A minimum liquidity policy that is not derived is a number somebody liked. Derived, it is three components — and the commonest way of testing cash against it counts one of them twice.
Add three things and you have a policy you can defend: the float the business needs to keep paying, a reserve sized on the measured error of your own cash forecast, and the seasonal peak your annual plan already knows about. On the company in this book that is 124,596,281.16 — and the .16 on the end is the first sign it was derived rather than chosen.
Worked in full in Treasury Management by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
| Component | Where it comes from | Amount |
|---|---|---|
| Operating float | 6 business days of disbursements at 5,232,000.00 a day | 31,392,000.00 |
| Forecast-error reserve | z = 2.326 on the standard deviation of the thirteen-week error | 31,204,281.16 |
| Seasonal peak | the third-quarter inventory build, net, from the annual plan | 62,000,000.00 |
| Minimum liquidity | of which 93,204,281.16 sits above the float | 124,596,281.16 |
The reserve is the only component that is not a policy choice, and it is the one most often invented. It comes from the cumulative thirteen-week error of the company's own rolling forecast, over the last twelve quarters, as a share of the quarter's disbursements.
| Statistic | Value |
|---|---|
| Quarterly disbursements | 327,000,000.00 |
| Mean error — the bias | 0.5750 per cent |
| Sample standard deviation | 4.1026 per cent |
| … in money | 13,415,426.12 |
| Reserve at one-sided 95 per cent (z = 1.645) | 22,068,375.97 |
| Reserve at one-sided 99 per cent (z = 2.326) | 31,204,281.16 |
| Reserve at one-sided 99.9 per cent (z = 3.090) | 41,453,666.72 |
The mean error is a bias of 1,880,250.00 a quarter, not noise. A bias is something to find and remove rather than to hold a reserve against — but until the line responsible has been found, the reserve should carry it: 33,084,531.16 on these figures.
Here is the error, and it is common enough to reproduce rather than to warn about. The float is held back from cash, giving available cash of 120,608,000.00. The float is also a component of the minimum, giving 124,596,281.16. Test the first against the second and the float has been counted twice: the company appears to be -3,988,281.16 short of its own policy, and the report says so.
The float is either in the cash or in the requirement, never both. Available cash already excludes it, so it is tested against the minimum above the float, 93,204,281.16. Accessible cash still contains it, so it is tested against the total, 124,596,281.16. The two bases give the same answer by construction.
| Basis: available cash | Basis: accessible cash | |
|---|---|---|
| Cash | 120,608,000.00 | 152,000,000.00 |
| Undrawn committed facility | 250,000,000.00 | 250,000,000.00 |
| Headroom | 370,608,000.00 | 402,000,000.00 |
| Minimum tested against | 93,204,281.16 | 124,596,281.16 |
| Over the minimum | 277,403,718.84 | 277,403,718.84 |
The minimum is not the interesting number; the margin over it without the conditional part of the liquidity is. Take the undrawn facility out and the margin is 27,403,718.84 — 5.2377 business days of the company's own payments. That is the line worth putting on the monthly report, every month, in the same place, so that the quarter in which it matters is not the first time anybody reads it.
Write the three components down with their sources, sensitise each of them, and say on the line which are policy choices: the 6 days, the confidence level, the treatment of the bias. Then test cash against the minimum on one consistent basis, and check which one you are on by asking whether the float appears on both sides.
Derive it from three components rather than choosing a round number: operating float (31,392,000.00, being 6 business days of disbursements), a reserve sized on the measured error of the cash forecast (31,204,281.16 at one-sided 99 per cent), and the seasonal peak from the annual plan (62,000,000.00). Here that totals 124,596,281.16.
Measure the cumulative error of your own thirteen-week forecast over the last twelve quarters, take its standard deviation (4.1026 per cent of quarterly disbursements here, or 13,415,426.12), and multiply by the z of the confidence level you choose: 22,068,375.97 at 95 per cent, 31,204,281.16 at 99.
Counting the operating float twice — deducting it from cash to get available cash and also including it in the minimum that available cash is tested against. On this company it turns a margin of 27,403,718.84 into an apparent shortfall of -3,988,281.16.
This article is one calculation from Treasury Management. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: corporate finance, valuation and markets → · All 1706 articles →
If this book helped — or didn’t — a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.