STUART THEOBALD | Informal lenders benefit most from new pension rules

We are almost two years into the biggest reform of the pension system for decades.

The two-pot system, introduced in September 2024, allows pension savers to withdraw a portion of their retirement savings without leaving their jobs. The question now is whether it is working as intended.

The opening act was dramatic. In the first nine months R57bn was withdrawn from retirement savings. Most of it went to paying down debt — particularly to nonbank lenders — with some covering education costs. The Reserve Bank did revise its consumption forecasts upward in late 2024, crediting two-pot payouts as a growth tailwind, and Black Friday spending that year was the strongest since before Covid-19.

However, the uplift was modest relative to the scale of the withdrawals, and formal retailers reported no meaningful surge. A large consumption splurge did not occur. What did happen was that withdrawals were taxable as income, and the revenue service collected about R15bn — almost three times what it expected. That helped the tax agency beat its targets handsomely.

The system’s long-term design is straightforward. One-third of contributions flow into a savings pot, accessible once per tax year with withdrawals taxed at marginal rates. Two-thirds go into a retirement pot, locked until retirement. The point was to end the destructive habit of workers cashing out their entire retirement savings when changing jobs. At the very least, that has been achieved. Two-thirds of contributions are now preserved for retirement. That is unambiguously an improvement on what came before.

The worry is what is happening to the savings pot. The tax logic strongly favours leaving it alone. Withdrawals are taxed at your marginal rate — potentially pushing you into a higher bracket. Money left invested compounds tax-free and can eventually be taken as a lump sum at retirement with considerably more favourable tax treatment. The savings pot should, in theory, function as a buffer that most members rarely touch.

It is not working that way. Alexforbes data shows 67% of members who withdrew in the most recent tax year are withdrawing again — half of them in the first month of the new tax year, the earliest opportunity — and taking the maximum available. This is not emergency access; it is habitual, and it is costing members substantially in tax and foregone compound growth.

This behaviour also reveals a flaw in my own criticism of the two-pot system. When it was tabled, I argued for a penalty tax approach instead of forced preservation — a model closer to the UK system, which allows withdrawals for clearly defined expenses but taxes them at 55%. That seems to work there.

It is not working here. The evidence shows that South African savers are so pressed for cash that a substantial tax penalty is not a deterrent. The rational utility optimisation the penalty approach assumes is not the behaviour displayed.

What is shown is hyperbolic discounting — applying a high discount rate to future cash flows relative to present ones. Studies consistently show that the lower your income, the higher your discount rate: the more urgently you want cash now relative to any future benefit. There is also evidence that this is more pronounced in fragile societies, where low trust in institutions and an uncertain future make long-term saving feel futile.

The debt repayment is not showing up in bank data, though. Banks have reported no meaningful paydown of loans. The working theory is that savers are repaying informal sector debt — to mashonisas and loan sharks.

The behavioural economics literature (Mullainathan and Shafir’s work on scarcity in particular) argues that financial stress actively narrows cognitive bandwidth, making long-term planning harder and reinforcing the very decisions that perpetuate scarcity.

The spending data supports this diagnosis. Research from Old Mutual and Sanlam shows about a third of withdrawals are covering living expenses — food, rent, electricity and transport — and between a quarter and half are going to service debt. No-one is buying flat-screen TVs. This is distress spending, not discretionary spending.

The debt repayment is not showing up in bank data, though. Banks have reported no meaningful paydown of loans. The working theory is that savers are repaying informal sector debt — to mashonisas and loan sharks. That would also explain why R57bn flowing out of retirement funds has produced so little measurable macroeconomic impact. The money is circulating in an informal economy largely invisible to standard data collection. So far, the most consistent beneficiaries of the two-pot system appear to be informal lenders.

A darker scenario lurks in that finding. If the members currently building retirement pot balances are also accumulating informal debt at high interest rates, they risk arriving at retirement with annuities that are immediately absorbed by creditors. The two-pot system may be improving preservation while doing little to address the underlying debt spiral that drives the withdrawals in the first place.

That said, the overall trajectory is probably positive. As the current generation of savers reaches retirement, they will arrive better funded than under the old system. That experience — of actually having retirement income — may over time shift the sense of financial futility that underpins hyperbolic discounting. Behaviour may settle down as trust in the system builds.

But the policy job is not over. The two-pot reform has improved preservation while exposing a different problem: South Africans at the bottom of the income distribution do not appear to have access to well-priced formal credit when they need it. They are borrowing expensively from informal lenders and then raiding their retirement savings to repay them.

More accessible formal-sector lending at the bottom of the market would address a root cause the two-pot system has revealed but cannot fix. More layers of the onion remain to be peeled.

• Dr Theobald is founder and chair of research-led consultancy Krutham.