While digitalization has been accepted worldwide as an absolute necessity in conducting day-to-day operations, companies are still at an early stage of establishing data-driven treasury departments. According to the latest global treasury report by PricewaterhouseCoopers (PwC), more than half of the surveyed treasuries have not allocated dedicated resources to data and analytics, and are not using enough third-party support for the purpose.
Basic systems such as spreadsheets were heavily used by treasuries for years, but these obsolete methods are no longer compatible with expanding businesses. “As the volume and complexity of data increases, treasurers must turn to integrated fit-for-purpose tools to help them manage increasingly complex treasury operations,” HSBC says in its report on data-led treasury.
Cash-flow forecasting, a crucial component of cash management for stable operations, still largely involves manual processing, with less than 3% of treasuries adopting predictive analytical tools, according to the PwC report. Meanwhile, surveyed treasury departments are determined to beef up data usage, mainly to enhance cash-flow forecasting accuracy and visualization.
Application programming interface (API) has emerged as a common cash-flow analysis tool, vastly improving on time-consuming and prone-to-mistake human processing. Gathering data from multiple sources including bank accounts, investment portals, and enterprise resource planning systems, API formats and visualizes data sets traced by different systems to improve the accuracy of business forecasts, shares open banking platform Trovata.
While visualizing treasury data, API pairs with artificial intelligence (AI) to optimize predictability. Top-tier banks such as J.P. Morgan have incorporated AI into their latest treasury tools.
Instant data adjustments have resulted in a stronger analysis of cash-flow patterns, boosting forecast accuracy to 94%, estimates Quadient, a French cash management solutions provider. AI can learn from humongous sets of data to be able to assess the impact of disruptive events such as supply chain disruptions or earthquakes, it notes.
“With access to much more granular data across their organizations, treasurers can model the impact of changes on the company’s future cash position at both a micro and macro level,” states The Association for Financial Professionals.
“Instead of the old model that used static data to forecast positions, liquidity planning incorporates dynamic data that recognizes that a change in one variable has implications for others.”
AI in fraud management
With its unrivalled capability to handle vast amounts of data and react at an unprecedented speed, AI has been deployed in other treasury roles, such as spotting granular glitches in payment patterns to forestall cyber attacks or frauds.
According to Deutsche Bank, common cyber security issues include fake invoices, compromised business e-mails, interrupted and manipulated transactions, and even impersonation of the CEO or a top corporate executive to access sensitive data and trigger unauthorized commands.
Following the shift from cash and cheques to electronic payments, fraud detection can be more challenging, Deutsche Bank projects, citing the rise of instant payment methods and trading of cryptocurrencies and other digital assets.
AI is, in fact, a double-edged sword because the generative branch can be taken advantage of to create cyber attacks. On the other hand, massive data-based training can be used against highly skilled hackers to fend off attacks.
The machine learning process is strengthened with artificial statistics, known as synthetic data. It adds more colour to the database by using AI-calculated threats that have not existed before, which ultimately “helps payments firms pre-emptively assess risks and block fraudulent transactions”, explains top AI chips supplier Nvidia.
Meticulous AI-enabled inspection is often used in know-your-customer verification, shares software and data analysis firm Infosys. AI can tell unusual changes in user names, passwords, and counterfeit signatures, and even decipher suspicious wordings in e-mails, thereby alerting treasuries of potential risks that humans may overlook.
Beyond corporate treasuries, the US government is also exploring the benefits to be derived from AI adoption. In February the US Treasury Department announced that it has recovered over US$375 million after it started using an AI-enhanced fraud detection process at the start of fiscal year 2023.
“The Treasury Department is committed to safeguarding taxpayer dollars through payment integrity – paying the right person, in the right amount, at the right time, and ensuring that Social Security payments, tax refunds, and other types of cheques, and people who are receiving them, are safe from fraud,” says US deputy secretary of the treasury Wally Adeyemo.
Released from the humdrum
Human errors are also something treasuries want to avert. Robotics process automation (RPA) can handle repetitive and manual grinds such as payment reconciliation, says automation solutions provider Automation Anywhere. RPA offers higher levels of accuracy and real-time calculation to handle larger sets of data across multiple accounts, and the ability to comply with different accounting standards. All these add up to lower cost with higher efficiency.
RPA can also accelerate the receivables collection process. As explained by Smart Bridge, a corporate digital service provider, a robot is capable of handling the entire chain of receivable processes, from copying and inserting invoice data, to payment validation and notifying responsible parties. As such, RPA releases manpower from mundane and repetitive tasks, and allows staff to focus on more meaningful decision-making.
Receivable patterns can reveal a great deal of credit management insights. While flagging late payers curbs unwanted partners, RPA can foster long-term relationships with credible counterparts, which leads to more stable and profitable operations.
In this regard, the digitalization of receivables data is essential. It enhances the visibility of relevant information on the receivables collection process, including recent bad debts, receivables concentration ratios, and days sales outstanding (DSO). At the same time, these records must be sorted according to due dates for easier tracking, says credit insurance company Atradius.
Atradius also notes that classifying payers by industry, geography and company size gives a broader picture of risk dispersion than conducting separate analyses of individuals. Then the treasurer can proceed to deploy suitable security arrangements on the red-flagged segments and deteriorated payment trends, including credit insurance and transfer of receivables, as well as amending clauses in payment terms and providing payment incentives.
Using external data
As internal data only show a limited perspective on payment trends, a more comprehensive analysis is supplemented with public data including court filings, third-party statistics, and information from banks.
Texas-based fintech company HighRadius notes that externally tracked elements like credit scores use “historical data and statistical analysis to predict future behaviour”.
“Third-party providers have access to vast quantities of reliable data that can provide a more complete picture than internal data or self-supplied references from the customer,” explains Dun & Bradstreet, a business data analysis company. It also suggests the use of a credit scorecard, which rates companies on a set of weighted variations.
Banks hold enormous amounts of payment details from numerous buyers, allowing treasuries to conduct peer-to-peer comparisons on their late payers. For instance, spotting shorter DSO and days payable outstanding with another supplier will justify more favourable payment clauses, Asset Benchmark Research learns from bankers participating in one of its surveys.
“Data is useful, but only with analytics to turn it into actionable insights that create value and better decision-making in treasury,” HSBC says in its data-led treasury report. “Analysis of historical data, cash flow patterns, and market trends can lead treasurers to enhanced cash management strategies, from funding to working capital, predictions of receivables or improved payment operations.”
