The Motu Institute recently published new research into the urban productivity premium in New Zealand, or the degree to which firms and workers in big cities tend to produce more and earn higher wages. This is an essential issue for urban and transport policy as it gets to the heart of why we have cities. As we’ve discussed in the past, cities offer opportunities for better connections between firms, workers, and customers, leading to better economic outcomes. (Economists usually describe this as agglomeration economies.)
In the paper – with the enthralling title of “Urban productivity estimation with heterogeneous prices and labour” – researcher Dave Maré sets out to update and extend some of his previous work on the topic. His new research investigates two issues that might bias estimates of the urban productivity premium:
- Imperfect competition in small markets: Firms in large cities face more competition and hence will tend to have less market power (ie ability to jack up prices) than firms in small cities. This tends to result in low estimates of the urban productivity premium.
- “Sorting” of skilled workers into cities: People with higher skill levels – which could mean more education, more experience, or better ideas – tend to gravitate towards cities. (Similarly, cities tend to have different mixes of firms and industries.) Not controlling for this can result in a high estimate of the urban productivity premium.
Even after controlling for these factors, Maré finds evidence of a non-negligible productivity premium in Auckland. That is,
We document an urban labour productivity premium, with Auckland firms having labour productivity that is 17.9% higher than that of firms in other urban areas, and 17.0% higher than firms in rural areas. Some of this premium is due to the mix of industries in different cities. Auckland has a disproportionately high share of employment in industries that have above average labour productivity. Adjusting for this composition reveals a smaller, but still sizeable, premium of 13.5% relative to other urban areas, and 11.3% relative to firms in rural areas.
Here’s a chart showing how other parts of New Zealand compare to Auckland in terms of productivity, after controlling for industry mix, workers’ skill levels, imperfect competition, and a range of other factors like firm size. (This chart is based on the first column in Table 3 of the paper.) As you can see, firms in Auckland are more productive than firms in other parts of NZ, with Wellington (4.2% less productive) and Tauranga (9.4% less productive) being closest to Auckland.
It’s worth noting that Maré’s new estimates of Auckland’s productivity premium are considerably smaller than his previous ones. In a 2008 paper, he estimated that firms in Auckland were around 51% more productive than firms elsewhere in New Zealand. These are obviously very different numbers! But, as explained in an appendix, the majority of the differences are due to different procedures regarding data selection and processing.
Notwithstanding the exact number, Maré’s new analysis raises a few important conceptual questions about the urban productivity premium. His analysis shows that a large share of the difference in productivity between big cities, small cities, and rural areas is down to the fact that skilled workers tend to sort themselves into cities. When we control for workers’ skill levels, we tend to get lower estimates of the urban productivity premium. Or, if you prefer that in economese:
The meta-analysis by Melo et al. (2009, Table 4) reports that studies that control for labour quality generally yield agglomeration elasticities that are 5 to 6 percentage points lower than studies that do not. In the current study, labour quality adjustment lowers the estimated agglomeration elasticity by 0.057 (from 0.079 to 0.022).
However, I’m not sure it is totally appropriate to adjust for skill levels, as it’s possible that cities’ ability to attract and retain talented, innovative, and motivated people (and productive firms) is in fact a type of agglomeration economy. In other words, we might be controlling away the effect of interest!
Open migration between Australia and New Zealand means that people who can’t find an appropriate place (urban and economic) in New Zealand can easily go to Australian cities. So the alternative for skilled people who are dissuaded from living in Auckland isn’t necessarily that they’ll go and start up a business in Hamilton. Instead, they might head across the Tasman, where their skills are totally lost to New Zealand.
What does this mean for urban policy in New Zealand? I’d tentatively identify two key ideas we might want to focus on in order to allow our cities to get better at attracting and retaining productive people and firms.
First, we need to think hard about whether our policies make it attractive for mobile people to come to (or stay in) our bigger cities. This is a key consideration for, say, urban planning reform, as high housing prices driven by constraints on housing development are an important barrier to people coming or staying. Evidence from the US suggests that, if left unaddressed, high house prices can systematically dissuade people from moving to productive places where they can put their skills to best use.
Second, we also need to think about how to preserve and enhance the amenities that are on offer in New Zealand cities. Our relatively clean air, reasonably well-preserved coastal environment (clean beaches, marine reserves, etc), and accessible forests and natural parks are important attractions, but other areas are letting us down. For instance, rural and small-town water quality is rapidly declining due to expanding dairy farming.
More relevant to transport, street design in New Zealand is pretty retrograde, leading to a lack of high-quality public spaces where people actually want to be. All too often, we insist upon shoving cars down corridors, heedless of the fact that some streets have higher value as places to be. But we know we can do better: places like O’Connell St and Wynyard Quarter give many people joy on a daily basis.
What do you think about the urban productivity premium? And how can we get more of it?
Last week, I took a look at the contribution of agglomeration to Auckland’s recent economic growth. Based on observed changes to employment density over the period, plus agglomeration elasticities calculated by David Maré and Daniel Graham, I estimated that 11-12% of Auckland’s recent productivity growth was due to increased urban scale and density.
The gains from agglomeration since 2000 are significant: Auckland’s GDP is approximately $1.4 billion larger as a result. Ultimately, productivity gains are good for everyone. If you’re retired, they help to pay your pension. If you’re in school, they help pay your teachers and living costs. In between, they help fund your health care and pay for your neighbourhood library.
But is agglomeration simply a consequence of urban scale, or does urban form also matter? In other words, are there any reasons that we should prefer one distribution of employment within cities to another?
There are a couple of ways we can address it. One would be to gather data on the spatial distribution of employment in a range of cities, and examine the impact on productivity. This is, implicitly, what Maré and Graham (and other economists studying productivity) have done by measuring effective job density and productivity at a highly detailed level and comparing outcomes within and between cities.
Other papers use a slightly different methodology but also come up with suggestive results. For example, a 2014 paper by Daniel Chatman and Robert Noland finds that public transport provision can encourage agglomeration economies in dense city centres. From the abstract:
Using data on US metropolitan areas, this paper traces the links from transit service to central city employment density, urbanised area employment density and population; and from these physical agglomeration measures to average wages and per capita GMP. Significant indirect productivity effects of transit service are found. For example, in the case of central city employment density, estimated wage increases range between $1.5 million and $1.8 billion per metropolitan area yearly for a 10 per cent increase in transit seats or rail service miles per capita. Firms and households likely receive unanticipated agglomeration benefits from transit-induced densification and growth, and current benefit–cost evaluations may therefore underestimate the benefits of improving transit service, particularly in large cities with existing transit networks.
Another approach would be to simulate the impact of alternative employment distribution on effective density and hence on productivity. Last week, I looked at how Auckland’s job density had changed from 2000 to 2015. This week, I’m also going to consider a simple simulation: What if the city had grown in a different way between 2000 and 2015?
According to Stats NZ’s Business Demography data, Auckland added approximately 174,000 jobs – a 33% increase – over this period. Here’s a chart showing how they were distributed at a local board level. Waitemata local board added the most jobs (almost 41,000) followed by Upper Harbour (22,000) and Howick and Maungakiekie-Tamaki (both around 19,000).
By contrast, the rest of the isthmus and lower North Shore saw exceptionally low rates of change. (A pattern that is matched in population growth: high growth in the city centre, Howick, and Upper Harbour, where planning rules have enabled growth, and low growth in other areas where they’ve prevented it.)
But what if Auckland’s recent employment growth had followed a different pattern? For example, what if we’d chosen to decentralise employment growth to a new “edge city”?
Let’s set aside, for a moment, the fact that creating new employment centres by planning fiat tends to disappoint. (As demonstrated by Manukau’s underwhelming history, and possibly also the new NorthWest mall.) If businesses don’t see an advantage in locating there, it won’t happen.
As a benchmark, consider a new “edge city” in Drury, which is currently a set of paddocks that are conveniently located along the Southern Motorway and the unelectrified portion of Auckland’s Southern Rail Line. Let’s assume that we succeeded in relocating a bit over 25% of Auckland’s recent employment growth to Drury – creating a new employment centre with 50,000 jobs. (Around half the size of the Auckland city centre.) All other areas of Auckland would have seen proportionately lower rates of growth.
What would that do to the city’s potential for agglomeration?
Here’s a map comparing the effective density of employment under this “edge city” scenario with the actual effective density of employment in 2015. (Remember, effective density is a measure of an area’s potential for agglomeration economies, as areas that are accessible to more jobs are likely to be more productive.)
Areas shaded in yellow would experience reductions in agglomeration potential, while areas in green and blue would be more accessible to jobs under the “edge city” scenario. Notice how most of the city is coloured yellow. The only places to benefit from this change are the areas immediately around Drury.
Overall, the job density gains around Drury are far outweighed by the “deglomeration” experienced by the rest of the city. Shifting employment growth to an “edge city” in Drury would have reduced the city’s overall job density by 9%. (In 2015, the average Auckland employee was accessible to around 92,000 other jobs. Under the “edge city” scenario, they’d only be accessible to around 84,000 jobs.)
This would in turn reduce the city’s economic productivity. Based on Maré and Graham’s measured agglomeration elasticity of 0.065, I estimate that Auckland’s productivity would be 0.6% lower under the “edge city” scenario. (Again, using an arc elasticity formula: (92,000/84,000)^0.065-1.)
Because Auckland’s GDP was $88.3 billion in 2015, the productivity losses from deglomeration would equate to roughly half a billion dollars a year. That’s a lot of money. For comparison’s sake, $0.5 billion is roughly equivalent to:
Of course, more centralisation is not always optimal. In a large city, there are good reasons for businesses to spread themselves around a bit. Retailers want to be close to local shoppers, warehouses want to be on cheap land close to transport infrastructure, and so on and so forth. And, from a policy perspective, adding peak transport capacity to enable existing centres to grow may be costly.
But, as this simulation of a Drury “edge city” shows, forcing decentralisation is likely to be highly sub-optimal. Auckland would be less productive if it had chosen to push employment growth into outlying centres, rather than accommodating it in the city centre and other locations throughout the city. Over time, that would translate into lower competitiveness for local businesses, lower wages for Auckland workers, lower living standards for residents, and worse public services and infrastructure.
Do you think urban form can contribute to a productive, happy city?
We’ve written quite a bit about agglomeration economies, as they’re one of the most important forces shaping urban life. Agglomeration economies refer to the benefits of proximity for economic and social interaction – when you’re around more people, it’s easier to meet the right person (for business or relationships!), easier to share knowledge, and easier to do things in general.
One “stylised fact” from the economic literature is that cities that are larger and better connected – i.e. denser and/or easier to get around – tend to be more productive. When it comes to economic performance, size matters. This benefits firms and workers, of course, but it is also good for consumers. For example, if you want to see a lot of rugby tests, you’re better off locating in Auckland than in Taupo, because test matches tend to go to where the people are. And if you want more restaurants and groceries, live in a denser neighbourhood.
However, economists have focused on agglomeration economies in production as they’re often easier to measure. For example, a few years ago the New Zealand economist David Maré estimated an “Auckland productivity premium” of around 50%. That means that firms located in Auckland are around 50% more productive than similar firms located in other regions. The premium was even higher for Auckland’s city centre.
In subsequent work, Maré and a collaborator, Daniel Graham, estimated that New Zealand had an “agglomeration elasticity” of around 0.065 (averaged across all industries). What this means is that places that are 10% denser tend to be around 0.65% more productive, all else equal.
But what does this mean in practice? How much does agglomeration contribute to the New Zealand economy? Is it a big deal, or not that important in the grand scheme of things?
To get a rough idea, I calculated changes in the “effective density” of jobs in Auckland over the period 2000-2015, taking into account the location of jobs within Auckland (from Stats NZ’s Business Demographics data) and the distance between job locations (calculated using GIS tools). I followed Maré and Graham’s formula for job density as a function of weighted distance to jobs in nearby areas – for the precise formula see equation 2 on page 12 of their paper.
Here’s a map showing how effective density of jobs changed over the whole period. Darker blues indicate higher percentage increases.
Almost everywhere in the Auckland urbanised area became more accessible to employment over this period – the rising tide of urban development lifted all boats. On average across the region, effective density rose by 29%. However, increases were faster around Albany and the upper North Shore, which saw rapid development, and slower in the western isthmus and west Auckland.
So things have changed quite a lot. The following chart shows that these changes happened incrementally over time. It shows the effective density of employment for the average job in Auckland. In 2000, the average job was proximate to around 71,000 other jobs. In 2015, that had risen to slightly less than 92,000 jobs.
So job density has gone up quite a lot over the last 16 years as a result of Auckland’s growth. What effect has this had on productivity?
As a point of comparison, I estimated changes in GDP per employee over the same period. (I used Stats NZ’s employment data, regional GDP statistics, and GDP price deflators for the whole country. This isn’t a perfect estimate, as I’ve excluded self-employed people and haven’t corrected for part-time employment, but it’s not miles off.) Here’s what that looks like. Over the entire period, GDP per employee has risen by approximately 14.4%. The city’s economy currently produces around $88.3 billion in output.
By comparison, Maré and Graham’s agglomeration elasticity of 0.065 implies that a 29% increase in job density is associated with a 1.7% increase in productivity (calculated using an arc elasticity formula: (92,000/71,000)^0.065-1). The true figure may be higher, as Auckland is specialised in industries that benefit more strongly from agglomeration economies.
- Roughly 11-12% of the total productivity growth in Auckland over the last 16 years is due to increased agglomeration economies
- In the absence of increased agglomeration economies from scale and density, Auckland’s economy would be around $1.4 billion smaller.
A wide range of other factors – increased skills, investment in capital goods, improved business practices, and changes to the composition of Auckland’s economy – also make important contributions to productivity growth. However, the contribution of agglomeration is significant – both in dollar terms and as a share of the city’s overall productivity growth. In the long term, those tenths of a percent add up to quite a lot. If we want a wealthier New Zealand, we need better, more productive cities.
Policymakers can do a lot to enhance – or undermine – agglomeration economies. For example:
What do you think about agglomeration economies in Auckland?