Ministry of Manpower Statistics lacks Credibility

Someone alerted me to the Ministry of Manpower's Statistics on Employment recently and I took a good look at it. The first doubt I have in mind is related to the "differentials". 

Let me explain. Increment in Jobs data for "Local"  may be subjected to a few factors which was explained by MOM in this news article at Straits Time.

MOM clarified that the increase mentioned by Dr Chee is not the total number of new jobs taken up by locals last year. The increase is actually the difference between the total number of locals taking up and leaving jobs, for example due to retirement. This difference - the "net" number of new jobs taken by locals - was 700 last year.

Locals here is defined as Singapore citizens + Permanent Residents. What this means could be expressed in the following simple equation:

Increment in Jobs for Locals = Increase in Citizen workforce (ICW) + Increase in PR workforce (IPRW) - Death (D) - Retirement of older workers (ROW) - "those left workforce but not looking for new jobs" (LW) - Retrenched (R)

Where the variables are affected by:

ICW     - Directly influenced by those born 20 years ago

IPRW  - Directly due to New Permanent Residents issued (Net growth of PR+New Citizens)

ROW  - Directly influenced by those born 55 years ago

LW     - Reflected in Long Term Resident Unemployment

R       - Reflected in changes in Unemployment figures

For those who are well trained in statistics would just take MOM's explanation just as it is. BUT, if you want to really understand the relevance of the whole statistics, you will have to look at each and every component. I shall leave out Death (D) but examine the others. 

Most importantly, one way to observe whether such statistics are relevant or "good data sets" or not, we will have to examine the "differentials" by looking at the comparisons between past and forward data, if any. 

I shall compare the data between 2015 vs 2014. 

Please refer to the following data set which is from MOM website:

2015 is a "abnormal" whereby there is a DRASTIC drop of the increment of jobs for LOCALS, from 96,000 to 700, as compared to a slight drop of increment of jobs for Foreign Labour, from 34,000 to 31,600! 

Such imbalanced drop in the increment of jobs for Locals vs Foreign is pretty alarming but MOM tried to explain it away by using the logic that there was more retirement of older workers than new entrants to the workforce due to "aging population". 

However, aging population is NOT a new phenomenon. Every year, there will be people retiring from the workforce while there will be new entrants. The differentials or "net effect" of this is reflected by the net difference between the two. 

However, such differential cannot varies in such a wide variance! How to examine or determine such variance?

 

Now, before we look at the variance, we shall look at the unemployment rate. According to MOM's statistics, the Resident (equivalent to Locals) unemployment rate is pretty stable for the past few years. Specifically for 2014 and 2015, there is only a slight increase of 700 unemployed which constitute an insignificant 0.1% to the overall unemployment rate.   

 

Thus what does this say? The differential must be due solely to difference in new entrants vs retired workforce.

New Entrants to workforce comes from two sources. One, increase in PRs and two, due to natural births happening twenty over years ago.

For simplicity, we compare the numbers born in 1960 (the earliest recorded data on birth) vs 1994 ( taking 21 years old). The result is 61775-49554= 12221 for birth differential. What does this mean? It means that we have a deficit of about 12,000 jobs if we depended on our natural growth of workforce alone. i.e. a negative growth of jobs for Locals of 12K if there are no supplement of PRs.

Please also note that for the period of 1990 to 1995, the total birth were pretty constant without much variation. 

However, what is the supplement from PRs and New Citizens? 

The following is the old chart I have used before on this blog. We must note that we must use data from this chart by minus 1 year basis in order to reflect on 2015 growth of local JOBS. i.e. for 2015 increment of jobs due to increase of PRs, we should look at 2014 new PR granted by ICA. 

Consistently for the past few years, the increase in PRs is pretty constant, along with New citizens. Please note that we should not include the data on New Citizens because these new citizens are from the original PR stock. The net effect should only be reflected by the total increase of PRs. 

The consistent potential maximum contribution of increase in PR + New Citizens is less than 30k.  

What do all these imply? It actually means that the new entrants for the past few years should be pretty CONSTANT.

Even if we take the variation of retirement of old workforce (ROW) into consideration, the variance should not exceed 5,000!

So how could MOM statistics come up with such a statistics with such a wide variation while maintaining a pretty constant unemployment rate?


From the 12K deficits of jobs, we add 30K impact of increase in PRs, we will get a net effect of about 18K of increase in jobs for Locals! Do a plus minus 5K or even 10K for the variation due to birth rate variance in the 1950s or 1960s, at most we should get about 28K of new jobs! How could we ever get such a huge 96K increase in jobs in 2014 for locals?

Does it mean that those previously out of job market aka in long term unemployment have re-entered the job market? No. The Resident (or Local) Long Term Unemployment numbers are pretty constant for the past few years as well! (No drastic decrease).

Thus, to me, the sum doesn't add up.

There are two question marks here. If we assume normal distribution (a technical statistical assumption), it is totally IMPOSSIBLE to have 96K increment of jobs in 2014 (which equivalent to about more than 100K new jobs created) or even 37,900 increment of jobs for Locals in 2011 IF we do not have at least 100K of new PRs in 2014 or 50K of new PRs for 2011!

And it is impossible to have such a minimal increase of unemployed Locals if there are nearly 30K of new PRs in 2014 which presumably will get into the workforce. There should be at least 15K of increment of jobs, not 700!

This is why I conclude that MOM's statistics doesn't really make any sense nor tally to differential between births in different time frame and the increase in PRs. Thus, MOM's statistics lack Credibility and totally nonsense in my view. Especially for period from 2010 to 2014 whereby New PR issued was curbed at constant about 29K but they still report such a high number such as 96K for increment of jobs!

It is either ICA's statistics is erroneous or MOM's statistics is totally rubbish. There is no other way around it.

In short, when Birth rates are pretty constant in the short period of 5 years from 1990 to 1994 as well as 1960 to 1965, new PRs numbers for last few years are pretty stable, unemployment numbers stable, long term unemployment stable etc but only increment of jobs for Locals fluctuates with WIDE variance, there MUST BE something wrong with the statistics. Such inconsistency is a very telling sign of rubbish statistics in the making.

Goh Meng Seng