This paper exploits the surge in Chinese exports from 1994 to 2004 as a natural experiment to evaluate the effects of a unilateral low wage trade and competition shock to producers in Mexico. We find that this shock causes selection at both firm and product levels as its impact is highly heterogeneous both on the intensive and extensive margins. Sales of smaller plants and more marginal products are compressed and are more likely to cease, while larger plants and products exhibit an opposite response. Similar results hold both for the domestic market and for competition facing Mexican exporters in a third market (i.e. the United States).

Using a novel dataset with transaction-level level exports from Malawi, Mali, Senegal and Tanzania, we explore the determinants of the survival of firm-product-destination combinations past the first year after entry on export markets. We find that survival correlates with diversification (or experience) at the extensive margin: a firm-productdestination combination is more likely to survive if the firm ships more * This paper is part of a World Bank research project initiated by Paul Brenton and Denisse Pierola. Financial support from the BNPP Trust Fund on Improving the Survival of African Exports is gratefully acknowledged. We are also grateful to the Malawi Revenue Authority, the Direction Générale des Douanes du Mali, the Direction Générale des Douanes du Sénégal, and the Tanzania Revenue Authority for their cooperation and willingness to share data. We also thank Frances Aidoo, William Baah-Boateng, Sidiki Guindo, Anthony Mveyange and Nelson Nsiku for their assistance in the collection of the customs data used in this paper. products to that destination, or if it ships that product to more destinations, suggesting synergies within the firm. Most strikingly, we find significant evidence of cross-firm synergies: a firm-product-destination is more likely to survive, the more firms from the same country export the same product to the same destination. These results may suggest a case for export promotion at the national level.

This paper presents new evidence on urbanization using sub-county data for the United States from 1880-2000 and municipality data for Brazil from 1970-2000. We show that the two central stylized features of population growth for cities -Gibrat's Law and a stable population distribution -are strongly rejected when both rural and urban areas are considered. Population growth exhibits a U-shaped relationship with initial population density, and only becomes uncorrelated with initial population density at the high densities found in predominantly urban areas. We provide evidence that the explanation for these patterns lies in different employment growth dynamics in the agricultural and non-agricultural sectors and the process of structural transformation away from the agricultural sector.

Do fixed geographic features such as coastlines and rivers determine town locations, or can historical events trap towns in unfavourable locations for centuries? We examine the effects on town locations of the collapse of the Western Roman Empire, which temporarily ended urbanisation in Britain, but not in France. As urbanisation recovered, medieval towns were more often found in Roman-era town locations in France than in Britain. The resetting of Britain's urban network gave it better access to natural navigable waterways, which mattered for town growth from 1200 to 1800. We conclude that history trapped many French towns in suboptimal locations.Our world is increasingly urbanised, with large cities located in many different environments. While some cities appear well situated, others are hampered by poor access to world markets or threatened by environmental hazards. But it is difficult to tell whether cities are in fact mislocated, because locational advantages that we cannot quantify could offset the costs that we do observe. To investigate whether cities are mislocated, one might examine whether cities that are devastated then relocate, and if so -where. But in recent years a growing literature, going back to Davis and Weinstein (2002), finds that cities are remarkably resilient and recover from shocks. This has generally been interpreted as evidence that locational fundamentals play an important role in pinning down the location of cities. At the same time, Bleakley and Lin (2012) show that once cities develop in particular locations, they persist even when the factors that led to their emergence are no longer important.We contribute to this literature by examining the impact on urban locations of a shock that was sufficiently large to eliminate an urban network. Specifically, we study the collapse of the Western Roman Empire around the middle of the first millennium CE, which wiped out the towns in what is now Britain, but not in France. 1 We are, to the best of our knowledge, the first to examine whether in the aftermath of such a devastating shock, an urban network that gets a fresh start reconfigures in new locations. We also examine whether such a reconfiguration improves towns' locations in terms of their first nature locational fundamentals. In order to carry out our analysis, we develop a new dataset, which tracks the locations of towns in parts of north-western Europe over two millennia.