1 | Page 1. Introduction This paper examines empirically the effect of auditor-provided non-audit tax services (NATS) on stock price crash risk, and the mechanisms through which the relationship, if any, manifests itself. Also examined, is whether tax service-induced crash risk varies according to the particular business strategy pursued by firms. The study is motivated by a long-standing debate in the history of auditing as to whether external auditors should provide non-audit services (NASs) to clients, and the recent regulatory ban on auditors’ providing a variety of NASs excluding the provision of NATS. There are competing arguments regarding the implications of NASs for audit quality. Opponents of NASs argue that auditor independence and, consequently, audit quality is impaired when auditors also provide NASs. This idea was formally developed by DeAngelo (1981) who relates auditor independence with client-specific future quasi–rents, defined as “...the excess of revenues over avoidable costs, including the opportunity cost of auditing the next-best alternative client” (italics in original) (p.116). She develops a two-dimensional definition of audit quality that sets the standard for addressing the auditor independence issue. An auditor will be considered independent if he/she can detect a material misstatement and then report it to the audit service consumers. If no client-specific quasi-rents are expected from a given client relationship, then the auditor should have no economic incentive to conceal financial misstatements and, therefore, should be considered perfectly independent from the client. However, if the client-specific future quasi-rents are high then auditors may compromise independence in order to retain them. It has been argued that NASs provide more quasi-rents than do audit services because of the higher margin derived from the former and, therefore, NASs are more likely to impair auditor independence [SEC, 2000], particularly independence in
44
Embed
Auditor-provided Tax Services and Stock Price Crash Risk · PDF filestock price crash risk, ... history of auditing as to whether external auditors should ... provided NATS are to
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1 | P a g e
1. Introduction
This paper examines empirically the effect of auditor-provided non-audit tax services (NATS) on
stock price crash risk, and the mechanisms through which the relationship, if any, manifests
itself. Also examined, is whether tax service-induced crash risk varies according to the particular
business strategy pursued by firms. The study is motivated by a long-standing debate in the
history of auditing as to whether external auditors should provide non-audit services (NASs) to
clients, and the recent regulatory ban on auditors’ providing a variety of NASs excluding the
provision of NATS.
There are competing arguments regarding the implications of NASs for audit quality.
Opponents of NASs argue that auditor independence and, consequently, audit quality is impaired
when auditors also provide NASs. This idea was formally developed by DeAngelo (1981) who
relates auditor independence with client-specific future quasi–rents, defined as “...the excess of
revenues over avoidable costs, including the opportunity cost of auditing the next-best
alternative client” (italics in original) (p.116). She develops a two-dimensional definition of
audit quality that sets the standard for addressing the auditor independence issue. An auditor will
be considered independent if he/she can detect a material misstatement and then report it to the
audit service consumers. If no client-specific quasi-rents are expected from a given client
relationship, then the auditor should have no economic incentive to conceal financial
misstatements and, therefore, should be considered perfectly independent from the client.
However, if the client-specific future quasi-rents are high then auditors may compromise
independence in order to retain them. It has been argued that NASs provide more quasi-rents
than do audit services because of the higher margin derived from the former and, therefore,
NASs are more likely to impair auditor independence [SEC, 2000], particularly independence in
2 | P a g e
appearance (Lindberg and Beck 2004). Proponents of auditor-provided NASs, however, argue
that, NASs provide considerable economies of scope. These economies of scope are broadly
categorized into knowledge spillover benefits (benefits from transferring information and
knowledge), and contractual economies (making better use of assets and/or safeguards already
developed when contracting and ensuring quality in auditing) (Simunic 1984; Beck et al. 1988;
Arrunada 1999).
The regulatory response to NASs, at least in the USA, favors the ‘impairment of auditor
independence’ argument. This response can be attributed to an independence concerns about big
audit firms arising from the provision of NASs to existing clients which, among other factors,
has been identified as one of the reasons for the massive corporate collapses experienced by the
US economy at the beginning of this century. Congress ratified the Sarbanes-Oxley Act (SOX)
in 2002. Section 201 of this act prohibits auditors from providing a variety of NASs for their
audit clients. However, the SEC determined that it would not prohibit auditor-provided NATS: a
specific type of NAS that auditors frequently provide to their audit clients. However significant
restrictions have been imposed on the provision of such services. In particular, (a) all auditor-
provided NATS are to be specifically pre-approved by the client’s audit committee; (b) fees paid
to auditors for NATS are to be identified separately in the company’s annual proxy statements;
(c) the auditor cannot audit his own work, function as a part of management, or serve in an
advocacy role for the audit client; and (d) SOX violations are subject to expanded criminal and
civil liabilities and penalties.
To address regulatory concerns, recent empirical studies have examined the relation
between auditor-provided NATS and earnings management, and find support for knowledge
spillover benefits from auditor-provided NATS outweighing an independence impairment
3 | P a g e
concerns. For example, auditor-provided NATS reduce the incidence of financial restatements
(Kinney et al. 2004) as well as tax-related restatements (Seetharaman et al. 2011). Paterson and
Valencia (2011) confirm the Kinney et al.’s (2004) result, but only for recurring NATS. Non-
recurring tax engagements, on the other hand, increase the probability of future restatements (a
threat to auditor independence). Also consistent with the knowledge spillover argument, Omer et
al. (2006) and Krishnan and Visvanathan (2011) find a negative association between auditor-
provided NATS and discretionary accruals. Gleason and Mills (2011) find that companies
purchasing tax services from their auditors do not manage tax reserves to meet/beat analyst
forecasts any more than other companies.
These studies, however, did not examine a more extreme form of adverse outcome
having direct economic consequences for investors: stock price crash risk. Stock price crash risk
at the firm level refers to the likelihood of observing extreme negative values in the distribution
of firm-specific returns after adjusting for the return portion that co-move with common factors
(Jin and Myers 2006; Kim et al. 2011a; 2011b). Conceptually, stock price crash risk is premised
on the notion that managers have a tendency to withhold bad news for an extended period of
time, allowing bad news to stockpile. When the accumulation of bad news passes a threshold, it
is revealed to the market at once, leading to a large negative drop in price for the stock (Jin and
Myers, 2006). Crash risk is an important characteristic of the return distribution that is relevant
to portfolio theories, asset-pricing, and option-pricing models. Sunder (2010) argues that crash
risk is undiversifiable, unlike the risk from symmetric volatilities. Harvey and Siddique (2000)
suggest that conditional skewness (one proxy for crash risk) is a priced factor.1 Because investors
1 Crash risk, a third moment of stock returns capturing negative skewness, is distinct from other measures studied in prior research, such as the average return (first moments), and the variance of stock returns (the
second moments). Several studies (e.g. Gabaix 2012; Kelly and Jiang 2014; Huang et al. 2012) have both
hypothesized, and empirically documented, that crash risk increases the required return of investors. Conrad et
4 | P a g e
are averse to stocks with greater crash risk due to the undiversifiable nature of those stocks,
determining whether auditor-provided NATS mitigate (exacerbate) crash risk by deterring
(aiding) managerial bad news hoarding behavior, would help investors better allocate their
resources.
In a recent empirical study, Kim et al. (2011a) document a positive association between
corporate tax avoidance and future crash risk. They argue that “Tax avoidance activities can
create opportunities for managers to pursue activities that are designed to hide bad news and
mislead investors…Perhaps more importantly, managers are able to justify the opacity of tax
avoidance transactions by claiming that complexity and obfuscation are necessary to minimize
the risk of tax avoidance arrangements being detected by the [taxing authorities]… To some
extent, these avoidance activities are shielded from the investigations of audit committees and
external auditors” (p. 640, italics added). Whereas Kim et al. (2011a) consider external auditor
monitoring of client tax avoidance to be lacking, we argue that the strength of the relationship
between tax avoidance and crash risk depends on auditor-provided NATS.
However, competing arguments exist as to whether auditor-provided NATS impair
auditor independence or generate spillover financial reporting quality benefits: i.e. insights
acquired from providing tax services can enhance audit effectiveness and client financial
reporting quality. In the former case external auditors could assist their clients by devising
complex tax avoidance strategies, thereby increasing the probability of a future crash. In the
latter case, however, the knowledge spillover created by auditor-provided NATS can enhance
overall audit effectiveness via better communication between the audit and tax sides, thus
mitigating auditor information asymmetry (Krishnan et al. 2013). Considered from this
al. (2013) find that more negatively skewed returns generate subsequent higher returns: a compensation for
greater crash risk. From an asset pricing perspective, Chang et al. (2013) find that negative skewness in market
returns is associated with a positive risk premium.
5 | P a g e
perspective, future crash risk is likely to be lower for firms with incumbent auditors providing
NATS. We test these competing arguments by including two channels, namely, tax expense-
based earnings management and tax avoidance, through which the association between auditor-
provided NATS and crash risk could be more meaningfully tested.2
We use a two-stage model to control for potential endogeneity associated with the
selection of the auditor for tax services. Our first-stage model includes determinants of auditor-
provided NATS identified in prior research (McGuire et al. 2012). We then examine the impact
of auditor-provided NATS on future crash risk and inquire whether tax expense earnings
management and tax avoidance moderate the relationship. We use earnings management in tax
expenses as opposed to the commonly used discretionary accruals as our earnings management
proxy. Using negative conditional skewness (NCSKEW) and down-up volatility (DUVOL) as our
crash risk measures, we find that auditor-provided NATS mitigates future crash risk by reducing
earnings management in tax expenses, and deterring tax avoidance behavior. We, therefore,
provide empirical support for the knowledge spillover benefits emanating from the provision of
NATS. We also find that auditor-provided NATS constrain tax avoidance and, hence, reduce the
probability of a crash for firms following innovator business strategies, a finding that has not
been documented before. We consider the strategy-tax avoidance link, as Higgins et al. (2014)
show that innovators engage more in tax avoidance than their defender counterparts.
2 Zang et al. (2013) find that auditors are more likely to resign from firms with aggressive ax strategies (high
tax avoidance). Relevant to our study, they also document a weaker effect of tax aggressiveness on auditor
resignation if the incumbent auditors provide tax services. They interpret this finding as NATS-induced
impairment of auditor independence. However, an alternative interpretation may suggest that knowledge
spillover benefits from tax services allow incumbent auditors to be more efficient in auditing financial and tax
accounting, reducing the chance of reporting irregularities which, in turn reduces auditor exposure to costly
litigation. This line of reasoning supports the contention that tax services provision lowers the chance of
auditor resignation.
6 | P a g e
We contribute to the literature in a number of ways. First, we provide evidence on the
desirability of allowing, further constraining, or banning auditor-provided NATS (PCAOB
2004). We do that by evaluating the extent to which provision of tax services increases or
decreases crash risk, a more acute form of outcomes having direct economic consequences for
investors. Unlike other studies that directly examine the effect of auditor-provided NATS on
earnings quality (Krishnan and Visvanathn 2011), we consider the joint effect of NATS and
earnings quality on crash risk. Second, we extend the growing literature on the determinants of
crash risk by investigating the effect of external auditors: an essential governance mechanism; on
crash risk. Since external auditors are directly responsible for verifying the authenticity of the
tax-related transactions reported in financial statements, it is logical to assume that a dominant
role is played by auditors. Robin and Zhang (forthcoming) find that industry specialist auditors
reduce future crash risk. They argue that industry specialist auditors have the expertise required
to reduce the managerial propensity to hoard bad news. Our study differs from Robin and Zhang
because of our focus on auditor-provided NATS and the role of tax expense-related earnings
management in moderating the relationship between NATS and future crash risk. Finally, we
contribute to the interface of strategy and crash risk research. An understanding of the
relationship between business strategy and crash risk, and of auditor-provided NATS moderating
such relationship, would assist investors in allocating resources carefully among companies with
differing business strategies.
The remainder of the paper proceeds as follows. Section 2 reviews the related literature
and develops testable hypotheses. Section 3 explains research design issues. The following
section provides our sample selection procedure and descriptive statistics. The main test results
are reported in section 5. The final section concludes the paper.
7 | P a g e
2. Review of related literature and development of hypotheses
Auditor-provided NASs have been, and still remain, a matter of serious regulatory and investor
concern. While all fees create economic bonds between the auditor and client, debate has
typically alleged that the provision of NASs provide incentives for audit firms to accept clients’
questionable accounting choices, thus reducing auditor independence and ultimately the quality
of financial reporting. The provision of NASs has been the focus of most recent attention,
presumably reflecting the widely accepted view that NASs are typically provided at a higher
profit margin than audit services (Ruddock and Taylor 2005). Such concerns prompted the SEC
to prohibit most types of NASs in Section 201 of the SOX-2002 act. However, SEC allowed
auditor-provided NATS subject to the pre-approval of audit committees.3
Auditors evaluate the validity of accrued taxes payable and tax contingent liabilities on
the balance sheet, income tax expenses on the income statement, and the related note disclosures,
in order to provide adequate assurance to the investing public about the appropriateness of these
items and disclosures (Barrett 2004). Although many accruals facilitate earnings management,
managers can use valuation allowances (Frank and Rego 2006), tax contingency reserves (Gupta
et al. 2011), estimates of accrued taxes (Dhaliwal et al. 2004), and the designation of
permanently reinvested earnings (Krull 2004) to achieve earnings targets. Because any material
information about risky tax transactions tends to be hidden in these accounts and disclosures,
auditors also have to assess whether their clients engage in potentially abusive tax transactions.
Whether auditor-provided NATS help auditors to discover these risky tax transactions
(knowledge spillover), or impair independence by aiding clients in developing abusive taxation
3 For a representative list of NATS that were routinely provided by auditing firms to their audit clients but are either banned by SOX, no longer offered by some auditing firms to their clients, or are unlikely to be pre-
approved by audit committees, see Appendix B in Seetharaman et al. (2013).
8 | P a g e
strategies (impairment of independence), has become an issue of immense importance for
regulators and investor community alike.
In a study of restatements involving GAAP violations, Kinney et al. (2004) find a
negative association between restatements of financial statements and tax services, but
Seetharaman et al. (2009) fail to find any such evidence. However, they do find a significant
negative association between auditor-provided tax services and tax-related restatements, which
would be consistent with knowledge spillover. Further evidence in support of the knowledge-
spillover benefits from NATS has been documented for the following: the propensity for auditors
to issue going-concern opinions (Robinson 2008); the maintenance of adequate tax reserves by
firms procuring tax services from incumbent auditors (Gleason and Mills 2010); lower debt
pricing costs for clients procuring tax services from the same auditor (Fortin and Pitman 2008);
and lower earnings management in tax expenses for firms having auditors providing NATS
(Lisic 2014). Krishnan and Visvanathn (2011) also find support for auditor-provided tax
services, constraining earnings management, but their proxy for earnings management is loss
avoidance, a proxy that which fails to capture the tax-related earnings management practices. In
a follow-up study, Krishnan et al. (2013) find that earnings are more value-relevant for firms
where incumbent auditors provide tax services. However, their study does not investigate the
possible channels that strengthen or weaken this relationship (e.g., information quality, external
monitoring and so on).
Evidence in support of a threat to auditor independence from auditor-provided NATS has
been documented by Cook et al. (2008). They find that incumbent auditors who also provide tax
services help clients to reduce 4th quarter effective tax rates (ETR), thereby increasing earnings
9 | P a g e
management. Harris and Zhou (2013) find that auditor-provided NATS reduce non-tax internal
control weaknesses, but not tax-related internal control weaknesses.
Since the decision to use an incumbent auditor for tax services is endogenously
determined, some studies examine the rationale for procuring tax services from incumbent
auditors. Albring et al. (2014) find that corporate governance attributes, such as board
independence, the audit committee’s accounting expertise, and separation of the CEO and
chairman of the board, play a role in a firm’s decision to switch to a non-auditor provider for tax
services. Lassila et al. (2010) provide evidence that firms with strong corporate governance and
relatively high levels of tax and operational complexity are more likely to retain their auditor for
tax services.
In a recent empirical work Kim et al. (2011a) find convincing evidence that tax
avoidance increases future stock price crash risk supporting the contention that aggressive tax
strategies and planning provide managers with a means to conceal negative information that
increases crash risk. Kim et al. (2011a), however, did not investigate whether auditor-provided
NATS aid or deter tax-related earnings management as well as tax avoidance behavior.
Therefore external auditors’ involvement in the tax avoidance-crash risk link remains
unexplored. Instead Kim et al. (2011a) considered strength of external monitoring, proxied by
analyst coverage and institutional shareholdings, as potential mechanisms for reducing crash
risk. We fill this void by arguing that incumbent auditors providing tax services are better able to
differentiate between value-enhancing versus value-destroying tax planning strategies. Firms that
engage in tax aggressiveness have a higher chance of issuing misstatements and restatements
because managers can use various accounts, such as valuation allowances, tax contingency
reserves, and estimates of accrued taxes to manipulate earnings (e.g., Hanlon and Heitzman
10 | P a g e
2010; Frank and Rego 2006; Gupta et al. 2011; Dhaliwal et al. 2004). Aggressive tax positions
involve complex and risky techniques, which require additional research, specialized audit
procedures, documentation, and consultations with tax professionals to audit (Donohoe and
Knechel 2013).4
Audit firms providing tax services to their clients package their reputation or technical
expertise in the provision of such services and, simultaneously, optimize their unique ability as
auditors to leverage these benefits into higher financial reporting quality (Seetharaman et al.,
2013). Insights gained from performing tax services enable auditors to be more familiar with
clients’ strategic decisions regarding tax planning: a feature that benefits the auditors in
uncovering tax expense-related earnings management and tax avoidance policies. That
familiarity also facilitates management’s consideration of the financial reporting implications of
various alternatives, and results in more certainty in the financial reporting of the company’s tax
positions, while constraining opportunities for management to use complex transactions for
obfuscating financial reporting opaqueness. Since financial reporting opaqueness has been
identified to be an important determinant crash risk (Jin and Myers 2006; Hutton et al. 2009), the
above argument on knowledge spillover suggests a negative relationship between auditor-
provided NATS and crash risk.
A competing argument exists, however. Auditor-provided NATS could threaten auditor
independence by increasing auditors’ economic dependence on their clients. This has become
4 For example, complex tax shelters created by Enron allowed managers to manipulate earnings while
preventing investors from understanding the sources (Desai and Dharmapala 2009). The tax avoidance
activities arranged by Tyco facilitated the centralization of power by management, and enabled them to
obscure their rent diversion through means such as unauthorized compensation, abuse of corporate funds for
personal uses, and insider trading (Desai 2005). In line with this view, studies document that aggressive tax reporting leads to lower earnings quality, a higher likelihood of managerial fraud, and higher stock price crash
risk (e.g., Badertscher et al. 2009; Ettredge et al. 2008; Frank et al. 2009; Hanlon 2005; Kim et al. 2011a; Zang
et al. 2013).
11 | P a g e
more of a concern after SEC’s simultaneous prohibition on other NASs. Seetharaman et al.
(2013) conjecture that this may have the unintended consequence of shifting the source of
independence problems from other categories of NASs to NATS. Such lack of independence will
reduce the incentives for incumbent auditors to detect and report tax-related earnings
manipulation, with the resulting consequence of a greater probability of future crash. Our first
hypothesis on the association between NATS and future crash risk is as follows:
H1: Auditor-provided NATS are associated with tax-related earnings management and client tax
avoidance with an implication for future stock price crash.
Our second hypothesis is built on the premise that there is an association between firm-
level business strategies, corporate tax avoidance, and subsequent crash risk. We argue that these
firm-level determinants of crash risk are determined, to a large extent, by the unique business
strategies pursued by individual firms. For example, firms following aggressive business
strategies may have to engage more in earnings manipulation in order to sustain overly optimistic
expectations by investors about the future performance of these firms (Higgins et al., 2014;
Skinner and Sloan 2002; Lakonishok et al. 1994).
Miles and Snow (1978, 2003) detail three viable business strategies that may exist
simultaneously within industries—Prospectors, Defenders, and Analyzers— because of
differences in the magnitude and direction of change regarding their particular products and
markets (Hambrick 1983). These strategies are positioned along a continuum, with prospectors at
one end and defenders at the other. Prospectors change their product market mix rapidly in order
to be innovative market leaders, while defenders focus more on a narrow and stable product base
in order to compete on the basis of price, service, or quality. Firms that constitute the middle of
12 | P a g e
the continuum are termed analyzers, and possess the attributes of both prospectors and defenders
(Miles and Snow 1978, 2003). Prior research on organization theory has demonstrated that
prospectors have a higher level of outcome uncertainty, are plagued with more information
asymmetry, and structure executive compensation that is primarily long-term and incentive-
based (Rajagopalan 1997; Balsam et al. 2011; Kothari et al. 2009). Higgins et al. (2014)
document an increasing propensity for innovators to engage in more tax avoidance compared to
their defender counterparts. Their finding is explained by the fact that firm strategies are, in part,
based on firms’ willingness to deal with risk and uncertainty, with prospectors being subject to
more uncertainty and, hence, requiring more tax planning. Habib and Hasan (2014) provide
empirical evidence that prospectors are more prone to crash. These two streams of literature,
therefore, suggest that the tax avoidance propensities of prospectors may contribute to future
crash. Auditor-provided NATS may strengthen (weaken) this relationship depending on whether
the independence impairment or the knowledge spillover argument dominates. Our second
hypothesis is as follows:
H2: There is an association between auditor-provided NATS and future crash for firms
pursuing differing business strategies.
3. Research design and sample selection
3.1 Stock price crash risk
In this study, two measures of firm-specific crash risk are used, as in Chen et al. (2001). Both
measures are based on firm-specific weekly returns estimated as the residuals from the market
model. This ensures that our crash risk measures reflect firm-specific factors rather than broad
market movements. Specifically, we estimate the following expanded market model regression:
to reduce taxable income without decreasing book income. BTDs have been shown to be associated with
earnings management activities (Phillips, Pincus, and Rego 2003), growth in future earnings (Lev and Nissim
2004), and earnings persistence (Hanlon 2005). 7 Both SHELTER and DTAX have their share of limitation, though. Both these measures rely on a very small
sample of firms (59 and 211 shelter-engaging firms in Wilson and Lisowsky respectively) whose shelter
behavior was detected by the tax authorities and litigated. Frank et al. (2009) DTAX measure relies on the
premise that permanent differences are more aggressive than timing differences. Although the shelter
probabilities and DTAX are likely correlated with aggressive tax planning, they may fail to satisfactorily
captures all potential aggressive tax planning strategies.
21 | P a g e
threatens auditor independence and, hence, accentuates future crash risk. In the former (latter)
case the coefficient on the interaction variable should be negative (positive).
Finally we test tax provision by incumbent auditors on crash risk for firms following
differing business strategies. The following regression equation is developed.
)9.........(..............................4||
****
*
18117116115
11411311211111019
181716
15141312110,
IMRBIGROADACLEVMTBSIZESDRETRETTURN
AVOIDSTRATEGYTAXAVOIDSTRATEGYSTRATEGYTAX
AVOIDTAXAVOIDSTRATEGYTAXCRASHCRASH
ttt
tttttt
ttt
tttttti
where, STRATEGY is categorized into prospector/innovator strategies (PROSPECT) and
defender (DEFEND) strategies. Details of STRATEGY score composition are given in 3.3 above.
All other variables are defined as before. Higgins et al. (2014) document an increasing
propensity for innovators to engage in more tax avoidance compared to their defender
counterparts. If auditor-provided NATS mitigate future crash risk then this effect should be more
pronounced for innovators willing to engage in more tax avoidance. Therefore, we expect a
negative coefficient on the three-way interaction variable TAX*STRATEGY*AVOID if
knowledge spillover benefits dominate.
To control for unobservable industry and year characteristics associated with firm tax
service provision and crash risk, we include year and industry dummy variables in all our
regression specifications. To take into account the time series and cross-sectional dependence in
the error terms of our regressions, we calculate t-statistics using standard errors that are clustered
by both firm and year (Petersen 2009).
22 | P a g e
4. Sample selection and descriptive statistics
Our sample selection procedure begins with a total of 163,266 firm-year observations with
available auditor-provided NATS and relevant audit-related data retrieved from Audit Analytics
from the 2000-2012 sample period. We begin our sample period in 2002, the year that Congress
ratified SOX. We then excluded firm-year observations from the regulated industries (two digit
SIC 49) and financial institutions (two-digit codes 60-69). This eliminated a total of 55,720 firm-
year observations. We further eliminated 15,879 non-US firm-year observations, yielding a US
sample with the required audit data of 88,897 firm-year observations. We then matched CIK
codes from COMPUSTAT with the Audit Analytics CIK codes and found a matched sample of
45,573 firm-year observations. Not all of these observations had the requisite crash risk measures
and related control variables, and this reduced the sample to 21,950 firm-year observations. Our
sample size further reduces for tax avoidance analysis because of missing tax avoidance data.
Panel A in Table 1 provides descriptive statistics for the variables used in the regression
analyses. The mean values of the crash risk measures, NCSKEW and DUVOL, are -0.07 and -
0.37 respectively. About 67% of the firm-year observations procure tax services from their
incumbent auditors. In terms of dollar values, average firms pay about $303,000 in tax fees
although there is substantial variation among companies as is evident from the high standard
deviation. Average tax avoidance measures are 0.42, 0.02, and 0.28 for SHELTER, DTAX, and
ETR measures respectively. The average STRATEGY score is 16.91 on a scale of 6 to 30, with
prospectors and defenders constituting 5% and 11% respectively of the sample observations
respectively. Seventy seven percent of the firm-year observations are audited by Big 4 audit
firms. Firm-year observations come from a wide variety of industries with two digit SIC codes
35-39 and 70-79 commanding the largest industry representation in our sample. Panel C in Table
23 | P a g e
1 presents the correlation analysis. NATS and the two measures of crash risk are correlated
positively. Both crash risk proxies as well as NATS are significantly positively related to firm
size.
[TABLE 1 ABOUT HERE]
5. Multivariate regression results
We begin our multivariate analysis by modeling the determinants of firms’ decision to procure
tax services from their incumbent auditors, in order to alleviate the endogeneity concern, since
the choice of incumbent auditors over other providers is not a random selection. Regression
specification (4) is a probit model whereby auditor-provided NATS is regressed on a set of
variables likely to determine a firm’s decision to engage the incumbent auditor as tax service
provider (McGuire et al. 2012). The model includes a number of audit-related variables, e.g.
BIG4 and second tier auditor indicators, audit fees, and a variable that captures auditor
independence. Also included are financial variables likely to explain a firm’s decision to choose
the incumbent auditor as NATS provider. We find that larger firms, firms audited by Big 4, firms
paying higher audit fees, and firms with more foreign income are more likely to procure NATS
from incumbent auditors compared to firms audited by second tier audit firms, and firms with
more equity income. We computed IMR from this regression model and used it as an additional
independent variable in all the subsequent regressions to address the self-selection concern.
[TABLE 2 ABOUT HERE]
24 | P a g e
Table 3 reports our results for H1, which investigates the association between auditor-
provided NATS and future crash risk (Models 1& 2), and whether tax expense earnings
management (change in ETR from 3rd to 4th quarter) acts as a conduit through which NATS
accentuates or attenuates future crash risk (Models 3 & 4). The baseline regression model
reveals that NATS, proxied by the dollar amount of tax service costs, is negatively related to
future crash risk for both the crash risk measures (coefficients of -0.000017 and -0.000014 for
NCSKEW and DUVOL respectively). These negative and significant coefficients suggest that
auditor-provided NATS reduce future crash risk, lending support to the argument that knowledge
spillover benefits are being derived from NATS. Among the control variables, the coefficient on
average returns (RET) and ROA is positive and that on return volatility is negative suggesting
that firms with better stock and accounting performance and lower volatility are more likely to
experience crashes. This also suggests that crashes are unlikely to be a manifestation of declining
business conditions, continuation of poor stock performance (i.e., negative stock momentum),
and/or high stock volatility. Larger firms and high M/B firms are more prone to crash risk. BIG4
audit firms appear to constrain future crash risk but the coefficient is significant in the baseline
model.
We then extend our baseline regression model by incorporating two additional
moderating variables, namely change in 4th quarter ETR (∆ETR) and the interaction between TAX
and ∆ETR (TAX*∆ETR). Since a decrease in 4th quarter ETR compared to 3rd quarter ETR is used
as a proxy for earnings management, the coefficient on the interaction variables is expected to be
negative (positive) if the knowledge spillover (impairment of independence) argument
dominates. The negative (positive) coefficient, in turn, would indicate a lower (higher)
probability of a crash. Our empirical result is consistent with the knowledge spillover argument
25 | P a g e
as the interaction variables for both crash risk measures are negative and significant at better than
the 5% level. The coefficient on the baseline TAX variable continues to be negative and
significant.
Overall, the test of H1 supports the auditors’ constraining effect on tax expense
management, courtesy of their more intimate NATS-derived knowledge, which has the potential
to reduce subsequent crash risk.
[TABLE 3 ABOUT HERE]
Next we examine the association between auditor-provided NATS and crash risk, and ask
whether clients’ tax avoidance strategies strengthen or weaken the relationship. As argued