Cash-Flow Forecasting and Working-Capital Controls for Malaysian SMEs

A cash flow forecast is only useful if it can be tested. Most SME forecasts fail not because the model is wrong but because the assumptions were never checked against what the business actually experiences — and nobody notices until the week the money is short.

Testability is a property of how the forecast is built. A forecast assembled from the ledger, on contractual payment terms, at monthly resolution, cannot be compared against anything meaningful. One built from the bank account, on observed collection behaviour, at weekly resolution, produces a variance every week that either confirms the assumption or corrects it.

Forecast cash, not profit

Profitable businesses run out of cash. The two diverge because of timing: revenue recognised before it is collected, inventory purchased before it is sold, tax and statutory payments falling due on their own cycle, and capital expenditure that never touches the income statement in the period it consumes cash.

A short-cycle forecast — typically thirteen weeks, updated weekly — works better than an annual budget for managing liquidity, because it operates at the resolution at which cash actually moves. Thirteen weeks is long enough to show a quarter-end statutory cluster and a facility maturity, and short enough that every line can be justified individually.

Build it on the bank account, not the ledger

Start from the closing balance on the bank statement, not the cash figure in the ledger. The difference between the two — unpresented cheques, deposits not yet cleared, transfers in transit — is exactly the timing that a liquidity forecast exists to capture.

Practical points that determine whether the model behaves:

• Direct method only. List expected receipts and expected payments. Do not start from forecast profit and adjust; the adjustments become the forecast, and they cannot be tested.
• One line per account. Where the business runs several accounts and an overdraft, model each, then aggregate. Cash sitting in an account that cannot fund a payment on Friday is not available cash.
• Show the facility separately. Distinguish the limit, the amount drawn and the headroom. A forecast showing a positive balance that is entirely borrowed is telling management the opposite of what it appears to say.
• Weeks run to a fixed day. Pick the same weekday each week and keep it, so that a payment run always falls in the same bucket and variances are comparable.

Derive collections from experience, not from terms

The single most common forecasting error is assuming customers pay to terms.

The correction is arithmetic rather than judgement. Take the last twelve months of invoices and compute, for each month of invoicing, the proportion collected in that month, the following month, the month after, and so on. That collection profile is the assumption. Apply it to the current debtor ledger and to forecast sales, and the resulting receipts line reflects what the business experiences rather than what its contracts say.

Refine it in three ways:

• Model large customers individually. An average built across a concentrated ledger conceals the exposure that matters. Identify the customers representing the bulk of the balance and forecast each on its own observed behaviour.
• Model deterioration, not the contract. Where a customer has been slowing, use the slower pattern. A customer who has moved from forty-five days to seventy-five will not return to forty-five because the forecast says so.
• Exclude disputed balances. A balance under query is not a receipt with a date; it is a receipt with a condition. Leave it out until resolved and record it separately, so its absence is visible.

Where invoices are raised late, the collection profile is measuring the invoicing delay as well as the customer. That is worth separating, because it is the one part of the cycle the business controls unilaterally.

Include what is committed, not only what is planned

Outflows are frequently understated because committed obligations sit outside the operating cycle and outside the person's view who prepares the forecast.

• Payroll-related statutory payments. EPF, SOCSO and the Employment Insurance System contributions, and monthly tax deduction, each fall due on the fifteenth day of the month following the month of deduction. They cluster on one date, and that date is not payday. Confirm current requirements with EPF, PERKESO and HASiL.
• Corporate tax instalments. Instalments under an estimate of tax payable fall due monthly on their own schedule, and any balance of tax arises separately after the return is filed. Both belong in the forecast on their due dates.
• Indirect tax. Where the business is registered, returns and payments follow a taxable-period cycle set by the Royal Malaysian Customs Department, independent of the business's own month end.
• Loan repayments — principal and interest separately — together with any balloon instalment, refinancing date or covenant test date.
• Trade facility maturities. Letters of credit, trust receipts and bankers' acceptances mature on fixed dates. A trust receipt falling due is a cash payment whether or not the underlying stock has sold.
• Lease payments under committed agreements, and capital commitments already contracted.
• Seasonal peaks such as bonuses, festive-period costs and the associated statutory contributions, which are larger in the month they fall than any average suggests.

Working capital is where cash is usually recoverable

Before seeking external funding, examine whether cash is already trapped in the cycle. Measure it rather than estimating it: days sales outstanding, days inventory outstanding and days payables outstanding, combined into a cash conversion cycle. A movement of a few days on any of the three, in a business of any size, releases or absorbs cash that is material relative to the facility being contemplated.

Three questions usually locate it:

1. Are debtor days lengthening, and on which customers? Ageing by customer, compared across quarters, identifies whether the problem is general or concentrated. The response differs entirely.
2. Is inventory holding stock that will not sell at cost? Where inventory is material, this is an accounting question as well as a cash one — stock carried at cost that will not realise cost overstates both the asset and the liquidity management believes it has.
3. Do supplier terms reflect actual purchasing volume? Terms agreed when the business was smaller frequently remain unrevised. Stretching payments without agreement is a different matter, and damages supply before it improves cash.

Each of these is typically faster and cheaper to address than raising finance, and none of them requires anyone's approval.

Model the downside and name the failure week

A forecast that only works on the expected case tells management nothing about risk.

Build a downside case by varying the assumptions carrying the most weight, one at a time: the largest customer paying thirty days later or not at all, a general lengthening of the collection cycle, a facility not renewed at its review date, a delayed capital receipt, a supplier withdrawing credit terms. Then identify the specific week in which cash, including undrawn facility headroom, runs out.

That week is the output that matters. It converts an abstract concern into a deadline, and it determines how much time is genuinely available to act. Where the downside shows a shortfall, act while options remain: the range of available responses narrows sharply as cash tightens, and the responses that remain latest are the most expensive.

Review variances weekly, and change the assumption

The discipline that makes the forecast reliable is the weekly comparison of forecast against actual, line by line, with each variance classified as either timing or permanent.

A receipt that arrived a week late is a timing variance; the assumption about that customer may need adjusting, but the amount is intact. A receipt that did not arrive because the customer disputes the invoice is permanent until resolved, and rolling it forward week after week produces a forecast that has quietly become fiction. The rule is to adjust the assumption that proved wrong, not the closing balance that resulted from it.

A working routine

1. Build a rolling thirteen-week forecast at weekly resolution, starting from the bank balance.
2. Derive collections from the measured collection profile, with large customers modelled individually.
3. Include every committed outflow, particularly the statutory cluster and facility maturities.
4. Measure the cash conversion cycle and address what is recoverable before seeking finance.
5. Maintain a downside case with a named failure week.
6. Compare forecast to actual weekly, classify each variance, and adjust the assumption.
7. Escalate early where the downside indicates a shortfall.

None of this requires software or outside assistance. A business with one or two bank accounts, a manageable customer list and someone numerate to maintain it can run a thirteen-week forecast in a spreadsheet, and doing so consistently is worth more than a sophisticated model reviewed occasionally. External support is worth considering where the position is genuinely tight, where a lender or a board requires an independently prepared forecast, or where the inventory and receivable judgements underlying it are material and contested — not as a matter of routine.

General-information limitation

This article is general information, not financial, accounting or insolvency advice for a particular organisation, and it does not determine any company's liquidity position or the appropriate response to it. Statutory payment dates, tax instalment obligations and indirect-tax cycles are set by the relevant authority and are periodically revised; confirm the current requirements against each authority's own published material. Obtain fact-specific advice where the position is material or time-sensitive.

To discuss your circumstances, contact Saifudin & Co.