Good pensions finance good infrastructure. Good infrastructure pays for good pensions. This crucial relationship only gets noticed when they both go missing — as the US, UK and several other countries are finding out. Having largely dismantled the defined benefit corporate pensions of yesteryear, they now struggle to turn fragmented individual pensions into the long-term investments their savers and their economies require. Fixing this is vital. It will not be easy.

Given the human lifespan, pension savings are the natural source of capital that can be tied up for 30, 40 or 50 years. In return, they earn the premium that comes from volatile or illiquid assets, which is all the more valuable when interest rates are low. But as Bank of England governor Andrew Bailey noted in a speech, something has gone wrong. “We live in a time where there appears to be no shortage of aggregate saving, but investment is weak,” he said. UK pension funds allocate just 3 per cent of their resources to unlisted equity.

The UK hopes to address this problem by easing regulations, allowing defined contribution pension funds — where individuals bear the investment risk — to hold more illiquid assets, and loosening caps on fees to allow for complicated investments such as infrastructure. But even if such changes have no unintended consequences, they will not address the fundamental challenge of a fragmented pension system, where decisions fall to individuals and it is hard to link their lifespan to the assets they own.

It is time to consider a more radical transformation, away from employer-based pensions towards large, permanent vehicles that can pour money into infrastructure and private equity if that makes sense. Traditional pension funds for public employees and sovereign wealth funds such as Temasek and the Government Investment Corporation of Singapore already do this. This is not a matter of who takes the investment risk — for better or worse, defined contribution is here to stay — but of how the money is managed.

A look at the options available to me as a Financial Times employee through its defined contribution pension plan makes the difficulty clear. There are about 200 different equity, bond and property funds, from a range of providers, actively or passively managed, covering different regions of the world. They all show yesterday’s market price. The choice is left to the individual.

This creates a series of problems. The best investment brains in the world spend their days trying to figure out which asset or region will outperform. An individual has no chance — although they may choose to avoid any option that seems risky.

Even if there was an infrastructure fund, or a venture capital fund, and the default channelled savers towards them, this structure would pose a problem. The fund managers have no idea who their investors are, or when they are likely to retire. They know it is pension money, and therefore likely to be “sticky”, but they still have to provide regular prices for the fund and keep cash on hand in case some investors suddenly cash out. The structure is simply not suitable for illiquid, long-term investments, to the detriment of savers and the economy.

It is worth asking, also, whether employer-based pensions still make sense. When employers took the investment risk, the arrangement was logical, but all it creates now is fragmentation. Every time people change jobs, they get new pensions; small plans have high fixed costs. There are economies of scale: the smaller and more numerous the schemes, the more gets wasted, and the harder it is to make sophisticated investments. Giving everyone their own personal pensions is a mistake for the same reason.

Consider, instead, the following structure. The government licenses a modest number of not-for-profit pension plans, perhaps based on existing endowments, trusts or public sector funds. Employers would decide their pension contributions, as they do today, but then make the payments to whichever plan their employee selects. The plan would decide how to invest the money, subject to regulations, and employees could not withdraw the funds until retirement.

This would still be a defined contribution system but it would work quite differently. The pension plans would quickly become large, giving them economies of scale, lower costs and the resources to handle sophisticated investments. They would know exactly when they needed to pay out pensions and could plan liquidity accordingly. The burden on companies and individuals would disappear. This is a structure for low-cost, long-term investment.

It may seem paternalistic. Certainly, individuals who want to manage their own investments should be allowed to. But consider also where the current set-up is heading. The OECD recently warned governments against tapping into private pensions to fund pet projects such as renewable energy. There is growing demand for state investment to build infrastructure and for public pensions because private provision is inadequate.

The old defined benefit pension plans were great economic institutions: sophisticated pools of private capital with a long investment horizon. Sadly, the security they offered pensioners is no longer achievable. We must strive to offer defined contribution pensions of similar quality instead.

Letter in response to this column:

Infrastructure investing is no pension fund free ride / From Bernard H Casey, Social Economic Research, London, UK, and Frankfurt, Germany