How Does Combined Assurance Affect the Reliability of Integrated Reports and Investors’ Judgments?

Abstract

As integrated reports evolve to meet the needs of stakeholders, it is increasingly important to establish reliable information. We examine the effect of a new technique to enhance the reporting reliability, referred to as ‘combined assurance’ (through which the audit committee coordinates the assurance roles of management, internal assurance and external assurance providers and concludes on the effectiveness of risk management, internal controls and reporting quality), on restoring investors’ willingness to invest when there are significant reporting reliability risks. We conduct a series of experiments with different operationalizations of reporting reliability risks. We find that the communication of combined assurance can restore investors’ perceived reliability of reported information and willingness to invest when there is bad news (Experiment 1); when there is a non-financial key audit matter (KAM) indicating estimation uncertainty in the non-financial information (Experiment 2); and when there is a financial KAM indicating risks of manipulated financial reporting (Experiment 3). Our study has implications for practitioners, standard setters, and regulators who are exploring possible mechanisms to enhance the reliability of financial and non-financial information in the integrated reports.

Keywords:

• Audit report
• Combined assurance
• Integrated report
• Investor judgment
• Key audit matters

Acknowledgements

We are very grateful to William Messier (the Editor) and the two anonymous Reviewers for the advice and insightful suggestions on our paper. We also appreciate helpful comments from Roger Simnett, Ken Trotman, Hun Tong Tan, Robyn Moroney, Kerry Humphreys, workshop participants at UNSW Sydney School of Accounting, Monash University Department of Accounting and 2019 AFAANZ conference. We are grateful to CPA Australia for their Global Research Perspective grant, UNSW Sydney School of Accounting and Monash University Department of Accounting for their financial support. We thank Fangbin Lin, Jingduan Li and Sarka Stepankova for their excellent research assistance.

ORCID

Hien Hoang http://orcid.org/0000-0003-4939-180X

Soon-Yeow Phang http://orcid.org/0000-0002-5647-1122

Notes

1 See best practice for combined assurance reports among South African companies in Appendix 1.

2 It is possible for companies to issue both traditional assurance report and combined assurance report. Zhou et al. (2019) control for the existence of CSR traditional assurance report and they find that the combined assurance report provides incremental value to the analysts’ decision making, suggesting a complementary effect of these two types of reports.

3 We also ran a principal component analysis on these two questions. The two questions load in the expected direction and a single component explains 85% of the variance in responses (eigenvalue = 1.67). We ran an additional analysis using the extracted common factor as the dependent variable, and no significant differences in inferences were found for our main results.

4 We asked participants to indicate the favorableness of the company's non-financial (natural capital) performance in accordance with the company's annual report. We find that participants rated the favorable performance as 7.47 (out of 10) under the good news condition and 7.02 under the bad news condition. These ratings are not significantly different. This result enhances our confidence that the perceived positiveness of the reported non-financial information is constant across conditions, and is not the factor driving our main results.

5 Prolific, located in the U.K., is an online subject recruitment platform, similar to Amazon's Mechanical Turk. Prior research indicates that Prolific is suitable for social science experiments, and Prolific respondents are diverse (Peer et al., 2017). Prolific has been used in other accounting studies (Murphy et al., 2019; Owens et al., 2019; Rennekamp et al., 2018).

6 We included a five-item quiz on participants’ knowledge of investment, accounting, and finance; 83% answered all questions correctly. We included a covariate (based on the marks they obtained on the quiz) and find no significant changes in our main results. When we excluded the participants, who did not have 100% correct answers, we find no significant changes in our main results.

7 76% of the total participants chose the correct end of the scale – 49 participants (78%) who read bad news chose ratings 0 or 1 ([0] Very unfavorable news) and 43 participants (74%) who read good news chose ratings 9 or 10 ([10] Very favorable news). There are no significant differences in inferences when we exclude the participants who failed to choose the correct end of the scale.

8 We conducted an additional analysis on whether our two independent variables affect the perceived reliability of financial information, perceived relevance of financial information and relevance of non-financial information, but find no significant differences (all p > 0.1, one-tailed). This enhances our confidence that it is the perceived reliability of the non-financial information (and not other factors) that is driving the investment judgment results.

9 As an additional analysis for a different operationalization of the non-financial assurance level, we conducted a 2 (CSR news: good, bad) × 2 (combined assurance communication: traditional assurance, combined assurance) between-subjects experiment in which the non-financial assurance report has a limited assurance level instead of a reasonable assurance level. We recruited 98 participants from Prolific as non-professional investors, and do not find significant differences in perceived reliability (mean = 7.23 vs. 6.79, p > 0.1, untabulated) or willingness to invest (mean = 6.55 vs. 6.58, p > 0.1, untabulated) between limited assurance and reasonable assurance conditions, irrespective of the presence of independent variables. When we reran the analyses using the limited assurance cells, we do not find significant differences in inferences in our main results. While we acknowledge that it may not be statistically appropriate to compare cells across two experiments, these findings offer some indication that the assurance level (reasonable or limited) does not make a significant difference to our main results.

10 We also measured the perception of management credibility (Mercer, 2005) on an 11-point scale ([0] Strongly disagree; [10] Strongly agree). We captured information on the following factors: (1) whether management is trustworthy and (2) whether management is competent. Factor analyses for these credibility questions indicate that these questions load together on one factor, with factor loadings 0.79 or higher. These questions were averaged to obtain the overall perception of management credibility measure. Our findings indicate that the communication of combined assurance does not significantly affect participants’ perceptions of management credibility, irrespective of the presence of good news or bad news (all p > 0.40). Participants were asked to assess the risk of investing in the company on an 11-point scale ([0] Not risky at all; [10] Very risky). We find that the communication of combined assurance reduces the perceived risks when there is bad news (mean = 6.27 vs. 4.67, t = 3.02, p = 0.008, one-tailed, untabulated), but does not have a significant effect when there is good news (mean = 5.17 vs. 4.79, t = 0.68, p = 0.451, one-tailed, untabulated). Last, participants were asked to assess their confidence in the investment decisions they made on an 11-point scale ([0] Not confident; [10] Very confident). We find that the communication of combined assurance increases investors’ confidence when there is bad news (mean = 6.17 vs. 7.36, t = 2.47, p = 0.035, one-tailed, untabulated), but does not have a significant effect when there is good news (mean = 6.79 vs. 6.69, t = 1.04, p = 0.361, one-tailed, untabulated). The main results are not affected by perceived management credibility, perceived risk, or perceived confidence as covariates. The additional analyses provide consistent findings with H1.

11 Similar to Experiment 1, we included a five-item quiz on participants’ knowledge of investment, accounting, and finance; 91% answered all questions correctly. When we exclude participants, who did not have 100% correct answers, we find no significant changes in our main results.

12 Of the participants, 61.6% were male, 90.2% owned shares, 93.8% planned on investing in the future, and 83.9% were below the 35–44 years age range; their investment experience was 7.0 years and full-time work experience was 13.5 years; 98.2% had university qualifications; they had taken an average of 4.04 accounting or finance units; and 89.2% had referred to a company's annual report to make investment decisions. Subject demographic variables were not significant when included as covariates in our analysis and did not affect the main results.

13 Our additional findings indicate that the communication of combined assurance does not significantly affect participants’ perceptions of management credibility, irrespective of the presence of KAM (all p > 0.40). We find that the communication of combined assurance reduces the perceived risks when a KAM is present (mean = 6.03 vs. 4.68, t = 2.70, p = 0.020, one-tailed, untabulated), but does not have a significant effect when a KAM is absent (mean = 4.84 vs. 4.81, t = 0.05, p = 0.499, one-tailed, untabulated). We also conducted an additional analysis on whether the independent variables affect the perceived reliability of financial information, perceived relevance of financial information and non-financial information, perceived management credibility, perceived investment risk, and confidence level, but we find no significant differences (all p > 0.10, one-tailed, untabulated). The main results are not affected when we include these investors’ perceptions as covariates.

14 One potential explanation for this non-significant mediation effect is that the non-financial KAM disclosure is a voluntary disclosure in the CSR assurance report. The lack of familiarity and understanding of this new disclosure could have resulted in weaker investors’ reactions. Also, there can be mediating factors other than perceived reliability of CSR information through which combined assurance influences investors’ willingness to invest.

15 In Experiment 1 and 2, we provided the participants with an integrated report with financial and non-financial information, financial audit report and non-financial assurance report. In Experiment 3, since we focus on the effect of combined assurance and KAMs in the financial reporting context, we provided participants with an annual report with financial information and financial audit report.

16 We restricted our subject pool to participants who did not participate in our previous experiments. 58.3% of the participants were male, 92.5% owned shares, 90% planned on investing in the future, and 79.2% were below the 35–44 years age range; their investment experience was 7.95 years and full-time work experience was 14.69 years; 93.3% had university qualifications; they had taken an average of 4.55 accounting or finance units; and 83.3% had referred to a company's annual report to make investment decisions. Similar to the first two experiments, we included a five-item quiz on participants’ knowledge of investment, accounting, and finance. 92% answered all questions correctly. Subject demographic variables were not significant when included as covariates in our analysis and did not affect the main results.

17 Our additional findings indicate that the communication of combined assurance did not significantly affect participants’ perceptions of management credibility, perceived riskiness, and confidence level, irrespective of the presence of a KAM (all p > 0.40). The main results are not affected by the perceived management credibility, perceived risk, and perceived confidence as covariates.

Additional information

Funding

This research is supported by CPA Australia Global Research Perspective Grant (number RG180370)