Sometimes the simplest tasks can be made problematic by the most obscure hurdles. This is particularly true for data management.
Imagine a Database Administrator is charged with sorting a range of historic texts by estimated publishing date. Let’s imagine these texts have estimated publication dates ranging from the early 18th century to the mid 20th, and that the headline information (including dates) for these texts are all stored in Excel. At first, you might assume this would simply require selecting the column containing the dates, and sorting by time.
However, in Excel the world began on the 1st January 1900 and is due to end on the 31st December 9999. After researching how to overcome this problem, the Database Administrator has installed a handy plugin that captures pre-1900 dates. Then the next hurdle presents itself, the eleven “lost days” between 2-14 September 1752. (Which arose from the Julian to Gregorian calendar switch.) By now the “simple task” has revealed itself to have far more steps than originally estimated.
A good data set is a large data set. Large data sets, however, come with increased complexities. As we saw with the example above, even fairly simple data sets can throw up difficulties. The larger the volume of data, the harder these problems are to unpick. The main issue with large data sets, though, is the increased likelihood of anomalies.
In short, the larger the data set becomes, the more difficult it becomes to work with. This catch-22 has led to increased investment in the field of data cleansing. A task that is necessary before any data mining activities, which is where the real value-add is found, can take place.
The data sets that are used in a cash forecasting process are no different. Just like all other forms of data analysis, cash forecasts suffer the same decreased output quality if the data integrity is compromised.
In the field of cash forecasting, having a clean set of historic data is enormously powerful as it forms a baseline from which accurate forecasts can be made. The data sources that feed into a cash forecasting process can be broadly broken into two categories, sources of actual data, and sources of forecast data.
In most cases the actual cash flow data is sourced from:
Whereas forecast cash flow data is usually sourced from:
The extraction of this data can either be automated, through the use of specialised cash flow forecasting tools, or it can be drawn manually. Because of the range of data sources that input into the process, it is important that the data is standardised into a common format before it is used. A common cause of data defects is that incorrect information has been input, possibly due to human error (if the process is manual).
As previously stated, this standardisation step can be an automated part of the process, but it is important that input feeds are mapped carefully, by someone with the appropriate expertise, to ensure the correct fields are captured.
Ultimately, the goal of cleaning a historical cash flow data set is to create to a representative view of what’s happened in the past, which in turn provides a reliable basis for building a view of the future. Key to this exercise is removing items and once off cash flows that will not be repeated in the future as they will ultimately pollute the modelled forecast output.
Things to watch out for when cleaning a cash flow data source include:
Any information exported from an ERP or accounting system could potentially contain accounting journals such as reversals or currency adjustments that will impact the quality of the underlying data that will be used for forecasting. These journals don’t represent underlying business activity and will need to be removed.
Acquisitions & Divestures
Acquisitions are likely to be the biggest cash out flow, or series of cash flows, leaving a business over the course of year. It’s highly unlikely that the amount spent and timing of an acquisition will be repeated in the future and therefore it will be necessary to remove these cash flows before using the data set for forecasting purposes. The same is true for divestures.
Investment Capital Expenditure
Aside from acquisitions, investment capital expenditure, particularly for spend on large once off projects, can be the lumpiest cash out flows over a period of time. Even if capital expenditure levels are expected to remain consistent with previous periods, the amount and timing of this expenditure will likely be very different to previous periods. Which in turn warrants their removal from the data set.
Debt drawdowns, repayments and refinancing can have a huge impact on total cash movements over a period of time. Typically, these movements aren’t representative of day-to-day business activity and should be removed when using the data set to model future activity.
In most mid to large size companies intercompany cash flows between business units can sometimes equal external cash flows in total volume over a period of time. Of course, intercompany shouldn’t have an impact on net liquidity but when analysing a particular business unit or segment it is important to remove non-trading intercompany movements that are unlikely to be repeated in the future.
In the regular course of business unusually large cash movements, be they payments or receipts, will occur for a number of reasons. The winning of a large new customer account could lead to a once off large cash receipt, for example, or the prepayment for raw materials to secure a discount could lead to large once off payment. The removal of these types of cash movements may require a little more judgement than the previous items but, in some cases, data is improved when they are removed.
Owing to the anomalies listed above, data sets often undergo a “cleaning” exercise before any in-depth analysis is performed. However, as noted, care must be taken not to conflate data anomalies with data defects. While it goes without saying that removing data defects improves data quality, callously removing all anomalies might mean removing some important signals from the data.
In cash forecasting, data cleansing is a continuous process. The various measures that a large organisation must take, such as those listed above, mean that there are regularly factors that cause sizable distortions to these data sets.
Specialised software simplifies the continuous data cleaning process, along with all other elements of the cash forecasting process. As referenced in the section on data sources above, automating the process expedites data cleaning. Elements such as intercompany cash flows should always net to zero. However, with a manual process this balancing task can be hugely time-consuming and inaccurate. Specialised software, with a dedicated counterparty driven intercompany tool, simplifies this task to a touch-of-a-button exercise.
This simplification enables the person analysing the data to quickly review and amend outputs and therefore take the right anomalies into consideration.
As we discussed in a recent article on the analytics skills gap, it is important to have people with the right skills at the right stages of the process. While not an absolute requirement, some companies may choose to recruit a Database Administrator (DBA) to ensure that all data is formatted and structured properly as it goes into the model.
In any event, the person charged with managing the database should be familiar working with large data sets and have a good knowledge of database theory and database design. This expertise will help with the identification and rectification of defects in the data, as well as managing the integrity and security of the data as a whole.
Once all checks have been made, appropriate actions have been taken, and everybody is confident that the data is of sufficient quality and integrity, it can be loaded into the forecasting model. The focus now switches to the output side of the process, i.e. on the reporting, forecasting, and analytics. Here, quality is improved through a process of careful measurement and adjustment. A gradual process, but one that becomes increasingly valuable to the company as these improvements are made.
Going back to our hypothetical Database Administrator, struggling to sort historical dates in Excel, their best option is to re-format the database. They’ll need to sort their dates into separate columns; one containing year, another for month, another for day. Then filter with a cascading sort, factoring in all three columns.
For Corporate Treasurers, however, we can be of far greater assistance. We provide dedicated cash flow forecasting software to large, multinational organisations and have extensive experience helping clients across a range of industries.
In addition, to assist those considering updating their old processes, we have written a whitepaper which outlines the steps involved in setting up a cash forecasting process. If you have any questions on this, or would like to see a demo of our software in action, please do not hesitate to contact us....
A well-designed cash flow forecasting template can dramatically improve the quality of reporting outputs. However, this is not a one-size-fits-all situation. For a template to be well designed it must be tailored to align with the business objectives of the cash forecasting process.
As one of the key stages in setting up a cash flow forecasting process, treasury teams often spend considerable time working out how their forecasting template should be designed. To assist, we have written this post which explores how a cashflow forecasting template should be designed.
In this post we will outline: what a cash forecasting template consists of, the key decisions that need to be made when in the design process, as well as providing some useful tips.
A forecasting template (also known as a cash forecasting model) is the reporting structure and associated logic that produces the required forecast output. A forecast template has two dimensions and typically collects two types of cash flow data.
The two dimensions of a forecasting template are:
The two types of cash flow data in the template are:
The level of granularity chosen for both the reporting periods and cashflow classifications will be determined by the overall objectives of the forecasting process.
For example, if the business objective is short term liquidity planning, the template will need will need a reporting granularity of at least daily. If the reporting periods are any broader than this, for example if reporting periods are presented weekly, short-term liquidity shortfalls may be missed.
These are important considerations as reporting granularity that is too fine can muddy the waters and disguise important trends in the data. Conversely, cashflow classifications or reporting periods that are too broad for the objectives might mean that precise but important signals are missed.
In addition, to support decision making in liquidity planning, the forecast will also need to be as accurate as possible. As forecast accuracy generally decreases the further into the future the forecast stretches, the forecast horizon will need to be relatively short. In our experience forecasts used for short term liquidity planning do not usually exceed a horizon of 10 business days.
To assist those currently making decisions on which is the appropriate time horizon for their objectives, we are in the process of releasing a short series of articles on the subject. The first reviews the practical uses for the 13-week cash flow forecast.
Model structure examples
The table below outlines some examples of illustrative business objectives. It includes which forecast horizon, reporting date granularity, cashflow classifications, and frequency of creation would be appropriate for each.
Reporting Date Granularity
Frequency of Creation
Short term liquidity planning
10 business days
High level flows and balances
Twice a week
Interest and debt reduction
Management reporting categories and flows
Covenant and key date visibility
To next significant reporting date (at least)
Management reporting categories and flows
Weekly, more frequent approaching key date
Liquidity risk management
Weekly for 13 weeks, then monthly for 3 months
High level flows and balances
While the above examples are illustrative, ultimately, the structure of the template chosen should be able produce the range of reporting outputs needed to meet the business objectives.
Capturing actual cash flow and balance data as part of the overall forecast reporting process is also an important consideration.
Similar to the granularity decisions, how much actual data should be included will be decided by the business objectives. Actual cash data allows the forecast model to produce both trend and variance analysis which are very useful pieces of supplementary analysis to the base forecast.
As business objectives help to define the scope of the various different elements of the cash forecasting template, having clearly defined business objectives is the first step in designing a forecasting template.
This post is an extract from the guide we recently produced which covers all aspects involved in setting up a cash flow forecasting process. Please follow this link to the cashflow forecasting setup guide, which we welcome you to download. The guide discusses: what is involved in setting business objectives, how to set the process up, as well as what comes after go-live....
Data visualisation is an important step in any form of data analysis. Presenting data in a graphical format often helps to highlight trends, identify anomalies, and uncover insights that will be missed when simply reading through raw data. The same is true when analysing the data produced as part of a cash forecasting process.
To illustrate how data visualisation can help with data analysis, this post will review three potential ways to visualise cash forecast data. These three options are;
However, it is important to note that these examples are not exhaustive, and that many other formats of visualising cash forecasting data are available.
A cash walk through visualisation breaks the journey from opening cash balance to closing cash balance into a series of steps that identify the most significant contributors/detractors to the net cash balance.
In the example below, we can see the three headline cash inflow categories highlighted in green (Customer Receipts, Investing Inflow, Dividend Receipts). The four headline outflow categories are highlighted in red (Supplier Payments, Tax, Payroll, Debt Payments).
One of the key benefits of this method of data visualisation is that it highlights the extent to which each category of cash flow affects the cash balance in an easy to understand visual. This means attention can be focused on the element(s) that will have the greatest impact. For example, in the graph above we can quickly see that supplier payments (which total $10million in outgoings) cancel out the positive contributions from customer receipts, investing inflow, and dividend receipts combined.
The graph below visualises a comparison of forecast versus actual data for a variety of reporting entities. It breaks the measure into two categories, percentage variance and amount variance. (Here, the different reporting entities’ variances are transposed to a common currency, US dollars).
One of the key benefits of this method of data visualisation is that it allows material variances to be quickly identified and put into context with other entities’ percentage accuracy. This means that attention can be paid to the variations that have the greatest impact. For example, in the instance above, Brazil had a major deviation between forecast and actual (92%) but this only equated to a variance amount of $1.3million. Whereas in China, where forecast accuracy was better (though still poor as China had an 83% divergence from the actuals), the value of this discrepancy was $4.0million. This means that an increase in the accuracy of forecasts produced by the Chinese entity (where there is considerable scope for improvement) would have a far greater impact on overall, company-wide forecast accuracy than focusing on the most inaccurate forecasting entity.
A Time Series Visualisation allows variance analysis to be carried out across multiple forecast versions and it is usually best displayed in a table such as the one below.
Here, the table shows how multiple forecast versions compare to the actual data (in bold):
The advantage of capturing the data in this format is that it enables easy comparison of multiple forecast versions. This means that trends in the data can be easily identified and, where appropriate, addressed to improve accuracy.
For example, the table above shows that while forecast accuracy broadly improved as the forecast horizon reduced, there remained a tendency to underestimate the actuals. If we review the submission made in Apr-18, we can see that the actual figure captured for the March 2018 closing balance was 154,000. Included in this submission were forecasts for the April 2018 closing balance (which it underestimated by 19,000), a forecast for May 2018 (underestimated by 100,000) and a forecast for June 2018 (underestimated by 134,000).
As there is enough data captured (in the table we can see 36 forecasts and six closing balances) it can by hypothesised that this is not a natural variance but rather a trend that is the result of a fundamental area of the forecast being miscalculated. This might be higher than expected sales volumes, lower overheads, perhaps a recent reduction in business rates was secured by the company but this wasn’t reflected in forecast calculations, for example. In any case, once this trend is identified its underlying cause can be investigated, identified, then corrected. Thereby improving forecast accuracy.
As mentioned at the beginning of this post, data visualisation is a necessary part of any data analysis. Presenting the information graphically enables an analyst to quickly spot trends, identify anomalies, and helps to uncover the underlying causes of any flaws in the process. For a treasury team, this means that these insights can then be presented back to the business in a clear and concise format for easy interpretation by senior management. This therefore positions treasury as a strategic department within the business.
As with all parts of a cash and liquidity forecasting process, data visualisation options can be greatly improved with the use of specialised software tools. We have helped many companies to automate their cash forecasting processes, to increase forecast accuracy, and set up a new cash forecasting process altogether. If you are looking to improve the impact of cash forecasting in your organisation, contact us to see a demo of our specialised software....
The CashAnalytics Summary Series takes major Central Banks’ extensive reports (often well over 150 pages), and summarises them into five minute reads.
This release summarises the 5th edition of the ECB Economic Bulletin for 2018, which was released on the 9th August.
The key points covered in the report include:
To read the summary of the ECB Economic Bulletin, please click here....
When setting up a new cash flow forecasting process, it is vital to select the right time horizon. This article will form part of a series we are producing which discusses how various different time horizons can be used in cash forecasting. This post will explore the 13 week cash flow forecast in detail.
Selecting the right time horizon for a forecast depends very much on the business objectives of the forecast. By exploring the uses, benefits and drawbacks of the 13 week cash flow model, you should be able to ascertain whether it is suitable for your business needs.
We chose to start this series of articles with the 13-week cash flow model because of its broad uses. Corporate senior management, including the CFO, are familiar with 13-week forecasts as they will use them in various forms of quarterly strategy and planning sessions. Similarly, investors often prefer the 13-week horizon as it gives them a reasonable view of the health of the company in which they are invested. The same holds true if a company finds itself in a distressed situation. In this instance a 13-week cash forecast should be able to identify when and how any potential liquidity shortfalls should hit a company. This therefore offers the ability to prepare for, if not rectify, any issues before they occur. A 13-week cash forecast is also often a requirement for bank reporting, or when submitting any lending requests. In short, while a 13-week cash flow forecast isn’t suitable for all requirements, it does have a broad range of applications. We will discuss these in further detail below.
Weekly cash forecast showing actuals (far left column) and forecast figures (columns to the right).
Striking a balance
The time horizon of the 13-week cashflow model is short enough to support agile, tactical decision making, but also takes a long enough view to drive longer term decisions. The 13-week cash flow forecast also helps strike a balance between accuracy and range. It is a universal truth in forecasting that the accuracy of a forecast degrades the further it extends into the future. 13 weeks provides enough sight to have an impact on strategic decision making, while remaining short-term enough to be able to provide a high degree of accuracy.
Planning for the medium-term without interrupting plans for the longer term
13 weeks of visibility allows a CFO, treasurer, or financial controller to make medium-term cash management plans. This could include debt drawdowns and repayments, short term investments, or other management decisions.
Taking a longer-term view, while 13 weeks is sufficient for medium term and month-to-month cash planning, it won’t overlap with longer-term plans which can extend years into the future. Because of this, there is little reconciliation required between plans. Medium and short-term planning decisions can be assisted by the 13-week forecast, whereas longer term planning remains in the purview of the overall business planning process.
Avoiding the short-term planning gap
Most planning processes in companies produce monthly forecasts and budgets, offering limited short-term visibility. A 13-week cash flow forecasting process is usually broken down into weekly reporting periods, therefore offering four times greater granularity. This means that any short-term planning shortfalls can be addressed with a detailed look-through.
Liquidity risk forecasting
Risk management is a key part of any cash forecasting process. Forecasting with a 13-week horizon should be accurate enough to identify any potential liquidity issues, while still offering enough time to take action to resolve those issues. For example, if a potential liquidity shortfall is identified with 10 weeks’ notice, the treasury team have ample time to prepare by arranging bank funding or reviewing intercompany lending options.
Matching business requirements
By covering a full quarter in scope, a weekly rolling 13-week cash flow forecast will always provide cash balance visibility on the next key reporting date or next quarter end. This is extremely valuable for the business, as senior management will always want reporting on cash balances that coordinate with their own key reporting dates.
Suitable for banks and investors
Depending on the ownership structure, debts, bank relationships, or investment status, a company may be compelled to produce a liquidity forecast with at least 13 weeks duration in place. It’s a key measure of good financial control and, from a bank’s or investor’s perspective, it provides clear visibility of the company’s working capital.
Short term liquidity planning
A shorter forecast horizon (perhaps of 10 business days) might be better suited if the business objectives are focused more closely on short-term liquidity planning. A shorter-term forecast could have a daily reporting granularity and therefore be able to report on high-level flows and balances with a high degree of accuracy at a more precise date in the very near term.
Longer term liquidity risk management
While medium term liquidity issues can be identified and actioned with a 13-week forecast horizon, a 6-month forecast horizon (broken into weekly for 13 weeks and then monthly for the following 3 months) might be better at flagging up longer-term liquidity risk issues.
Take an “objectives first” approach
As referenced at the beginning of this article, it is important to identify the right time horizon in order to produce the desired forecasts. Considering what the objectives of the forecasting process are, then selecting a time horizon which produces forecasts that meet those needs, is the right approach.
The benefits of software
Using specialised cash flow forecasting software can improve the accuracy and quality, remove the administrative burden, and increase confidence in a company’s cash and liquidity forecasts. For example, with rolling 13 week flow model, the weekly forecasts produced can be analysed at the touch of a button. Quick and easy forecast vs forecast, and forecast vs actual, accuracy measurement enables biases to be identified and adjustments and corrections made as necessary.
Setting up a new process?
If you are considering setting up a new cash forecasting process, we have written a cashflow forecasting setup guide, which we welcome you to download. The guide discusses; the practicalities of designing a forecasting model, what is required in preparation for the project launch date, as well as what comes after go-live....
For a head office treasury or finance team, setting up new cash forecasting process or refreshing an existing process can bring many benefits. Being able to accurately forecast a company’s future cash position reduces risk of future liquidity issues and minimises the opportunity cost of holding uninvested cash. Also, having fast access to reliable cash flow data allows the treasury and finance team to consistently contribute to the strategic decisions being made within their organisation.
However, designing and rolling out any new reporting process in a large organisation can be challenging. Especially if adequate preparation is not done in advance. Dealing with multiple peoples in multiple places using multiple systems might seem like a daunting task, but taking a structured approach to the design and roll-out of a new process, as well as carefully managing the communication with all stakeholders, will enable a smooth roll-out and ongoing operation.
CashAnalytics has created an in-depth guide to assist people who want to set-up a cash forecasting process in their company. This guide follows a number of straightforward stages and mirrors the successful roadmap we use with our own clients. This roadmap covers five key areas:
1. Setting business objectives
2. Designing a forecasting model
3. Scoping and planning
4. Process set-up
5. Communication and roll-out
The business objectives are the reason the new cash forecasting process is being put in place. Cash forecasting supports a range of high value business activities, and the business case for putting a process in place can have a number of dimensions.
The headline activities supported by cash forecasting often include:
Working Capital Management
Ensuring the short-term cash and working capital needs of the business are adequately planned and provisioned for.
Debt and Interest Reduction
Gaining the required forecast visibility to confidently and safely use excess cash to reduce debt levels and interest costs.
Covenant and Key Date Visibility
Being able to accurately project expected cash levels on key reporting dates, thereby understanding the impact on covenant levels.
Liquidity Risk Management
Cash forecasting may not instigate a direct action but, in many cases, it is used simply as an early warning signal of future problems.
The design of the forecasting model itself is a critical task. At a basic level, the forecast model outlines the level of reporting detail required and the forecast time horizon.
Often the level of reporting detail in a cash forecast mirrors the level of detail in management reporting packs. This, at the very least, is a useful starting point.
It’s always a good idea to start with the simplest model first. One that provides the base level of required information. This can then be scaled to a more detailed or complex model over time.
Once the business objectives have been set and the forecast model designed, the next step is to scope the project and set a plan.
This planning process won’t be much different to planning for any other project, but it remains an important consideration. A planning document should identify:
The process set-up is a practical step that brings the requirements outlined in the design, scoping and planning stages to life. Whether using a manual tool such as excel or a dedicated cash flow forecasting software solution designed to automate the manual parts of the process, the three steps outlined below will need to be covered.
Step 1: Tool Configuration
Configure the chosen tool so that it mirrors the required reporting structure and maps to the correct master file data (business unit names, etc.)
Step 2: Map to Data Sources
Mapping to other data sources such as ERP systems is a key automation step than needs to be considered, but is often left until after the base reporting process is in place.
Step 3: Testing
The testing phase ensures that the implemented structures produce the required reporting output, and that the data interfaces are working correctly.
Clear communication to all stakeholders will be required throughout the project. However, communication is particularly important in and around the time the process is ready to be rolled out to the wider organisation.
It is a factor that is sometimes overlooked but clear and consistent communication with stakeholders, particularly those people contributing forecast information, is key success factor in both the initial roll-out and ongoing operation of the forecasting process.
During the roll-out part of the project, training will need to be provided to everyone involved. This typically takes the form of webinars supported by training documentation. On an ongoing basis, dialogue and communication will centre on feedback, business analysis and the continuous improvement of forecast quality.
It is the responsibility of any business leader to focus resources on the highest value activities. Lower value activities, on the other hand, are to be automated or eliminated completely. CFOs, Finance Directors and Financial Controllers in companies of all sizes have been driving technology projects over the last decade aimed at removing administrative overheads and freeing resources to invest in these higher value activities.
The vision of most senior finance leaders is to have their teams focused on analytical tasks. Tasks that feed directly into strategic planning and decision making, as opposed to spending time on recurring, manual processes. This focus on analysing and interrogating data, rather than simply collecting and organising it, is a key differentiator between high and low performing teams. Automation plays a key role in realising this vision but the other headline consideration, that can sometime be overlooked, is deciding which tasks and jobs help a company move to this higher value focus. And, do the people moving into these new roles have the skillsets needed to fulfil them?
A common challenge faced by most companies is that the time freed up by the removal of slow/manual/administrative tasks, is not always allocated to the strategic/analytical ones they had planned. Instead, one set of administrative tasks is replaced with another. There are a couple of potential reasons for this; the psychology of returning to the familiar perhaps, or maybe the demands of the business mean that as soon one burden is lifted it is replaced before any repositioning can occur.
The traditional skills required to be a top finance professional are well established. Gaining accounting qualifications, for example, requires a vast amount of work. It is a competitive field to get into, therefore the bar is set high. When embarking on their careers, accountants will then pick up further skills. Over time, as accounting the professional builds his or her career, developing these skills and rising in seniority, they become established, an expert at the top of their game. Now, senior in their post, they are increasingly being asked to review, remodel and analyse the data they once helped produce. Their job therefore requires skills that are separate from the ones that helped them to succeed through their careers so far.
To excel at the new, analytical, advisory, strategic responsibilities that automation has unlocked, the CFO or Financial Controller will realise that there are new areas of skills required for their team. In our experience, those departments that truly succeed are the ones with broad analytical skill sets to supplement their traditional financial ones. It is important to note that it is a skill set that is required, rather than one, highly refined, skill.
Initially, it is the technical skills that need to be developed. Though a good professional is likely to be highly proficient in some elements of excel from extensive spreadsheet work, they will now need to develop enhanced database and data modelling skills. After this, it is the analytical skills (the core of the upskilling requirement) that need to be developed. Analytical skills involve the ability to read, interpret and understand complex data sets. A good understanding of the business then helps to turn these understandings into actionable insights. Finally, communication skills come into play. Without these skills the value of the new insights may be lost in communication.
A range of jobs and roles exist within the data analysis sphere. In order for a CFO or Financial Controller to understand the skillsets they need on their team, it’s useful to look at the different types of data roles and what skills are needed for each role.
There are now many ways to both upskill and keep your skills sharp. Aside from traditional degree or certificate-based qualifications, offered by established universities and colleges, there are now a broad range of options to help with upskilling and keeping skills sharp. These include:
Numerous Massive Open Online Course (MOOC) websites have been launched in recent years that make courses offered by Universities such as Harvard and Stanford available in an online format. Some of the most popular MOOCs are Coursera and edX which offer numerous analytics courses including:
Short courses, structured on or offsite, can be a great to gain a deep understanding of a new topic in a short period of time. These courses typically last no longer than a few days and can often be spread over a number of weeks. Short courses in the area include:
There are a number of business analysis professional qualifications offered by various bodies. While more of a commitment than an online or short course, they may suit someone interested in building a career in the area. Professional qualifications are offered by the following bodies:
For a CFO or Financial Controller seeking to increase the analytical capabilities of their team, hiring considerations may be explored alongside upskilling options for the current team. When this is the case it is important the person building out the job spec doesn’t simply dust off an old version and tack on a couple of bullet points on data modelling and analytics. Instead they should be looking at the core of what they are seeking to achieve with the new position.
After a review, it may become apparent that the best approach is to go for someone with a hybrid, analytical/accounting skill set, or it might be that the current team has enough of the accounting skills in place already to seek out a pure analytical candidate, and/or a purely technical one. As automation increases the capabilities and strategic input of the team, the business case for greater investment in the department becomes clear. Making sure that this investment, and the subsequent increase in headcount, is deployed effectively, then solidifies the departments position as a strategic business function.
Beyond analytics, every department in every company in the world is looking at what skills they need to include to stay ahead of (or even perhaps keep pace with) an ever-evolving world. Because of this, the buzzwords of today shape the job specs of tomorrow. The need for a firm to keep abreast of the potential that Artificial Intelligence (AI) and Machine Learning (ML) can bring, means that every department is seeing the need to incorporate data scientists in their teams. It is no longer just the technology department that is seeing these changes, now that the potential has been seen they are being rolled out across a company. The finance and account departments are no different. They too will feel the need to be bolstered by these new skills.
The CFO’s vision is clear. Use automation to remove low-value, manual processes and change focus to high-value, strategic, analytic tasks. However, as mentioned earlier in the article, there are traps a department can fall into once the automation is in place. The way to avoid those pitfalls, and to truly see the benefits of automation, is in the preparation. This is the upskilling groundwork done before the newly automated processes are implemented. It is only when the skills gap has been bridged that the firm can truly progress.
This article, written by Conor Deegan – Managing Director of CashAnalytics, originally appeared in the International Accounting Bulletin....
It has been over six months since President Trump signed his Tax Cuts and Jobs Act (TCJA) into law. The act reduced a range of business and personal taxes, however they most eye-catching was the reduction of the headline corporate tax rate from 35% to 21%. Before this cut, the U.S. had the highest tax rate of any OECD country. It now ranks in the lower half of the corporate tax table.
Few areas of the business world have felt the impact of these tax cuts, from a practical perspective, more than the corporate treasury departments of large multi-national companies. While much of the media focus since the cuts has been on the cash returned to investors via share buy backs and dividends, there has been huge amount of behind-the-scenes activity in addition to these headline grabbing developments. The resultant fundamental structural changes will shape corporate balance sheets and Corporate Treasury for years to come.
Corporate treasury and tax departments were readying themselves far in advance of the TCJA coming into law in December. This is borne out by the numbers since then, which has shown a significant uptick in certain activities:
It is estimated that U.S. companies hold over $2 trillion offshore. The TCJA imposed a once off charge on cash repatriations, but JP Morgan estimates that $217 billion has been brought back to the U.S. in the first quarter of this year alone. The U.S. Federal reserve estimates that foreign earnings held abroad fell by an annualised $633 billion in the same period, an almost threefold increase on 2017.
Liquidating Offshore Bond Holdings
As early as February, Bank of America Merrill Lynch was reporting an increase in corporate bond spreads. They put this down to companies liquidating their offshore bond holdings in anticipation of repatriation. This has led to the bond holdings held by companies dropping significantly since December, with big players such as Apple shrinking their corporate bond portfolio for the first time in years.
Share Buy Backs & Cash Purchases
Last month, CNN reported that U.S. companies completed a record $201.3 billion of stock buybacks and cash takeovers in May. This means that repatriated cash is being returned to investors and put to work quickly by companies. Share buybacks alone hit $178 billion in the first quarter of 2018, topping the previous record set in 2007, just before the economic crash, according to the New York Times.
While it’s interesting to note the pick-up in certain activities since December, it’s perhaps more interesting to analyse how the TCJA will impact Corporate Treasury into the future. Below we’ve outlined some things to expect:
The headline impact of the TCJA from a Corporate Treasury perspective is that companies’ will now hold less cash. The TCJA reduces the need to hold foreign profits offshore, with the expectation that cash will now be put to a more productive use. Share buybacks, dividends, acquisitions, and capital expenditure will all reduce the aggregate cash balances held by companies.
Less Tax Risk
The management of tax risk is a big driver of day-to-day Corporate Treasury Activity. Treasury teams typically work very closely with their tax colleagues to map out funding routes designed to minimise tax expense. While this collaboration will continue, the penalty for getting this wrong will be lower due to the greater alignment between European and U.S. tax rates.
More fluid capital flows
Bringing cash into the U.S. from certain foreign locations is now less expensive and therefore the flow of cash between foreign entities within the same business will become more fluid. The lower risk and administrative overhead is expected to grease the wheels of cross border capital flows.
Lower Debt Levels
Most corporate debt is rolled over on maturity by issuing a new debt instrument or raising bank funding. Now U.S. companies, flush with cash, plan to repay debt fully, either before it matures or on maturation. A survey in 2017 by Bank of America Merrill Lynch showed that paying down debt was expected to be the number one use of repatriated cash, ahead of share repurchase and capital expenditure.
All of the above is good news for Corporate Treasury. The imbalances created by disparate tax rates in Western economies will be eroded as balance sheets adjust to the new environment. Corporate Treasury is now in a position to drive significant investment into their operations, safeguarding its role as a strategic partner. To do so, it must reinvigorate its way of operating, and question what’s been taken for granted in the past. Businesses, especially those in the U.S., are now more dependent on Corporate Treasury than ever before. This provides a massive opportunity that must not be missed....
Working with treasury and finance teams across the globe, we see a large number of cash forecasting processes with varying degrees of automation. Some are fully automated, “no touch” processes, some deploy automation at key steps in the process, while others are fully manual, labour intensive processes. In this post we look at what forecasting automation is, what the benefits of automation are, which components of the process can be automated, and how cash forecasting automation can be achieved.
There are many workflows, data inputs, and communications and reporting exercises that make up a cash forecasting process. Cash forecasting automation may refer to automating the process in its entirety or may refer to the automation of the constituent parts of this process.
As mentioned above, the extent to which the process is automated, and indeed how the process is automated, fall on a spectrum. Some parts of the cash forecasting process may be automated with Machine Learning (ML) or Artificial Intelligence (AI) technologies, or those technologies may be used at the end of the process and deployed more on the output. This post focuses on how cash forecasting processes can be automated without ML/AI. For further reading on how ML/AI can be used in cash forecasting, we have written an article discussing how ML/AI methods of forecasting compare with traditional statistical ones, which explores the topic in further detail.
Automating the cash flow forecasting process brings many benefits, namely:
The result of these benefits is that automation doesn’t just mean fewer mistakes, when applied well, it leads to a roundly improved, best practice forecasting process.
Typically, there are two sets of data collected in a cash forecasting process, the forecast data and the actual data. Automating the input of this data to the process encompasses both the collection and classification of this data. This be achieved by either algorithmic or rules-based classification.
Forecast cash flow data
Actual cash flow data
Without automation a cashflow forecasting process for head office teams can involve significant manual work, therefore automating key workflows can alleviate much of this administrative burden.
As discussed above, a cash forecasting process can be automated in its entirety, or it can be separated into its constituent parts, which can each be automated in isolation. In either case, automation is difficult if not impossible to achieve without the use of specialist cash flow forecasting software.
If you are setting up a new cash flow forecasting process, please read our Cashflow Forecasting Setup Guide.
If you have any questions, or would like any advice on how to automate any or all of the elements of your cash forecasting process, please do not hesitate to contact us....
Behavioural scientist Philip Tetlock groups forecasters into two categories; hedgehogs and foxes. He introduces these categories – borrowed from Isaiah Berlin – in his 2015 book Superforecastering: The Art and Science of Prediction. Tetlock, who currently teaches at the Wharton School of Business, has spent 20 years of studying the impact of human judgement of political decision making and uses the book to make the assertion that forecasting, at least in the short term, can be performed with a high degree of accuracy.
He categorises hedgehogs as forecasters whose judgement is based on big theories, such as the outbreak of a war or a natural disaster, and foxes as basing their judgement on a series of smaller theories which are continuously adjusted as new information becomes available.
Tetlock found that foxes, on aggregate, were far superior forecasters to hedgehogs. While their predictions were initially less satisfying to narrative demanding audiences, such as government policy makers or a mainstream press looking for sensational headlines, they ultimately proved considerably more reliable.
Forecasting in a business environment, and cash forecasting in particular, is subject to a similar dynamic. Although, for day to day business and cash forecasting, the dynamic is shaped less by the judgement of the forecaster and more by the range of skills they possess. The hedgehogs of the business forecasting world have a single, highly refined, forecasting skill, while the foxes have a range of skills.
Forecasting is a specialist skillset that is, in essence, a combination of other skills. The skill of the person or analyst creating the forecast is the primary factor influencing the quality of the forecast itself. The fact that good forecasting requires multiple skills is often overlooked. As a result, the task is often outsourced to someone with a high degree of skill in one area – for example data modelling – but without the required skills in other areas.
In CashAnalytics, we’ve worked with many companies over the years to help them establish cash forecasting processes that continuously generate a high-quality output. During this time, we’ve identified the following as the core skills needed to be a good forecaster:
1. Business understanding
An excellent knowledge and understanding of the underlying dynamics of a business are required to be a good forecaster. Without an understanding of what drives and influences the business, such as customer payment behaviour or future sales projections, a forecast won’t be anchored in reality.
2. Technical knowledge
Forecasting in large companies typically requires the gathering of information from a number of sources, such as ERP and Treasury Management Systems. To do this in an efficient way, and on an ongoing basis, requires a base level of knowledge of these systems and a technical understanding of how to extract the required data from them.
3. Data modelling skills
A forecaster will need the ability not only to collate the data collected, but also to model future events and scenarios. Alongside the application of business knowledge, this is perhaps where the most value is added to a forecasting process. However, simply focusing on data modelling alone won’t allow the forecaster to generate a high-quality forecast.
Forecasts are neither created nor used in isolation. The ability to communicate and interact with a range of people within an organisation is an essential skill required by any forecaster. If a large amount of time and effort is invested in creating a high-quality forecast, but the output and findings are poorly communicated to key decision makers, the forecasting process will ultimately fail.
As CFO or Treasurer, you may question who on your team possesses all of these skills? Or, as the forecaster you may feel you lack a particular skill or simply don’t have the time to carry out every task effectively.
However, when each of the forecasting tasks is broken down into its constituent parts it becomes clear that much of the data gathering, consolidation and base reporting tasks can be automated using specialist software, freeing up time that the forecaster to spend on the highest value activities such as data analysis and interpretation.
This hybrid approach, of combining specialist forecasting software with the refined skillsets of a forecaster, will ensure that finance and treasury teams are able to contribute to the strategic conversations within their organisation.
As Tetlock suggests, the skills needed to be a super-forecaster can be learned. However, to continue his metaphor, the fox knows that assumptions must change in light of new data. With that in mind, perhaps is it time to consider your companies approach to forecasting?...
The CashAnalytics Summary Series takes major Central Banks’ extensive reports (often well over 150 pages), and summarises them into 5 minute reads.
This release summarises the 4th edition of the ECB Economic Bulletin for 2018, which was released on the 28th June.
The key points covered in the report include:
To read the summary of the ECB Economic Bulletin, please click here....
In 1979, the Greek academic Spyros Makridakis published an article in the Journal of the Royal Statistical Society. The article showed that showed simple (traditional statistical) methods of forecasting outperformed more complex methods. At the time, the increase in readily accessible computing power had allowed newer, computational, and apparently more accurate methods of forecasting to proliferate in business and academia. Makridakis, who was by this stage a Professor at Insead, was sceptical about the true value of these more complex forecasting methods, which prompted him to embark on the extensive empirical study behind the article.
The results of Makridakis’s study were widely criticised at the time, primarily due to the belief that complexity and greater computing power must produce more accurate results. The subsequent disagreement led to the creation of the “M” series of forecasting competitions which set out to find the most accurate forecasting methods for different types of predictions. There has been a total of four “M-Competitions” since 1982 with the most recent concluding in May of this year. The winner of this competition will be announced in October.
Fast forward 40 years and a similar debate has been sparked by a paper released by the same author. In his latest study, which was released in March of this year, Makridakis and his colleagues compared the effectiveness of eight Machine Learning (“ML”) methods versus eight simpler statistical methods, used for time series forecasting.
The most recent study was prompted by the increasing popularity of ML methods in forecasting, despite the lack of sufficient evidence of their superiority over traditional, statistical methods. For this reason, the authors set out to test whether their original assertion, that simple forecasting methods are more accurate than complex computational ones, was still true.
In broad terms, the authors sought to measure forecast performance against two metrics, accuracy and computational requirements. To help define these metrics, the authors use a number of key terms that outline how these measurements were made.
The paper itself stretches to over 20 pages with a detailed breakdown and comparison of the accuracy of the different forecasting methods analysed. Some of the key findings of the study conclude:
1. Simple is Still Better
The research contained in this study showed that the simpler statistical forecasting methods “dominated” ML methods across both accuracy measures used for all forecasting horizons examined. The simplest method of statistical forecasting was the naïve method, which is essentially a rollover of historical data. The naïve method performed better than all but three of the ML methods for a one-step ahead forecast, using both measures of accuracy, with far less Computational Complexity.
2. Computational Complexity
The authors highlight the excessive computational complexity of some of the ML methods as being be a barrier to their practical application. For these methods to be used in business and other fields, “their computational requirements must be reduced considerably.” The report suggests deseasonalizing data, using simpler models and limiting the number of training iterations as ways to reduce computational complexity.
3. The Best Fitting Models Don’t Produce Best Forecasts
ML forecasting techniques typically “fit” a line or a model to historical data and use this to extrapolate into the future. One measure of the effectiveness of an ML technique is how closely it can fit this model to historical data. This study shows that methods or models that best fit a data set did not necessarily result in more accurate forecasts.
4. Need to Crack Open the Black Box
One of the key suggestions in this paper is that for ML forecasting methods to be become useful in practical business applications, the way they work and how they produce results needs to be clearer to users. The researchers stated that “obtaining numbers from a black box is not acceptable to practitioners who need to know how forecasts arise and how they can be influenced or adjusted to arrive at workable predictions.”
5. Automate Preprocessing Tasks
The preprocessing of historical data is time consuming and requires decisions by the user that add to the complexity of the overall forecasting process. The automation of these preprocessing tasks is seen as key to ML forecasting techniques becoming useful to users, on a day-to-day basis.
While the paper shines a light on the shortcomings of ML forecasting methods, Makridakis and his colleagues do highlight the “great potential of ML for forecasting applications.” The conclusion of this paper reiterates this point and mentions that “specialised algorithms”, unique to forecasting, may be required to justify ML as a viable forecasting technique. For now, however, the most effective methods of time series forecasting are the simple, statistical ones.
In CashAnalytics we believe, as is alluded to in the paper, that forecasting is like no other business discipline or task. Unlike many other business processes, the end result is measured in degrees of accuracy (and other factors), rather than a binary right or wrong.
As a technology company, and as dedicated liquidity forecasting specialists, it is our duty to be at the cutting edge to ensure our clients benefit from the best that technology has to offer. This means thoroughly testing the application of new technologies, and measuring their advantages and disadvantages. Like Makridakis, we believe that empirical analysis is the best measure of effectiveness....
Monthly updated working capital of hundreds of industry leading companiesLearn More
Measuring the accuracy of forecasts you use for decision making reporting purpos...Cash Forecasting Accuracy Measurement (429 downloads)