I am responding here to Aidan Morrison's brilliant deep dive into the ABS population statistics entitled "The Missing Million: Is Australia's migration rate actually high?".
1. The ABS always strings together their data after series breaks. It is really bad in the economic data where the whole concept of what they are trying to measure can change (like the 1998 CPI revision). You could do a similar deep dive into every ABS dataset and finish with more questions than answers.
2. Many people who use the data (including most of those people in your example) know there is a series break in 2006. But as you have shown, it doesn’t really make much of a difference in the end. The divergence with the raw arrivals/departures data that started in the early 2000s remains. The new 12/16 rule applies symmetrically to those who leave and arrive, and although the ABS notes that it increases the NOM estimate during their 2003-06 test period, this was kind of the point. They were classifying many arrivals as short-term (i.e. not in ERP) when they stayed more than 12 out of 16 months, but not consecutively. This applies to much of the international student cohort, of which there are now over 600,000, compared to 200,000 in 2001.
3. My point is, the discrepancy you point out between your physically present population (PPP) and the estimated resident population (ERP) is not primarily because of this change in methodology in 2006. The difference is simply the number of residents currently abroad, short term, at any point in time, minus short term visitors (under the new definitions of short-term). The census also won’t catch the million residents who are abroad that night.
And while short term departures are growing, the number abroad at any point in time will grow as well.
4. The new methodology means that the ERP is higher than the old methodology because it makes sense to count people who reside in Australia more than 12 out of 16 months as residents who need the type of long-term facilities that residents need. In my view, ERP is the more important concept economically than physical bodies on the ground on any day. It also matches more closely the way these things are measured internationally.
5. So if the whole issue boils down to net short-term movements what is causing the change since 2001-02? From ABS 3401 we see that this is mostly from the rise in short-term resident departures for holidays!
Also, mostly these trips are for less than a couple of months.
We can see that the increasingly popular destinations are NZ (big boost in 2003-04), Indonesia (likely Bali) and the US. Less so to China or India, as many might expect.
anecdotes to back this up. And it’s now a hot button political issue that pensioners are getting the pension while on cruises and travelling abroad.
The data also shows that the number of people over 65 travelling abroad has tripled in the 10 years to 2015-16, with a relative decline in the share of short-term resident departures coming from people aged 30-60.
The other point to note is that both the baby boomers and their children (the baby-boomer echo) were, in the past ten years, at the stage of life where they are more likely to travel abroad.
We can see below that what data we have shows that people over age 60 have been a larger share of short-term departures recently, with people aged 30-60 being a declining share of the total.
6. Which brings us full circle. The 2006 methodology change meant that some inward immigration which was previously considered short-term, like international students, is now considered long term (i.e. residents), while there has been a rise in short-term departures from baby-boomers and their children holidaying in Bali!
Should we subtract retirees on holidays in Indonesia and on cruise ships from our resident population figures? I suspect not. Do their holidays mean that inbound residents residing more than 12 out of 16 months don't need to be considered in terms of housing and infrastructure needs? Probably not.
Overall, after a lot of thinking on this topic, I suspect that the ABS data, although imperfect, and although hiding a series break, is probably the ‘better’ metric to use in planning and economic analysis.