What is the rise of big data and AI doing to audit? Quite a bit, as Caroline Biebuyck finds out.
Audit is facing what could be its biggest shake-up ever. Automation is changing the nature of audit work, replacing much of the low-level ticking and bashing with higher value added work, concentrating on analysis and judgements.
And it is something all audit firms need to know about. While the largest firms are leading the way with introducing new technology, these techniques are starting to filter down and will eventually have a place in all audit practices.
The biggest game changer to date is data analytics. This can be used to review the ordinary transactions of a business, downloading transactions from client companies and interrogating them. Nick Frost, KPMG head of audit technology, gives the example of a global client that had more than 3.5 billion transactions across nearly 100 countries last year. “You need to know what’s material and important and where the risks are to ensure you don’t get carried away. We could interrogate this to the nth degree. But we don’t: our tools interrogate in a way that’s appropriate for the audit.”
This interrogation starts to get interesting when machine learning – in which the technology continuously teaches itself – is applied. For instance, PwC is using machine learning to audit journals. “We’ve not told it what to look out for, such as journals posted on a Saturday night or unusual amounts or unusual account combinations,” says Gilly Lord, head of audit strategy and transformation at the firm. “The more journals and the bigger population you feed it, the better it gets at identifying what is a real anomaly.”
One of Lord’s colleagues piloted the machine learning in parallel with the regular human investigation of the data analysed on a recent audit. The results, she says, were fascinating. “The machine is identifying fewer anomalies for investigation – but it turned out that these were the real anomalies, the ones we needed to spend time on.”
Some analytic tools are being applied to judgements that are based on predictions, such as asset impairment. These tools can review a client’s forecasts for recovery of an asset and then apply predictive algorithms against those forecasts to come up with a probable value of the cash the asset will generate. What used to be a manual process is transformed into a multiple-scenario model that can be used to challenge the client’s judgement.
These tools are incredibly powerful, says Frost. “Now I can sit in front of an audit committee and say I am 90% sure that this asset is worth what you think it is. That’s something I could not have done without the predictive algorithm.”
There’s also a lot of work going on around natural language generation and natural language processing. This has become more pressing with the introduction of new accounting standards, such as those for revenue recognition, which require audit teams to review large numbers of contracts. The Big Four firms have machines that use natural language processing to “read” the contracts, looking out for triggers that might make a difference when the new rules come into play.
BEYOND THE BIG FOUR
Much of the growth in data analytics and artificial intelligence in the Big Four firms over the past few years can be put down to one factor: competition. The change in audit firm rotation rules and resultant explosion in tenders for listed company audits has sent the largest firms into a technology arms race. “If you don’t turn up at the tender beauty parade with something to say about technology, you’re less likely to win the contract,” says Henry Irving, head of ICAEW’s Audit and Assurance Faculty.
As the technology trickles down, every audit firm, regardless of its size, needs to decide on its innovation strategy: whether it should be a first mover, fast follower, mass implementer, or laggard. None of these choices are bad, Irving points out – it’s all a question of what suits a firm’s client base. “Firms need to balance the risks and benefits,” he says. “Being a first mover comes with high costs and high risks; that’s of limited benefit to firms further down the size ranking.”
Medium-sized firm UHY Hacker Young is taking its time in evaluating its strategy. The firm has looked at three different software providers, all of which take a slightly different approach to using data analytics in the audit. “The difficulty is in establishing how we can actually get better audit quality through the use of data analytics,” says partner Martin Jones. “We will invest but I think like many mid-tier firms we will wait and see how the technology progresses before jumping in.”
As is usual with firms of its size, most of the UHY’s audits are based on substantive testing rather than testing and then relying on controls. This means that the key risks tend to be around judgement calls and non-routine transactions. “I’m not sure to what extent data analytics programmes lend themselves to that – you still need a human to consider the grey areas,” says Jones.
While there are cloud-based technology options being marketed to cover substantive tests, the critical factor is cost. One solution Jones looked at carried an approximate fee of several thousand pounds for each audit. “If the audit fee is £1m then that’s not a problem, but if it’s £10,000 then it’s not worthwhile.”
Small firms are the slowest at taking the developing technologies on board. However, small audit practices could find data analytics useful in risk assessment and analysis, says Nigel Hughes, managing director of Totteridge Associates and chair of the ICAEW Practice Committee.
“Audit focuses on risk assessment, planning and so on, and that hinges on asking the right questions. Those questions could be generated from more intelligent accounting systems or other tools, and this will have an impact on the way all practices do their audits. In addition, digital techniques will affect more judgemental issues, either because of how the questions are structured or because they will highlight judgements that need to be made.”
Hughes is optimistic that while the owners of some small practices are deciding audit automation is a step too far and are selling up, some will take the plunge and invest in the new tools available. “A good number of those I have talked to see the opportunities and commercial benefits in audit technology. Especially since they can rent cloud-based software, so do not need to make a large investment.”
TRANSFORMATING THE AUDIT
What impact is this technology having on the overall audit product? Firms cite improved audit quality: machines can simply do things quicker and more accurately than humans. Other improvements are making the audit more effective (“previously when we had to do something manually it might have taken us down a rabbit hole,” as Frost says) and giving better insight into the historical accuracy of a company’s forecasts and the volatility of assumptions in those forecasts.
Irving, however, points out that the new technology is still a tool. “Questions around audit quality and scepticism are the same as they always were. A practice might have different tools and be operating in a different environment but it still needs to be aware of what’s going on and make a judgement on where it takes its practice and skills set to do what’s expected of it.”
David Herbinet, global head of audit and assurance at Mazars, has long been concerned that the current audit service does not deliver what the market expects or needs. “We believe that the collaboration of audit professionals with stakeholders in audit and technology experts can bring results that are capable of disrupting a market that has not evolved much over the years,” he says.
“Everything is changing, from the way technology is used and is impacting business to new societal expectations on business conduct and the paramount importance of trustworthiness. It’s imperative for our profession to find alternative ways to deliver our service.”
WHAT GOES ON INSIDE THE BLACK BOX?
As software develops, more of the audit process will likely be performed on automated platforms. Clients may also use these platforms themselves. ICAEW’s Henry Irving says this has implications on the use of such platforms in the statutory audit. It also raises the question of whether the platforms should be regulated as well. “Do the platforms reach certain criteria and do they comply with the international standards on auditing? If there is less auditor involvement in the mechanics of what’s being done, the auditor might need to show he has correctly discharged his duty to make sure the tool he’s using is appropriate and up to standard.”
This could include considering whether there is any bias pre-programmed into the system. Consider Microsoft’s unfortunate experience with Tay, its AI chatterbot launched in 2016. Microsoft had to shut Tay down within just 16 hours; learning from its users, the bot’s persona had turned from wholesome teen to offensive bigot.
Meanwhile, a project sponsored by the US Defense Advanced Research Projects Agency is looking to get intelligent machines to explain to humans how deep learning systems come up with the answers that they do – in other words, how the machines’ minds work. The research is expected to run until 2021.
BLOCKCHAIN: THE BIG DISRUPTOR?
A number of multinational companies are looking at putting their supply chains on blockchain – and more are thinking of joining them. Records that go on blockchain can be permanent, immutable and public, explains Richard Anning, head of ICAEW’s IT Faculty. “A lot of audit work could become redundant on blockchain, such as reconciling balances of parties that are on it. The larger firms are working on this, as are established tech companies like IBM together with a variety of tech start-ups.”
Although many people anticipate blockchain having a dramatic impact on audit, Anning points out that the technology is at an early stage. “We are looking at how accountants and auditors can use the technology rather than assuming it will spell the end of the profession; we will need to work with this and adapt.”
Originally published in Economia on 7 September 2017.