Every city tells a story. More specifically, the geographical shape of an urban area is key to understanding its economic development. Is it compact and circular, making life more efficient for residents and businesses? Or is it marked by spatial sprawl, resulting in longer commutes? Mariaflavia Harari, Wharton professor of real estate, explores this notion in her forthcoming paper, “Cities in Bad Shape: Urban Geometry in India.” Harari uses satellite images that show the shapes over time of 450 cities in India and compares that with economic indicators, such as housing rents and wages, to paint a revealing portrait about urban growth that has implications for policy makers as they plan for the future.

Tweaking the Circle

I work on urbanization and developing countries, and in my recent research I have looked at the spatial patterns of urbanization in India, trying to understand the economic implications of different city structures. If you look at urban areas from an airplane, for example, you can see that cities come in different shapes. Some have outlines that are roughly circular; others appear more fragmented, and yet others are constrained by their geographies. For instance, cities on islands or peninsulas end up having even more irregular shapes. My research question was what influence, if any, does a city’s shape have on the location choices of consumers and firms? Do consumers and firms benefit in terms of welfare or productivity from locating in cities with particular shapes?

Why would we expect city shape to matter at all from an economic standpoint? As urban planners have long recognized, city shape is one of the determinants of urban commuting efficiency, along with more widely studied factors, such as infrastructure or travel demand. More compact, more circular city geometries are conducive to shorter, within-city trips. To give you a sense of how this works, think of a circle. Now imagine that, holding the area constant, you start tweaking the circle, turning it into some irregular shape. As the shape of this polygon departs from that of a circle, becoming less and less compact, the average distance between points within this polygon will tend to increase. The same thing happens to cities as they expand in space.

“As urban planners have long recognized, city shape is one of the determinants of urban commuting efficiency, along with more widely studied factors, such as infrastructure or travel demand.”

A good, real-world example of these types of interactions is the comparison between Calcutta and Bangalore. Of the top cities in India, Calcutta is the least compact one. It kind of looks like an upside down giraffe head with a long, north-south tentacle. Bangalore, on the other hand, is more like a pentagon, and it’s the most compact one. I calculate that, all else being equal, if Calcutta had the same compact shape as Bangalore, the average within-city trip, proxied by the average distance between points within the city, would be shorter by 6.2 kilometers (3.8 miles). This is not trivial if you think that the average commuting speed in the top cities in India is around 15 kph.

The View from Far Above

If you look at a picture of the Earth at night taken from a satellite, urban areas are clearly visible as patches of light. I looked at the outlines of over 450 cities in India based on how they appear in nighttime satellite imagery. For each city, for each of these patches of light, I computed a number of indicators for geometric shape that are basically telling us the extent to which a city’s shape departs from that of a circle. I did this for multiple years to have a sense of the dynamic evolution of city shapes over time. And finally, I linked city shape to economic outcomes measured at the city level, such as population, average local wages and average housing rents. I found that households located in more compact cities are better off. First, more compact cities attract larger populations. Second, households that are located in more compact cities are implicitly paying a premium in terms of higher housing rents and net of local wages in order to live in cities with better geometries. This tells us that consumers are valuing city compactness as an urban amenity that they are willing to implicitly pay for. And this implied premium that they are paying is actually quite large. I estimate that households are trading off implicitly 5% of their income in order to live in cities that are more compact by one standard deviation. Now for the average city, a one standard deviation improvement in compactness implies that your average round-trip commute decreases by 720 meters. These are potential commute estimates based on the distribution of points within the city; these are not observed commutes that I can see in my data.

“Consumers are valuing city compactness as an urban amenity that they are willing to implicitly pay for.”

On the other hand, firms are not affected in their productivity by city shape. Compact cities are just as productive, and compactness appears to be a pure consumption amenity. Now, you may worry that compact cities are systematically different from others in some dimension other than geometry. What if these more compact cities are also more successful, more developed cities that consumers like for reasons that are just correlated with geometry, but have nothing to do with it? To address the fact that these confounding factors might be biasing my results, I’m not simply making a crude comparison between compact and non-compact cities in my analysis. I look at what happens when a given city becomes less compact as a result of hitting some topographic obstacle, like a mountain or a lake, as it expands in space. What I find is that when a city ends up with a worse geometry because of hitting some geographic constraints, its population growth slows down, and housing rents and local wages tend to decrease.

Growing up Versus Growing Out

India is predicted to urbanize very rapidly in the next decades, and local policy makers need options to accommodate this urban growth. My findings suggest that accommodating this urban expansion in a more or less compact way can have potentially relevant implications for household welfare and urbanization patterns. Clearly, not all cities can be circular due to topography. In fact, at a descriptive level, one of the patterns that I find is that cities tend to become less and less compact over time as they expand. But there is room for land-use regulations to promote more or less compact development. As an example of this, in my work, I consider a regulatory tool that is quite controversial in India right now – vertical limits. These are restrictions on building heights that in India are considered very strict, relative to international standards. A lot of commentators and local policy makers are concerned that these stringent vertical limits are inducing sprawl and making cities spread out more. What I find is that cities that have more restrictive building height regulations end up having footprints that are not only larger but also less compact relative to what their topography and predicted growth would suggest. Relaxing these constraints would be a way at least to slow down the deterioration in city shape that city growth entails.

A second policy implication could be related to investments in urban infrastructure. One of my findings is that that negative consequences of bad city geometry are more pronounced in cities that have a poor road network. So, providing road infrastructure might be a way to compensate consumers for the poor shape of the city that they live in.

Digging Deeper into the Data

So far in my research, I have looked at city-level aggregate outcomes and compared the outlines of cities, looking at cross-city comparisons. What I’d like to do next is take this to a more disaggregated level and look at what happens within cities, look at the internal structure of cities. As a follow up to this project, it would be important to understand who bears the costs of longer commutes in cities that are not compact. What categories of consumers are affected the most? Perhaps it is low-income households that don’t have access to individual transportation. What consequences, if any, does city shape have on residential patterns, and in particular, residential segregation by socio-economic status? Are some kinds of city shapes more conducive to segregation and others more conductive to integration?