Facebook Google Menu Linkedin lock Pinterest Search Twitter



Feb 25, 2013

Agglomeration economies: are they exaggerated?

We're increasingly being told we need higher job densities in our cities to exploit 'agglomeration economies' associated with the 'knowledge economy'. But are the benefits exaggerated?


The city centre's % share of total metropolitan employment (from Charting Transport). Note that cities can't be compared due to differing definition of the centre

Promoting ways to increase the economic benefits from “agglomeration economies” is now standard fare in planning strategies for cities. It’s the new orthodoxy.

For example, see the Federal Government’s newly released State of Australian cities 2012 report, or the Victorian Government’s current metropolitan strategy discussion paper.

Another example is this recent short article in the Australian Financial Review, Heart of the city remains the lure, by Jane-Frances Kelly and Peter Mares of the Grattan Institute.

The authors quote Federal Infrastructure Minister Anthony Albanese, who says housing continues to “creep outwards”, even though Australian cities are “shrinking in on themselves” as knowledge-intensive industries create inner-city jobs.

They say “the rise of knowledge-intensive industries is clustering high-value jobs around city centres.” That’s because a “growing body of research shows that the more knowledge-intensive our economy becomes, the more proximity matters.”

The authors say the reality is most people want to live near their inner city jobs, yet despite this trend we:

continue to spread our cities outwards, as if the future of employment lay in manufacturing plants on the suburban fringe.

I think many involved with cities often get a little hot and breathless about agglomeration economies. That’s in no way unusual – urban policy is inherently political and very susceptible to “the next big thing” syndrome.

It’s true firms value agglomeration economies – that’s primarily why they overwhelmingly locate in cities. They can draw on a bigger pool of labour; they can reduce transportation costs by being closer to suppliers and markets; they learn from each other through “knowledge spillovers”; and more.

Within metros, very dense locations – like the centre – amplify these benefits for some firms. But not for all, or even most.

Many don’t like the high rents and congestion that come with very high density. So most locate further from the centre where rents are lower.

Although the Grattan article assumes it’s all about the city centre, the fact is only a minority of firms locate there. And it’s a small minority at that.

In Melbourne, for example, only around 10% of all metropolitan jobs are in the CBD i.e. the “Golden Mile” bounded by Flinders, Spencer, La Trobe and Spring Sts. Include the new areas of Southbank, Carlton and Docklands – what I call the extended CBD – and it’s still only around 15%.

In fact less than 20% of jobs are within the boundaries of the City of Melbourne, which covers an area of 36 sq km. Only 28% are within a 5 km radius of Melbourne Town Hall (area 78 sq km).

And half of all the jobs in metropolitan Melbourne are more than 13 km from the CBD (more detail on the geography of employment in Melbourne here).

Not only is the proportion of firms who value the high density of the city centre quite small, but those firms are also very particular.

They include corporate and government administrative headquarters and highly centralised institutions like the Supreme Court, opera houses, and Parliament.

The sector, though, that really benefits most from the high densities of the CBD is commercial services. It comprises the high-end functions of the finance, insurance, business services and property services industries.

This sector accounts for a massive 54% of all jobs in the extended CBD (61% in the CBD proper) but just 19% of those in the rest of the metro area. It is far and away the most geographically centralised sector in Melbourne.

These CBD commercial services firms are also highly productive in terms of the value of their output, but that’s no doubt also true of a mine in outback WA. What matters in the context of planning is density, and firms in this sector are prepared to pay big for the benefits it gives them.

What all this tells us is that only a small minority of firms value high density and they’re drawn from a narrow of range of sectors and institutions. Most firms place a much lower premium on density.

That contention is borne out by estimates that it would require a doubling of metropolitan job density in Melbourne to get a circa 9% increase in metropolitan productivity. That’s a sizeable increase, but generating a 100% increase in metropolitan job density would be incredibly hard.

The meme that demand for greater agglomeration in the centre is growing rapidly is questionable too. As I’ve discussed before – and as the exhibit shows – the share of metropolitan jobs in the CBDs of most Australia’s capitals barely changed between 2006 and 2011 (and in two capitals went down).

The share captured by Melbourne’s CBD was static, although the (extended) CBD increased its share. That might have been primarily because increased land supply (e.g. Docklands) lowered the cost of a CBD location, rather than because structural changes in the economy mean firms are placing a much higher value on density.

For the great bulk of firms, it’s being in a metropolitan area that counts most, not being at high density.

Indeed, while nearly all firms value some degree of agglomeration, most only want a modest level because for them (and their customers), the costs exceed the benefits at even relatively low densities.

The density of employment in inner city Melbourne – let alone in the suburbs where most jobs are located – is an order of magnitude lower than that in the CBD.

So CBDs are characterised mainly by the sorts of knowledge-intensive jobs that Mr Albanese cites. But there’re too few, and the rate of growth is too modest, to sustain the claim that Australian cities “are shrinking in on themselves”.

The Grattan Institute’s argument that “(the) reality is that people want to live near their inner-city jobs” might well be true for the elite who read the Financial Review, but it’s not true of the vast bulk of Australians.

The Institute needs to understand that most people and most jobs are in the suburbs. Most workers can’t afford to live in the inner city – only 8% of Melbourne’s population lives within 5 km of the CBD.

It should also be noted that the attractions of density aren’t the only forces driving firms and organisations to locate in the city centre. Some of the most highly centralised industries – like culture – value the CBD primarily because of its accessibility rather than its density.

There’s a host of other interesting questions associated with this topic, like: do cities really need higher job densities in the centre in order to flourish and be sustainable?; Who really benefits from the high density of the CBD – is it, for example, shareholders, or is it executives?; Do implicit subsidies for CBD firms – like subways – make the rest of us better off, or worse off?

And in particular, do the benefits of subsidising high density-loving firms to locate in the CBD really trickle down to the rest of the population? I’ll look at some new research on the latter question shortly (see here).


We recommend

From around the web

Powered by Taboola


Leave a comment

Your email address will not be published. Required fields are marked *

8 thoughts on “Agglomeration economies: are they exaggerated?

  1. Scott Elaurant


    My apology if I have misinterpreted your remarks. In terms of meeting the demand for supporting infrastructure and services, I fully agree we cannot be too CBD-centric.

    I think our perspective on this issue also depends on the city you live in. In Melbourne there is an obvious need to increase services, particularly public transport, in outer suburbs. Here in Adelaide there has been large recent spending on freeways in the outer suburbs, whereas PT services to the city still need capacity upgrade.

  2. Alan Davies

    Scott Elaurant #6:

    I don’t know why you think that. The article is quite clear that high density locations like the CBD benefit some industries, esp high-end commercial services.

    The article is not anti-density or anti-CBD. The key point it makes is those firms (while they contribute disproportionately to metro income) are a minority of all firms/jobs in the metro. In fact half the jobs in the commercial services sector in Melbourne are in the suburbs.

    City managers concerned with practical issues like providing infrastructure have to consider all firms.

    The article also makes the point that the share of jobs in the CBD is not increasing significantly (even if their share of metro income is). That is, the claim that the demand for very high density is increasing is over-stated. That has important planning implications.

  3. Scott Elaurant

    I think this post somewhat misunderstands the nature of the claim for agglomeration economies. The claim is not that they generate more jobs, but that the jobs in higher density locations are more valuable. In this regard Melbourne would surely prove the theory corret, because the jobs in the Melbourne CBD are obviously more valuable per job (higher income) than say those in Sunshine.

    Paul Buchanan did a lot of empirical research on this topic to evaluate the London Crossrail project. He was able to establish that there are reliable elasticities of productivity for urban density for many industries. See:

    I have applied this theory in Adelaide and it held true for the Adelaide CBD as well. i.e. the highest value jobs were in the CBD. This was true not only overall, but within each industry. E.g. the highest value finance jobs are in the CBD, the highest value service jobs, and so on. It takes a bit of work to calculate because ABS only publish income by place of residence, rather than by place of work.

    It is true that there are some industries for which there is no evidence of any density/productivity benefit – e.g. manufacturing. However, generally these are not the high value jobs that cities are trying to attract. For most of the latter category of jobs, aggoleration elasticies are significant.

  4. 2perdean

    Could you factor in the effect shift work has on the jobs distribution? most people in the cities work 9-5 while a lot of companies in industrial areas have 2 shifts or are 24/5. including restamurants and hospitals? if CBD companies started going 24/5, would they have higher job densities?

  5. hk

    Are the unit costs for commercial services; which comprise the high-end functions of the finance, insurance, business services and property services industries lowered by agglomeration? My hunch is probably not. The benefits are exaggerated!
    The staging and managing of existing agglomeration is more about protecting and expanding the existing value of institutional and developer investments than it is about planning for the net community benefit to the urban population.

  6. Jack Guilmartin

    Just want to let you know the title is mispelled

  7. Dylan Nicholson

    Urban density is already increasing in most (all?) Australian cities…to me it’s really neither here or there whether that’s a good thing economically, what matters is whether it can occur in such a way that our quality of life is not just maintained but continues to increase. Transport seems to be by far the biggest challenge in managing this, followed closely by how to preserve/enhancement “neighbourhood character” but not fall victim to constant NIMBYism that prevents necessary addition of living-space in inner city areas, especially for lower-income households that can’t afford the sorts of things developers these days have decided are the only things worth building.

  8. Strewth

    Thanks Alan – lots of good argument and supporting data here, though I’m not sure it really tells us much that we didn’t already know. It’s fairly well understood that the industries that employ most people in Australia are those like health, aged care, education and retail that don’t benefit so much from metropolitan-level agglomeration as they do from locating closer to where people live. This kind of employment will always be spread across the greater urban area and will account for a large portion of total employment.

    As you say, the real beneficiaries of agglomeration are the ‘knowledge economy’ sectors like commercial services, R&D and high-end administration. They represent a smaller share of employment, but people focus on them because they are considered vital to getting big improvements in economic productivity. If in future we are to keep getting more for less in something like primary health care, it won’t be because GPs agree to do the same work for less pay – it’ll be because GPs get together with other GPs, medical researchers and other ‘knowledge workers’ to improve treatments and the underlying processes and technology. Agglomeration and close linkages between institutions are of major importance here – even if the GPs themselves work mostly on their own or in small clinics in the suburbs.

    And in Melbourne at least, it’s never been just about the CBD but the inner-urban region extending out to 10 or even 15km: this region still has the highest concentrations of employment and is the focus for most growth in employment. Outside this region, jobs will largely follow the people because of their nature: which also means that if the population becomes more concentrated in some suburban centres, jobs will become more concentrated as well.


https://www.crikey.com.au/2013/02/25/aggomeration-economies-are-they-exaggerated/ == https://www.crikey.com.au/free-trial/==https://www.crikey.com.au/subscribe/

Show popup

Telling you what the others don't. FREE for 21 days.