Key Takeaways
- Historical attrition is evidence, not the answer. Setting the assumption equal to last year’s exit rate is the single most common error we see.
- It is a long-term average. The assumption should reflect expected attrition over the duration of the liability, not the next twelve months.
- Actuarial attrition is not business attrition. The actuarial assumption applies only to the existing workforce and excludes future hires, so it is usually lower than the HR planning number.
- The impact of attrition assumption depends on the gap between salary escalation and the discount rate. Higher attrition brings payments forward. Where salary escalation exceeds the discount rate, that reduces the liability; where the discount rate is higher, it increases it. This holds whatever the scheme.
- Vesting conditions often reduce the liability as attrition assumption increases. Where a benefit is forfeited or scaled down depending on how and when an employee leaves, higher attrition pushes the liability down. How much depends on the scheme’s rules and how much of the workforce sits below the threshold.
- Stability is a feature. Change the assumption when management’s view of the future changes — not because one year’s experience was noisy.
Table of Contents
- What the Standards Actually Require
- Historical Attrition Is a Guide, Not the Assumption
- Business Attrition vs Actuarial Attrition
- How Attrition Moves the Liability
- Using Staggered Attrition Rates
- When to Change the Assumption
- Common Mistakes and What Auditors Ask
- FAQs
The employee attrition rate is one of the assumptions that a reporting entity — not the actuary — is responsible for setting. Get it wrong and the liability recorded in the balance sheet is wrong, often materially.
The principles in this post apply to any defined benefit scheme valued under AS 15, Ind AS 19 or IAS 19 — gratuity and compensated absences in India, and End of Service Benefit (EOSB) schemes across the Middle East. Where the mechanics differ between schemes, they are set out separately in section 4.
Companion posts cover the discount rate and the salary escalation rate, and our governance framework for setting actuarial assumptions sets out who should own each decision.
1. What the Standards Actually Require
Ind AS 19 and IAS 19 classify employee turnover as a demographic assumption — para 76(a)(ii) names “rates of employee turnover” directly — and require it to be the reporting entity’s best estimate of future experience. AS 15 (Revised 2005) carries the equivalent requirement at paras 73–74. Two conditions matter in practice:
- The assumption must be unbiased (Ind AS 19 / IAS 19 paras 75 and 77) — “neither imprudent nor excessively conservative”. There is no prudence margin in a defined benefit valuation.
- The assumption must be mutually compatible with the other assumptions (para 78), particularly salary escalation and the discount rate. An attrition assumption implying a workforce that turns over every four years sits badly with a salary escalation assumption built on long-service promotion patterns.
The ownership point is worth restating because it is still misunderstood. The predecessor standard, AS 15 (1995), was silent on who set the assumptions, and in practice the actuary’s recommendation was usually adopted without challenge. AS 15 (Revised 2005) para 74 made the assumptions explicitly the enterprise’s best estimates, and that position carried through unchanged into Ind AS 19. The actuary advises; management decides and signs off.
Who signs off? The attrition assumption should be approved by someone who genuinely has a view on future workforce behaviour — typically the CHRO or head of HR, with the CFO confirming consistency with financial plans. An assumption approved solely by finance, with no HR input, is difficult to defend in audit.
2. Historical Attrition Is a Guide, Not the Assumption
The standards require an assumption about future exits. Past experience informs that view; it does not determine it.
A company may have experienced attrition of 12.58% over the reporting period. If management expects future attrition of 5% — because a retention plan has been rolled out, or because the prior year included a one-off restructuring — then 5% is the correct assumption, not 12.58%. What is required is the reasoning and evidence behind the 5%, documented at the time.
The assumption is a long-term average, not a spot rate
This is the point most often missed. The attrition assumption is not a forecast for the next twelve months. It is a time-weighted average over the duration of the liability, which for a terminal benefit scheme is commonly a decade or more, though it varies widely with workforce profile.
Historical attrition is volatile by nature. Using a single year’s experience as the sole basis imports that volatility straight into the balance sheet, producing actuarial gains and losses that reverse the following year and tell readers of the accounts nothing useful.
A practical test
Ask: “Over the next ten years, what proportion of our current employees do we expect to leave in an average year?” If the answer differs materially from the rate you are about to adopt, the assumption is being set on the wrong basis.
3. Business Attrition vs Actuarial Attrition
Most companies already make attrition assumptions for recruitment planning, budgeting and workforce modelling. The actuarial assumption should be consistent with these — an actuarial valuation assuming 6% attrition alongside a hiring plan built on 18% is not defensible.
But consistent does not mean identical. The two numbers measure different populations:
| Business attrition assumption | Actuarial attrition assumption | |
|---|---|---|
| Population | Entire future workforce, including employees not yet hired | Only employees on the books at the valuation date |
| Purpose | Recruitment, budgeting, capacity planning | Measuring the obligation arising from past service |
| Typical level | Higher | Lower |
Attrition falls as employees accumulate service. The existing population is, by definition, the group that has already survived its high-turnover early years, so the actuarial assumption is generally lower than the corresponding business number. New joiners — who will drive much of the future headline attrition — are irrelevant to the valuation because they have no past service to value. The relationship is not universal, though: a workforce with a heavy long-service tail alongside an aggressive hiring plan can invert it, so reconcile rather than assume.
Capture every source of exit
By convention, attrition in an actuarial valuation covers all exits other than death, disability and normal retirement, which are modelled separately. That includes:
- Voluntary resignation
- Termination and dismissal
- Planned redundancies and restructuring
- Non-renewal of fixed-term contracts
- Contract localisation or nationalisation programmes
Several of these will not appear in past attrition data at all. A redundancy programme approved by the board but not yet executed is a known future exit that the assumption must reflect.
The mode of exit matters as much as the rate
Attrition is not a single homogeneous decrement. What the employee is actually paid depends on how they leave, and most schemes draw those distinctions somewhere — a resignation, a dismissal for cause, a redundancy and a contract non-renewal can each attract a different entitlement under the same plan.
Two consequences for assumption-setting:
- The data needs to record exit reason, not just exit counts. Most HR systems capture this. Many actuarial data extracts drop it, and the actuary is then forced to adopt a default split and disclose it — a worse answer than the data the company already holds.
- A single blended rate implicitly assumes a fixed mix of exit reasons. If the mix shifts — a restructuring replaces voluntary churn with redundancy, say — the liability moves even though the headline attrition rate has not.
Where the scheme rules make no distinction by exit mode, this collapses to a single rate and none of it matters. Where they do, it matters a great deal. Section 4 sets out which is which.
4. How Attrition Moves the Liability
Finance teams frequently assume that higher attrition means a lower liability. That intuition is unreliable. Two distinct forces are at work, and they can pull in opposite directions.
Force one: timing
Higher attrition brings the expected payment forward. Whether that raises or lowers the present value depends entirely on the gap between the salary escalation rate and the discount rate:
| Relationship | Effect of an earlier expected exit | Effect of higher attrition |
|---|---|---|
| Salary escalation > discount rate | Less salary growth is projected, and the discounting saved does not offset it — present value falls | Reduces the liability |
| Discount rate > salary escalation | The payment is discounted over a shorter period, and that dominates — present value rises | Increases the liability |
This applies to any scheme where the benefit is linked to final salary and paid on exit. It is also why the attrition assumption cannot sensibly be set in isolation from the other two — the same attrition rate moves the liability in opposite directions depending on where the discount rate sits relative to salary escalation.
One qualification: the logic assumes the accrued benefit is not capped. Where a ceiling applies, it truncates the salary projection for employees near the cap and can flatten or reverse the effect for that cohort.
Force two: forfeiture and reduced entitlement
Where the scheme pays less — or nothing — to an employee who leaves early or leaves in a particular way, higher attrition means more of that reduction is expected to happen, which pushes the liability down.
The size of the effect depends on two things: how severe the reduction is, and how much of the workforce sits below the relevant threshold. A vesting condition is irrelevant to a workforce that has already comfortably passed it, and dominant for one that has not.
Where a scheme has no forfeiture of any kind, force two disappears and the timing rule alone determines the direction.
How this plays out by scheme
The two forces above are general. What differs between schemes is the forfeiture rules that drive force two.
| Scheme | Forfeiture / reduced entitlement rules | Net effect of higher attrition |
|---|---|---|
| Gratuity (India) | Five years’ continuous service required for resignation and retirement, waived on death or disablement. Gratuity may be forfeited in whole or part on dismissal for specified misconduct. Reduced threshold for fixed-term employees — see callout below. | Usually reduces the liability for a workforce weighted below the vesting threshold. For a largely vested workforce, forfeiture falls away and the timing rule governs — which can mean an increase. |
| Leave / compensated absences (India) | Policy-driven rather than statutory. Many policies allow unrestricted encashment on exit, in which case there is no forfeiture at all; others restrict encashment on resignation or cap accumulation. | Small in either direction. These liabilities are short-duration, so the timing effect has little room to operate, and forfeiture often does not apply. The availment assumption frequently matters more. |
| EOSB (GCC) | Depends on jurisdiction. Saudi Arabia scales the award down on resignation (see below); the UAE removed its equivalent reduction in 2022. Most jurisdictions also withhold the award where the contract is terminated for cause. | Depends on the exit mix. Where a resignation scale applies, resignation-driven attrition reduces the liability materially while termination-driven attrition largely does not. |
Our post on sensitivity of gratuity valuation to assumptions and employee profile works the gratuity case through in detail, with charts by age, salary and past service.
Saudi EOSB: the resignation scale
Saudi Arabia is the clearest example of entitlement varying by mode of exit. Under Article 84 of the Labour Law the full award is half a month’s wage for each of the first five years and one month’s wage for each subsequent year, on the last wage, pro-rated for part years. Article 85 then scales it down where the employee resigns:
| Service at resignation | Proportion of Article 84 award |
|---|---|
| Less than 2 years | Nil |
| 2 years to 5 years | One-third |
| More than 5 years, less than 10 | Two-thirds |
| 10 years or more | Full award |
Article 85 preserves the full award notwithstanding resignation in defined cases — force majeure, and a female worker leaving within six months of marriage or three months of childbirth. Article 80 removes the award entirely where the contract is terminated for the listed causes.
The practical consequence: two companies with identical 15% attrition can carry materially different obligations if one loses people mainly through resignation and the other mainly through employer-initiated termination.
A common EOSB data error
Article 84 is based on “wage” as defined in Article 2 — basic pay plus periodic and regular allowances — not basic salary alone. Supplying basic-only data understates the obligation, and the error compounds with every other assumption.
Labour codes change the gratuity calculus for fixed-term employees
India’s four labour codes came into effect on 21 November 2025, and the final Central Rules were notified on 8 May 2026. The Code on Social Security itself provides only that gratuity is payable to a fixed-term employee on a pro rata basis, disapplying the five-year requirement, without specifying a minimum period. The one-year threshold sits in the Code on Social Security (Central) Rules, 2026, together with a rounding rule under which a subsequent period of six months or more counts as a further year.
Two caveats. The Central Rules apply where the Central Government is the appropriate authority — banking, insurance, telecom, mines, railways, central PSUs and similar — so for most other establishments the operative position depends on state rules, which remain unevenly notified. And “fixed-term employee” is a defined term under the Industrial Relations Code: it does not cover contract labour engaged through a staffing agency, where the liability sits with the contractor.
Where it applies, the effect is direct. Force two largely disappears for that population, so the timing rule governs and attrition feeds through to the reported liability far more strongly than before. See our analysis of the new gratuity rules under the labour codes.
5. Using Staggered Attrition Rates
A single flat attrition rate applied to the whole workforce is the crudest defensible approach. Where the data supports it, a staggered assumption improves both accuracy and stability. Common bases:
| Basis | When it helps |
|---|---|
| Years of service | Most valuable basis. Attrition is typically much higher in the first two to three years. A service-based scale prevents the liability being distorted by applying a headline rate to long-service employees whose observed exit rates are far lower. |
| Attained age | Useful where the workforce spans a wide age range, or where older employees are effectively locked in by proximity to retirement. |
| Grade or salary band | Where junior and senior populations behave very differently — common in IT services, BPO and retail. |
| Location or cost centre | Where labour markets differ materially between sites, or where one location is being wound down. |
| Nationality | Relevant in the GCC, where expatriate and national employee populations frequently show very different exit patterns and exit reasons. |
Staggering is particularly valuable for smaller companies where a handful of long-service employees drive most of the liability, and a flat rate would either overstate or understate their expected exit dates significantly.
The constraint is data credibility. Splitting a 200-employee workforce into twelve age-and-service cells produces cells with two or three observations, which is noise rather than information. Stagger only as finely as the data will support, and document the basis.
6. When to Change the Assumption
As far as possible, the attrition assumption should not change year on year. It should change when management’s view of long-term future attrition changes — not when a single year’s experience comes in high or low.
Legitimate reasons to revise:
- A structural change in the business — new sector, major acquisition, significant change in workforce mix
- A retention or restructuring programme that management expects to change exit behaviour durably
- Several consecutive years of experience diverging from the assumption in the same direction
- A move to a staggered basis, or a material improvement in the quality of the underlying data
- Regulatory change altering exit economics — for example the labour code treatment of fixed-term employment
Not legitimate reasons:
- One year of higher or lower attrition than assumed
- A desire to move the reported liability in a particular direction
- Mechanically resetting the assumption to last year’s actual experience each cycle
An assumption that tracks realised experience each year is not a best estimate of the long term — it is a lagging indicator dressed up as a forecast, and it will be challenged.
7. Common Mistakes and What Auditors Ask
| Mistake | Why it is wrong | What to do instead |
|---|---|---|
| Assumption set equal to the reporting period’s actual attrition | Confuses a spot rate with a long-term expectation; imports volatility into the balance sheet | Set a long-term average and document the reasoning |
| Using the HR or business planning rate directly | That rate covers future hires, who have no past service to value | Adjust for the existing-employee population and reconcile to the business number |
| Ignoring planned redundancies or restructuring | Known future exits are absent from historical data | Reflect approved plans explicitly in the assumption |
| Single blended rate where the benefit varies by exit mode | Locks in an implicit mix of exit reasons that may not hold | Split the assumption by exit reason; capture the split in HR data |
| Flat rate across a workforce with a wide service spread | Overstates exit probability for long-service employees, who drive most of the liability | Use a service-based scale where data supports it |
| Setting attrition without reference to the other assumptions | The direction of the impact depends on the salary escalation and discount rate gap | Set and review the three together for mutual compatibility |
| Changing the assumption every year | Signals the assumption is being derived from experience, not set as a best estimate | Change only when the long-term view changes; document the trigger |
| No documented basis or approval | Management, not the actuary, owns the assumption | Record the rationale, evidence and sign-off each cycle |
Auditors will typically ask for the historical attrition analysis, the reconciliation between that analysis and the adopted assumption, evidence of management approval, and an explanation of any year-on-year change. Our post on issues to consider when auditing actuarial valuation reports covers the wider review process.
8. FAQs
Can we just use our actual attrition rate for the year?
No. The standards require a best estimate of future experience over the duration of the liability. Actual experience for one year is an input to that judgement, not a substitute for it. Where the adopted assumption happens to equal recent experience, the file should still show why management believes that rate is sustainable long term.
Should the attrition assumption match our HR attrition number?
It should be consistent with it, but it will usually be lower. The HR figure covers the whole future workforce including new joiners, who experience much higher turnover. The actuarial assumption applies only to employees already on the books at the valuation date, who have already survived the high-attrition early years. If the two numbers are far apart, be ready to explain why.
Does higher attrition always reduce the liability?
No, and this is the most common misconception. Higher attrition brings payments forward, which reduces the present value only if the salary escalation assumption exceeds the discount rate. Where the discount rate is higher, earlier payment increases the liability. Forfeiture rules push in the other direction, so the net effect depends on the scheme and on how much of the workforce sits below any vesting threshold. For a mature, largely vested population the liability can rise with attrition.
Why does the reason for leaving matter, not just the rate?
Because most schemes pay different amounts depending on how the employee leaves. A resignation, a redundancy, a dismissal for cause and a contract non-renewal can attract quite different entitlements under the same plan. A single blended attrition rate bakes in an assumed mix of those reasons; if the mix changes, the liability changes even though the headline rate has not.
How is the attrition assumption different for EOSB schemes in the GCC?
The principles are the same. What differs is the forfeiture rules. Saudi Labour Law scales the award down on resignation — nothing below two years’ service, one-third from two to five years, two-thirds from five to ten, full award only beyond ten — so the resignation-versus-termination split drives the liability as much as the overall rate. Other GCC jurisdictions differ; the UAE removed its resignation-based reduction in 2022, so the Saudi logic should not be transplanted without checking local law.
How much does attrition matter for leave encashment valuations?
Usually much less than for terminal benefits. Leave liabilities are short-duration, so the timing effect has little room to operate, and where the policy allows unrestricted encashment on exit there is no forfeiture effect either. In that case the availment assumption — how much accrued leave employees are expected to take rather than encash — often has a larger impact. Check the policy first, though: where encashment is restricted on resignation, attrition does create a forfeiture effect. Note also that accumulating compensated absences expected to be settled beyond twelve months are other long-term employee benefits under Ind AS 19, so remeasurements go through profit or loss rather than OCI.
How often should we review the assumption?
Review it every valuation cycle; change it only when the long-term view changes. Reviewing and concluding “no change” is a valid and well-documented outcome, and is usually the right one.
Do the new labour codes change how we set attrition?
They do not change the principles, but they can change the impact. Where the Code on Social Security (Central) Rules, 2026 apply, fixed-term employees become eligible for gratuity on a pro-rata basis after one year of service rather than five, so forfeiture no longer absorbs early exits in that population and the attrition assumption feeds through to the liability more directly. Companies with a significant fixed-term workforce should consider modelling that cohort separately. Whether central or state rules govern depends on the establishment, and state rules remain unevenly notified — so confirm the applicable position before changing the basis.
Download our whitepaper on transitioning to Ind AS 19 and Ind AS 102. Click on the picture below:
Need an Actuarial Valuation You Can Rely On?
We provide employee benefit valuations for companies across India and the GCC — gratuity, leave, EOSB and pension schemes — under AS 15, Ind AS 19 and IAS 19. Accurate, compliant, and built to stand up to audit scrutiny.
July 30, 2025 at 4:55 pm, MAHADEVA said:
We have experienced that 3 Nos of drivers leave the job every month and it is 36 per year (3*12). The total No. of employees as given in the active sheet is 194. Therefore, the Attrition Rate can be considered as 36/194*100 = 18.55%.
December 31, 2025 at 8:38 pm, 5 ways in which COVID-19 impacts actuarial valuation • Numerica said:
[…] to this post for more information on how to set the employee attrition assumption for actuarial […]