Calls to limit retirement balances to $2.5m are getting wide support. It sounds fine in theory, but how will it operate and will it have an impact?

I can see two broad ways of applying the cap. Firstly, it could operate like the old Reasonable Benefit Limit system whereby the benefits would be measured against the cap whenever a superannuation benefit (ie lump sum or pension) was initiated. This approach would be relatively simple but would permit capital growth within the pension phase to go well beyond the $2.5m cap. As such, this approach doesn't appear to achieve what proponents of the cap desire.

The alternative would be to measure against the cap at regular intervals. If the cap is exceeded, presumably the excess would need to be withdrawn.

The alternative approach seems simple and addresses concerns about post-pension growth. However, there are a number of issues that need to be addressed if this approach is applied:

- what is measured against the cap? Is it merely the current super balance or should previously taken benefits (eg lump sums) be considered as well? If the mischief is to put a cap on the use of the tax-free status of pension earnings, then it should be the former. Otherwise, the compliance burden will be onerous

- who's job should it be to administer? Many fund members will have multiple superannuation interests, so presumably it will be the fund trustee's job to administer particular interests if they exceed $2.5m alone but with the member to have some responsibility if aggregated interests exceed the cap

- should the cap be administered absolutely? We have seen balances slaughtered during the GFC. Imagine the double impact on individuals and the overall impact on retirement savings if members affected had been forced to make cash withdrawals in a falling market. Market forces will cause balances to rise & fall quite suddenly, depending on how funds are invested. In light of this, I would propose a yellow and red card system where a one-off cap breach puts the balance on watch and corrective action is only required for multiple breaches

- how often should the cap be applied? Compliance costs would be better managed if the cap was applied no more than annually. However, this is probably too frequent. In this regard, one of the main influences on the balance will be the performance of investment markets. The measurement of the cap would be more meaningful if applied according to investment cycles, say once each 3 or 4 years

- how should the forced withdrawal be taxed? I propose that the withdrawal should be taxed like any other withdrawal - eg currently tax-free if over age 60. The reason why the cap is proposed in the first place is to ensure that the superannuation system is sustainable and that a member's balance is not out-of-step with the policy settings of the system. Members should not be penalised if they have invested well and part of their capital base needs to be shifted outside super to be taxed like ordinary investments

Like many reforms, things that sound simple are often deceptively so.