The problems of cash forecasting
The ability to accurately forecast future cash inflows and outflows has always been one of the biggest challenges for treasurers and finance professionals the world over. The situation has been further exacerbated by the tightening of capital markets in the last decade which has made cash a scarcer resource!
In this blog, we will explore how one of the hottest emerging technology in recent years – Artificial Intelligence – is being leveraged to improve this process.
Problem components of cash forecasting
There are various key drivers of cash flows, but many of the large ticket items like CAPEX, business development, debt servicing, etc. are planned well in advance and generally easier to forecast several weeks out. The biggest challenge is with operational flows – by far the biggest drivers of cash – which are often volatile and hard to forecast with any reasonable accuracy beyond a couple of weeks! This volatility and uncertainty is because more often than not, these flows are influenced by external and/or macroeconomic factors that are outside of the organization’s control!
The biggest conflict is between shared service teams that are unable to commit to firm forecasts typically beyond 4-6 weeks out and treasury departments that generally like to have an average of 3-6 months’ view on cash so they can plan capital position efficiently!
AI driven solutions for cash forecasting problems
So what role can Artificial Intelligence (AI) play in solving this puzzle?
The key element is DATA – using various data points over multiple periods to try to forecast future events – following the popular maxim "The best predictor of future behavior is … past behavior"!
Most manual or semi-automated forecasting models use limited data points and are also influenced by the sentiments of the individuals creating and using the forecasts – and with employees leaving or changing roles, a lot of the intelligence in the process is reset!
With the computing power of today’s computers, AI models are able to slice, dice and analyze large volumes of relevant data efficiently and effectively to create more effective and data driven forecasting models – and the data adds over time making models increasingly intelligent.
Let us take a quick look at some of the main data points that would be relevant in this scenario:
- Internal data – this would include items like sales data, purchase history, committed orders, cash cycles, payment history of customers, etc.
- Market data – this could include macroeconomic indicators, industry specific factors, currency rates, etc.
- Stakeholder data – this could include credit ratings of key stakeholders, stock prices of critical supply chain partners
The above list is merely indicative – our AI solutions assimilate many more data points within the models!
The various specified data points are blended together to run complicated and complex computations to factor in seasonal swings, identify and adjust any aberrations – all with an aim to provide more accurate data driven forward looking views!
Before concluding, I would like to state that AI systems are not intelligent enough (not yet at least!) to replace human intelligence, but exist to serve and assist humans in making more informed decisions based on facts.
At the end of the day, AI systems are merely computational tools and do exactly what they are programmed to do! What data points to use – how to use them – what to output – is all driven by the functional process owners. You get to decide how intelligent your AI system is and you get to define what intelligence means for you!
Moreover, even though numbers don’t lie, a successful professional with years of experience can often see things that sometimes gets hidden or doesn’t show up in the data – computers still haven’t developed a sixth sense! Many AI models actually have inbuilt functionalities to add specific manual adjustments or override some components of the calculated logic. Bottom line is that the onus is eventually on the subject matter experts to interpret the AI results and decide how it is used!
To sum up, the use of AI on treasury and finance is still evolving, but there is a tremendous potential for its use especially in cases where data and past behavior can be analyzed to provide views on future behavior.
Techno Functional Speak
The AI algorithm is a rule-based system developed with the particular firm and industry in mind. The statement of changes in cash flow & market conditions provides another accounting model that has been used by teams to optimize cash flow. The format of this statement can be modified for decision-making purposes. For example, the changes can be grouped by changes in assets, liabilities, stockholders' equity, cash inflows, and discretionary and nondiscretionary cash outflows. Placing the assets in these categories allows the system to make recommendations based on the information that describes that category of outflows. As with the ratios, the actions and the parameters on which the parameters are based are a function of the firm's policies and unique needs. Accordingly, LeanCash™ Algorithms would have to be adapted to the unique situation.
Medullus LeanCash™ is a tried-and-tested solution that draws upon lean concepts to streamline Corporate Treasury by eliminating waste in processes and functions which helps to manage liquidity and risks in the most optimal way.
Our products and services are geared towards making improvements in Treasury & Finance processes to bring big gains with minimal impact. The low hanging fruit is sometimes difficult to find in a large organization with myriad of processes and systems. Take the 1st step in identifying these high value/low risk improvements – get your LeanCash score, benchmark against other organizations that have obtained their score and improved them. It does not cost anything to find out where you are as compared to industry-best-practices!
Blog post by: Tejnain Singh. Reach Tej @ email@example.com
Tej is a seasoned treasury and finance professional with over 17 years of extensive experience across organizations like GE, Deloitte, SABIC and Valeant Pharma. He has held senior positions in treasury operations and cash management and has led multiple strategic initiatives in the areas of payment automation, treasury system implementations, business process simplification and shared financial services among many others. Tej is a Chartered Accountant from India.
Warning: count(): Parameter must be an array or an object that implements Countable in /homepages/12/d529376851/htdocs/medullus2018/wp-content/themes/vamtam-consulting/templates/post/meta/tags.php on line 7