The Structure of Reform

In the countries that are undergoing structural reforms, a multi-pillar structure is developing. This also happens to be the structure recommended by the World Bank, and we are glad that many countries are moving in that direction.

The new structure consists of two mandatory parts. One mandatory part--or pillar--handles people's retirement saving, the funded accounts that people are required to have. This pillar is a defined contribution instead of defined benefit, thereby linking benefits closely with contributions. Benefits are funded rather than pay-as-you-go, and the funds are privately managed. That is really the centerpiece of the structural reforms.

This pillar is always buttressed by another mandatory pillar, which is a publicly managed tax-financed arrangement designed to provide a social safety net to protect low wage earners.

That's the basic system we find in countries that are instituting structural reforms.

In addition, there is always a third, voluntary pillar for people who want more consumption and income in old age. But I am going to talk mainly about the mandatory part of the system. Despite the fact that this is the general model that reforming countries are following, they are following it in many different ways, and I will talk about the major variations that we see developing around the world.

Defined Contribution vs. Defined Benefit

Let me explain briefly the arguments for the characteristics of that middle pillar--that middle, mandatory saving pillar, the pillar that handles people's individual savings accounts.

It is defined contribution rather than defined benefit. The rationale is that this links benefits closely to contributions. Therefore, it should reduce the incentive for evasion. Evasion is a big problem in many countries. Evasion can run up to 50 percent of the labor force. But the smaller the tax element, and the closer the link between benefits and contributions, the smaller the incentive for evasion.

Perhaps more important, it discourages early retirement and makes the system less financially sensitive to early retirement decisions. Furthermore, it adjusts the retirement age upward, or the benefit rate downward, automatically as longevity increases.

This is really important, because a characteristic of defined-benefit programs in most countries is that people can retire early at rates that are not actuarially fair. So if you retire early, over your lifetime you collect a larger lifetime benefit than you do if you retire later. This is a bigger incentive for people to retire early. David Wise has done a study which shows that this incentive, in fact, is very closely linked to people's decisions to retire early. In some cases, this produces a very high implicit tax on continuing to work, and people respond to that high implicit tax. In some countries, the implicit tax is as high as 70 or 80 percent because people are foregoing a large benefit if they continue working.

This imposes a huge financial strain on the social security system. And it's bad for the economy because it reduces the experienced labor force.

When you retire, in a defined-contribution plan, you turn your capital accumulation into an annuity on actuarially fair terms. So if you retire early, you get a lower benefit. This has two advantages. It makes the system financially sustainable and, since you're not passing the cost on to others, you're discouraged from retiring early.

In addition, as longevity increases in a defined-benefit system, politicians face the very difficult task of raising the retirement age continually to keep up with the longevity increases. In a defined-contribution system, this process happens automatically without a difficult political decision. Either people voluntarily retire later, or they find themselves faced with lower annual benefits, which in turn leads them to postpone their retirement.

These are some of the reasons why we see this partial shift to defined contribution as desirable.

The Shift to Funding

We also see a shift to pre-funding, as compared with pay-as-you-go. This avoids unaffordable promises, which most developed countries have made by now. But the developing countries that we're dealing with haven't made them.

At the World Bank, we are urging them not to make those unaffordable promises. That avoids the large payroll tax increases that you get as populations age in a pay-as-you-go system. It avoids the intergenerational transfers that automatically take place--income transfers to the early cohorts that retire away from their children and grandchildren. And it helps to build long-term saving.

Many countries, including the United States, feel that they have a shortage of long-term saving. It is true that we have global capital markets and to some extent we can import capital, but it is also true that domestic investment seems closely tied to the rate of domestic saving, and this in turn is a source of economic growth. So funding can be an important source of long-term national investment and growth if it is part of a set of policies designed to increase public and private saving.

Higher Rates of Return Through Private Management

Why should the middle pillar of mandatory saving be privately and competitively managed? The important point here is that this maximizes the probability that investment decisions will be based on economic rather than political considerations. It therefore maximizes the real rate of return to society and the monetary rate of return to the pension fund.

At the World Bank, we have assembled data on rates of return to publicly and privately managed pension funds, and we show that publicly managed pension reserves around the world have fared poorly. In most countries, in fact, they have earned negative real rates of return.

The reason is that the investments have been politically dictated. Often they have to be invested exclusively in government bonds--we see that in the Social Security Trust Fund. But sometimes they are invested in special-issue government bonds that yield negative real rates of return. In addition, sometimes they have to be loaned to failing state enterprises that are going bankrupt. The pension funds are used to bail them out.

In any event, for a variety of reasons, if the funds are publicly managed, it is very difficult to avoid political objectives creeping in. And these can conflict with the economic objective of maximizing the economic rate of return. Decentralized competitive management is most likely to get you an efficient allocation of capital.

In addition, for the middle-income countries that the World Bank deals with, funded, privately managed pension plans can be a very important way of developing financial markets, and in fact this has been shown to have played a very important role. Econometric studies of Chile's rapid rate of economic growth during the past 15 years, for instance, indicate that the funded pension plans played a major role, in particular in developing financial markets. The development of financial markets is given credit for an increase in total factor productivity of about 1 percent per year.

The Latin American Model

These pension reforms are taking somewhat different forms in different parts of the world. In particular, I want to talk about three types of reforms.

The Latin American model more or less follows the model set by Chile in 1980. Chile was, as you know, a pioneer in social security reform. But that system has spread now throughout Latin America to Argentina, Mexico, Peru, Bolivia, and Uruguay. It is going to sweep South America and is moving to Central America.

Interestingly, it has just been adopted by Hungary and by Kazakhstan. I would expect that, in one form or another, in the next three to five years this kind of reform is going to be sweeping throughout Eastern and Central Europe and the former Soviet Union, always with variations dictated by country conditions.

The key features of the Latin American model are:

1. Workers choose the investment manager.
2. Workers have individual accounts and can move their individual accounts from one pension fund to another.
3. The pension funds are very much like mutual funds, but they operate subject to regulations that are set up by the countries.

Another key feature of the Latin American model is that there are transition costs. Pre-reform, a large contribution was being made to a pay-as-you-go system. When some of that contribution is diverted to the individual account, you still have the obligations remaining from the old pay-as-you-go system. Somehow, you have to cover those transition costs.

This is a big problem, and it has stymied a lot of countries. It really slowed down the reform process in Eastern and Central Europe. But, in fact, it turns out that the transition cost problem can be solved, and I will talk for a minute or two at the end about how it has been solved in these countries.

A major criticism of the Latin American model that has been raised--and I mention it because it is sure to surface in the United States as an important factor--is the issue of administrative cost. Chile has been criticized for having high administrative costs, stemming in large part from high marketing costs. I just want to say three things about that.

First, I think the administrative costs in Chile have been overstated, because there were heavy start-up costs. There always are when you're instituting a new system.

Second, in Chile all the fees are front-loaded. That is, you pay when you make your contribution, but you don't pay annual expenses based on your accumulated assets. Naturally, in the first few years of the plan, when contributions are high relative to assets, this looks like a big subtraction from your rate of return. But as time moves on and your assets grow relative to your annual contributions, it will be a much smaller subtraction from net return. These costs have been coming down over time as a percentage of assets. Simulations show that for a worker who contributes for 40 years, the fee structure in Chile will amount to less than 1 percent of assets annually, which is more or less comparable to what mutual funds in the U.S. charge. So this is in part a red herring.

Nevertheless, my third point is that even 1 percent is pretty high for a mandatory system. It is something we have to think about seriously, and we have to develop ways to bring it down.

The OECD Model

The second main model that we find developing is what I call the OECD model. Australia is a key example of this model. You also find it in Switzerland, Denmark, and the United Kingdom.

A key feature of the OECD model is that rather than having individual accounts with individual choice, you have group choice. That is, the employer and/or the union trustees choose the investment manager for the company or the occupational group as a whole.

This course was taken in these countries for historical reasons. They had a large number of collectively bargained plans; and when they added their mandatory savings tier, it was easier and more politically acceptable to build on these plans and simply say "We are going to have group choice."

The advantage of group choice is the possibility of economies of scale and expertise, which may keep costs down. But the possible disadvantage is the principal agent problem. The investment manager your employer chooses for you may not be the investment manager you would have chosen for yourself. This manager may not use the investment strategy that you want. You may not think it's the best investment manager for you.

As a result, we see in these countries some choice for workers beginning to emerge--as we in the U.S. have in 401(k) plans, for example. The employer may say, "You can choose any fund you want from among this family of funds." Furthermore, we see in the United Kingdom, and also to a smaller extent in Australia, policies that allow people to opt out of the employer's plan into their own personal account.

So I think that pure group choice is not really a viable political equilibrium. Pressures develop to give people some individual choice. But these OECD countries started with group choice, and still are predominantly group choice.

The other really important feature of the OECD countries is that these countries typically had very modest first pillars. Australia just had a means and asset-tested publicly financed safety net, and Switzerland had a modest pay-as-you-go system. So when they decided they needed a second funded tier, they just added it on. They said, "Well, we had this small tax-financed public pillar. It's clearly not going to be enough as the population ages, so we're going to add on a mandatory saving component." Since they did not divert contributions, there was no transition costs problem in these countries.

The Swedish Model

The third model--which I give to you with a question mark--is the Swedish notional account system. It has been adopted also in Italy and Latvia. To some extent, it is on the verge of being adopted in Poland. China would like to have a funded individual account system, but they can't figure out how to finance the transition; so far, they have a notional individual account system.

The basic idea of the notional individual account system is that, ostensibly, it is defined-contribution instead of defined-benefit but it remains pay-as-you-go. People have individual accounts. They have bankbooks that show their accumulations and the interest that they are earning on the accumulations. But there is actually no money in those accounts. There are no assets. It's virtual; it's notional.

The theory behind this is that you get the advantage of defined contribution and the closer link between benefits and contributions. Yet the system basically remains pay-as-you-go.

On the one hand, this means the countries don't face transition cost problems. In fact, the countries that have adopted this system have done so precisely to avoid facing the transition cost problem. The money keeps flowing into the pay-as-you-go system.

However, it also means that the countries don't get the benefits from funding. They don't get the buildup of long-term national saving or financial market development. They don't avoid payroll tax increases in the future, or intergenerational transfers.

From my way of looking at it, the minuses outweigh the pluses, but these countries have decided that is the way they want to go. Typically, they will have a small funded component, such as 2 or 2.5 percent. In Sweden, collection and record-keeping for the funded component will be centralized, workers will choose the investment manager from among a long list of mutual funds, and the money will be moved in large blocs. Poland is planning to have a larger funded component as well as the notional account.

Basically, the notional account is another way of setting up a large first pillar, the public pillar, and most of these countries are going for a small funded, privately managed second pillar.

Covering Transition Costs

These are the major variations that you see around the world. Let me now explain the ways the Latin American countries, and now Hungary and Poland, are going to cover their transition costs. There are four main methods.

1. Downsizing
   The most universal is to downsize the old system. Usually, the old system is too generous to begin with. It is unsustainable, and it has to be downsized. You cut the benefit rate; raise the retirement rate; change the indexation method, often from wage indexation to price indexation; and eliminate the abuse of disability.

This is almost universal. We find this in practically every country that has reformed. And I would say that it's essential.