Summary / tl;dr

Increasing urban density could be an important cause to work on for people engaged in urban planning and policy. However, uncertainty is large regarding the magnitude of the effect.^[1]

Density is important

The geographic organization of humanity can take many forms, with profound implications. At one extreme, we could all live in places as dense or denser than Manhattan. At the other extreme we could all live in endless suburbs that would cover a large and ever growing fraction of the habitable land area of the planet, replacing vital ecosystems in the process. How densely people live affects their energy use, time use, wellbeing, productivity and access to services.^[2] It is also path dependent - once land is converted to residential use, it tends to remain in residential use as long as it remains possible to live there.^[3] Therefore, current choices will have strong effects on the future geographic organization of humanity. 

Policy determines density

Where people live is determined by demand and supply: where people want to live and where housing is affordable and available. In many nations, the construction of new housing in existing cities is restricted through various legal means, primarily for the benefit of current homeowners in these cities. Restricted supply is thereby displaced to lower demand areas on the outskirts of cities, where dense construction is less profitable. This may result in a suboptimal geographic organization which may justify policy involvement. 

There is reason to believe the resulting organization is far from optimal. To see why, note that there is a strong positive correlation between population density in a city, population size in a city and rent prices in a city. In particular, rent on a typical apartment in one of the densest cities is far above the economic costs of supplying an additional such apartment. This implies that the supply of such apartments is restricted to be well below demand for such apartments.^[4] Therefore, in a standard model of supply and demand without externalities, increasing the supply of such apartments would greatly increase total welfare. 

Density benefits others

Such a calculation only considers private benefits by assuming there are no externalities. That is, that the choice of living in a dense city has no effect on the welfare of other people. However, recent empirical evidence suggests that the externalities of density are positive. That is, adding a person to a dense city makes other people better off compared to not adding that person, all else equal.^[5] The external benefits from density seem to stem primarily from 4 factors:

• lower energy usage, 
• lower time usage, 
• lower land usage, 
• and increased innovation. 

In particular, the fact that increasing density reduces both distance travelled and travel time implies that congestion, the main negative externality of density, is dominated by the increased transportation efficiency of density. The other external benefits thus constitute pure gain to society. 

An important caveat is that current estimates of the external benefits of density are based on comparing more and less dense cities in the current distribution of city density among the largest cities in advanced economies, and mostly in the US. This has two main implications. 

1. It is very likely that there is some level of density beyond which this would no longer hold. However, existing estimates suggest that we are not there yet as discussed above in relation to congestion. This implies that even if every city in the US was made to be as dense as NYC, externalities from density would still be positive.
2. In addition, large cities in advanced economies tend to have high quality infrastructure to support density. While such infrastructure tends to pay for itself quite quickly, some countries that lack the knowhow and access to finance to build such infrastructure may have greater difficulty in supporting greater density. This implies a need for caution when advocating for density in poorer nations but also implies a scope for philanthropic work in developing the needed capacity. 

An important cause area?

Both the average individual and the public as a whole would benefit from letting many cities be much denser than they currently are. However, all the above does not mean that EA should get involved. To argue for EA involvement, we need to show that such involvement would be effective and existing evidence does not suggest a high level of effectiveness, for several reasons

1. The external benefits are not very large, at least based on estimates from the US. Doubling density in the US would reduce net carbon emissions by about 7% and increase growth by at most 1.76% but probably by only 0.037% to 0.22%.^[6] 
2. The problem is not very tractable because of entrenched political opposition from insiders. 
3. Wealthy nations should be able to increase density and capture a large part of the benefits. For density to be beneficial in poor nations, their organizational, legal and physical infrastructure needs to be far better compared with current levels. 

To summarize, there are many reasons to be cautiously optimistic that increasing urban density could be an important cause to work on for people engaged in urban planning and policy. However, it is still not clear what exactly is the magnitude of that effect, and whether it should be generally recommended.

1. ^{^}

   A previous post on this topic emphasized the GDP lost to land use reform and outlined several courses of action. In contrast, we emphasize externalities such as effects of urban density on innovation and on GHG emissions, which are informative about the long term net benefits to humanity of higher density.

2. ^{^}

   Duranton and Puga 2020 and Ahlfeldt and Pietrostefani 2019 are two recent reviews of the accumulated evidence.

3. ^{^}

   Davis and Weinstein 2002 find that the distribution of population is highly persistent over the 6 thousand years and that even the devastation of Japanese cities during the last year of WWII had no lasting effects on where people live. They report that the populations of the typical Japanese city completely recovered within 15 years of the end of the war and that Hiroshima and Nagasaki both exceeded prewar population levels within a decade of being bombed with nuclear weapons. More prosaically, Diamond 2016 finds that people in the US have a strong preference to live near where they grew up. Gleaser and Gyourko 2005 document an almost perfect correlation between the number of housing units and the number of residents in a US city.

4. ^{^}

   Glaeser and Gyourko 2018

5. ^{^}

   Duranton and Puga 2020 and Ahlfeldt and Pietrostefani 2019

6. ^{^}

   The net carbon emissions estimate is from Ahlfeldt and Pietrostefani 2019; The growth estimates are arrived at by using the estimated effect of density on patents from Ahlfeldt and Pietrostefani 2019 and the estimated effects of patents on productivity from Kogan et al 2017 (0.037% to 0.22%) and Balasubramanian and Sivadasan 2011 (1.76%). I give greater weight to Kogan et al 2017’s estimates because they have a credible causal specification and because they calculate the effect of patents on total factor productivity, while Balasubramanian and Sivadasan 2011’s strategy is not causal and they calculate the effect of patents only on labour productivity which is difficult to related to growth without further details which they do not provide.