Identifying the information sources that are available as input to the cash flow forecast can help determine the methodology used to build the forecast. Typically, short-term cash flow forecasts are built using one (or a combination) of three different methods—a receipts and disbursements methodology, sometimes referred to as a working capital approach; a bank data approach; or a business intelligence or statistical modeling approach.
Receipts and Disbursements Methodology
In this methodology, the practitioner creates separate schedules of projected cash receipts and disbursements. These schedules are created on a cash basis rather than an accrual basis, as the object is to forecast actual cash flow and not projected earnings. The receipts schedule includes projected collections form customers, as well as other cash inflows such as interest and dividend payments. The schedule should also include non-recurring cash flows such as financing and sales of assets. The disbursement schedule projects cash disbursements for purchases, payroll, taxes, interest, dividends, rent and any projected debt repayments.
This data typically is taken from accounts payable data but may need to be adjusted to reflect the fact that payments often do not clear on the day they are issued. The two schedules are then combined with the beginning cash balance (ending cash balance from the prior period) to create a projected ending cash balance. This methodology works best for short term forecasts and is only as reliable as the underlying data sources.
Bank Data Approach
The bank data approach is similar to the receipts and disbursement methodology but is based on the company’s bank statements. This methodology works backward from the current cash position on an entity-by-entity or account-by-account basis. It takes a snapshot of the current bank statement and combines it with the best estimates of anticipated changes on a category-by-category (e.g., lockbox, credit cards, wires, checks, ACH, etc.) basis. It then combines them to create a short-term forecast. This approach is best suited for a very short-term forecast such as the daily cash position, and is the basic format used by many bank-offered systems.
Statistical and Business Intelligence Methodologies
A statistical methodology uses historical data combined with projected sales and other known factors to forecast future cash flows. A variety of statistical tools such as moving averages, regression analyses and Monte Carlo simulations are used to identify the relationship between historical trends and projected cash flows. The historical data used may be as simple as prior period disbursements and receipts adjusted for seasonal variances, or can be very complex and factor in such things as projected sales, commodity pricing and other market trends.
Business intelligence methodology expands the use of historical data to include information on known business drivers. The drivers of business results need to be incorporated into the cash forecast methodology and design. Here is where the treasurer needs access to corporate planning and strategy sessions where these items are discussed. Many other aspects of the company materially impact cash flows and need to be incorporated into the methodology and design of the cash forecasting process. This includes periodic cash peaks and troughs, seasonality and sporadic big-ticket items such as expansion or the divestment of business lines.
Making the Choice
The choice of methodology used is typically determined by the quality of the available forecast data and the time horizon for the forecast. When reliable data is available from internal systems, there is little need to manipulate it using statistical methods and a simple receipts and disbursement method may be all that is needed. When the data is not as reliable or is highly variable, more sophisticated statistical tools may be needed to create a satisfactory forecast.
In selecting a methodology and building a model, it is important to keep the KISS (keep it simple, stupid) principle in mind. The goal should be to keep the model as simple as possible, while still achieving the necessary forecast quality. A model that achieves 90% reliability but is easy to maintain and use may be more valuable than one that achieves 95% reliability but needs constant modification and attention.
Mark Webster is partner with Treasury Alliance Group.
For further insights on how treasury and finance professionals are managing the upheaval caused by the coronavirus pandemic, visit AFP's Coronavirus Resource Center.