Retention Is a Capital Decision, Not an HR Program
Board note — for decision
To: Board of Directors From: Chief Financial Officer Date: 14 September 2026 Subject: Funding retention, or budgeting replacement Evidence base: Ministry of Manpower (MOM) — Labour Market Report, First Quarter 2026
Purpose of this note
The board is asked to decide whether retention is funded as an investment, or replacement is budgeted as a recurring cost. The note does not ask for a larger sum of money. It asks the board to choose, deliberately, which of the two it funds.
1. The recorded position
MOM's Labour Market Report for the first quarter of 2026 records an average monthly resignation rate of 1.0%, 73,300 job vacancies, and an unemployment rate of 2.0% (MOM, 2026).
Read alone, the resignation rate is reassuring, and previous meetings have read it that way. The rate measures exits that have already occurred. It provides no information about staff who have not left, and it does not tell the board what a replacement would cost if one did.
The vacancy figure requires more attention. 73,300 vacancies against unemployment of 2.0% (MOM, 2026) describes a market where the roles we need to refill are contested, and where the candidates we lose are drawn from the same pool our competitors are drawing from. In that market, our budget does not set the price of replacement. The market sets it at the moment we need it.
2. Why the current comparison does not hold
The board compares retention spend with replacement cost every year. That comparison is presently invalid, for a structural reason rather than an analytical one.
Retention spend is budgeted, annual, visible, and sits inside a line the board approves. Replacement cost is not budgeted. It arrives after the fact: as a resignation, then a requisition, then a period in which the role is unfilled, then a ramping period for whoever fills it. It spans departmental lines and more than one financial period. It is rarely presented as a single figure, because it was never created as one.
As a result, the board chooses between a cost it can see and a cost it cannot. That is not a decision; it is an accounting artefact. It is also why retention spend is held to a standard of proof that no replacement cost has ever been asked to meet.
3. The exposure created by a low exit rate
A low resignation rate is a lagging indicator, and it is most misleading when it is lowest. Two effects follow.
First, a low exit rate removes the visible symptom that would otherwise prompt a review. The attrition line stays flat while the conditions that produce attrition — pay position against the market, the absence of a credible internal route for staff who want to move — remain unaddressed, because nothing in the monthly accounts shows them.
Second, the same market conditions that keep exits low make replacement expensive when they do happen. With 73,300 vacancies and 2.0% unemployment (MOM, 2026), the search does not shorten because our own exits are low; the role is filled from the same constrained pool, on the same terms.
This note does not argue that the 1.0% monthly resignation rate will rise. It states that the board holds no measure of its exposure, and that the absence of a measure has been read as the absence of a risk.
4. The options in front of the board
Option A — Fund prevention. A retention line targeted at the roles where replacement is demonstrably most expensive, approved against a costed model and measured against the same model.
Option B — Budget replacement. Treat resignations as a recurring cost, carry them as a provision, and approve them at the level the model supports.
Option C — Neither. Retention is not funded as an investment and replacement is not budgeted as a cost. This is the default position, and it is the only option that cannot be defended on the evidence above.
5. What is not in this note
No methodology for the replacement cost model; that model is a separate paper and will be tabled before any allocation is treated as settled. No assessment of named individuals or teams. No assumption about the length of a vacancy period. No external benchmark has been used: the position above rests only on the official labour market statistics for the first quarter of 2026 (MOM, 2026).
The decision
The board is asked to approve the following, and only the following.
Reclassify retention. Administer retention as a capital decision, competing for the same cash as any other investment on the board's list and assessed on the same terms.
Withhold new funding at this meeting. No incremental retention budget is approved today. A costed replacement model is tabled at the next meeting, built on assumptions this board can interrogate. The allocation is decided then, against that model, rather than against the current reading of a stable market.
Bound the redirection. Where the model supports prevention, the retention allocation is funded by redirecting the sum currently carried as provision for replacement, and it must not exceed that sum. Where the model does not support prevention, the provision is restored in full, and the retention line does not stand. In either case the board receives a figure, and the trade is made explicitly rather than by default.
The decision is not whether the company can afford retention. It is whether the board will keep funding the option it can see over the option it has not yet measured.
Frequently asked questions
Is retention really a CFO decision rather than an HR one?
Yes. Every resignation is a dated cash event, and the choice in front of the board is which of two costs the company funds: prevention, which is budgeted and visible, or replacement, which arrives after the fact. That is a capital allocation question, and it is the CFO's to frame.
Why is a low resignation rate not reassuring?
Because it is a lagging indicator: it measures exits that have already happened, and says nothing about the exposure that remains. A low rate removes the visible symptom that would otherwise prompt a review, while the conditions that produce attrition stay unaddressed.
Why does the usual retention-versus-replacement comparison not hold?
Because the two sides are not measured the same way. Retention spend is budgeted, annual and visible; replacement cost arrives as a resignation, then a requisition, then a vacancy, then a ramp-up, spanning departmental lines and more than one financial period. The board ends up comparing a cost it can see with one it cannot, and holding retention to a standard of proof no replacement cost has ever been asked to meet.
What does the labour market data actually show?
The Ministry of Manpower's Labour Market Report for the first quarter of 2026 records an average monthly resignation rate of 1.0%, 73,300 job vacancies, and unemployment of 2.0% (MOM, 2026). The vacancy figure is the one to read: the roles we need to refill are contested, so the budget does not set the price of replacement, the market does.
What is the board actually being asked to approve?
Three things and no more: withhold new funding at this meeting; table a costed replacement model at the next, built on assumptions the board can interrogate; and where that model supports prevention, fund it by redirecting the sum currently carried as provision for replacement, never above it.
Comments