Wednesday, April 27, 2011

When in doubt, turn to the data (Immigration)

Census 2010 was completed last year. A census is a rich data source to mine for information. Since the government mouthpiece media is so completely useless in reporting news that I actually want to read, that might affect my voting decision, I have to obtain the data sets and perform the analysis myself.

If you want something done...

Readers may consider this post and my next few posts a public sharing service. All data sets obtained from Statistics Singapore.

Data for the following 3 charts obtained from here; pages v and 3. Note that SingStat considers Singapore residents to be comprised of Singapore citizens AND permanent residents.

Needless to say, foreigners as a percentage of our population have skyrocketed in the last 6 years. As of last year, 1 out of every 3 persons (36%)  you see on the street is a foreigner.

The annualized growth rate for foreigners is even more astounding. See below.

Now let's turn to PRs and naturalized citizens.

SingStat does not distinguish between Singapore citizens and PRs in much of the data it reports, so it is difficult to get a handle on what characteristics differentiate citizens from PRs, such as age, race, sex, income level etc. I think it likely that it has this kind of data, but for whatever reason, it does not disclose this to the unwashed public. Too incendiary sensitive, I suppose.

Nonetheless, I can reconstruct data sets and make an educated guess at the age and sex profile of new PRs in the last 10 years by employing some actuarial tricks.

Data for this section of the post is obtained from Census 2000 Demographics Table 3 and Census 2010 Demographics Table 3. These tables break down the Singapore resident population into single years of age.

It is possible to use the Census 2000 data to project the expected breakdown of the Singapore resident population in 2010 by single years of age. To do this, we adjust the Census 2000 data by applying mortality rates, calculated from the 2005 complete life tables. I chose the 2005 tables for as an average as they fall midway between years 2000 and 2010, when the Census was held.

For example, a group of say, 21166 male babies aged between 0 and 1 year of age back in year 2000 should become a group of 21088 ten year olds due to the effects of mortality. However, there were in fact 24463 male ten year old Singapore residents in Census 2010.

I compared expected and actual numbers for both female and male Singapore residents for all ages starting from 0 to 89. I truncated the data beyond age 89 due to the small numbers of individuals involved and the difficulty in dealing with aggregated data.

The chart below shows (Actual - Expected Number of Singapore residents) plotted against age.

The difference in expected and actual numbers is accounted for by random fluctuations in mortality, citizens and existing PRs travelling abroad/returning to Singapore from abroad, and of course, new citizens and PRs.

In relation to travel for example, many students head to overseas universities to study. So, Singapore residents aged 8 to 15 in year 2000 are likely to show a larger than expected dip in numbers, more than mortality can solely account for, when they become the group of Singapore residents aged 18 to 25 in year 2010. They could be captured in the 2000 Census, but be abroad at the time of the 2010 Census.

Similarly, the group of Singapore residents aged 18 to 25 in year 2000 who were not captured in the 2000 Census may become a bigger group of residents aged 28 to 35 in year 2010 due to return from overseas studies.

However, we do not see this pattern in the data. There is a consistent positive divergence between actual and expected number of Singapore residents from age 10 all the way to age 50 (for males), and age 47 (for females).

What about Singapore citizens who head abroad to work and then subsequently return?This is unlikely to have great effect given that people are moving away and coming back all the time; the inflow and outflow should be matched, if not exactly equal. 

The likeliest assumption to account for most of the divergence between actual and expected numbers is the increase in new PRs. This is not unreasonable given that between 2000 and 2010, we know for a fact that we added about 254000 PRs to our population (Census data 2000 and 2010). In comparison, my chart above shows that the sum of (Actual - Expected) for ages 10 through 50 add up to 269000, a difference of only about 6% from 254000.

Thus, my interpretation of the data is this:

There have been approximately 250000 new PRs in the last 10 years. Considerably more new PRs in the last 10 years are female than male (we're talking ~28% more females than males).

Most new PRs in the last 10 years are today, aged in their late 20s to early 40s. There is a small bump in the LHS of the chart that indicates that significant numbers of new PRs are kids born in Singapore to foreigner or PR parents. Either that, or many Singapore citizens had their kids abroad and brought them back in the last 10 years (also possible).

There is a persistent trend of (small) negative numbers in the chart above the age of about 50 (males) and 47 (females). Given the consistent negative numbers across ages above these two ages, this is unlikely to be a random fluctuation in mortality. This indicates that a number of Singapore residents (most likely PRs) leave the population through emigration rather than mortality.

PRs aren't necessarily leaving only after the ages of 50 and 47. More generally, the rate of PR departure starts to exceed the rate of PR arrival at these ages. The mean age at which PRs give up their PR is likely to be lower.

Note that Singapore isn't necessarily retaining large numbers of PRs even though the negative numbers past the age of about 50 are small relative to the big positive surges in the ages of 20 through 50. It is important to note that the departing PRs and the surge in new PRs relate to different ages and hence cohorts. Our new PRs who are now aged in their 20s and 30s today could well leave en masse five, ten or twenty years from now.

I reiterate the assumptions: I am ignoring new citizenship take-up, and emigration/travel abroad/return from abroad by Singapore residents, and attributing the divergence between actual and expected numbers of Singapore residents to PRs alone. These are not trivial assumptions to make, but I believe that these are reasonable assumptions in light of the data I have.

And all that is certainly more information than you're ever going to be able to get out of the government.


Ponder Stibbons said...

Thanks for crunching the numbers.

Ray said...


Also, did you get the SingStat data for free? Last I check, it was behind a firewall.

mjuse said...

To Ray:

As far as I know, the SingStat data is free and in the public domain.

Chieh Schen Teng said...

it is good that there are others looking into this, as I have been concerned for the longest time, esp the impact in the middle.

sgcynic said...

Strange: the annualized growth rate of PRs was declining across 2000 to 2008 and it spiked in 2009. This coincided with the sharp faill in annualized growth rate of non-residents. Is that how creative accounting and implementation are executed?