Review of Industrial Organization (2006) 29:193–212 DOI 10.1007/s11151-006-9117-5
© Springer 2006
A Basic Quantification of the Competitive Implications of the Demise of Arthur Andersen EMILIE R. FELDMAN Harvard Business School WYSS HALL 302-A, MA 02163, Boston E-mail:
[email protected]
Abstract. Andersen’s exit from the already-concentrated market for auditing services is shown to have increased market concentration and audit fees. Changes in market concentration are found to be significantly related to changes in audit fees, suggesting that the structure-performance hypothesis is applicable to the post-Andersen accounting industry and that the “Final Four” accounting firms may have exercised market power in this environment. The paper concludes with a discussion of the implications of Andersen’s exit from the market. Key words: Arthur Andersen, accounting, audit fees, market concentration, market power, structure-performance hypothesis.
I. Introduction On March 15, 2002, after months of negotiations, the Department of Justice (“DOJ”) obtained a criminal indictment against Arthur Andersen, LLP (“Andersen”) on the felony charge of obstructing justice in connection with its conduct as Enron’s auditor. Three months later, a jury convicted Andersen of that crime.1 Under the Rules of Practice of the Securities and Exchange Commission (“SEC”) (SEC, 1995), a firm with a criminal conviction would be barred from practicing before that agency. The conviction put Andersen out of business, following a few months in which major clients were defecting daily to the other accounting firms.2 1
The United States Supreme Court reversed Andersen’s criminal conviction on May 31, 2005 on procedural grounds (Supreme Court, 2005) and remanded the case to the lower courts. On November 22, 2005, the DOJ filed a motion to dismiss the criminal case against Andersen, a move that “wasn’t surprising in light of the Supreme Court ruling and Andersen’s demise as an accounting firm” (Emshwiller, 2005). 2 The attrition of Andersen’s client base was so steady that the New York Post satirized it in a daily feature called “Andersen Watch,” which juxtaposed a list of the day’s defections with a picture of Paul Volcker (the Andersen-appointed head of an Independent Oversight Board created to effect change within the firm).
194
EMILIE R. FELDMAN
Figure 1. Accounting Industry Consolidation Since 1988.
Consequently, the structure of the accounting industry went from five dominant firms to four (respectively, the “Big Five” and the “Final Four”), exacerbating a consolidation trend that began in the late 1980s (see Figure 1).3 The primary finding of this paper – a positive and signifi-
3 The accounting industry has a long history of mergers and corresponding increases in market concentration, which was examined shortly after Andersen’s demise in a statutorily-mandated report (GAO, 2003) by the then-Government Accounting Office (since renamed the Government Accountability Office and referred to herein as the “GAO”). In 1989, two mergers among Big Eight firms reduced the number of dominant players to six. Two more mergers were proposed in 1998, only one of which was consummated, resulting in the Big Five. Price Waterhouse combined with Coopers and Lybrand, forming PricewaterhouseCoopers, but the proposed merger between Ernst & Young and KPMG was quickly abandoned in the face of opposition by the European Commission, which rejected it for its potential competitive consequences “before [DOJ] staff made a recommendation” (antitrust.org, 2001). Accordingly, that merger, which would have reduced the number of major firms to four, was never officially reviewed for its domestic antitrust consequences. It was Andersen’s demise in 2002 that ultimately reduced the major players in the accounting industry from the Big Five to the Final Four. Figure 1 is a graphic depiction of the industry’s consolidation since 1989, derived from the GAO Report.
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
195
cant relationship between percent changes in market concentration and in audit fees between 2001 and 2002 – suggests that Andersen’s exit from the already-concentrated accounting market enabled the Final Four to exercise market power, in accordance with the structure-performance hypothesis. This paper proceeds as follows: Section II discusses the academic literature; Section III describes the data; Section IV motivates the analysis; Section V develops the model; Section VI presents the results; Section VII addresses the issue of auditor capacity; and Section VIII concludes.
II. Literature Industrial Organization research has found that, above some threshold, industry consolidation is expected to result in higher profitability for the remaining players in a market; this positive concentration-profitability relationship can be interpreted in one of two ways. The structure-performance hypothesis holds that an increase in concentration imparts market power to the surviving firms in an industry, allowing them to set higher prices and therefore, to attain higher profits (Bain, 1951,1956); the revisionist hypothesis, in contrast, maintains that industry consolidation creates cost efficiencies and consequently, higher profits (Demsetz, 1973; Peltzman, 1977, 1979; Scherer, 1979). Numerous Industrial Organization papers have examined the concentration-profitability hypothesis in the context of horizontal mergers (Barton and Sherman, 1984; Berger and Hannan, 1989; Borenstein, 1990; Werden et al., 1991; Kim and Singal, 1993; Prager and Hannan, 1998). The accounting industry has provided a rich setting for empirical research. Early papers examined the mergers that occurred throughout the 1980s and 1990s (Simunic, 1980; Danos and Eichenseher, 1981,1982, 1986; Palmrose, 1986; Simon and Francis, 1988; Tonge and Wootton, 1991; Maher et al., 1992; Doogar and Easley, 1998; Ivancevich and Zardkoohi, 2000; Menon and Williams, 2001; Sullivan, 2002, while recent papers have analyzed the implications of the Andersen situation (Callen and Morel, 2002; Chaney and Philipich, 2002; Krishnamurthy et al., 2002; Doogar et al., 2003; Eisenberg and Macey, 2003; Fuerman, 2003; Krishnan, 2003; Rauterkus and Song, 2003; Asthana et al., 2004; Chi, 2004; Barton, 2005; Blouin et al., 2005; Fuerman, 2005; Lai, 2005; Schloetzer, 2005). Additionally, in Section 701 of the Sarbanes-Oxley Act (SEC, 2002a), Congress mandated that the GAO analyze the increase in concentration in the accounting industry resulting from both the industry consolidation and
196
EMILIE R. FELDMAN
Andersen’s exit from the market.4 That study (GAO, 2003) was released in July 2003.5 This paper extends these streams of literature by empirically analyzing the impact of Andersen’s collapse on the relationship between market concentration and audit fees. It finds support for the structure-performance hypothesis by showing that the consolidation of the accounting industry was related to an increase in audit fees, signifying that the Final Four exercised market power. III. Data The analyses that are made in this paper are derived from an initial sample of 1,800 publicly traded companies purchased from Board Analyst, a database offered by The Corporate Library.6 Corporate proxy statements that were filed on SEC Forms 14A were then accessed from the SEC’s EDGAR database, to generate audit fee and auditor information for each of the 1,800 companies in the years 2000, 2001, and 2002. The dataset was further refined to exclude non-U.S. companies7 and companies for which audit fees were unavailable in any of the three years.8 In so doing, the sample size was reduced to 1,071 companies. Financial data was then gathered from Compustat and SDC Platinum. The major limitation of a study of audit fees, for this and any sample of companies, is that changes in audit fees after 2002 are affected by “noise” arising from the Sarbanes-Oxley Act (“SOX”). SOX, enacted in May 2002 with the purpose of regulating corporate governance and imposing stricter 4 The introduction to the GAO Report (2003) describes its purpose: “Mergers among the largest firms in the 1980s and 1990s and the dissolution of Arthur Andersen in 2002 significantly increased concentration among the largest firms, known as the ‘Big 4’ . . . This consolidation and the resulting concentration have raised a number of concerns. To address them, the Sarbanes-Oxley Act of 2002 mandated that the GAO study the factors contributing to the mergers; the implications of consolidation on competition and client choice, audit fees, audit quality, and auditor independence; the impact of consolidation on capital formation and securities markets; and barriers to entry faced by smaller accounting firms in competing with the largest firms for large public company audits.” 5 Refer to Footnotes 3, 4, 10, 17, 20, 23, 25, and 39 for further discussion of the GAO Report. 6 On its public website, Board Analyst describes its universe of companies as follows: “Our universe currently includes over 2,100 public US companies including the S&P Super- Composite 1500, Russell 1000, public companies of the Fortune 1000, and the DSI 400.” () 7 Although this paper examines the U.S. accounting industry, the international market was comparably affected by Andersen’s dissolution. White (2001) discusses the international market for accounting services. 8 For example, a company that merged in 2001 (making its audit fees unavailable in that year) would be excluded from the sample altogether.
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
197
requirements on boards of directors and on the preparation of financial statements, predictably caused an increase in audit fees to compensate the accounting industry for (a) the higher costs of satisfying the new requirements imposed by SOX and (b) assuming a higher degree of audit risk.9 Unfortunately, these SOX-impelled increases overwhelmed the increase in audit expenses following the post-Andersen increase in concentration,10 leaving a small window in which to examine the direct effect of Andersen’s exit from the market.11
IV. Motivation This section of the paper will motivate the analysis of the post-Andersen relationship between market concentration and audit fees. Subsection 1 presents descriptive statistics on concentration and audit fees in the domestic accounting market. Subsection 2 refines the concept of “markets” to determine the applicability of the structure-performance hypothesis to the industry. 1.
THE U.S. MARKET
Andersen’s indictment compromised its viability as an auditor. Clients, affiliates, and professionals began defecting in increasing numbers after the announcement of the indictment, and the trend accelerated after the first quarter of 2002,12 when “many companies wrapped up their 2001 books” (Ahlberg and Whitman, 2002). Andersen’s conviction put the firm out of business: the SEC Rules of Practice barred any “person” (including
9 Section 802 of SOX describes the punishments associated with fraudulent behavior on the part of accounting firms in their audits of companies (SEC, 2002a). Consequently, the Final Four have become increasingly careful about which companies they take as clients. According to Ernst & Young chairman James S. Turley, “Our client acceptance and reacceptance processes . . . have been re-engineered with an increased focus on determining which companies we really want as audit clients and culling out those that we do not believe have adapted to the new environment and demands on a public company . . . No auditor wants to go the way of Arthur Andersen” (Browning, 2005). 10 The GAO concurred: “Many . . . believe prices will increase further due to the implementation of the Sarbanes-Oxley requirements and related changes in the scope of certain audit services and possible changes in auditing standards. Because of these important changes and the potential for market power, it would be difficult to isolate the portion of any price increase resulting from noncompetitive behavior” (GAO, 2003, p. 35). 11 As discussed in Section VII, SOX requirements affected the number of Final Four employees, making it difficult to isolate Andersen’s impact on these firms’ capacities to produce auditing services. Refer to Footnote 36. 12 Refer to Footnote 2.
198
EMILIE R. FELDMAN
entities) with a criminal conviction from practicing before the SEC,13 and so all of the firm’s remaining clients had to find new auditors. A result of the defection of clients away from Andersen was a substantial increase in concentration in the accounting industry.14 Table I presents the yearly auditor market shares and the corresponding Herfindahl-Hirschman Index (“HHI”)15 for the U.S. market, as derived from the Public Accounting Report in Panel A and from the sample16 in Panel B. The comparability of the data between the sources indicates that the sample is representative of the domestic market. In Figure 2, the HHI levels of the accounting industry are compared to the level that attracts DOJ concerns in the advent of a proposed merger.17 Publicly traded companies also faced higher audit fees in the year immediately following the redistribution of Andersen’s market share. The GAO found that average audit fees for Standard & Poor’s 500 companies increased by 27% in 2002; it attributed this result “primarily to new requirements and changing audit practices in the wake of recent accounting scandals” (GAO, 2003, p. 35). Table II presents univariate comparisons “ . . . any person who has been convicted of a felony or a misdemeanor involving moral turpitude shall be forthwith suspended from appearing or practicing before the Commission” (SEC, 1995, D.102.e.2). 14 The high concentration was not limited to the U.S. accounting market; for example, the Final Four has “the audit work for 97.5 per cent of [the leading] 350 companies . . . [and] the audit work for the top 100 publicly-quoted businesses in Britain – the FTSE [Financial Times Stock Exchange] 100 – completely sewn up” (Flanagan, 2004). 15 HHI, a measure of market concentration, is calculated by summing the squares of the market shares of each of the firms operating in a given market. 16 Auditor market share is calculated by dividing the total audit fees paid to each auditor by the sum of the audit fees of all the companies in the sample. HHI is then calculated by summing the squares of the market shares. Auditor market share and HHI can alternatively be calculated on the basis of client revenues or number of clients audited. 17 The post-Andersen increase in concentration in the accounting industry did not occur via horizontal merger and so would not have been evaluated under applicable DOJ policies, as expressed in the DOJ’s 1992 Horizontal Merger Guidelines (a longstanding policy tool used to assess the competitive implications of proposed mergers) (DOJ, 1992). However, like the GAO Report, this paper assumes that the DOJ’s Merger Guidelines, while legally inapplicable, provide policy guidance as to levels of unacceptable market concentration. Figure 2 (reproduced from the GAO Report) juxtaposes (a) the HHI for the domestic auditing market from 1988 to 2002 with (b) the DOJ’s notion of acceptable levels of market concentration, suggesting that (c) the redistribution of Andersen’s market share among the Final Four would have been deemed by the DOJ to have high potential for the exercise of market power had it occurred through (d) a transactional acquisition of Andersen’s business. Similarly, the magnitudes of the HHIs computed in Table I would have also contravened the DOJ’s policy, as expressed in the Guidelines, thereby confirming the GAO Report’s suggestion that the increase in concentration resulting from the redistribution of Andersen’s market share among the Final Four was a problem for public policy. 13
199
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
Table I. Concentration in the U.S. Market for Auditing Services Firm
Panel A: Public Accounting Report Market Share (%) 2000 2001 2002
Panel B: Sample Data Market Share (%) 2000 2001 2002
Arthur Andersen Deloitte & Touche Ernst & Young KPMG Peat Marwick PricewaterhouseCoopers BDO Seidman Grant Thornton Other Total HHI
13.00 21.09 15.43 17.06 29.97 1.49 1.50 0.46 100.00 2,046
18.42 18.56 18.48 14.70 29.27 0.25 0.19 0.14 100.00 2,098
15.86 22.61 16.54 11.70 29.72 1.55 1.40 0.62 100.00 2,061
0.00 31.34 23.98 17.00 22.61 1.87 2.12 1.08 100.00 2,366
14.88 19.55 18.93 14.96 31.12 0.24 0.20 0.13 100.00 2,154
0.00 22.63 24.76 18.71 33.33 0.28 0.20 0.10 100.00 2,586
Totals exceed 100% due to rounding. Panel A Sources: PAR 02/28/01, 02/28/02, 02/28/03 Panel B Source: Board Analyst dataset.
Figure 2. U.S. Accounting Industry HHI vs. Boundaries of DOJ Merger Guidelines
of yearly average audit fees for the sample; similar to the GAO finding, average fees in the sample increased by over 32% in 2002. 2.
MARKET DEFINITION
The appropriate definition of a market in the accounting industry is the provision of auditing services to companies operating within specific
200
EMILIE R. FELDMAN
Table II. Univariate Comparison of Audit Fees Summary Statistics
Mean ($) Standard Deviation ($) Observations Null Hypothesis 2001 Fee = 2002 Fee
Audit Fees ($) 2001
2002
1,348,781 2,473,093 1,071 t-statistic 3.50
1,785,704 3,256,293 1,071 p-value 0.00***
Significant at *** 1%, **5%, *10% Source: Board Analyst dataset.
industrial segments, as the high entry barriers and switching costs in such markets facilitate the exercise of market power. This is consistent with the Merger Guidelines’18 definition of “economically meaningful markets – i.e., markets that could be subject to the exercise of market power” as “a product or group of products and a geographic area in which it is produced or sold such that a hypothetical profit-maximizing firm . . . that was the only present and future producer or seller of those products in that area likely would impose at least a ‘small but significant and nontransitory’ increase in price” (DOJ, 1992). Entry barriers into accounting markets are high due to the dual impacts of reputation and specialization: the major players in each market rely on their reputations and the expertise of their personnel to attract and retain clients,19 and it is expensive (especially for non-Final Four firms20) to 18 Again, while the Merger Guidelines would be inapplicable to the Andersen-impelled consolidation of the accounting industry, they are relevant because Andersen’s dissolution had the same effect on the industry as if its share had been purchased by its Final Four competitors. Accordingly, the Guidelines are useful in understanding how such a consolidation would have been evaluated had it been subject to antitrust scrutiny. Refer to Werden (1993, 2000), Kwoka and White (1999, ch.4), and Federal Trade Commission v. Staples, Inc. and Office Depot, Inc. (FTC, 1997) for further discussion of the Guidelines. 19 An example of a market in which reputation and specialization are crucial is the auditing of casinos. Auditing casinos requires much specialized knowledge: The volume of cash handled by casinos is large, which necessitates the formulation, imposition, and administration of very strict financial controls and unique accounting standards and practices, with which auditors must be comfortable. Reputation is also critical in this market, evidenced by the prompt response of the casino industry to Andersen’s indictment: The New Jersey attorney general “asked the Casino Control Commission to bar Andersen from auditing, consulting, or otherwise working with the state’s casinos based on the firm’s federal indictment . . . ” Commission chairman James R. Hurley justified the decision: “Simply stated, we can harbor no reservations as to the propriety of an audit” (Binkley, 2002). 20 The GAO confirmed this point: “Although firms of all sizes [have] some difficulty attracting staff with specialized audit or industry-specific expertise, smaller firms [find]
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
201
acquire personnel with the requisite industry knowledge and experience to become a specialized auditor. Due to the presence of entry barriers, auditor choice is quite limited for clients operating within a given industrial segment, thereby raising the potential for a profitable exercise of market power. Andersen’s dissolution further increased this potential by reducing clients’ choices of viable auditors down to two or even one in alreadyconcentrated markets and/or in markets where Andersen was a dominant player.21 For a number of reasons, switching costs are substantial in the accounting industry,22 making clients reluctant to change auditors. In equilibrium, as long as switching costs exceed any increase in audit fees, it is in a client’s best interest to remain with its current auditor.23 However, when Andersen went out of business, the tradeoff between incurring switching costs and Footnote 20 Continued this particularly difficult. [They] have difficulty keeping talented employees, especially those with sought-after expertise, from leaving for jobs with the Big Four” (GAO, 2003, p. 47). 21 This situation occurred in the mortgage industry, when Fannie Mae dismissed KPMG. Fannie Mae’s choices for a replacement auditor were quite limited: Ernst & Young was advising the company’s audit committee, Deloitte & Touche was providing advisory services to Fannie Mae’s regulator, and PricewaterhouseCoopers served as a competitor’s (Freddie Mac’s) auditor (Weil, 2004). This was also confirmed in the British accounting market: “You may find that instead of having a choice of a Big Four if you want to rotate your auditor, that can come down to just two if one rival is doing the audit for a major corporate competitor and another does not provide a specialist service” (Flanagan, 2004). 22 First, there is the expectation of reciprocal loyalty in longstanding auditor-client relationships; second, continuity of audit representation creates an orderly environment in which to conduct business; third, the audit-client must report both the fact of and the reason for the change in auditors to the public in a potentially embarrassing SEC filing; fourth, there are professional relationships that have developed between the client’s internal audit personnel and the accounting firm’s audit team; fifth, there is a fixed cost incurred by both the new auditor and the client that is associated with the “breaking-in” of a new audit team that is unfamiliar with a company; and, sixth, since financial statements must be reported on a multi-year comparative basis, a change in auditors requires the full cooperation of the terminated firm. This sixth point needs elaboration. While the terminated firm will undoubtedly conduct itself in a professional manner, in the crush of the audit season, (i) assisting in the audit of a terminated client will have a lower priority than performing auditing services for continuing clients, and (ii) the cost of that final work may be quite expensive. The exiting firm even has last minute leverage over its former client in that the firm’s written permission is required in order for the client to publish necessary audit data for comparative year(s) (SEC, 1972, 2-02.a,e). The SEC was forced to address the issue of obtaining consents following Andersen’s demise in an amendment to the Rules of Practice, in which it waived the requirement that Andersen approve the use of its audit results for clients that moved to other auditors SEC (2002b). 23 The GAO confirmed this point: “if switching costs are prohibitively expensive . . . companies will not switch auditors and price competition will have no impact on the Big 4’s market share” (GAO, 2003, p. 60).
202
EMILIE R. FELDMAN
paying higher audit fees became irrelevant for its clients because they had no choice but to change auditors. Particularly in markets where a small number of Final Four firms operated, the new auditors of former Andersen clients could raise audit fees above the market-clearing level without competitive repercussions because those clients had few alternative options.24 Thus, while accounting firms individually hold much market power over their clients because of switching costs, the potential for an exercise of market power is greater in markets with fewer viable auditor choices. Having defined markets as the provision of auditing services to clients operating in a given industrial segment, we can now analyze the relationship between auditor concentration in these markets and audit fees, to determine the extent to which the structure-performance hypothesis is applicable to the post-Andersen accounting industry.25 For the purposes of this paper, surrogates for specific industrial segments were achieved through the use of 4-Digit SIC codes.26 V. Model The model that is used to explore the relationship between changes in market concentration and audit fees is as follows. Summary statistics and descriptions of the variables appear in Table III. %fee = β0 + β1 [%HHI] + β2 [moderate] + β3 [AA switch] +β4 [FF switch] + β5 [ln(TA)] + β6 [loss] +β7 [leverage] + β8 [restatement] + β9 [invrec] + β10 [MA] +β11 [financial] + β(12−17) [auditor] + ε The primary variable of interest is %H H I , the percentage change in market concentration between 2001 and 2002. This variable is derived 24
This would also have occurred in markets where one or more of the Final Four were conflicted from representing the company seeking a new auditor. Refer to Footnote 21. 25 The GAO alluded to the possibility that the structure-performance hypothesis might be applicable to the accounting industry: “While evidence to date does not appear to indicate that competition in the market for audit services has been impaired, the increased degree of concentration coupled with the recently imposed restrictions on the provision of non-audit services by incumbent auditors to their audit clients could increase the potential for collusive behavior or the exercise of market power” (GAO, 2003, pp. 25–26). 26 The choice of 4-Digit SIC codes, instead of a broader definition of industrial segment (such as 3-Digit SIC codes), was driven by the Guidelines’ definition of a relevant market: “a group of products and a geographic area that is no bigger than necessary to satisfy [the above-described] test [of that which constitutes a market]” (DOJ, 1992). It should be noted, however, that the results of the model derived using 3-Digit SIC codes as markets are functionally identical to the results derived using 4-Digit SIC codes. Refer to Section VI and especially Footnote 31.
%fee %HHI moderate AA switch FF switch ln(TA) loss leverage restatement invrec MA financial auditor
Variable
0.3630 0.0956 283.0608 0.1849 0.0103 7.8250 0.2586 0.2388 0.0177 0.2750 0.3365 0.1699 –
Mean 0.6294 0.2000 957.8055 0.3884 0.1009 1.7754 0.4381 0.1850 0.1321 0.1998 0.4727 0.3758 –
Std. Dev. 1,071 1,071 1,071 1,071 1,071 1,061 1,071 1,060 1,071 1,040 1,067 1,071 1,071
Obs. percent change in audit fees percent change in HHI of 4-Digit SIC code market HHI increase in highly-concentrated markets 1 if firm switched from Andersen 1 if firm switched from other Final Four ln(total assets) 1 if net income < 0 debt/total assets 1 if firm filed restatement with SEC (inventory + accounts receivable)/total assets 1 if firm engaged in M&A transaction 1 if firm is in financial services SIC code (60–64) auditor indicator variable
Description
Table III. Summary Statistics and Variable Descriptions
2001—2002 2001–2002 2001–2002 2001–2002 2001–2002 2002 2002 2002 2002 2002 2002 2002 2002
Time Period
price change concentration change moderating variable Andersen switchers other auditor switchers size risk risk risk risk, complexity complexity complexity auditor
Measure
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
203
204
EMILIE R. FELDMAN
from a calculation of auditor market shares and HHIs in both 2001 and 2002 in each market defined by a 4-Digit SIC code.27 If the structure-performance hypothesis is applicable to the post-Andersen accounting industry, β1 should be positive: the change in concentration in a given market between 2001 and 2002 should be positively related to the change in a client’s audit fees between 2001 and 2002 (%f ee) if the Final Four exercised market power following the Andersen-impelled consolidation of the industry. Moderate is derived by multiplying the %H H I variable by the 2001 HHI of each market. The underlying logic behind this variable is that clients in markets in which (a) concentration was already high; and (b) concentration increased substantially after Andersen went out of business; may have faced relatively larger increases in their audit fees. If this hypothesis holds, β2 is expected to be positive. AA switch and FF switch are indicator variables taking the value “1” for clients that changed auditors between 2001 and 2002 from Andersen and from a Final Four firm, respectively. These variables are included to measure the impact on audit fees of being forced to change auditors (for former Andersen clients) as compared with choosing to change auditors (for clients that switched within the Final Four). Consequently, β3 should be positive: Audit fees of former Andersen clients probably increased because their new auditors could charge higher fees without competitive repercussions. β4 should either be negative or small and positive, since a new auditor would presumably have to offer a better value proposition to induce a client to choose to switch auditors.28 The remaining variables in the model are controls that have been shown in previous empirical work to affect audit fees. ln(TA) represents client size; loss indicates whether a client incurred a net loss; leverage measures the level of the client’s indebtedness; restatement indicates whether a client was forced to restate an SEC filing; invrec is the ratio of a client’s inventory and accounts receivable relative to its total assets and measures the complexity of the audit; MA indicates whether the firm underwent an M&A transaction; financial indicates whether the client is a financial services firm, as determined by its SIC code; and auditor is a series of indicator variables for the client’s auditor.29 With the exception of auditor, the coefficients on these variables are all expected to be positive; auditing larger, risker, and/or 27 Analogously to the domestic “market” discussed in Section IV.1, market share can also be measured at the 4-Digit SIC code level by dividing each auditor’s fees by the total audit fees in that market. Market-level HHI can then be calculated by summing the squares of the market-level auditor shares. 28 Refer to the discussion of switching costs in Section IV.2. 29 Table III presents greater detail on the derivations of these variables and the influence that they are intended to measure.
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
205
more complex clients is more work for an auditor, so these clients should face larger yearly increases in their audit fees than others. VI. Results The model presented in Section V was fitted using a least squares regression,30 and its results are presented in Table IV. As predicted, the coefficient on %H H I was positive and statistically significant at the 5% level, suggesting that the structure-performance hypothesis is applicable to the post-Andersen accounting industry and the Final Four firms did exercise market power at the expense of their clients.31 Contrary to its hypothesis, moderate is negative and statistically significant at the 10% level; fees actually decreased for clients in markets that (a) were initially highly concentrated, and (b) faced large increases in concentration. However, the magnitude of the coefficient on moderate (−0.00) compared to that on %H H I (0.50) indicates that moderate had a relatively small impact on %f ee. While the coefficient on AA switch was positive and the coefficient on FF switch was negative, both as predicted, neither coefficient was statistically significant. A test of the null hypothesis that the two coefficients are identical cannot be rejected.32 Following the hypotheses in Section V, the coefficients on ln(TA), loss, leverage, and MA were positive, and the coefficients on ln(TA) and MA were significant at the 1% level. Contrary to prediction, however, the coefficients on restatement, invrec, and financial were negative, with invrec statistically significant at the 1% level.33 Although the overall explanatory power 30
Industry control variables are not included in the model because they should be captured by the first-differences methodology. To confirm this hypothesis, a likelihood ratio test is used to test the null hypothesis that the industry indicator variables are jointly equal to zero. The LR test statistic is 72.21, and the critical value for X2 (59) is 0.1159. The null hypothesis cannot be rejected, so it is appropriate to exclude the industry controls from the model. 31 Following the discussion in Footnote 26, these results were derived by calculating market shares and HHIs within markets defined by 4-Digit SIC codes. The results of the model are essentially identical if markets are instead defined by 3-Digit SIC codes. 32 F(1, 1,017) = 0.42 and p > F = 0.5187. 33 Some additional control variables were tested in the model. A variable representing the number of each client’s business segments was tested as a measure of complexity. This variable was excluded because its coefficient was statistically insignificant and its inclusion substantially limited the sample size. The natural logarithm of each client’s revenues was also tested as an alternative measure of size. While the coefficient and significance of this variable were similar to those of ln(TA), the latter was used because it is the standard measure of client size in the literature. Finally, a variable for each client’s EBIT was tested. The coefficient on this variable was negative and insignificant; it was excluded without any change to the other coefficients or to the overall significance of the model.
206
EMILIE R. FELDMAN
Table IV. OLS Regression Results Dependent Variable: Exogenous Variables+
%fee Coefficient++
%HHI
0.4998** (0.2282) −0.0001* (0.0000) 0.0332 (0.0502) (0.0907) (0.1881) 0.0408*** (0.0130) 0.0675 (0.0452) 0.0457 (0.1134) (0.1495) (0.1440) −0.2862*** (0.1064) 0.1628*** (0.0408) (0.0576) (0.0638) 0.0643 (0.2915) 0.0595 1,035
moderate AA switch FF switch ln(TA) Loss Leverage Restatement Invrec MA Financial Constant R2 Observations +
Model includes auditor indicator variables. Standard errors in parentheses. Significant at ***1%, **5%, *10% ++
of the regression is not great, the structure of the regression – cross-section first-differences – would be expected to yield low levels of explanatory power. VII. Capacity The positive and statistically significant coefficient on %H H I described in Section VI reveals that the post-Andersen increase in market concen-
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
207
tration was related to an increase in clients’ audit fees. However, the ultimate effect of the Andersen-impelled increase in market concentration on audit fees also depends on Andersen-related changes in capacity among the Final Four; a capacity increase among the Final Four would be expected to put downwards pressure on audit fees, potentially offsetting the positive relationship between concentration and fees. Table V presents annual numbers of audit partners and professionals of the Big Five and Final Four firms.34 The short-term impact of Andersen’s collapse was a decrease in auditor capacity: the increases in total Final Four partners and professionals between 2001 and 2002 were smaller than the declines in the numbers of Andersen partners and professionals. This decrease in auditor capacity is probably related to the increase in audit fees between 2001 and 2002.35 Longer-term, the 2001–2003 and 2001–2004 increases in Final Four partners exceed the decline in Andersen partners, and the increases in Final Four professionals approach the decline in Andersen professionals. The long-run trend towards increased Final Four capacity will probably reduce audit fees, though it should again be emphasized that any changes in the accounting industry after 2002 cannot be attributed solely to Andersen, due to SOX’s impact.36 VIII. Conclusion Against a backdrop of corporate governance scandals, the DOJ’s punishment of Andersen for its prominent role in the Enron matter took the form of a criminal indictment.37 That the government obtained a conviction initially validated its approach.38 However, at the time that 34 Annual data on the numbers of partners and professionals (the sum of the numbers of partners and non-partner employees), as well as the percentage of each firm’s practice in accounting and audit services, were available from the Public Accounting Report. The numbers of audit partners and professionals were derived by multiplying the total numbers of partners and professionals by the percentage of each firm’s practice in auditing services. These data represent the firms’ capacities to produce since labor is the primary input to the provision of auditing services. 35 Because only aggregate numbers of employees were available (as opposed to numbers of employees working with each client in the sample), it was not possible to quantify the impact of capacity on fees using regression analysis. 36 Refer to the discussion of the impact of SOX on the accounting industry in Section III and especially Footnote 11. 37 In Michael Chertoff’s words: “Our concern is not with honest mistakes, but with those that are deliberately manipulating or disseminating false information. Public confidence is promoted when people see that we are willing to prosecute [these cases]” (Sommar, 2002). At the time, Chertoff was the head of the Criminal Division of the DOJ. 38 Refer to Footnote 1.
7,072 8,840 7,011 12,710 35,633 8,133
7,509 9,167 7,384 12,683 36,743 8,980
8,083 10,146 7,876 11,469 37,574 0
9,009 9,920 8,811 11,564 39,304 0
2003
Numbers of Professionals+ 2000 2001 2002
2003 1,019 1,240 1,087 1,204 4,550 0
753 1,122 912 974 3,762 697
2002 942 1,250 967 1,236 4,396 0
668 1,109 713 922 3,413 565
Numbers of Partners 2000 2001
Includes partners Source: Public Accounting Report, various issues
+
Deloitte & Touche Ernst & Young KPMG Peat Marwick PricewaterhouseCoopers Total, Final Four Arthur Andersen
Auditor
Deloitte & Touche Ernst & Young KPMG Peat Marwick PricewaterhouseCoopers Total, Final Four Arthur Andersen
Auditor
Table V. Auditor Capacity
9,136 12,388 9,685 12,237 43,446 0
2004
1,027 1,340 1,141 1,229 4,738 0
2004 266 118 175 230 788 −697
274 218 229 255 976 −697
2001–2004
574 979 493 −1,215 831 −8,980
1,500 753 1,428 −1,120 2,561 −8,980
1,627 3,221 2,301 −446 6,703 −8,980
Changes in Numbers of Professionals 2001–2002 2001–2003 2001–2004
189 128 55 262 634 −697
Changes in Numbers of Partners 2001–2002 2001–2003
208 EMILIE R. FELDMAN
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
209
the DOJ decided to pursue a criminal prosecution strategy, a competing DOJ policy was overridden: maintaining competition, in this case, in the already-concentrated accounting industry.39 The consequences of Andersen’s prosecution were foreseeable: The SEC’s Rules of Practice always provided that the criminal conviction of an accounting firm would bar it from auditing public companies, its core business. Furthermore, in a concentrated market characterized by high entry barriers and switching costs, it was also predictable that Andersen’s collapse would cause practically all of its major clients to migrate to the Final Four, which would further increase market concentration. Then, in accordance with the structure-performance hypothesis, the Final Four would be expected to capitalize on their newly-augmented market power by adjusting prices. This paper has shown that to be the case through its finding of concomitant increases in market concentration and audit fees. Acknowledgements I am extremely grateful to Professor Richard Caves for his extensive and invaluable support throughout the iterations of this paper. I also sincerely thank Professor Felix Oberholzer-Gee for his assistance with the statistical model; Erika McCaffrey for her help in collecting data from Compustat and SDC Platinum; the Economics Department at Harvard University for having generously funded the purchase of the basic dataset; and partners from various accounting firms, Andersen alumni, and the executives of several public companies who provided me with their insights into the real-world consequences of Andersen’s collapse. There are many different lenses through which one can view the circumstances and consequences of Andersen’s demise, and in earlier drafts of this paper I used most of them. But for the patient and productive comments of Professor Lawrence White and two anonymous referees, the paper would not have come into being in its present form. References (2001) Cases Related to Merger Policy. http://www.antitrust.org/cases/mergers.htm# accounting. (2001–2005) Public Accounting Report. Atlanta: Strafford Publications, Inc. Ahlberg, E., and J. Whitman (2002) ‘Bankruptcy Looms Large Again at Andersen’,The Dow Jones Newswires Asthana, S., S. Balsam, and S. Kim (2004) ‘The Effect of Enron, Andersen, and Sarbanes-Oxley on the Market for Audit Services’, SSRN Working Paper. 39
According to the GAO, “It is unclear whether and to what extent the Antitrust Division was consulted and to what extent the DOJ’s Antitrust Division had input into the decision to criminally indict Andersen” (GAO, 2003, p. 19).
210
EMILIE R. FELDMAN
Bain, J. S. (1951) ‘Relation of Profit Rate to Industry Concentration: American Manufacturing, 1936–1940’, The Quarterly Journal of Economics, 65(3), 293–324. Bain, J. S. (1956) Barriers to New Competition. Cambridge, MA: Harvard University Press. Barton, D. M., and R. Sherman (1984) ‘The Price and Profit Effects of Horizontal Merger: A Case Study’, The Journal of Industrial Economics, 33(2), 165–177. Barton, J. (2005) ‘Who Cares About Auditor Reputation?’ Contemporary Accounting Research, 22(3), 549–586. Berger, A. N., and T. H. Hannan (1989) ‘The Price-Concentration Relationship in Banking’, The Review of Economics and Statistics, 71(2), 291–299. Binkley, C. (2002) ‘New Jersey Regulators Order Andersen to Halt Its Casino Business in the State’, Wall Street Journal. Blouin, J., B. Grein, and B. Rountree (2005) ‘The Ultimate Form of Mandatory Auditor Rotation: The Case of Former Arthur Andersen Clients’, SSRN Working Paper. Borenstein, S. (1990) ‘Airline Mergers, Airport Dominance, and Market Power’, The American Economic Review, 80(2), 400–404. Browning, L. (2005) ‘Sorry, the Auditor Said, but We Want a Divorce’, New York Times. Callen, J. L., and M. Morel (2002) ‘The Enron-Andersen Debacle: Do Equity Markets React to Auditor Reputation’? SSRN Working Paper. Chaney, P. K., and K. L. Philipich (2002) ‘Shredded Reputation: The Cost of Audit Failure’, Journal of Accounting Research, 40(4), 1221–1245. Chi, W. (2004) ‘The Effects of the Enron-Andersen Affair on Audit Pricing’, SSRN Working Paper. Danos, P., and J. W. Eichenseher (1981) ‘The Analysis of Industry-Specific Auditor Concentration: Towards an Explanatory Model’, The Accounting Review, 56(3), 479–492. Danos, P., and J. W. Eichenseher (1982) ‘Audit Industry Dynamics: Factors Affecting Change in Client-Industry Market Shares’, Journal of Accounting Research, 20(2), 604–616. Danos, P., and J. W. Eichenseher (1986) ‘Long-Term Trends Toward Seller Concentration in the U.S. Audit Market’, The Accounting Review, 61(4), 633–650. Demsetz, H. (1973) ‘Industry Structure, Market Rivalry, and Public Policy’, Journal of Law and Economics, 16(1), 1–9. Department of Justice (1992) 1992 Horizontal Merger Guidelines. Washington, DC: Department of Justice. Doogar, R., and R. F. Easley (1998) ‘Concentration Without Differentiation: A New Look at the Determinants of Audit Market Concentration’, Journal of Accounting and Economics, 25, 235–253. Doogar, R., T. Sougiannis, and H. Xie (2003) ‘The Impairment of Auditor Credibility: Stock Market Evidence from the Enron-Andersen Saga’, SSRN Working Paper. Eisenberg, T., and J. R. Macey (2003) ‘Was Arthur Andersen Different? An Empirical Examination of Major Accounting Firm Audits of Large Clients’, Journal of Empirical Legal Studies, 1(2), 263–300. Emshwiller, J. R. (2005) ‘Moving the Market: Andersen Figure Files to Withdraw His Guilty Plea’, Wall Street Journal. Flanagan, M. (2004) ‘Audit Market Held by Big Four’, The Scotsman. Federal Trade Commission (1997) Federal Trade Commission v. Staples, Inc. and Office Depot, Inc., 970 F. Supp. 1066. Washington, DC: Federal Trade Commission.
A BASIC QUANTIFICATION OF THE COMPETITIVE IMPLICATIONS
211
Fuerman, R. D. (2003) ‘Audit Quality Examined One Large CPA Firm at a Time: Mid-1990’s Empirical Evidence of a Precursor of Arthur Andersen’s Collapse’, SSRN Working Paper. Fuerman, R. D. (2005) ‘Differentiating Between Arthur Andersen and the Surviving Big Four on the Basis of Auditor Quality: An Empirical Investigation of the Decision to Criminally Prosecute Arthur Andersen’, SSRN Working Paper. Government Accounting Office (2003) Public Accounting Firms: Mandated Study on Consolidation and Competition. Washington, DC: Government Accounting Office. Ivancevich, S. H., and A. Zardkoohi (2000) ‘An Exploratory Analysis of the 1989 Accounting Firm Megamergers’, Accounting Horizons, 14(4), 389–401. Kim, E. H., and V. Singal (1993) Mergers and Market Power: Evidence from the Airline Industry, The American Economic Review, 83(3), 549–569. Krishnamurthy, S., J. Zhou, and N. Zhou (2002) ‘Auditor Reputation, Auditor Independence, and the Stock Market Reaction to Andersen’s Clients’, SSRN Working Paper. Krishnan, G. V. (2003) ‘Did Houston Clients of Arthur Andersen Recognize Publicly Available Bad News in a Timely Fashion’? SSRN Working Paper. Kwoka J. E. Jr., and L. J. White (1999) The Antitrust Revolution: Economics, Competition, and Policy, 3rd ed., New York: Oxford University Press. Lai, K.-W. (2005) ‘Do Succeeding Auditors to Arthur Andersen After the Financial Scandals Report More Conservatively’?, SSRN Working Paper. Maher, M. W., P. Tiessen, R. Colson, and A. J. Broman (1992) ‘Competition and Audit Fees’, The Accounting Review, 67(1), 199–211. Menon, K., and D. D. Williams (2001) ‘Long-Term Trends in Audit Fees’, Auditing: A Journal of Practice and Theory, 20(1), 115–136. Palmrose, Z.-V. (1986) ‘Audit Fees and Auditor Size: Further Evidence’, Journal of Accounting Research, 24(1), 97–110. Peltzman, S. (1977) ‘The Gains and Losses From Industrial Concentration’, Journal of Law and Economics, 20(2), 229–263. Peltzman, S. (1979) ‘The Causes and Consequences of Rising Industrial Concentration: A Reply’, Journal of Law and Economics, 22(1), 209–211. Prager, R. A., and T. H. Hannan (1998) ‘Do Substantial Horizontal Mergers Generate Significant Price Effects? Evidence from the Banking Industry’, The Journal of Industrial Economics, 46(4), 433–452. Rauterkus, S. Y., and K. Song (2003) ‘Auditor’s Reputation, Equity Offerings, and Firm Size: The Case of Arthur Andersen’, SSRN Working Paper. Scherer, F. M. (1979) ‘The Causes and Consequences of Rising Industrial Concentration’, Journal of Law and Economics, 22(1), 191–208. Schloetzer, J. D. (2005) ‘Auditor Switching After Andersen and SOX: An Oligopoly Model and Empirical Analysis’, SSRN Working Paper. Securities and Exchange Commission (1972) Regulation S-X.: Title 17, CFR 210. Washington, DC: Securities and Exchange Commission. Securities and Exchange Commission (1995) Rules of Practice: Title 17, CFR 201. Washington, DC: Securities and Exchange Commission. Securities and Exchange Commission (2002a) Sarbanes-Oxley Act of 2002, Public Law 107–204, 116 Stat. 745. Washington, DC: Securities and Exchange Commission. Securities and Exchange Commission (2002b) Temporary Final Rule and Final Rule: Requirements for Arthur Andersen LLP Auditing Clients, Release No. 33–8070 [Amendment to (SEC, 1972)]. Washington, DC: Securities and Exchange Commission. Simon, D. T., and J. R. Francis (1988) ‘The Effects of Auditor Change on Audit Fees: Tests of Price Cutting and Price Recovery’, The Accounting Review, 63(2), 255–269.
212
EMILIE R. FELDMAN
Simunic, D. A. (1980) ‘The Pricing of Audit Services: Theory and Evidence’, Journal of Accounting Research, 18(1), 161–190. Sommar, J. (2002) ‘Chertoff Out for Bad Guys’ Blood’, New York Post. Sullivan, M. W. (2002) ‘The Effect of the Big Eight Accounting Firm Mergers on the Market for Audit Services’, Journal of Law and Economics, 45 (2), 375–399. Supreme Court of the United States (2005) Arthur Andersen LLP v. United States, 04–368. Washington, DC: Supreme Court of the United States. Tonge, S. D., and C. W. Wootton (1991) ‘Auditor Concentration and Competition Among the Large Public Accounting Firms: Post-Merger Status and Future Implications’, Journal of Accounting and Public Policy, 10(2), 157–172. Weil, J. (2004) ‘Fannie’s Dismissal of KPMG Shows Dwindling Choices Among Big Four’, Wall Street Journal. Werden, G. J., A. S. Joskow, and R. L. Johnson (1991) ‘The Effects of Mergers on Price and Output: Two Case Studies from the Airline Industry’, Managerial and Decision Economics, 12(5), 341–352. Werden, G. J. (1993) ‘Market Delineation under the Merger Guidelines: A Tenth Anniversary Retrospective’, The Antitrust Bulletin, Fall 1993, 517–555. Werden, G. J. (2000) ‘Market Delineation under the Merger Guidelines: Monopoly Cases and Alternative Approaches’, Review of Industrial Organization, 16(2), 211–218. White, L. J. (2001) Reducing the Barriers to International Trade in Accounting Services. Washington, DC: The AEI Press.