fsd

The Impact of Mandatory versus Voluntary Auditor Switch on Stock Liquidity in the Korean Market

Ferdinand A Gul Monash University, Sunway Campus, Malaysia

Woo-Jong Lee

The Hong Kong Polytechnic University

Abstract

Using Korean-listed firms subject to the auditor designation rule, this paper shows that, 1)

on average, stock liquidity is negatively related to auditor switches, and that 2) the negative

liquidity effect of the auditor switch is concentrated in firms that switch from high-quality

auditors to low-quality auditors. Meanwhile, firms with designated auditors exhibit even

higher stock liquidity when they have higher audit risk, consistent with enhanced audit

quality as a result of auditor designation.

Data Availability: The data are available from the sources identified in the paper.

JEL Classification Code: G14, M48

Keywords: Auditor switch; stock liquidity; auditor designation

_________________ * We gratefully acknowledge comments from Sunhwa Choi, Simon Fung, Lee-Seok Hwang, Yongtae Kim, Jay Junghun Lee, Se-Yong Lee, Jeffrey Pittman, Bahlgeun Roh, B. Charlie Sohn, Wilson Tong, and Cheong H. Yi. We also thank participants at Monash University, Kuala Lumpur seminar workshop for their helpful comments. We thank Cheong H. Yi for generously providing auditor designation data. We acknowledge financial support for this research from The Hong Kong Polytechnic University. All errors are our own. ** Correspondence: Woo-Jong Lee, School of Accounting and Finance, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong; Tel: +852 2766 4539; Fax: +852 2330 9845; Email: [email protected].

1

The Impact of Mandatory versus Voluntary Auditor Switch on Stock Liquidity in the Korean Market

1. Introduction

This paper investigates the association between auditor switches and stock liquidity.

Auditors play an important role in determining the quality of financial reporting and are

responsible for its effect in the stock market (Pitman, 2004). Chief Justice Burger in a

unanimous decision of the U.S. Supreme Court had this to say about the responsibility of

auditors to the public (Wild et al. 2001, p. 129):

By certifying the public reports that collectively depict a corporations financial status, the independent auditor assumes a public responsibility transcending any employment relationship with the client. The independent public accountant performing his special function owes ultimate allegiance to the corporations creditors and stock holders, as well as the investing public..Public faith in the reliability of a corporations financial statements depend upon the public perception of the outside auditor as an independent professionif investors were to view the auditor as an advocate for the corporate client, the value of the audit function might itself be lost.

Such a role of auditors in independently verifying financial statements could be more

important in emerging markets where the quality of financial reporting is generally poorer

(Francis and Wang, 2008). In addition, there is a growing demand for an investigation of the

audit quality in emerging market countries because investors are increasingly looking to these

markets to diversify their investment portfolios (Levich, 2001). At a general level, however,

the literature on emerging markets has relatively neglected the role of auditors in those

markets. We aim to fill this gap by examining how stock liquidity is related to firms decision

to switch their incumbent auditors.

2

In this paper, we focus on Korean-listed companies for the several reasons. First, the

Korean audit market enables us to compare the effects of voluntary auditor switches and

mandatory auditor switches on liquidity. The Korean regulatory authority, the Securities and

Futures Commission (SFC), mandates firms that have a propensity to manage earnings to

replace their incumbent auditors with designated auditors (a.k.a., auditor designation rule).

This in turn enables researchers to examine differential impacts of two types of auditor

switches. Second, the Korean stock market is characterized by high market liquidity.

According to Lesmond et al. (2004) and Lesmond (2005), while U.S. firms listed in NYSE

and AMEX have approximately 23.5 percent zero returns over an annual trading period,

Korean-listed firms exhibit about 13 percent zero returns, indicating that the Korean stock

market is more liquid than the U.S. market.1 Third, there is a high frequency of auditor

changes in the Korean audit market. In our sample spanning from 1995 to 2006, about 20

percent of our sample switch auditors, which is relatively high compared with U.S. firms.2

Collectively, both high liquidity and frequent auditor switches provides us with a better

setting to investigate the relation between auditor switches and stock liquidity.

1 This high liquidity of Korean stocks, at least in part, can be attributed to the significant presence of foreign investors in the Korean stock market. Unlike other emerging markets, the Korean capital market is widely open to foreign investors that demand high quality financial reporting and provide sufficient liquidity, especially in developing markets (Doidge et al., 2009). For instance, in 2006, foreign investors held 35.16 percent of the total shares in the Korean Stock Exchange, while domestic institutional investors held only 20.08 percent. The average foreign ownership of the 20 largest listed companies (based on market capitalization) was 52 percent as of 2006. 2 Highly frequent auditor switches may be attributable to the discrepancy between high demand and low supply for high quality audit services. First, the market share of the Big 4 auditors in Korea is very low compared with that in the U.S. According to Choi and Wong (2007), the market share of the Big 4 auditors in Korea is 66.36 percent, while that in the U.S. is 95.79 percent. Second, as Fan and Wong (2005) indicate, where legal institutions are less developed, firms with serious agency problems tend to employ the Big 4 auditors to alleviate them. Further, the Korean capital market is largely open to foreign investors that demand high quality financial reporting (Doidge et al., 2009). Such high demand for high-quality audit services, accompanied by the low supply of the Big 4 audit services, creates a unique opportunity for these Big 4 auditors to rebalance their audit client portfolios with ease.

3

In this paper, we first analyze the differences in stock liquidity across firms that switch

from high-quality to low-quality auditors (and vice versa) and firms that do not switch

auditors. In doing so, we employ three stock return-based constructs of stock liquidity: the

information cost as measured by Lesmond et al. (1999), the proportion of zero returns, and

the effective spread as measured by Roll (1984). We expect these stock return-based

constructs to reflect credibility of financial reporting. Using factor analysis, we identify a

liquidity factor, which captures the information contained in our three individual liquidity

variables. We then compare the liquidity of switching firms and non-switching firms. 3

Second, we examine whether or not the effect of auditor switches on liquidity is also

pronounced in firms that are required to change auditors under the auditor designation rule.

While auditor switches generally reflect the tendency of managers for opinion-shopping, Kim

and Yi (2009) document that mandatory auditor switches by auditor designation improves

accounting quality. We thus posit that investors reaction to mandatory auditor switches will

be different from that to voluntary auditor switches.

The premise of this paper is that auditor quality is positively associated with credibility

of corporate reporting, and hence stock liquidity. This premise is consistent with the notion

that high-quality auditors produce reliable information about firms that may efficiently

3 Another set of conventional liquidity measures may include Amihuds (2002) price impact measure and the bid-ask spread. Because the use of these measures drastically reduces the sample size, we report the results of these tests only as supplementary evidence under the additional tests section at the end of the paper. Meanwhile, there is no reason to believe that the use of such high-frequency measures bring about different results. Specifically, Goyenko et al. (2009) have shown that low-frequency measures (i.e., stock return-based measures) capture high-frequency measures (i.e., transaction-based measures) well.

4

resolve contracting problems (e.g. Jensen and Meckling, 1976; Watts and Zimmerman, 1986;

Francis et al., 1999; Pittman and Fortin, 2004). For example, high-quality auditors can

enhance the credibility of financial statements by improving the precision of the earnings

reports of firms (DeAngelo, 1981; Balvers et al., 1988, Kanagaretnam et al., 2010), thereby

lowering the firms cost of capital (Fortin and Pittman, 2007). Both theory and empirical

evidence are consistent with this link (Easley and OHara, 2004; Francis et al., 2005; Leuz

and Verrecchia, 2007). Furthermore, the literature also documents that earnings quality is

positively associated with stock liquidity. Lang, Lins, and Maffett (2011) document greater

liquidity for firms with less evidence of earnings management. Therefore, the quality of

auditors in assuring the quality of financial reporting is to be reflected in stock liquidity.

However, both theory and empirical evidence on the economic effects of an auditor

switch are somewhat incomplete and difficult to interpret. Theories suggest that the direction

of the capital market reaction to auditor change cannot be clearly anticipated. First, given that

auditor choice is an optimal firm decision, only auditor changes that enhance firm value

should be initiated, suggesting that the market should react positively to such changes. If a

change in auditor is fully anticipated by market participants, the market price of the firms

stock will then reflect its value. Therefore, the market price should be, at worst, zero in an

efficient market (Johnson and Lys, 1990). Second, if managers change auditors solely to

maximize their own utility rather than to maximize shareholder wealth, market reactions to

5

auditor changes will be negative. Furthermore, if a firm changes its incumbent auditor to

signal its future prospects, only a change from a high-quality auditor to a low-quality auditor

will reveal negative private information on firm prospects, leading to a short-term negative

impact on the value of a firm in a market with weak informational efficiency. The upshot of

these different arguments is that it is difficult to predict the market reaction to an auditor

switch.

Furthermore, the findings of empirical studies based on U.S. data remain inconclusive.

Earlier empirical evidence shows a negative or insignificant market reaction to auditor

change (Fried and Schiff, 1981; Nichols and Smith, 1983; John and Lys, 1990). By focusing

on a shorter event-window, later studies succeeded in finding a negative market reaction (e.g.,

Whisenant et al., 2003; Beneish et al., 2005; Knechel et al., 2007). On the other hand, other

studies document a positive or non-negative market reaction (Sankaraguruswamy and

Whisenant, 2004; Whisenant, 2006).4 In sum, evidence on the economic consequences of an

auditor switch remains insufficient.5

Our empirical tests use two sets of data to address the implications of an auditor switch.

First, using a large sample of Korean-listed firms that exhibit a high frequency of auditor

4 As for Korean evidences, since Korean firms started to disclose auditor switch news after 2002, it was practically impossible to examine stock market responses around auditor switch news in Korea before 2002. Using 201 Korean firms in 2002 and 2003, Sohn and Kim (2004) find that stock market negatively responds to auditor switch. 5 It is also argued that the economic consequences of an auditor switch vary with the type of the switch. For instance, regulators expect switches from a low-quality auditor (i.e., a non-Big 4 auditor) to a high-quality auditor (i.e., a Big 4 auditor) to enhance the credibility of the firms financial statements, improve corporate transparency, and hence benefit investors (e.g., Securities and Exchange Commission, 1988). This therefore implies that the economic consequences of a change in auditor are asymmetric to the type of auditor change (e.g., Eichenseher et al., 1989). Extant evidence fairly consistently supports that investors interpret switches negatively when the analysis more narrowly isolates those involving downgrades (i.e., from Big N to non-Big N).

6

switches, we analyze the effects of auditor switches on stock liquidity. Second, by

categorizing our sample into firms subject to mandatory auditor change (i.e., where the

auditor is designated by the SFC) and those that change their auditor on a voluntary basis (i.e.,

where the auditor switch is initiated by the client firm), we examine whether or not auditor

switches that are mandatory also have an effect on stock liquidity. The SFC has assigned

statutory auditors to firms suspected of accounting fraud under a prescribed set of conditions

since 1990. Given that this type of regulatory designation forces firms that would not

otherwise switch auditors, we expect an auditor switch initiated by the rule to have a non-

negative effect on stock liquidity. More details on the institutional uniqueness of the auditor

designation rule in Korea are presented in Section 2.

In this work, we run cross-sectional firm-level regressions using these two datasets and

find lower stock liquidity for firms that switch auditors than those that do not. Further

analyses reveal that this effect is concentrated on BN firms (i.e., firms that switch from Big 4

to non-Big 4 auditors).6 In one of our specifications that use zero return frequency as a proxy

for stock liquidity, the percentage of days without trades is higher in BN firms by 1 percent,

which represents about a 3 percent deterioration in liquidity relative to firms that do not

switch auditors. Even when we aggregated all three liquidity proxies into a single liquidity

factor for parsimony, the tenor of the results does not change. 6 We used NB, BN, BB, and NN to designate firms that switch from non-Big 4 to Big 4 auditors, firms that switch from Big 4 to non-Big 4 auditors, firms that switch from one Big 4 to another Big 4 auditor, and firms that switch from one non-Big 4 to another non-Big 4 auditor, respectively. Likewise, our industry specialty tests also used similar terms, such as NS, SN, SS and SS, where N (S) denotes a non-specialist (specialist) auditor.

7

We also relax the assumption that a firms decision on switching auditors is exogenous.

In reality, the auditor choice decision is likely to be endogenous. We use both Heckman

(1979)s 2 stage regression approach and propensity score matching to address the concern,

thereby ensuring that our findings are not due to the endogeneity of the auditor switching

decision, but are more likely to represent a causal relationship.

While the liquidity effects for firms switching auditors are statistically and

economically significant, these effects are generally smaller for firms whose auditor switches

are mandated by the regulatory body than firms voluntarily switching auditors. This suggests

that investors are more concerned about voluntary auditor switches than mandatory auditor

switches. Interestingly, we find that firms with designated auditors do not exhibit lower

liquidity, consistent with the market perception that auditor designation results in enhanced

audit quality. When a new auditor is assigned to a firm suspected of earnings manipulation

under certain conditions, such as being subject to high default risk, investors expect the

auditor designation to improve the credibility of the firms financial statements, which is

consistent with the view that designated auditors improve audit quality (Kim and Yi, 2009).

This study makes several important contributions to the literature. First, while prior

studies have examined different dimensions of liquidity in different markets (Bacidorea et al.,

2005; Frino et al., 2008; Kryzanowski et al., 2010), in this study we focus on Korea, a rapidly

developing economy. In this way we extend the vast literature on liquidity issues, in general,

8

and the links between financial reporting and liquidity, in particular. Second, prior studies

regarding the link between the economic consequences of an auditor switch and stock returns

are inconclusive and mixed. In this paper, we instead focus on stock liquidity around the time

of the auditor switch as an alternative proxy for the capital market effect. As the only

exception, Boone and Raman (2001) find that liquidity decreases for firms whose auditors

resigned, but not for firms whose auditor is dismissed. While they focus on auditors

resignation to adjust client portfolios, we discuss clients incentives to switch auditors and

extend their findings. Third, the concern of Korean regulators in auditor switches is that

auditor changes are motivated by management opportunism. Hence, the findings should be of

interest to regulators and policy makers in Korea as well as in any other countries considering

the adoption of similar regulation. Fourth, our study makes a novel contribution in that it

examines the effect of a mandatory auditor switch on liquidity. This is particularly interesting

because the economic consequences of the auditor switch depend on who actually initiates

the decision to switch auditors. Finally, our paper adds to the general body of knowledge

concerning the role played by high quality audits in reducing information asymmetry in the

market (Autore et al., 2009; Pittman and Fortin, 2004).

The remainder of the paper is organized as follows. Section 2 develops our hypotheses

and reviews the literature. Section 3 specifies the research design we use for hypothesis

testing and Section 4 presents the empirical results. Section 5 summarizes and concludes this

9

paper. In the Appendix, we provide additional details on the construction of the three

liquidity proxies.

2. Literature review and hypothesis development

2.1. Auditor switch and stock liquidity

Rational managers may maximize their own utility at the expense of the shareholders

rather than boosting the shareholders wealth by enhancing firm value. However, it would be

still fair to assume that the objective of the auditor change decision is the maximization of

firm value. Empirical evidence in this line of research is largely inconclusive. Most

importantly, while some of the prior literature shows that opportunistic managers switch

auditors to obtain more favorable audit opinions (Lennox, 2000), other empirical evidence on

opinion shopping is not supportive of this finding because post-switch opinions are no more

favorable than pre-switch opinions (e.g., Krishnan and Stephens, 1995). Numerous papers

that use stock returns to proxy for the capital market effect document that investors react

negatively to auditor changes and that stock price drops vary cross-sectionally with auditor or

firm characteristics (Smith, 1988; Wells and Loudder, 1997; Shu, 2000; Whisenant et al.,

2003; Beneish et al., 2005; Knechel et al., 2007). While the prior studies noted above report a

negative response to auditor switch announcements, some studies document a positive or

non-negative market reaction (Sankaraguruswamy and Whisenant, 2004; Whisenant, 2006),

thereby raising questions over the opportunistic auditor switch assumption.

10

While tests based on stock returns conducted in prior studies are aimed at reflecting

perceived changes in firm value surrounding the auditor switch, our tests attempt to

determine whether or not the auditor switch implies the willingness of investors to face risks

stemming from changes in the credibility of the firms financial statements. An important

distinction between stock return and liquidity tests is that the former reflects changes in the

expectations of the market as a whole, while the latter reflects changes in the expectations of

individual investors. Although news of an auditor switch may be neutral in that it does not

change the expectations of the market as a whole, it may greatly alter the expectations of

individual investors. In an extreme case, we can envisage a shift in portfolio positions, which

is captured by a change in liquidity without any price reaction. If so, a return-based test may

be less sensitive to news of an auditor switch than a liquidity-based test. In a similar vein, we

expect stock liquidity to be more sensitive to auditor switch news than stock returns to the

extent that we assume informed and uninformed traders react differently to such news.

Informed traders are likely to be indifferent to information about the credibility of financial

statements because of their greater access to private information about the firm, while

uninformed traders may indeed be affected by such an announcement. Stock liquidity then

better captures the heterogeneity in the expectations of informed and uninformed traders than

do stock returns. Hence, to better examine the market perception of auditor changes, we

11

employ stock liquidity instead of stock returns and test the following hypothesis in alternative

form:

H1A: The stock liquidity of firms that switch auditors is significantly different from that of

firms that do not switch auditors, other things being equal.

The premise underlying this hypothesis is that an auditor switch conveys new

information to capital market participants about the firms future prospects. We expect that

switching from a high-quality auditor to a low-quality auditor (from a low-quality auditor to a

high-quality auditor) makes corporate reporting less (more) credible, increases (decreases)

information asymmetry, and hence decreases (increases) stock liquidity.

2.2. Effect of auditor designation

In Korea, the regulatory authority designates external auditors for firms deemed to have

strong incentives and/or great potential for fraudulent accounting and orders these firms to

replace their incumbent auditors, appoint new auditors, and retain them for a certain period.

This regulatory regime is known as auditor designation or mandatory auditor assignment.

(Kim and Yi, 2009; Jeong et al., 2005).

12

The auditor designation rule in Korea is unique and different from mandatory auditor

rotation in three ways. By designating auditors for problematic firms, the auditor designation

rule aims to alleviate alleged problems arising from long-term relationships between auditors

and clients while mandatory auditor rotation typically requires auditor changes for all firms

after a certain period of an initial engagement. Furthermore, the auditor designation rule

mandates retention of newly designated auditors for a certain period to protect designated

auditors from early dismissal, and thereby mitigating a major threat to auditor independence.

Finally, by requiring the regulatory authority to choose incoming auditors, the auditor

designation rule enhances auditor independence, and thereby improving audit quality. To our

knowledge, no country other than Korea has ever introduced or considered an auditor

designation policy. We take advantage of this unique institutional feature to examine whether

differences in stock liquidity are less pronounced in firms with designated auditors.

Empirical evidence suggests that the appointment of a designated auditor may result in

higher audit quality. Jeong et al. (2005) find that designated auditors charge significantly

higher audit fees than others. More recently, Kim and Yi (2009) find that the auditor

designation rule is effective in deterring managers from engaging in income-increasing

earnings management. They also find that firms subject to mandatory auditor changes

effected through the auditor designation regime report significantly lower discretionary

13

accruals than firms that make voluntary auditor changes, suggesting that the appointment of a

designated auditor enhances audit quality.

To the extent that the auditor designation rule is effective in improving auditor

independence and thus audit quality, designated auditors are likely to be less lenient towards

managerial opportunism than are non-designated auditors. As a consequence, firms with

designated auditors should exhibit higher stock liquidity, or at least they should not exhibit

lower stock liquidity. We expect the liquidity effects of mandatory auditor changes (i.e.,

auditor designation) to be less pronounced than those of voluntary auditor changes and thus

test the following hypothesis.

H2: The stock liquidity effects of auditor switches are less pronounced when firms are

mandated to switch auditors.

3. Research design and sample composition

3.1. Research design

To test our hypotheses, we specify the following regression model linking stock

liquidity with auditor switch and other control variables.

Liquidityit = 0 + 1 SWITCHit + 2 SIZEit + 3 LEVit + 4 ATURNit + 5 ROAit + 6 SGROWit + 7 QUICKit + 8 RETVARit + Year + Industry + it, (1)

where for firm i and year t:

14

Dependent Variables Liquidity stock liquidity proxies; LOT, ZERO, ROLL, and FACTOR LOT = The natural logarithm of a yearly estimate of the transaction costs implied

by the trading behavior of investors developed by Lesmond et al. (1999). The estimation period spans month -5 through month +7 so that the firms annual reports are publicly available for a few months, which should enable markets to impound the auditor switch effect.

ZERO = The proportion of zero daily returns out of the maximum potential number of trading days in a given year. The measurement period spans month -5 through month +7 relative to the firms fiscal year-end.

ROLL = Rolls (1984) effective spread based on the bid-ask bounce-induced negative serial correlation in returns. The measurement period spans month -5 through month +7 relative to the firms fiscal year-end.

FACTOR = a liquidity factor based on factor analysis using LOT, ZERO, and ROLL

Independent Variables Switch = an indicator variable which is one if a firm switches its auditor and zero

otherwise BN = an indicator variable which is one if a firm switches from a Big 4 auditor to

a non-Big 4 auditor and zero otherwise NB = an indicator variable which is one if a firm switches from a non-Big 4

auditor to a Big 4 auditor and zero otherwise BB = an indicator variable which is one if a firm switches from one Big 4 auditor

to another Big 4 auditor and zero otherwise NN = an indicator variable which is one if a firm switches from one non-Big 4

auditor to another non-Big 4 auditor and zero otherwise Control Variables

SIZE = the natural logarithm of the book value of total assets LEV = financial leverage measure by the ratio of total debt to total assets ATURN = asset turnover measured as sales deflated by assets ROA = net income deflated by total assets SGROW = sales growth measured as the difference in sales deflated by lagged sales QUICK = current assets deflated by current liabilities RETVAR = return variability measured as the standard deviation of daily returns

To conduct the empirical tests, we create a binary indicator variable, Switch, that takes

the value of one for firms that switch auditors. This variable should capture the average

liquidity effect around the time of the auditor switch as the key variable of interest. We

introduce four separate indicator variables for firm-year observations from firms that switch

15

auditors. We distinguish BN, NB, BB, and NN switchers, respectively, depending on the types

of the former and successor auditors.

For dependent variables, we use four proxies LOT, ZERO, ROLL, and FACTOR for

stock liquidity.7 We construct the first three proxies using daily individual stock returns and

aggregate market returns. Assuming that daily returns are normally distributed, we estimate

all three measures for each firm and year over the period from month -5 through month +7

relative to the firms fiscal year-end.8 The first dependent variable, LOT, is a yearly estimate

of the total round trip transaction costs implied by the trading behavior of investors developed

by Lesmond et al. (1999). The second dependent variable, ZERO, is the proportion of zero

daily returns out of the maximum potential number of trading days in a given year. Lesmond

et al. (1999) argue that zero-return days occur when the transaction costs of trading outweigh

the value of new information not yet contained in prices and it is better not to trade. The zero-

return metric commonly serves as a proxy for illiquidity (e.g., Bekaert et al., 2007). The third

measure is ROLL, which is designed to capture the bid-ask bounce-induced negative serial

autocorrelation in returns to estimate the effective spread. If the serial autocovariance is

positive, we force it to be negative and use the Roll estimate as if a negative serial

autocovariance were being estimated (Harris, 1989). Harris (1990) explains that positive

7 In Table A1, we also report the results based on the price impact measure and the bid-ask spread. We avoid using these two measures in the main analyses because of limited data availability. 8 Korean listed firms have not been required to disclose auditor switch news before 2002. Without the exact date of auditors being replaced, we are unable to examine short-term market responses around auditor switch dates. Admitting potential confounding effects can exist in our long-horizon test, we also use two different estimation periods for our liquidity proxies (month -6 to month +6, and month -4 to month +8). Our results hold robust to the changes in estimation periods.

16

autocovariance can result from closing prices that cluster at the ask, violating Rolls

assumption of trade independence. Due to the strengths and weaknesses of each liquidity

measure and proxy, we employ all three measures to construct FACTOR using factor analysis

and use it as a separate dependent variable in the analyses.9 Note that higher values of all of

our liquidity proxies indicate lower stock liquidity by definition. In common with prior

studies, we control for firm size, leverage, asset turnover, return-on-assets ratio, sales growth,

the quick ratio, and return variability in the regressions (Chordia et al., 2000; Leuz and

Verrecchia, 2000).

3.2. Sample composition

The initial sample for our study consists of all firms listed on the Korea Stock Exchange

(KSE) and the Korea Securities Dealers Automated Quotation (KOSDAQ). We retrieve

financial data from the KIS-Value database developed by the Korea Information Service

(KIS), the top credit rating agency affiliated with Moodys. This database has been used in

several recent studies on accounting and finance (Kim and Yi, 2009). We obtain monthly

stock returns data from the 2006 Korea Securities Research Institute database (KSRI 2006).

The sample period spans from 1995 to 2006.10

9 Factor analysis is initially run on three measures of stock liquidity and the factor obtained is used to test the hypotheses. Factor analysis is a data reduction technique that simplifies complex and diverse relationships that exist among a set of variables by generating new variables, the factors of which extract the main sources of variation among the original variables (Dillon and Goldstein, 1984). Specifically, we combine three liquidity proxies (i.e., trading costs based on Lesmond et al. (1999), zero returns, and effective spreads based on Roll, 1984) by factor analysis with one oblique rotation. The liquidity factor is the factor score from the first (and only) factor with an eigenvalue greater than one. 10 We start from 1995 because this is the first year for which stock return data are available in the KSRI database.

17

We begin the sample collection procedure with all observations for which we have the

necessary data to compute the variables used in the firm-year regressions described above.

We exclude financial companies from our sample because the financial reporting

environment for financial institutions differs from that for companies in other industries. The

final sample includes 9,241 firm-year observations after we delete observations with missing

financial data. For our auditor designation sample, we obtain a list of 583 firm-year

observations for 291 distinct firms whose auditors were designated in the 19952002 period

from the Korea Financial Supervisory Service. 11 For the analysis testing our second

hypothesis, the data consist of 5,093 firm-year observations including 583 auditor-designated

observations.

4. Empirical results

4.1. Sample description

Table 1 summarizes our variable definitions and Table 2 describes the summary

statistics. Panel A of Table 2 provides a breakdown of the number of observations and shows

the composition of the samples used for our analyses. The full sample consists of 7,445 non-

switching firm-year observations and 1,796 switching firm-year observations. The second

dataset, the auditor designation sample, spans the period from 1995 to 2002 and consists of

583 firm-year observations in which mandatory auditor switches take place. The number of 11 We thank Professor Cheong H. Yi for providing the list of firms subject to auditor designation during the 1995-2002 period. We dont extend the sample period because SFC practically stopped designating auditors for listed companies. Since 2003, only a small number of firms subject to delisting have been mandated to change auditors.

18

observations in our non-designated sample during the same period is 4,510, yielding 5,093

firm-year observations in total.12

Panel B of Table 2 shows that the mean of SWITCH is 19.4% for the final sample,

indicating that about one-fifth of our firm-year observations are switchers, a higher

proportion than those observed in U.S. samples. Among these switching firms, about half

(8.5%) change from one Big 4 auditor to another Big 4 auditor, suggesting that caution is

required when drawing inferences based solely on SWITCH as a proxy for audit quality. The

means of LOT and ZERO are -4.835 and 0.089, respectively. The 8.9% of trading days with

zero returns equates to 23 days in each year without price changes. The ROLL measure is

2.3% on average, which is a fairly low effective spread cost. As previously discussed, the

Korean stock market has better liquidity than other emerging markets, a feature we expect to

provide us with reasonable test power in the following analyses. We winsorize all the

variables other than those with natural lower and upper bounds at the 1st and 99th percentiles.

The correlations reported in Panel C of Table 2 show that all three liquidity variables are

positively associated with the auditor switch indicator (SWITCH), and similarly, return

variability (RETVAR) is positively correlated with the switch indicator. The three liquidity

proxies are also significantly and positively related with one another, suggesting the effective

internal validity of these measures. Firm size (SIZE), financial leverage (LEV), asset turnover

12 The availability of audit fee data used to construct fee-based industry specialization variables restricts our final dataset to 4,850 observations since audit fee data is publicly available since 2001.

19

(ATURN), the return-on-assets ratio (ROA), sales growth (SGROW), the quick ratio (QUICK),

and return variability (RETVAR) are significantly related to liquidity measures, thereby

confirming that we need to control for these variables in examining the incremental liquidity

effect of an auditor switch.

4.2. Auditor switch and stock liquidity

4.2.1. Multivariate analyses

We now test the association between auditor switches and stock liquidity for the full

sample. We estimate Eq. (1) using our full sample of 9,241 firm-year observations covering

1,677 unique firms. In Table 3, we report the regression coefficients and their t-values

computed using standard errors adjusted for clustering at the firm level for our pooled OLS

regressions to control for serial correlation problems.

The results reported in Panel A of Table 3 show that the coefficient for SWITCH is

significantly positive in all cases in which a stock liquidity measure is used as the dependent

variable. These findings are consistent with the view that auditor switches have negative

economic consequences and suggest that auditor switching firms are less liquid than their

non-switching counterparts.

We then run regressions using separate indicators of the auditor switch types BN, NB,

BB, and NN in Panel B of Table 3. We find that the negative stock liquidity effect of an

auditor switch is pronounced only in BN firms. In our specification, the percentage of days

20

without trades increases by 1% for BN firms, which represents a deterioration in liquidity of

about 3% (= 0.009/0.274) relative to the liquidity of firms that do not switch auditors. The

coefficients of both LOT and ROLL are also significantly positive (t = 3.94 and 2.18,

respectively). When we aggregate all three liquidity proxies into a single liquidity factor for

parsimony, we also find a significantly lower liquidity only for BN firms (coefficient = 0.196,

t = 3.69).

Turning to the control variables, we observe several significant relations. Using the

liquidity factor as a dependent variable, we find that consistent with the prior market

microstructure literature (e.g., Chordia et al., 2000; Leuz and Verrecchia, 2000), larger firms

and firms with higher profitability, higher sales growth, a higher quick ratio, and higher

return variability exhibit higher stock liquidity.

4.2.1. Accounting for endogenous auditor switching decisions

The reverse causality problem may not be a big concern in Table 3 because it is unlikely

that firms facing the prospect of lower stock liquidity will tend to replace high-quality

auditors with low-quality auditors. Rather, as Fan and Wong (2005) argue, firms with a

higher level of information risk will hire high-quality auditors to mitigate investors concerns

about information risk. In this regard, the potential reverse causality issue, if any, is unlikely

to explain our results because it works against finding significant results in our analyses.

21

On the other hand, because auditor switching decisions are unlikely to be exogenous,

this may raise some concerns over our previous findings on the liquidity effect of an auditor

switch. To address the concern, we use matched sample techniques that have been developed

in the literature (e.g., Clatworthy et al., 2009). Specifically, we estimate the probit model of

the probability of switching auditor using a set of firm characteristics as explanatory variables,

as described below:

SWITCHit = 0 + 1 Xit-1 + Year + Industry + it, (2)

where SWITCH is an indicator of whether firm i switches auditor in year t, X is the vector of

firm characteristics including KSE (an indicator of whether a firm is listed on the Korean

Stock Exchange), Size (the natural logarithm of total assets), ROA (the return-on-assets ratio),

LOSS (an indicator of a firm reporting negative net income), the interaction of ROA and

LOSS, and NSWITCH (the frequency of auditor switches during the sample period).13 We

take the predicted value from Equation (2) to construct the propensity score of auditor

switches for each firm i in year t. Next, using this propensity score, we match auditor-

switching firm-years with non-switching firm-years that have the closest propensity score to

the matched switching firm for the same year.14 The resultant sample thus consists solely of

firms that have a similar probability of switching observations. By doing so, we control for

13 At the first stage, all the explanatory variables other than SIZE are statistically significant at the 5% significance level. The model also has reasonable fit as its log likelihood is -3,676.49. The coefficients are estimated as: SWITCH=-1.30-0.51*KSE-0.01*SIZE+1.44*ROA-2.37*ROA*LOSS+0.11*LOSS+0.47*NSWITCH. 14 The mean value of the differences in propensity scores between the two samples is 0.003, with a standard deviation of 0.013, indicating our propensity matching exercise is reasonably successful.

22

firm characteristics that could possibly drive auditor switch decisions. In such a sample, the

coefficient on our SWITCH indicator variable can be reliably interpreted as a causal effect of

the auditor switch on stock liquidity.15

Before analyzing the matched sample, we first use observations for firms that switch

auditors only once during our sample periods shown in columns (1) and (2) of Table 4. This

sample consists of 2,819 observations for 542 firms. We choose one-time switchers only

because we expect they exhibit a clearer difference in stock liquidity after changing auditors

than firms that switch auditors repetitively. By doing so, we can compare stock liquidity of

pre-switching periods with that of post-periods for such firms. In column (1), the coefficient

of SWITCH is positive but insignificant. But in column (2), the coefficient of BN remains

significantly positive, suggesting that firms experience lower liquidity after they switch high

quality auditors to low quality auditors.16 This reconfirms our previous results that stock

liquidity is related to the quality of auditors.

15 Propensity score matching has several advantages over instrumental variable (IV) methods as a solution of the endogeneity problem (Core, 2010). First, it does not require the identification of reliable IVs which is very challenging in many empirical settings. For example, Clatworthy et al. (2009) and Francis, Lennox, and Wang (2010) find that the inferences from Heckmans two-stage estimation are sensitive to the choice of IVs. Furthermore, Larcker and Rusticus (2010) show that if IVs are not appropriate, ordinary least squares estimates can be better than two-stage or three-stage least squares (2SLS or 3SLS) estimates. Second, it is robust to misspecification of the functional form because it is not subject to parametric assumptions of IV methods (Armstrong et al., 2010). However, propensity score matching does not address endogeneity directly as do IV methods. We thus discuss the robustness of our results using Heckmans two-stage estimation in footnote 20. 16 Examining changes in liquidity around the event of auditor switches is known as an effective way in controlling for potential biases from omitted variables. However, the event date cannot be identified in our setting. Even though auditor switches can occur at any time during the fiscal year, in Korea, firms switching auditors have not been required to announce the news at the event before 2002. As a result, it is practically impossible to figure out when the auditor switch news is released to the public. Therefore, it may not be appropriate to investigate the impact of auditor switches on the change in liquidity variables over time. It is noted that matching samples can be an alternative in such a case (Mayhew and Mihov, 2004).

23

In columns (3) and (4), we match all 1,796 switching year observations to the same

number of non-switching year observations, yielding 3,592 firm-year observations in total.

The matched non-switching observations should have the closest propensity score in the same

year of the switching observations. Specifically, a switching firm-year observation is matched

with a non-switching firm-year observation having the most similar likelihood of switching

an auditor in the same year, which means the benchmark group in columns (3) and (4)

consists of non-switching observations in the same year of switching observations. For

completeness, in columns (5) and (6), we choose the switching year observations of 542 one-

time switchers that are used in columns (1) and (2) and match them to the same number of

observations for firms that have never changed auditors, yielding a total of 1,084

observations. The matching also requires the closest propensity score in the same year.

The results in columns (3) to (6) corroborate our previous findings that auditor switches

have negative implications for stock liquidity. The findings suggest that, even after we

control for firm characteristics that might affect auditor switch decisions, the negative

liquidity effect of auditor switch holds robust. Especially, the within-firm approach in

columns (1) and (2) confirms that the results in Table 3 are not due to the potential bias of

auditor switching decisions, but are instead likely to reflect a causal relationship.17

17 Despite several advantages of the propensity matching technique, its major limitation is that matching is based only on observable variables and therefore it cannot control for selectivity based on unobservables. We thus conduct Heckmans (1979) 2SLS test for robustness check and find similar results. With the inclusion of the inverse Mills ratio estimated from the first stage probit model in the second stage OLS, the coefficient of SWITCH is positive but becomes insignificant (coeff. = 0.102, t = 0.98), while the coefficient of BN remains significantly positive (coeff. = 0.238, t = 2.10). However, we are also cautious in interpreting this result because of the difficulty of identifying an exogenous instrument variable in the first stage.

24

4.3. The liquidity effects of a mandatory auditor switch: evidence from an auditor designation

sample

4.3.1. The effect of a mandatory auditor switch on stock liquidity

In this section, we compare the liquidity effects of mandatory auditor changes (i.e.,

auditor designation) with those of voluntary auditor changes to investigate further whether

the effect of mandatory auditor change differs systematically from that of voluntary auditor

change. In so doing, we construct a subsample in which auditors are switched on either a

mandatory or a voluntary basis from 1995 to 2002. We use 5,093 firm-year observations

consisting of 583 firm-year observations for designated auditor switches and 4,510 firm-year

observations for non-designated auditor switches.

Table 5 reports the subsample results for the liquidity effects of auditor designation. For

simplicity, we report only the results based on the liquidity factor (FACTOR), although the

results obtained using the three different individual measures are qualitatively similar to those

reported for the composite factor. Column (1) shows the result of Eq. (1) using the subsample

and reports a positive coefficient of SWITCH. We include the auditor designation dummy

(DSG) and define an interaction term, DSG*SWITCH, that captures any incremental liquidity

effect for firms subject to the auditor designation rule. As shown in column (3), the

Note that it is essential to identify exogenous independent variables from the first stage choice model that can be validly excluded from the second stage regression to successfully implement the Heckman (1979) procedure (Francis et al., 2010). We thus consider NSWITCH as an exogenous variable to capture firms tendency to frequently switch auditors. We also expect NSWITCH not to be related to the error terms in Equation (1) as it is counted over the sample period and there is no reason to expect it has an impact on stock liquidity in a particular year.

25

coefficient of DSG*SWITCH is significantly negative, which indicates that the negative

liquidity effect of auditor switching is less pronounced if the switch is driven by the auditor

designation rule. As the rule aims to improve auditor independence and thus audit quality,

switching to an incoming designated auditor is perceived to have positive economic

consequences in the capital market. The F-test result confirms that the negative impact of an

auditor switch is no longer observed when the switch is mandatory. As the auditor is

designated when a firm is suspected of fraudulent accounting, this finding is consistent with

the view that designated auditors improve audit quality (Kim and Yi, 2009).

Columns (3) and (4) show that the negative liquidity effects of auditor-switching firms

are concentrated in firms that switch from a Big 4 auditor to a non-Big 4 auditor and that such

effects are less pronounced when the switch is initiated by the rule. The positive liquidity

effects of auditor designation, which we report in Columns (2) and (4), are mostly driven by

firms switching from a non-Big 4 auditor to a Big 4 auditor.

4.3.2. The effect of audit risk on the relation between a mandated auditor switch and stock

liquidity

If the non-negative liquidity impact of a mandated auditor switch is attributed to the

fact that investors are well aware of the regulatory intervention objective of improving audit

quality, this impact is likely to be greater when the firm is subject to higher audit risk. We

26

thus further examine whether the liquidity effects of designated auditor switches vary with

the level of audit risk. As shown in Panel A of Table 6, Article 10 of the Financial

Supervisory Service (FSC) Regulation under the Act on External Audit stipulates that listed

firms will be subject to an auditor designation requirement if they meet one of the 10

specified conditions. Following Kim and Yi (2009), we consider that the first four conditions

describe instances of lower client risk, while the last six describe circumstances of higher

client risk. Among the 583 auditor designation sample observations, 184 (399) belong to the

low-risk (high-risk) group.18 Panel B shows that the liquidity effect is pronounced only in

NB firms subject to higher risk. Difference tests between the two risk groups show that

liquidity effects can be observed in firms switching to Big 4 auditors, regardless of the type

of auditor formerly engaged.

The results imply that the positive liquidity effect of a mandatory auditor switch is

related to the potential impact of auditor designation on the audit risk of client firms. The

finding that there is a positive liquidity effect of auditor switches in firms with higher audit

risk suggests that investors expect designated auditors to play a more significant role in such

firms. Accordingly, the results reported in Table 6 corroborate our previous findings about

the implications of mandatory auditor switches.

4.4. Additional tests 18 We follow Kim and Yi (2009) in classifying risk. Of the ten criteria for auditor designation, four are regarded as indicating low client risk and the other six are regarded as indicating high client risk. See Table 6 for details.

27

4.4.1. An alternative proxy for audit quality: industry specialization

Prior studies have employed auditor brand name or auditor size (i.e., Big 4) as a proxy

for audit quality. However, audit firm brand name is only one possible indication of audit

quality (e.g., Knechel et al., 2007). To mitigate the concern over using an empirical proxy for

audit quality, recent studies have hypothesized that auditor industry specialization also

contributes to audit quality. Knechel et al. (2007) find that firms switching between Big 4

auditors experience significant positive abnormal returns when the successor auditor is an

industry specialist and experience significant negative abnormal returns when the successor

auditor is not a specialist. Balsam et al. (2003) report that client firms of specialist auditors

have lower discretionary accruals and higher earnings response coefficients than client firms

of non-specialists. On the whole, the industry specialist indicator is perceived as an effective

proxy for audit quality. We thus corroborate our analyses by focusing on industry

specialization. Given that the market perceives audit quality differences based on industry

specialization (Knechel et al., 2007), this test will confirm that differences in the credibility

of financial statements are relevant to stock liquidity.

We identify auditor expertise by considering each auditors industry market share in a

specific year. To identify industry specialists, we manually collect audit fee information

disclosed by Korean listed firms. We code an indicator variable (Specialist) equal to one if

the firm is audited by an industry specialist that reports the greatest market share in terms of

28

the audit fee in each two digit industry. We also categorize switchers as SN, NS, SS, or NN

based on the industry specialization of their successor auditors. In so doing, we examine

whether a switch from an industry specialist auditor to a non-specialist auditor has a

significant impact on stock liquidity. If the auditor switch has a discernible effect on stock

liquidity, we expect the firm switching from a specialist (a non-specialist) auditor to a non-

specialist (specialist) auditor to exhibit a lower (higher) level of stock liquidity than a firm

that does not switch its auditor. This test is carried out using 4,850 firm-year observations

during the 20012005 period, a timeframe for which audit fee data are publicly available.

Table 7 confirms that the negative liquidity effect of an auditor switch remains constant

in the subset of data for the limited sample period from 2001 to 2005. Similar to the previous

result showing a concentrated liquidity effect among BN firms, the liquidity effect of auditor

switches is most pronounced among SN firms (i.e., firms switching from an industry

specialist auditor to a non-specialist auditor). NN firms also exhibit lower stock liquidity,

suggesting that market participants penalize firms that switch to non-specialist auditors

regardless of the type of auditor formerly engaged. Given that the market perceives audit

quality differences based on industry specialization (Knechel et al., 2007), these results

suggest that employing industry specialists to improve the credibility of financial statements

29

is relevant to stock liquidity. In this regard, our findings can be regarded as robust regardless

of the measure of audit quality employed.19

4.4.2 Other measures of stock liquidity

Following prior studies, we also use the price impact measure of Amihud (2002) and the

bid-ask spread as additional measures of liquidity. However, trading volume and transaction

data, both of which are essential to construct these two measures, are available only for KSE

(Korean Stock Exchange) firms and not for KOSDAQ (Korean Securities Dealers Automated

Quotation) firms. For example, the bid-ask spread can be calculated for only 5,089 firm-year

observations from the KSE, which represents about 55% of our total sample. To avoid such a

substantial loss of data, we have reported our main results without relying on these two

measures. However, when we repeat our main analyses based on these two measures for

completeness, the tenor of our main results does not change. These results, which are

reported in Table A1, give us more confidence in our conclusions.

5. Concluding remarks

19 We also use alternative measures of industry specialization such as auditors industry share based on auditees sales or assets (Balsam et al., 2003; Knechel et al., 2007). Although we find directionally consistent results, the coefficient on SN is not statistically significant (t=0.45 when we use asset-based industry leadership and t=0.21 when we use sales-based industry leadership). Rather, the coefficient of NN becomes positive and marginally significant. (t=1.86 when we use asset-based industry leadership and t=1.88 when we use sales-based industry leadership). It is possible that an auditor can increase the number of clients (thus the summation of sales or assets of the clients) by proposing a discounted price (Ettredge and Greenberg, 1990; Pearson and Trompeter, 1994). Alternatively, another auditor can charge higher audit fees and at the same time increase the market share in an industry. The latter case is possible only when the auditor provides higher-quality audit service enough to justify the higher price than the other auditors charge. In contrast, it is hard to say that the auditors in the former case are the industry expertise auditor. Thus, with caution, we argue that the industry specialist measure based on audit fee is a better proxy for the auditor industry specialist. Consistent with this argument, Craswell et al. (1995) document the existence of significant industry specialist auditor fee premium.

30

In this paper, we examine the effect of auditor switches on stock liquidity. We find that

other things being equal, firms that switch auditors are more likely to exhibit lower stock

liquidity than firms that do not switch auditors, thereby supporting the agency risk

explanation of auditor switches. We also find that the negative liquidity effect is driven by

firms switching from a high-quality auditor to a low-quality auditor.

We further examine whether auditor switches mandated by a regulatory body under the

auditor designation rule in Korea are also associated with stock liquidity. Consistent with the

notion that auditor designation improves audit quality, we find a non-negative association

between stock liquidity and a designated auditor switch. Furthermore, we document an even

positive association between them when firms with newly designated auditors have higher

audit risk. Overall, our results are consistent with the notion that the auditor designation rule

in Korea enhances audit quality and thus the credibility of financial reporting.

Our study is subject to several limitations. First, as many previous studies point out,

auditor switches often coincide with other substantial changes in the firm. It is possible that

our results reflect the joint effects of these changes and hence cannot be attributed solely to

the auditor switch. Second, if the stock market is efficient enough to anticipate the

announcement of the auditor switch, this is likely to reduce the power of our tests. Finally,

the results are based solely on data on Korean firms and therefore cannot be easily

generalized to other countries.

31

Despite these limitations, the study makes meaningful contributions to the literature. In

particular, we provide evidence regarding the economic consequences of an auditor switch in

terms of the significant capital market costs incurred around the time of the announcement. In

contrast, the results of prior studies based on stock returns have been inconclusive and mixed.

Given that regulators interest in auditor switches is due to the concern that auditor changes

are motivated by management opportunism, our findings are likely to be of interest to

regulators and policymakers. Our study is also novel in that it examines the liquidity effect of

mandatory auditor switches required by regulation. It implies that investors are concerned not

only with changes in audit quality, but also with who initiates the change. Overall, our paper

contributes to the ongoing debate on the economic costs and benefits of voluntary versus

mandatory auditor changes and the literature on stock liquidity.

32

Appendix Table A1. Regression results based on other liquidity proxies This table reports the regression results for the relation between auditor switch and stock liquidity using PRCIMPT and SPREAD. Panel A (B) regresses the stock liquidity measures on the auditor switch indicator (each auditor switch type). FACTOR5 represents the scores for single factors extracted from all five liquidity variables (LOT, ZERO, ROLL, PRCIMPT, and SPREAD) employing factor analysis. We include industry- and year-fixed effects based on the two-digit SIC code, but we do not report the coefficients. The table reports the OLS coefficient estimates and t-statistics based on robust standard errors that are clustered by firm. See Table 1 for variable definitions. (Panel A) Results based on auditor switch indicator

Dependent var. = (1)

PRCIMPT (2)

SPREAD (3)

FACTOR5

Parameter Coef. t-stat. Coef. Coef. Coef. t-stat.Intercept 12.186 9.95 -3.064 -12.39 6.190 10.93

SWITCH 0.117 2.21 0.012 0.70 0.051 1.36SIZE -0.527 -12.55 -0.101 -11.98 -0.238 -12.10LEV -0.141 -0.46 0.139 2.07 0.025 0.17

ATURN -0.179 -1.73 -0.049 -2.07 -0.008 -0.16ROA 1.271 2.81 -0.473 -4.72 -0.399 -1.96

SGROW -0.199 -2.13 -0.013 -0.51 -0.040 -0.74QUICK -0.049 -1.22 0.011 0.95 0.015 0.62

RETVAR -44.528 -11.38 3.106 2.61 -8.767 -3.67Year Yes Yes Yes

Industry Yes Yes Yes Adj.R2 0.334 0.167 0.296 #Obs. 4,852 5,089 4,147

(Panel B) Results based on auditor switch type

Dependent var. = (1)

PRCIMPT (2)

SPREAD (3)

FACTOR5

Parameter Coef. t-stat. Coef. t-stat. Coef. t-stat.Intercept 12.309 10.08 -3.064 -12.28 6.197 10.92

BN 0.230 2.14 0.082 2.23 0.182 2.42NB 0.062 0.60 -0.013 -0.39 0.010 0.13BB 0.199 2.71 0.005 0.20 0.035 0.65NN -0.224 -1.51 -0.030 -0.73 -0.021 -0.21

SIZE -0.531 -12.66 -0.101 -11.84 -0.238 -12.08LEV -0.152 -0.49 0.136 2.03 0.021 0.14

ATURN -0.179 -1.72 -0.050 -2.09 -0.010 -0.19ROA 1.286 2.85 -0.469 -4.67 -0.378 -1.87

SGROW -0.199 -2.13 -0.013 -0.50 -0.040 -0.75QUICK -0.051 -1.28 0.011 0.93 0.014 0.56

RETVAR -44.492 -11.34 3.112 2.61 -8.773 -3.67Year Yes Yes Yes

Industry Yes Yes Yes Adj.R2 0.335 0.167 0.296 #Obs. 4,852 5,089 4,147

33

References Autore, D.M., Billingsley, R.S., Schneller, M.I., 2009. Information uncertainty and

auditor reputation. Journal of Banking and Finance 33 (12), 183-192. Armstrong, C.S., Jagolinzer, A.D., Larcker, D.F., 2010. Chief executive officer equity

incentives and accounting irregularities. Journal of Accounting Research 48: 225-271.

Bacidore, J.M., Battalio, R., Galpin, N., Jenning, R., 2005. Sources of liquidity for NYSE-listed non-US stocks. Journal of Banking and Finance 29: 3075-3098.

Balsam, S., Krishnan, J., Yang, J., 2003. Auditor industry specialization and the earnings response coefficient. Auditing: A Journal of Practice & Theory 22: 71-97.

Balvers, R.J., McDonald, B., Miller, R.E., 1988. Underpricing of new issues and the choice of auditor as a signal of investment banker reputation. The Accounting Review 63: 605-622.

Bekaert, G., Harvey, R.H., Lundblad, C., 2007. Liquidity and Expected Returns: Lessons from Emerging Markets. Review of Financial Studies 20: 1783-1831.

Beneish, M.D., Hopkins, P.E., Jensen, I.P., Martin, R.D., 2005. Do auditor resignations reduce uncertainty about the quality of firms financial reporting? Journal of Accounting & Public Policy 24: 357-390.

Boone, J.P., Raman, K.K., 2001. Auditor resignations versus dismissals: an examination of the differential effects on market liquidity and trading activity. Advances in Accounting 18: 47-75.

Choi, J-H., Wong, T.J., 2007. Auditors governance functions and legal environments: an international investigation. Contemporary Accounting Research 24: 13-46.

Chordia, T., Roll, R., Subrahmanyam, A., 2000. Co-movements in bid-ask spreads and market depth. Financial Analysts Journal 56: 23-27.

Clatworthy, M., Makepeace, G., Peel, M., 2009. Selection bias and the Big Four premium: new evidence using Heckman and matching models. Accounting and Business Research 39: 139-166.

Core, J., 2010. Discussion of chief executive officer equity incentives and accounting irregularities. Journal of Accounting Research 48: 271-287.

Craswell, A., Francis, J., Taylor, S., 1995. Auditor Brand Name Reputations and Industry Specializations. Journal of Accounting and Economics 20: 297-322.

DeAngelo, L., 1981. Auditor size and audit quality. Journal of Accounting and Economics 3: 183-199.

Dillon, W.R., Goldstein, M., 1984. Multivariate analysis: methods and applications. John Wiley & Sons: New York, US.

Doidge, C., Karolyi, G., Stulz, R., 2009. Has New York become less competitive in global markets? Evaluating foreign listing choices over time. Journal of Financial Economics 91: 253-277.

Easley, D., OHara, M., 2004. Information and the cost of capital. The Journal of Finance 59: 1553-1583.

Eichenseher, J.W., Hagigi, M., Shields, D., 1989. Market reaction to auditor changes by OTC companies. Auditing: A Journal of Practice & Theory 9: 29-40.

34

Ettredge, M., Greenberg, R., 1990. Determinants of Fee Cutting on Initial Audit Engagements. Journal of Accounting Research 28: 198-210.

Fan, J.P.H., Wong, T.J., 2005. Do external auditors perform a corporate governance role in emerging markets? Evidence from East Asia. Journal of Accounting Research 43: 1-38.

Fortin, S., Pittman, J.A., 2007. The role of auditor choice in debt pricing in private firms. Contemporary Accounting Research 24: 859-896.

Francis, J.R., Khurana, I.K., Pereira, R., 2003. The role of accounting and auditing in corporate governance and the development of financial markets around the world. Asia-Pacific Journal of Accounting and Economics 10: 1-30.

Francis, J.R., Maydew, E.L., Sparks, H.C., 1999. The role of big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice & Theory 18: 17-34.

Francis, J., Lafond, R., Olsson, P., Schipper, K., 2005. The market pricing of accruals quality. Journal of Accounting and Economics 39: 295-327.

Francis, J.R., Lennox, C.S., Wang, Z., 2010. Selection models in accounting research. Unpublished Paper.

Fried, D., Schiff, A., 1981. CPA switches and associated market reaction. The Accounting Review 56: 326-341.

Frino, A., Gerace, D., Lepone, A., 2008. Liquidity in auction and specialist market structures: Evidence from the Italian bourse. Journal of Banking and Finance 32: 2581-2588.

Goyenko, R.Y., Holden, C.W., Trzcinka, C.A., 2009. Do liquidity measures measure liquidity? Journal of Financial Economics 92: 153-181.

Harris, L., 1989. A day-end transaction price anomaly. Journal of Financial and Quantitative Analysis 24: 29-45.

Harris, L., 1990. Statistical properties of the Roll serial covariance bid-ask spread estimator. Journal of Finance 45: 568-579.

Heckman, J., 1979. Sample selection bias as a specification error. Econometrica 47: 153-161.

Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: managerial behavior, agency costs, and ownership structure. Journal of Financial Economics 3: 305-360.

Jeong, S.W., Jung, K., Lee, S-J., 2005. The effect of mandatory auditor assignment and non-audit service on audit fees: evidence from Korea. The International Journal of Accounting 40: 233-248.

Johnson, W.B., Lys, T., 1990. The market for audit services. Journal of Accounting and Economics 12: 281-308.

Kanagaretnam , K., Lim, C. Y., Lobo, G.J., 2010. Auditor reputation and earnings management: International evidence from the banking industry. Journal of Banking and Finance 34: 2318-2327.

Kim, J-B., Yi. C., 2009. Does auditor designation by the regulatory authority improve audit quality? Evidence from Korea. Journal of Accounting & Public Policy 28: 207-230.

35

Knechel, W.R., Naiker, V., Pacheko, G., 2007. Does auditor industry specialization matter? Evidence from market reaction to auditor switches. Auditing: A Journal of Practice & Theory 26: 19-45.

Krishnan, J., Stephens, R.G., 1995. Evidence on opinion shopping from audit opinion conservatism. Journal of Accounting & Public Policy 14: 179-201.

Kryzanowski, L., Lazrak, S., Rakita, I., 2010. Behavior of liquidity and returns around Canadian seasoned equity offerings. Journal of Banking and Finance 34: 2954-2967.

Lambert, R., Leuz, C., Verrecchia, R.E., 2007. Accounting information, disclosure, and the cost of capital. Journal of Accounting Research 45: 385-420.

Lang, M.H., Lins, K.V., Maffett, M.G., 2011. Transparency, liquidity, and valuation: International evidence on when transparency matters most. Journal of Accounting Research, forthcoming.

Larcker, D.F., Rusticus, T.O., 2010. On the use of instrumental variables in accounting research. Journal of Accounting and Economics 49: 186-205.

Lennox, C.S., 2000. Do companies successfully engage in opinion-shopping? Evidence from the U.K. Journal of Accounting & Economics 29: 321-337.

Lesmond, D., 2005. Liquidity of emerging markets. Journal of Financial Economics 77: 411-452.

Lesmond, D., Ogden, J., Trzcinka, C., 1999. A new estimate of transaction costs. Review of Financial Studies 12: 1113-1141.

Lesmond, D., Schill, M., Zhou, C., 2004. The illusory nature of momentum profits. Journal of Financial Economics 43: 4421-1142.

Leuz, C., Verrecchia, R.E., 2000. The economic consequences of increased disclosure. Journal of Accounting Research 38: 91-124.

Levich, R., 2001. The importance of emerging capital markets. Brookings-Wharton Papers on Financial Services: 145.

Mayhew, S., Mihov, V., 2004. How do exchanges select stocks for option listing? Journal of Finance 58, 447-471.

Nichols, D.R., Smith, D.B., 1983. Auditor credibility and auditor changes. Journal of Accounting Research 21: 534-544.

Pearson, T., Trompeter, G., 1994. Competition in the Market for Audit Services: The Effect of Supplier Concentration on Audit Fees. Contemporary Accounting Research 11: 115-135.

Pittman, J.A., Fortin, S., 2004. Auditor choice and the cost of debt capital for newly public firms. Journal of Accounting and Economics 37: 113-136.

Roll, R., 1984. A simple implicit measure of the effective bid-ask spread in an efficient market. Journal of Finance 39: 1127-1139.

Sankaraguruswamy, S., Whisenant, J.S., 2004. An empirical analysis of voluntarily supplied client-auditor realignment reasons. Auditing: A Journal of Practice & Theory 23: 107-121.

Securities and Exchange Commission (SEC), 1988. Disclosure Amendments to Regulation S-K, Form 8-K and Schedule 14a Regarding Changes in Accountants and Potential Opinion Shopping Situations. Financial Reporting Release NO. 31. Washington, D.C.: Government Printing Office.

36

Sengupta, P., 1998. Corporate disclosure and the cost of debt. The Accounting Review 73: 459-474.

Shu, S.Z., 2000. Auditor resignations: clientele effects and legal liability. Journal of Accounting & Economics 29: 173-205.

Smith, D.B., 1988. An investigation of Securities and Exchange Commission regulation of auditor change disclosures: the case of accounting series release no. 165. Journal of Accounting Research 26: 134-145.

Sohn, S., Kim, Y.H., 2004. Information Content Study of the Auditor Switch. Korean Management Review 33: 685-702 (in Korean).

Watts, R., Zimmerman, J.L., 1986. Positive accounting theory. Englewood Cliffs, N.J.: Prentice-Hall, Inc.

Wells, D.W., Loudder, M.L., 1997. The market effects of auditor resignations. Auditing: A Journal of Practice & Theory 16: 138-144.

Whisenant, J.S., 2006. An analysis of the information content of auditor change announcements. The Grant Thornton Commentary Series Investor Perceptions of Audit Firm Parity 1 (1).

Whisenant, J.S., Sankaraguruswamy, S., Raghunandan, K., 2003. Market reactions to disclosure of reportable events. Auditing: A Journal of Practice & Theory 22: 181-194.

Wild, J.J, Bernstein, L.A., Subramanyam, K.R., 2001. Financial Statement Analysis, McGraw Hill Book Company.

37

Table 1 Variable definitions

Variable Description Dependent variables LOT The natural logarithm of a yearly estimate of the transaction costs

implied by the trading behavior of investors developed by Lesmond et al. (1999). The estimation period spans month -5 through month +7 so that the firms annual reports are publicly available for a few months, which should enable markets to impound the auditor switch effect.

ZERO The proportion of zero daily returns out of the maximum potential number of trading days in a given year. The measurement period spans month -5 through month +7 relative to the firms fiscal year-end.

ROLL Rolls (1984) effective spread based on the bid-ask bounce-induced negative serial correlation in returns. The measurement period spans month -5 through month +7 relative to the firms fiscal year-end.

PRCIMPT Amihuds (2002) illiquidity measure which captures the price impact of trades. We measure illiquidity as the median daily price impact over the year and compute the price impact as the daily absolute price change in percentage terms divided by trading volume (measured in million KRW). To avoid the misclassification of days with no or low trading activity, we omit zero-return days from yearly median calculations. We take the natural logarithm of the value for expositional purposes. The measurement period runs from month -5 to month +7 relative to the firms fiscal year-end.

SPREAD The bid-ask spread. We obtain the closing bid and ask prices for each day from the intraday transaction data of the Korean Stock Exchange and compute the daily quoted spread as the difference between the two prices divided by the midpoint. We then compute the median daily spread over the year. The measurement period runs from month -5 to month +7 relative to the firms fiscal year-end.

Auditor switch variables SWITCH 1 if a firm changes its auditor in the period; 0 otherwise BN 1 if a firm changes from a Big 4 auditor to a non-Big 4 auditor; 0

otherwise NB 1 if a firm changes from a non-Big 4 auditor to a Big 4 auditor; 0

otherwise BB 1 if a firm changes from one Big 4 auditor to another Big 4

auditor; 0 otherwiseNN 1 if a firm changes from one non-Big 4 to another non-Big 4

auditor; 0 otherwise Industry specialization variablesSN 1 if a firm changes its auditor from an industry specialist to a non-

specialist; 0 otherwiseNS 1 if a firm changes its auditor from a non-specialist to a specialist;

0 otherwise SS 1 if a firm changes its auditor from one specialist to another

specialist; 0 otherwiseNN 1 if a firm changes its auditor from one non-specialist to another

38

non-specialist; 0 otherwise

Other variables SIZE The natural logarithm of total assetsLEV The leverage ratio, calculated as the sum of short-term and long-

term debt divided by total assetsATURN The asset turnover ratio, calculated as sales divided by total assetsROA The return-on-assets ratio, calculated as net income divided by

total assets SGROW Sales growth, calculated as the change in sales divided by sales in

the previous yearQUICK The current ratio, calculated as current assets divided by current

liabilities RETVAR Return variability, calculated as the standard deviation of daily

returns in the year. The measurement period spans from month -5 to month +7 relative to the firms fiscal year-end.

DSG 1 if a firms auditor is designated by regulation; 0 otherwise FO Percentage of foreign ownership

39

Table 2 Descriptive statistics Panel A presents the number of observations for each auditor switch type. Panel B presents the summary statistics for the full sample. Panel C reports the Pearson correlation matrix for the key variables. The full sample comprises a maximum of 9,421 firm-year observations between 1995 and 2006 with non-missing financial data. See Table 1 for variable definitions. (Panel A) Number of observations by auditor switch type

Type Full sample Auditor designation sample (1995 to 2006) (1995 to 2002)

Designated Others Non-switch 7,445 286 3,839

Switch 1,796 297 671 BN 353 75 124 NB 347 57 149 BB 786 117 262 NN 310 48 136

Total 9,241 583 4,510 (Panel B) Full sample (N = 9,241, 1995 to 2006)

Variable Mean Std Dev Min. Q1 Median Q3 Max.SWITCH 0.194 0.396 0.000 0.000 0.000 0.000 1.000

BN 0.038 0.192 0.000 0.000 0.000 0.000 1.000 NB 0.038 0.190 0.000 0.000 0.000 0.000 1.000 BB 0.085 0.279 0.000 0.000 0.000 0.000 1.000NN 0.034 0.180 0.000 0.000 0.000 0.000 1.000 LOT -4.836 0.551 -6.123 -5.199 -4.852 -4.516 -3.175

ZERO 0.089 0.053 0.016 0.052 0.076 0.111 0.310 ROLL 0.023 0.014 0.000 0.013 0.020 0.030 0.113 SIZE 25.648 1.458 23.035 24.593 25.410 26.423 30.057 LEV 0.405 0.240 0.003 0.230 0.396 0.556 1.300

ATURN 0.951 0.514 0.140 0.620 0.855 1.162 3.249 ROA -0.015 0.168 -0.993 -0.013 0.019 0.055 0.216

SGROW 0.089 0.297 -0.655 -0.049 0.065 0.189 1.476 QUICK 1.950 1.914 0.235 0.930 1.361 2.116 12.022

RETVAR 0.039 0.014 0.015 0.028 0.038 0.049 0.074

40

(Panel C) Pearson correlation matrix

LOT ZERO ROLL SIZE LEV ATURN ROA SGROW QUICK RETVAR SWITCH 0.062 0.024 0.070 -0.007 0.047 -0.011 -0.011 0.005 -0.020 0.079

0.00 0.02 0.00 0.52 0.00 0.31 0.28 0.61 0.06 0.00 LOT 0.754 0.126 -0.194 0.145 -0.008 -0.146 -0.040 -0.064 0.114

0.00 0.00 0.00 0.00 0.42 0.00 0.00 0.00 0.00 ZERO 0.183 -0.020 0.076 0.022 0.022 0.016 -0.080 -0.372

0.00 0.05 0.00 0.04 0.03 0.12 0.00 0.00 ROLL -0.203 0.151 -0.075 -0.227 -0.039 0.000 0.650

0.00 0.00 0.00 0.00 0.00 0.99 0.00 SIZE 0.137 -0.063 0.208 0.063 -0.270 -0.294

0.00 0.00 0.00 0.00 0.00 0.00 LEV 0.064 -0.388 -0.023 -0.513 0.171

0.00 0.00 0.03 0.00 0.00 ATURN 0.105 0.150 -0.154 -0.078

0.00 0.00 0.00 0.00 ROA 0.227 0.070 -0.302

0.00 0.00 0.00 SGROW -0.054 -0.055

0.00 0.00 QUICK 0.029

0.01

41

Table 3 Regression results for the relation between auditor switch and stock liquidity This table reports the regression results for the relation between auditor switch and stock liquidity. Panel A (B) regresses the stock liquidity measures on the auditor switch indicator (each auditor switch type). FACTOR summarizes our liquidity proxies and represents the scores of a single factor extracted from liquidity variables (1) through (3) employing factor analysis. We include industry- and year-fixed effects based on the two-digit SIC code, but we do not report the coefficients. The table reports the OLS coefficient estimates and t-statistics based on robust standard errors that are clustered by firm. See Table 1 for variable definitions. (Panel A) Results based on auditor switch indicator

Dependent var. = (1)

LOT

(2)ZERO

(3)ROLL

(4) FACTOR

Parameter Coef. t-stat. Coef. t-stat. Coef. t-stat. Coef. t-stat.Intercept -1.765 -9.43 0.275 15.44 0.056 17.53 4.792 13.62

SWITCH 0.041 2.83 0.003 2.23 0.001 2.35 0.067 2.61SIZE -0.127 -19.08 -0.007 -10.56 -0.002 -15.71 -0.186 -14.93LEV 0.133 2.46 -0.011 -1.95 0.007 8.64 0.004 0.04

ATURN -0.003 -0.21 0.003 1.63 -0.001 -4.69 0.026 0.83ROA -0.618 -9.69 -0.030 -5.15 -0.009 -5.68 -0.880 -7.57

SGROW -0.061 -3.18 -0.002 -1.01 -0.001 -2.42 -0.075 -2.13QUICK -0.015 -3.31 -0.001 -2.63 0.000 1.26 -0.026 -3.10

RETVAR -3.661 -5.37 -1.037 -16.11 0.188 13.23 -14.551 -11.64Year Yes Yes Yes Yes

Industry Yes Yes Yes Yes Adj.R2 0.161 0.305 0.334 0.242 #Obs. 9,241 9,241 9,241 9,241

(Panel B) Results based on auditor switch type

Dependent var. = (1)

LOT

(2)ZERO

(3)ROLL

(4) FACTOR

Parameter Coef. t-stat. Coef. t-stat. Coef. t-stat. Coef. t-stat.Intercept -1.780 -9.48 0.274 15.40 0.056 17.38 4.771 13.54

BN 0.115 3.94 0.009 3.15 0.001 2.18 0.196 3.69NB 0.011 0.40 0.002 0.74 0.000 0.72 0.027 0.58BB 0.026 1.31 0.002 1.15 0.001 1.19 0.046 1.27NN 0.024 0.73 0.000 -0.14 0.001 1.25 0.017 0.28

SIZE -0.126 -18.95 -0.007 -10.52 -0.002 -15.55 -0.185 -14.85LEV 0.132 2.44 -0.011 -1.98 0.007 8.62 0.001 0.01

ATURN -0.003 -0.19 0.003 1.65 -0.001 -4.69 0.026 0.85ROA -0.617 -9.68 -0.030 -5.15 -0.009 -5.66 -0.880 -7.57

SGROW -0.060 -3.16 -0.002 -0.99 -0.001 -2.41 -0.074 -2.12QUICK -0.015 -3.31 -0.001 -2.63 0.000 1.27 -0.026 -3.10

RETVAR -3.654 -5.37 -1.035 -16.10 0.188 13.20 -14.528 -11.63Year Yes Yes Yes Yes

Industry Yes Yes Yes Yes adj.R2 0.218 0.305 0.334 0.242 #Obs. 9,241 9,241 9,241 9,241

42

Table 4 Regression results after controlling for the likelihood of switching auditors This table controls for the likelihood of switching auditors in the regression results for the relation between auditor switch and stock liquidity. In columns (1) and (2), we use firm-year observations for firms that switch auditors only once during our sample period. In columns (3) to (6), we first construct the propensity score as a predicted probability of auditor switch. Next, using this propensity score, we match auditor-switching observations with non-switching observations in the same year in columns (3) and (4). In columns (5) and (6), we match switching year observations for firms that switch auditors only once during the sample period with the same year observations for firms that have never changed auditors. We use FACTOR and the scores for single factors extracted from the three liquidity variables - LOT, ZERO, and ROLL - as dependent variables. We include industry- and year-fixed effects based on the two-digit SIC code, but we do not report the coefficients. The table reports the OLS coefficient estimates and t-statistics based on robust standard errors that are clustered by firm. See Table 1 for variable definitions.

Dep. Var = FACTOR

Pre-Post analysis: One-time switchers only

Switching and non-switching pairs matched by the propensity score

Switching and non-switching pairs using one-time switchers and never-

switching firms (matched by the propensity score)

(1) (2) (3) (4) (5) (6)Parameter Coef. t-stat. Coef. t-stat. Coef. t-stat. Coef. t-stat. Coef. t-stat. Coef. t-stat. Intercept 4.284 7.48 4.200 7.34 4.794 8.32 4.757 8.18 5.395 6.24 5.209 6.05

SWITCH 0.039 0.97 0.082 2.29 0.121 1.81 BN 0.275 2.95 0.220 3.68 0.370 3.35 NB -0.070 -0.88 0.028 0.54 -0.005 -0.06 BB -0.027 -0.48 0.071 1.55 0.059 0.71 NN 0.009 0.10 0.017 0.25 0.096 0.90

SIZE -0.169 -7.81 -0.166 -7.68 -0.188 -9.47 -0.187 -9.31 -0.212 -7.08 -0.205 -6.90 LEV 0.135 0.72 0.130 0.69 0.000 0.00 -0.008 -0.05 0.201 0.90 0.194 0.87

ATURN 0.002 0.04 0.006 0.12 -0.006 -0.12 -0.005 -0.11 0.002 0.03 0.007 0.12 ROA -0.829 -3.97 -0.839 -4.06 -0.662 -3.67 -0.664 -3.69 -0.686 -2.53 -0.701 -2.59

SGROW -0.080 -1.64 -0.079 -1.63 -0.096 -1.30 -0.093 -1.27 -0.157 -1.89 -0.156 -1.91 QUICK -0.039 -3.05 -0.037 -2.97 0.001 0.09 0.002 0.11 -0.024 -1.58 -0.021 -1.39

RETVAR -14.253 -7.71 -14.155 -7.71 -14.933 -6.67 -14.870 -6.65 -20.308 -6.35 -20.067 -6.27 Year Yes Yes Yes Yes Yes Yes

Industry Yes Yes Yes Yes Yes Yes Adj.R2 0.258 0.260 0.223 0.225 0.251 0.258 #Obs. 2,819 2,819 3,592 3,592 1,084 1,084

43

Table 5 The effect of auditor designation on the relation between auditor switch and stock liquidity This table reports the regression results for the auditor designation effect on the relation between auditor switch and stock liquidity. The sample comprises 5,093 firm-year observations between 1995 and 2002. The auditor designation rule is applied in 671 of these observations. DSG is an indicator that equals 1 if a firm is audited by a designated auditor and 0 otherwise. We use FACTOR and the scores for single factors extracted from the three liquidity variables - LOT, ZERO, and ROLL - as the dependent variables. We include industry- and year-fixed effects based on the two-digit SIC code, but we do not report the coefficients. The table reports the OLS coefficient estimates and t-statistics based on robust standard errors that are clustered by firm. See Table 1 for variable definitions.

Depvar.=FACTOR (1) (2) (3) (4) Parameter Coef. t-stat. Coef. t-stat. Coef. t-stat. Coef. t-stat.Intercept 2.524 4.59 2.512 4.58 2.453 4.49 2.439 4.46

SWITCH 0.085 2.65 0.092 2.28 BN 0.141 2.19 0.169 2.05NB 0.107 1.82 0.141 1.97BB 0.058 1.23 0.045 0.85NN 0.057 0.72 0.068 0.84

DSG 0.288 4.31 0.349 4.75DSG*SWITCH -0.170 -2.50

DSG*BN -0.327 -2.38DSG*NB -0.355 -2.56DSG*BB -0.203 -1.70DSG*NN -0.292 -1.54

SIZE -0.091 -4.99 -0.090 -4.98 -0.087 -4.79 -0.086 -4.75LEV -0.232 -1.69 -0.235 -1.70 -0.317 -2.32 -0.335 -2.44

ATURN 0.068 1.65 0.067 1.64 0.072 1.76 0.072 1.76ROA -0.987 -6.07 -0.987 -6.07 -0.981 -6.04 -0.970 -6.00

SGROW -0.130 -2.86 -0.129 -2.83 -0.127 -2.84 -0.126 -2.81QUICK -0.003 -0.21 -0.003 -0.22 -0.006 -0.46 -0.007 -0.48

RET