CityReads│How Disaster Influences the Urban Redevelopment
How Disaster Influences the Urban Redevelopment
Using the destruction from the 1906 San Francisco Fire as a laboratory, this paper examines residential density across razed and unburned areas between 1900 and 2011 and try to explore how disaster influence the urban redevelopment. Thriving cities face substantial redevelopment frictions in the form of durable buildings and that large shocks can greatly alter the evolution of urban land-use outcomes over time.
Siodla, J. (2015). Razing San Francisco: the 1906 disaster as a natural experiment in urban redevelopment. Journal of Urban Economics, 89, 48-61.
Source: http://www.sciencedirect.com/science/article/pii/S0094119015000479
Picture source: http://news.nationalgeographic.com/news/2006/04/photogalleries/san_francisco_quake/
As clusters of durable capital, cities can be slow in adapting their forms and structures to changing economic conditions. Buildings, which are costly to remove and replace, act as frictions that delay redevelopment, the timing of which is dependent upon many economic factors. This timing can be disrupted by a large shock, which could then have substantial short-run and long-run effects.
Using the destruction from the 1906 San Francisco Fire as a laboratory in which thousands of buildings were destroyed, this study compares development across razed and unburned areas both before and after the disaster and studies the impact of a localized disaster to provide unique insight into urban dynamics.
This study seeks to understand two facets of urban redevelopment: the role of durable buildings in delaying redevelopment in the face of growing demand, and the extent to which exogenous shocks to the timing of redevelopment last through time. All else equal, a significant difference in San Francisco’s form upon reconstruction implies that the durability of urban capital is an important friction in redevelopment. Furthermore, a long-lived impact of the fire suggests that shocks can greatly alter the evolution of urban land-use outcomes over time.
As a large shock, the fire greatly disrupted the timing of redevelopment in San Francisco. For some cities, shocks may have little long-run impact on development patterns, thus suggesting that natural location-specific attributes dictate long-run urban growth. Other cities may thrive because of self-reinforcing agglomeration economies. Evidence for this phenomenon is seen in cases where urban areas that were initially developed for reasons that no longer matter can persist for long periods of time, thus perpetuating development patterns even well after original advantages are gone.
In San Francisco, these frictions were greatly reduced when much of the city’s land was cleared by the fire in 1906. For thousands of parcels in the razed area, the fire reduced demolition costs by reducing buildings to ashes. Transaction costs were reduced as agreements on the value of structures themselves were no longer necessary in order to exchange properties. Further, the removal of old buildings likely bolstered expectations that rents would not suffer by the presence of poor-quality buildings upon redevelopment, assuming that newly constructed buildings were of better quality on average than the buildings that burned. The fire also encouraged passive property owners to alter their land since current rents earned were made equal to zero. The removal of these frictions—a cleansing effect of the fire—led to large-scale redevelopment reflecting new economic conditions.
By providing a clean slate in San Francisco, the fire reduced redevelopment frictions and thus greatly accelerated the rate of capital replacement in the razed area while the unburned area did not experience the same impetus.
This study estimates the causal effect of the fire in San Francisco on residential density. Evidence of a significant post-disaster increase in density is found in areas razed by the fire relative to unburned areas, an effect which persists to the present day.
Historical Background
San Francisco experienced significant population growth leading into the early twentieth century, thus causing housing rents to escalate with assessments increased at an average rate of roughly 2 percent per year between 1900 and 1906. As shown in Table 1, this growth has continued over the last century, thus suggesting that the pressure to become denser through redevelopment has not disappeared with time.
The 1906 San Francisco Fire followed a tremendous earthquake and consumed more than 28,000 buildings. The property damages inflicted by the fire greatly exceeded those caused by the earthquake, representing at least 80 percent of total damage from the disaster. The earthquake left buildings in moderate disrepair, whereas the fire often completely razed them. Furthermore, San Francisco’s capital markets were substantial and well-functioning at this time. Thus, capital was largely available in reconstruction.
Reconstruction was fairly rapid. As Table 2 shows, the number of buildings constructed in San Francisco between 1906 and 1914 was roughly equal to those destroyed in the fire, thus suggesting that new density patterns in the razed area had likely emerged within a decade of the disaster. Furthermore, the table shows that new buildings constructed in the city were largely made of the same materials used in those that were destroyed by the fire. A study found developers also opposed new building codes, such as height limitations and fire-resistant walls, which were either defeated or ignored in reconstruction. These episodes suggest that private interests dominated other interests in rebuilding the city. Though regulation initially played a very minor role, after the 1930s, however, other factors such as strict land-use regulations may partially confound the results.
Data
The study utilizes both historical and modern data for a sample of city blocks. The source of the historical data is maps produced by the Sanborn-Perris Map Company, which were created for companies that insured buildings in cities and towns against the risk of fire. They contain detailed information on the residential, commercial, and industrial buildings that existed at a particular time. And the text will refer to the years 1900, 1905, 1914, 1931, and 1950. Modern data from 2011 were provided by the San Francisco Planning Department.
Data were gathered for 421 city blocks. Blocks that have edges or corners immediately adjacent to the fire’s boundary are identified as boundary blocks and make up a subsample that is used in the analysis.
Net residential density, which equals the number of residential units per residential acre, is the outcome of interest. In this measure, residential units are the sum of single-family dwellings, multi-family units, flats, and apartment units, while residential acreage excludes all vacant, nonresidential, and mixed-use land.
Fig. 1 displays two fire-map sheets—a total of four city blocks (eight polygons)—from 1905. The maps show the number of residential units on each block
Razed and unburned blocks were identified using a map created in 1908 by the State Earthquake Investigation Commission (SEIC, 1908), shown in Fig. 2. The map depicts a strongly delineated fire boundary along which there are very few partially burned blocks.
Fig. 3 shows the blocks for which data were gathered. These blocks are located in close proximity to the fire’s boundary. Blocks that have edges or corners immediately adjacent to the fire’s boundary are identified as boundary blocks and make up a subsample that is used in the analysis.
Estimation and Results
Fig. 4 shows average density from 1900 to 2011 for each of the samples utilized in the paper. In the restricted sample, razed and unburned areas exhibited similar density levels before the fire, as well as similar trends in density growth leading up to 1906. After the fire, density in razed areas increased substantially relative to unburned areas, an effect that persists to the present day. For the full sample, the graphs depict similar trends in density growth leading up to the fire. After redevelopment, there is an increase in density among razed blocks relative to unburned blocks. After 1915, the groups follow similar density trends until around 1950, when evidence of convergence in density becomes apparent, especially in the boundary blocks.
The results of the regression analysis using the full sample suggest that overall, the boundary block results are similar to the results for all blocks: the full specifications show a large initial impact of the fire on residential density, one that dampens after 1950 in levels and after 1914 in logs. Furthermore, the full sample shows a relative decline in log density by 2011.
As for the restricted sample, the regression analysis tells the fire’s effect is significant from 1914 but had weakened by 1931 and was nonexistent by 1950 in the log specification. Overall, the fire caused a large relative increase in density in the razed area by 1914, and the level of density in the unburned area has yet to catch up.
Conclusion
The 1906 San Francisco Fire provides a unique setting in which many redevelopment frictions were suddenly and exogenously eliminated through the destruction of thousands of buildings. A thriving city at the time, San Francisco was rebuilt in a manner that diverged significantly from its previous structure. After the fire, developers constructed much denser housing in razed areas relative to unburned areas where frictions were still in place. This result is consistent with the notion that the frictions associated with adjusting land use in cities are substantial, even in relatively free-market ones. The implication of dynamic urban models is that capital durability produces different development patterns than does a static model with malleable capital.
Not only was the initial effect of the fire strong, but its impact isstill apparent today. In prominent residential neighborhoods, a density differential still exists across razed and unburned areas where it did not exist just before the fire. While many factors may contribute to this persistence, large shocks that disrupt the timing of redevelopment in a city can have long-lived effects on urban land use. Overall, the evidence presented in this paper suggests that a significant legacy is present in the form and structure of urban areas, one which is greatly influenced by external shocks.
Related CityReads：
45. How San Francisco Beat L.A?
（Enter Official Account and Reply number“6"“11”or "12”）