Jane O’Sullivan has made some simple estimates of how much it costs Australians to maintain a high rate of immigration (200,000 per year) and “natural” increase (nearly as large). Her results are sobering.

For Australia with a population growth rate around 1.5%, the cost is more than 10% of GDP. That’s a huge impost on our national effort, around $15,000 per taxpayer. Say $150 billion per year.

Jane’s arguments are outlined here because the original papers are not very accessible.

Estimating the cost

The argument has been explained in a paper by Jane O’Sullivan^*. Infrastructure refers to shared facilities like roads, electricity grids, water supplies, hospitals, schools and so on. In the context of the present argument, we need also to include housing, shopping centres and other “private” facilities, because these are also needed by new immigrants, hence the broader term “durable assets”.

The essence of the argument goes like this. A lot of infrastructure lasts a long time, typically around 50 years. If our population were steady we would need to replace about 2% of infrastructure per year, so after 50 years it would all have been renewed. Then we’d start again.

If the population increases by 1% in a year, the infrastructure should be expanded by about 1%, to accommodate the new arrivals without degrading the service. That means we’d have to build 2% (replacement) plus 1% (expansion), for a total of 3% of infrastructure in a given year.

Infrastructure is expensive. Whatever we would pay for replacement with a steady population, we would pay half as much again with the growing population. That’s a 50% increase in cost.

Putting numbers to it, Australian Gross Fixed Capital Formation (roughly, public infrastructure) cost $283 billion in 2010. GDP was about $1308 billion, so the infrastructure  cost was about 22% of GDP. If two thirds of that was for replacement of old infrastructure and one third was for capacity expansion, the latter would be about $94 billion or 7% of GDP. For a steady population the cost would have been about $188 billion, versus $283 billion with the growing population.

Those numbers are rough estimates using a simplified argument, for the sake of demonstrating the idea, and getting a feel for the magnitude of the cost: $94 billion. A more refined estimate takes account of different lifespans of different kinds of assets, and of some additional factors that increase or decrease the estimates.

For example the lifespan of household furnishings would be around 15 years. That would mean replacing about 7% every year to keep them up to date with a steady population. With population growth of 1% in a year, we have to buy 8% of the total stock. So population growth has a smaller proportional effect for shorter-lived assets. In 2010 we spent $24 billion on household furnishings, so one eighth of that, or $3 billion, would be for new arrivals.

A compensating factor is that, with a steadily growing population, by the time an asset is due to be replaced there are more people to pay for it, so the cost per capita is diluted. This reduces the cost as a fraction of GDP by almost half. On the other hand some infrastructure may have to be replaced or moved before it reaches its nominal lifespan because of increasing density of the population. Higher density may also require more expensive infrastructure, such as tunnels and overpasses. These will increase costs over the basic estimate.

O’Sullivan compiled a more detailed estimate for the total cost of providing durable assets of several kinds, taking account of these additional factors, and arrived at $160 billion for 2010, a year in which the population grew by 1.45% or 325,500 people. This is 12.2% of GDP, or $490,000 per person added. For a 1% population increase the cost would have been 8.4% of GDP.

In a later study for the UK, O’Sullivan refined the estimates a little more and got 6.9% of GDP per 1% increase in population²³³. Still she argues that her estimate errs on the side of being conservative. These numbers can certainly be refined, and that ought to be a priority task of governments, using their more complete data. However it is clear the cost is large, closer to 10% than 1% of GDP for a 1% increase of population per year.

Immigrants, children and old people

In Australia the population issue usually arises in two main ways. First is our high immigration rate, among the highest of developing countries. Second is a concern that over the next few decades the ageing of the baby boomers will impose a heavy burden on the working part of the population.

The debate about the “ageing” population is skewed and incomplete. First, O’Sullivan notes that people are dependent at both ends of life: children have to be supported, just as old people do. As the proportion of old people increases, the proportion of children decreases (assuming there is not another baby boom), so the one offsets the other. During the baby boom of the 1950s and 1960s, there were relatively more child dependants than old dependants. Now, as baby boomers retire, the proportions reverse. The relative costs will not be identical, but they will still substantially cancel.

O’Sullivan proposes that immigration creates a third category of dependants, those who have not yet arrived. If, ideally, durable assets are expanded before they arrive, then the cost is borne entirely by the present population. If, more realistically, durable assets are expanded during or after the arrival of the immigrants, then the assets are overloaded and everyone pays indirectly through reduced services. Unfortunately governments are happy for the costs of reduced services to be left out of their accounts and rendered invisible.

One of the justifications claimed for a high immigration rate is to counter the ageing of baby boomers by bringing in people who, on average, are younger. It turns out the effect of immigration on the age profile is quite minor. The cost of supporting old people is anticipated to rise slowly to 2.4% of GDP in 2050, assuming a 1.2% population growth rate in the meantime. The additional cost due to the higher proportion of older people is not given, but will be only a fraction of this total, let’s say roughly 1% of GDP. Thus the additional cost of the aged would be around one tenth of the cost of population growth, so it is a quite secondary concern.

Another claimed justification for high immigration is the need to compensate for supposed skills shortages. O’Sullivan argues that immigration may have little effect on skills availability, because adding a person with a specialist skill also increases demand for many other skills. Whether the net effect is significant, or even negative, is not clear and would require very detailed analysis to demonstrate.

An underlying reason used to justify high immigration comes from economic analyses that purport to show a net economic benefit of a larger population. These are based on “snapshot” portraits of the economy now and in, say, 2050, without careful modelling of how to get from here to there. Thus these studies overlook the large cost of getting there, around 10% of GDP per year. Even so, a number of studies have failed to show a significant increase in GDP per capita in a larger population, even including one by Australia’s Productivity Commission²³⁴.

Thus the claim that a larger population is beneficial is based on effects that are quite small (around 1% of GDP) or possibly absent. The overlooked cost of expanding durable assets is much larger, around 10% of GDP, and completely dominates the assessment. A high immigration rate imposes a very large cost on the existing population.

Reasons for overlooking immigrant costs

O’Sullivan notes several reasons why the costs of expanding durable assets have been overlooked. Economists have used snapshot before-and-after models that do not include the process of change. This is symptomatic of mainstream economics’ reliance on static, equilibrium models that are incapable of addressing processes of change, whether they be population growth or a market crash.

Another major reason for the neglect is that infrastructure costs are presumed to be an investment and therefore obviously necessary and good. With a steady population that would be true, but with a growing population the extra cost does not yield any benefit to those paying it. It is the cost of running to stay in the same place. O’Sullivan argues, sensibly, that the cost of expanding durable assets should be treated as a recurrent cost, not an investment.

There are other reasons as well. The costs are spread around many sectors and portfolios, and at different levels of government: for example housing, education, health, transport, telecommunications, water, electricity, waste management and so on. No-one collates the total cost. The obsession with increasing the GDP biasses policy to paths that increase GDP, or increase it faster, even if GDP per capita does not increase, and even if a smaller population with a smaller GDP would yield a higher GDP per capita.

Finally, if durable assets were expanded in time for new arrivals, so as to maintain services, it would be more obvious the cost was due to the increasing population. However the expansion of durable assets typically lags the new arrivals, so it seems as if the cost is just normal maintenance for the benefit of everyone, rather than because of the new arrivals and with no net benefit to existing residents.

Population policies

Population growth has been much discussed, but more often in the context of “developing” nations. A high population growth rate has long been recognised as an impediment to “development”. In one of the few precursors to O’Sullivan’s work, Lester Thurow estimated that expanding durable assets would cost around 12.5% of GDP per 1% pa increase in population²³⁵. He concluded that no country was likely to progress economically with a population growth rate above 2%. It is salutary that in 2009 Australia’s population growth rate peaked briefly at 2.1%.

A widespread view has been that as economic development proceeds women become more educated and have fewer babies, so reducing the rate of population growth. This is called the demographic transition. O’Sullivan disputes this interpretation, arguing that a strong family planning program has been a pre-requisite for rapid growth of wealth, which gains speed as fertility drops below 3 babies per woman. She argues that education and other factors are not sufficient to explain development, and that the overlooked factor is the cost of durable asset expansion.

Turning to Australia, Engineers Australia’s Infrastructure Report Card for 2010²²¹ rates our infrastructure at C+: “major changes are required to enable infrastructure to be fit for its current and anticipated future purposes”. They estimate a shortfall costing $700 billion. We may conclude that congested city traffic, crowded city trains, rapidly rising electricity charges, struggling public hospitals and schools and many other symptoms of overloaded infrastructure are substantially (though not exclusively) the result of the high immigration rate.

O’Sullivan’s estimates come in at around $400,000 per person added. If our immigration rate were reduced from, say, around 230,000 to 80,000 per year, a reduction of 150,000, we would save, conservatively, $60 billion per year. That could buy a lot of schools and hospitals. It could pay for the original National Broadband Network in a year or so. It could enable a fast train network to be built through eastern Australia within a few years.

I have written mostly of immigration, but of course the total population increase is due also to local “natural” increase, the net of births over deaths. However this is smaller than present immigration rates, about 150,000 per year, and it is likely to decline slowly. The natural increase rose around 2006, evidently responding to Government encouragement, which included a baby bonus payment. It could, if desired, be reduced by analogous persuasion, without any need for draconian policies.

We should always look for the beneficiaries of any particular policy, and the beneficiaries of a high immigration rate are not hard to find. The construction and related industries get an obvious and immediate benefit. Those many big businesses that want to ensure a steadily increasing customer base are also prominent supporters. Politicians who want to ensure the GDP increases, ensuring good headlines, are also strong supporters, even though the GDP per capita may be static; in other words the growth obsession is a major factor in our population growth. So these groups benefit at the expense of the rest of us. Here is yet another way in which the flow of wealth is biased to the wealthy.

There is a vigorous, though marginalised, argument that Australia should stabilise its population. The main argument is that we are exceeding the carrying capacity of the land. Though this argument is very sensible, it is hotly disputed, and it is difficult to establish it with definitive arguments. As well, population increase is commonly regarded as a slow issue, mainly affecting the future, and therefore one that lacks urgency and can always be put off.

The cost of expanding durable assets, on the other hand, is large and immediate, and the benefits of reducing it would be quickly evident. This is true whether the cost is paid in dollars in advance or in congestion and poor service due to the failure to keep up with or ahead of demand. It is therefore a far more potent argument for reducing our immigration rate and, over time, also reducing our rate of natural increase.

It is a great credit to Australia that we have absorbed such a diversity of people over the past 60 years. Our immigrants have moved into all parts of the society. They are innovative, productive and creative, and our culture is enriched in much more diverse and substantive ways than just food, although that has added many pleasures to our experience. We may keep spicing our population with new people, but we will serve everybody better if we do it in a more informed and considered way.

• O’Sullivan, J. N. (2012), The burden of durable asset acquisition in growing populations, Economic Affairs, 32(1), 31-37.