Overcoming the Challenge of Cash Flow Forecasting
Although global treasurers are increasingly focusing on cash flow forecasting, the majority still use spreadsheet models for their forecasts and therefore fail to achieve much-needed accuracy. The quality of cash forecasts is sometimes so low that organisations decide to stop performing them and therefore lose their immense value. The key to improved forecasting ability is systems support and integration of source systems.1
Cash forecasts are all about information so it is important to understand what that actually means. Information can be divided into four levels that must be understood when setting up system support for cash forecasting:
System support must enable an organisation to collect data from source systems and provide key personnel with knowledge based on accurate and trustworthy information so they can take the required actions.
When implementing a system, it is easy to simply pave over the current infrastructure instead of building new efficient systems. It is necessary to question the current processes connected to forecasting when implementing system support, in order to benefit from the full impact of the change that the system will enable. A system implementation project should begin with the evaluation of the current forecasting models if they exist, and go on to evaluate what are the best, and less successful, aspects of the current model and to what extent the forecast model fulfils its targets. Such targets may concern reducing the use of credits or a better use of forwards, and should be formulated so that they can be measured and followed up.
As mentioned earlier, most treasurers tend to use spreadsheets for cash forecasts, which causes problems. The first, and perhaps most significant, is uncertainty in the accuracy of the information, which undermines the value of the forecast or yields faulty decisions that cost money. When information is manually read into source systems, e.g. accounts receivable, and then manually consolidated into spreadsheets, many errors occur. After this stage, the information is sent to the treasury which aggregates several subsidiaries into one forecast, giving rise to yet more errors.
The second problem is that manual processes cost money in the form of highly skilled personnel who are highly paid performing simple tasks.
The third issue is related to incorporating subsidiaries in the forecasting process. Once subsidiaries send in a spreadsheet to the treasury department, they often won’t receive relevant feedback or have any possibility of following up on the forecast.
A fourth problem is lack of flexibility in forecasts performed on a spreadsheet. If data is fed in on a daily basis, creating like-for-like figures on a weekly basis requires a great deal of work.
Finally, the quality check of previously performed forecasts is difficult to obtain, which is an obstacle because quality checks of forecasts are vital to the forecasts’ value.
The key to solving the first two problems described above is the integration of source systems. IT departments often hesitate when the word ‘integration’ crops up because this process can be costly but this does not have to be the case. Such integration should be evaluated when looking for a software supplier of cash forecasts, as suppliers vary in their ability to handle integration.
The question of what should be integrated should be evaluated through a cost/benefit analysis. In general, you should assume that as much as possible should be integrated. All accounts receivable and accounts payable with significant cash flows need to be integrated to achieve high quality forecasts. The 80/20 rule often applies and can provide a good basis. If a high proportion of incoming cash flows is retrieved through cash or cards, a budget system may need to be integrated or built into your forecasting system.
Bank information should be integrated as extensively as possible, in order to obtain an updated starting balance for your forecast. Bank information can also be used in follow-up analyses, ensuring that your forecasts are accurate. Follow-up analyses should be performed continually, to detect areas where the forecast quality is not as high as it might be. Action can then be taken to find the errors and correct them, based on which a new quality analysis will be performed during the following period. Quality will then be raised continually and the fifth problem outlined earlier addressed.
Certain information may be hard or impossible to collect from another source, e.g. taxes or salaries. There can be other ways of supporting the automatic update of such information though, such as the system supporting the automatic creation of recurring payments that could be used for salaries.
When integrating source systems, you should be aware that information in the source may not be up to date, or may even be faulty, and this can be handled in various ways. You should begin by addressing the question of why the information is faulty and improving the process that makes it so, for example, errors in source information can occur if an invoice paid earlier is not ticked off in the ledger. In some cases this might not be enough though. You need a way of correcting information in your forecast when information is inaccurate due to incorrect data in the source system. For instance, this can be performed through a web interface.
As mentioned earlier, it can be difficult to include subsidiaries in the forecasting process, which lowers motivation and therefore the quality of reports, if you use a spreadsheet model. This means the forecast system needs to support the distribution of information to subsidiaries, an issue that can be resolved using a web interface. A web interface can also be used for reporting data that cannot be integrated and to report corrections that are not up to date in the ledger.
Furthermore, a web interface can be used to provide subsidiaries with the chance to follow their own forecasts and perform quality checks on them by comparing them with bank data.
A system-supported and integrated cash forecast will give the treasurer more time to focus on core duties. Time used for forecasting will shift from data collection and consolidation to the more interesting task of analysis. Of course, this will increase the value and interest of the forecast while engendering higher accuracy and providing a simpler and more helpful analysis.
When setting up a forecast model with system support, the treasurer should start by asking which benefits they want. When this has been documented, subsidiaries should be incorporated in the process, and may even be able to contribute to the goal setting. Perhaps not all subsidiaries should be involved in the process, but it would be of great value if some could give their input to how the forecast process should work.
If integration and automation is the key to a successful forecast, some issues can be an impediment to the forecasting process if they are not addressed at an early stage. Information should be standardised, i.e. the organisation needs to find a common language for the cash flow terms in the cash forecast. A forecast model should not contain too many cash flow terms. Examples of this are salaries, bonuses and social taxes. Would a forecast containing all of these items be of any value or could all be incorporated in the cash flow term ‘salaries’? The value of each term incorporated must be questioned.
When a forecast is in use, the quality of the forecast should be evaluated regularly and it is important to achieve as much value as possible with respect to the forecast. If forecasts are performed without any quality evaluation, the source of any errors will never be found. Furthermore, errors will arise through bad decisions taken on the basis of incorrect information, e.g. by a deposit placed with no balance to cover it. Consequently, support for the quality follow-up of a forecast is a mandatory function of any forecasting system in order to provide substantial value.
While more attention is being paid to forecasts and more suppliers of forecast systems are entering the market, the solutions available vary greatly. Treasurers must be clear about what they want to achieve when choosing a supplier. When the goal of a forecast has been defined and a requirement specification documented, a follow-up can form a good start for a requirement specification.
Of course, there are many other issues to cover but the ones detailed above are the most essential. The treasurer can now investigate whether a standardised system fulfils existing needs or whether a customised system is preferable.
For the majority of corporations with the need for more efficient cash flow forecasting, system support can be of great value. When the decision to implement systematic forecasts is made, it is important to consider whether the chosen system fulfils the organisation’s needs. With the right system support, forecasting can become a relatively painless task for all concerned.
1 Most of the issues covered in this article are related to short-term forecasts but they may also apply to long-term forecasts and budgets.