Conference: IGC cities research group

The IGC Cities research programme held a conference from Thursday 21 – Friday 22 May 2015 at the London School of Economics and Political Science (LSE), led by Edward Glaeser, an IGC Cities Research Programme Director.

The conference showcased the best current work on cities and development, and provided a forum in which leading academics and policymakers could come together to set an agenda for future research.

Day 1 (Thursday 21 May)

Leading researchers presented seven academic papers, concerning cities and urbanisation in developing countries. These papers are listed below:

Time Paper Presenter
0830-0930 “Creative destruction: Barriers to urban growth and the great Boston fire of 1872” Richard Hornbeck (Harvard University and NBER)
0930-1030 “Location, search costs and youth unemployment: A randomized trial of transport subsidies in Ethiopia” Simon Franklin (Oxford University)
1100-1200 “Economic development and the spatial allocation of labour: Evidence from Indonesia” Gharad Bryan (LSE)
1200-1300 “Shelter from the storm: Upgrading housing infrastructure in Latin American slums” Sebastian Galiani (University of Maryland)
1400-1430 “Political Economy of Cities” Abhijit Banerjee (MIT)
1430-1530 “Cities in bad shape: Urban geometry in India” Mariaflavia Harari (MIT)
1530-1630 “Transport infrastructure, urban growth and market access in China” Matthew Turner (Brown University)
1700-1800 “The geography of development: Evaluating migration restrictions and coastal flooding” Klaus Desmet (Southern Methodist University)
1800-1930 Expert panel discussion: Methods, experiments, and data Nathaniel Baum-Snow (Brown University)
Marcel Fafchamps (Stanford University)
Stephen Redding (Princeton University)

 Day 2 (Friday 22 May)

Time Paper Presenter
0830-1030 Expert panel discussion: South Asia Giles Duranton (Wharton, University of Pennsylvania)
Mushfiq Mobarak (Yale School of Management)
Ijaz Nabi (Pakistan and IGC)
Taibur Rahman (Bangladesh)
1100-1300 Expert panel discussion: Africa Vernon Henderson (LSE)
Alexander Moradi (University of Sussex)
Tony Venables (University of Oxford)
Hugh Wenban-Smith (Independent researcher)
Mayor Sam Gibson (Sierra Leone)
Robert Kyukyu (Uganda)
1400-1600 Expert panel discussion: Public finance and urban growth Tim Besley (LSE)
David Albouy (University of Illinois)
Adnan Khan (IGC)
Mihaly Kopanyi (World Bank)
Meembo Changula (Zambia)
1630-1830 Expert panel discussion: Threats to urban life Nava Ashraf (Harvard Business School)
Rudiger Ahrend (OECD)
Edward Glaeser (Harvard University)
Esteban Rossi-Hansberg (Princeton University)
Guy Michaels (LSE)


The expert panel discussions brought together a range of leading academics and policymakers to discuss African and Asian urbanisation issues, public finance and urban growth, and threats to urban life. The panels consisted of short research presentations, before a more general discussion of the key themes, and the way forward for policy and research.

Please find below a final programme.



“Creative destruction: Barriers to urban growth and the great Boston fire of 1872” (Richard Hornbeck)

In the second session Richard Hornbeck (Harvard University) presented his paper “Creative Destruction: Barriers to Urban Growth and the Great Boston Fire of 1872”. The paper analyses new plot-level data in the aftermath of the Great Boston Fire of 1872 to study externalities and transaction costs related to the transformation of outdated durable buildings. He estimates substantial economic gains from the created opportunity for widespread reconstruction by the fire.


In November 1872, a small fire spread through a large section of Boston’s business district, eventually destroying 776 buildings over 65 acres of the downtown Boston area. Boston firefighters were unable to stop the fire before it spread, due partly to sickness amongst the fire department’s horses that prevented the rapid deployment of equipment to the burning area. The fire burned for 22 hours, eventually stopping with the arrival of massive firefighting resources from surrounding areas. The fire killed 20 people and caused approximately $75 million in damage, or 11% of the total assessed value of all Boston real estate and personal property.


The empirical analysis in the paper uses a new detailed plot-level dataset, covering all the plots burned area and surrounding areas in 1867, 1872, 1873 and 1893.


  1. The land values increased substantially from 1872 to 1873 in the burned area, relative to the unburned area. These estimates imply economically substantial gains from the opportunity for widespread reconstruction, as individual landowners previously had the opportunity to replace their own building.
  2. Land values also increased immediately in nearby unburned areas, relative to further unburned areas. The nearest unburned areas received an increase in land value similar to the burned area, and the estimated impact declines until leveling at around 1400 feet (approximately 5-6 blocks or 25-35 buildings away).
  3. Building values also increased substantially in the burned area, following reconstruction, and converged over time. These impacts were greatest at the lowest quantiles of building values, reflecting replacement of the worst building stock, but building values increased even at the highest quantiles. Seen through the lens of the model, these results suggest that even the most recently constructed (and, therefore, the highest value) buildings were replaced with substantially better buildings, consistent with neighborhood externalities.

Policy Implications:

The historical episode of Boston fire highlights the challenges of maintaining economic growth with increasingly heterogeneous vintages of capital stock, to such a degree that the destruction of durable capital generated substantial economic gains. Positive externalities from capital replacement may be lower in other contexts, due to differences in the economic or policy environment, but these impacts will be salient in contexts with a return to coordinated investment.

By Vikas Dimble, Country Economist, IGC India Central

“Location, search costs and youth unemployment: A randomized trial of transport subsidies in Ethiopia” (Simon Franklin)

Do transport costs hinder young people in the developed world from finding good jobs? Cash constraints and large distances from jobs could be locking some individuals out of labour markets. In this presentation Simon Franklin uncovered the labour market frictions and inequalities that transport costs can create when job seekers are cash constrained. His results are evidence that reducing transport costs leads to better employment outcomes in the context of his study, and further provide an argument for policymakers to support dissemination of vacancy information as well as match-making institutions.

The study

Franklin studied the urban labour market in Addis Adaba, Ethiopia using a randomised control trial. Unemployment in Addis Adaba is high, currently at 24% for those under 30. Most young poor job seekers live outside of the city centre, while firms that offer skilled and professional jobs are clustered within the centre, and so is information about these jobs. Job search is therefore expensive, and costs applicants about 1$ a day. Applicants living far away from the centre have to run down their savings to finance their job search. Given cash constraints, these costs might be high enough to lead them to under-invest in search. To test this, Franklin randomly assigns non-fungible transport subsidies that, twice a week for 9-11 weeks, cover the cost of a return fare to the city centre for job applicants living further than 5km away. This allows him to analyze whether the applicants receiving the subsidy will on average (1) search more and (2) find better jobs. He formalises these hypotheses in a dynamic optimisation model that predicts large treatment effects even for very low levels of risk aversion.


  1. Reported savings decline over time for job seekers
  2. Transport subsidies prevent job seekers from giving up search
  3. Unemployed who were not actively searching for jobs in the absence of transport subsidies are induced to start search
  4. While subsidies are in place, job seekers who had been actively looking for jobs before the subsidy started are less likely to divert their efforts and engage in temporary work to finance their job search
  5. Impacts are largest among the poorest/ most cash constrained
  6. Impact on search persist after subsidies have ended
  7. No evidence for Hawthorne effects that might invalidate findings

Policy Implications

Franklin’s results carry policy implications of great import, particularly in the context of the big transport overhaul that the government of Addis Adaba is undertaking at the moment. For labour market outcomes to improve, it will be important for authorities to consider ways to keep transport costs low. In this study Franklin tests the effect of subsidies, but, as the discussant Bryce Millett Steinberg highlighted, other solutions could also comprise of loans or cash drops. Complementary to this, the government can improve the functioning of the labour market by providing match-making services of different kinds, starting with better and wider dissemination of job vacancies and going to counseling, guidance and job fairs.

By Sebastian Wolf, Country Economist, IGC Uganda

“Economic development and the spatial allocation of labour: Evidence from Indonesia” (Gharad Bryan)

Gharad Bryan (LSE) presented his research on the spatial allocation of labour and economic development, showcasing evidence from Indonesia. He began by observing the wide differences in nominal wages across space, and posed two questions. Firstly, would moving people across space imply increases in productivity? Here, the two main issues to consider are selection and the quantitative effect on productivity of such a move. Secondly, if the answer to the first question is positive, the question becomes about why people do not make such moves. Bryan referred to two possible explanations, namely amenity differences and the associated costs of movement.

Bryan highlights that the facts show that both selection and movement costs are present. He presented a model, consistent with these facts, to answer the questions, using US and Indonesian census data. The structural estimates show that movement costs have decreased over time (1995-2012) in Indonesia and that such costs are much higher than in the US. Bryan also finds that amenity is negatively correlated with productivity. Quantitatively, the results show that reduced movement costs and changes in amenity both separately account for approximately 25% of growth. Bryan explained that higher movement costs explain approximately 5% of the US-Indonesia gap.

Bryan concluded by explaining that if the heterogeneity observed in wages across space is due to costly migration or amenity differences, this may represent an opportunity and a role for policy. If, however, selection accounts for most of this difference there is less room for policy to have an effect.

Daniel Sturm (LSE), the discussant of the session, began by emphasising the importance of the issues discussed in this paper, stating that the implications of distortions in the spatial distribution of economic activity are under-researched but that this is a first-order question. Sturm relayed concerns regarding the way in which the model presented amenities, arguing that the model is unable to capture the empirical regularity that larger locations pay higher wages. According to Sturm, the key policy question is how government policy can reduce the estimated migration costs to achieve better job matches. He stresses the importance of understanding what underlies the effect of distance on migration costs prior to developing and implementing costly policy initiatives, such as building highways to reduce migration costs.

By Michelle Jacob, igc Hub Economist

“Shelter from the storm: Upgrading housing infrastructure in Latin American slums” (Sebastian Galiani)

What are the welfare impacts of providing urban slum-dwellers with decent, basic houses? In this session, Sebastian Galiani present results from an impact evaluation conducted in El Salvador, Mexico, and Uruguay, which aimed to answer this question.

The house contruction was led by the NGO “Un Techo Para Mi Pais”, or “TECHNO”, which held a lottery for slum dwellers, and provided ‘winners’ with simple, 18 square meter, houses. The houses had no indoor bathroom or kitchen, but were a step up in terms of hygiene, comfort, and durability.

The researchers measured slum dwellers’: satisfaction with physical features of their houses (floor, walls, roof, protection against rain, etc); health indicators (respiratory diseases, diarrhoea instances, etc); economic activity; and life satisfaction. First measurements were taken at baseline, second immediately after residents moved into the homes, and thirdly eight months after residents moved into the houses.

Effects on health were mixed- no impact on respiratory diseases was identified, but in El Salvador and Mexico there were apparently large impacts of instances of diarrhoea (a 30% decline overall, but which was insignificant in regressions at the country level). No impact on employment or income was found. Perhaps most interestingly, the study found large positive effects on the treatment group’s satisfaction with their houses, and life satisfaction, in the first post-treatment survey, but 60% of these gains had been lost by the time of the follow-up study 8 months later; the researchers hypothesise that this is due to hedonistic adaptation, a fairly well-documented phenomena in which people appear to ‘get used to’ material improvements in their lives (e.g. a new house, car, jacket, lottery win), causing any initial, post-purchase, spike in happiness or satisfaction to recede fairly quickly, and life satisfaction levels to return to pre-purchase levels after (usually) a few weeks or months.

The study also uncovered additional, note-worthy, cross-country difference in the lives and preferences of the urban poor. Firstly, whereas in Mexico and Paraguay, the slum-dwelling poor were richer than the urban poor, in El Salvador (by far the poorest country of the three) this pattern was reversed. Secondly, slum-dwellers in Paraguay endured lower quality housing than those in Mexico or El Salvador, but to counter this, receive higher wages despite lower levels of education. In Mexico, some lottery winners moved out of their free TECHNO houses, despite the fact that they would receive land titles if they resided there for five years.

Discussants suggested the author provide figures on the cost-effectiveness of the housing provision, in order that it can be compared with ‘sites and services’ approaches (in which the government provides neighbourhood plans and basic infrastructure); ‘sites and services’ methods are generally perceived to be more cost-effective and scalable than direct provision of houses.

By Sally Murray, Country Economist, IGC Rwanda

“Transport infrastructure, urban growth and market access in China” (Matthew Turner)

Professor Matthew Turner (Brown University) presented a paper on the role of infrastructure investments in linking local markets and its growth effects. This is an important growth topic because, as the presenter noted, a very large amount of development lending and government budgets are spent on infrastructure every year in the developing world. The goal of this work is to address three points. The first is to provide more evidence on the relevance of the scale of cities and urban areas to productivity and growth. The second, which is a technical but very important point, is to provide a method for defining how infrastructure improves growth in areas that it links up. This focuses on the important distinction between reorganisation, where it appears that an area has received growth from a new road or infrastructure link, but it is really a relocation of economic activity, and actual growth, which is genuine new economic activity. The third is to provide a study in the effects on infrastructure in urban areas, which is helpful as a lot of thinking about infrastructure focuses on its potential as a rural development tool. The paper focuses specifically on the unprecedented massive expansion of the highway networks in China in between 1990 and 2010 in order to explore these questions.

The main mechanism by which roads appear to have increased growth in China during this period has little to do with the roads themselves, but is the power that these roads have to link urban centres with each other. These infrastructure investments increase the potential market for each urban centre, as products can now be shipped farther in the same amount of time. This allows for higher integration in markets and trade between these centres, and ultimately leads to higher efficiencies through this coordination. The initial results are suggestive that this effect of extending infrastructure networks can have powerful effects on economic growth. This also has strong implications for the potential of infrastructure investments, as strong economic benefits of these infrastructure networks only exist insofar as they link up previously unconnected productive economic centres. Simply building roads that provide no such links will achieve little and cost much, whereas well-targeted roads that create such linkages have the potential to generate genuine growth in economic activity.

By Aaron Weisbrod, Country Economist, IGC Myanmar

“The geography of development: Evaluating migration restrictions and coastal flooding” (Klaus Desmet)

Where a person lives determines wellbeing, productivity and income. But choice of location is not a permanent or free choice. Desmet et al look at how migratory restrictions shape the welfare and global distribution of economic activity over time. The researchers develop a model for the world economy in form of dynamic spatial growth model incorporating geography, trade, migration, innovation and diffusion of technology. Using data for the world economy, they split the world into 1×1 degree cells and study the economic impact of completely or partially relaxing the migration restrictions that exist today. The model indicates that with unchanged migration restrictions the high population density places will gradually become high productivity places over time. Liberalizing migration fully would increase world welfare more than threefold and there would be substantial changes in the distribution of economic activity. The study also models the welfare impact on the world economy of a demographic shock in form of a 1 or 6 meter rise in sea levels. The results indicate that coastal flooding has important dynamic and geographic implications for welfare in the world economy and these vary depending on the global migration restrictions.


  • Endogenous changes in fundamental amenities?
    • Are amenities time-invariant?
    • Are they likely to capture other endogenous characteristics?
  • Non-legal Barriers of migration?
    • Studies suggest that there are substantial non legal migration costs across countries. How would incorporating these impact on the gains?
  • Pace of sea-level rise?
    • The paper simulates the sea-level rise as a sudden shock. However, sea-levels are expected to rise gradually and the question is how incorporating these would affect the results.
By MILJAN SLADOJE, Country Economist, IGC Zambia

Expert panel discussion: Methods, experiments, and data

This session discussed methodological issues for conducting studies related to cities and urban development. Nathaniel Baum-Snow (Brown University) said that a key constraint for research on cities is the difficulty in finding appropriate control groups for comparison, and to scale-up interventions in a manner that provides rigorous results of impact. This is because in cities there are many linkages between the housing market, labour market, transport system etc. He said that one way to push forward is to generate more data on households living in cities.

Marcel Fafchamps (Stanford University) spoke about the issues of rapid urbanisation in Africa. According to UN estimates, the population of Africa is expected to reach 4 billion by 2100 and a large part of it will be living in urban areas. Economists must conduct new research on cities to contribute to improving the spatial efficiency of Africa’s urbanization. He presented existing theory on understanding the economy of cities and their economic relationship with the surrounding countryside.

Stephen Redding (Princeton University) promoted the use of quantitative spatial models, which have made a lot of progress in the past few years. He is using these to estimate commuting, migration and local employment elasticities. His research aims to assess the impact of local changes in the economic environment and their ability to attract factors of production – especially labour. He said that commuting of workers between locations is important and this spatial aspect must be accounted for while estimating impacts of interventions on the labour market.

By SOHAIB ATHAR, Country Economist, IGC Pakistan

Expert panel discussion: South Asia

Dr Gilles Duranton (Wharton, University of Pennsylvania) presented its research on the misallocation of factors of production in India. The research aims to document the misallocation of factors in India and understand the determinants and the policy implications. This paper quantifies the misallocation of manufacturing output and factors of production between establishments across Indian districts during 1989–2010. It first distils a number of stylized facts about misallocation in India, and demonstrates the validity of misallocation metrics by connecting them to regulatory changes in India that affected real property. With this background, the study next quantifies the implications and determinants of factor and output misallocation. Although more productive establishments in India tend to produce more output, factors of production are grossly misallocated. A better allocation of output and factors of production is associated with greater output per worker. Misallocation of land plays a particularly important role in these challenges.

Professor Mushfiq Mobaraq (Yale University) presented his research on improving sanitation in Bangladesh. This research was conducted in relatively dense rural areas of Tanore district in northwest Bangladesh, the poorest region of the country. Although sanitation coverage has increased dramatically in rural Bangladesh in recent decades, progress in Tanore has been slower. Prior to the start of the study, 31 percent of households reported that they either lacked a latrine or used an unimproved latrine. Only 50 percent had regular access to an improved sanitation facility, defined as a toilet or latrine that separates human excreta from human contact.

The study, covering more than 18,000 households in 380 neighborhoods, aimed to systematically investigate these alternative hypotheses. In order to make sure that the results were in no way contaminated by pre-existing differences among the neighborhoods, a lottery was employed to assign different interventions to different neighborhoods. Some neighborhoods got a version of the highly touted Community Led Total Sanitation (CLTS) campaign, where, over two days, the community participated in graphic demonstrations of how diseases travel from the feces left in the field to their bodies, along with extended harangues and appeals to their better nature. Another randomly chosen set of neighborhoods got help with supply. Latrine supply agents were appointed to provide assistance with the installation and maintenance of toilets. Both these interventions failed comprehensively: at the end of the study period people in these neighborhoods showed no more interest in toilets than their counterparts in neighborhoods where absolutely was nothing done. On the other hand, there was a very substantial increase in toilet construction and a correspondingly large reduction in open defecation, in those (randomly chosen) localities where the demand generation campaign was combined with a discount of 75% on the cost of installing a toilet, with the subsidy targeted only to the landless poor in the community.

Dr Ijaz Nabi (IGC Pakistan) highlighted the role that urbanisation is taking in shaping economic growth in South Asia. The presenter explored the significance of urbanisation, hubs, competitive exports, and dynamic cities as drivers of economic growth and what this could mean for South Asia’s development. Dr. Nabi also stressed that for Pakistan to achieve a sustain growth path, their cities country will likely have to undergo rapid urbanisation. This will require better infrastructure to support not only commercially critical operations, such as air freight, roads, and telecommunications, but populations in low-income urban settlements that lack basic household sanitation and energy sources.

Dr Taibur Rahman (Ministry of Planning) presented the case of urbanisation in Bangladesh. The presenter highlighted the role of Dhaka city as Primate City, as the urban area in Bangladesh that has been growing the most. In the last 20 years, Dhaka has been attracting most of the rural-urban and urban-urban internal migration in Bangladesh. While other cities have been struggling to achieve the same economic growth, Bangladesh alone contributes to 15% of the GDP, gathering in its boundaries 30% of the urban population of Bangladesh. Dr. Rahman also evidenced how this success is been politically induced through the disproportionally large investment that Dhaka has been enjoying compared to the other urban areas of Bangladesh.

By FILIPPO SEBASTIO, Country Economist, IGC Bangladesh

Expert panel discussion: Public finance and urban growth

The Public Finance and Urban Growth session was chaired by Tim Besley (LSE, IGC) and and opened with David Albouy’s (University of Illinois) presentation of “Are Cities Ever Too Small”, which addressed the topic of optimal city size and revenue redistribution schemes. Albouy’s theoretical model considers the planning incentives at the city, federal, political, and competitive level, identifying the optimal size of cities, considering variances in positive and negative externalities across and within cities. Policy ideas that emerged for debate were the role of government in actively coordinating movement to minor cities and the possibility of subsidising agglomeration, balanced by a congestion tax.

Adnan Khan (LSE, IGC) followed, discussing “Evidence of Property Tax Collection in Pakistan”, during which he reviewed past and ongoing research to assess incentives among revenue agents. These projects sought ways to work around the political constraints that affect property tax collection, the incidence of corruption, and tax morale. One key point he emphasised was the political norm of government accountability for public service provision facilitated via the implementation of property taxation.

Then Mihaly Kopanyi (World Bank) reviewed the experience of Property Tax Reform in Rwanda, highlighting Kigali’s narrow property tax base, free rider problems, and short-sightedness in system design. Kopanyi’s recommendations included the ex-ante consideration of alternative tax systems, simulation of potential impacts, and analysis of institutional and implementation capacity. Examining these factors ex-post for Rwanda showed a three-fold increase in Kigali’s estimated tax-take and for Kopanyi to recommend the revision of the property tax legislation.

The session concluded with Meembo Changula (Ministry of Local Government and Housing, Zambia) who presented “The Case for National Urban Policy”, who outlined the historic and expected development of cities in Zambia. This economy has seen urban concentration along its mineral and rail corridors and is one of the fastest urbanising populations in Africa. Zambia is working to identify and formulate a cohesive National Urban Policy (NUP), which will be implemented as the country simultaneously plans and undertakes reforms to decentralise fiscal and revenue activities.

By ANNE LASKI, Country Economist, IGC Tanzania

Expert panel discussion: Threats to urban life

The Threats to Urban Life session was chaired by Ed Glaeser (Harvard, IGC) and opened with his own overview of some of the pertinent threats to urban life. Glaeser outlined a number of issues which arise for government when considering the downsides of density, primarily with respect to the large externalities associated with rapid urban growth. He went on to expound on some of the potential public policy tools, by way of subsidies and regulation, which can offer solutions to some of these concerns. Gleaser addressed ideas around the balance between the upsides and downsides of density and in particular the ways in which we might want to think about how public policy can make the downsides less harmful. He also iterated the importance of this research agenda for the IGC, noting that there is little by way of knowledge in understanding how to make agglomeration work more effectively. This is thought to be particularly important in the development context where countries have urbanised much more quickly, relative to national wealth, than countries have historically.

Nava Ashraf (Harvard Business School, IGC) followed with discussion of her project, working alongside Glaesar, on ‘Water, Health and Wealth in Zambia’. In this presentation Ashraf discussed how vital infrastructure developments have been in supporting urban growth and thus the needs of a growing urban population. In particular, the research focuses on the impact of a reliable municipal water supply in Lusaka, looking specifically at the economic and health impacts. The preliminary results from the research highlight the negative consequences resulting from pipe breakages in terms of economic and health outcomes, pointing to the importance, in this policy space, of stable water supply.

Following Ashraf, Esteban Rossi-Hansberg (Princeton) considered ‘The Need to Shrink Effectively’, with a discussion of the challenges associated with shrinking cities. This presentation recognised the need to think about the level of flexibility of our urban systems, particularly in the event of economic or local shocks. This discussion was placed within the context of the potential for climate change to exert its own shocks and specifically in the developing country context. Using Detroit as an example of a shrinking city, Rossi-Hansberg highlighted the need to consider, from a policy perspective, the management of shrinking cities and what this means in terms of their adaptability and structural transformation.

Guy Michaels (LSE) followed Rossi-Hansberg’s discussion by presenting his research on flooded cities. This cross-country study examined the effect of large-scale flooding on urban areas and found that there was no evidence that people moved to higher ground following a large-scale flood event. Michaels suggested that the reconstruction process should encourage movement to areas less prone to flooding.

Finally, Rudiger Ahrend (OECD) presented research completed by the OECD on metropolitan governance. This considered the importance of stronger governance mechanisms in urban areas in terms of decision-making and externalities. The OECD research highlighted that urban areas with smaller numbers of municipalities were generally more productive and better at making efficient decisions on behalf of the city as a whole. Aligned to this finding, cities with governance bodies were also associated with better outcomes by way of productivity.

BY ELLA SPENCER, igc Evaluation manager