag is magnified by growth opportunities. There is a negative three-way interaction among FCS, Res_ln(Audit fees), and Total Q (β = -0.0912, p < 0.10), suggesting that the inverse relation between audit lag and the interaction of financial constraints with residual audit fees is stronger for firms that have more growth opportunities. Specifically, the simple slope for FCS is – 0.4276 (t = -3.60, p < 0.01) when both Res_ln(Audit fees) and Total Q are one standard deviation above their respective means. 

16 In an additional analysis, we find that ESCS interacts negatively with Res_ln(Audit fees) to predict audit lag (β =-0.1708, p < 0.05). That is, when firms pay higher, but not lower, audit fees, audit lag is negatively associated with equity-seeking constraints (b = -0.2751, t = -2.50, p < 0.05). There is no significant results for DSCS.

17 As a sensitivity test, we estimated a changes model in which we replaced all the continuous variables in the audit fee and audit lag models with their first differences and removed the fixed effects. The variable of interest is FCS change dummy, which equals 1 for firm years whose annual change in FCS is in the upper quartile of the sample and 0 otherwise. Consistent with our main results, Δln(Audit fees) is positively associated with change in FCS (β = 0.0153, t = 2.85, p < 0.01), whereas ΔAudit lag is negatively associated with it (β = -0.6853, t = -2.20, p < 0.05).

18 For additional analysis, we replaced ln(Audit fees) with Res_ln(Audit fees) as the dependent variable in our test of the audit fee hypothesis, removed the control variables, and reran our test. Our conclusions remained unchanged. The estimated coefficient on FCS is again positive and significant (β = 0.0092, t = 2.72, p < 0.01). We also reran our audit lag test after removing the control variables and obtained consistent results (β = -0.2517, t = -3.14, p < 0.01).

19 We also tested the FCS x Early audit completion dummy interaction instead of the FCS x Audit lag interaction. As expected, the results reveal a positive coefficient on the interaction term (β = 0.0071, t = 1.99, p < 0.05).

20 If high audit fees were simply an indication of high audit risk, one would have expected to see an opposite (or no) interaction effect. To shed some light on the risk issue, as an additional analysis, we tested whether equity-seeking constrained firms that paid higher audit fees are associated with greater future litigation risk. We found no such evidence. In fact, ESCS interacts negatively with Res_ln(Audit fees) to predict future litigation likelihood (β = -0.0027, p < 0.01; Table OA4), suggesting that the likelihood of litigation for equity-seeking constrained firms is lower when audit fees are higher. Specifically, there is a negative relation between ESCS and the likelihood of litigation when residual audit fees are one standard deviation above the mean (b = -0.0031, t = -1.96, p < 0.05).

21 Big 4 auditors charge higher fees than small auditors and are typically referred to as “high quality” auditors in the prior literature (DeFond & Zhang, 2014). To ensure that our costly audit signal argument is not confounded by a possible Big 4 effect, we added the interaction between ESCS and Big 4 dummy to the equity financing model. The estimated coefficient on the interaction term was insignificant (β = -0.0035, t = -0.79, p = 0.432), whereas the interactions of ESCS with audit fees and audit lag remained significant at the 1% level.

22 SA Index = -0.737 * Size + 0.043 * Size² – 0.040 * Age, where size is the natural logarithm of inflation-adjusted total assets (capped at $4.5 billion) and age is the firm age in Compustat (capped at 37 years). The SA index is positively correlated with both ESCS (ρ = 0.20, p < 0.01) and FCS (ρ = 0.14, p < 0.01).

23 WW Index = -0.091 * Cash Flow / Assets – 0.062 * Dividend Dummy + 0.021 * Long-Term Debt / Assets – 0.044 * Log of Assets + 0.102 * Industry Sales Growth – 0.035 * Sales Growth. The WW index is positively correlated with both ESCS (ρ = 0.11, p < 0.01) and FCS (ρ = 0.05, p < 0.01).

24 In untabulated results, we find that in the face of higher financial constraints, firms produce more reliable financial statements with lower restatement likelihood and fraud risk. In terms of economic significance, the restatement likelihood is lower by 3.6 percentage points for firms in the top decile of FCS than those in the bottom decile of FCS (i.e., a 21.6 percent decrease in the average restatement likelihood). Similarly, moving from the bottom to the top decile of FCS is associated with a 3.0 percentage points decline in fraud risk (i.e., a 7.0 percent decline in the likelihood of fraud, on average).

25 If a firm issues multiple SEOs in a fiscal year, we keep the earliest SEO issued in the fiscal year.

26 Calendar-time portfolio regressions are commonly used in prior research as this method yields well-specified test statistics (e.g., Fama, 1998; Brav et al., 2000).

27 Also, there is a significant difference in alphas between the two portfolios (diff = 0.873%, t = 2.61, p < 0.01), suggesting that the latter group outperforms the former group during the post-SEO period.

28 Our interpretation is consistent with prior research, which differentiates between priced and unpriced risk (e.g., Core et al., 2008; Mohanram & Rajgopal, 2009). After controlling for priced risk factors (i.e., market, size, value, profitability, investment, and momentum), the calendar time portfolio regression model alphas capture abnormal returns due to other factors.