WHEN is a debt not a debt? When the money is promised to prospective pensioners. That appears to be the underlying message of a package unveiled this month by the Polish government.

The Polish pension system, set up in 1999, has two tiers, in which workers make mandatory contributions into a state system (ZUS) and into funds run by private fund managers (OFEs), with their retirement income determined by a complex formula. The authorities are planning to absorb the government bonds held by OFE schemes, cancel them and thereby reduce the state’s debt-to-GDP ratio by around eight percentage points. The government says the pension rights of OFE scheme members will be unaffected.

The headline debt-to-GDP ratio will indeed fall, which might impress the markets. But in effect, Poland intends to replace an outright liability (the bonds) with a contingent liability (the state-pension promise). Fitch, a ratings agency, dismissed the reforms as “broadly neutral” for the country’s debt rating.

But if Fitch is right, what is the point of the reform in the first place? One possible explanation is that the Polish government was approaching a debt-to-GDP ratio of 55%, a threshold it had legally promised not to cross. But the government has said it will lower the threshold to take account of the impact of the pension reform.

Polish workers have a right to be cynical. Defaulting on a government bond, even in a small way, is a highly public and market-moving act. Reducing the value of a state pension is a much easier trick to pull off. A standard sleight-of-hand is to adjust the inflation-linking formula behind pension payments so that the real value of the pension falls over time.

The assumption that governments will default, at least in part, on pension promises is one reason why public-debt-to-GDP numbers do not include the present value of future pension promises. If they were included, the numbers would look a lot worse. A report from Fitch earlier this year suggested that the pressure of ageing populations could push up the average debt-to-GDP ratio of European Union nations by almost seven percentage points by 2020 and a staggering 111 points by 2050.

The sheer scale of this liability means that pension “reforms” (in plain English, benefit reductions) will have to be pushed through to prevent fiscal meltdown. But that leaves politicians in the awkward position of promising voters benefits that simply cannot be delivered.

Many governments have sought to head off the fiscal burden of their pension promises by encouraging the establishment of private schemes. In theory, workers on private pensions are less likely to become dependent on the state; in addition, private pension funds create a potential investor base for the country’s capital markets. It might seem odd, therefore, that many private pension schemes invest in government bonds, particularly at current low yields. Why set up an elaborate pension-delivery system if retirement incomes still end up, in effect, being another claim on the taxpayer?

The reason is that a pension is in essence an annuity—an agreement to pay an individual an income for the rest of his or her life. And annuities are bond-like: the income they generate falls as bond yields fall. As more of their members approach or pass retirement, pension schemes tend to buy more bonds in order to hedge their liabilities. Some public schemes do the same. The Bank of England’s scheme is mainly invested in inflation-linked government bonds. America’s Social Security trust fund is entirely invested in Treasury bonds.

Governments issuing debt to pay their own IOUs looks to some people like a Ponzi scheme. But all pension schemes have a pyramid-like element in that they represent a claim on the incomes of future workers. A fund invested entirely in equities depends on future workers to generate the profits needed to make those equities valuable. If, as seems likely, the size of the workforce in European countries is set to decline over the next 20 years, that is a problem whether pension benefits are paid via the tax system or via a separately funded scheme.

The best way to absorb the cost is for workers to retire later. Since 1971 life expectancy has risen across the (mainly rich-country) OECD by four to five years, yet only now are retirement ages being adjusted upwards. Such changes will barely keep pace with projected longevity gains. It would be far better for citizens to work longer and get a decent pension for a shorter time than to see their benefits eroded by cash-strapped governments.