1	Urban Structure in a Climate of Terror Stephen Sheppard Williams College Guns and Butter – The Economic Causes and Consequences of Conflict 9-10 December 2005
2	Terrorism and urban structure Why worry about urban structure? Pace of urban expansion Doubling of developing country urban population in next 30 years Enormous investment Durable investment – distortion generates costs over time Impact on economic performance Factor productivity Distribution of non-market goods
3	Terrorism and urban structure Why worry about impact of terrorism? Policy concern regarding impact New technologies enhance impact General climate of terror Affect large and anonymous population Distribution of costs of terror Test and distinguish between theories of urban structure in extreme conditions Prospect for corrective public policy Two perspectives Empirical Theoretical
4	Empirical evidence – analogy with war
5	Empirical evidence – city comparison Find comparable cities with different exposure to terrorist incidents
6	Empirical Evidence – cross country model
7	Theoretical Perspectives Three approaches to analysis: New economic geography Harrigan and Martin (2002) Dynamic model Rossi-Hansberg (2004) Traditional urban model Each models terrorism as a tax or distortion Different implications for public policy If data exist – potential for test to distinguish
8	Theory – new economic geography Based on Fujita, Krugman and Venables Increasing returns and monopolistic competition led to agglomeration Terror attacks more likely in agglomerations Terrorism acts like a tax on production for firms in agglomeration No analytic solution – numerical simulation Modest amounts of terrorism leave agglomeration unchanged Higher levels destroy rationale for agglomeration and lead to dispersion of production For many parameter values dispersal is an alternative stable solution – end of terror does not restore agglomeration
9	Theory – dynamic model Agglomeration supported by production externality Identifies a steady-state allocation of land use and productive capital Terrorism implies a risk of loss of structures (capital) at any location where density exceeds a fixed level K₀ With no adjustment costs – Terrorist attack implies lower steady state capital at all locations Capital density gradients have reduced range Public policy Subsidy to support agglomeration If public sector has private knowledge about attack risk – can improve efficiency
10	Theory – traditional urban model Terrorism can be modeled as one of three distortions Increased transportation costs Reduced productivity of land in housing production Reduced productivity of land in export good production Impacts on density and maximum extent of urban area Adapt the model of Brueckner (1987)
11	Modeling urban land use Households: L households Income y Preferences v(c,q) composite good c housing q. Household located at x pays annual transportation costs t·x The transportation costs increase in direct proportion to the expected incidence of terrorism In equilibrium, we must have: for all locations x
12	Modeling urban land use Housing producers Production function H(N, l) to produce square meters of housing N = capital input, l=land input Constant returns to scale and free entry determines an equilibrium land rent function r(x) and a capital-land ratio (building density) S(x) Land value and building density decline with distance Combining the S(x) with housing demand q(x) provides a solution for the population density D(x,t,y,u) as a function of distance t and utility level u The extent of urban land use is determined by the condition:
13	Modeling urban land use Equilibrium requires: The model provides a solution for the extent of urban land use as a function of Population Income Agricultural land value Transportation cost If we generalize to include an export sector, then urban land use will also depend on MP of land in goods production World price of the export good
14	Hypotheses
15	Data – a global sample of cities
16	Data
17	Measuring urban land use
18	Change in urban land use: Cairo
19	Model estimation Cross-country model Total Urban Land Use Urban area population National GDP per capita Terrorist incidents in preceding 10 years Agricultural output per hectare arable land Groundwater availability Air linkages (city) and IP address share (country) Environment type Endogeneity? Additional variables?
20	Variables used in analysis
21	Dichotomous variables in analysis
22	Urban rank and the impacts of terror
23	Estimates: terrorism and urban expansion
24	Hypotheses consistent with data
25	Concluding remarks The models perform surprisingly well Almost all parameter estimates significant at 10% level or higher All parameter estimates correct sign Terrorism has an impact on urban structure Reduces amount of land where capital is located Consistent with both Rossi-Hansberg and simple urban model Estimated impact is robust to different specifications Correctly signed but not significant in “differenced” model Limited number of observations? Explore alternatives when all data available
26	Future directions Endogeneity? Correlation between RHS variables and model error Potential problem with income and terror Reduced problem by use of national variables Problem with air linkages Instruments: Data for neighboring cities Physical conditions Data being collected by field researchers Compare differenced and non-differenced models Other variables Regional and regime fixed effects? Better measures of transport costs? Infill versus peripheral development Test prediction of flatter density gradient Distinguish between simple urban and dynamic urban model