In times of extreme volatility, accurate cash flow forecasting is more important than ever for treasurers. At the height of the COVID-19 pandemic, a company’s forecasting could mean the difference between surviving the storm or going under. We spoke with experts to discover just how accurate cash flow forecasting is in reality, and how it can be improved.
Cash flow is the “lifeblood” of any company, according to Ray Suvrodeep, Global Head of Deposit & Investments Product Management at HSBC. In the first two weeks of the COVID-19 pandemic hitting Europe, ratings agency S&P reviewed 20% of its portfolio, downgrading or putting on negative watch or outlook some 75% of the reviewed companies. It became evident that companies that couldn’t demonstrate a steady cash flow would suffer not only because of dried up demand, but also from ratings scars.
Unfortunately for treasurers, there is no way to forecast future cash flows with 100% accuracy. As Shafiq Abdul Jabbar, CFO at Astro Malaysia Holdings Bhd (AMH) explains, “The future outlook is always fraught with many variables, some controllable and some not.” He also notes that there is an added dimension with this, as volatilities are not constant across time.
Dino Nicolaides, Managing Director, Head of Treasury Advisory UK&I at Redbridge Debt & Treasury Advisory, agrees, explaining that even with effective forecasting tools in place, a company can never be sure where it will land. One example, he says, is the airline industry. Airlines are particularly sensitive to a lot of factors. They produce forecasts on passenger numbers, fuel costs, staffing, lease costs of planes, to name a few. “All of a sudden, a pandemic like COVID-19 happens, and everything is grounded – so all forecasts go out the window,” he explains.
Of course, pandemics aren’t the only thing that can have this effect. For example, a terrorist attack or a global recession may stop many people travelling, Nicolaides adds. The speed with which these events unfold can put treasurers in a difficult position where forecasting is concerned.
For Abdul Jabbar, every forecast is just a “best estimate” at a point in time. The key for him is keeping a closer eye on the controllable items, whilst simulating possible scenarios for the uncontrollable. He adds that it’s also essential to manage the residual risks to an acceptable level within the available capacity of the organisation, using third-party tools where possible.
Nicolaides agrees and says that the question of what is accurate enough is a common one in his line of work. “It all depends on what decisions you make off the back of your forecasts,” he says. Forecasts have so many uses (for example, liquidity management, foreign exchange hedging, setting strategic objectives), and so treasurers need to establish exactly what they’re using the forecast for before they can establish its efficacy.
Meanwhile, Suvrodeep explains that the most common issue with cash forecasting comes from the “single version of truth”, whereby the treasury’s understanding of the forecast is different to that of the business units. But, he says that the goal around accuracy begins with improvement from current, unsatisfactory levels, and transitions into maintaining the desired levels in a cost-effective fashion. “The actual target is very much dependent on the individual company,” he says. Treasurers, and CFOs, need to assess if they can rely on the forecast to make decisions.
Suvrodeep adds that in practical terms, there is a natural trade-off between accuracy and the time horizon of the forecast. “Treasury strives for greater accuracy of its short-term forecast – say end of current week up to 13 weeks – as granular data and business outlook is more readily available,” he explains. “For forecasts over longer horizons, directional accuracy and visibility over material movements are the areas to focus on.”
When it comes to overcoming challenges and improving forecasting models, Nicolaides says there are four main things he advises treasurers to do. The first is to regularly re-evaluate the forecasting data that is required after business needs change. “People at the centre should always think about and review what information they really need to get out of their business units in the various countries and continents,” he says.
The second – perhaps the biggest forecasting challenge that corporates face – is educating the various global teams that are necessary to accurate forecasting. “Treasurers rely on all these teams to give the information needed to forecast, but unless these teams are educated on all the reasons the information is needed and is valuable to the organisation, and what exactly is done with it, they’ll be less inclined to ensure it’s accurate,” he explains.
That then leads into his third point, which is to reinforce the positive behaviours around forecasting by monitoring the forecasts that are given. For example, a treasurer sitting in Europe who asks for a forecast from a team member in Asia every month should be following up on the results of that forecast. “If they never ask, ‘last month the forecast was 60% out, why was that?’, and never give that feedback, then the treasurer in Asia will never have the incentive to give credible forecasts,” he says. But if the Asia treasurer knows that they’ll be asked about it, they will think twice when putting in the forecasts.
This not only ensures every team member involved in the forecasting process is working accurately and effectively, but also gives valuable learning experiences. “I always say that monitoring should be done not just by comparing forecasts and actual data, but also by comparing originally forecasted data and revised forecasted data,” he adds. This is especially important for long-term forecasts which can be significantly affected by rapid changes such as COVID-19.
Lastly, Nicolaides believes that it’s important to build forecasting accuracy into people’s key performance indicators (KPIs), as another incentive to make them care about making the forecasts as accurate as possible. “If it has no impact on their KPIs, they will probably never spend the relevant attention on it,” he says. In order to do this, of course, management support is essential. “Management need to give the relevant support to forecasting and dictate within the organisation that this is a very important part of the business that needs to be as accurate as possible.”
As with many aspects of treasury, technology is an essential part of forecasting. Abdul Jabbar explains that cash flow visibility at AMH was significantly improved after the establishment of an in-house bank (IHB) – which saw AMH named Highly Commended Winner in the First Class Relationship Management category of the 2019 Adam Smith Awards Asia. Key to the IHB is the automated reconciliation structure that simplified liquidity management and allowed for improved cash visibility. “By leveraging the IHB, we have managed to advance our forecasting models by having precise information of liquidity positions, improving both capital and liquidity management,” he explains.
The automation of processes is a key step in adopting more technology to improve forecasting, says Nicolaides. The automation of manual processes can allow treasurers to regain valuable time. But more important, he says, is the adoption of artificial intelligence (AI) and machine learning (ML).
ML has the ability to identify data anomalies and ignore them, recognising that they are the exception to the rule and not the norm. ML can also identify flexibility in payment times. For example, if a company makes an annual payment every January, but is slightly short of cash one year, through ML the system might identify that in the previous five years the payment has actually had flexibility between January and March. ML can also help to identify committed and avoidable outflows, which may help when companies are strapped for cash. Lastly, it can identify surplus cash for investment over a specified period of time.
Indeed, ML has enormous potential in the treasury space more generally: in Deloitte’s 2019 Global Treasury Survey, 61% of respondents said ML was ‘critical’ or ‘important’ to treasury, and 56% believed they were ‘well versed’ in or had a ‘general understanding’ of the technology.
Despite these encouraging figures, 66% of respondents said that they weren’t using ML at all, and there was an equal 15% for both having ‘a whole dedicated team or a few in-house experts’, and ‘currently educating existing staff or looking for new talent’. Of those that have adopted AI and ML, 46% said they had either already implemented or were very likely to implement it, believed that it was especially relevant to ‘forecasting, financial planning, back and stress testing, and model validation’.
“In this space, the approach we are taking is evolutionary,” says Suvrodeep. “It involves breaking down the forecasting value chain into its constituent parts and then assessing where AI or ML can solve real-world problems, whilst not expecting treasury staff to be data scientists themselves!”
Suvrodeep notes that AI and ML still has a way to go, but that companies with strong and consistent seasonality in their cash flows can find that the technologies offer a significant forecasting edge. But, says Abdul Jabbar, it’s important to note that history is not always the best indicator of the future. In order to ensure the most accurate and effective forecasting data provision and analysis, it’s prudent to also look at unexpected scenarios.
The COVID-19 pandemic, and the resulting turmoil, has increased the focus on technology, says Abdul Jabbar. “It has highlighted the need for greater digitisation, real-time transactions and data to monitor cash flows at any point, and the ability to react immediately when conditions warrant a change.”
Suvrodeep agrees and says that the pandemic has drawn attention to the importance of contingency planning, which goes hand-in-hand with liquidity planning exercises such as cash forecasting.
Nicolaides believes the future will be a mixture of both. “For me, the future is having a cash flow forecasting process using technology and with a level of accuracy that cannot just give you what you need for making day-to-day decisions, but can also act as a very useful tool when something unprecedented happens, like COVID-19.”
Regardless of how cash forecasting develops, it’s clear that utilising technology to improve the forecasting process is going to become unavoidable for treasurers. Adopting it sooner will allow companies to progress with technological advancements, rather than having to catch up after.