Keep the record-keeping and other administrative demands simple, and concentrate formal pension systems in urban areas and big firms
The efficacy of a pension scheme depends greatly on its administrative costs. Other things constant, high costs mean lower benefits in retirement or higher contributions during working years. This is true for any pension scheme, whether managed publicly or privately, and whether funded or pay-as-you-go. But some managerial and financing arrangements may incur higher costs than others.[1]
It is not always easy to measure administrative costs. Privately managed companies have an incentive to keep track of costs. But publicly managed programs may omit some inputs and may receive others at below-market rates. The pension agency may not pay rent for its premises and may purchase its mail and telephone services at subsidized rates. Depreciation and fringe benefits often do not appear on the agency's budget. The costs of collecting the payroll tax may be passed on to another agency. Private pension plans charge a risk premium for the longevity, inflation, and interest rate risk they assume. Public pension plans do not charge this risk premium explicitly but instead pass it along to the rest of society, implicitly. For all these reasons, publicly managed old age programs may appear much cheaper than they are, relative to privately managed plans.
Most private schemes accumulate pension reserves that they invest, usually in a combination of public and private, debt and equity securities---a process that raises their administrative costs. Most public plans do not have large pension reserves, and even if they do, they do not carry out the same function of evaluating alternative investments, because they invest primarily in government bonds. Furthermore, to attract savings, the private sector incurs marketing costs. But balanced against these higher costs is the incentive of private pension funds to behave efficiently, and to perform a capital allocation function that would otherwise have to be performed elsewhere in the economy. For all these reasons, it is difficult to compare the costs of publicly and privately managed pension plans, and few such studies have been conducted.
It is somewhat easier, but still problematic, to compare the costs of different publicly managed pay-as- you-go plans. Two frequently cited measures are biased against developing countries with immature systems. The first is the ratio of administrative costs to benefit expenditures in a given year, a measure extremely sensitive to the maturity of the scheme and to demographic changes. In Belize, a young country with an immature pension scheme, the ratio of costs to benefit expenditures in the late 1980s was more than 50 percent, because few workers had been in the system long enough to qualify for pensions.
The second indicator used is the ratio of administrative costs to contributions. One obvious problem with this indicator is that contribution revenues will be higher, making the expense ratio appear lower, in countries with higher contribution rates. Because these countries typically have more mature schemes and older populations, this measure is also biased against pension schemes in developing countries.
A better way to compare administrative costs across publicly managed pay-as-you-go plans is to calculate the cost per participant (worker plus pensioner) in the program. The most important functions for these plans are collecting contributions and paying benefits, so the number of participants should determine efficient costs.
For some purposes it is useful to normalize the administrative cost by the average wage or income per capita---that is, to examine the ratio of administrative cost per participant to income per capita across countries. The average wage or income per capita indicates a country's capacity to pay for old age pensions and their administrative costs, once the dependency rate is given. The higher the administrative cost per participant, the less funds will be available for beneficiaries. If unskilled labor is the predominant administrative input into pension plans, one might expect this ratio (of administrative cost per participant to income per capita) to be positively related to per capita income, because poor countries have plentiful unskilled labor and low wages. But if skilled labor and capital inputs predominate, one would expect this ratio to be negatively related to per capita income.
Examination of the data shows the latter to be the case (see figure). Apparently, educated labor, computers, communications infrastructure, and other capital-intensive inputs are important elements in the pension production function, so the administrative cost per participant relative to per capita income is much higher in low-income countries, where these inputs are scarce and expensive. This number is 65 times higher in Zambia than in the United States. It is 84 times higher in Burundi than in Switzerland. Further analysis indicates that administrative costs fall as the number of participants increases, for both public and private plans, suggesting significant economies of scale. If this is the case, low-income countries, especially small ones, should think twice before they establish complex formal systems of old age security. Industrial countries started their present systems at much higher levels of per capita income, education, and infrastructure than many developing countries today. If low-income countries initiate formal old age systems, they should keep the record-keeping and other administrative demands simple and should concentrate the systems in urban areas and in large firms, where transaction costs will be low.
