Pensions Talk Interview: Kim Gubler, Managing Director, KGC Associates

This month, we caught up with Kim Gubler, Managing Director, KGC Associates who offered valuable insight into the Government’s pensions reform agenda, improving retirement outcomes, addressing the self-employed pensions gap, and unlocking productive investment. The conversation also examined the future of DB surpluses, the transformative potential of pensions dashboards, and the growing role of AI in pension administration, alongside the governance and oversight needed to support innovation.

1. 1. With Andy Burnham set to be Prime Minister and expected to pursue a more interventionist economic agenda, what do you believe this means for the pensions industry? Which areas of policy from pension fund investment and adequacy to surplus extraction and member outcomes are most likely to see significant change over the course of this Parliament? 

Andy Burnham may bring a more interventionist economic philosophy, particularly around regional investment and the role of institutional capital. However, I do not expect another wholesale redesign of workplace pensions during this Parliament. 

The Government’s updated roadmap, published this week, already contains a substantial and highly sequenced programme of reform. Value for Money, consolidation, small pots, guided retirement, CDC and the new scale requirements will reshape the market and potentially have a profound effect on people’s eventual retirement incomes. The immediate priority should be delivering those reforms well rather than continually adding new ones.

The next major policy debate will be adequacy. The Pensions Commission has already estimated that 15 million people are undersaving for retirement and will make its final recommendations in early 2027. The difficulty is that the need to increase saving is becoming clearer at precisely the point when many households are struggling with day-to-day costs and employers have absorbed significant increases in employment costs. Simply requiring both parties to contribute more may be economically and politically difficult.

That means the Government will need to consider not only contribution rates, but how and when they increase, who bears the cost and how lower earners are protected. Phasing, better targeting and using behavioural defaults will all be important. 

The greatest unresolved gap remains the self-employed. Only a very small proportion are contributing to a pension. Too many are relying on the future sale of a business, property or some other distant windfall which may never materialise. Finding a practical mechanism which makes pension saving normal and easy for the self-employed should be one of the next major areas of reform. 

I would therefore expect the emphasis under a Burnham government to be less on creating an entirely new pensions structure and more on adequacy, productive investment and ensuring that the reforms already enacted produce tangible benefits for members.

2. 2. The Government's proposals to make it easier for well-funded DB schemes to release surplus have sparked significant debate across the industry. How should trustees balance protecting member benefits with the opportunity to unlock value for sponsors, and what does good practice look like when deciding how surplus should be shared? 

Member benefits must remain the starting point, but protecting members does not necessarily mean that surplus can never be used. The new legislation gives trustees greater flexibility, but it does not turn surplus extraction into an automatic employer entitlement. Trustees must still act in accordance with their scheme rules and fiduciary duties, consider the scheme’s funding and investment risks and understand the strength of the employer covenant. It has to be noted though that employers have been paying millions of pounds into their schemes for decades – money that could have been invested in their businesses. So if they do claim some of their surpluses, this could redress that balance. Employees and the economy need strong growing businesses. But of course, this needs to be worked through carefully.   

From the arrangements I have seen developing, the most credible approaches use clearly defined upper and lower funding buffers. Surplus may become available above an agreed level, but distributions stop if funding falls towards the lower threshold. The buffer should be sufficiently robust to withstand adverse investment experience, changes in longevity and covenant deterioration without putting members’ benefits at material risk.

Trustees should also consider the full range of possible uses. A payment to the employer is one option, but surplus could potentially support discretionary member improvements, meet the employer’s DC contributions or fund other agreed benefits. There is unlikely to be a single sharing formula which is right for every scheme.

Good practice requires:

• an independently tested long-term funding and investment strategy
• clear triggers for starting, stopping and reviewing payments
• proper stress testing and actuarial certification
• consideration of member interests, including whether members should share directly in the benefit
• transparency about why the proposed allocation is fair
• appropriate member communication
• a documented decision-making process

Most importantly, trustees and employers should work collaboratively from the design stage. A strategy presented to trustees as a completed employer proposal is far less likely to build confidence than one developed openly, with the risks, protections and potential benefits understood by both parties.

3. 3. After years of planning and preparation, pensions dashboards are moving closer to becoming a reality. In your view, will dashboards genuinely transform member engagement and retirement planning, or is the industry placing too much expectation on them? What are the biggest hurdles schemes and administrators still need to overcome to ensure a successful rollout?  

Dashboards will be transformational, but not necessarily immediately and probably not in quite the way some people expect. Their first achievement has already happened. The requirement to connect has forced schemes to examine data which might otherwise have remained untouched for years. The overall condition and accessibility of pension scheme data is undoubtedly better than it would have been without dashboards.

That does not mean it is uniformly accurate. Data quality remains one of the principal risks, particularly because matching depends on the personal data held by schemes being sufficiently complete and current. A pension which is not matched may appear to the user not to exist. Conversely, weak matching criteria create a risk that information is disclosed to the wrong person.

The other major uncertainty is member behaviour. We do not yet know how many people will use the MoneyHelper dashboard, how many will contact schemes because a pension has not appeared, or how many will query information which looks unfamiliar or inconsistent. Administrators need to plan for those contacts now, including how they will distinguish between a dashboard issue, a data problem and an underlying benefit query.

Trustees also need to recognise that connection is not the end of the project. There are demanding ongoing requirements around availability, response times, record keeping, reporting, complaints and remediation. Schemes are expected to target 99.5% availability and retain detailed records of dashboard-related member contacts and operational failures. That level of compliance oversight is not yet fully understood everywhere.

This is only version one. We have to start somewhere. Over time, going to a public or commercial dashboard to see all your pensions in one place should become entirely normal. That is when the Government’s ambition of helping people to “find, understand and act” can truly be realised.I am not convinced that dashboards will trigger an immediate wave of consolidation. Some people will consolidate, but many will simply gain a clearer understanding of what they have. Combined with the Value for Money framework and better retirement support, that increased visibility should ultimately help members make better decisions and improve retirement outcomes.

4. 4. Artificial Intelligence is rapidly moving from theory to implementation across financial services. Where do you see the most immediate opportunities for AI within pensions administration, and what governance, security and data-quality safeguards need to be in place to ensure innovation doesn't come at the expense of member trust? 

AI is not something which is coming to pensions administration at some distant point. Many administrators are already using it, although the sophistication and governance of that use varies considerably. The most immediate opportunities are in augmenting people rather than replacing them. AI can classify incoming correspondence, extract information from documents, summarise case histories, draft member communications, support call handlers and identify the next appropriate step in an administrative process. It can also help detect anomalous transactions, possible fraud, recurring errors and cases which are at risk of missing service standards.

Used well, that should reduce the amount of time skilled administrators spend searching systems, manually rekeying information or processing straightforward cases. It creates more capacity for the difficult work which genuinely requires judgement, technical expertise or empathy. There is also considerable potential to improve member communication. AI can help translate complex pension terminology into clearer language, tailor information to a member’s circumstances and make communications more accessible. However, it must not drift into unregulated financial advice or present uncertain information as fact.

Trustees are right to be curious and appropriately sceptical. Blanket resistance to AI is unlikely to protect members, because it may simply mean that trustees do not understand how their administrators are already using it. The better approach is active governance. Trustees should know:

• Where AI is being used across the service
• What member or scheme data is being provided to it
• Whether that data is retained or used to train external models
• Who remains accountable for the output
• How errors, bias and hallucinations are identified
• Which decisions require human review
• How models and controls are tested over time
• What happens if the technology or supplier fails

There should be strong access controls, data minimisation, clear contractual protections, cyber resilience and a complete audit trail. Administrators should also be able to explain an AI-supported decision rather than simply stating that the system produced it.

Data quality is fundamental. AI can process poor data more quickly, but it cannot turn it into reliable information. In fact, deploying AI over inaccurate or inconsistent data can industrialise an existing problem.

The objective should therefore not be to use AI because it is fashionable. It should be to solve a defined service or member problem, with a clear owner, measurable benefit and proportionate human oversight. Innovation and member trust are not competing objectives. Properly governed, each should reinforce the other.

5. 5. If you could ask the Government to prioritise just one pensions reform over the next 12 months, what would it be and why?

I would prioritise creating an effective pension-saving mechanism for the self-employed. Automatic enrolment has transformed pension participation among employees because it makes saving the default. The self-employed have been largely excluded from that success. Only around one in 25 wholly self-employed people is currently saving into a pension, leaving millions at risk of reaching later life with little beyond the State Pension.

The answer cannot simply be another communications campaign telling people that saving is important. Most already understand that. What is missing is a frictionless mechanism which fits the way self-employed people earn and pay tax. Government should develop and test a default contribution arrangement linked to the tax or National Insurance system, with the flexibility to accommodate irregular income and a clear right to opt out. Contributions could be collected periodically rather than through a conventional monthly payroll model. 

There would be difficult questions around affordability, tax treatment and the genuinely low-paid self-employed. However, unless the system moves from encouragement to an effective default, this group is likely to remain the largest structural gap in UK pension provision. We should finish implementing the reforms already under way, but adequacy will not be solved while millions of self-employed people remain outside pension saving altogether.