While not disputing this conclusion, a recent roundtable hosted by the Academy of Social Sciences in Australia and the Australian National University's Tax and Transfer Policy Institute noted the importance of the Parliament confirming the system's objective and addressing outstanding issues so our maturing system actually delivers on its potential.

The review suggests the system's objective be "to deliver adequate standards of living in retirement in an equitable, sustainable and cohesive way".

Based on this objective, as the report makes clear, it is high time to settle the pensions phase. People need to understand how they can convert their accumulated resources into products that "deliver adequate standards of living in retirement". For too long the public debate has focused only on the accumulation phase, in particular the superannuation guarantee.

Much of that debate has been ill-informed, not properly linked to the system's objective. On the left there are those who pretend the superannuation guarantee can be increased without any cost to wages, and on the right there are those who pretend increasing the guarantee is all cost and no benefit. But "adequate standard of living" means for most maintaining their pre-retirement living standard: the system is therefore all about spreading lifetime earnings. The challenge is to set the cost during working life commensurate with the benefits of maintaining living standards in retirement.

There is a legitimate debate about the appropriate level of the superannuation guarantee but the precise level now required - 9.5 per cent, 10 per cent or 12 per cent - depends on what retirement income products should be on offer and on exactly how the means test should operate. Unfortunately Callaghan does not provide much useful guidance on either of these key issues.

The proposed new retirement covenant for fund trustees, while delayed to July 2022, offers better guidance than Callaghan on appropriate retirement income products.

According to the government's 2018 position paper, the covenant will require funds to offer retirees up to three "flagship products". For example, those likely to be eligible for full-rate pensions will already have protection from outliving their savings so their "flagship product" is unlikely to include a significant life annuity. Those unlikely to be eligible for any age pension, on the other hand, should be encouraged to direct a significant proportion of their savings into a life annuity to maintain their standard of living; those in between should consider devoting some lesser amount towards a life annuity.

It is evident that too many people are unaware that trying to manage on their own the risk from not knowing when they will die is highly inefficient: a life annuity represents far better value for money. It can also give them confidence when choosing how to allocate the rest of their saving, for example between an allocated pension and a capital reserve whether for discretionary spending, a contingency against other uncertainties or a planned estate.

Roundtable participants endorsed Callaghan's view that funds regularly advise members about the incomes their savings are likely to be able to deliver in retirement - a requirement under the proposed covenant.

Callaghan rightly identifies "cohesion" as a central element of the system's objective. Current trends suggests that about 60 per cent of retirees will be eligible for some age pension when the system matures, about half being eligible for a part pension, so most retirees will rely on a mix of superannuation and pension.

The central cohesion challenge therefore is to have a means test people can plan around with a degree of certainty and one which, while concentrating assistance on those most in need, retains effective incentives.

Australia is alone in operating separate income and assets tests, the two are not consistent and, as demonstrated in 2017, they may change for short-term political purposes.

The post-2017 assets test taper may allow, as Callaghan suggests, some marginal increase in retirement income from a marginal increase in savings, but only if the savings are drawn down at very high rates; even then the marginal increase in retirement income is not commensurate with the savings involved. Where the extra savings are compulsory superannuation guarantee contributions, the taper effectively turns the superannuation guarantee into a form of taxation.

A simpler, more equitable and more stable means test could be introduced. My own preference is:

In the 1960s the conversion rate was 10 per cent: a lower rate would be appropriate today given increased life expectancy. A rate slightly above the indexed annuity a retiree could purchase, say 8 per cent, would encourage the purchase of annuities. With the 50 per cent income test taper, this would have a similar impact to the 3.9 per cent assets test taper that existed prior to 2017. To limit costs, the assets thresholds should be reduced; if necessary, the income test taper could also be slightly increased.

The other priority concern for the pensions phase is the inadequate support for those in private rental accommodation and those forced to retire early. Increasing rental assistance would be by far the best use of taxpayer dollars in terms of alleviating poverty, despite Callaghan's confusing comments. Newstart also needs a more substantial increase than the government has proposed, and should be indexed by wage movements as the pension is. If this is too hard, then at least make the improvements for recipients over 50.

While there was no consensus at the roundtable about tax arrangements, most rejected Callaghan's use of a comprehensive income tax as the benchmark for identifying superannuation "tax concessions". Taxing superannuation that way would impose a huge penalty on long-term savings.

Like farmers' income averaging arrangements, superannuation tax arrangements take into account how superannuation spreads incomes (over lifetimes). The post-Turnbull tax regime for superannuation is complex but about right in my view and does not unduly favour the rich (the exception being the use of self managed super funds for tax avoidance purposes).

Not surprisingly, the roundtable of experts preferred to see housing assets treated like other assets for tax and means test purposes. That is probably a bridge too far but what could be done now is to make it easier for people to draw on those assets if they wish, including through the pensioner loans scheme.

A package of reforms is still needed for the pensions phase of the retirement income system - the retirement covenant and the products funds offer, a merged means test and improved rent assistance. Get these right, and perhaps the superannuation guarantee does not need to increase to as high as 12 per cent.