Like compulsory voting, compulsory pensions have not taken off to a great extent: Australia practices both, Switzerland has had mandatory supplementary pensions since the 1980s, and pensions are compulsory for most workers through collective labour agreements in the Netherlands.

Much has been written in recent years about the ingredients for successful pension design – but two essential levers for a good system are very simply high contribution levels and high rates of coverage.

In the UK it looks like auto-enrolment might capture most of the benefits of compulsory contributions without the political risk that the contributions will be viewed by voters as a tax. Two years into the auto-enrolment roll-out programme, the opt-out rate is around 10%, according to figures compiled by the Pensions Policy Institute, meaning that 90% of individuals have stayed in their pension scheme. But the early phase of auto-enrolment has covered larger employers, most of whom will already have a relatively generous pension fund, making it much more attractive to stay in the scheme.

It is over-optimistic to think that the average opt-out rate will stay as low as 10% in the later phase of auto-enrolment, which includes thousands of SMEs. Most of these small businesses will not already offer a generous pension, and will probably be less motivated to encourage staff to stay enrolled. Taking all this into account, the opt-out rate could climb to as high as 25% overall.

Where the UK scores less well is on contribution rates. The auto-enrolment minimum increases from a derisory 1.8% (not including tax relief) to 7% by 2018, made up of 4% from the employer and 3% from the employee. But this will secure low average pension savings for a large number of people, which risks undermining the long-term viability of the auto-enrolment concept.

Dutch pension fund contributions, which are set by collective agreements, frequently exceed 20%, split between employer and employee. Australian superannuation contributions have increased successively over the past 20 years and are now around 9.5%, although they are set to increase further.

Notwithstanding the debate around contribution levels, auto-enrolment will still add more than £10bn in annual contributions in the UK, according to government estimates. According to the Pensions Policy Institute, defined contribution (DC) assets are set to increase sharply over the next 15 or so years and total DC assets could overtake the stock of DB pension assets by 2036. The likes of NEST, Standard Life, the People’s Pension and Legal & General look set to take considerable share of that extra £10bn.

Currently these providers are digesting the long-term implications of the government’s decision to end compulsory annuitisation from 2015. The annuity market was worth about £12bn a year, but Legal & General thinks it could plummet by as much as 75%.

The withdrawal needs and habits of UK pension savers will probably not become clear for several years. In the meantime, there is potential for considerable thought and innovation on providing for those likely needs. The lowest earners may well want to withdraw their assets as cash, and the highest earners will have access to more sophisticated financial planning. For the middle-income earners, there will be considerable demand for some form of stable, long-term retirement income.

This leads to an important question: what do we demand of a pension fund? Is it to be little more than a long-term savings plan, with individual responsibility for translating assets into income? Or is there a role for pension funds in servicing income needs? Master trusts like NEST and others have plenty of potential to earn their stripes by combining retirement income and innovative annuity products.