Friday, December 3, 2010

Are we being a bit dense about productivity?

Taking the long view
There has been a lot written about the virtues of a high density city and a thriving CBD.  I have no issues with either of these notions, but let’s at least understand the dynamics before we concoct policy to try to bring them about. 
High density does not necessarily mean a city is more productive, or wealthier, or more sustainable.  It just means more people live or work in a smaller spaces.  This may be because of a land shortage - think Japan and its cities.  (I hope this is not what planners have in mind for Auckland).  Traditionally, high density living is associated with less wealth, lesser lifestyles, and public health challenges. 
The long view of the west since the Middle Ages has been one of increasing real wealth and falling densities.  Especially since the 18th century greater personal mobility and enhanced material standards have been implicated in healthier, longer lives and greater social equity. 
Much as I rail against conservative planners, reversing this trend is something we should be wary about, especially when crowded cities are associated with social disparity and ageing urban infrastructure with greater risk of disruption.
The transport tail wagging the density dog
Yet today’s planners and policy-makers not only try to restrict development to increase densities, but also promote transport investments in the belief that they will make it happen. 
Instead of designing a public transport system that might serve our particular form of settlement in Auckland, we are designing one which is intended to change the shape of the city.  The New Zealand Herald (3 December 2010) quotes transport expert from the University of Sydney, David Hensher, as “absolutely amazed” that we would even consider rail given Auckland’s relatively low density.  It would be far more cost effective, he thinks, working on extending our busways.  At least that way we could get public transport to where people want to live and work and we could keep it flexible to meet changing needs.
Of course, one of the interesting things about rail-based public transport is that if it works it encourages long-distance commuting, and lower densities.  This, in turn, increases the distances travelled for non-work purposes. 
And then the transfer of travellers from private to public transport increases capacity on key roads, encouraging additional travel and promoting further suburban and exurban growth.  That’s why sweeping statements that “transport investments serve to facilitate higher densities” [1] have to be taken with a grain of salt. 
The density myths
I want to challenge what I see as some of the myths about city density.  I want to deal in this blog with the proposition that actively increasing density improves aggregate (or individual) productivity.  There are a couple of myths that I will tackle subsequently.  One is that increasing density will increase the health of the urban economy.  The other is that to save the city we must preserve the CBD. 
The productivity story
I have already queried the role of city size and density in urban growth in earlier blogs, mainly because it hasn’t apparently helped Auckland over the past decade.  Being concentrated in the region is more likely to be a disadvantage than an advantage to business. 
Despite this, the policy argument goes along these lines:
·         Large cities grow faster than small cities;
·         The higher densities associated with larger cities foster higher productivity (value added per worker);
·         Therefore if we force higher employment densities by regulating land use we will boost productivity and cause the economy to grow;
·         We can also increase employment densities in certain parts of the city by delivering public transport commuters to them;
·         These transport nodes will become fast growth areas and this growth can be attributed to  both public transport investment and consequent higher employment densities.
This reasoning supports policies to: (1) limit land availability for business and the view that business would be better off redeveloping “brownfield” sites or intensifying on existing sites because that will improve productivity;  (2) subsidise otherwise uneconomic public transport.
The evidence – marginal at best
Those are big – and potentially very costly – policy calls.  Just how much confidence can we have in them?  What does the supporting analysis say?
In New Zealand we relied for some time on overseas analyses, but since the mid 2000s we have developed a local evidence base using econometric research.  David Maré’s 2008 study is the most widely cited, and suggested a substantial productivity “premium” accruing to Auckland and, within the region, the CBD.  [2] This work has been called on, for example, to support the recent business case for an underground rail loop in Auckland.[3]  But just how robust is it for policy purposes?   
At best the evidence is weak with respect to the possibility that higher densities bring about higher productivity.  At worst, while the analyses may be technically correct, the policy assumptions that follow may be plain wrong.
The complication of composition
The growth of a national, regional, or urban economy over time naturally leads to a “sorting” of businesses.  This “composition” effect means that some places have businesses within them – for all sorts of historical reasons – that produce higher value goods or services than other places.  The composition effect based simply on classifying firms to sectors explains half of the apparent productivity premium identified by Maré. 
Even this may be conservative, though.  The two digit classification of firms he used still hides considerable diversity among firms in the same sector.  These differences are likely to explain a lot more than areferring simply to the generalised sector each belongs to. 
Maré also estimated the relationship between density and productivity in Auckland. His analysis siggested that 10% higher density is associated with 0.86% higher productivity.  This is not a huge gain when we consider what might be entailed by way of public spending and regulation to bring about a 10% increase in density.  There have got to be easier ways to increase productivity.
In any case, is this really a cause and effect relationship?  Maré's estimate came from a static analysis.  Establishing a dynamic relationship, such that when one item changes another follows, is more difficult.  Maré could not estimate a “robust” positive relationship over time, and called for more work to be done. Even if he could have demonstrated a dynamic relationship, we are still stuck with chicken or egg ambiguity – does higher productivity in fact lead to higher densities?
It gets shakier
Maré and Daniel Grahams' subsequent work for NZTA is even more circumspect.  Based on comparisons across New Zealand regions, it suggested that firms at locations with 10% higher density are 1.7% more productive.  But firms in different regions tend to do different things.  Observable differences based on 15 categories to allow for industry mix actually accounted for 70% of the density “premium”, dropping the productivity gain aacross a 10% density increment rom 1.7% to 0.5%.  (Incidentally, the figure for Auckland is down from 0.86% to 0.76% in this analysis).
Making sure we are doing the right things in the region is far better, I would have thought, than increasing densities and hoping this makes companies stronger.  The harder it becomes it invest in new activity, the more likely we are to be locked into yesterday's industries.
Oops - maybe Auckland is the wrong place
Maré and Graham explored different ways of estimating productivity effects, allowing for differences among enterprises and for sector mix within regions.  I drew the following conclusions from their analysis – although I concede some of the subtleties eluded me:
(1)    The predominant impact on the productivity-densities relationship is sector mix in a region;
(2)    Within Auckland, the stronger relationships and therefore the purported potential for gains from increasing employment densities are associated with the predominantly rural, low density areas of Rodney and Franklin, followed by Manukau, not the old Auckland city;
(3)    There are diminishing returns to agglomeration: productivity gains associated with density are higher in low density areas and lower in high density areas;
The results suggest that we might better concentrate on connecting and intensifying development in areas of relatively low density rather than trying to shoehorn more into areas that are already relatively dense.  When we take into account the relative costs – public and private – of trying to lift already high density localities this makes even more sense. 

Perhaps we need to revisit the notion I advanced to the Metropolitan Auckland in 2006 - let's lift our sights and consider the economic potential of the northern North Island, and begin to think about connectivity and opportunity on a wider canvas.

And shakier
The evidence regarding “productivity premia” in some places rather than others appears just too weak to support land use policies designed to enforce higher densities or to justify lifting urban densities as a rationale for transport investment.   
Daniel Graham is quite explicit on this matter in subsequent studies:
A key conclusion is that we are unable to distinguish agglomeration effects from other potential explanations for productivity increases, most notably functional heterogeneity.  Consequently, the agglomeration effects of transport investments cannot be interpreted causally.  [4]
Digging into heterogeneity – the occupational dimension
Heterogeneity- - a fancy word for diversity – almost inevitably overwhelms any econometric attempt to measure a relationship as weak as that between density and productivity.  The analysts try to control for the sector a firm is in, but that still hides a whole heap of heterogeneity.
Sectors are not the only things that vary between places.  Occupational mix is critical, for example.  The 2006 New Zealand Census showed that 19% of employment in Auckland region fell into the category “Legislators, Administrators, and Managers” compared with only 14% in the rest of New Zealand and 21% in the CBD.  Another high value category, Professionals, made up 18% of Auckland’s employees, 25% in the CBD, but only 15% in the rest of New Zealand.  Similarly, Technicians and Associate Professionals made up 19% of employees in the CBD, 15% in the region, but only 12% in the rest of New Zealand.
Jointly these high value added occupations make up 66% of Auckland’s CBD workforce, 50% across the rest of the region, and 42% across the remainder of New Zealand.  Such pronounced labour market stratification will account for differences in value added per worker even within firms and sectors.
Conclusion –make it easier to do business to make Auckland grow
Quite simply, the higher order, management, control, and negotiation jobs tend to end up in big centres.  This facilitates a range of local, national, and international transactions that they are associated with.  The challenge for planners is to make it easier, though, and not harder for them to locate in the region.  Imposing congestion and over-the-top infrastructure costs associated with promoting a high density city might just have the opposite effect, discouraging local investment and employment.
The benefits that firms might be looking for by heading to the principal commercial centre in a region or nation can be delivered by the quality of transport and communication links for the conduct of business, and the quality and cost of living to sustain a healthy labour market.  But if we make it too hard for them to locate at a suitable site in Auckland – whether or not that is in the CBD – the continuous improvements in transport, mobility and communications that we are witnessing today might just make it easier for them to leave or bypass the region.

[1]  Maré D and Graham D (2010) Agglomeration elasticities in New Zealand, NZ Transport Agency Report 376
[2]               Maré D (2008) Labour productivity in Auckland firms Motu Research Paper 8-12
[3]               APB&B (November 2010) Business Case Auckland CBD Rail Link for KiwiRail and ARTA
[4]               Graham DJ and Ven Dender K (2009) Estimating the agglomeration benefits of transport investment: Some tests for stability Discussion Paper 2009-32, Joint Transport Research Centre, OECD


Andrew D Atkin said...

Hi Phil,

Here I go again...

I know the focus of your post is the question 'does higher density lead to greater productivity', but I would like to make the point that the (supposed) productivity gains of intensification cannot justify forced intensification, regardless. Even if we were more productive in a "human zoo" that would still in no way justify forced intensification, because at the end of the day free men and woman are not sombody elses "human resource" to be forced onto a planners treadmill.

Some other points of interest:

-Low density living in itself is a "product" that thousands of people want to consume. Did the planners measure and balance that fact? It would seem a bit crazy to have everyone with money in their pockets yet no power to buy the key thing that they want.

-And then there is the fact that productivity can only be statistically measured in terms of the taxable economy(?) what about the domestic economy? You can either make your own lunch or buy it in town - so which functionality is more "productive"?

-And what exactly is the relationship between production/consumption and real living standards? I've often thought that the most direct way to improve real living standards is to reduce the very need to consume in the first place eg. don't consume so much power by living somewhere warmer. Obviously increasing consumption is not the only way to improve real living standards, and that's one of the (many) things to consider before prescribing people's lifestyles.

...and a final point: clearly density in itself does not achieve an economic positive, but efficient access does. Common sense tells me that increasing density will not have much of an effect because you can stretch a city out significantly before having any concerning impact on average travel times. Making the roads say 30% longer between the stops won't do much to extend average urban travel times (nor the fuel bill). And this is not to mention where cyberspace is taking us with the telecommute - we know of nothing that can achieve better access-efficiency than that.

Phil McDermott said...

Thanks Andrew. Always good to get your feedback.

I take your point on enabling residential choice. Robert Bruegmann in "A Sprawl: A Compact History" (University of Chicago Press, 2005) demonstrates just how pervasive the desire for space is.
Of course, there will always be those who can afford to purchase housing in environments where they have good access to open space - either in their yard or in the public domain. They usually accept higher commmute costs (and these may go higher if we ETS) as the price of the greater benefit they derive from their preferred environment. And they may well make offsetting travel savings on their recreational and social trips.
Those that cannot afford "low density" housing end up in apartments or high rises in areas that tend to be heavily trafficked and lacking in public resources like parks, etc. we know the social consequences in the housing estates that have arisen from this type fo development in Europe and the US in the late 20th century. Hopefully we can avoid the worst outcomes of increasing housing densities without regard for who will live in them and how.

There's nothing wrong with medium or even high density living if it is based on choice and comes with a decent level of amenity. I have happily lived in high and medium rise apartments in overseas cities and made full use of public transport because it suited my circumstances at the time. My choice now is to live out of town with more space, work from home, and reduce my transport demands.
I'm lucky, I can make that choice. But I don't see why we should make it too hard for others who may want to operate outside some planning-prescribed norm. (I have always found it ironic when the advocates of high density living themselves live in places like the Waitakere Ranges, Waiheke Island, or the Hibiscus Coast on Auckland's fringes).

Treating the housing market as undifferentiated and apartments, units, terraces, and bungalows as perfectly substitutable is hardly a sound basis for policy making.

Andrew D Atkin said...

Thanks for your reply Phil. And very well said.