The Auto-Enrolment Completion Loop
Auto-enrol the easy 90%, then go hunt the missing 45% deliberately
Auto-enrolment was the headline UK pension policy of the last 20 years, and Bell celebrates that 9 in 10 eligible employees now save. But he insists the work is half done: 45% of working-age people still save nothing, because auto-enrolment only catches the standard employer-employee relationship. The framework is a two-stage loop — first lock in the easy default, then deliberately design coverage for everyone the default missed.
The second stage is the harder one. Self-employed saving has collapsed from around half to one in five, lowest earners are uncovered, and some ethnic minorities save at lower rates. The framework treats these not as residual problems but as the next mandatory phase of the policy — without them, headline coverage stats are misleading.
The operator's job is to resist celebrating the 90% number, identify exactly who the default cannot reach, and build a parallel mechanism that doesn't rely on the missing employment contract.
- Defaults beat education for the population the default can reach.
- Coverage stats are lies until you publish the share of the working-age population, not just the eligible.
- Every default leaves a population it cannot touch — name them explicitly.
- Self-employed saving requires a non-employment-relationship mechanism, not a tweaked default.
- Phase the default, but put a deadline on the gap-fill stage.
- Set the default and lock the opt-outBuild a hard nudge into the existing payroll or transactional flow. The user must take action to opt out, not opt in. This is the cheap win that delivers the 90% number.WarningA soft nudge (suggestion, prompt) gets a fraction of the result. Auto-enrolment works because the cognitive cost is on opting out.
- Publish coverage as % of total population, not % of eligibleForce yourself and stakeholders to look at the denominator that includes everyone the default missed. UK debate confused 90% of eligible with overall coverage — the real saving rate is closer to 55%.Pro tipReport two numbers side by side every time: eligible coverage and total-population coverage.
- Segment the missing populationBreak the uncovered group into actionable segments — self-employed, low earners below the qualifying band, ethnic minority cohorts with lower take-up, gig workers. Each needs a different mechanism.
- Design a non-employment saving mechanismFor the self-employed, the employer-payroll anchor doesn't exist. Build the saving moment around the closest equivalent transactional event — tax filing, invoicing platform, banking app — so the nudge attaches to a recurring action they already take.WarningDon't just write another report saying self-employed don't save. The hard yards are in mechanism design.
- Solve decumulation in parallel with accumulationMost DC schemes have no built-in drawdown route. Build a default decumulation path so savers don't have to become hyper-engaged investors at retirement. Otherwise some over-consume and run out, others under-consume and live too thin.Pro tipTreat decumulation as the next auto-enrolment moment — build a default, allow opt-out.
- Set a public deadline for stage twoGive the second stage the same forcing function as the first: a named commission, a named year, a published report. Without a deadline, the gap-fill stage gets kicked indefinitely.
Hard nudge into employer schemes 'more or less doubled the number of private sector employees who are eligible who are saving.' But the headline obscured that self-employed coverage collapsed in parallel.
Started earlier, took longer to reach today's contribution levels. Bigger pots but underdeveloped decumulation — Bell sees retirees consuming too little because no default drawdown exists.
Auto-enrolment applies to earnings between roughly £6k and £50k. Higher-rate taxpayers must claim extra 20% relief themselves; many don't. Coverage looks good, real contribution is lower than reported.
Bell traces the model to the Turner Commission, which in the mid-2000s recognised that voluntary workplace saving was dying and that only a hard nudge — auto-enrolment — would reverse it. He now argues the commission's logic was always two-stage: stage one was the default, stage two was always going to be filling the gaps it left. The 2025 revived Pensions Commission is, in his framing, simply executing stage two on schedule.