Taking the Guesswork Out of Liquidity Buffers

Setting up a cash buffer can be a vital part of liquidity planning, but the difficulty in knowing precisely how much to set aside can lead to cautious over-estimation, despite the inefficiency of doing so. Is it simply better to be safe than sorry, or can technology be used to reduce or even remove the need for buffers? TMI explores the options.

Cash buffers seem like the only practicable way to mitigate the risks inherent in an unstable trading and market environment. Yet every treasurer understands the potential inefficiency of holding ‘just in case’ cash. So is there a more effective means of calculating and holding buffers while keeping risk under control?

In practice, the need for a buffer is relative to each company’s cash position, says Hubert Rappold, Chief Sales Officer, Nomentia. “For companies that are already cash rich, just having a stand-by credit facility may be sufficient, then they could place their excess cash in longer-term investments with better interest rates because a real cash buffer in the current inflationary environment will be losing its value quite drastically over time.”

Many of the companies Nomentia works with try to avoid having cash buffers, instead focusing on “really accurate cash flow forecasting” and then seeking “the best possible horizons” for placing cash in short-, medium-, and longer-term deposits with their banks or other financial providers.    

The need to achieve better cash flow forecasting is a view also espoused by Jeannot Jonas, Assistant Treasurer, Sonoco, and a global treasury executive with more than 30 years’ corporate treasury experience under his belt. But he tempers his belief with the understanding that the key to better forecasting is data quality and timeliness, which he says is not always forthcoming.

“From my previous experience, the data has either not been as good as it could be or has been difficult to obtain because the company had multiple ERP systems,” recalls Jonas. When this is the case, he says treasury is necessarily reliant on information provided by other functions that may not have the same goals. 

Indeed, if it’s financial planning and analysis (FP&A) data, it will be skewed more towards cash flow forecasting than cash forecasting, Jonas making the distinction that the former is intended for longer-term planning, while the latter aims to provide a handle on the immediate cash position. 

While under legacy conditions the potential for error in buffer calculation is “practically baked in”, Jonas believes technology could bring sharper focus. However, he says treasurers will have to wait for the right tools. “When technology such as AI is used to extract the information needed to create a robust cash forecast, then those buffers can probably be reduced. But until that happens – and I haven’t seen it yet – it remains a best guess as to how much we think we need to run the business with no disruptions over a specified period.”

For Veikko Koski, Founder and CEO, FinanceKey, the common legacy of siloed systems and lack of collaboration create a need to hold extended cash buffers. “Often treasuries lack visibility over payment outflows as separate teams take care of the day-to-day payment operations,” he explains. “Batch payments are often processed via an ERP, going directly to the bank. When that task is outsourced to an external shared service organisation [SSO], its often rapid turnover of staff can limit knowledge of payment orders and currency cut-offs. It’s why some level of buffer is always needed for urgent ad hoc payments.” 

One of the main drawbacks to holding a substantial cash buffer is that treasury cannot optimise interest income and efficient liquidity allocation, continues Koski. “When holding cash in foreign currencies, depending on the FX carry, treasurers could earn even more for swapping balances to their home currency and investing in, for example, MMFs.”

Cash clarity

Of course, there is a matter of defining ‘cash’. Does this include short-term investments or is it solely cash in the bank on deposit? Arguably cash has to be highly liquid and for Jonas, while mutual funds such as MMFs may be deemed as such, he is not convinced they are as liquid as some think. 

However, with bank deposits, he is fully aware of the spectre of counterparty failure that is raised from time to time, as recently happened with Silicon Valley Bank (SVB) and Signature Bank in the US. Of course, SVB had a highly targeted client base in the tech sector, nonetheless both banks’ difficulties stemmed from investment policy and governance issues that could befall any bank.

“Cash protection is a matter of sound investment policy,” Jonas reminds other treasurers. “It should clearly dictate your bank exposure limits.” For many businesses, he believes this will mean banking partners having at least one A-rating from the main agencies. And given current uncertainty within the banking sector, an immediate revisit of related policy, certainly to protect buffers, should be in progress now.

Cash sitting in a bank account may, as interest rates continue their upward trajectory, generate some useful interest (the European Central Bank’s deposit rate moved to 3.50%, as of June 21). But rather than just sitting in an account doing very little, treasurers should be thinking about time (or term) deposits, says Jonas. These can be overnight or longer and will be liquid. But, he warns, if the cash is needed before maturity, an early withdrawal fee may be payable, although the power of a strong banking relationship may override this.

The next consideration is where to hold the cash buffer. A global business will probably establish regional buffers, says Jonas. As an example, in a previous treasury role, in addition to cash held in the respective currencies of the UK, the US and Hong Kong, he says the company also held a euro account in Europe. As with most buffers, the levels were maintained as low as possible to keep within limits agreed with the banks. This created a target balancing act for treasury, which for the euro account became notably challenging.

“We had  negotiated with the banks that up to a certain amount of cash deposited there would be no charge while negative interest rates applied. These limits were initially OK, but then the banks decided to lower them because they said they were ‘not meeting their business objectives’. That forced us to redirect large amounts, with much of our euro cash being converted and repatriated to the US.”

The regional model for cash buffers clearly comes with potential issues, albeit not insurmountable, and for an MNC, it’s usually beneficial to keep the cash close to where it will be needed. “But it’s an active process,” notes Jonas. “Reaching target balances can be automated, but monitoring them to ensure you’re not exceeding or dropping under your targets requires someone to review them and make a decision.”

This is one of the reasons why Koski recommends to clients that buffers are centralised as much as possible, based on agreed limits and thresholds per currency and bank account. “Central treasury teams are in the best position to invest excess cash effectively, based on investment limits and policies, while regional treasuries can be given a mandate to manage their cash positions,” he suggests. “Required cash buffers can then be invested in MMFs, deposited short term or held in an investment account with a competitive yield.”

Educated guess 

Liquidity buffers can be defined in several ways, for example as ‘time buckets’, setting target liquidity amounts available in five, 30, and 90 days, say Koski. “Estimated and prior outflows are a good guide when setting buffer metrics, with cash availability and effective risk management prioritised over yield optimisation.”

Short-term cash buffer calculations can easily be automated where there is visibility over banking data, payment outflows and treasury transactions. From a mid- to longer-term perspective, high-level investment adjustments might be needed to cover potential activities such as M&A activities and dividend payments.

While helpful mechanisms can be implemented, reaching a precise buffer figure is rarely possible because there are so many intervening variables, says Jonas. He believes that calculation has always been more of “an educated guess than anything scientific”. That being said, his is an approach that has evolved progressively, using actual flows across the enterprise to anticipate disruptions. 

“As a treasurer, the worst thing that can happen is that suddenly there is not enough cash for an entity to settle its payroll first, and its suppliers second,” explains Jonas. With this in mind, he says payroll can be taken as a generally steady figure. To this can be added the typical values of payment runs. The most significant unknown is the value of collections. “This depends on factors such as how efficient the credit function is, and the quality and predictability of the customer base in terms of their payment behaviours, and as we know, market conditions can sway this considerably.” Even here, historical figures can be manipulated to provide a working buffer but, as Jonas comments, “it’s not very sophisticated”.

Koski, having already cited no or limited visibility over outflows as the driver for setting cash buffers, all too often sees “siloed systems, a lack of internal collaboration, and under-resourced treasuries causing bottlenecks in driving more efficiencies investment allocation”.

With Jonas recalling multiple times where provided outflow data was either significantly over or under what his bank data suggested, treasury can indeed be frustrated by the efforts of others. Withholding or delaying the flow of information has often been the likely cause, but he says treasurers experiencing this might not get to the root of the issue without a determined enquiry.  

“It usually comes down to systems,” he states. In an ideal world, a global company will be running the same ERP, and the same instance of that platform in every location. This affords treasury a better grip on its data because it can acquire it without relying on anyone else. “As treasurers, we still need to match what our systems tell us about payables and receivables with bank data – a close match, say within 5%, is acceptable – but this is the real world, and companies often have more than one ERP, and multiple instances of these.”

Superior technology  

If legacy system architectures are generally at fault, is more technology the answer? Rappold says treasurers are actually in a really good position today because the technology to assist is already available. AP/AR and order backlog data within an ERP can be extracted using automation tools, he states. “This is a big first step where basic integration of ERP systems using modern APIs or even old-fashioned secure file transfer protocols [SFTP] can be an enabler, giving access to this data for use in forecasting processes to generate accurate results.” This could be the base point for the subsequent application of predictive analytics.

Here, Rappold explains that historical payment or order patterns can be combined with industry-relevant, forward-looking macro-economic factors to give greater clarity. Technology, he notes, “can take on a lot of the painstaking work required to gather and model data, and produce accurate and trusted forecasts”. 

Trust may be reinforced by deeper core system integration and the application of predictive analytics tools. The capabilities of analytics tools to work with large volumes of data and automate processes has improved considerably over the past few years, comments Rappold. He also sees products such as Microsoft’s Power BI as having enhanced the whole reporting aspect by being deeply integrated into multiple company functions. 

“It’s data that treasury can access to create powerful cash flow reporting output. Where previously much of the time was taken up with background processes, leaving little time for analysis, now the reverse is true. Forecasting quality can improve, which in turn enables a more accurate view of how much cash is needed, and how investment horizons should be best set.”   

Koski believes the correct technology can help teams “to collaborate better, flag cut-offs for payment orders, and provide a single view of short-term liquidity and longer-term forecasting”. The starting point, for him, involves banking APIs. From a pure cash management perspective, he says these can deliver real-time cash visibility and data aggregation, even across a globally distributed group of entities. “It can lead to live dashboards, set up to provide structured views on daily liquidity needs, and help treasury to plan liquidity and optimise their short-term cash buffers.”

With the major TMS platforms all now integrated with a host of banking API libraries, Jonas also sees the greatest improvement to cash buffer calculation as coming from this direction. Of course, this assumes the TMS is integrated with, or at least capable of extracting data from, existing ERPs. It also must connect (via APIs or SWIFT) and aggregate transactional data from as many banking partners as possible into a single or limited panel of primary banks, which may be regionally or centrally placed, as established above. 

In addition, while Jonas has already stated that AI is not a technology he has seen in action in this context yet, he nonetheless sees it as a feasible solution, with many TMS vendors now incorporating this into their platforms for other purposes. Indeed, as Koski notes, AI/ML capabilities can “flag inefficiencies within liquidity management and help optimise cash buffers driving higher returns on cash”.

However, with some treasury departments still relying heavily on spreadsheets, would even the mention of AI or predictive analytics be too much to bear? “The good thing about predictive analytics is that you can start small,” assures Rappold. “Companies can scope one or two categories of line-item data, and can arrive at a deep dive in just a few days,” he says. “Then they will know if it is for them or not. But once they do know it works, and that it could put their cash forecasting several levels above what they are currently achieving – with much less time and effort invested –there is no way back for them, they will want to use it!”

When approaching predictive analytics, the bottom line for Rappold is the absolute need for at least three years’ clean historic cash flow data to feed the algorithm. This data set should reveal most likely anomalies or outliers and any seasonality, which can subsequently be accounted for. Unfortunately, incidences where this span of data is not available are common, he says, with M&A activity or a prior lack of focus on cash flow data often forcing fractures in its continuity or accessibility.

Although without this range of data the results will not yield the expected forecasting improvements, even this may not be a complete show-stopper as it can often be reconstructed from ERP or accounts systems, says Rappold. “It just makes the task harder. But any vendor suggesting that only a couple of months’ worth of data are needed to generate a cash flow forecast will produce figures that are unusable for treasury. If any company is told that this minimal level of cash flow data is OK for an AI application, the best reaction is to walk away.”

The end of buffers?

So, does technology have the capability to completely remove cash buffers? “No,” states Jonas. While he recognises that better system integration and connectivity – and advanced tools such as AI – have great potential to help optimise buffers, cash ‘events’ can happen at any time. 

“During an M&A or divestiture, for example, sometimes a payment has to be executed before a certain deadline. If your cash is locked into an MMF or a time deposit, you will not get those funds until the end of the day. Where does the cash come from? In the US, there are no overdraft facilities. You could go to Europe but that will incur a major unplanned interest expense. In my view, this is where cash buffers are needed.”

For Koski too, cash buffers will remain for now. “Human adjustments might still be needed with cash buffers, especially where fragmented systems and data flows are not enabling the adoption of intelligent technologies just yet,” he observes. “As systems become more integrated, and AI/ML is applied to treasury and banking innovations, the need for buffers will be considerably lessened,” he believes. “But at some level, companies will still need buffers because the complexity of treasury and payments operations is increasing in parallel.”

And while Rappold believes companies now have better technology options and “should be able to almost get rid of cash buffers”, he too accepts that there will always be some level required. “Try to keep it at a minimum, and keep it centralised, but be fully aware that it will potentially cost you quite a lot of money to hold, and that you need to calculate whether it’s cheaper to have the buffer or a committed credit facility with your bank.”