The occupational pillar of the South African pension system

Abstract

In a study on pension reform, the World Bank recommended a multi‐pillar pension system to provide for pension needs. The South African pension system rests on three pillars: an occupational pillar, a voluntary saving pillar, and a redistributional pillar. The main focus of this article is on the first pillar. South Africa has a well‐developed occupational pension system, but several problem areas remain, such as limited coverage, lack of competition between funds, and the taxation of funds. To improve the occupational pension pillar it is recommended that licences should be made available to private pension fund administrators (PPFAs) to launch new open pension funds, as was done in Chile. Such PPFAs could extend coverage, promote competition between funds, and also cater for the specific needs of lower‐income individuals. If PPFAs do not perform satisfactorily, their licences may be withdrawn and made available to new bidders.

1 INTRODUCTION

Pensions are an important aspect of the social security system, and form an integral part of the social insurance and social assistance components of social security in South Africa. Social insurance usually refers to programmes that provide some insurance against life's contingencies, such as disability, unemployment and old age (e.g. through unemployment insurance and occupational pensions). These programmes are usually financed with the aid of contributions made by employees and employers, and only those who have made contributions qualify for benefits. Social assistance programmes (e.g. child support grants and social pensions), on the other hand, provide support to needy individuals. They are usually financed through general government revenue and do not require any contributions. Measures such as the means test are usually employed to determine who are eligible for benefits or not.

To limit the scope of this article the focus will mainly be on occupational pensions. Occupational pensions are important for the modern economy not only because they provide an income for the aged, who are at a very vulnerable stage of their life, but also because in most countries, contributions to pension funds are responsible for generating a substantial part of personal contractual savings. Occupational pensions are also a significant contributor to the development of financial markets and, as a result, fulfil a very important function, especially in developing countries that often do not have well‐developed financial markets.

Though pensions funds can be distinguished from provident funds (see Section 5.2), in this article the term ‘pension funds’ is frequently used as a collective noun for all retirement funds, unless otherwise stated. The two concepts, ‘pensions funds’ and ‘retirements funds’, are also used interchangeably.

Pension reform is currently receiving much attention internationally, inter alia, because changing demographics are endangering the financial viability of pay‐as‐you‐go national pension schemes. During the last couple of years, many countries were forced to amend their pension systems, or are considering changes to ensure sustainability. This is often referred to as the so‐called ‘crisis’ in social security, and is one of the reasons why the World Bank (1994) undertook a research project on pension reform. This study proposed, among other things, a three‐pillar pension system, a concept that sparked much interest and debate by stakeholders. The World Bank, however, acknowledged that the study does not serve as a blueprint, but rather as a benchmark for pension reform. Furthermore, it also admitted that conditions might vary between countries due to different historical developments, cultural differences, the social environment and the approach to social security issues. It is important to note at the outset that ideology plays an important role in the development of social security systems, as can be seen in the differences between the welfare states of some European countries and the United States.

Although the South African pension system is relatively advanced for a developing country, it has several shortcomings. One problem is that many South Africans will never have the opportunity to become members of occupational pensions, because they might not have the opportunity to find employment in the formal sector of the economy. Moreover, some of those who do secure formal employment might not become members of a pension fund, due to being employed by a sector or firm that does not have an occupational pension fund in place. Such individuals might not make any (or sufficient) provision for their retirement and, as a result, may eventually be unable to maintain their living standards after retirement or become dependent on social pensions. Social pensions are relatively low (R640,00 per month per individual in 2002) and provide for only a very basic minimum living level. As they are financed through the national budget, any increase in social pensions also raises the level of government expenditure.

New developments in occupational pensions, such as the replacement of conventional pension funds with provident funds, tend to have detrimental effects on the insurance motive. Members of occupational pension funds might end up with an insufficient pension to maintain their living standards, due to market volatility at the time of retirement, and might even exhaust their savings before they die. Another problem is that funds that are withdrawn by members due to job changes or retrenchment are often not reinvested for pension purposes.

As indicated earlier, this article focuses mainly on the occupational pillar of the South African pension system. The next section examines why government involvement in the area of pensions may be necessary. This is followed by a discussion of the characteristics of a well‐functioning pension system, in which the recommendations of the World Bank receive special attention. The next section provides a brief overview of the country's occupational pension system, highlighting some of the advantages and problem areas. The article concludes with recommendations that may improve the South African pension system.

2 WHY IS GOVERNMENT INVOLVEMENT NECESSARY?

In most countries the government is involved in pensions either by way of provision or regulation. But should the government still be involved in this area at a time when a market‐oriented economy and its efficiency gains are widely accepted and where privatisation has become paramount? The market is more than capable of providing insurance against the contingencies of old age to individuals who wish to purchase such insurance voluntarily. However, according to the World Bank (1994:5), if pensions are left to the private sector and the voluntary choices of individuals only, several problems may be encountered:

1.	Some individuals may be short‐sighted and may not make any or sufficient financial provision for their old age. This contradicts the implicit rationality assumptions of life cycle models, which assume that people take their full life cycle into consideration when they plan their finances (e.g. saving and consumption behaviour). Myopic behaviour in this regard opens the door to paternalistic behaviour by the government, as individuals may not always take the possibility of reaching old age sufficiently into account when they decide to maximise their welfare and, as a result, the government has to intervene. This is a disturbing idea for neoliberalists, who argue that an individual must have the freedom to choose to be poor in old age. Freedom is a complex concept and a distinction can be made between positive and negative freedom (Berlin, 1969:122–34); however, a definition of the concept does not fall within the scope of this article. Friedman & Friedman (1982:188) argued that supporters of freedom must also allow people the freedom to make mistakes. The problem with such an approach, however, is that it may become rational to be myopic, because it may benefit individuals not to make provision for their old age. Most modern societies find it unacceptable to let people remain in poverty and, as a result, introduce programmes to support the poor. Depending on the levels of benefits provided, it may serve as an incentive to the poor to become free riders. Given their limited information and time horizons, it may be very difficult for young individuals to understand the full implications of poverty during old age, or as Samuelson (1975:543) stated: ‘People live miserably in old age because they do not realize when young what are the consequences of their private savings habits. So by democratic fiat, they paternalistically impose on themselves a within‐life pattern of consumption that favors old age at the expense of young.’ Most societies have therefore decided to target individuals who have the ability to sacrifice present consumption for future consumption.
2.	Saving opportunities may be limited because of underdeveloped capital markets. This is, however, much less of a problem in South Africa than in some other developing countries, as this country has a relatively well‐developed financial sector. Nevertheless, financial institutions may not be all that eager to accommodate low‐income individuals due to the high costs associated with saving accounts that have relatively low balances and are characterised by frequent small deposits and withdrawals. Low‐income individuals may also be excluded from other contractual saving opportunities due to the requirement of relatively high minimum monthly contributions. Unstable macroeconomic conditions can also discourage saving. Periods of inflation and negative real interest rates could, for example, make saving unattractive. Furthermore, the taxation of revenue derived from interest may also be a disincentive to saving.
3.	Two types of market failure – adverse selection and moral hazard – also hamper insurance markets. Adverse selection occurs when low‐risk individuals are not prepared to purchase insurance at actuarially determined prices (Hemming, 1991:22). This implies that high‐risk individuals will be more prepared to purchase such insurance, which will endanger the financial viability of that particular insurance market. Moral hazard refers to a situation where people might not protect themselves from risk as well as they would have if they were not insured at all (World Bank, 1994:xxii). Due to these market failures, insurance markets are not prepared to provide insurance against risks such as investment, inflation, depression, longevity and disability (World Bank, 1994:6). Insurance against disability could be provided by the market, but is usually subject to stringent medical tests to eliminate high‐risk individuals as far as possible or, if insurance is provided, to provide it at exorbitant prices.
4.	Contrary to the assumptions of a perfectly competitive market, perfect information is not readily available, and information may be expensive. In the modern world, saving instruments have become very complex and numerous, and individuals may find it difficult to identify the best saving instrument for their purposes or to determine the solvency of insurance and private saving companies. The Saambou Bank saga in 2002 serves as a case in point. Many pensioners ran the risk of losing all or a large proportion of their savings. To assess the productivity of the multitude of investment opportunities is also no easy task, and even specialists often make mistakes. If individuals discover only late in life that they have made a big mistake with respect to the company or the saving instrument they have chosen, it is usually too late to rectify the situation (World Bank, 1994:6).
5.	Some people are poor and are unable to generate sufficient savings for their old age, which may require a redistributional policy to ensure at least minimum standards of living during retirement. Beckerman (1979:268) reminded that ‘it should be remembered that one of the reasons why state pensions were introduced at an early stage in the evolution of most countries’ national insurance programmes was the failure – usually on account of inability rather than lack of normal prudence and foresight – of most people with low incomes during their working lives, not to mention some with higher incomes, to make adequate provision for their old age' [emphasis added].

There are, therefore, good reasons why the government is still involved in the provision of pensions at a time when market‐oriented policies are in vogue.

3 SOME CHARACTERISTICS OF A WELL‐FUNCTIONING PENSION SYSTEM

As indicated earlier, the World Bank (1994) has taken a keen interest in pension reform, an interest which has both a social and an economic basis. Aged individuals often find it difficult to generate sufficient income and, as informal security or the role played by the extended family is being eroded by economic development, the private sector and the government have stepped in to meet this need. Changing demographics in developed and some developing countries may endanger the viability of publicly provided pension funds, especially if they have implemented pay‐as‐you‐go pension funds rather than fully funded systems. As mentioned, a pay‐as‐you‐go pension fund usually uses current revenue generated from contributing members to finance the pensions of present retirees, and therefore seldom has sufficient accumulated funds to comply with the actuarial requirements of a fully funded scheme. The revenue for such funds is normally generated through an earmarked tax or a pay‐roll tax.

Economic growth is an important requirement for a sustainable pension system, as it can enable both young and elderly people to consume more. Higher consumption can again promote economic growth, which could lead to increases in the real income of the younger generation and also provide scope for the aged to at least maintain their real economic position if their assets are well invested. It is therefore of the utmost importance that the pension system should promote economic growth and not be detrimental to the economy. Pay‐as‐you‐go systems could encounter shortfalls and may become unviable as a result of changes in demographics. Efforts to rectify this situation could hamper economic growth, as the present workforce will have to pay higher taxes.

In its study on pension reform, the World Bank (1994:10) identified the following requirements for a well‐functioning pension system. It must:

1.	At least provide an opportunity to save for one's old age (i.e. redirect present consumption for future consumption)
2.	Address redistributional issues, as poor individuals are usually unable to make sufficient provision for their old age
3.	Provide insurance to protect participants against the occurrence of: a. Bad investment decisions or recessions, which may wipe out a large part of a person's savings b. Inflation, which may erode the real value of pensions c. The possibility that individuals may outlive their savings d. The failure of public programmes

A further requirement that could be added is a relatively stable exchange rate. Large fluctuations in the exchange rate could have a significant impact on investments overseas.

To comply with these requirements, the World Bank (1994:15) recommended that a pension system should rest on three pillars: a mandatory publicly managed pillar, a mandatory privately managed pillar, and a voluntary pillar. In the following section the main characteristics of the South African pension system are highlighted to determine to what extent it meets these requirements.

4 THE SOUTH AFRICAN PENSION SYSTEM

The South African pension system is relatively well developed and complex. It consists of occupational pensions, social pensions, and voluntary saving, such as contributions to annuity funds or other forms of investment.

4.1 Occupational pensions

Occupational pensions in South Africa make a major contribution towards contractual savings. In 2001, total assets of the retirement industry amounted to R836 billion, or approximately 86 per cent of the gross domestic product (GDP), while the net assets of self‐administered funds were approximately 40 per cent of GDP (Registrar of Pension Funds, 2001:6). According to the Mouton Committee (1992:7), South Africa ranks high in this regard compared with other countries, while the Taylor Commission (2002:93) also estimated that the private pensions and insurance sectors of South Africa are the largest in the world as regards gross national product (GNP). One reason for this phenomenon is that South Africa has a relatively long history of pensions. The first pensions in this country were introduced as far back as 1837, although occupational insurance for retirement only became institutionalised in the 1920s (Mouton Committee, 1992:37–8). According to Van der Berg (1994:18), the Law on Pension Funds of 1956 was an important landmark in regulating the financial responsibilities of pension funds.

If the South African pension system is compared with those of other countries and tested against the requirements identified by the World Bank, one important limitation is that South Africa lacks a mandatory publicly managed pension scheme. This issue has often received attention in the past. Both the Mouton Committee (1992) and the Smith Committee (1995) concluded that South Africa should not take this route, mainly because only those people employed in the formal sector of the economy, the majority of whom already belong to occupational pension schemes, would contribute to such a scheme (Smith Committee, 1995: Annexure 5:36–7). Membership of a pension fund is only compulsory for persons employed by industries or sectors that have a pension fund in place. Many employees in the formal sector therefore do not have the opportunity to become a member of an occupational pension fund. These issues will be raised again later.

4.2 Social pensions

A rather unique characteristic of the South African pension system (especially for a developing country) is the existence of social pensions. This social assistance programme has been devised to assist needy elderly persons older than 60 years (women) and older than 65 years (men). Social pensions ensure that the South African system complies with the World Bank requirement that redistribution should be addressed, as many individuals are unable to contribute towards retirement provision. Social pensions in South Africa do not require contributions, and beneficiaries are usually identified with the aid of a means test. Social pensions are relatively low (R640 per month per person in 2002). The fact that the redistributional function is separated from the other pillars may be regarded as an advantage of the South African pension system. In most countries this function is the responsibility of the mandatory public pillar, which implies that higher income groups will receive a lower pension than justified by their contributions. This is one of the reasons why a mandatory public pension system is unpopular.

The history of social pensions reveals the impact of racial discrimination on the South African social security system. Social pensions were introduced in the 1920s for whites and coloureds, and during the 1950s coverage was gradually extended to include blacks (Lund, 1993:9–10). The levels of pensions originally differed with regard to geographical area and racial classification. For example, in the early years of the programme, lower pensions were paid to whites in rural areas compared with those in urban areas, while three levels of pensions were introduced for blacks, coloureds and Asians, depending on the geographical area in which they lived. Furthermore, the rates of pensions also differed between the eight most important urban areas in South Africa, in an effort to remove incentives for blacks to move to urban areas. Between 1960 and 1968, some of these differences were gradually phased out (Pollak, 1981:154–5). The levels of pensions for the different racial groups were, however, not equal. Parity was only achieved in 1993 (Van der Merwe, 1996:400). This was engineered over a period of several years, by keeping adjustments in the pensions of whites below the inflation rate (at approximately 60 per cent of the inflation rate), thus reducing their real value, while the increases in the levels of pensions for the other racial groups exceeded the inflation rate.

Social pensions are the most important component of the South African social safety net, because the extended family of elderly individuals usually also benefits from these grants, making it the most important programme of poverty alleviation in South Africa. Van der Berg (1994:27) indicated that social pensions circulate income to the broader poor community, a view which is mirrored by Le Roux (1995:7), who concluded that social pensions are exceptionally well targeted to benefit poor households. If it is assumed that all members of a household benefit from the pension of an elderly member, it can be argued that five to six other individuals may stand to gain from one person's social pension (Le Roux, 1995:11). Case & Deaton (1996:1) also concluded that social pensions are an effective instrument of redistribution and that they benefit poor households the most. The question whether or not this is indeed the most appropriate way of alleviating poverty falls beyond the scope of this article. One drawback of such an approach, however, is that only households with elderly individuals benefit from these transfers. Given their importance as part of the social safety net, social pensions cannot easily be changed or phased out, without replacing them with at least an equally well‐targeted programme, such as a basic income grant.

4.3 Voluntary saving

Voluntary saving enables persons who may or may not be members of an occupational pension fund to make (additional) provision for their retirement. Voluntary saving may be the only opportunity to accumulate retirement funds for those persons employed in industries without occupational pension schemes, for the self‐employed, for low‐income groups and for those employed in the informal sector. Some of these groups may, of course, not have the means to make provision for their retirement on a regular basis.

If the government wants to promote voluntary saving and capital accumulation, it must create a stable macroeconomic environment and see to it that the following criteria are met (World Bank, 1994:247):

1.	The economic environment should be stable and non‐inflationary.
2.	The regulatory framework should promote confidence in financial institutions.
3.	Citizens should be educated to realise that a relatively high savings ratio is required to maintain living standards later in life.
4.	The necessary tax incentives should be provided.

With regard to the first requirement, South Africa has made some progress, although the sharp depreciation of the rand in 2001 created renewed inflationary pressure. The strict monetary policy adopted since this episode, and the consequent appreciation of the rand, may enable the South African Reserve Bank (SARB) to reach its inflationary target of between 3 and 6 per cent in the near future. The position with regard to the second requirement is less favourable. The dismal way in which the Saambou Bank fiasco in 2002 was handled by the authorities did not contribute to confidence in banks or the financial sector. Smaller banks, in particular, are suffering as a result. South Africa also did not perform that well with regard to the third requirement, with nominal savings by households decreasing from R5 718 million in 1991 to R1 437 million in 2001 (SARB, 2002:S‐131). Expressed as a percentage of disposable income the amounts were 2,7 and 0,2 per cent respectively. Tax incentives (or exemptions) to promote saving have recently been increased to R6 000 of interest and dividends for persons under 65 years, and to R10 000 for persons 65 years and older (SARS, 2003:11). Although these increases should be welcomed, it is doubtful whether they will have a significant impact on people's saving behaviour.

Voluntary saving instruments available to South Africans range from the rather primitive way of keeping savings in a ‘safe’ place at home to sophisticated instruments such as unit trusts, bonds, equities and hedge funds. As mentioned earlier, South Africa, unlike many other developing countries, has a well‐developed banking and insurance sector that offers a whole range of saving instruments. At present, banks are regulated by the SARB, while other financial institutions are regulated by the Financial Services Board.

To high‐income earners the number of saving opportunities is almost unlimited, ranging from low‐ to high‐risk options. The relaxing of foreign exchange control also allows persons to make offshore investments up to a certain limit (R750 000 at the time of writing). Low‐income earners, however, are not as fortunate and are usually restricted to more conventional methods of saving, such as an ordinary savings account at a bank. These accounts enable people to save on a regular or irregular basis, usually accept relatively small amounts and in this way meet some of the saving needs of low‐income earners. However, in an effort to cut administrative costs, banks usually pay very low nominal interest rates (negative real interest rates) on relatively small balances in savings accounts. Furthermore, the banking costs (i.e. service charges) of such accounts are also usually exceptionally high. Banks levy high service charges to penalise individuals (usually the low‐income earners) who keep relatively low amounts in these accounts, as they often use saving accounts for making deposits and withdrawals on a regular basis. These excessive costs have received some attention in the popular press, when people complained that an amount of more or less R100,00 deposited in a savings account at a bank would decrease rapidly due to high banking costs, and that the account holder could even end up with a negative amount. More sophisticated saving instruments usually require a relatively large lump sum or monthly payment, which effectively puts them out of the reach of low‐income groups who may only be able to afford small or irregular contributions. Instruments of voluntary saving should, therefore, receive more attention to ensure that the needs of all citizens are met.

Summary

As stated earlier, the guidelines of the World Bank (1994:15) indicate that the best way to address the retirement needs of a country is through a multi‐pillar system of old‐age income security. In its study, the World Bank recommended two mandatory pillars – one which should be publicly managed and financed through taxation, and one which should be privately managed and financed. These two pillars should be complemented by a voluntary pillar. It is possible to argue that the South African pension system already consists of three pillars. However, there are important deviations between the South African system and the proposals of the World Bank, which does not necessarily constitute a problem. The World Bank noted that each country should implement a system that will serve its needs best, and indicated that its recommendations should only serve as a benchmark. The first two pillars of the South African pension system are privately managed and financed, while the third is publicly managed and tax‐financed. The first pillar – which is also the focus of this article – consists of occupational pensions, which is a social insurance programme and mandatory to employees if the firm or industry has a pension scheme in place. The second pillar consists of voluntary saving by individuals at private financial institutions, while the last pillar is redistributional in nature and consists of social pensions.

5 OCCUPATIONAL PENSIONS

As pointed out earlier, occupational pensions are an important component of the South African pension system. In broad terms, occupational pensions refer to funds that are managed by the private sector (e.g. industrial councils, insurers and private firms) and also include funds managed by institutions such as the central government and local authorities. According to the Pensions Funds Act 24 of 1956 (South Africa, 1956), the Registrar of Pension Funds – a subdivision of the Financial Services Board – is responsible for:

1.	The supervision of certain aspects of foreign funds for members who are resident in South Africa
2.	Underwritten funds, which are funds that ‘operate exclusively by means of policies of insurance issued by registered insurers in the Republic’ (Registrar of Pension Funds, 2001:4).
3.	Self‐administered funds, which are funds that ‘invest their assets with bodies and institutions in the public and private sectors of the economy on their own behalf’ (Registrar of Pension Funds, 2001:4).

Certain funds do not fall under the jurisdiction of the Registrar of Pension Funds, for example:

1.	Funds that are founded by special law for employees of the state and some of the parastatal institutions
2.	The fund established for the employees of Transnet
3.	The fund founded for the employees of Telkom SA Limited

The Post Office Pension Fund

1.	Funds set up by collective agreements and which have been concluded by councils in terms of the Labour Relations Act 66 of 1995 (South Africa, 1995). In these cases, the Department of Labour supervises the funds (Registrar of Pension Funds, 2001:4).

From the above discussion it is clear that the regulation of pension funds is not centralised, as two institutions, the Registrar of Pension Funds and the Department of Labour, are responsible. Ideally, one institution should be responsible for the regulation of all pension funds, to ensure that standards are uniformly applied.

5.1 Membership, beneficiaries and coverage

In South Africa, membership of occupational pension funds is not mandatory and neither are companies obliged to start their own funds. However, if firms and industries already have a pension fund in place, new employees complying with certain requirements have to become members. According to Van der Berg (1994:18), pension funds experienced a growth in membership, even in the absence of compulsion, because several Industrial Council agreements committed employers to maintain pension funds.

The number of retirement funds has increased substantially, from 2 771 in 1958 to 15 023 in 2001 (Smith Committee, 1995: Annexure D2.3; Registrar of Pension Funds, 2001:21). Membership of pension funds has also increased substantially, from 923 000 in 1958 (Smith Committee, 1995: Annexure D2.4a) to 9 533 846 in 2001 (Registrar of Pension Funds, 2001:5). These figures must, however, be treated with caution as double counting occurs, due to some individuals belonging to more than one fund. In 2001, membership consisted of 8 252 092 active members and 1 281 754 pensioners, deferred pensioners and dependants (Registrar of Pension Funds, 2001:5). Despite their limitations, these figures can be used to determine what percentage of the economically active population are members of occupational pension funds. The figures imply that, in 2001, more or less 51 per cent of the economically active population were active members of such a fund. Active members of pension funds make up more or less 30 per cent of the South African population between the ages of 15 and 65 years. A large proportion of the South African population is thus still not covered by occupational pension funds and will eventually be dependent on a social pension when they reach retirement age. Despite the progress that has been made, it is estimated that only 6 per cent of South Africans are self‐sufficient when they reach retirement age (Downie, 2002:1.7).

In 2001, the total benefits paid by pension funds in South Africa were R111 206 million. Table 1 shows that 23 per cent of benefits paid in 2001 went towards pensions, while 77 per cent went towards lump sum payments. Approximately 23 per cent of lump sums were paid at retirement or death, while almost 73 per cent were paid at resignation or termination of membership. These latter payments receive more attention in Section 5.6, as they may have a negative impact on the final levels of pensions if people do not reinvest these accumulated funds for retirement purposes.

Benefits paid by all pension funds, 2001

Type of spending	R million	% of lump sum payments	% of total benefits
Pensions	25 536		23,0
Lump sum payments	85 670		77,0
• On retirement or death	19 897	23,2	17,9
• Resignation and termination	62 291	72,7	56,0
• Other	3 482	4,1	3,1
Total	111 206	100	100
Source: Registrar of Pension Funds (2001).

Between 1960 and 1993, the number of beneficiaries of social and occupational pensions increased from 354 400 to more than 2,4 million (Smith Committee, 1995:4). In 1997, this number exceeded 3,2 million. The number of beneficiaries of occupational pensions increased from 14 000 in 1960 to 896 900 in 1993 (Smith Committee, 1995: Annexure D2.5). By 1997, this number, which includes pensioners, deferred pensioners and dependants, had almost doubled to 1 612 488 (Registrar of Pension Funds, 1997:3). Note that after 1997, the Registrar of Pension Funds stopped publishing these figures.

Coverage of workers in the agricultural services, domestic services and trade, catering and accommodation sectors is still very low. According to an estimate done for the Mouton Committee (1992) for an unspecified year, 27 per cent of persons employed in the formal sector were not members of a pension fund. Some of these employees had such meagre incomes that it could hardly satisfy their basic needs, let alone allow for contributions to a pension fund. Furthermore, almost 55 per cent of the population in the age group 20 to 64 years (an estimated 9,9 million) are not employed in the formal sector of the economy and therefore do not have the opportunity to become members of an occupational pension fund.

Changes in labour practices due to globalisation may also have a negative impact on occupational pensions. In a world characterised by more open markets with greater competition, there is an incentive for manufacturers to cut labour costs, as it may be more difficult to cut other costs. This is manifested in the shift from formal full‐time employment to contract work. Contract work enables employers to decrease their salary and wage bill, as they need not employ individuals during quieter periods. Furthermore, they need not contribute towards medical and pension funds and other social security programmes, and they also do not need to pay employees when they are on holiday, as the principle of ‘no work, no pay’ applies. A popular criticism of a comprehensive social security system is that it limits the ability of welfare states to compete with developing countries that do not have such a system. On the other hand, the lack of a comprehensive social safety net implies hardship to individuals when they face contingencies such as unemployment.

The growing popularity of contract work also limits the coverage provided by occupational pensions, because it shifts the responsibility of retirement provision entirely to individuals, who may be myopic or lack the necessary know‐how or information to make a sound decision. The only real option available to contract workers is to participate in a voluntary saving plan, which may not provide them with a sufficient income or pension to maintain their living standards when they retire. Due to short planning horizons, they might even make no provision for retirement at all. When they eventually realise the need for such contributions it may be too late to accumulate the necessary funds. In the latter case, people may become the responsibility of the government, which has to support them by means of a social pension. This puts pressure on government expenditure and may also imply a dramatic decrease in the living standards of the persons concerned.

Occupational pensions in South Africa are not designed to attend to the needs of low‐income groups. These schemes usually do not accept relatively small and irregular contributions, which may be the only avenue open to lower income groups. Occupational pensions are also not very popular with low‐income groups, because at retirement the monthly value of their occupational pension will probably be more or less equal to that of social pensions. In such a case it does not make much sense for low‐income groups to be members of an occupational pension fund, as the latter will disqualify them from receiving a social pension (which does not require any contributions).

5.2 Characteristics of occupational pensions

Occupational pensions are usually financed by contributions by employers and employees, paid into a fund in accordance with a contractual agreement, which grants certain defined or undefined benefits. Other benefits, such as withdrawal rights when employees resign, retrenchment benefits and insurance benefits (for disability and dependants of contributors who have died), are usually included in the package (Van der Berg, 1994:18). Occupational pension funds in South Africa fall into the following broad categories (Smith Committee, 1995:68):

1.	Defined contribution funds
2.	Defined benefit funds
3.	A mix of both

A defined contribution scheme usually determines the value of benefits according to the contributions made on behalf of the insured by the employee and the employer, plus the accumulated returns on accumulated funds. Such a scheme therefore cannot experience a shortfall, which explains its current popularity with employers. A defined benefit scheme determines the value of final benefits according to a formula, which usually includes the number of years of membership, multiplied by a certain percentage of the final salary (which may be that of the last month, last year or last couple of years) (Van der Berg, 1994:21). A defined benefit scheme may experience a shortfall, which may be rectified by increasing the rate of contributions or even by requiring a transfer from employers, hence the unpopularity of this type of fund. Different combinations of the two schemes also occur and may vary from fund to fund. Some benefits, such as those for retirement, may for example be determined on the basis of defined contributions, while death and disability benefits may depend on defined benefits (Smith Committee, 1995:69).

A further possible distinction is between a pension fund and a provident fund, although both funds may be either a defined benefit or a defined contribution scheme. The main difference is that with a pension fund only one third of accumulated amounts may be withdrawn as a lump sum when a person retires (the remaining two‐thirds must be taken as a pension), while the total accumulated amount of a provident fund may be withdrawn at retirement. Pension funds can be divided into ordinary pension funds (where membership is restricted to employees) and retirement annuity funds (where membership is open to self‐employed persons) (Katz Commission, 1995:56). For the purpose of this article, retirement annuity funds are classified as an example of a voluntary saving instrument and therefore are beyond the scope of this article.

As in many other countries, there has been a notable shift towards defined contribution schemes in South Africa in recent years. As indicated earlier, one reason is that defined benefit schemes may experience shortfalls, which create financial liabilities for employers and employees. One drawback of defined contribution schemes, however, is the increased risk for participants. The timing of retirement becomes crucial, because retirement during a period of volatile or bear markets may have a significantly negative impact on members' benefits. Funds may counter these problems by offering members the opportunity to switch to less risky investments (away from the stock exchange to money markets, for instance) as they approach retirement age. A further drawback of such schemes is that there may be a relatively large deviation between an individual's last salary and his or her monthly pension, with the result that persons may find it difficult, if not impossible, to maintain their living standard after retirement. The investment decisions of the financial managers of defined contribution schemes are thus very important, as they affect people's accumulated assets. This does not imply that investment decisions will not hamper the viability of defined benefit schemes; however, present employees and employers run the risk of losses, not retirees. An advantage of a defined contribution scheme is that it removes the incentive to promote individuals at the last minute before their retirement commences, which could happen with defined benefit schemes, because such a step (depending on the formula used) could have a decided impact on a retiree's pension.

As pointed out earlier, provident funds allow people to withdraw the full amount of accumulated funds at retirement. This could be an advantage to individuals who have the know‐how to reinvest these funds wisely, as they are no longer at the mercy of fund managers to make wise investment decisions on their behalf. Unfortunately, some individuals might make unwise investment decisions, with the result that their accumulated funds may decrease or, in extreme cases, are wiped out completely. The pyramid schemes (e.g. the Krion scheme), which received a great deal of attention in the press in 2002, showed that persons might be ignorant or greedy and frequently lose their accumulated funds when such a scheme inevitably collapses. Those affected could become poor and dependent on government support. On the other hand, those who reinvest their lump sums with reputable companies could see an improvement in their financial position or at least protect themselves against the erosive powers of inflation over the longer term.

Defined contribution pension schemes thus violate one of the requirements of the World Bank (1994:10), as set out in Section 3, as they do not insure beneficiaries against:

1.	Bad investment decisions or recessions, which may wipe out a large part of their savings
2.	Inflation, which may erode the real value of pensions
3.	The possibility that individuals may outlive their savings

The move to defined contribution schemes has shifted the risk to the members and, in the process, some of the advantages of social insurance have been jeopardised.

5.3 Investment by self‐administered funds

A well‐developed retirement sector makes an invaluable contribution to the development of well‐functioning money and capital markets. It is also important to note that the contractual savings generated by the retirement industry are an important source of capital to the government, local authorities, deposit‐taking institutions and public companies. In the previous section it was pointed out that the investment decisions of fund managers have a decisive impact on the levels of defined contribution pension funds. This section focuses mainly on the investment decisions in the retirement industry.

Regulation 28 of the Pension Funds Act of 1956 sets limits to the proportion of assets that may be invested in different investment asset classes. The main stipulations are summarised below (Ketola, 2003:172–3):

1.	A maximum of 75 per cent of assets may be invested in equities.
2.	A maximum of 15 per cent may be invested in foreign assets.
3.	A maximum of 90 per cent of assets may be invested in a combination of property and equities.
4.	A maximum of 20 per cent of assets may be deposited at a single registered bank; however, 100 per cent of assets may be held as cash.
5.	A maximum of 10 per cent of assets may be invested in Kruger rands.
6.	All assets may be invested in bonds; however, a maximum of 20 per cent may be invested at a singular issuer of bonds (e.g. Eskom). This limitation does not apply if the issuer is the Republic of South Africa itself – in such a case, funds may invest all their assets in government bonds.

Table 2 sets out the investment instruments utilised by self‐administered funds and shows how the situation changed between 1997 and 2001. From the table it is apparent that the four most important investment instruments during this period were equities, insurance policies and bills, bonds and securities, and foreign investment. It is also apparent that important changes occurred during the period. Investment in equities – ordinary shares and preference shares that are listed or unlisted, as well as shares in participating employers, subsidiaries or holding companies – expressed as a percentage of total investment decreased from 43 per cent in 1997 to 31,1 per cent in 2001. The share of insurance policies – linked policies, guaranteed and other policies – increased from 26,2 to 32,6 per cent. The share of bills, bonds and securities – mainly those issued by the central government or provincial administrations – decreased from 14,8 per cent in 1999 to 11,2 per cent in 2001. The share of foreign investment increased sharply from 2,1 per cent in 1997 to 10,1 per cent in 2001, indicating that the retirement industry utilised the new opportunities to invest overseas in order to acquire better‐balanced portfolios. At the time of writing, the government's relaxation of foreign exchange control enabled the retirement industry to shift 10 per cent of the net inflow of funds in the previous year overseas, subject to an overall limit of 15 per cent of assets under management (see Manuel, 2000:12).

Investment instruments utilised by self‐administered funds, expressed as a percentage of total investment, 1997, 1999 and 2001

Investment instruments	1997	1999	2001
Immovable properties	4,1	3,6	2,7
Equities	43,0	34,3	31,1
Unit trust schemes	2,9	5,8	5,1
Debentures	0,4	0,2	0,01
Loans	0,5	0,6	0,8
Participating mortgage bonds	0,0	0,0	0,0
Bills, bonds and securities	14,8	12,0	11,2
Deposits and Kruger rands	5,4	7,0	6,3
Insurance policies	26,2	28,2	32,6
Derivative market instruments	0,03	0,08	0,02
Other investments	0,4	0,2	0,1
Foreign investment	2,1	8,0	10,1
Total	100	100	100
Note: The total in each case might not add up due to rounding off. Sources: Registrar of Pension Funds (1997, 1999, 2001).

Table 2 also shows that the share of immovable properties decreased from 4,1 per cent in 1997 to 2,7 per cent in 2001, while investment in unit trust schemes doubled between 1997 and 1999, but subsequently declined to 5,1 per cent in 2001. Interestingly enough, the share of derivative market instruments did not experience much of a change during the period under consideration, probably due to the high risks and complexity of these instruments. Ketola (2003:173) indicated that the regulation of derivatives is an important omission from Regulation 28, which stipulates the maximum proportion of pension fund assets that may be invested in certain asset classes. High risk and uncertainty about regulation therefore appear to be disincentives to investing in derivatives.

As emphasised earlier, the investment performance of defined contributions funds has a large impact on the eventual level of benefit members will receive. Any evaluation of the investment performance of pension schemes must be conducted carefully, with every effort being made to ensure that ‘apples are compared with apples’. The investment returns of funds may, for example, deviate substantially because of the period under consideration and the fund's risk profile. If markets are favourable, a fund with a high‐risk profile could generate higher returns than one with a lower risk profile. If, however, markets experience a slump, a high‐risk fund could have worse returns than a low‐risk one. Market uncertainties and differences in risk profiles may have an important impact on results.

According to Downie (1997:2.16), it makes more sense to evaluate and compare the performance of funds over ‘a reasonable period of time’, but he does not indicate how much time this should be. A period of at least three to five years is probably required, bearing in mind that historical performance does not necessarily serve as a good guide to future performance. With regard to the minimum performance of a fund, members may require that the returns at least exceed the average rate of inflation over such a period.

Asset managers in South Africa have agreed to comply with the requirements of the Investment Performance Standards of South Africa, an approach driven from the United States through the Association for Investment Management and Research (Downie, 2002:3.2). These standards put forward a stringent set of procedures and protocols for asset managers to comply with, while the results have to be verified by an independent third party. After full implementation of the approach, comparisons of the performances of funds should be easier and more reliable.

The Assetbase International Performance Survey (which uses unaudited figures that are accepted on the basis of honour) illustrates how performances of retirement funds may differ. This survey includes only the portfolios of managers of fully balanced pension and provident funds that have been given full discretion by trustees over their investment decisions, and therefore should provide a good indication of a manager's ability to produce the necessary results (Downie, 2002:3.2). The results of one such survey are shown in Table 3. In a group of 30 funds with standard risk portfolios, the fund with the best performance for the year ending 30 June 2002 showed a return of more than 31 per cent, while the worst‐performing fund had a return of approximately 5 per cent. The top quartile of funds experienced a return of over 14 per cent. The median was more than 12 per cent, while the bottom quartile of funds experienced a return of more than 10 per cent (Downie, 2002:3.5).

Investment performances of funds provided by the Assetbase International Performance Survey for standard risk portfolios, aggressive risk portfolios and conservative risk portfolios, for the one‐, three‐ and five‐year periods ending 30 June 2002

	Percentage change
	One year	Three year	Five year
Number of managers (including universe)	31	29	24
Highest	31,86	32,51	27,39
Top quartile	14,52	16,73	13,18
Median	12,70	15,40	12,10
Bottom quartile	10,11	13,94	10,42
Lowest	4,79	11,79	8,33
Aggressive risk portfolios
Number of managers (excluding universe)	10	9	5
Highest	16,50	18,84	17,29
Top quartile	14,07	16,82	16,61
Median	13,34	15,23	14,38
Bottom quartile	8,26	12,84	10,57
Lowest	1,75	9,50	9,48
Conservative risk portfolios
Number of managers (excluding universe)	4	4	3
Highest	14,14	16,73	14,78
Top quartile	13,33	15,81	14,60
Median	12,21	15,11	14,45
Bottom quartile	9,07	14,31	14,03
Lowest	8,35	11,86	13,01
Source: Downie (2002:3.5–3.9).

For the three‐year period ending on 30 June 2002, the best‐performing fund had a return of more than 32 per cent, while the fund with the worst performance had a return of about 12 per cent. Over the five‐year period ending 30 June 2002, the fund with the best performance had a return of more than 27 per cent, while the worst‐performing fund had a return of more than 8 per cent (Downie, 2002:3.5). The findings of this survey indicate that, the longer the period under consideration, the smaller the differences in the performances of funds become. For the one‐year period, the difference in the performance of the best and worst‐performing fund was approximately 27 percentage points; for the three‐year period more or less 20 percentage points and for the five‐year period about 19 percentage points. This reaffirms the notion that short‐term results should not be overemphasised. However, the differences are still large, showing how important it is that members should monitor the performance of their pension funds closely to determine whether or not the fund managers are performing satisfactorily. The difference between the best‐performing fund and the worst performer over a period of five years (more than 19 percentage points) translates into massive differences in the final level of pensions for members of the respective funds.

Compared with conservative risk portfolios, aggressive risk portfolios carry a higher risk. The best performer of aggressive risk portfolios outperformed the best performer of the conservative risk portfolio for all three periods (see Table 3). However, if the worst‐performing funds for the two categories are compared, the aggressive risk portfolios performed worse than conservative risk portfolios for all three time‐frames. The worst performer of aggressive risk portfolios had shown a very poor return relatively to the worst performer in the conservative risk portfolio for the one‐year period. The difference between the three‐ and five‐year periods is smaller, but still significant. Although the best‐performing quartile of aggressive funds has outperformed the conservative funds over all three time‐frames, the bottom quartile of conservative risk portfolios has outperformed the bottom quartile of aggressive risk portfolios.

In an effort to prevent a massive outflow of capital from South Africa and to protect the value of the local currency, the government sets limits to the amount of money that funds can invest overseas. These limitations can have an important effect on the final value of the pension or lump sum a person receives. If the rand continues to depreciate, it will erode the eventual purchasing power of pensions. To enable pensioners to maintain their living standards, it is therefore important that they be protected against depreciation of the rand. One way of doing this is to allow pension schemes to invest some of their funds overseas. Exchange control may, therefore, hamper the effectiveness of funds, and also violates the investment principle of not putting all one's eggs in one basket (South Africa, in this case). Unstable exchange rates also make it very difficult for fund managers to make good short‐term investment decisions, which once again emphasises the importance of a stable macroeconomic environment to the retirement sector.

The investment decisions of retirement funds are sometimes criticised for contributing towards a ‘paper chase’ on the JSE Securities Exchange instead of contributing towards socially desirable investment. Although retirement funds do indeed have a social responsibility, critics usually ignore the secondary spin‐offs of current investment decisions. The prime responsibility of retirement funds is to seek the highest returns for their members, while limiting the risks involved. If their social responsibilities are overemphasised, it may defeat the original intention of these funds to provide an adequate pension to members upon retirement. If funds cannot provide adequate pensions, it may create an additional burden to the government, as more people may eventually become dependent on social pensions and other social services. Furthermore, Table 2 shows that the share of investment by pension funds in equities and unit trust schemes decreased from 45,9 per cent in 1997 to 36,2 per cent in 2001. This trend does not support the ‘paper chase’ hypothesis.

5.4 Tax incentives and the taxation of the retirement industry

Most countries have implemented tax concessions or tax expenditures to induce people to make provision for retirement. (The general advantages and limitations of tax incentives are a comprehensive topic which is beyond the scope of this article.) In South Africa, tax incentives are applicable to both the occupational and voluntary pillars. Contributions to certain approved pension funds (excluding some provident funds) are deductible from gross income, the maximum deduction allowed being the largest of either R1 750 or 7,5 per cent of a person's earnings during a financial year (SARS, 1999:8). This policy corresponds with the recommendation of the Katz Commission (1995:64) that the overall contribution rate that should be deductible must be limited to 22,5 per cent (7,5 per cent per employee and 15 per cent per employer).

According to an estimate done for the Mouton Committee (1992) for an unspecified year, tax expenditure on pensions already amounted to R5 billion at that stage (Katz Commission, 1995:203). Tax incentives can be criticised not only on efficiency grounds, but also on equity grounds, because they mainly benefit higher income groups. According to the cash‐flow principle of taxation, however, tax allowances for pensions merely represent deferred taxation. This is, however, only true if the full pension, including all lump sums, is fully taxed when it becomes due. The present system does not tax the full lump sum of pensioners and as a result violates this principle. As the Mouton Committee (1992:9–10) warned, exemptions and concessions relating to lump sum benefits may result in a significant portion of the deferred tax never actually being collected. The tax‐free component of retirement income is calculated with the aid of formulas that take into account the number of years worked and the ‘highest average salary actually earned during any five consecutive years in the service of the employer by whom [the individual] was employed during his membership of the fund’ (Downie, 2002:8.3).

This was one of the reasons why the Third Interim Report of the Katz Commission (1995:60) recommended the taxation of the retirement industry, which was previously excluded from the tax base. The purpose of this new tax would be to:

1.	Protect the integrity of the tax system
2.	Promote uniformity
3.	Limit the scope for tax arbitrage

The Katz Commission (1995:58) also argued that the previous system was regressive, because high‐income groups with their higher marginal tax rates benefited most. Retirement funds in South Africa were previously taxed on the ‘Exempt, Exempt, Taxed’ (EET) basis – the first ‘E’ referring to the exemption from tax of contributions, the second ‘E’ to the exemption of fund income, and the ‘T’ referring to the taxation when benefits are received by the insured (Katz Commission, 1995:57).

From 1 March 1996, a 17 per cent tax was levied on the gross interest and net rental returns of pension funds, provident funds and retirement annuity funds (National Treasury, 1996:5.5). This changed the taxation of the South African pension system to an ETT (‘Exempt, Taxed, Taxed’) basis, as the returns of the fund and benefits are now taxed (Katz Commission, 1995:62). On 1 March 1998, the tax was increased to 25 per cent (Manuel, 1998:24). Although this tax addresses certain inefficiencies in the tax system, it met with severe criticism due to its negative impact on the returns of pension funds and, as a result, on the levels of pensions. Furthermore, some critics of the tax argued that it may discourage people to provide for retirement. The second ‘T’ and third ‘T’ also imply a double taxation of retirement funds, if the partly exempted lump sums are ignored. According to the Registrar of Pension Funds (1997:16), this tax increased general administration expenses (including this tax) by 59 per cent (or R1,1 billion) between 1996 and 1997. (This figure does not include the additional increase of 8 percentage points in the tax from 1 March 1998.) This tax needs to be researched further to ensure that it does not have a negative impact on saving for retirement. If more people become dependent on social pensions and social services as a result of this tax, it may create an additional burden to the fiscal authorities. Moreover, it may also limit people's ability to maintain their living standards after retirement.

5.5 Freedom to choose, mobility, specific needs and transparency

Unfortunately, most employees in the formal sector of the economy do not have the opportunity to select an occupational pension fund of their own choice. People cannot switch between funds of their own accord and usually have to change jobs to change their pension fund. This lack of choice of, and mobility between retirement funds are limitations of the occupational pension system in South Africa. Individuals are usually not granted the opportunity to choose a fund that will serve their interests best and are obliged to stay with a pension fund, even if its overall performance is poor. In the past, retirement funds were also rigid and did not cater for specific needs. Fortunately, this has now changed. Young people may, for instance, prefer higher‐risk investments with possible higher returns, while older persons may be more risk averse and more concerned with capital retention. However, catering for the specific needs of individuals may have administrative cost implications. Freedom to choose one's preferred fund and to shift accumulated funds to a new retirement fund with a better investment performance could also promote competition, and discipline fund managers to keep costs low and make sound investment decisions.

Another problem encountered by many members of retirement funds is a lack of transparency, which often causes resentment against paternalistic attitudes during the decision‐making process (Smith Committee, 1995:21). An amendment in 1996 of the Pension Fund Act of 1956 provides for a Board of Trustees for each pension fund, consisting of the necessary stakeholders who must protect the interests of members. Fifty per cent of these stakeholders must be elected by the members of the fund. In addition, people are also not always adequately informed about the performance of their pension fund. One way of addressing this problem would be to provide members with regular statements (e.g. quarterly or annually) of the actuarial value of their fund or their personal account (in the case of defined contributions schemes), and of the administrative costs incurred. Greater transparency may force fund administrators to improve their efficiency (e.g. cut administrative costs) and to be careful with their investment decisions. Ways of promoting competition between funds receive further attention in Section 6.

5.6 The withdrawal of funds

Another important limitation of the present occupational pension system is that individuals are allowed to withdraw their accumulated funds when they are retrenched or change jobs. Although one could argue that people should have access to their accumulated funds when the need arises, this may result in individuals having an insufficient pension when they retire. In 2001, approximately 77 per cent of the total benefits paid by all pension funds went towards lump sum payments (see Table 1 in Section 5.1). In the same year, only about 23 per cent of lump sum payments were spent at retirement or at the death of the insured. Approximately 72 per cent of lump sum payments, or 56 per cent of the total benefits of pension funds, were paid out upon resignation or termination of membership. The application of these funds is left to the discretion of individuals, who may or may not decide to transfer the funds to alternative pension options. According to a sample drawn for the Mouton Committee (1992:145), fewer than 10 per cent of individuals who leave a pension fund before reaching the age of retirement actually transfer their accumulated funds to another pension fund. It is, of course, possible that they invest these funds in other forms, which could benefit them when they retire. Poor investment performances by pension funds would also discourage people from reinvesting funds with pension or provident funds. Many people might, however, succumb to the temptation to use these funds for consumption expenditure, such as an overseas trip or a new car.

The Smith Committee (1995:51) concluded that many pension funds still penalise their members when they withdraw funds due to a job change or retrenchment, which exacerbates the problem. Members may, for example, receive only their own contributions plus a very low rate of return on their accumulated funds. They often have to forfeit the contributions made by their employers and the largest share of the returns on their accumulated funds. Some funds, however, allow members to withdraw the full accumulated amount when they change jobs.

The implications of these penalties and early withdrawals could be that individuals end up with an insufficient pension when they retire. An effort by the previous South African government to rectify this situation through the Preservation of Pension Interest Bill failed miserably, mainly because black unions viewed it as an attempt to prevent them from gaining access to their money (Kruger, 1992:213), which of course was precisely the objective. The accumulated pension funds often serve as a safety net to members during periods of unemployment. It is difficult to convince people of the importance of preserving their accumulated funds for retirement if they are in dire straits due to having lost their job. If one has to choose between food now or in the future, present consumption will always be preferred. The high levels of unemployment in South Africa and the limited benefits granted by the Unemployment Insurance Fund (UIF) compelled the Smith Committee (1995:48) to propose that people should have access to their accumulated pension funds when they become unemployed. The Smith Committee (1995:48) felt that ‘this makes compulsory preservation of total benefits inappropriate now and for some time to come’. However, allowing accumulated pension funds to serve as a general social safety net to members during times of crisis will inevitably be at the expense of the pension's final value, unless people increase their marginal propensity to save once their situation improves. Possible ways to rectify this problem may be to improve the benefits provided by the UIF (although this may imply higher contribution rates by employees and employers, and even by the government), or to force persons to make larger contributions towards a pension when they are employed again. These issues receive further attention in Section 6.

There are also other reasons why unions resisted the government's proposal to protect pensions. One reason was that social pensions (see Section 4.2) are subject to a means test, which implies that anybody who receives an occupational pension (even at relatively low levels) may be partly or totally excluded from the benefits of social pensions, due to the limit on free income granted by the means test (i.e. the amount of income an individual may earn and still qualify for the full social pension). Persons with a relatively low occupational pension may, therefore, end up with a retirement income that falls below the level of social pensions (Kruger, 1992:213). The resistance of unions against the proposed legislation eventually forced the government to withdraw the Preservation of Pension Interest Bill in its entirety (Van der Berg, 1994:19). One way of addressing the problem may be to increase the free income level of the means test, and in this way reward those persons who have made provision for their retirement.

The fact that members of provident funds are allowed to withdraw the full amount of their accumulated funds as a lump sum when they retire, may also create a problem. There is no guarantee that some members will not squander their assets and eventually become dependent on government support. Tax incentives, such as an increase of the tax on a lump sum, could be implemented to serve as a disincentive to take a lump sum payment. This may also be in accordance with the cash‐flow principle of taxation. On the other hand, the exemption of a part of a lump sum may alleviate the burden of double taxation created by the ‘ETT’ principle (see Section 5.4). The taxation of the full lump sum will, however, penalise individuals who act responsibly with their lump sum payments, for example those who reinvest it to maintain their living standards and to leave a bequest to their heirs.

5.7 The levels of pensions

The levels of occupational pensions in South Africa are still relatively low. For example, 40 per cent of the pensions paid in 1995 were less than R410,00 per month, which is equal to the level of social pensions in that year (Smith Committee, 1995:51). One of the reasons for these low levels of pensions is a lack of indexation, which resulted in the annual adjustments in pensions often being below the rate of inflation. Furthermore, the period of membership of a retirement fund in South Africa is often too short to build up sufficient accumulated funds for retirement. More than half of the present pension funds were only established after 1970. A further reason was suggested in the previous section, namely that only a very limited number of individuals transfer their accumulated funds to other pension funds when they change jobs or are retrenched. Actuarial surpluses may also contribute to this problem.

5.8 Actuarial surpluses

In the past, there was a tendency not to transfer the investment reserve portion of the total actuarial reserve to the receivers of pensions, but to keep it as a reserve that could eventually be transferred to employer companies. During the shift from pension funds to provident funds, which became very popular during the last decade or so, some employer companies refrained from transferring a substantial part of the reserves to the new funds. One notorious case was Sentrachem, which retained 40 per cent of the assets of employees' pension fund and did not transfer it to the new provident fund (Sake Beeld, 2000). One of the negative consequences of such a practice is that members might not benefit from surpluses at all.

The Pension Funds Second Amendment Act, No. 39 of 2001 (South Africa, 2001a) was implemented to rectify some of the problems encountered. According to this law, the actuarial surplus of a fund is the difference between:

1.	‘the value that the valuator has placed on the assets of the fund less any credit balances in the member and employer surplus accounts; and
2.	the value that the valuator has placed on the liabilities of the fund in respect of pensionable service accrued by members prior to the valuation date, together with the value of the contingency reserve accounts which are established’ (Downie, 2002:7.5).

According to this Act, the actuarial surplus belongs to the fund. Once this actuarial surplus has been allocated to members or employers through the member surplus account or the employer surplus account, they acquire rights over these surpluses. Employers may only benefit from a surplus in their employer account (e.g. through a contribution holiday), while members must enjoy the benefits of any surplus on member accounts. If the board of a fund decides to apportion the actuarial surplus between employers and employees, it must inform the Registrar of Pension Funds of its intention within 18 months after the date of implementation (Downie, 2002:7.29). Surpluses in the member accounts may, according to this Act, only be used for improving the benefits of members and of previous members, decreasing the current contributions of members, or reducing expenses that would have decreased members' contributions which would be invested (Downie, 2002:7.30). These amendments should protect the rights of members and employers and prevent any further abuses of the system. Furthermore, it should also enable the Registrar of Pension Funds to perform its regulatory function satisfactorily.

5.9 Employees in small industries

The Labour Relations Act provides for pension funds for specific industries, but smaller businesses are usually not members of such a fund. The Smith Committee (1995:49) recommended that unions, employers and employees should be encouraged to start additional industrial or umbrella pension funds to enable small‐scale employers and their employees to benefit from economies of scale. This issue is dealt with in Section 6.

5.10 The age of retirement and benefits to dependants

Although people usually qualify for pensions at the age of 65, allowance is often made for early retirement. However, individuals are usually penalised if they retire early, qualifying for lower benefits than would have been the case if they had retired at the prescribed age. Early retirement has become a worldwide phenomenon, due to changes in labour markets and efforts to provide younger people with job opportunities. Relatively high unemployment levels could therefore promote early retirement. In South Africa, affirmative action has also forced many white people to select this option.

HIV/Aids may also have serious implications for retirement funds, as members may die before they reach the age of retirement. Dependants of such members may thus qualify for benefits at an earlier stage and for longer periods than would have been the case in the absence of the disease. This may have a negative impact on the financial viability of pension funds, but not so for provident funds, unless death benefits are of a defined benefits nature. If benefits are determined by defined contributions, dependants will have to make do with smaller benefits, as the deceased members would have contributed to provident funds for a relatively shorter period.

5.11 A brief evaluation

The number of occupational pension funds in South Africa increased sharply in past decades, but has recently decreased, mainly due to the cancellation, amalgamation and conversion of some pension funds. The numbers of members and beneficiaries are also rising, but there is still a substantial number of people who are not covered by occupational pensions, such as persons employed in industries without pension schemes, self‐employed individuals, lower income groups, those employed in the informal sector, and the unemployed. Those who are excluded from coverage must use their initiative to make provision for retirement through the voluntary saving pillar and, if they fail to do so, they may eventually become dependent on social pensions for survival.

Membership of a pension fund is not always mandatory, and people who change jobs or are retrenched often use accumulated funds for other purposes (e.g. as a safety net), rather than reinvesting it for retirement purposes. The system also lacks competition, as individuals are not free to choose a fund that suits their needs best. With regard to the performance of pension funds, the degree of transparency needs to be improved, particularly because it is often difficult to identify the administrative costs involved. The double taxation of pension funds may also hamper the performance of funds and may eventually serve as a disincentive to participation, leading to lower levels of pensions. Moreover, HIV/Aids may have a negative impact on pension funds, as the dependants of members may qualify for benefits far sooner but at lower levels than would otherwise have been the case. Given the large number of people who are not members of a pension fund, one could argue that South Africa needs some kind of a mandatory national pension scheme. In Section 6, an alternative approach to such a scheme is proposed.

6 RECOMMENDATIONS

This article concludes with two recommendations which, if implemented, can improve the coverage provided by occupational pensions and maintain accumulated capital.

6.1 The implementation of private pension fund administrators

To improve the coverage of occupational pensions it is proposed that membership should be made mandatory for all full‐time and partially employed people earning an income above a certain minimum level, which could also take the size of households into account. As membership of occupational pensions is not open to everyone, the government will have to initiate new kinds of private pension funds. This could be done by making available a number of licences to private pension fund administrators (PPFAs). The most suitable bidders will each receive a licence that will enable them to start a new fund. Such an approach corresponds with the Chilean Administradoras de Fondos de Pensiones (AFP), which are private institutions responsible for the administration of pension funds (SPFA, 1995:17). These PPFAs are not meant to replace existing occupational pension funds but, among other things, to broaden individuals' choice. Members of other pension funds should therefore be allowed, if they wish, to convert their membership to a PPFA that suits their needs. Such an approach should increase competition between funds and result in efficiency gains.

PPFAs will also provide self‐employed persons and persons in the informal sector with the opportunity to become members, and could also afford lower income groups the opportunity to make small and irregular contributions. To promote participation, additional tax incentives may be considered. Employers of domestic workers may, for example, be allowed to subtract part, or the full amount of their contributions to a pension fund that benefits their employees, from their own taxable income. Regulation of the system may be left to the Registrar of Pension Funds, who has the necessary know‐how and capacity to oversee the performances of funds. The Registrar must protect members against the default of a PPFA and ensure that investment occurs according to the rules. If a PPFA does not perform well, it may lose its licence, which then becomes available to new bidders. To promote economies of scale, the number of PPFAs should be limited.

The benefits of PPFAs thus include the extension of coverage and the introduction of more competition. This may improve efficiency, provided that administrative costs are kept at relatively low levels. Diamond (1996:95) concluded that the administrative costs of privately administered funds may be substantially higher than those of a well‐regulated and unified public‐managed system. However, a mandatory national public pension system may not be very attractive to a developing country such as South Africa, with its high levels of unemployment and poverty. Furthermore, a large number of formal sector employees are already members of occupational pension funds. If such a fund is to be a pay‐as‐you‐go system, it may encounter the same problems experienced in other countries if demographics change. PPFAs may not only represent an improvement in occupational pensions, but could eventually also benefit the redistributional pillar or social pensions, as some people who previously depended on the government may now provide in their own needs. It may, however, require a more liberal implementation of the means test to ensure that there is an incentive for individuals to participate.

6.2 Maintaining accumulated funds

People should be obliged to maintain their accumulated funds for pension purposes. To ensure that the system is flexible and complies with people's needs, members may be allowed to withdraw part (e.g. up to a certain fixed percentage) of their accumulated funds, subject to certain criteria, such as unemployment and other life‐changing experiences. As mentioned earlier, these funds often serve as an important safety net. However, such an approach may require higher contributions later when an individual's position has improved, for example, by becoming employed again. A flexible system may also allow persons to become affiliated members. This implies that they do not lose their membership when they stop making contributions for bona fide reasons, and may start contributing again after an interruption. An alternative is to improve the UIF and increase the levels and periods for which benefits are paid. Insured individuals qualify for benefits up to 238 days (34 weeks), depending on the period of employment and contributions made (usually one day's benefits for every six days of work) (South Africa, 2001b). However, given the unemployment levels in South Africa, and possible disincentives on labour supply, such a system may not be viable.

Additional information

Notes on contributors

Theovander Merwe 1

Professor, Department of Economics, University of South Africa, Pretoria, South Africa. The author wishes to express his gratitude to Professor PJ Mohr and two anonymous referees for valuable comments made on an earlier draft of the article.

Notes

Professor, Department of Economics, University of South Africa, Pretoria, South Africa. The author wishes to express his gratitude to Professor PJ Mohr and two anonymous referees for valuable comments made on an earlier draft of the article.