Within each mandatory pillar are numerous policy options—some of which are much better than others. For the public pillar, one important choice is between earnings-related benefits, on one hand, and those that are flat or means-tested on the other. Although earnings-related benefits may have political appeal, the need to pay higher benefits to high-income workers while also trying to alleviate poverty among low-income workers creates numerous problems, including high taxes and excessively large transfers to the first generation of beneficiaries, many of whom are well-to-do. For these and other reasons, earnings-related benefits are a poor choice for the public pillar.

The next choice, between flat and means-tested benefits, is less clear-cut and will depend on such variables as institutional and taxing capacity (flat benefits are costlier but easier to administer) and income distribution (means-tested benefits redistribute to the poor more efficiently, provided that the administrative requirements can be met at low cost). A variant of the means-tested scheme, a minimum pension guarantee, may be the best option for the public pillar in countries that already have a mandatory saving scheme to which the guarantee can be added.

For the mandatory funded pillar, governments must choose between public management and private management. Most publicly managed funded schemes, also known as provident funds, have had poor results. Publicly managed funds are usually required to invest in government securities or the securities of quasi-government entities such as state enterprises or public housing authorities--frequently at below-market interest rates. These funds thus earn less than they would on the open market and must charge higher contribution rates or pay lower benefits than would otherwise be the case. In contrast, private, competitively managed schemes—in which workers or employers choose their fund managers—are rarely required to accept below-market returns and are less likely to be used as disguised forms of government revenue. They have incentives to invest in stocks and bonds that offer the best risk-yield combinations, and can tap the benefits of international diversification and management expertise. The resulting superiority of privately managed funds over publicly managed funds is strikingly apparent from a comparison of rates of return of selected funds in the 1980s.

Higher returns to contributors aside, mandatory, privately managed funded schemes offer economy-wide advantages. They can be part of a national policy to develop new financial institutions and deepen capital markets by mobilizing long-term saving and allocating it to the most productive uses, including uses in the private sector. For these reasons, the report strongly recommends that the funded pillar be privately managed.

For privately managed funded schemes, still another choice must be made between personal saving plans and occupational pension plans set up and run by employers. Occupational plans have the seeming advantage that they can start up on a voluntary basis before the market and the government are ready for a mandatory plan. They can be implemented through payroll deductions with low record-keeping and marketing costs. And they use the financial expertise of employers and prearranged groups of workers to overcome insurance market problems. However, these advantages often prove illusory. Without mandatory participation and adequate regulation, occupational plans tend to be spotty in coverage, to be offered mainly to middle- and upper-income workers, to involve large regressive tax expenditures, to he underfunded and therefore default-prone, and to restrict vesting and portability of benefits, impeding labor mobility and economic restructuring. In contrast, coverage under personal saving plans can be broad, and benefits are fully portable. Because of these distributional and labor market effects, personal saving plans are probably preferable to occupational plans for the funded pillar, except for countries that already have substantial coverage under well-functioning employer-sponsored schemes. A privately managed mandatory personal saving scheme was pioneered in Chile and is now being incorporated into new systems in Argentina, Colombia, and Peru.

Despite the advantages of privately managed, mandatory personal savings schemes, governments should not rush to establish one. Rather, they need to assess carefully marker and regulatory capacities before deciding to go ahead. A banking system, rudimentary stock and bond markets, and the capacity to develop these further in response to demand from pension funds are essential preconditions. In addition, because workers may lack the education and experience necessary to choose effective investments, careful government regulation is needed to keep the investment companies financially sound and workers' exposure to investment risk within reasonable bounds. Where market and regulatory capacities are lacking, governments should proceed slowly and cautiously.

How should countries start this process: and how can those that already have large public pillars make the transition? Although the ultimate goals are similar for all, the paths and time frame depend on country circumstances (chapter 8).

Consider first a country with a young population, a low per capita income, and only a small public pillar or publicly managed provident fund, primarily one covering government sector employees. Many countries in Africa and South Asia are at this stage. The weakening of informal systems of old age support and the absence of reliable capital and insurance market instruments are prompting political pressures for an expanded public pillar. These countries typically do not have the financial markets or regulatory capability necessary to establish a decentralized funded pillar. But they should be creating an enabling environment for voluntary and, later, mandatory saving and pension plans by:

• Keeping inflation down

• Avoiding interest rate and exchange controls

• Establishing reliable savings institutions that are accessible to people in rural as well as urban areas

• Developing a regulatory framework that gives people confidence in banks, insurance companies, and other financial institutions

• Instituting an effective tax policy and tax administration system

• Building the human capital essential for the effective management of financial and regulatory systems.

These basic conditions, important for old age systems, are also necessary for continuing economic growth.

These countries should also be taking steps especially geared to providing old age security, using methods that avoid the problems of large pay-as-you-go plans and that will eventually fit into a multipillar system. 'I'hey should:

• Keep the existing contributory public pillar small, flat, and limited to urban areas and large enterprises in which transaction costs are relatively small, fraud is easiest to detect, and the informal system breaks down first.

• Provide social assistance (in cash or in kind) to the poorest groups in society, including the old poor who are not covered by contributory plans, taking into account their vulnerability stemming from their diminished ability to work.

• Carry out simulations of the long-run impact of alternative public plans (coverage, benefit level, retirement age) on taxes and the distribution of transfers across and within generations. This requires making assumptions about wage growth, interest rates, labor force participation, unemployment, and evasion—and recognizing that the choice of system will affect these parameters.

• Phase out (or convert to voluntary status) centrally managed provident funds, which are often misused.

• Set up the legal and institutional framework for personal saving and occupational pension plans, requiring full funding, portability of benefits, and disclosure of information for the latter. Give equivalent tax treatment to occupational and personal retirement plans that meet prudent standards.

• Avoid crowding out informal support systems and offer incentives to families to continue taking care of their older relatives.

• Avoid the pitfalls overgenerous pensions, early retirement, benefit-contribution structures that encourage evasion or discourage saving, perverse redistributions to high-income groups in public plans and unregulated, unfunded, nonportable occupational plans--that are so tempting, especially in young countries with immature schemes and limited regulatory capability.

The next set of countries, also with young populations, is aging and often growing rapidly—since rapid economic growth is associated with falling fertility rates and rising longevity. Many East Asian economies are at this stage. In addition ro accelerating all the actions just mentioned, these economies should:

• Begin designing and introducing a mandatory decentralized funded pillar. Preconditions for this pillar are government regulatory capability; a banking system, a secondary government bond market, and an emerging stock market----or the ability to develop these institutions quickly in response to demand from new pension funds.

• Start by setting up a strong regulatory framework, determining the required contribution rare, and deciding whether saving or occupational plans should be used for the mandatory funded plan. Establishing this structure and phasing in the second pillar could take several years. Governments should not rush ahead too fast, beyond their institutional capabilities. But if they do not move ahead fast enough, strong political pressures will otherwise develop, from middle and high-income workers, for a dominant earnings-related public pillar—and all its associated problems.

• Gradually expand coverage for the public pillar, keeping it modest and redistributive while satisfying workers' saving or income smoothing needs through the privately managed funded pillar. Otherwise these economies will face the much more difficult task of restructuring later.

• Initially use a payroll tax for the public pillar, to avoid inefficiencies from excise taxes and transfers from uncovered to covered groups, but shift to a broader tax base as coverage becomes universal, the government's ability to collect general income and consumption taxes increases, and the redistributive function can be emphasized.

The third set of economies comprises those that are already middle-aged, are growing older rapidly, and have substantial public pension programs that provide widespread coverage and whose costs will soar, with dependency rates, over the next three decades. This set includes OECD and Eastern European economies and several Latin American economies. Although the degree of urgency varies, all these economies face imminent problems with their old age systems. Rather than relying on an ever more costly public pillar to do it all, at high tax rates that inhibit growth and bring low rates of return to workers, the time is ripe for these economies to make the transition to a mandatory multipillar system.

• The first step is to reform the public pillar by raising the retirement age, eliminating rewards for early retirement and penalties for late retirement, downsizing benefit levels (in the frequent cases in which they are overgenerous to begin with), and making the benefit structure flatter (to emphasize the poverty reduction Function), the tax rate lower, and the tax base broader.

• The second step is to launch the second pillar by setting up the appropriate contribution and regulatory structures. The transition can be accomplished by:

(1) Downsizing the public pillar gradually while reallocating contributions to a second mandatory pillar or

(2) Holding the public benefit relatively constant (in cases in which it is low to begin with) but raising contribution rates and assigning them to the second pillar or

(3) Recognizing accrued entitlements under the old system and agreeing to pay them off while starting a completely new system right away. This involves designing the new system, calculating the implicit social security debt that is owed under the old system, and figuring out how to finance it all in a way that is both politically and economically acceptable.

Several OECD countries are engaged in the gradual transition (alternative 1 or 2). Several Latin American countries have already introduced a radical transition (alternative 3). And many former socialist countries are trying to decide which way to go.

A mandatory multipillar arrangement for old age security helps countries to:

• Make clear decisions about which groups should gain and which should lose through transfers in the public mandatory pillar, both within and across generations. This should reduce perverse or capricious redistribution and poverty.

• Achieve a close relationship between incremental contributions and benefits in the private mandatory pillar. This should reduce effective tax rates, evasion, and labor marker distortions.

• Increase long-term saving, capital market deepening, and growth through the use of full funding and decentralized control in the second pillar.

• Diversify risk to the Fullest because of the mix of public and private management, political and market determination of benefits, the use of wage growth and capital income as the basis for finance, and the ability to invest in a wide variety of securities--public and private, equity and debt, domestic and foreign.

• Insulate the system from political pressures for design features that are inefficient as well as inequitable.

The broader economy should be better off in the long run as a result. So should both the old and the young.

The right mix of pillars is not the same at all times and places. It depends on a country's objectives, history, and current circumstances, particularly its emphasis on redistribution versus saving, its financial markets, and its taxing and regulatory capability. The kind of reform needed and the pace at which a multipillar system should be introduced will also vary from quick in middle and high-income countries whose systems are in serious trouble to very slow in low-income countries, which should avoid these same mistakes. But one simple recommendation is clear: all countries should begin planning now.