PRTVATE, decentralized pension plans are sometimes criticized as more expensive to administer than centralized public plans. Such criticism may, however, be based on inaccurate measurements of both the costs and efficiency of public plans.
The efficacy of a pension scheme depends in part on its administrative costs. Other things held constant, high costs mean less income in retirement or higher contributions during working years. This is true for any pension scheme, whether it is voluntary or mandatory, managed publicly or privately, and financed on a funded or a pay-as-you-go basis. But when governments force people to save for their retirement, the cost of administering those accounts becomes a public policy concern. The pension reform debate now taking place in many countries has focused attention on this issue.
Until recently, all publicly mandated retirement plans were managed by the central government or parastatal agencies. Since the early 1980s, however, reforms in at least seven countries have allowed private management of mandatory retirement savings. Critics of these private pension arrangements cite high administrative costs as a major disadvantage. In particular, they point to the high costs of the Chilean Administradoras de Fondos de Pensiones (AFPs) [Pension Fund Administrators], comparing the AFPs with the US Social Security Administration, which appears to have administrative costs that are only a fraction of those of Chile’s AFPs ($14 versus $49 per member, in 1989 dollars).
Are private, decentralized plans really more expensive to operate than centralized public plans? This runs counter to the common belief that competition brings efficiency. Which factors external to the pension system determine costs? For example, are costs higher in developing or industrial countries? Which system features will minimize costs? How can the relative efficiency of alternative systems be evaluated, when both external and internal factors influence operating costs, and when benefits also vary?
Administrative cost comparisons among different pension schemes are fraught with difficulties, owing to the enormous differences in country conditions, the kinds and quality of services provided, and cost-accounting techniques. Upon careful examination of the evidence, it appears that external factors, such as the country’s per capita income and the number of workers covered by pension plans, are major determinants of administrative costs. Decentralization may add to administrative expenses, but the resulting benefits, in terms of improved quality and higher investment returns, may outweigh the costs.
Which costs should be counted?
Administrative cost measurements should include all costs, including those borne by various government agencies, employers, and individuals. In practice, the true administrative costs of pension plans are practically never measured. Public pension plans, in particular, systematically understate their costs. For example, in the United States, the Internal Revenue Service collects and audits the Social Security payroll tax and investigates recalcitrant employers, so these administrative expenses do not appear on the books of the Social Security Administration, although they add to expenses incurred by the Internal Revenue Service. In many countries, the public pension agency may not pay rent for its premises, is not charged for depreciation or fringe benefits, and gets preferential rates on mail and telephone services. Private pension plans, which are generally funded, must charge consumers a risk premium for the longevity, inflation, and interest rate risks such plans assume, while public pension plans, which are usually run on a pay-as-you- go basis, do not charge such a risk premium as an explicit administrative cost but, instead, raise taxes later on to cover benefit expenditures as they grow.
a US citizen, is Lead Economist in the Poverty and Human Resources Division of the World Bank’s Policy Research Department.
a US citizen, is a Consultant to the Country Operations Division of Country Department II of the World Bank’s Europe and Central Asia Regional Office.
Pecuniary and nonpecuniary costs to employers and workers are also important, yet are regularly ignored. In countries with malfunctioning public bureaucracies, new pensioners spend many hours tracking their applications through a labyrinth of government offices. Delayed payments imply interest and inconvenience costs to pensioners. In some cases, workers bribe bureaucrats to expedite the claims process—another unreported cost. In competitive systems, where workers have a choice among service providers, they can more easily escape these bureaucratic costs—but competitive systems require regulation, which is costly. Taking all these factors into account, on balance, it appears that publicly managed old- age programs tend to understate their true administrative costs and overstate their efficiency relative to privately managed plans.
Which factors influence costs?
In explaining cost differences between countries, it is important to distinguish among external influences, pension system features, and simple waste that affects administrative costs. Two countries may have very different costs, regardless of their systems or efficiency, because of fixed country conditions. It is also important to understand which system features add administrative expenses and how much of a system’s operating cost simply results from waste that could be eliminated by better management.
External influences on costs. A pension scheme operating in a poor country faces very different resource constraints from one operating in a rich country. On the one hand, administrative costs may be higher in the rich country, because wages are higher. On the other hand, weak communications infrastructure and banking systems in the poor country may raise the cost of collecting contributions or paying benefits. A scarcity of computers and the skilled personnel needed to use them could raise the cost of record keeping. One way to take these factors, which are external to the pension system, into account is to control for per capita income when making cross-country comparisons. (Per capita income also captures the effects of some internal factors that raise costs. See the subsequent discussion of pension quality.) Multiple- regression analyses done for a sample of 50 pension schemes show that administrative costs rise with per capita income, but not as fast: for every 100 percent increase in per capita income, administrative costs rise only 43 percent.
Since administrative costs are incurred by collecting contributions from workers, keeping records, and paying benefits to pensioners, total costs are higher in pension systems with many members. If economies of scale are important, however, costs will not rise proportionately with the number of members, so a large country will have lower administrative costs per participant. The statistical analysis mentioned previously shows that, for every 100 percent increase in members, total administrative costs rise less than 80 percent and average costs fall 10–20 percent.
It may appear that average costs can be reduced by increasing coverage rates. This is only the case if countries have the infrastructure and enforcement capacity to reach workers in rural areas and in small firms, as well as to keep the informal sector small. This capacity, in turn, is strongly related to the level of income. Given the presence of multiple public schemes in many countries, public policies that unify the management of existing schemes would be more effective in achieving economies of scale. In sum, system size (number of contributing workers plus pensioners) and per capita income explain much of the variation in administrative costs and set the limits within which the system’s features and managerial skills can influence costs.
Systemic influences on costs. Features of the pension system, such as the scope and quality of the services provided, the amounts and kinds of investments made, and how it is marketed may also influence costs. Benefits as well as costs must be considered. For example, most developing countries pay only a lump sum upon retirement or provide an annuity that is not indexed for inflation, while industrial countries typically provide indexed annuities that have greater insurance value but are more expensive to administer.
Complaints about the low quality of services—such as inaccurate records, late payments, and long processing times—are common in many developing countries’ public pension schemes. Costa Ricans requesting old-age pensions from the Caja Costaricense de Seguridad Social can expect to wait 3–5 months for their claims to be processed, as can Argentines or Chileans who still belong to their countries’ old state systems. In contrast, the average turnaround time in Chile’s new privatized system is 17 days. If the operating cost of a system is higher, but the quality and scope of service are higher, too, this does not indicate less efficiency; in fact, it may indicate greater efficiency, since workers are often willing to pay more to get better service. Superior service undoubtedly accounts for part of the higher administrative costs in high-income countries.
An important function provided by some pension systems is the investment function. Most mandatory pension plans are run on a pay-as-you-go basis without substantial reserves, but some accumulate surpluses. In publicly managed plans, any surplus is usually required to be invested in government bonds or in the bonds of state enterprises—that is, on the basis of political, rather than economic, objectives—in which case their investment expenses are negligible but their investment returns are low or even negative. In contrast, privately managed plans are usually allowed to invest in a broad range of private, as well as public, securities. They incur higher expenses, but they also allocate capital to more productive uses and earn a higher rate of return, therefore allowing a lower contribution rate to finance a given pension. In the United States, many workers voluntarily pay administrative fees to mutual funds to invest in private sector securities, so they clearly value this function. When one adds in the annual fees that the average US investor pays for an amount of private investment equivalent to what the average Chilean has, this largely eliminates the Chilean cost disadvantage.
Finally, pension systems that are organized along private competitive lines incur marketing costs and transaction costs when workers switch from one pension fund to another. Approximately 30 percent of the administrative costs of Chilean AFPs are attributed to advertising and sales expenses, which may provide useful information to consumers but are primarily a form of economic warfare. Administrative costs in competitive systems may also be higher because they do not enjoy the same economies of scale that accrue to centralized systems. Balanced against this is the pressure for greater efficiency that may be exerted by competitive market forces.
Administrative efficiency. After taking account of external factors, such as per capita income and system size, some public pension schemes cost more to administer than would be predicted based on their characteristics, while others cost less. Statistical analysis shows that national schemes in Austria, Chile, Finland, and Kuwait fall in the former category, while schemes in Canada, Denmark, and Mauritius that feature universal flat benefits fall into the latter category. Countries that spend significantly more than would be expected may be offering additional or better- quality services in a cost-effective way, or they may be administratively inefficient.
For example, the decentralized character of the Chilean pension system leads to higher marketing expenses and failure to exploit scale economies. But it also leads to more and better service, which people value (discussed previously). The alternative cost of this service and the higher returns earned by Chile’s pension system (compared with those earned by a large number of publicly managed schemes in other countries during the 1980s) far exceeded its higher marketing costs. If Chile were to switch to a publicly managed system, its administrative costs might fall, but its investment performance would probably suffer, so the contribution rate required to fund the same pension level would probably increase substantially. This is why such a move would be unpopular in Chile and was one of the considerations that recently led Singapore to move toward private, decentralized management of its public pension fund, despite the higher administrative costs this might entail.
Developing and industrial countries compared Administrative costs per participant of publicly managed pension plans
Source: World Bank, Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth, Oxford University Press, New York, 1994.
Note: Data are for years ranging from 1988 to 1992.
Alternative cost ratios
The preceding discussion shows that there are multiple external and internal influences on costs and that all of these must be considered to determine whether a country is administratively efficient. In practice, such influences have generally not been systematically taken into account. Instead, simplified cost ratios, which are often misleading, are used as indicators of administrative efficiency.
Two ratios that are often presented are administrative expenses as a percentage of total contributions taken in or of total benefits paid out. In general, these ratios will be high in countries with young populations and immature systems (which are usually poor countries), whether or not their systems are run efficiently. In such countries, there are few pensioners, so both benefits paid out and contributions taken in to cover these expenditures are small, but the system must still incur all the costs associated with tax collections and record-keeping. Indonesia and Kenya spend 30 and 72 percent of contributions, respectively, on administrative costs, and this is more than they pay out in benefits. In contrast, Japan and the United States spend 1 percent or less of benefits and contributions on operating expenses. But, actually, statistical analysis indicates that both sets of countries are spending approximately what would be expected, given their per capita incomes and numbers of covered workers and pensioners. These ratios raise questions about the wisdom of starting old-age security programs in small, poor countries, but they do not tell us anything about the internal efficiency of these plans.
A ratio that avoids some of these problems is the administrative cost per plan member—that is, per worker and pensioner covered by the plan. This ratio avoids the bias against young countries with immature systems that is inherent in other measures. However, while this ratio is better than the others, because it adjusts for the number of system participants, it, too, is not an adequate measure of administrative efficiency, because it does not control for the quality of services provided, the price of labor and capital, and economies of scale. Small, high-wage countries such as Luxembourg and the Netherlands may have high administrative costs per member even if they operate their plans efficiently.
Finally, it is useful for some purposes to divide the administrative cost per member by income per capita, which gives us a ratio that is independent of national currency. This ratio shifts the cost per member downward more for high-income countries than for low-income countries, which can be viewed as a crude adjustment for the higher input prices that the former countries face and the higher-quality services they provide. The problem here is that it overcorrects, since costs rise at a much lower rate than per capita income, owing to higher productivity. Thus, it makes schemes in high-income countries look more efficient than they really are. Nevertheless, it is a good measure of the “administrative burden” that the system imposes in terms of local incomes. The accompanying chart shows that this burden is much greater in small, poor countries than in industrial countries. In Burundi and Tanzania (not included in the chart), the average cost per member is 7–8 percent of per capita income each year, while in Japan and the United States, it is less than 0.1 percent of per capita income.
Per capita income and pension system size exert a strong effect on administrative costs. Total costs rise as income and size rise, but not proportionately. Several conclusions follow from this empirical observation. First, the right way to measure administrative efficiency across countries is to compare actual costs with expected costs, based on the countries’ characteristics. Second, simple ratios of costs to benefits paid out or to contributions taken in are misleading as efficiency measures. The best simplified index of adminstrative efficiency is the average cost per plan participant, but this should be used as an efficiency measure only when comparing countries with roughly equal system sizes and per captia incomes.
Third, the higher administrative cost of the decentralized Chilean system relative to centralized systems appears to be overstated and is outweighed by the former’s high quality of service and strong investment performance. If higher administrative costs are part of a system that yields higher benefits and returns, they may lead to lower costs overall. Although decentralization may initially involve a loss of scale economies, the development of regional pension markets may eventually enable small countries to benefit more from scale economies under private systems than under public systems. As capital markets become globalized, large pension and insurance companies will increasingly operate across national borders (as is beginning to occur in parts of Latin America and Asia)—provided they are permitted to do so by governments and by harmonized regulations.
Fourth, the ratio of administrative costs per member to local incomes is much higher in small, poor countries than in large, rich ones, regardless of the type of pension system adopted. Even when the former countries are operating their systems as efficiently as possible, workers may find their contributions going mainly to support administration rather than pensions. Such countries should think twice before establishing or expanding formal old-age-security systems, especially if their informal systems are operating reasonably well. If they do establish formal systems, these should be kept simple to minimize collection, record-keeping, and payout costs.
Finally, all countries can institute practices to lower their administrative costs. For competitive systems, governments might dispense consumer information in order to minimize marketing costs, allow pension companies to keep transfer costs down by charging lower fees to long-tenured members, and permit these firms to form centralized clearinghouses for collection and record keeping on a voluntary basis. (None of these practices are currently used in Chile.) For publicly managed systems, collection of payroll and general income taxes should be integrated and computerized, and duplicate facilites avoided. Multiple industry or occupational plans run by separate public sector agencies should be avoided, or unified if they already exist. All countries, but especially those whose pension schemes cost significantly more than expected, should consider these techniques.
Further reading: S. Valdés-Prieto, “Administrative Charges in Chile, Malaysia, Zambia, and the United States,” World Bank Policy Research Working Paper No. 1372, 1994; O. Mitchell, A. Siinden, and P L. Hsin, “Social Security Costs in Latin America, the Caribbean and OECD Nations,” Journal of International Compensation and Benefits, Vol. 2, January–February 1994; G. Reid and O. Mitchell, “Social Security Administration in Latin America and the Caribbean,” World Bank Technical Department, Latin America and the Caribbean Regional Office, forthcoming in 1995; and E. James and R. Palacios, “Administrative Costs and Efficiency in Public and Private Pension Plans: An International Perspective,” World Bank Policy Research Working Paper, forthcoming in 1995.
The International Monetary Fund presents
Balance of Payments Compilation Guide NEW
The Balance of Payments Compilation Guide is a companion document to the Balance of Payments Manual, the fifth edition of which was published by the IMF in 1993. The primary purpose of the Guide is to provide practical direction for using sources and methods to compile statistics on the balance of payments and the international investment position. The Guide reflects the emergence of new data sources and adaptations in the application of statistical methodologies to changing circumstances.
Available in English. 380 pp. (paper) 1995. ISBN 1-55775-470-5 US$32.50. Stock # BOPCEA.
Balance of Payments Manual, 5th Edition
Available in English, French, or Spanish, (paper) ISBN 1-55775-339-3 US$27.50. Stock # BPM
To order, please write or call:
International Monetary Fund
700 19th Street, N.W.
Washington, D.C. 20431 U.S.A.
Telephone (202) 623-7430
Telefax: (202) 623-7201
Cable Address: INTERFUND
American Express, MasterCard, and
VISA credit cards accepted.
Estelle James and Robert Palacios