Paradox #20
The Year One in Six Walked Away
→ 4.8x spike in 3 years. Almost 1 in 6 civil servants left in 2023 alone.
Referenced as sidebar in Chapter One
Civil service separations 2020
1,084
normal year
Civil service separations 2023
5,202
= 17% of entire civil service in one year
The full numbers (from RCSC CSS 2025):
| Year | Separations | % of CS workforce |
|---|---|---|
| 2020 | 1,084 | ~3.5% |
| 2021 | 1,504 | ~4.8% |
| 2022 | 2,646 | ~8.4% |
| 2023 | 5,202 | ~17.2% |
| 2024 | 2,013 | ~6.6% |
| 2025 | 1,948 | ~6.4% |
| 6-year total | 14,397 | ~half the workforce turned over |
| In 2025 alone, voluntary resignations: 1,180 (60% of total separations). Plus 359 contract completions, 138 contract terminations on regularization, 143 superannuations, 44 deaths, 32 administrative terminations, 22 compulsory retirements, 24 cancellations. |
Imagine this
A senior civil servant in a ministry returns from his summer holiday in 2023 to find that his department of 12 has lost 4 people while he was away. Two went to Australia. One joined an NGO. One started a private consulting firm. He looks around the office and realises he doesn’t recognise most of the new faces — they’re contract hires brought in to fill gaps. The institutional memory of his department has been hollowed out in a single quarter. This happened across the entire civil service in 2023. 5,202 people left. Roughly one in six government employees walked out the door. Many were mid-career professionals (P4 and P5 levels) — the engine room of the civil service. They took with them years of institutional knowledge, project context, relationships with stakeholders, understanding of historical decisions. The colleagues who stayed were left to absorb the work, train the replacements, and continue delivering on the 13th FYP — all without the people who had been doing the work the year before.
Where this came from
The 2023 spike is multi-causal. The Civil Service Reform Act of 2022 introduced new performance management frameworks and tightened tenure expectations — pushing some out. Economic stress from the post-COVID slow recovery limited civil service salary adjustments. The Australia migration pipeline (paradox #13) reached its peak draw. The combination created a perfect outflow window: people who had been considering leaving for years suddenly found the conditions aligned. They left.
Why this matters now
Two years later (2025), separations have moderated back to ~6.4%, but the underlying conditions haven’t fundamentally changed. The institutional damage from 2023 is still being absorbed — promotions accelerated to fill gaps (4,183 promotions in 2025, the highest in 5 years), contract hires expanded to plug knowledge holes. The civil service that exists today is not the same civil service that existed in 2022.
What it should be
Healthy government workforces lose 5-7% of staff per year — the natural turnover that brings in new energy without destroying institutional memory. 17% in a single year is workforce restructuring territory.
How others do it
- OECD government average attrition — 5-7% per year.
- Singapore civil service (premium standards) — about 5%. Highly competitive recruitment + retention.
- India — about 3%. Very rigid tenure; the opposite problem (under-turnover).
- UK Civil Service — about 9% (post-Brexit + post-COVID). Higher than usual but recovering.
- Crisis-era Asian civil services — rarely exceed 10% in a single year, even in countries undergoing serious economic stress.
- Bhutan 2023: 17% — 2-3x even crisis-era norms.
The question we should be sitting with
When 1 in 6 of your colleagues left in a single year, what does that say about the system you’re inside? What did the people who stayed conclude — and what did the people who left conclude? Have we learned what to fix, or have we just stabilised at a new lower baseline?