(Reprinted from HKCER Letters, Vol. 66, July/August 2001)


MPF: An Unfinished Business

Alan Siu



The Mandatory Provident Fund (MPF) Schemes Ordinance enacted in 1995 provides the legal framework for the establishment of a system of employment-based, privately managed, defined-contribution retirement savings schemes. A comprehensive pension system has three important functions: saving, redistribution and insurance. The MPF system is just a forced saving scheme, and is not designed to play any role in redistribution or insurance.1

After much preparation, the MPF system commenced operation on December 1, 2000. The estimated annual MPF contributions will amount to $10 billion in the initial years of operation and expand to $60 billion when the system matures after 30 years. This source of funds will provide impetus to the further development of the capital market, and is good for the already sizable fund management industry in Hong Kong. The intermediation role of banks will be under pressure. Instead of relying on bank financing, medium-sized firms will have more access to the capital market in the future.

The population of Hong Kong is relatively young when compared with many developed economies, but is getting older quite rapidly. At present, people aged 65 and above account for about 10 percent of the population. This proportion is projected to be 13 percent by 2016 and to 20 percent by 2036. As the population gets older, there will be more retirees and fewer workers. In order to maintain the same standard of living, the economy will need to be more productive. This can be brought about by additional savings to increase the capital available to the economy in the future. The MPF system is a forced retirement saving program. A typical worker is required to contribute 10 percent of his earnings to an investment fund, with 5 percent of the contribution is paid by his employer. Whether the system will succeed in raising the saving rate is unclear. On the macro level, Hong Kong already has a high saving rate of more than 30 percent. There are no statistics on household savings. It is unknown how many families have a saving rate of less than 10 percent, and how many will be induced to save more. People can undo the forced saving by increased borrowing. As reported in The Economist (1998), In the first five years of the Australia mandatory superannuation system, workers contributed US$110 billion, but increased their borrowing by almost as much. In Hong Kong, credit card debts have been mounting since the Asian Financial Crisis, but it is not known how much this is related to the MPF system.

After nine months of operation, the Government will launch its first review of the MPF Ordinance in September. The review will likely concentrate on administrative issues, and may not pay sufficient attention to long-run fundamental issues. This article will address three key issues, namely, MPF retirement benefits, annuitization, and the interaction between MPF and the welfare system. We offer some concluding remarks in the last section.

MPF Retirement Benefits

The goal of the MPF system is to accumulate a fund to support a target income for members during retirement. The target is commonly expressed as a replacement rate that gives the ratio of real benefits to real earnings at the time of a worker's retirement. Given a 10 percent contribution rate and retirement at 65, the replacement rate depends on five uncertain factors: the length of working life, the real wage growth, the real rate of return on investment, the real interest rate at retirement and the number of years in retirement. The replacement is thus a random variable, and it is difficult to characterize its distribution. Nevertheless, it would be useful to have some idea of what the replacement rate would be under different sets of assumptions.

In order not to complicate matters, assume continuous contribution at 10 percent of income since joining the system, a 5 percent return on investment, a 5 percent interest rate, retirement at 65 and 15 years in retirement. Table 1 in Siu (2000) gives the real retirement benefits under these assumptions. Three scenarios are given, assuming a 1 percent, 2 percent and 3 percent real wage growth.2 Two measures are used to gauge the level of retirement benefits. The second column gives the number of monthly salary accumulated by age 65, while the third is the replacement rate defined as monthly pension benefits as multiples of final salary.

Ambitious Coverage

A couple of points can be drawn from Table 1. The logic of compounding means that it is better to start saving early. The replacement rate declines with real wage growth, other things the same. Obviously, a higher real wage growth is better than a lower one. Hence, focusing on the replacement rate and ignoring the level of retirement benefits can be misleading. The replacement rates for workers older than 50 are lower than 20 percent. These workers cannot count on the 10 percent mandatory savings alone to provide them with a decent retirement income. They must draw on other forms of support during their retirement.

Even though it is beneficial to start saving early, it does not make economic sense to coerce young workers to save for their retirement. Their earnings are low. They are often liquidity constraint, and may even be debtors. To the extent that they save at all, rather than worrying about life after retirement, they have more pressing concerns, like saving for their marriage or for the downpayment of their home. The 10 percent mandatory contribution rate would mean that they have to borrow money at high interest rates, or to postpone their plans, like getting married or having their first child. Both effects are counter-productive to retirement protection. The median age of first marriage for men is 30, and for women is 27. The total fertility rate is 951 per 1,000 women in 1999 which is among the lowest in the world. Further drop in the marriage age will reduce the number of births which will then reduce the number of workers 20 years down the road.3 It is better to let young workers, say less than 30, opt out of the MPF system.


One major criticism against a defined-contribution system with individual accounts is that it has high administrative fees (Diamond 1993). An annual management fee of 1 percent can reduce the value of accumulated benefits by 19.6 percent after a 40-year work career, when compared with no fees (Diamond 1999). The point is that a high administrative fee can reduce accumulated benefits substantially.

Under the MPF system, trustees are free to set their fees. They are only required to make the fees and their structures transparent. There are different types of fees, like front load, bid-offer spread, administrative and investment management fees. At present, in the initial launch phase when schemes are competing for members, most of the fees are waived, and fees are quoted in terms of percentage of net assets under management.4 The fees now range from 1 to 2 percent. Large employers can negotiate better rates for their members.

The current fees are lower than the fees commonly charged for mutual funds offered in Hong Kong. Whether the fees will rise when the industry settles in a couple of years is something to watch out for. The industry has economies to scale and scope. It is predicted that the market will eventually support only 6 companies. With reduced competitive pressure, each remaining trust might want to take advantage of their market power by raising fees.


Under the MPF system, the workers have to shoulder the risks associated with fund accumulation. The ultimate benefits will depend on the growth in their earnings, on the returns earned on the portfolio held in their account, and on the length of their working lives. The risks involved can be quite substantial. Uncertainties about future economic conditions can cause realized benefits to fall one-third below the planned target, or to be twice as high as the target.5 The actual year to retire can also be crucial. Retiring when the asset markets are depressed can hurt badly.

In addition to the risks during the accumulating processes, workers also face risks when they annuitize their accumulated benefits. The risks involved in the annuitization process are interest and mortality risks. For a given accumulation, workers who retire when real interest rates are low will receive a smaller real annuity than those who retire when real interest rates are high. This risk can be hedged by increasing the weights of long-term bonds in the investment portfolio when approaching retirement.

The mortality risk refers to the length of life after retirement. Mortality improvements have been uneven over decades in the past, and projections for the future are highly controversial. As mortality rates fall, a given accumulated retirement fund has to be spread over a longer time, thus shrinking the replacement rate. Take the case of Japan. Between 1953 and 1990, life expectancy at age 65 for Japanese men increased from 11.8 years to 16.7 years. This 40 percent improvement in life expectancy would reduce the retirement benefit by about 30 percent (Thompson 1998).

To assess the risks involved under the MPF system, workers will need expert pension advice. Complicated computation is needed to tell a given worker what saving rate is needed to have a probability of say 95 percent in attaining any given replacement rate at retirement. The MPF Authority can consider providing such a useful service to the public at its web site.

Knowing the risks is one thing, preparing for them is another matter. Insurance is not available for many of the risks inherent in the MPF system. Workers have to insured against the risks themselves. Prudence dictates a higher saving rate than the mandatory 10 percent.6


The MPF system provides a structure for accumulation in individual accounts, but none for the provision of retirement income flows. The system permits retirees to take their accumulated benefits as a lump sum. This arrangement is laudable in allowing retirees maximal freedom to decide on how to deccumulate their retirement benefits accrued under the system, but is rather odd. The rationale for having a forced saving program is that some people are myopic. This lack of foresight suggests that people would consume too rapidly out of their lump sum withdrawals, resulting in low incomes and poverty among the very old; thus defeating the very purpose of setting up the MPF system in the first place. Furthermore, given the current welfare programmed for the aged, people with only a modest nest-egg would have a powerful incentive to consume rapidly so as to be eligible for the public pension programmed. To guard against myopic and opportunistic behavior, the Chilean pension system does not allow for lump sum withdrawal; accumulated benefits have to be used to purchase an indexed annuity or to be drawn down periodically subject to a maximal rate of withdrawal (OECD 1997).

It is understandable that the government has taken a minimalist approach to the withdrawal of retirement benefits. Lump sum withdrawal is the simplest option, and accords well with the notion that the retirement benefits are owned by the retirees who should have the freedom to dispose of their accumulated funds as they see fit. If retirees want to purchase an annuity, they can do so in the market. Annuities provided by the market tend to be expensive, due to the adverse selection problem.7 Left to their own devices, people do not take much advantage of annuities. The current individual annuity market in Hong Kong is extremely tiny, with only 247 annuity policies in 1998.8

The demand for annuity products in Hong Kong will grow in the future, and the market will supply such products to meet the demand.9 An important issue is whether products will be available to protect against inflation. It is unlikely that the market will provide indexed annuity because of the lack of indexed bonds dominated in local currency. As a small open economy with its currency linked to the U.S. dollar, the Hong Kong government can do little about inflation.10 The government would not voluntarily want to bear any inflation risk by issuing indexed bonds.

Instead of buying nominal fixed annuities, retirees can buy variable annuities as a partial guard against inflation. The underlying assets of variable annuities are securities that act as hedges against inflation.11 Instead of bearing inflation risk, buyers of variable annuities have to shoulder investment risk. There is no free lunch.

The market will attempt to screen out bad risks. It is likely that a substantial number of retirees will be priced out of the annuity market, and run the risk of outliving their retirement funds. The policy question is whether to require all retirees to buy annuity and to coerce insurance companies to use a single life table for all members of a retiring cohort. Government intervention is needed to overcome the problem of adverse selection, thus resulting in a more efficient pooling of mortality risks. For insurance products, group choice is always cheaper than individual choice. Good risks will then be asked to subsidize the bad risks. For example, treating men and women in the same risk class will on average redistribute from men to women. Whether such redistribution is called for is a value judgment that the government might be asked to make in the future.


The MPF system will do little for workers whose lifetime earnings are low, and for those who have long spells of unemployment. The current stance of the government seems to be that a tiny nest egg for such workers is better than nothing, and that the current CSSA programmes for the aged poor will continue to provide a safety net. This assessment of the peaceful coexistence of the two systems need to be reexamined. The old-age CSSA programme can undermine the MPF system for low and medium income workers.

Currently, a person aged 60 or above can collect $4,000 each month from CSSA, subject to income and asset tests. A poor worker near the point of eligibility for CSSA has little to lose and much to gain from holding a high-risk portfolio. The government has legislated every MPF scheme to offer a money market fund to its members. The aim is to provide a relatively low risk investment vehicle for low-income workers. Based on economic consideration alone, a low-income worker will be ill advised to choose a low risk investment portfolio.

The option to withdraw accumulated retirement benefits in a lump sum also provides for opportunistic behaviour. Suppose a worker retires with $100,000 saved up in his retirement account. What is the best way to dispose of this sum of money? A cynical economist would advise the worker to live like a king for a short while or to give his children a handsome gift, and then rely on the welfare system to take care of him for the rest of his life. The public will not take kindly to this kind of behaviour.

Means-tested welfare programme can also create a major disincentive for low and middle-income workers to save on top of their mandatory contributions for their retirement. It also discourages older workers to continue working after attaining the retirement age. It can also be inequitable. Suppose two workers, A and B, have the same lifetime earnings. Worker A does not save for retirement and consumes all his disposable income, while worker B saves by investing in his children. Worker B's children would wind up supporting the retirement of both A and B. This is patently unfair.

Every redistribution programme has disincentive effects. The question is whether the MPF system and the current old-age welfare program can coexist. Given the obvious disincentive effects, and likely opportunistic behaviour, public support for the old-age CSSA program may dwindle in the distant future, and the government has to revamp the system.

Waiting for the system to unravel and then searching for remedies will be costly. The government can take a proactive approach. One solution is advance funding. The government can commit to phase out the old-age CSSA programmes after say, twenty years, and start making top-up contributions for low-income workers now. In the first quarter, there are about 390,000 workers earning less than $5,000 a month. The government can top up the MPF accounts of these workers by say, $3,000 a year. This will cost the treasury $1.2 billion a year. This is a non-trivial sum of money, and would account for about 10% of the total MPF contributions, which are estimated to be $10 billion in the first year of operation of the system.

The top-up contributions are not spending, but are savings for future obligations under the current CSSA system. The difficulties are how to define the cut-off point for making top-up contributions, and to decide on the level of the top-up. Whether the government can really commit to phase out the current old age CSSA programme is uncertain. But the major hurdle for advance funding for future welfare obligations is to convince policy-makers that there is really a need for policy initiative.

Concluding Remarks

The MPF system has many good features. The government is relegated to play the regulatory role, leaving investment decisions to the private sector. The system will help the development of the capital market, and is designed to have minimal adverse effects on the labour market.

The coverage of the system may be too ambitious. Young workers are often liquidity constraint, and have other pressing needs. Forcing them to save for retirement would be counter-productive. It is worth considering letting them opt out of the system.

The MPF is a minimalist saving program for retirement. Its focus is on saving for the long term. It has no minimum guarantee on the investment returns, no annuity option, and allows for a lump sum withdrawal. The system does not cater for redistribution and insurance. Workers have to bear all the risks involved in accumulation and annuitization. Private markets are relied upon for investment and insurance. Prudence dictates the purchase of life and disability insurance, as well as saving more than the mandatory rate.

CSSA plays the role in providing a safety net to the needy, but its old-age program will offer disincentives to low and medium income workers, and encourage opportunistic behavior. The old-age CSSA program may unravel due to the lack of public support. The government may want to consider advance funding of its CSSA old-age obligations.


Diamond, Peter (1993). "Privatization of Social Security: Lessons from Chile." NBER Working Paper 4510.

-- (1999). "Administrative Costs and Equilibrium Charges with Individual Accounts." NBER Working Paper 7050.

OECD (1997). "The Chilean Pension System." Aging Working Papers AWP 5.6.

Poterba, James M. (1997). "The History of Annuities in the United States." NBER Working Paper 6001.

The Economist (1998). "Retiring the State Pension" 22 October 1998.

Thompson, Lawrence H. (1998). "Predictability of Individual Pensions." OECD Aging Working Papers, AWP 3.5.

Siu, A. (2000). "Hong Kong's Mandatory Provident Fund." http://www.hiebs.hku.hk/working_paper_updates/pdf/wp1021.pdf


1 This article is a shortered version of Siu (2000). For a more detailed description of the MPF system, please refer to the full version.

2 The real wage grew at a trend rate of 1.04 percent between June 1987 and June 2000, while real payroll increased by 3.1 percent a year between 1994 and 2000. Payroll covers wages, bonus and other cash payments.

3 Technically, the aged-dependency ratio, commonly defined as the ratio of the population over age 65 to the population of working age, 15 to 64, will worsen. The ratio is a key factor driving retirement costs.

4 This implies cross-subsidization of low-income contributors, because of the consideration fixed cost component of the fund schemes.

5 See Thompson (1998).

6 The additional saving can be invested in one's children. The family has played the role of risk sharing for a long time. The investment in children, though also risky, can have a very high rate of return.

7 Actuarially fair annuities are non-existent. Poterba (1999) found that for a 65-year old male annuity buyer in the U.S., the expected present discounted value of the benefits of a typical policy available in 1998 was approximately 85 percent of the purchase price.

8 In contrast, in 1998 the number of individual life insurance policies is 3,781,201 which is more than half of the size of the population in Hong Kong.

9 The growth of defined contribution pension plans in the United States has spurred the development of the annuity market. See Poterba (1997).

10 Between 1981 and 1999, the inflation rate averaged out to be 7.2 percent, with a standard deviation of 3.8 percent.

11 The underlying securities tend to rise in value with the nominal price level. The payout of a variable annuity is higher when inflation rate goes up. According to Poterba (1997), variable annuities grew rapidly in recent years in the U.S.

Alan Siu is the Deputy Director of the HK Institute of Economics and Business Strategy of The University of Hong Kong.


| Index | Research Projects | HKCER Letters |
| Speaker Program / Conference | Index of Economic Freedom |
The Hong Kong Centre for Economic Research
School of Economics and Finance
The University of Hong Kong
Phone: (852) 2547-8313 Fax: (852) 2548-6319
email: hkcer@econ.hku.hk