How much should treasurers fear robot replacements?

Increasing levels of automation are reducing the need for manual input – as they should. But are treasury roles at risk? Lesley Meall investigates

It was the best of times, it was the worst of times. Over the past decade, crises in banking, credit, the euro and sovereign debt have formed some very dark clouds, but their silver linings have highlighted the tactical and strategic importance of efficient cash and liquidity management, the mitigation of risk and the need to optimise resources, save time and improve transparency – accelerating trends towards greater use of specialist software and systems in many areas of corporate treasury.

“Increased automation in treasury has become a strategic objective for Cairn Energy,” says Robert Scriven, the energy company’s group treasurer and planning manager; and Cairn is not alone.

 Does this mean that treasurers are heading down the slippery slope towards extinction? Can all treasury activities and complex processes potentially be automated? 

“We want to ensure that our treasury processes are as efficient and as automated as possible,” says Guy Morgan, treasurer for Australia’s Export Finance and Insurance Corporation, which recently opted for a cloud-based treasury management system (from Reval). “Our goal is to have end-to-end workflow on a single platform to eliminate manual intervention,” says Morgan, enabling straight-through processing (STP) of payments.

Meanwhile, Cairn Energy has improved STP on the back of its installation, identifying a need for trade through trading, confirmation, payment and so on, says Scriven. But there are challenges. “Suppliers are not well integrated and this is particularly difficult outside FX, where auto quoting and trading are less developed,” he explains.

Intelligent software

At MAHLE Industries, the entire multifaceted processing of remittance advices – many of which arrive by email – has been automated. The global automotive supplier has implemented a solution that uses specialist software (eFLOW from Top Image Systems) to automatically identify, extract, classify and verify the relevant information from these payment advice emails.

It then delivers that data electronically, directly into the FinanceSuite AutoBank Automatic Cash Application (an SAP add-on from Hanse Orga), which has built-in intelligent reconciliation logic for bank statements, remittance advices and lockbox information.

“We are happy to have been able to automate these processes,” says John Nielsen, MAHLE Industries’ treasurer. The project has enabled the company and its finance function to save time and improve efficiency, he says. “The previous remittance advice processing was dominated by tedious manual tasks,” he recalls. “The time savings we have experienced due to the project have been so significant that our finance staff now has time to focus on more value-adding tasks,” he adds.

As software and systems become smarter and able to exchange information more easily, the potential for replacing manual and semi-manual processes that once required people, with entirely automated processes, is increasing. We are already seeing use of software-based robot process automation (see the ‘What are software robots?’ item below) in some treasury activities and more may follow suit.

But does this mean that treasurers are heading down the slippery slope towards extinction? Can all treasury activities and complex processes potentially be automated? Might smart software one day replace the entire treasury function? As with many such big questions, perspectives and answers tend to vary based on who you ask.

Robot replacement

Daniel Wong, formerly global treasury manager at British American Tobacco (BAT) and now BAT business development manager (M&A), says: “I think there is definitely potential to replace some activities in the near future, but not all.”

Recent research from Oxford University and Deloitte, From Brawn to Brains (PDF download), predicts a bleak future for finance and accounting professionals in many areas, and treasury seems unlikely to be immune. But can all treasury processes follow accounts payable and receivable down the road towards robot automation?

“Some activities are already being performed by robots today,” Wong says. “It’s easy to see this in the cash management and payment space where you can have robots recognising patterns that do cash allocation and general ledger clearings and postings. In the risk management space, it’s also easy to see that robots or programs can consolidate and analyse data very quickly.”

 Software robots are good in automating simple repeatable processes where you deal with masses of transactions 

In the near future, there may be little need for big teams of people to manually manipulate data. “In the dealing space, there can be algorithms that can be run to manage small positions as and when they arise,” he adds.

So perhaps the direction of travel is clear; perhaps not.

“I think there is a lot of complexity in treasury that makes it hard to completely remove a human’s touch,” says Wong. For example, managing sanctions, responding to changing regulations and making strategic decisions on funding and risk appetite.

“All these things will require judgement," he adds, "although these decisions can be aided by robots. Time will tell: it may be the next generation of automation that radically reshapes treasury, rather than current incarnations of robot process automation.

Dawn of the machine

“We are seeing the dawning of machine learning in corporate treasury,” says Jukka Sallinen, a cash management domain expert at OpusCapita, a financial process automation specialist. “Software robots are good in automating simple repeatable processes where you deal with masses of transactions,” says Sallinen. “Many treasury tasks are more complex by nature,” he adds.

The combination of robot process automation and machine learning now emerging will create many new possibilities, because machine learning can help with the automation of complex decisions that cannot be done solely using rule-based process automation.

Machine learning uses computer algorithms that can learn and make predications of data. “It’s not a new concept. We’ve already seen many applications around automated stock markets and in the medical industry, to name but a few,” says Sallinen.

Potential applications for corporate treasury vary from self-learning cash forecasting to detection of payment fraud anomalies. The latter is a good example: treasury has a key role in mitigating cash-flow-related crime attempts, but to mitigate the risks, you need high-quality payment processes and the capacity to monitor flows, then make necessary adjustments to the strictness of your payment processes.

This undertaking is a mountain for a person, but it’s a molehill for a robot.

Vital components

“Machine-learning algorithms can be introduced to alert and stop exceptional flows,” says Sallinen. In fraud detection, no matter how knowledgeable or experienced a person is, they are pitting their wits against an algorithm that is built on a model containing huge numbers of past transactions and variables, along with their potentially myriad correlations.

“This means the model is much more powerful and able to catch suspicious flows at an earlier stage than a human,” Sallinen adds. So the model has a much greater potential than a person or a team to help with the mitigation of risks in corporate payment processes.

Where does this leave treasurers? “Technology is advancing at a rapid pace and it’s important to keep on top of the trends,” says Robert Waddington, a director in PwC’s corporate treasury and commodities group.

When the group recently identified eight technology trends that could impact future treasury teams, the list was topped by robotic processing and data analytics, vital components in the next generation of smart systems.

Services around machine learning and artificial intelligence in treasury and cash management remain largely at pilot stage. Treasury professionals with an eye on the future will want to watch this space.

What are software robots?

Hollywood has helped shape our perceptions of the future and robots. However, as Raheem Hasan, president of the Institute for Robotic Process Automation, explains, the reality of software robots is rather different.

“A software robot or robotic process automation is the application of technology or computer software that can be configured to capture, interpret and interact with existing applications for processing transactions, manipulating data, triggering responses and communicating with other digital systems,” he explains. Their potential scope is vast.

Large-scale users of knowledge workers can leverage software robots or robotic ‘full-time equivalents’ to perform high-volume transactional processes, at a fraction of the cost of people. “Just as industrial robots are remaking the manufacturing industry, software robots are revolutionising the way that knowledge work gets done,” he adds.

This is just the beginning…

About the author

Lesley Meall is a freelance journalist specialising in technology and finance

Scroll to top