I’ve just read an article on AI and Audit, asking when (or if) Artificial Intelligence will take over the Auditing of companies financial statements. (“Distinguishing Hype from Reality about the Future of Automated Audits” By Gregory P. Shields, CPA, CA http://thinktwenty20.com/images/Issues/Winter_2020.pdf)
While the article provides a fine survey of the technical issues and impediments, it misses one key element; AI is a threat to the Auditing and consulting firms' business model. Until they can determine how they incorporate this without erosion of fees and profit, there will be limited uptake.
It is an interesting article, if you are into this kind of thing, and considers three scenarios.
|
Arguably, predictions fall into
three overall categories: 1. AI will soon displace human
auditors. Clairvoyants in this category might be referred to as chicken
littles, predicting the auditing sky will soon fall on human auditors. 2. AI will complement the work of
human auditors, but never entirely displace them. These clairvoyants might be
referred to as eternal optimists, looking at the future of human auditors
through rose-coloured glasses. 3. AI will ultimately displace human
auditors. These clairvoyants may be the realists. But, what does “ultimately”
mean and what are the implications for decisions that have to be made now? (Gregory is a member of the board of
directors of the University of Waterloo Centre for Information Integrity and
Information Systems Assurance (UWCISA). He is also a member of CPA Canada’s
Audit Data Analytics Committee. Before his retirement, he was CPA Canada Director,
Auditing and Assurance Standards. His recent projects include developing
non-authoritative auditing guidance on data analytics, cryptocurrencies, and
accounting estimates.) |
The conclusion is not surprising; that AI still has a long way to go, that regulators will take convincing, and that there remain too many cognitive processes that cannot yet be replicated by AI. These are the “value-added” that the human accounting specialist brings to the audit.
But what is missing from the article is the change required in the economics of audit to allow AI to play a meaningful role in disintermediating the human auditor.
The Accounting profession and Auditing firms exist in and rely on an economic model that will delay the introduction of full AI support for the audit. Auditing firms are highly profitable ‘consulting’ companies with a market niche that they will exploit and protect from competitors, and AI is, even if used as an ‘internal’ tool, a competitor.
Almost all major accounting firms are partnerships. While this is mostly a structural and governance mythology, they are bodies of self-feeding units that rely on a “Borg” like assimilation into a greater single entity which looks out for the collective interests. Let’s break that down. Auditing firms are like ant hives, all working together, but ready to cull the non-performing elements, and perfectly happy to jettison worker ants to protect the hive.
Each Partner owns a number of shares in the company, and those shares pay a dividend to the partners. The more shares, the greater the percentage of profits. Junior Partners “buy” a starter set of shares, and as they progress in profitability and seniority, they are allocated more shares. There are no external shareholders, and therefore no wider investor community to answer to. Certainly they talk stakeholders, but what they means are those that can impact their business model for the better or worse.
Each Partner has a team (or shares a team with other partners for efficiency) and ‘sells’ those resources to clients on specific engagements. Basically, people-hours for an agreed person-hour rate, frequently merged into a single overall rate or price for the job. At the core is that simple economics of the cost of inputs and the income generated by those inputs. Technology helps those inputs (people hours) be more productive and increases the potential price of that input on an hourly basis – because fundamentally, hours are the product.
Just as a metal stamping business many (does) use percentage utilisation of metal stamping machines as a measure of productivity, the accounting firms carefully monitor individual human utilisation rates as the core to determining productivity and potential profitability. People-hours utilised (billable) is the core of the Auditing (and associated consulting services) profession's productivity and profitability.
All other functions within the Auditing and consulting firms are paid for by the numbers of hours of human time that are sold and can be seen by the client as being delivered. The cost of delivering those hours if high. Premises, technology, HR services, and ongoing training of staff are significant costs. But there is another cost; the cost of ensuring that the profession continues to exert control over companies and their financial supply chains. After all, without a ‘clean’ audit, most companies cannot access capital at economic rates.
As an aside, brothels work in much the same way; the fundamental product is people-time, and the more time your product can spend ‘billing’ the more profitable the enterprise. Just like the brothels, if the young ones are not producing enough revenue, the young ones and the madams suffer.
A non-profitable partner or business unit within an Audit and consulting firm will be culled, and quickly. This means that a major element of the Senior Manager through Partner ranks, and especially anyone aspiring to Partner rank, is the selling of new business and the protection of and ideally expansion of fees with existing clients.
Unlike the brothels, Auditing and consulting firms have the additional costs of sustaining an industry body whose purpose is to ensure that they retain their dominance in the financial supply chain. (although I guess one could equate this to the cost of bribes to police and local officials?)
Audit firms carry the additional costs of lobbying at the international, national and state level, and through ‘ownership’ (capture is probably a better word) of the process for standards-setting. And standards that will threaten their business model will be managed. The International Accounting Standards Board (as one example) would find itself significantly less effective without the ongoing support from the major Auditing and accounting firms. The accountants and auditors provide much of the technical resources required to draft and review standards, and technical resources are frequently ‘seconded’ at no cost from the firms. And for a good reason; such secondments ensure that their people contribute to the development of the standards, it also ensures that the firms have resources with in-depth knowledge to trot out to their clients.
An AI-heavy audit is a threat to their business model. It is difficult to add a premium to the audit fee for a reduction in the amount of work that will be performed by humans. After all, if the human time is the primary cost factor, when a reduction in the total human inputs should translate into a reduction in the audit fee. Firms have been down this road before. Audit automation and through automated work-papers contributed to a reduction of the audit fee, but was offset by increases in the costs of labour and the total amount of effort required.
Sarbanes-Oxley section 404 was an excellent example. Following the Enron and Worldcom frauds, the additional requirement for a CEO/CFO certification statement resulted in a windfall to the Auditing and consulting firms, boosting total numbers of hours on audits, and the cost-per-hour of now scarce resources. First year accounting graduates were suddenly seeing offer letters at 15% to 20% higher than their friends received the year before. Internal Audit consulting firms found that they were able to raise their rates by more than 50% to 70% in a year, with that particular gravy train lasting almost a decade. It took that long to build a population of internal auditors sufficient to service the new need, in order to once again begin to push rates down.
This is after ignoring the fact that two Auditing firms facilitated the frauds at Enron and Parmalat. Ironically, such frauds, while costing individual firms dearly, actually raised the overall revenues across the industry. In the case of Anderson’s, it destroyed the firm. In the Parmalat case, the local accounting major’s firm was expelled from the network, and their international clients in Italy were instructed to contact EY on their first day back in January 2004.
Another area where the business model will not be permitted to be undermined by actual practice is the relationship with Internal Audit. Long considered the slightly dim cousin in the auditing world, Internal Audit arguably plays a more critical role in the actual success of companies. Not only does Internal Audit provide real comfort to the Board (along with Risk Management and Management), it can also contribute to process improvement and organisational efficiency. IA should also be able to contribute to a reduction in the cost of External Audit.
I’ve long argued that improved Internal Audit should enable the External Auditors to place greater reliance on the system of internal controls, and therefore reduce the amount of external effort and cost required to provide an audit opinion. Yet year after year, regardless of the depth or coverage of Internal Audit, the External Auditors provide trite comments about the need to confirm the quality of IA’s work, and that perhaps next year there could be a reduction in External Audit activity. Next year never comes, regardless of how thorough IA has been, or how effective management has been at the implementation of IA recommendations.
After all, effective IA is a threat to the External Auditors’ business model.
And the business model trumps any other consideration.
Artificial Intelligence will encroach on the audit, but it will not successfully do so until the Audit and consulting firms have figured out how they can incorporate it into their business models, or until their business models are changed..
• In my view, you are quite correct in pointing out that increased use of AI is likely to change the economics of the audit.
ReplyDeleteConcerns about how to appropriately price an audit has come up in the context of possible increased use of automated audit techniques (data analytics) even without making use of AI. As you note, human hours (the basis most commonly used now for determining audit fees) are reduced by use of automation. Fear and uncertainty about how and when to change the audit business model are likely key inhibitors of increased use of automation in audits, including AI.
One proposed solution is to move to “value-based” invoicing, but that is easier said than done: audits are commonly viewed as a commodity. That view may change, though, if some firms take the bull by the horns and see a competitive advantage (or more importantly, a billable competitive advantage) in making use of automation to better detect fraud and provide much improved evaluations of management’s assessment of an entity’s ability to continue as a going concern (two big aspects of the continuing audit expectations gap).
• Regarding the structure of firms and its effect on audit partner remuneration, rules of professional conduct have evolved to help ensure that provision of non-audit services by a firm does not impair the independence of an audit partner. Also, in the near future, UK regulations seem likely to go even further and require the Big 4 accounting firms to separate there audit service arms from their business advisory arms. This is, in part, in response to huge audit failures relating, for example, to Carillion and Wirecard. Whether other jurisdictions such as Canada will follow the UK example is difficult to predict.
• I think you may be underestimating the effects of internal audit on the external audit. I recall the debates held when ISA/CAS 610, Using the Work of Internal Audit was being revised back in 2012. Auditors from western countries (including Canada) maintained that audits of many large financial institutions would not be practicable without using the work of internal audit. Auditors from other countries (notably Japan) were strongly opposed to even allowing use of such work because of lack of internal audit independence. The IAASB (and AASB) decided to take a balanced view as noted in the Basis for Conclusions of the standard which states: “Not only should the pitfalls of over and undue use of the work of the internal audit function for purposes of the audit be highlighted, but also the advantages to audit quality of a constructive and complementary relationship between external and internal auditors including, where appropriate, the possibility of coordination and cooperation.”
I used to be involved with developing non-authoritative guidance on continuous automated audits. At various conferences over the years, it was apparent that while some internal auditors were moving to adopt continuous automated audits, external auditors were not. However, it was also apparent that many internal auditors were just a reluctant as external auditors to automate their approaches. That may be changing (I’m not sure).
• Some may question the validity of your comparison of audit firms to brothels, although I have heard auditing being referred to as the second oldest profession!