This post is a follow-on from my second post on migration, where I made the proposition that the true level-shift which warranted discussion was actually in international mobility, rather than Australia’s inward migration. I described conceptually how the definition of migration (which I’d discussed in detail in my first post) would inevitably lead to Australia registering a high migration rate while mobility was increasing globally, since we were naturally more likely to be a long-trip destination than a short-trip destination. I argued that this was likely to be linked to our cultural, political, economic, and overwhelmingly geographic circumstances more than policy, since we are competing with Asia for cheap holidays and business destinations, but could out-compete other Western Countries as an accessible, safe location to learn English and get a Western education, or base a back-packing adventure in the Pacific.

I used a cute analogy about an old bar/restaurant in a newly-gentrified area, that was losing drinkers of coffee and cocktails from their bar area to new hip start-ups, whilst attracting more seated diners, who tended to turn over faster, since they preferred to have drinks and desert elsewhere. My point was that by classifying only the seated diners as ‘patrons’, and insisting that this is the most important metric for tracking the business, one could be at once both technically right and stupidly unhelpful, since this might obscure other important trends in the data that could better point causes of the change, as well as challenges and opportunities moving forward.

In this post and some subsequent I’ll unpack some data to support these broad ideas, and expose some of the complexity in Australia’s migration and movement patterns which ought to drive policy-making decisions.

Stay long, but leave a lot

The following one shows clearly how Australia has become a place to arrive long-term, and leave from short-term, which is the core contention of my thesis. It also shows how Net Overseas Migration tends to very closely resemble the long-term movements, whereas the actual total movements net out to a far lower figure.

Figure 1: Net Movements diverged after 2001, with Short Term Movements becoming strongly negative, and long term movments more positive.

Up until about 2001 Short Term Movements netted out to zero. Long Term Movements roughly equalled Net Overseas Migration, which roughly equalled Net Movements. So what changed after about 2000? I’m arguing it’s the scale and pace of movements globally, which has served Australia a strange and unique hand, being hard to get to and from. So we’re great to visit long-term, and leave short term. Whilst we define our population as being here for the longish-term, our population will appear to grow, but we will have missed a large part of the story if we don’t also acknowledge that the population, whilst theoretically long-term, is increasingly “part-time”. This has led to a substantial gap opening up between the number of people we regard as being ‘resident’ and the number of people who are physically here.

Figure 2: The cumulative discrepancy between official NOM and the population actually in Australia has exploded

Forest for the trees

To put things into perspective, since 2000 there have been over 200 million arrivals, and over 200 million departures. The total difference between the two, or cumulative net arrivals, is just over 2 million during that period. About 1%. I think that’s worth just keeping in mind when we dive into this data, as it’s easy to miss the forest for the trees. For any and all definitions of migration, it is tracked through movements, which is two orders of magnitude larger. We’re already looking through a microscope at movements when we start to discuss migration.

In the following chart I suggest you select the view option that looks like two tabs above one another, and read off the precise numbers for these two series at any point along the line. The difference between these two numbers represents the total increase in Australia’s physically present population, since we started recording in 1976.

Figure 3: The amount of travel taking place has skyrocketed

But of course, it’s important to get a sense of how the movements actually happen, month to month:

Figure 4: Monthly Movements have risen, and are showing far greater variance

This series shows two things which the official Net Overseas Migration (NOM) figures fail to capture. The first is the rate of increase in movements, and the second is the tremendous increase in the volatility of travel, which shows increasingly seasonal patterns.

Speed of movement

Let’s start with just the rate of growth of movements. Adding the two together, and fitting an exponential model, we find a growth path that looks pretty impressive:

Figure 5: Movements Exhibit long-term growth about 6% per annum

However, it should be noted that the fit isn’t that great, with what looks like overestimation recently, and under estimation in the late 1990s. Taking a break in 2004, a tighter fit can be achieved, which curiously demonstrates that the growth rate has actually slowed in recent times.

Figure 6: Historically movements have grown at a staggering 7%

Figure 7: Recently growth in movements has cooled, to a smoking 5.4%

However, this hasn’t changed the most important conclusion, which invariably is that the growth rate is still extraordinarily high when compared to the growth of the population. Expressed here as an annual number of movements divided by the Estimated Resident Population, we can see that the rise in ‘mobility’ is still rising relentlessly.

Figure 8: Movements Per Capita demonstrate continuous growth

Officially, we’ve had more than one overseas movement per person since 2005. Whilst it’s important to consider the distribution of these movements amongst the different types of people coming in and out of the population, the potential for us to have a very significant ‘part-time’ population is clear. Whilst it’s tempting to simply say that this only reflects short-term visits, Figure 1 demonstrates how short-term and long-term visits are no longer separable in terms of their impact on migration. A significant and increasing fraction of our population is only here part-time, and attempts to get a clear grasp on our demographics that don’t appreciate that will be persistently miss their mark.

Seasonality of movement

This unremitting increase in mobility should be cause for consideration for policymakers and economists. The pace of movement is picking up, and this is likely to create pressures that are quite independent from the gradual accumulation of more ‘resident’ people. It is more like heat, or higher temperature in a physical system causing pressure, rather than compression due to the presence of more particles.

Perhaps one of the clearest ways in which this is likely to manifest itself in in seasonality. As can be seen from earlier charts, the dots are spreading out, and intra-year variance is increasing. The following graph shows net movements, (arrivals less departures) by month, for every year we have data on.

Figure 9: Net Movements by month show increasingly extreme seasonal fluctuations

The unmistakeable trend is that net seasonal fluctuations are increasing massively. In both July and December, Australia physically drains itself of more than 1% of its population. To see how that pans out cumulatively in any given individual year, I’ve replicated the previous December result as month zero, to track how the population actually fluctuates throughout the year.

Figure 10: Australia’s Physically Present Population peaks every February and October

This tells an important, and a very different story about Australia’s physically present population, which has been largely overshadowed by superficial discussions of a single, monolithic migration figure (NOM). The fluctuation in our population is rising dramatically, far more dramatically than migration itself. At years end, only a couple of years have actually resulted in a net cumulative movement which is noticeably higher than we’ve had for a very long time. 2008 and 2016 are those two years, but almost every other recent year has landed back down at a net cumulative movement that is not altogether unlike we’ve had in the 80s and 90s.

To see how that stacks up over time, let’s look the data we have in two time-periods, 2000 and later, compared to everything before (from mid 1976).

Figure 11: The recent historical accumulation people is increasingly seasonal

Figure 12: Past accumulations of people were less seasonal

Before 2000 it took about 23 years to add two million people to the population through net movements. Since then it’s taken 17 or 18, or maybe as few as 13 or 14, depending on whether you measure to peaks or an arbitrary equilibrium level. The point is that that we didn’t have peaks and troughs significant enough to care about in the past, but now we do.

If we run all of that together in a time series, the best fit we can find is a straight line, running at just over 100,000 people per annum.

Figure 13: Long-Term Cumulative Net Movements shows steady linear growth

I should point out that the line has an R squared value over 0.98, not to be dismissed lightly in terms of its fit to the data. You can see here that the impression that Australia’s migration rate has suddenly ramped up to extraordinary new heights relies on a definition that doesn’t incorporate the impact of our increasingly part-time population, and a pretty selective view of history that focusses on a short-lived ‘Golden Era’ of particularly low net movements during a few years in the early 2000s.

It also brings into focus the real complexity around things like infrastructure. Is there likely to be heavier traffic in February than in June? I’m haven’t yet looked to find out, but it’s quite possible that there isn’t. Our sense of traffic is dominated by the intensity of the peaks and troughs in its fluctuations on a daily or weekly basis, rather than its long-term aggregated average. It’s perfectly plausible that the people making up the net increase aren’t predominantly adding to peak-hour commutes, as they might be on holiday, or staying in hotels close to their work, or not yet enrolled in their course, or on their way out of town for an adventure, or seasonal work in the country. It’s also possible that enrolling in a course, or getting out of town, might put especially concentrated demand on certain pieces of infrastructure at exactly the worst times.

My point is that it’s really impossible to discuss this intelligently without diving a little further into the data to discover what really constitutes the movements, and the behaviour of people making them. Insistent claims that “Australia will need to add a Canberra-worth of infrastructure and housing every year” because NOM is over 200,000 people, (see here) are pure simplistic nonsense. In real terms, Australia adds a Canberra-and-a-half worth of people in a matter of weeks between December and February every year. But over the course of any given year we average much less than a third of Canberra from net movements. The infrastructure requirements for meeting that could be trivially easy, completely impossible, or anywhere in-between, depending on the more detailed sub-structure of these flows.

To speak intelligently about how/whether our population is growing, how/whether we should try to change that type of growth, one really has to understand more about our increasingly mobile part-time population. In later posts I’ll follow Dr Cameron Murray’s lead and further explore where in the world they are in any given time of year, whether and how they’ll collectively come back, and what we can or should do to prepare for it.