Agglomeration - diverting attention from the basics

A bit more on agglomeration economies

In my last posting I pointed out that the advantages of business concentration are not evident in the relative performance of Auckland in New Zealand or Sydney in Australia.  Even if agglomeration economies benefit individual firms or sectors, they are not being translated into superior regional growth.

Of course this could be a measurement problem.  Perhaps productivity gains in the major cities are such that they achieve superior output growth even as employment performance falls behind smaller centres. 

And of course, the gross figures used hide the "composition effect": history means cities are made up of different mixes of business.  Cities with an abundance of growth industries do better than those with too many sunset industries. In this way cities go through cycles of growth, stagnation, and decline just like products or firms.  Maybe agglomeration economies exist but Auckland happens to have an “inferior” mix of activities compared with the rest of New Zealand or metropolitan Australia.

Sounding a Warning

So I have done some more analysis, summarised below.  Because it is little dense – and most of this stuff on agglomeration is – here is a summary of my conclusions. 

There is no evidence at all that business sectors have been advantaged by being concentrated in Auckland over the past decade (whether or not individual firms are).  In fact, there may be a (weak) tendency for location here to be a disadvantage: growth prospects might be better for the less concentrated sectors and for activities located outside Auckland.

This raises questions over policies promoting business concentration in Auckland, especially policies that presume higher land use densities will generate higher productivity and therefore growth.

A focus on such policies may have deflected attention from the basics: the cost of doing business in Auckland, its attractiveness to investment, and its attractiveness to skilled people.  It means we may have underplayed the diseconomies of agglomeration, the environmental, social and economic costs associated with boosting densities on ageing infrastructure, for example.  These arise from congested and costly services prone to disruption (including but not limited to transport), inefficient labour markets (because of the commuting distances and housing costs, for example), and overpriced land. These may be he areas most in need of policy attention.

Agglomeration economies offer Auckland no silver bullet. Rather, such things as the quality of labour, the quality of services and infrastructure, the quality of investment, and the quality of life all require at once a more responsible, responsive, and perhaps light-handed approach for Auckland to prosper.

Data on individual sectors

I have explored whether agglomeration in Auckland has conferred advantages on business by examining whether individual business sectors concentrated in the region (1) do better than other sectors less concentrated there and (2) do better there than the same sectors elsewhere in New Zealand. 

For this I analysed employment data (from the Statistics New Zealand website) for sectors defined at the four digit level of the New Zealand and Australian Standard Industrial Classification. I looked at the years 2000 to 2007 (a period of sustained growth) and 2007 to 2010 (a period of employment decline).  Primary sectors (agriculture, forestry, fishing, and mining) were omitted as they are not urban activities. Sectors with fewer than 200 employees in Auckland in 2010 were also omitted.[1] 

For each of the remaining 168 sectors I calculated a score known as the “Location Quotient” (or LQ) indicating how heavily they were concentrated in Auckland in 2000, and another for 2007.  A LQ of 1.0 means that the share of a sector in Auckland is the same as the region’s share of all New Zealand’s economic activity (measured as employment).  Anything over 1.0 means the sector is concentrated in the region.  Anything less than 1.0 means it is under-represented. 

The proposition

If a sector is concentrated in Auckland relative to the rest of the country this suggests that firms in the region should reap economic benefits from shared services, skills, supplies, and information. Consequently, these sectors should outperform other sectors in Auckland that are less concentrated and, more importantly, should do better in Auckland than elsewhere in the country.

Put simply, we would expect a high LQ to be associated with faster growth rates in the city. 

The big picture

To explore this proposition the 168 sectors are plotted on two axes in the following diagram. The horizontal axis measures their concentration in 2000 (increasing from left to right).  Anything with a LQ of more than around 1.3 can be considered significantly concentrated in Auckland.  58 sectors fell into this category, and another 30 had some degree of concentration (a LQ between 1.1 and 1.3).

The vertical axis measures their employment performance relative to the rest of New Zealand.  The higher the score, the better the sector performed in Auckland compared with the same sector outside the region.

Guess what? There is no relationship evident between concentration in Auckland in 2000 and how sectors performed over the next ten years. Sure, a few were concentrated and have done well (in the upper right area of the graph), but plenty of others have were concentrated and have done badly (lower right).  Equally, a number of the sectors that are less concentrated in Auckland have nevertheless performed well (upper left). 

Concentration and growth – or decline?

But there's more. If we want to measure the strength of any relationship between sectors concentrated in Auckland and their growth, we can estimate the correlation coefficient between the two sets of scores.  The correlation coefficient would be close to 1.0 if concentration does increase the likelihood that a sector concentrated in Auckland will perform better there than elsewhere. If the score is close to zero no such relationship exists.  If it is close to -1.0, then concentration in Auckland is actually associated with under-performance, the converse of what agglomeration theory and policies would lead us to expect.

A number of coefficients have been calculated to see if the level of concentration in Auckland in 2000 and 2007 is associated with how employment changes subsequently.  They have also been calculated separately for different groups of sectors.  The results are given in the following table.

They provide no proof that agglomeration economies favoured Auckland employment at all.  If anything, employment performance in Auckland over the decade tended to be worse than the rest of New Zealand (r=-0.22) if a sector was concentrated in Auckland the city at the beginning of the decade, particularly in commercial services (r=-0.31).  Although these relationships are weak, the point is that they are in the wrong direction!   

And the post 2007 shakeup had a tendency to be more severe among Auckland businesses in the industrial and distribution sectors, shown in the second part of the table.

The Correlation between Sector Concentration in Auckland and Employment Growth, 2000-2010

	All Sectors	Industry	Distribution	All Services	Commercial Services
(1) Relationship with 2000 LQ
Auckland Growth 2000-07	-0.09	-0.07	-0.09	-0.04	0.07
Auckland Growth  2007-10	-0.16	-0.07	-0.24	-0.11	-0.11
Relative Growth 2000-07	-0.18	0.05	-0.12	-0.35	-0.35
Relative Growth 2000-10	-0.22	-0.08	-0.26	-0.27	-0.31
(2) Relationship with 2007 LQ
Auckland Growth 2000-07	-0.14	0.08	-0.22	-0.20	0.01
Relative Growth 2007-10	-0.15	-0.21	-0.33	0.06	0.06

Note: Relative growth is Auckland employment growth rate minus rest of New Zealand employment growth rate

Proving the existence of agglomeration economies has become a major past-time for academics and policy analysts.  It’s all pretty arcane stuff.  Even the simple enough analysis outlined here contains limits and qualifications. 

But the results raise serious doubts over the value of policies that rely on the benefits of agglomeration to boost growth.  Or, worse, policies that manipulate land use to increase employment densities in the belief that this will somehow advantage businesses. These results, at least, suggest that at best such policies have no effect and at worst could disadvantage the region’s economy.

It may well be time to address the fundamentals about what makes a city an attractive place to invest and live in over and above aspirations to be the best simply by being the biggest. 

[1] There were 19 sectors with less than 200 people employed in 2000, accounting for just 1,420 jobs.

---

Is bigger better? Trying to reshape Auckland’s lagging economy

Getting Auckland growing – a national policy imperative

It has become conventional wisdom that New Zealand’s performance depends on Auckland’s economy, and that this in turn depends on agglomeration – the supposed economic advantages associated with large cities. 

But is Auckland really the driver of New Zealand’s growth?  It would be nice if it was, if innovation and investment in urban business was the main source of our prosperity.  And hopefully it will be a major contributor one day.  But the truth is, as I pointed out earlier, the country remains dependent on the fortunes of our non-urban sector. 

So how are we going to wind Auckland up from being a place where consumption and income redistribution are concentrated, to be the focus of national economic growth?

Promoting agglomeration to promote economic development

 Policy advisors make much of the virtues of agglomeration – sheer city size – as a driver of growth. [1] By concentrating in our largest urban area, businesses are presumed to reap productivity benefits.  Is this right?  And is it sufficient to drive national economic performance?

The benefits to firms of locating in large urban areas were first discussed late in the 19^th century by economist Alfred Marshall. The low cost of transactions with nearby businesses, shared access to specialised services and suppliers, a common labour market, and information sharing explain the advantages enjoyed by firms in big cities.

Arguments underpinning agglomeration economies are firmly rooted in the industrial age.  So they are questionable when wealth is associated with services as much as goods, logistics companies integrate international transport, communications are seamless, and capital moves easily to most places in the world.

There is still some statistical evidence to suggest that firms in bigger cities (or states) might grow faster than in smaller ones.  The difficulty, though, is in separating differences in performance from differences in the fine mix of products, services, and occupations that naturaly concentrate in larger cities in response to the larger markets there in a cumlative process that is not necessarily economically rational. 

But does it lend itself to policy?

Despite this, agglomeration theory has been adopted by policy makers in two areas. First, has been promotion of concentrations or clusters of businesses as a growth policy.  Second, compact city advocates argue that a higher density of employment in large cities is a path to higher productivity. 

Leaving aside possible flaws in data, methods, and assumptions, the statistical evidence reveals at most 10% gain in aggregate productivity for a 100% increase in employment density.  More commonly the estimated gain is half that.  Anyway, any estimate is so influenced by method and measurement as to undermine any confidence in policy based on it, as demonstrated in a recent report by Motu Research for the New Zealand Transport Agency. 

Surely there are easier, less costly ways to gain a few percentage points in productivity and output: to do with the quality of the labour force, for example, the quality of investment, or commitment to research and development?

Living in policy dreamland

As far as I can see from reading many international studies on agglomeration, the cost of lifting the capacity of ageing infrastructure within existing urban areas rarely gets a mention.  In trying to reveal cause and effect from within a tangled web of associations, the studies assume away the actual costs of increasing land use densities.  They ignore the quite different circumstances and histories of the cities and states bundled into their samples.  And they often mistakenly infer investor and firm behaviour on the basis of trends among cities. [2]

Transforming the buildings and infrastructure that make up a city to increase land use densities in the hope that businesses will screw more out of their investments is a very costly exercise.  The impact on the quality of urban life is also likely to be unacceptable.  Crowding and congestion, the unreliability of over-capacity and dated infrastructure and network services, and increased risk of service disruption mean that significant increases in density are unlikely to be sustainable even if they can be achieved without a fiscal blow-out.

Quite apart from the economic and political naivety of jumping from quantitative analysis to policy prescription, is bigger better anyway?  A simple look at recent history does not show that it is.

Auckland is already bigger – but is its performance better?

A focus on Auckland’s growth presumes that it is our only city big enough to have an international presence, so that we should focus policy and plans on reaping agglomeration benefits there. 

To explore this presumption we look at how well Auckland has done recently.  In 2000 the region had 32.6% of New Zealand’s employees, nearly two and half times Canterbury’s employment. [3] Surely this spectacular concentration of economic activity would become apparent in Auckland’s superior performance?  if agglomeration economies are to be evident anywhere, they would show up here.

Well, no.  Look at the graph.

Auckland’s employment growth lagged Canterbury’s through to 2010. It lagged non-metropolitan New Zealand (areas outside Auckland, Canterbury and Wellington).  And when the global financial crisis hit in 2007, Auckland’s employment fell faster.  Even Wellington, trailing early in the decade, was not hit as hard, down just 0.7% from 2007 to 2010 compared with Auckland’s negative 2.8%.

Naturally Auckland still dominated national employment in 2010 (32.4% of the total), but according to the theory that dominance should have increased not diminished.

So how do we explain this?

First, maybe the economic benefits of agglomeration are overstated.  Tthe empirical evidence tends to be generalised and static.  Despite the growing complexity and apparent precision of analyses, the causal impact on firm performance of lifting urban densities remains fuzzy and trivial. They are a weak foundation for policy.

Second, I suspect that all those factors assumed away in the analyses are actually important in the real world.  We cannot exclude the physical environment of individual cities, their history, the condition of sunk capital (public and private), different regulatory environments firms, and stocks of human capital, skills, and experience. To assume that proximity and density have a predominant influence over the investment and operation of firms seems naive.  And we cannot ignore the external costs of over-concentration.

Third, what if it really works? What if the secret of economic growth is simply being the biggest city? And at the sam time, what if breaking down economic barriers means that the advantage of size is measured across nations, not just within them?  Where does this leave Auckland?

New Zealand economy’s is increasingly integrated with Australia’s. Labour, capital, goods and services move relatively easily between the two countries.  Where does New Zealand stand in an Australasian urban system?  The second graph gives us some clues. Look at where Auckland stands in the bigger picture.

Australia’s metropolitan areas outperformed their New Zealand counterparts between 2000 and 2007. [4] In percentage terms, Sydney was a bit of a laggard.  But this simply confirms that being biggest is no guarantee of top performance.  Melbourne outstripped both Auckland and Sydney. 

More interesting were the second tier cities, Brisbane and Perth. With just 12% of Australasian metropolitan employment in 2000, Brisbane accounted for 18% of its growth!  Compare this with Sydney: with 29% of employees in 2000, it accounted for just 22% of subsequent growth.

Auckland had 7.8% of Australasian metropolitan employment in 2000 but accounted for only 7.1% of growth, significantly less than Perth (11.8%) and a little more than Adelaide (6.8%). 

Two simple conclusions 

One, in a global economy we cannot reap greater economic benefits from urban agglomeration than our Australian (or, especially, our Asian) neighbours. Being the biggest city in New Zealand offers no particular advantage in an international economy.  We have to find a different way.

Two, in both Australia and New Zealand, being biggest does not guarantee topping the economic performance tables.  Let’s do what needs to be done to make Auckland a great place to do business.  But size is certainly not everything in these stakes.

Sure, these are broad comparisons based on general data, but if we think that promoting agglomeration holds the policy key to Auckland’s – and New Zealand’s – economic performance, we need to think again.

[1]        E.g., Ascari Partners (2007) Assessing Agglomeration Impacts in Auckland: Linkages with Regional Strategies, Report to Auckland Regional Council

[2]           It is a mistake to infer the behaviour of individuals from the behaviour of populations, the so-called ecological fallacy. There are exceptions, e.g., David Mare’s Labour Productivity in Auckland Firms deals with micro-level data in an attempt to reveal the impact on individual firms of locating in Auckland

[3]           Based on employment data from the Statistics New Zealand Business Directory. 

[4]           Labour market data is for the major statistical regions in February so that the figures are broadly comparable with the New Zealand Business Directory figures.

---