The Adequacy of the Central Provident Fund in meeting post-retirement consumption expenditure across age and income groups

Transcription

1 The Adequacy of the Central Provident Fund in meeting post-retirement consumption expenditure across age and income groups Renbao Chen and Bibi Jean Christophe Nicolas Correspondence Address: Associate Professor Renbao CHEN Department of Finance and Accounting Graduate School of Business National University of Singapore 10 Kent Ridge Crescent Singapore Tel: (65) Fax: (65) Dated: July American Risk and Insurance Association Conference 2006 Dr. Renbao Chen currently he is Associate Professor in the Department of Finance and Accounting of the National University of Singapore. Mr. Bibi Jean Christophe Nicolas is studying his B.B.A. (Honours) degree in the Department of Finance and Accounting at the National University of Singapore. They wish to thank Dr. Angelique Chan W. M. for providing the data used in this study.

2 The Adequacy of the Central Provident Fund in meeting post-retirement consumption expenditure across age and income groups Abstract After 50 years since its creation, the Central Provident Fund (CPF) of Singapore stands as one of the most important pillars in the economic and political development of the nation. With the changing demographics towards a grayish society, questions arise as to its adequacy in supporting a particular lifestyle, with particular concerns given the rising medical care costs, and the shrinking family size. Our main findings are that CPF benefits are mostly inadequate for most consumption patterns under most conditions of earnings growth, across age and income groups. Withdrawals will further worsen the situation. Only subsistence consumption pattern is achievable under inflation rate below 3% for most cases. 1. INTRODUCTION Singapore has one of the world s fastest ageing populations and by 2030, 18% of the population is expected to be over 65 years old. In a speech by Prime Minister Lee Hsien Loong on September 25 th 2005, he mentioned the twin challenges of longer lifespan and smaller families, resulting in an ageing society and workforce. The changing demographics of Singapore towards an alarmingly ageing population, with low birth rate and low death rate, have prompted the review of several policies designed to sustain retirement income. This stresses the importance of the individual s savings to meet post-retirement consumption expenditure, with the issue being further aggravated by the rising costs of medical care. Given the paternalistic political philosophy of the Singapore government, an individual has to ensure, on his own, that his future income streams will be sufficient in meeting his postretirement consumption expenditure pattern. Despite the philosophy that one has to stand on one s own feet, the Singapore government, as a wise father, should conduct appropriate policies to guide the individual in sustaining his future needs. Such policies can take various forms such as policy reforms to fine tune the Central Provident Fund (CPF) of Singapore, or through informational campaigns to promote savings, purchase of life annuities or implementation and usage of reverse-mortgaging option. As we will review later, the pension scheme of Singapore is very different from the U.S or U.K pension schemes which are based on the Pay-As-You-Go system. In Singapore, the CPF is based on the defined contribution model and each Singaporean should be well informed and made responsible for his own retirement. 1.1 Objectives of this study This study has the following objectives. Firstly, the overall goal of the study is to examine the adequacy of the CPF benefits in meeting post-retirement consumption expenditure for different age and income groups in Singapore, using the accumulation and de-accumulation framework. This analysis will be done for the current retirement age of 62 1, and the expected retirement age of 65. Secondly, we wish to investigate the adequacy of the CPF benefits under different withdrawal rates 2. An individual is permitted to use the CPF as a tool to purchase houses and for educational purposes, amongst others. Thirdly, we allow for flexibility in the post-retirement consumption patterns, carrying out analysis based on current 1 Retirement Age Act. - The minimum retirement age for employees is prescribed to be 62 years of age. See Retirement Age (Prescribed Retirement Age) Regulations 1998 (S 591/98 w.e.f ). 2 Withdrawals from the CPF can be made under Central Provident Fund Act S15 and other schemes. 1

3 consumption expenditure, subsistence consumption expenditure, and consumption expenditure, using replacement rates of 40% and 60% of last drawn salary. Fourthly, special attention will be paid to expenses such as housing, medical fees or other major expenses as allowed by the various CPF schemes. Medical care and old age-related care costs have become a major concern in Singapore, given the changing population demographics. Hence, it is very important that individuals save enough for their own retirement and health costs, as less support will be available, with the shrinking family size. Fifthly, we also consider the option of reverse-mortgaging and the use of life annuity insurance to supplement CPF benefits in the retirement period. In addition, we will assess whether members will meet the CPF Minimum Sum at the age of 55, given that this is the earliest when they can made withdrawal under Central Provident Act S15. And lastly, we will carry out sensitivity analysis of the CPF benefits to initial salary, earnings growth rate and to retirement age. 1.2 Justifications for this study Through its various schemes, the CPF has enhanced the standard of living of the inhabitants, and contributed to the overall economic development of Singapore. The government is not ready to adopt a completely new form of pension scheme, and it has made this point clear in various press releases. Therefore, the adequacy of the CPF is a very important issue that needs to be addressed in the new millennium, given the graying of the population and the increasing cost of living. The adequacy test will allow members to be better prepared in their retirement period. Although it is argued that the CPF is not solely a retirement tool, the question of its adequacy still needs to be made known to the public. With this information, the government agencies or even private companies may supply new retirement products to meet up with the future consumption expenditure of Singaporeans. This study performs analysis across age and income groups, and incorporates the medical care expenditure in the consumption pattern. Moreover, it provides members with an appropriate judgment on the amount of life annuity insurance to purchase to sustain different lifestyles. 1.3 Organization of study The remainder of the thesis is set out as follows. The next section will be devoted to the discussion of past studies pertaining to the CPF and various categories of research that have been done. Our focus will be mostly on the tests of adequacy that researchers have attempted to carry out. Following that, we will list out the main aspects of the CPF as well as some observations made concerning residential property prices, amongst others. We also provide in-depth description of the situation in Then, we proceed with the methodology that we will use to carry out the various adequacy tests, and we present the results and findings. We conclude with a summary of the study and list out the possible limitations, while providing avenues for further research. 2. LITERATURE REVIEW Under the People s Action Party (PAP) regime, the CPF serves as both a macroeconomics and social tool. It has been used for their 100% Housing policies, where amendments have been made over the years to encourage the use of the CPF by members to buy housing or 2

4 upgrade their current homes. In the 1970s to early 1980s, it served to complement the budget deficits. At present time, with the multitude of schemes offered, one can use the CPF to pay for education fees at tertiary institutions (although the benefits are in the form of a loan that has to be repaid one year following graduation), invest in CPF approved stocks and pay for medical expenses. Asher (1995) finds that the CPF is designed in such a way to remain a competitive and affordable pension system. Until now, it seems that the main priority of the CPF has been housing, education or health needs rather than serving retirement purposes. More recently, CPF contribution rates have been revised downwards to support the competitiveness of the business arena in Singapore. Hence, it is felt that the adequacy of the CPF for retirement may be jeopardized at the expense of other needs. Many researchers in the past have studied the CPF from various perspectives, such as its effects on society, housings, savings, consumptions, financial markets and adequacy. The relationship between CPF and savings was studied by James and Vittas (1994); Datta and Shome (1981); Beck (1984); Wee and Han (1983); Asher (1985); Loh and Veal (1985), Lim et al (1988), Hoon and Teo (1992); and Hoon (1991). While earliest research found no effect of the CPF on savings, later studies, for example Lim et al (1998), found evidence of negative effects of CPF on private savings for the period Various researchers have investigated the effects of CPF on housing in Singapore since the creation of the Housing scheme. Lim (2001) demonstrated that the CPF scheme can distort an individual s consumption choices, and stimulate higher housing demand from Singaporeans than would otherwise be needed. In addition, a study by Phang and Wong (1997) showed that the timing of government policies relating to the availability of HDB or CPF housing financing has an impact on the prices of housing in Singapore. Ng (2000) looked at CPF and the financial markets and found that CPF is the biggest participant in the bond markets, but is a serious dead weight. In addition, he found that CPF makes little contribution to the equity market, despite liberalization of the CPF investment schemes. Studies on the attitudes towards CPF have been conducted by Fernandez (1994), Chen et al (1997) and Chan (1997), amongst others. They give the general idea that Singaporeans are satisfied with the present arrangement of the CPF and reject various forms of direct social assistance from the government. However, Chen (1997) found that members are not totally happy with the low rate of return of the CPF. From Chan (1997), among those with a CPF account, 38.1% of them think that they have adequate fund for old age, while 61.9% think otherwise. Furthermore, from Chan (1997), elderly Singaporeans with a CPF account do not appear to be relying on these savings for monthly living expenses. However, we interpret from these above findings that the members may not be fully aware of the inadequacy of the CPF in supporting their current consumption patterns and lifestyles. While they may currently reject any shift towards the Western pension system, they may think otherwise if they realize that they may well have to considerably downgrade their lifestyle in post-retirement period. As such, the question of adequacy is still very important. Asher (1991) argued that any set of social security arrangements should satisfy social adequacy in three areas, which are poverty alleviation, income maintenance and health care. He also commented that there are three general factors which affect the adequacy of the CPF. 3

5 These factors are (i) macroeconomics management, and international monetary and trading system, (ii) prudent management of the CPF balances and (iii) cultures and attitudes of Singaporeans. From his paper, he found that the present arrangements are not enough and the problem will only worsen with Singapore s affluence and demographics characteristics. Chen and Wong (1998) analyzed the adequacy of the CPF at individual level for a representative agent who starts work at age 25. In addition, they examined the adequacy of the CPF with respect to the age of retirement which is variable at 60, 62 and after 65 years old. Scenario analysis with salary at $800, $1600 and $2400 was used with CPF interest rate of 4%, 5% and 6%, and salary growth ranging from 4-8%. They found that the CPF benefits will be sufficient during the post-retirement period of 20 years for most members and under most conditions. Only those with low starting salary ($800) may have financial problems after retirement. While the study provides the backbone for the current paper, we make several changes to the way the CPF balance is accumulated, and the various adequacy tests used for different consumption patterns. Lim (2001) performed simulation study for age groups of 40-44, and at five monthly wage levels, with withdrawal rates of 45% and 60%. She found that for the 45% withdrawal rate, at least 60% of them will not meet the minimum sum of S$80,000 by When calculating the replacement rates of the CPF balances, Lim (2001) calculated the replacement rates of the annuity payments of the CPF balance for the age groups as a percentage of their last-drawn monthly salary, and the benchmark for adequacy set at twothirds of last-drawn monthly salary to maintain the standard of living. The best scenario result is only a 40% replacement rate for the age group of with a withdrawal rate of 45%. However, the area of investigation was only on the Minimum Sum balance and we felt that the whole CPF should be taken together to obtain a clearer picture. Furthermore, Ngee and Tsui (2003) assessed the adequacy of the CPF Minimum Sum. Unlike Chen (1998), they did not take the Minimum Sum as given, but instead worked backwards to find the minimum amount that will sustain a standard of living, measured however by current monthly annuity payout. Their results indicated that the actual Minimum Sum (as of year 2002) is inadequate in meeting future consumption needs of the female elderly. The inadequacy becomes more severe when medical expenses are set at higher growth rates. The methodology employed was a Monte Carlo simulation study, with three major components, which are (i) calibration of subsistence and medical expenses of the elderly, (ii) forecast of cohort survival probability by age and by sex, and (iii) generation of yield curves to discount the future cash flows. Thus it seems that the medical care cost is an important factor in determining the adequacy of the CPF. Ewing et al (2003) examined the time properties of medical net discount rates and provided evidence that the standard auto regression moving average forecasting model may be improved by modeling the time series volatility characteristics of the medical net discount rates. Gerdtham and Lothgren (2001) carried fundamental analysis on the determinants of medical care costs. However, given the unavailability of data on the medical care costs in Singapore, we have to resort to past studies, which used data from a survey carried out in 1995 and Ngee and Tsui (2005) provided a framework to assess the viability of a reverse mortgaging (RM) market in Singapore in providing retirement income for home-owners. By estimating the probability of loss for both the private and public suppliers, they found that a replacement rate of about 50% is likely to be achieved, but there will be a missing market if the target is 4

6 set at 70%. Furthermore, they pointed out that although most public housing homeowners are assets rich in housing equity, the value of their property is inadequate to meet a 70% replacement ratio. Finally, they suggested a transforming role for the Housing Development Board, from a public housing provider to a reverse mortgaging provider. To relate to the reverse mortgaging topic, a working paper by Mitchell and Piggott (2004), the authors evaluated the factors that are needed to implement a reverse mortgaging (RM) in Japan. They suggested that for a successful RM market, policies that exempt RM from capital gains, transaction tax and annuity income tax are required. In addition, they suggested housing reforms to enhance the information flows to provide all parties with information on housing trades, securitization of housing loans and lines of credit. Furthermore, improvements in financial markets, especially to the availability of long-term bonds, are quintessential to the smooth functioning of the RM market. Finally, they pointed out two factors that might slow down RM market, which are bequest motives and a need for precautionary savings. 3. CPF AND VARIOUS TRENDS This section gives a very brief overview of the CPF structure and operations, as well as some of the trends for withdrawals rates, residential housing evaluation and appreciation rates, and the cost of medical care for the elderly. Also, we perform some descriptive analysis on a survey conducted in 1999, with most of our analysis reported for the first time in this paper. 3.1 Structure of CPF CPF is a statutory board under the government of Singapore, and under the CPF scheme, mandatory contributions by employees (Singapore citizens and Permanent Residents) and employers are deposited into three accounts, namely, Ordinary Account, Special, Retirement and Medisave Accounts. The pension funds under the CPF scheme are financed from the compulsory contributions from both the employers and the employees. As such, the government does not draw on its reserves to pay for retirement benefits. Under the mandatory contribution scheme, the monthly contributions by both the employee and employer towards the CPF account would be accumulated over time. In addition, the latter will earn an annual market related interest rate, (12-month fixed deposit and month-end savings rates of the four major local banks and revised every three months). Savings from the Ordinary Account can be used to buy a home, pay for CPF insurance, investment and education purposes. The Special Account is mostly for investment in retirement-related financial products. And the last component, Medisave Account, is used for hospitalization (medical) expenses and approved medical insurance. 3 Under the CPF scheme, the CPF Board is the trustee of member s CPF savings, which invests the members contributions in government bonds and other unit trusts. This is to ensure a fair market return at minimal risk, given the low risk nature of bonds and other trusts. Furthermore, CPF earnings are not taxable 4 and the government guarantees the payment of CPF savings. This assurance from the government aids in improving the quality of CPF. 3 Information obtained mostly from the Central Provident Fund of Singapore Website Retrieved on 25 th September 2005 from 4 In 2005, the tax range for the individual was between 3% - 28%. With an extensive use of tax incentives for business and social goals for the individuals, Singapore does not tax capital gain. 5

7 Plans offered include withdrawals by the member for retirement, permanent disablement, home ownership, health care and asset enhancement, amongst others. 3.2 Operations of CPF Our focus will be on Table A of the contribution rates which concerns the private sector employees, government non-pensionable employees, non-pensionable employees in statutory bodies and aided schools and Singapore permanent resident (SPR) employees from their third year onward. Table 1 lists the current rates, while the long term target rates are spelled out in Table 2. Contribution By Employer Table 1: Current contribution rates Contribution Total By Employee Contribution Ordinary Account(%) Employee Age Credited Into Special (years) (% of wage) (% of wage) (% of wage) Account (%) 35 & below 13.0% 20.0% 33.0% 22.0% 5.0% 6.0% Medisave Account (%) Above % 20.0% 33.0% 20.0% 6.0% 7.0% Above % 20.0% 33.0% 18.0% 7.0% 8.0% Above % 19.0% 30.0% 15.0% 7.0% 8.0% Above % 12.5% 18.5% 10.5% 0.0% 8.0% Above % 7.5% 11.0% 50.0% 0.0% 8.5% Above % 5.0% 8.5% 0.0% 0.0% 8.5% Source: CPF Board Table 2: Long-term target CPF contribution rates Age Total Employee Employer 50 and Below 30-36% 20% % Above 50 to % 18% 6-12% Above 55 to % 12.50% 6% Above 60 to 65 11% 7.50% 3.50% Above % 5% 3.50% Source: Ministerial Statement, PM Goh Chok Tong, 28 th August 2003 The salary ceiling for the workers in the private sector is now at $5000 at year Withdrawal can be made at age of 55 in a lump sum after setting aside the CPF Minimum Sum in the retirement account. The CPF Minimum Sum can be used to either buy a life annuity from a participating company, placed in deposit with a participating bank or left in your retirement account with the CPF Board. The Minimum Sum is currently set at $90000 as of 2003, and will be increased to $ gradually by Currently, 50% of your retirement account can be withdrawn even if you fail to meet the Minimum Sum. However, this percentage will be reduced gradually to zero by As at 2005, the interest rate in the ordinary account was 2.5% p.a. and an additional 1.5% for Medisave, Special and Retirement accounts. 3.3 Annual average withdrawal rate Table 3 shows the total housing withdrawals and the total withdrawals under section 15 and other schemes as a percentage of the annual contributions. 6

8 Total Contributions Table 3: Withdrawal rates as a % of Total Annual Contributions Withdrawal under public housing scheme Withdrawal under residential properties scheme Total housing withdrawals Total Withdrawal under Section 15 & CPF schemes Total housing withdrawal as a % of Total Contribution Total Annual Withdrawal as a % of Total contributions Year ($m) ($m) ($m) ($m) % % ,623 5,256 3,261 8,517-58% ,873 5,933 3,114 9,047 11,457 57% 72% ,999 8,505 3,448 11,953 13,610 75% 85% ,826 10,235 5,486 15,721 12, % 100% ,093 8,208 4,238 12,446 14,556 88% 103% ,322 8,331 3,677 12,008 18,860 66% 103% ,166 8,220 3,694 11,914 14,821 74% 92% ,870 6,850 3,039 9,889 11,817 62% 74% ,320 6,792 3,246 10,038 10,310 66% 67% Total housing withdrawals vary between 57% and 123% of annual contributions, confirming the high withdrawal pattern from the CPF. Over the period from 1996 to 2004, the mean annual percentage is 74%. Data collected are from CPF annual reports for the period from 1996 to Residential property valuation and appreciation rates Table 4: Valuation by Types of Property, 4th Quarter Room 2-Room 3-Room 4-Room 5-Room Executive Mean $73,500 $111,467 $164,236 $244,323 $322,235 $380,068 Table 4 shows the residential property value in Singapore for the 4 th quarter of Data is collected from the Housing Development Board (HDB) website. With most Singaporeans staying in 3-room apartment and above, property is likely to be more than $160,000 in value. In Tables 5A and 5B, we compute the property appreciation rates from year 1990 to The appreciation has been in double-digit number in the period from , and slowly picking up again as from 2001 onwards, after the Asian Financial Crisis. The figures give a rough idea of property value and appreciation rate to use when deriving the annuity income from using these properties as collaterals in reverse mortgaging contracts. Table 5A: Yearly appreciation rate Table 5B: Period appreciation rate Year Index, Base 1998 Rate % % % % % % % % % % % % % % % % % 7

9 % 3.5 Medical care costs for the elderly Ngee and Tsui (2005) use A. Chan s longitudinal data and calculate that average annual medical care costs in the post-retirement period for the year Table 6 shows the annual value and monthly value of medical care costs for the different age groups, estimated at age 62. Table 6: Medical expenditure ($) Age groups Annual Mean* Monthly Mean* $1,643 $ $1,351 $ $1,110 $ $912 $ $750 $ $616 $ $507 $ $416 $35 Growth Rate: 4% *Figures are at PV at age 62 We will make use of these figures as monthly medical care costs at the 4% growth rate level when computing the total monthly consumption expenditure in the post-retirement period. The following analyses are conducted using the longitudinal data from the 1999 survey of Transitions in Health, Wealth, and Welfare of Elderly Singaporeans. 3.6 Type of dwelling and ownership of property It is estimated from the data that around 42% of elderly Singaporeans stayed in 3-room HDB flat, followed by 33% in 4-room HDB flat, while only 10% stay in 1-room HBD flat. This confirms Chen s (1997) and other researchers results that majority of Singaporeans stay in 3- room flats and above. Analyses of the data further indicate that most elderly are owners of their residence as 22% is owned by respondents, while 29% is in joint ownership with spouse. 3.7 Gross monthly income, in 1999 Despite the high percentage of elderly who are not working, it should however be noted that, among the Chinese male elderly population, most of them fall in the income group of $1500- $1999, while Malay males in the income group of $500-$999, Indian males in $1000-$1499 and others in $500-$999. For analysis on female gender, most Chinese and Malays earn in the income group of $500-$999, while Indians in the income group $1000-$ Problems foreseen after retirement, money in CPF and lump sum Table 7 gives a rough idea of the problems foreseen after retirement. There is an equal percentage of elderly who foresees no problem and those who foresee financial problems, followed by those who foresee boredom in retirement. However, 73% refused to reply, did not know or did not apply to that question. We feel here that there might be another problem of short-sightedness among the elderly in Singapore. 8

10 Table 7: Problems foreseen after retirement, 1999 Types of problems Gender % of elderly population No problem Male 7.18% Female 2.05% Financial problem Male 7.17% Female 2.09% Boredom Male 4.51% Female 0.81% Family problem Male 0.29% Female - Loss of authority Male 0.92% Female 0.33% Become dependent Male 0.72% Female 0.14% Others Male 0.13% Female 0.06% Health problem Male 0.43% Female 0.14% NA % In addition, 70% of elderly has had a CPF account, with female accounting for about two third. And among those with a CPF account, around 45% has made withdrawal. From these, we feel that the adequacy problem exists, but goes unnoticed by the elderly, either because they are unable to plan ahead in the future, or are relying on other sources of income. 3.9 Use of CPF for home, educational or medical expenses We now look at the percentage of the population who have used their CPF for the monthly mortgage or lump sum payment on flat or private residence, educational purpose of child/children and for medical expenses. About 30% of elderly across ethnic groups have used the CPF for mortgage purposes. This shows the high attractiveness of the CPF housing scheme. Chinese are the largest users with 22%, followed by Malay 5%. For educational purpose withdrawals, we note that an overall of 1% only has made use of that fund. However, the generalization might suffer from lack of data bias. For medical purposes withdrawals, less than 10% has made use of that fund. Detailed results are available at request Fund in CPF, ordinary, medisave and special accounts From Table 8 on Ordinary Account, across all age groups, the majority of elderly will either have none in CPF balance or less than $5000. Table 9 on Medisave Account shows that it is common across all age groups that the majority of elderly has less than $5000 in Medisave Account. And lastly, Table 10 shows that a similar conclusion is reached for the Special Account, where across age groups, most elderly falls in either no fund or less than $5000 left. This is quite alarming given that the objective of this fund is to provide them with adequate income during the post-retirement period. However, they may be getting their income from other sources. 9

11 Table 8: Amount in Ordinary Account Age group % of Elderly population % of age group None 9.19% 22.39% <$ % 17.86% $ $ % 2.48% $10000-$ % 3.17% $20000-$ % 0.76% $ % 1.06% $ % 0.85% $ % 0.21% Refused 0.16% 0.38% Can't remember 6.12% 14.92% NA 9.35% 22.80% None 7.73% 34.32% <$ % 23.21% $ % 0.29% $10000-$ % 0.68% Can't remember 3.08% 13.68% NA 6.26% 27.82% None 5.25% 30.49% <$ % 14.54% $ % 0.69% $10000-$ % 0.33% $20000-$ % 0.33% Can't remember 1.90% 11.05% NA 7.33% 42.57% None 2.97% 28.43% <$ % 7.89% Refused 0.09% 0.84% Can't remember 1.14% 10.94% NA 5.42% 51.89% None 2.88% 32.18% <$ % 7.24% $ % 0.27% Can't remember 0.58% 6.50% NA 4.82% 53.82% 85+ None 1.47% 28.22% <$ % 8.79% $10000-$ % 0.36% Can't remember 0.29% 5.52% NA 2.98% 57.12% 10

12 Age group Table 9: Amount in Medisave Account % of Elderly population % of age group None 0.37% 1.05% <$ % 34.39% $ $ % 5.39% $10000-$ % 10.80% $20000-$ % 2.58% $30000-$ % 0.21% $ % 0.42% Refused 0.08% 0.21% Can't remember 6.67% 18.70% NA 9.35% 26.24% None 0.23% 1.04% <$ % 43.89% $ $ % 4.61% $10000-$ % 7.64% $20000-$ % 0.50% Refused 0.11% 0.49% Can't remember 3.15% 14.01% NA 6.26% 27.82% None 0.07% 0.40% <$ % 37.26% $ $ % 3.09% $10000-$ % 1.86% $20000-$ % 0.33% $ % 0.33% Can't remember 2.44% 14.15% NA 7.33% 42.57% None 0.14% 1.33% <$ % 32.56% $ $ % 1.50% Refused 0.09% 0.84% Can't remember 1.24% 11.87% NA 5.42% 51.89% None 0.07% 0.83% <$ % 35.83% $ $ % 0.49% $10000-$ % 0.54% $20000-$ % 0.27% Can't remember 0.73% 8.21% NA 4.82% 53.82% None 0.04% 0.76% <$ % 32.86% $ $ % 0.28% Can't remember 0.47% 8.99% NA 2.98% 57.12% 11

13 Age group Table 10: Amount in Special Account % of Elderly population % of age group None 6.34% 17.78% <$ % 27.88% $5000-$ % 1.67% $10000-$ % 1.55% $20000-$ % 0.21% $30000-$ % 0.35% Refused 0.16% 0.44% Can't remember 8.51% 23.88% NA 9.35% 26.24% None 5.14% 22.84% <$ % 31.08% $5000-$ % 0.89% $10000-$ % 0.75% Refused 0.06% 0.25% Can't remember 3.69% 16.37% NA 6.26% 27.82% None 4.02% 23.33% <$ % 18.15% $5000-$ % 0.40% Can't remember 2.68% 15.55% NA 7.33% 42.57% None 2.07% 19.86% <$ % 15.84% Refused 0.09% 0.84% Can't remember 1.21% 11.57% NA 5.42% 51.89% None 2.46% 27.43% <$ % 9.99% $5000-$ % 0.27% $10000-$ % 0.34% Can't remember 0.73% 8.15% NA 4.82% 53.82% None 1.28% 24.49% <$ % 10.19% $10000-$ % 0.36% Can't remember 0.41% 7.84% NA 2.98% 57.12% 12

14 3.11 Life annuity It seems that the annuity option under the CPF is still not popular among the elderly, as only 16% have heard of it. Chinese has bought more annuity, followed by Indian, Malay, others and Eurasian. And overall, more males than females are buyers of life-annuity CPF Top-up, bequest motives Less than 10% of the elderly has their CPF top up by children. This is a relatively new option, and should become more popular in the future. The majority of top-up is less than $1000. And lastly, we note that about 30% of elderly plan to give their CPF account as bequest, while around 20% has no intention of doing so. Hence the assumption of no bequest motives may not longer hold, thus aggravating the adequacy of the CPF. 4.1 Data Source 4. DATA AND METHODOLOGY For the following adequacy tests, data is obtained from various sources. Current CPF balances, annual contribution, contribution rates for various age groups, CPF interest rates, information on Minimum Sum as well as other miscellaneous CPF statistics and the description of the schemes are obtained from the CPF Board and the CPF website. Data on earnings, household expenditure, earnings growth rates and other key statistics of Singapore is obtained from the Singapore Department of Statistics. All figures used are brought up to 2004/2005 levels. When data is not available, we use proxy and we explain the construction in the methodology section. We feel there is a need to do more micro survey on the pattern of elderly consumption and lifestyle for Singapore. 4.2 Methodology: Design Using the annuity theory, we construct actuarial models to calculate the monthly CPF benefits for the different age and income groups. The overriding goal is to examine the adequacy of the monthly CPF benefits in supporting a retiree s post-retirement consumption expenditure under different patterns. Our initial assumption is that the individual is expected to live for 20 years after retirement. However, this assumption will be relaxed for further analysis. Individuals are classified into various age groups: 20-24, 25-29, 30-34, 35-39, 40-44, 45-49, and These groups are further sub-divided into high, medium and low income categories. The methodology section is organized as follows. We start off with the computation of the future value of the CPF balance, using variables such as current CPF balance, wages and earnings growth rates, contribution rate, withdrawal rate and interest rates. Next, we will compute the value of the monthly benefits for a retirement period of 20 years, under various withdrawal rates. This 20-year post-retirement period will be later relaxed to allow for a post-retirement period of 20, 18, 43 and 30 years, given that the maximum life span is taken to be 105 years, under mortality tables. We then forecast the consumption expenditure under different lifestyles. Special consideration will be given to the treatment of medical fees as well as the option of reverse mortgaging. 13

15 4.2.1 Computing the future value of the CPF balance, FV CPF After meeting the CPF Minimum Sum and the required Medisave Minimum Sum, the earliest withdrawal can be made when the person reaches the age of 55. Thus, we will compute the future value of the CPF balance at age of 55, 62 and 65 years. However, for most of the analyses in the Results and Findings chapter, the balance at age 62 will be used given that this is the mandatory retirement age in Singapore, effective as of Speculative talks prompt that the retirement age will be revised upwards and the most plausible target that one can think of is 65 years. First of all, we estimate the future value of the CPF balance when the individual reaches age of 55, 62 and 65 under Equation 1. This is the sum of the future value (FV) of the current amount in CPF and FV of future contributions until retirement. Equation 1: FV CPF = FV IniCPF + FV Contr where; FV CPF = Value of Total CPF Savings at retirement FV IniCPF = Value of the current CPF balance at retirement FV Contr = Value of the contribution at retirement Apportioning into high, middle and low income groups As CPF data is usually confidential and not released to the public at the micro level, the CPF annual report only gives us an aggregate balance of the current CPF balance for its members. Hence, to obtain the mean balance for the individual in the respective age groups, the total aggregate balance is divided by the number of members in that age-group. This will give us a mean CPF balance which we assume for the middle income group. Furthermore, using a report on the occupational wages from Ministry of Manpower, we classify managers and professional as high income group, followed by tech and assoc prof, clerical workers, sales and services workers and production craftsmen as middle income group, and plant and machine operators and cleaners and laborers as low income group. Average wage, based on simple average methods, are calculated for the high, middle and low income groups, and the difference in percentage of the high and low income groups from the middle income group is computed. This ratio will serve as the basis for apportioning the current CPF balance into high and low income groups. Hence, while we derive an average for the representative person who will fall in the medium group using the CPF annual report, we apply the difference in percentage in wages to obtain a proxy for the high and low income groups. We feel this is appropriate since CPF contribution is a percentage of wages and salaries, and therefore the earnings differentials between high earners and low earners will reflect in the CPF balance for the different age groups. Allocation of CPF initial (current) balance to accounts Since different interest rates apply to the Ordinary Account (2.5%) and the Medisave, Special and Retirement Accounts (4%), it is wise to divide the balances obtained into their respective accounts. Of course, things would have been simpler if the data were released by the CPF board regarding the different accounts balance for each age group. 14

16 However, to proxy for such levels, we compute the ratio of the Ordinary Account and Medisave, Special and Retirement Accounts to the total balance on an aggregate level over a few periods, and apply this ratio to determine the level of the current CPF for each age group for each type of account. It should be noted that such methodology constitutes only an approximation method. One pitfall is that it is not fair to assume the same ratio for each age group. Short of other alternatives, it still serves as a good proxy for such a study. Past studies on CPF adequacy did not even attempt to separate into the different accounts, and thus applied a constant interest rate of 4%, which is even more misleading. Details are available at request. The result shows that from September 2003 to March 2006, the ratio was roughly fifty-fifty. CPF interest rates CPF members receive a market-related interest rate, based on the 12-month fixed deposit and month-end savings rates of the major local banks, on their CPF savings. Under the CPF Act, the minimum interest rate that has to be paid on Ordinary Account is 2.5% annually, and the Medisave, Retirement and Special Accounts enjoy an additional 1.5%, bringing the interest rate to 4% on these accounts. It is expected that the rates will remain at such level for the foreseeable future. Computing future value of initial (current) CPF balance, FV IniCPF From the above sub-sections, the future value of the current balance can be calculated at age of 55, 62 and 65 by using a simple future value formula. Here, with the assumption of no withdrawal of the current balance, the future value is given by Equation 2. Equation 2: n n FVIniCPF = 0.5IniCPF x ( ) + 0.5IniCPF x ( ),where IniCPF: Current CPF balance allocated to high, middle, low income groups n : number of years to age of 55, 62 or 65. Calculating the Future Value of CPF contribution, FV Contr The next step is to calculate the future value of the CPF contributions from year 2004 until the member reaches age 55, 62 and 65 years old. The variables used in this computation include current earnings (wages and salaries), earnings growth rates, salary ceilings, contribution rates and CPF interest rates. We will, as in the first case, assume no withdrawal for housing, medical claims and education purposes. Later the no-withdrawal case will be relaxed in line with the most current and relevant trend of using the CPF money for purposes other than retirement. The premise assumption will be that wages and salaries are paid at the end of the month. A simplistic formula to compute the future value of the contributions is given by Equation 3. Equation 3: FV Contr = Sum of future value of (CPF Contribution rates x Min [Monthly Salary t, $4,500]) CPF contribution rates We use the middle range values from the long-term contribution rates from Table 2. Earnings and earnings growth rates 15

17 We use the average wages computed from previous sections. Each age group will be divided into high, medium and low income earners. For the earnings growth rates, we deem it more useful to apply a scenario analysis such that a range of growth rates can be used. As such, in line with the general earnings growth pattern observed from which is 4.5% and from which is 5.5%, we forecast that annual growth rate will be in the range of 4% to 8%. Hence computations and analyses will be carried out using 4%, 6% and 8% earnings growth rates. CPF salary ceilings The CPF contribution is limited by the CPF salary ceiling, which in 2004 was $5,500. In the computations, we assume that the salary ceiling will remain at $4,500 from year 2006 onwards. A downward revision will lead to the CPF being more inadequate. This is because current CPF savings (future consumption) are being sacrificed for current consumption. As mentioned previously, the CPF interest rates are expected to remain at 2.5% for the Ordinary Account and 4% for the Medisave, Retirement and Special Accounts. It is important to note is that the monthly contributions collected will be allocated into accounts paying different interest rates, and it has been a weakness in previous research to apply a standard interest rate of 4% on all monthly contributions to CPF. This paper appropriately computes the monthly contribution, allocating them to respective accounts following the current allocation percentage in force at the CPF Board. Appropriate CPF interest rates are then applied. Detailed calculations are available at request. Withdrawal Patterns We use varying degrees of withdrawals to proxy for withdrawals from the CPF account for housing purposes, medical cost expenditure, as well as financing education of children. However, we will still exclude the topping up option. This is because we feel that this will have offsetting effects on the adequacy of the CPF on an inter-generational basis. Past academics have used a withdrawal rate between 45% and 60% when testing the adequacy of the CPF. Some of them even felt that the withdrawal rate of 50% is optimistic for some members. Because of the uncertainty in the amount and time of withdrawal for housing, medical and educational purposes or other schemes as permitted under the CPF, it is not feasible to give an exact estimate of the final future value of the CPF balance at retirement. In this paper, we use a 40% and 55% withdrawal rates. The annual withdrawals as a percentage of total contributions are quite significant, and it justifies our rates of withdrawal used above. The pattern of medical care expenditure will be dealt in the section when calculating the present value of post retirement consumption expenditure Calculating monthly CPF benefits The summation of the initial balance and the value of future contribution will give us an idea of how much savings an individual member will have at age of 55, 62 and 65 as given by Equation 1 above. For ease of comparison, we will derive the monthly benefits from the balance computed above in FV CPF to obtain the monthly benefits that retirees will be enjoying from their CPF balance. We assume that the retirees will live 20 years after retirement in the basic case, allowing that figure to vary from 20 to 43 years if we assume maximum lifespan to be 105 under the mortality table. 16

18 Again, in previous research, a simple annuity formula was applied to get the monthly benefits. It can be misleading to apply a single interest rate to the balance. This is because the CPF only pays interest of 4% on the Minimum Sum balance and Medisave, Retirement and Special Accounts while paying 2.5% on the Ordinary Account. Hence, we will assume that the balance in the Ordinary Account will continue to enjoy a 2.5% interest, given that the money can be either left with the Board, invested in time deposit savings accounts in banks or used to buy annuities. The case for annuities will be discussed later Testing the adequacy of the monthly benefits Subsistence Consumption Pattern Elderly consumption expenditure consists mostly of subsistence consumption and healthcare consumption. The subsistence consumption refers to the consumption of basic necessities such as food, clothing, transportation and other miscellaneous expenditures. In 2002, the CPF Board set the minimum amount for subsistence living at $282 per month. From the 1998 Household Expenditure Survey, data indicates that most elderly were in the lowest quintile with an average expenditure of $378 per month. We will then a 1%, 3% and 5% inflation range to forecast the subsistence value of this consumption at retirement. Consumption pattern at 40% and 60% replacement rates However, it is very doubtful whether most Singaporeans would want to live at such low standard of living in post-retirement period. Thus it is worth considering the current or most appropriate replacement ratio. We use a replacement ratio of 40%, and 60% for analysis based on last drawn salary prior to retirement. Consumption based on current consumption pattern We again categorize consumption into high, medium and low to reflect the difference in consumption pattern between different classes of population. Data for consumption for age group of 60 is not available. 5 As such, we use the average monthly household expenditure by quintile. First of all, monthly consumption expenditure of household with 1 person is selected. Then, using monthly household expenditure consumption by quintile, we use the highest 20% to form our high consumption group, 3 rd quintile for the medium group and lowest 20% for the low consumption group. As such, the 1-person expenditure is increased/decreased by a difference factor (obtained from the difference between the quintile). Since the data is available for year 2003, we compute the monthly data to form yearly data, and then project it to year Using the 2004 figures, we compute the value of the yearly expenditure for each age group for the high/medium/low categories at time of retirement. Inflation rate is variable at 1%, 3% and 5% for the scenario analysis. The formula is as follows: AnnCon 62 = [MonthlyConsum 2003 * 12 * (1+Rate)] * (1+Rate) ^ #year to retirement AnnCon 62 = [YearlyConsum 2004 ] * (1+Rate) ^ #years to retirement (Appendix 10) Where MonthlyConsum 2003 High Medium Low 5 We have approached the department of statistics of Singapore, as well as NUS arts faculty economics department for the information. However, they informed us that data is usually not available at such level. An suggestion for future research would be to approach the CPF Board regarding the data. 17

19 Average Monthly Expenditure by quintile $6,160 $2,797 $1,259 Factor Difference MonthlyConsum 2003 $2,425 $1,101 $496 Reverse mortgaging and life annuity insurance For the reverse mortgaging option, we use one finding from Ngee Choon Chia and Albert T.C.Tsui (2005) to supplement CPF benefits in the post-retirement period. The issue of reverse mortgaging is not to be ignored. Singaporeans are asset rich, cash poor, and if their property is pledged as collateral for an annuity loan, it may help alleviate any problems of adequacy in old age. In that case, we assume that the supplier will be a government agency, such that the probability of loss will be lower from the lower costs of funds faced. For the case of life annuity insurance, the value of the insurance policy to be bought is calculated only after analyzing the deficits of CPF benefits over consumption expenditure. The individual will thus be able to optimize his expenditure, given that he can purchase according to his needs. 5. RESULTS AND FINDINGS Based on the methodology described in the previous chapter, we run simulations to test the various hypotheses set. We only present the main findings in this chapter. The Central Provident Fund (CPF) is a major pillar in the Singapore economy, and its importance to retirees is more than ever felt, given the rising cost of living and medical costs. Analyses will be carried across various age and income groups as mentioned earlier. 5.1 Hypothesis 1: CPF balance at age 55 versus minimum sum As the earliest withdrawal can be made at age of 55, we compute the CPF balance when the members will reach 55 years old. This will assess whether members are able to meet the Minimum Sum required. We assume that withdrawal rate from CPF is 0% on the future contribution in this particular setting. From our findings, when the rate of inflation is 1%, the majority of members will be able to meet the Minimum Sum required at age 55. Those in age group and low income group will have some difficulties to make up if earnings growth is at 4%. The failure to meet minimum balance increases for the age group 55-59, since both the middle and low income groups are not able to satisfy the requirements at all earnings growth rates. At the 3% inflation level, for the age group of and above faces more difficulties to meet the Minimum Sum requirements. At 5% inflation level, the case is seriously worsened as the middle and low income groups are not able to meet the Minimum Sum. One possible explanation for these findings is that, those in the age group 45 and above has already made withdrawals from CPF under various schemes such as housing and education. As such, their future monthly contributions are inadequate to bring their total CPF balance at the required level. Furthermore, for age group of 45 and below, they only face difficulties when inflation rate is 5%, and that is for the low income groups only. 18