Building the city

20th Sep 2017

Author: Tanner Regan

A large share of a nation’s capital stock is in its buildings. Kenya’s largest city, Nairobi, is rapidly reconfiguring its built capital and doing so in astonishing quantities – a process that comes overwhelmingly with issues of land market informality. Interesting questions arise in such a context. How does the built environment of a city evolve? Is there a role for slum-style building technology in effective urban development? What are the costs of informality and poor institutions in land markets? In this paper we provide both novel empirics and develop a theoretical framework to answer these types of questions. 

The goals of our research are three-fold:

First, to analyse Nairobi’s built capital in 2003 and 2015 using unique data on buildings and to document the stylised facts on development and redevelopment for a major developing country city. Between 2003 and 2015, the volume of the city core increased by 50-60%, largely due to increasing heights of redeveloped buildings. We breakdown where the action is and in what ways so much volume is being added to these places.

Second, we develop a theoretical framework of the built environment over time and space that incorporates issues of informality. Here we consider two important aspects of slums: construction technology and tenure issues. Our main distinction between slum and formal sectors is differences in building technologies, rather than tenure status. Slums are characterised by collections of tin huts, a construction technology that can be efficient even under well-functioning land markets. We introduce tenure issues by considering that the formalization process requires a fixed cost that can vary across space and inhibit the conversion of building technology from slum to formal.

Our third and final goal is to calibrate the model, allowing us to formally calculate the welfare cost of institutional frictions present in Nairobi. Fixed costs to formalization are shown to delay formal development, preventing high value areas from being redeveloped to a more intensive use of capital, and leading to huge losses in the efficiency of land use. We estimate that slum land within 6 kilometres of the centre of Nairobi has passed its optimal date for formal development if the formalization process were to be costless. Overcoming the formalisation costs today, the conversion of slum land 3-4km from the city centre corresponds to a gain of roughly $9,200 for each household currently living on this land.

Facts from a city in flux

Our data is based on two cross sections that detail each building’s footprint in Nairobi for 2003 and 2015. For the later year, heights are measured and we augment the early period by interpolating the heights of buildings that were knocked down from nearby buildings that were unchanged. To examine the type of change a building undergoes, we apply an algorithm to the raw data to classify buildings as either unchanged, redeveloped, infilled, or demolished. A more in depth discussion can be found in this earlier blog post. In Figure 1 we map the extent of the city in each time period based on a minimum level of built land cover. This is the first attempt that we know of which uses city-wide data on individual buildings to detail the changes in the urban landscape.

 

fig1

 

As mentioned earlier our data shows massive changes in the built volume of the city. In particular, total built volume increases by 48% on the intensive margin (inside the dashed black line in Figure 1), and 120% on the extensive margin (in between the dashed and solid black lines in Figure 1) for a total increase of 59% across the 2015 extent of the city. Interestingly the annualized growth is about 3.9% which is very similar to annual population growth between the 1999 and 2009 censuses. There are also large changes with respect to the number of buildings; for instance, inside 3km from the city centre, 35% of building were torn down in the last 12 years - an astonishing number compared to developed country standards where even 10% would be high. Over the study period, the formal sector increased by 60 percentage points, modestly more than the 55 percent growth of slums; however, slums account for a much smaller share of Nairobi’s total area and only account for about 9.2% of the increase in total volume. When we compare areas based on distance to the city centre, we see that until around 9km out, relative increases in the formal sector generally dominate those from slums - demonstrating the greater role of the formal sector in the main part of the city. Instead, slum changes tend to dominate at the city edge where land is cheaper.

We also break down volume increases based on the process of construction: whether buildings were torn down and redeveloped, infilled on empty land, or demolished. As can be seen in figure 2, formal volume change from redevelopment dominates that from infill until about 4.5km after which infill begins to account for more. Finally, demolished buildings occur throughout the city at low levels of relative volume change.

In the 2015 cross section of the city, volume intensity is very similar across sectors; both slum and formal have volume gradients that are high near the centre and decline towards the fringe. However, each sector delivers volume in different ways; formal by height and slums by crowding short buildings.  Figure 3 reveals that similar differences arise in the dynamics; formal sector buildings that are knocked down tend to be redeveloped into taller buildings, while in slums the increase in volume comes primarily from an increase in cover. In the formal sector, heights of new builds are increasing across the city, but most prominently in the range of 1-5km from the city centre where increases are over 50%. While the average height for slums does increase, the change is only slight, and never above two stories. To make up the remaining volume increases, slum buildings must crowd heavily in empty areas.

 

fig2

Most existing urban theory models are not well suited to the types of cities we are interested in. They cater to the developed world; they are static, and they fail to account for informality. Cities in the developing world, however, particularly African ones, are extremely dynamic and often have large and complex informal sectors.  We build on previous theory starting with a dynamic monocentric city model similar to Braid (2001) with durable capital by introducing the idea of informality with two different building technologies and formalization costs. The city is monocentric, with higher prices near the centre, and assumed to be growing so land prices at every location are rising. A profit maximizing developer will then allocate more built volume to higher valued locations with a choice of building either slum or formal property.

One aspect of informal settlements is the physical construction materials. In Nairobi’s slums the majority (about 55%) of housing walls are corrugated iron sheets, while a large share is made from mud and organic materials (about 20%). Formal sector housing, on the other hand, is predominately (over 90%) made of stone, brick, or block. In our model, formal sector buildings are analogous to ‘putty-clay’ technology: they can be built high, but are durable meaning height cannot be added unless the building is torn down and built from scratch. Slum buildings are much more flexible; their production can be thought of as the rental on Meccano parts that can be reassembled without cost, but because of their weaker materials cannot be built tall. In the model, formal sector neighbourhoods increase volume by building taller, while slums do so by crowding and increasing cover.

Another important aspect of slum settlements is the informality of the land market. Without formal title, large investments in durable structures are risky and hard to finance. Establishing property rights and formalising the land can be difficult and costly due to institutional constraints and the particular history of a piece of land. In our model, the first instance of formal development on a piece of land requires a fixed cost depending on the location. A higher cost of formalisation delays the conversion of land use from slum to formal since more time is needed for growing fundamentals to push up formal sector rents and trigger formalisation. When the delayed development finally occurs, it will occur at a higher rent and hence volume as well.

What can we conclude from the theoretical framework? As the city and land prices grow, areas developed with informal technology continuously increase cover. On the other hand, formal sector buildings are periodically demolished and redeveloped to higher heights as the city grows. Absent formalisation costs (D=0), new slums are all built at the city edge where land is cheap and interior slums are redeveloped at appropriate points into formal sector buildings. Figure 4A, details the expansion of volume per unit land in both sectors along distance from the centre and time.

Figure 4B shows how introducing formalisation costs delays the conversion of certain areas, we impose costs on the land from 10-15km from the city centre, delaying the length of time that slums are built on this land. A final experiment imposes random formalisation costs across the city creating what we call the ‘hotchpotch’. Figure 4C shows volume in a city with random formalisation costs. The pattern resembles iconic photographs of side-by-side formal high rise and slum low rise buildings that often appear alongside discussions of urban inequality.

 

fig3 new

 

Formalisation costs and welfare analysis

Combining our data and empirical framework we estimate parametres from the model and calculate the implied welfare cost of institutional frictions. In table 1 we consider different dates (z) of conversion to formal land along kilometre wide rings from the city centre. The final row gives the optimal date of conversion based on our formulation of the present value of land rent. Each ring of slums inside 6 kilometres has an optimal date of conversion before 2015, the point in time which the data is calibrated. The largest losses in welfare are in the 3-4km ring from the city centre. For example, the cost of perpetual informality as compared to the optimal date (z=2000) of conversion in this band is $790-$409=$381, that is about half of the maximal value is lost. In this band there are 1.13 km2 of slum land, redeveloping 1km2 to formal sector (leaving 10% to be used as public space like roads and schools) results in an aggregate gain of $268 million. This corresponds to roughly $9,200 per household of the 29,000 households living in slums in this band. These households are typically spending $300 a year on housing, leaving a lot of room for sharing these gains.

Table 1: Present value of land rent  ($/m2)?

Date of formalisation, z

3-4 km

4-5km

5-6km

6-7km

7-8km

z = 2015

767

645

542

456

384

z = 2065

600

525

459

402

353

z = ∞

409

369

332

299

270

z = optimal z

790

652

543

457

387

Optimal z (year)

2000

2005

2011

2017

2023

 

Concluding remarks

This paper examines building development and redevelopment in a growing city and the welfare costs of institutionally created land market frictions. Our model considers a growing city where formal buildings are durable, and informal are not. Building volumes rise towards the centre and over time; discretely in the formal sector as redevelopment involves building taller with durable technology and continuously in the informal sector where the flexible technology can be added incrementally. We calibrate the model with a unique data set on the built environment of Nairobi for 2003 and 2015 by estimating key parameters, finally formulating a measure of the welfare costs of institutional frictions in land markets, which plague many cities in the developing world.

In Nairobi there has been major redevelopment of 2004 formal sector buildings into taller new buildings, driving 50-60% increases in volume. There is high intensification of land use with infill of new buildings further from the city centre.  However, while this dynamic development is occurring, there remain persistent slums in the mid-city with little change in the way of heights in these areas. Applying our calibrated model indicates that the cost of this inefficient land-use is high. Poorly functioning land market institutions dramatically alter the built fabric of the city, creating a building hotchpotch and significant welfare losses. Nairobi continues to grow along with many emerging urban centres in sub-Saharan Africa, making institutions and urban policy today critical for hundreds of millions of people in the future.