Impression Management in Public Sector Audit Proposals: Language and Fees

Abstract

We adapt impression management theory to investigate: (1) whether governmental audit proposals include the values of ‘professionalism’ (i.e. independence and competence) and ‘commercialism’ (i.e. client cooperation, auditor trustworthiness, client satisfaction, commercial interests), and, (2) the relations of audit proposal language and contextual influences on proposed audit fees and proposal success. The results suggest that proposals accord with audit firms’ market strategies. Compared to other firms, proposals from small, non-specialist firms use more commercial language, while proposals from governmental specialist firms use more competence language. The results also suggest that the market for US public sector audits includes many low cost, low quality firms; specifically, the modal proposal (41.2%) includes high use of commercialism language and low proposed audit fees. We also find that audit firms propose higher audit fees in US states with a weak governmental internal auditing office, and, that greater use of commercial language correlates with higher rates of proposal success, but only in US states with weak governmental internal auditing functions. The finding of inter-relations between proposal language, the strength of governmental internal audit functions, and proposed fees suggests the value of considering multiple influences on the quality of the public sector financial controls.

Acronyms : American Institute of Certified Public Accountants (AICPA) ; Certified Public Accountant (CPA) ; Corporate social responsibility (CSR) ; European Court of Auditors (ECA) ; European Union (EU) ; fixed effects (FE) ; General Accountability Office (GAO) ; Governmental Financial Officers Association (GFOA) ; Independent Regulatory Board for Auditors (IRBA) ; Institute of Internal Auditors (IIA) ; International Accounting Education Standards Board (IAESB) ; International Federation of Accountants (IFAC) ; International Organization of Supreme Audit Institutions (INTOSAI) ; National Association of State Auditors ; Comptrollers and Treasurers (NASACT) ; not-for-profit (NFP) entities ; natural language processing (NLP) ; Office of Management and Budget (OMB) ; Request for proposal (RFP) ; Two-stage least squares (2SLS) ; United Kingdom (UK) ; United States (US)

Keywords:

• Audit proposals
• Impression management
• Audit firm strategy
• Professionalism
• Commercialism

Disclosure statement

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

Supplemental data

Supplemental data for this article can be accessed https://doi.org/10.1080/09638180.2021.1969260.

Notes

1 On condition of anonymity, one accounting firm shared an archive of submitted audit proposals for private audit engagements. Review of these written proposals indicated high similarity (in some cases identical content) in marketing language of the public and private audit proposals. Of course, the technical content of the proposals varied with the nature and scope of the engagements.

2 The National Association of State Auditors, Comptrollers and Treasurers (NASACT, 2012) report that the state audit agency (CPA firms) serves as the primary auditor of the basic financial statements in 34 (11) states and the single audits in 13 (10) states. In most states, a mix of state auditors and CPA firms conduct the single audits. In 2012, in states in which a mix of state auditors and CPA firms conduct single audits, the percentage of government audits conducted by CPA firms ranged from 0.4% (in Alabama) to 98.0% (in New Mexico).

3 We conducted six semi-structured interviews to better understand audit procurement process for US governmental audits. Interviewees included a state controller (Ross, 2013), the Chief Operating Officer of a state’s Auditor of Public Accountants (Giesler, 2013), a deputy state auditor (Carlin, 2013), a state archivist (to better understand the documents submitted and retained as a part of procurement - Moore (2013)), and executives at the NASACT (Poynter, 2013, 2020; Rowland, 2013).

4 In the US states where external audit firms conduct audits, the responsibility of selecting CPA firms goes to different parties across states. In 23 states, the state audit agency is fully responsible for initiating an RFP process to contract out the single audits; in the other states, the entity to be audited selects its own auditor. In the latter circumstance, the state audit agency may or may not be required to supervise or monitor the audit procurement process conducted by the auditee (NASACT, 2012).

5 The evaluation process sometimes includes interviews between evaluation team members and ‘short listed,’ typically three of fewer, audit firms (Poynter, 2020). Interviews of audit firms are more frequent for larger government audits but are rarely mandated. For example, the AICPA (2014a) guide to procuring governmental services and the New York state procurement requirements (https://www.ogs.state.ny.us/procurecounc/pdfdoc/ProcGuidelinesDraft2009-0722.pdf) list interviewing prospective candidates as optional. Our investigation focuses on the written procurement process. Interview data are unavailable for the engagements that we study.

6 For example, KPMG has multiple independence policies and processes, including mandatory independence training, annual independence confirmations of professionals, online systems for managing and facilitating compliance with independence requirements (i.e., KPMG’s web-based application SentinelTM), and independence programs promoting leadership and guidance on independence issues (i.e., Global Independence Group, Ethics, and Independence Partner). See KPMG international Ethics and Independence (EandI) polices at https://home.kpmg.com/xx/en/home/insights/2016/12/our-culture.html. Sentinel^TM is an automated tool used by KPMG for every prospective engagement to identify potential independence (and conflict of interest) issues.

7 The sample is collected from states in which the state audit agency issues an RFP. Data from states in which client management is fully responsible for selecting external audit firms are unavailable.

8 The NASACT includes 56 audit agencies in which state officials are elected or appointed to the Office of State Auditor, State Comptroller, or State Treasurer in the 50 states, the District of Columbia, and U.S. territories.

9 These states include CO, FL, HA, IL, IN, KS, KY, MI, MO, NC, ND, NH, NJ, NY, OH, OK, TX, VA, VT, and WY, of which five states (i.e., FL, HA, MO, NY, and TX) are omitted in the analysis due to incomplete data. States are excluded from the study, because they either do not: (1) hire external CPA firms for governmental audits or (2) maintain a central repository of relevant documents.

10 There are 2,469,214 words in the 368 audit proposals. This domain-specific corpus is large compared to some previous domain-specific corpora; for example, Conrad (2017) compares 60 professional civil engineering reports containing 201,700 words with 50 student reports and 50 civil engineering journal articles which include about 200,000 words each.

11 MANOVA test results indicate that, during the sampled years, these five states do not statistically differ from the (control sample) 45 states in population, modified Ranney’s political competition Index, tax revenue, state revenue per capital, long-term public debt, number of state and local employees, general expenditure, spending from state and federal funds as a percent of state personal income, and state economic freedom scores.

12 We obtained the data on A-133 audits from Federal Audit Clearinghouse (https://harvester.census.gov/facdissm) to determine the category of ‘Small Specialist.’ The category of ‘Small Non-Specialist’ is used for comparisons with the first three categories (i.e., BIG4, regional, small specialist) in the regression analysis.

13 We also utilize NVivo for text similarity analysis to address a concern that audit proposals may consist of standardized, boilerplate language, particularly within audit firms (see Online Appendix 1).

14 Source: Thesaurus. 2014. Accessed September 2. https://www.thesaurus.com.

15 Following reviewers’ constructive suggestions, we created a new dictionary to capture auditors’ language reflective of their commercial interests. The development of this dictionary is mainly based on Gendron (2001) while not following the Pennebaker et al. (2007) validation procedures.

16 The extracted factor has an eigenvalue of 2.02, which accounts for 51 percent of the total variance. The Kaiser–Meyer–Olkin (KMO) value is 0.72, which is acceptable (Kaiser, 1974). The KAO statistic indicates sufficiently correlated variables that are appropriate for a PCA. In addition, all measures of sampling adequacy (MSA) exceed the threshold value (0.5). We run the PCA with varimax rotation. The results are not changed by data standardization (i.e., normalizing before PCA).

17 The term weighting calculation is: ${\omega }_{i,j}=\{\begin{array}{ll}{\displaystyle \frac{\left(1+\log \left(t{f}_{i,j}\right)\right)}{\left(1+\log \left(a_j\right)\right)}}\log \frac{N}{d{f}_i}& ift{f}_{i,j}\ge 1\\ 0& \text{otherwise}\end{array}$ where N equals the total number of audit proposals in the sample; dfi the number of documents containing at least one occurrence of the i^th word; tf_ij the raw count of the i^th word in the j^th document; a_j the average word count in the document (Loughran and McDonald (2011). The unweighted measures (i.e., from Pennebaker et al., 2007) provide alternative variables in supplemental analysis reported in the later section. We report the analysis results using weighted measures, though Henry and Leone (2016) find that weighting, compared to unweighted analysis, does not influence their results.

18 The Center for Public Integrity, a nonpartisan, nonprofit organization, surveys experienced journalists in each state and conducts state integrity investigations. The details of the scoring methodology can be found at https://www.publicintegrity.org/2015/10/14/18316/how-we-investigated-state-integrity. We utilize the assessment results, which regard a state’s internal audit practices (i.e., the existence and effectiveness of a state’ independent audit institution, the independence of audit agency’s investigations, public access to audit reports), as a proxy for state internal audit quality. Online Appendix 3 lists the relevant survey questions.

19 A paired sample t-test indicates that the mean value of INDP is less than that of COMP (t=−26.84), Trust (t=−2.65), and CS (t=−21.61) but greater than that of CML (t=4.68).

20 The average score for US state internal audit quality does not differ from that of the out-of-sample US states (mean=79.3; t=0.26).

21 The Kentucky Auditors of Public Accounts allows CPA firms to submit one proposal for multiple engagements. After excluding the KY data from the sample, the average number of bids per RFP is four. Excluding KY proposals (n=38) from the main analysis does not change the inferences from the results.

22 When an RFP is issued for a multiple-year audit, we calculated the average of the proposed audit fees per year. However, most (57%) sampled RFPs are for one year.

23 We evaluated the coding schema and modified the coding based on the agreement among the PI, co-author, and second coder. The inter-rater reliability of the coding (i.e., agreement between coders) is above acceptable levels as evidence by our Cohen’s Kappa coefficients all > 0.9, which is greater than the suggested criteria of 0.81 (Landis & Koch, 1977).

24 We follow Chang and Stone (2019) in the selection of control variables. WC is the natural logarithm of total words in a proposal; RDFRE is a readability measure calculated using the Flesch reading ease formula. All other control variables are as previously defined.

25 To aid interpretability, we create a dummy variable IAQ_H that equals 1 if a state’s internal control quality is equal to or greater than the mean IAQ in our sample. The regression results are invariant to those reported when the median IAQ is used as the cut-off value.