Appraising NUS study concluding CPF sufficient for young Singaporeans

I refer to the NUS Department of Economics study concluding that CPF is sufficient for the retirement needs of young Singaporeans joining the workforce today [1].

Adequate income replacement doesn’t mean adequate income
The study strives to show that pre-retirement earnings can be adequately replaced in retirement. But it says nothing about whether ours is one of high pre-retirement earnings replaced by high retirement income or low pre-retirement earnings replaced by low retirement income. Inadequate pre-retirement earnings, no matter how adequately replaced, will still be inadequate retirement income.

30th to 70th percentile covers only 40% of young cohort
The study concludes that CPF is sufficient for the retirement needs of the 30th to 70th percentile range of young workers today. The 30th to 70th percentile range encompasses only 40% of the young worker cohort which leaves us wondering how the other 60% will cope. Just because the 30th to 70th percentile range can cope doesn’t mean the 71st percentile can cope. The study emphases that the 30th, 50th and 70th percentile young couples must buy 3-room, 4-room and 5-room BTO flats respectively in order to have sufficient CPF for retirement. Beyond the 70th percentile however, couples could exceed the income ceiling and therefore not qualify for BTO flats. So the whole premise of couples buying BTO flats in order to have sufficient CPF falls apart for this group of people falling outside the 30th to 70th percentile range who would then be forced into the HDB resale market or even the condominium and executive condominium markets. The study shows that by buying just one BTO flat size bigger than recommended, retirement income falls to 56%, 58% and 48% respectively for the 30th, 50th and 70th percentile couples respectively which are dangerously close to the 53% minimum recommended by the World Bank. The fall will be even greater when couples are forced into the resale or condominium markets.

Possibly only half of middle income cohort covered
The study is modelled on the situation of a 30-year old male and a 28-year old female forming a family unit and purchasing a BTO flat because the median age of purchasing BTO flats is 30 for males and 28 for females. If the median age at which a male citizen buys a BTO flat is 30, it means that for the cohort of 30 year old males, only half of them would have bought a BTO flat, half of them would not have. As the latter half delay marriage, their incomes rise and they may end up not qualifying for a BTO flat so the whole premise of sufficient CPF based on purchasing BTO flats falls apart for them. As such, the study’s claim that the 30th to 70th percentile range of young workers has sufficient CPF to retire doesn’t refer to each and every one in that range but possibly only half of them. Thus, perhaps only half of the 40% middle income cohort or just 20% of the whole cohort is covered in the study. A more appropriate conclusion might have been that at least 20% of today’s young cohort has sufficient CPF for retirement.

The study can be improved by taking into account the marriage age distribution to ascertain for each income group, the percentage that would qualify for BTO flats by the time they get married. A rough calculation [2] shows for example that the 68% to 70% income group will hit the BTO income ceiling just two years after the assumed marriage ages of 28 and 30 respectively for bride and groom and by then, only 54% of males and 62% of females would have gotten married. In other words, for this income group, 46% of males and 38% of females will not qualify for BTO flats by the time they get married so the whole premise of sufficient CPF through purchase of BTO flats falls apart for them.

Counting rent
The study also claims that Singapore retirees own their homes unlike their Western counterparts and thus save on rent, so rent that our retirees save on is added to their retirement incomes to be comparable to Western nations. The following table [3] shows the owner occupation rates for several European countries:

2010 2009
Spain 85.00% 85.00%
Norway 85.00% 85.00%
Italy 80.00% 80.00%
Belgium 78.00% 78.00%
Ireland 74.50% 74.50%
Luxembourg 70.40%
USA 66.90% 67.40%
UK 66.40%
Sweden 66.60% 66.30%
Finland 59.00%
France 57.80% 57.80%
Austria 56.00% 56.00%
Netherlands 55.50%
Denmark 53.60% 53.70%
Germany 43.20% 43.20%

Countries like Spain and Norway have home ownership rates comparable to our 88.6%. It is unfair that we add rent to our retirement income but not them. France’s home ownership rate of 57.8% is actually closer to our rate than further from it. So a statement that says for example that Singaporean retirees don’t pay rent whereas French retirees do is 57.8 / 88.6 = 65% false and 35% true only. Of the countries listed, only Germany has a rate slightly further from ours than closer to ours. The assumption that retirees in the West pay rent while ours don’t is highly unsubstantiated.

Furthermore, retirees in the West, even if they were to rent, often has access to cheap social housing [4]:

Social housing as % total housing
Netherlands 32%
Austria 23%
Denmark 19%
UK 18%
Sweden 18%
France 17%
Finland 16%
Ireland 8.70%
Belgium 7%
Italy 5.30%
Germany 4.60%

In contrast, our 46,000 HDB rental flats is just 0.04% of our total housing stock. So if the study truly strives to be comparable, the rent to be added to Singapore retirement income should be cheap subsidised rents, not HDB open market rents.

Our high home ownership rate should also be taken with a pinch of salt. Firstly, our so-called home ownership is only 99-year leasehold whereas ownership in the West means owning the land on which the house is built. Furthermore, many Singaporeans are forced to sell their homes to fund retirement so much so that home ownership rates for retirees could be lower than national average. This is in contrast to the situation in the West where the older generation tends to have higher home ownership rates [5].

Retirees in Western nations are also often given free or subsidised medical treatment. Why does the study not consider deducting medical costs from our retirees’ retirement income to be comparable to Western nations? Why add rent to our retirement income in the name of comparability but not subtract medical costs? The study’s believability can be improved if retirement income adjustments comprise both positive and negative adjustments instead of just positive adjustments only.

Accounting for layoffs
Based on CPF contribution history from 1981 to 2011, the study concludes that those between 25 and 54 are employed 85% of the time and those between 55 and 65 are employed 78% of the time. It is presumptuous to conclude that just because CPF is not being contributed to, a person has been laid off. The person could have become a taxi driver, a private tutor or have taken up some other jobs that don’t pay CPF.

Long run inflation rate
The report uses a long run inflation rate of 1.8% based on the 20 year consumer price index (CPI) from 1991 to 2011. It’s strange that the study traces CPF contribution history all the way back to 1981 to account for layoffs, but only goes as far back as 1991 to calculate long run inflation rate. If we consider CPI from 1980 to 2011, the long run inflation rate is 2.0%, slightly higher than that used in the study.

Other considerations
The fact that over the years, CPF had to be tweaked again and again to meet rising retirement needs makes us wonder how it won’t be tweaked a few more times between now and 40 years later? So how can anyone guarantee that CPF as it is now, without any more tweaks along the way, would definitely meet retirement needs 40 years down the road? If NUS professors are so confident, perhaps they should put their money where their mouth is by guaranteeing our retirement needs 40 years down with their own money.


[2] First, approximately extract wage data from the study’s graphs. Then interpolate wages between the 50th and 70th percentiles to find wages for each percentile between 50 and 70. Add the corresponding percentile wages for males and females to get each percentile couples’ combined wage for all age combinations (males two years older than females as assumed in study). For each percentile, find the age combination at which the HDB income ceiling of $10,000 is reached. Next get the marriage age distribution from Singstats. Must combine Muslim and non-Muslim marriage data. The data is in 5-year intervals so must estimate for each year using cubic spline. Then convert the data to cumulative frequency. Finally, cross reference the age at which income ceiling is reached with the cumulative percentage of males and females that have gotten married by that age.

[3] European Mortgage Federation, Hypostat 2010, November 2011

[4] Housing Europe Review 2012, The nuts and bolts of European social housing systems

[5] CESifo Working Paper No. 3280, Residential mobility of the European elderly, Dec 2010
For this reason it is not surprising to find that very high proportions of elderly households own their home all over Europe (homeownership rate is above 70 percent for those aged 50-79 in most countries)


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: