Ownership illusions: Retirement income policy
Part 3: "Pre-funded" retirement systems only appear that way because there are no priced ownership rights for future "pay-as-you-go" pensions.
This is Part 3 of a four-part series. Please consider also reading Part 1 and Part 2.
In case you missed it, I have a new working paper with co-author Tim Helm, entitled Ownership Illusions: When ownership really matters for economic analysis. In the paper, we look at four situations where failure to recognise the structure of ownership leads economic analysis astray.
Today, I want to expand on our third example.
Part 3: Pre-funded retirement systems
Another ownership illusion arises in the analysis of retirement income, or pension, systems. Increasingly, these systems rely on ownership of financial assets to fund the incomes of retirees, by way of compulsory savings used to purchase assets like listed company shares, corporate and government bonds, and cash. These can be in individual accounts, like Australia’s superannuation system and Singapore’s Central Provident Fund (CPF) accounts, or in jointly owned social security funds.
In some countries, the value of assets in “pre-funded” retirement systems is significant. The Netherlands, for example, has retirement funds valued at more than 200% of GDP, while Canada, Australia, Switzerland and the United Kingdom each have pre-funded pension systems holding assets valued at over 100% of GDP (OECD, 2020).
The alternative system is called “pay as you go”, or sometimes an “unfunded” system. Here a country’s Treasury pays pensions from its account. This reduces the government budget balance, instead of reducing the balance in individual accounts, as in a pre-funded system.
However, the notion that a compulsory saving system “pre-funds” retiree spending whereas a pay-as-you-go system does not is another ownership illusion. It is based on the idea that society’s capacity to pay retirement incomes depends on capitalisation, rather than real economic flows, which is wrong. The illusion treats capitalised asset wealth as a funding source but the state’s uncapitalized power to raise revenue otherwise – and in so doing inserts a normative assumption about who should pay into the ostensibly-neutral economic concepts of “funded” and “unfunded” systems.
To illustrate this we consider the value of ownership rights in each system, and what it means to fund retirement income payments.
What does the value of a financial asset in an individual pension account represent? Simply what a buyer is willing to pay for the future stream of income from the asset.
Just as the value of a house comes from the future occupancy it provides, the value of a financial asset comes from the goods and services its income stream can buy. The occupancy, and the goods and services, are real. The lump-sum valuations of the housing rent or income stream are not. They are the agreement of a seller and a buyer as to the exchange value these rights to occupancy (or income) are worth.
An economically consistent approach to comparing retirement systems must compare either the capitalised values of each system or the income streams available to fund ongoing retirement payments.
But what is the capitalised value of the future taxes, bonds and seigniorage that funds retirement incomes in a pay-as-you-go pension system? It is not measured, because there are no markets valuing it, and no routine practices of non-market valuation.
The capitalised value of public revenue is as (un)real an economic phenomenon as the capitalised value of asset income, yet our rituals of capitalisation do not let us see it.
In principle, valuation is possible, however. The right to a pension payment could be converted to an asset by creating an ownership right. We could call these financial instruments pension bonds, give one to each eligible pensioner, and allow them to be traded. The total market capitalisation would represent the asset wealth of pension bondholders available to fund future income (retirement or otherwise). Equivalently, it would represent the capitalised value of the public revenue stream that in a pay-as-you-go system funds these payments. There would be no change in economic flows, but a capitalised ownership value would exist where it did not before.
Australia’s pay-as-you-go age pension, which exists alongside compulsory retirement savings, distributes around AUD$55 billion per year. At the 3-4% yields applying to other government funding (such as Treasury bonds), these pension bonds would have a capitalised value of $1.2-1.6 trillion. By comparison, the market value of assets in compulsory retirement savings accounts is AUD$3 trillion. Without the ownership illusion, Australia’s pay-as-you-go system seems very well funded.
One might raise two objections to the argument that pay-as-you-go pensions are equally meaningfully described as “funded”.
First, do asset values in retirement accounts not indicate real production, which provides capacity to pay income, but capitalised taxation only the power to take this, which does not?
No – the distinction here is semantic, not real. Asset ownership is only ever a legal claim on real product. Tax powers are also a legal claim on real product. Property is a bundle of rights, but so is taxation, and to treat the value of those differently is an ownership illusion.
All that matters for capacity to pay is real production. The value of assets in specially labelled accounts does not measure that. Whether compulsory saving increases real production is the relevant question, and it is ultimately an empirical and context-specific one (discussed further below).
Second, does it not matter who funds and receives retirement incomes? Compulsory saving means each individual funds their own income. Whatever the fairness of this, to describe different systems as funded or unfunded on this basis not only stretches language, but smuggles a premise of distributional justice into concepts of economic analysis purporting to be ethically neutral.
Another aspect of the ownership illusion in retirement income systems is that the value of financial assets in “pre-funded” systems often merely represents an ownership rearrangement – not additional productive capacity.
When a “pre-funded” system “saves” by buying assets, it usually buys those from a prior owner, rather than investing in new buildings and other real capital assets. Ownership simply moves from outside the retirement system to inside it.
Figure 4 illustrates this point. The left panel shows direct ownership by individuals of an asset class such as listed company shares, bonds, or property. To the extent compulsory saving does not increase savings, or change its composition, the effect is identical to each owner selling some of their assets to their retirement fund. The change is simply one of ownership structure, by inserting a retirement fund intermediary (right panel).
This is why the value of retirement funds cannot be measured against the value of the share market or property market without substantial double-counting. About 37% of the publicly traded share market in Australia is owned by retirement funds, for instance (Myer, 2021).
A “pre-funded” system boosts the capacity to fund retirement incomes only to the extent that compulsory saving increases the stock of real capital assets, via more spending on new capital equipment or other productive capacities. But there are theoretical reasons to doubt this will occur: if capital formation is demand-driven, then reducing circulation of spending in the real economy decreases capital investment (a “paradox of thrift”).
In sum, it is not clear either that compulsory savings increase society’s capacity to fund retirement incomes, or that the concepts and language of “funded” and “unfunded” systems have any real economic grounding. Both ideas are based on ownership illusions: in the simple assumption that retirement accounts represent funding capacity, and the subtler assumption that capitalised asset values have some economic meaning different in essence to the uncapitalised right of the sovereign to raise tax.
The normative premises embodied in these concepts and language can be seen as part of the power struggles over the ownership, allocation, and control of economic assets. Ownership illusions, in this light, contribute to concealing such power struggles behind the façade of objective, ethically-neutral analysis.
Dear Cameron
Taxation and the common law are strangers. Taxation is not part of property law. It is the raising of an unsecured debt by the Crown against a contributor who is deemed to have given his consent through his representatives in the Commons. Pitt the Elder's speech on the American Revolution explains this. As for capitalising pension rights funded by taxes, this is unworkable since people can decide not to have babies to pay for other people's pensions or move their capital or labour out of the jurisdiction. What does matter is the deadweight loss of taxes on labour and capital. There is no deadweight loss if adult children voluntarily support their aged parents or relatives. That was the "pay as you go" plan of Nature before Bismarck. It would be sensible to allow unlimited income transfers and means test people out of the aged pension accordingly. The tax system fails to recognise voluntary income redistribution and thereby substitutes inefficient public redistribution far more than is really needed. So people having babies are producing taxpayers for others and meanwhile being taxed to keep pensioners in asset test free homes they themselves can't afford to buy to house their growing families. There is actually perverse redistribution in the current pay as you go tax-funded age pension.
Kind regards
Terry Dwyer
If you're updating the working paper, the URL on the Myer (2021) reference is incorrect.