MANY workers depend on their employers for their retirement income. But, for defined-benefit schemes, this is an explicit bet that their employer will still be around several decades later: quite a gamble. So governments in Britain and America have set up insurance schemes designed to protect workers against the risk that their companies go bust. These bodies, Britain’s Pension Protection Fund (PPF) and America’s Pension Benefit Guaranty Corporation (PBGC), are funded by levies on employers.

Now that Tata Steel’s loss-making British operations are up for sale, the chances are that the company’s pension fund will end up in the PPF’s clutches before too long. Buyers are likely to cherry-pick the steel company’s assets, which may leave the scheme without a viable sponsor—contributions into the fund were £155m ($219m) in the most recent financial year.

And because the scheme is a legacy of the old nationalised British Steel, it is huge, relative to the existing business: its assets are almost £14 billion compared with annual turnover of just £8.1 billion at Tata Steel’s European operations.

Fortunately, the scheme is pretty well-funded, thanks to a policy of buying inflation-linked government bonds to match its liabilities closely. As of March 2015 it had a deficit of £485m (or 3% of assets). That would still be a lot for a private buyer to agree to fund. But it is not that big a burden for the PPF, which had a surplus of £3.6 billion at the time of its last report.

Things don’t look quite as healthy over at the PBGC, which has been around for much longer.

As of its 2015 report it had total assets of $87.7 billion and liabilities of $164 billion—a deficit of $76.3 billion, a record (see chart). The big problem is the multi-employer bit of the PBGC’s responsibilities, where the deficit is $52.3 billion.

Multi-employer schemes cover industries such as mining and trucking, in which a number of companies contribute to a collective pot. As the industry shrinks and individual employers go bust, the financial position of such schemes deteriorates. They end up in the arms of the PBGC when they run out of money to pay benefits (normally, when a single company goes out of business, there are enough assets to cover most of the liabilities). So when a multi-employer failure occurs, the PBGC’s liability is accordingly huge. It makes provision on its balance-sheet for funds that it expects to run dry over the next decade. But Congress has provided for the PBGC to get a levy of only $27 per employee per year. That adds up to an annual payment of around $270m, woefully inadequate to cover a $52 billion liability.

On March 31st the PBGC warned that it will require significantly higher premiums to keep its multi-employer scheme running. And last year it said, with respect to the scheme, that “the risk of insolvency rises over time, to exceed 50% in 2025.” It added that the risk increases to more than 90% within 20 years. “When the programme becomes insolvent, PBGC will be unable to provide financial assistance to pay guaranteed benefits in insolvent plans,” it concluded.

The PPF is in a much stronger financial position because it has been better funded. But the fundamental problem for such insurance schemes is the changing nature of industry and of the type of pensions promised by employers. Defined-benefit promises, where the pension is linked to a worker’s final salary, are the most expensive for companies to fund. They were the norm in the 1960s and 1970s when developed economies had a much bigger focus on heavy (and unionised) industry. Those sectors have now shrunk in size but they still must bear the legacy cost of the pensions promised to former workers. British Steel’s pension fund has fewer than 17,000 working members but more than 86,000 claiming retirement benefits.

Meanwhile, the businesses that have emerged in the past 20 years have tended to be non-unionised and to offer defined-contribution pensions, in which retirement income is not guaranteed by the employer. Such schemes are not part of the PGBC or the PPF; there is no promise for an insurance scheme to back. So there are no new companies to pay the levy; an ever-smaller number of employers are funding a huge historic liability.

Congress needs to pull its finger out. The PBGC can hardly cut benefits any further: a worker with 30 years’ service in a multi-employer scheme will get less than $13,000 a year, with no inflation protection. If Congress doesn’t want to charge employers more, it should fund the PBGC directly.

Setting up an insurance scheme, and then failing to fund it adequately, is a betrayal of its constituents.