Since 1980, the US has witnessed a huge decline in the number of defined benefit pensions in the private sector. In a defined benefit plan, the employer invests all contributions and promises to deliver a defined schedule of benefits. But most companies have ended their defined benefit plans due to accounting and funding regulations.

By contrast, defined benefit pension are still prevalent in the public sector — in cities and states.

Unfortunately, the unfunded liabilities of these defined benefit pensions in the public sector are huge — in 2014, for example, the average state pension plan had only 70% of the assets needed to pay future retirement benefits.

Nevertheless, public sector unions understandably resist the move to a defined contribution plan, such as a 401k plan, since it would shift the risk of underfunding from the public employer to its employees. In such a plan, contributions from employees and the employer are put into individual accounts, where the investment funds are chosen by each employee.

Thus, the challenge is to develop a third type of retirement plan, which manages risk better than a defined contribution plan, while operating at the lower costs of a defined benefit plan. I call this third way — a collective retirement plan.

The key to a collective retirement plan is to invest all employer and employee contributions in one collective pool — with the investment strategies of defined benefit plans, and hence with much lower costs than most defined contribution plans.

This type of collective retirement plan would have several key advantages.

First, the investing of a collective pool will cost a fraction of the expenses of the average mutual fund offered in a 401k plan. That’s because collective investing can be done by pension experts at institutional rates for one large, diversified pool.

Second, the costs of administering a collective retirement plan will be a fraction of those costs for a defined contribution plan. That’s because the plan could avoid the administrative expense of educating participants about the 20 investment choices typically offered by most 401k plans.

Third, the investment choices of a collective retirement plan will be superior to those of the average participant in a 401k plan. Many participants put all of their plan contributions into inappropriate vehicles for long-term investing for retirement, or change Investments too frequently to get optimal returns.

Instead, in a collective retirement plan, independent experts would construct an investment portfolio with a reasonable return at a relatively low risk. In my view, that portfolio should be a balanced fund, with roughly 60% of its assets in a diversified US stock index fund and the other 40% in a diversified US bond index fund — rebalanced each year. Such a portfolio has earned a real return over 6%, with relatively low volatility, over successive long periods.

Finally, a collective retirement plan would not require the employer, a state or city, to recognize an unfunded pension liability on its balance sheet. Rather, the investment risk associated with the plan would be managed by the trustees of the collective retirement plan based on a contingent reserve. In specific, the trustees would establish a “probable” benefit schedule for retirees based on the amount and timing of attributed contributions. That schedule would assume a 5% annual return on plan investments and a cost-of-living adjustment.

How would the collective retirement plan mitigate the risk that it could not deliver the “probable” benefit schedule? To begin with, the assumed Investment return of 5% is below the 6+% historical return of the balanced portfolio described above, and much below the 7+% assumed return of most defined benefit plans in the public sector. At the same time, the collective retirement plan would have much lower costs than any defined contribution plan.

As a result, the trustees of the collective retirement plan could gradually establish a reserve of 3% to 7% of the plan’s asset as a contingent reserve. That reserve could be drawn down if needed to meet the “probable” benefit schedule in the event of a shortfall . If the portfolio’s return plus the reserve were still insufficient to meet the “probable” benefit schedule, the trustees could waive the cost of living adjustments for several years.

In short, most defined benefit plans in the public sectors are going down an unsustainable path. However, if we want to avoid pension crises like the ones in Detroit , we cannot expect public sector unions to accept the investment risks and high costs of the typical defined contribution plan. Instead, we need to create a new type of pension plan.

A collective retirement plan is much less risky and much less costly than the typical defined contribution pension.

While a collective retirement plan cannot provide “guaranteed” benefits at retirement, it can offer reasonable assurances that the benefit schedule

would be met in most instances. And the promised benefits of many public sector plans are not in practice “guaranteed.”

Bob Pozen is currently a Senior Lecturer at the MIT Sloan School of Management, and former President of Fidelity Investments. He has extensive experience in business, government and academia, and has written extensively on financial policy issues

Originally published at