"Universality and the cost of basic pensions", by Larry Willmore, 30 January
2012.

Notes for presentation at a meeting convened by HelpAge International on 23-24
February in the UK.

A typical reaction to universal pensions is “That’s a great idea, but we can’t
afford it, so prefer to give pensions only to those who need them”. This reaction
appears to be common sense, but it is wrong. Means tests shift costs, but do not
lower them because means tests are taxes on income or assets. They are paid by
elderly citizens and, sometimes, their families. It is more efficient to fund
basic pensions with general taxes paid by everyone. All taxes, with the exception
of head taxes, distort choices. Of particular concern are taxes that discourage
work and saving. Low taxes levied on an entire population are less distorting
than high taxes levied on the elderly.

This is the efficiency argument for universality. It is valid even with costless
and perfect targeting, with no stigma, no exclusion errors, no erosion of
political support. It is a simple argument, yet often ignored because means tests
are usually recorded as expenditure reductions rather than as tax collections.
Framing is important. A message that means tests are taxes could appeal to voters
across the political spectrum.

It is important to recognize that, even though universality is optimal, all means
tests are not equally bad. Clear, simple rules are preferable to complex
regulations that leave discretion to government bureaucrats. Rules matter more
than whether the number disqualified by targeting is small (an ‘affluence test’)
or large (a ‘means test’?). The term ‘affluence test’ is imprecise and adds
nothing to our understanding of means tests.

To show that means tests are taxes, even when disguised, we examine some typical
tests - first simple rules, then complex tests. Our base scheme is one in which
everyone who reaches a specified age has a right to a pension, regardless of
income, wealth or employment history. It is an age pension, not a retirement
pension, because a beneficiary need not retire from work, or have ever worked for
pay. This is the universal pension that the World Bank praised in 1994 (p. 240):

	“Administratively, this is the simplest structure, with the lowest
	transaction costs, for the public pillar—an important advantage in
	developing countries with limited institutional capacities and incomplete
	record-keeping systems. It avoids the disincentive to work and save
	inherent in means-tested plans. Its universal coverage helps ensure that
	the poverty reduction objectives are met, provides a basic income for all
	old people (coinsuring against low investment returns or high longevity),
	and might receive broad political support.”

The cost of this pension is the number of beneficiaries times the size of the
pension, plus administrative expenses. Governments sometimes include benefits in
taxable income, as New Zealand and Mauritius do, lowering the net fiscal cost.
This is a very mild means-test, one that ensures that pensioners and non-
pensioners with equal incomes pay the same tax. The test is simple, transparent
(not disguised), and it has a ‘fairness’ rationale. But it does penalise paid
work by the elderly if the taxable pension moves them to a higher tax bracket.

The next test to consider is the pension test. Anyone with other pension income
receives a smaller basic pension or none at all.  A pension-tested benefit is a
“universal minimum pension”.  Sweden in 1913 implemented the world’s first
universal minimum pension and simultaneously mandated contributions to an
earnings-related scheme. This was very forward-looking legislation, because
retirement pensions were almost unknown in Sweden at the time. For many years,
there were few earnings-related pensions so, for older retirees, the pension was
actually universal.

The pension test is a tax on other pension income, which discourages
contributions to earnings-related schemes. For this reason it is seldom applied
to benefits from voluntary pensions. When the test applies only to compulsory
schemes, it is feasible to “claw back” the pension from other pension income at
the rate of 100%, as Sweden has done since 1913. Governments can achieve the same
effect, as Lesotho does, by allowing pensioners to receive no more than one
government benefit. For some workers, a 100% rate of clawback taxes away all
benefits from further contributions to an earnings- related plan, which
transforms mandated contributions into a tax on earnings. As an employment
incentive, Norway, which had a pension test similar to that of Sweden, in 2010
lowered its clawback rate from 100% to 80%.

A broad income test is also possible, and is equivalent to a tax surcharge on
earnings and savings of the elderly. The test works best when integrated with the
income tax system, clawing back the basic pension from all other income. Canada
and Iceland are examples of this type of system. Canada claws back its basic
pension at the rate of 15%, Iceland at the rate of 25%. Both countries have very
modest basic pensions, and allow a large ‘exempt’ amount of income before the
surcharge begins. Any basic pension left after clawback is taxed as ordinary
income.  This test is a tax surcharge on retirement savings (more correctly, on
income from accrued assets and on withdrawals from tax-subsidised retirement
savings accounts) and on earnings of those who continue to work beyond the state
retirement age.

Finally, comprehensive tests of income and assets are common, especially in low-
income countries.  Such a system, Amartya Sen (1995) wrote, “rewards cheating and
penalizes honesty”. I would add that it also facilitates corruption. Even when
well administered, the tests discourage work beyond retirement age and saving for
retirement. The United States basic pension (known as SSI), for example, provides
maximum benefits of $698 a month for an eligible individual, or $1,048 a month
for an eligible couple. The rules are stringent. To be eligible, a person cannot
own assets (real estate, bank accounts, cash, stocks and bonds) worth more than
$2,000 ($3,000 for a couple).  Benefits are reduced 50 cents for each dollar of
wage income (after $65 a month), and dollar for dollar for other income (after
$20 a month). Elderly pensioners living in poverty are subjected to high taxes
(clawbacks) on incomes that attract no tax at all for non- pensioned workers.

In developing countries, means tests typically consider the income and assets of
entire households, even of adult children who live independently. Pensions
intended for households in poverty are often very small, so represent a small,
lump sum tax on those above the cut-off point for benefits. With the heroic
assumption of perfect targeting at no cost, this tax might be relatively
efficient (little effect on choice), but it is ineffective in providing income
security.

Means tests can help to finance basic pensions, but we should not ignore their
consequences. Almost always, general taxation is a better option.

Further reading:

N. Rowe & F. Woolley, “The benefits of universality", Policy Options (June 1999),
pp. 57-60. http://www.irpp.org/po/archive/jun99/rowe.pdf