By Renee Zahnow
In recent years the world has experienced a number of extreme disaster events. The 2004 Indian Ocean Tsunami devastated countries from East Africa to Thailand, and in the United States, Hurricane Katrina in 2005 had far reaching and long lasting consequences for New Orleans. In 2011, Christchurch, NZ was severely damaged by one of the largest earthquakes recorded in the country. This event was closely followed by a tsunami that ravaged large areas of Japan. Natural disasters resulting from rapid onset hazards such as floods, bushfires, and storms incur wide ranging impacts on individuals and communities. There is some consensus that disaster events are increasing in frequency and severity (Gencer, 2013; World Bank, 2010). Further evidence suggests that natural disaster costs are increasing and will continue to rise in the future, as a function of population growth, an aging housing stock, and growing concentration of assets in disaster prone-areas (Australian Government Productivity Commission, 2014; World Bank, 2010). These costs are direct (e.g. damage to public and private property and infrastructure) and indirect (e.g. flow on effects as the community responds to the disaster). A potential indirect cost of disaster is increased crime.
Although some studies suggest disaster increases community capacity for social regulation by bringing residents together, through their shared experience of trauma (Drabek & McEntire, 2003; Quarantelli, 2005), others find disasters produce anomic conditions that encourage people to panic. This can then hinder normative behaviors that are necessary for informal social regulation, leading to an increase in crime (Erikson, 1994). Further disasters alter the routine activities of residents and can increase the number of properties left unguarded, which in turn increases opportunities for crime (Leitner and Guo, 2013). To mitigate the negative effects of a disaster event it is imperative to understand what makes a neighborhood vulnerable to negative outcomes, including crime increases in the post disaster context.
In January 2011 Brisbane, the state capital city of Queensland, Australia, experienced an extreme flood event. It impacted over 175,000 people, and resulted in $7.5 billion dollars worth of damage, making it the most costly natural disaster in Australia’s history. Thousands of homes were damaged or completely destroyed, and the composition of neighborhoods changed as a result of the flood. This study investigates whether these changes to the neighborhood structure influenced crime trends across 390 communities in Brisbane, following the flood event. It examines the association between post-flood changes in property crime and neighborhood structural characteristics, and changes in those characteristics and crime in nearby neighborhoods. In so doing it addresses three questions: What pre-disaster neighborhood characteristics indicate vulnerability to post-disaster crime increases?; how do neighboring communities influence post disaster changes in property trends; and do disaster related changes in neighborhood characteristics influence property crime trends? This research advances earlier work on disaster and crime by examining changes in crime trends post-flood and by incorporating spatial effects.
Geographical Information System (GIS) was used to spatially integrate Australian Bureau of Statistics census data and property crime count data with maps depicting flood extents. ARIMA times series analysis was employed to model neighborhood crime using pre-flood monthly crime count data (1996-2010) and to forecast expected post-flood crime trend. The forecast trend was then compared to actual crime counts to establish whether or not post-flood property crime trend was significantly different than forecast. A neighborhood was found to experience post-flood property crime that diverged from pre-flood trend if post-flood property crime fell outside of the 95 percent confidence interval for the forecast trend (for example see Figure 1). After conducting ARIMA times series analysis for all Brisbane neighborhoods each was assigned to one of three categories: post-flood crime not significantly different than forecast; post-flood crime lower than forecast; post-flood crime higher than forecast. Multinomial logistic regression was employed to assess structural characteristics associated with increased likelihood of deviation from forecast crime post-flood.
The results of the ARIMA time series analyses indicated that in most neighborhoods (70%) property crime trend did not deviate from forecast post-flood (Figure 2). Both flooded and non-flooded neighborhoods were among those that experienced deviation from forecast property crime trend post-flood. While there was no significant spatial clustering in deviation from forecast crime trend (Moran’s I= 0.055), pre-flood spatial context was a predictor of post-flood deviation from crime trend. Specifically, neighborhoods that were surrounded by low crime areas pre-flood were more likely to experience post-flood crime that was lower than forecast. Other neighborhood characteristics associated with greater likelihood of lower than forecast crime post-flood were higher population density and lower residential mobility. Higher residential mobility, percentage households renting and percentage single parent households were associated with greater likelihood of a neighborhood experiencing higher than forecast crime post-flood.
This research uses spatially integrated data to examine the effects of disaster on crime and to identify neighborhood characteristics associated with greater vulnerability to post-disaster crime increases. This study found that residential mobility, population density, percentage households renting and percentage single parent households predicted greater likelihood of post-flood deviation from forecast crime trend. It also found that pre-flood crime in the neighborhood of interest and in surrounding neighborhoods was associated with likelihood of deviation from forecast crime trend in the disaster context. These results have a number of implications for future research and policy regarding disaster preparedness and response. Pre-disaster identification of places most vulnerable to crime increases can assist with resource allocation and ensure preparedness and prevention actions are focused on these neighborhoods. The findings suggest that following a disaster event it may be necessary to increase police patrols in particular types of places to prevent unexpected increases in property crime. It may also be necessary to educate residents in vulnerable neighborhoods on the increased risk of property crime in the post-disaster period. There is potential for this analysis to be expanded to incorporate indicators of density of crime attractors and generators to assess whether the presence of particular physical characteristics make neighborhoods more vulnerable to experiencing changes in crime trend in the event of a disaster. A limitation with existing research on the effect of disaster on crime trends is that it does not control for spatial variations in key structural and social characteristics that are known to influence patterns of crime in times of disaster quiescence. This research begins to redress this gap in the existing scholarship.
Australian Government Productivity Commission (2014) Natural Disaster Funding Arrangements: Productivity Commission Draft Report Volume 2. Commonwealth of Australia: Canberra. http://www.pc.gov.au/projects/inquiry/disaster-funding/draft
Drabek, T.E. and McEntire, D.E., (2003),Emergent phenomena and the sociology of disaster: lessons, trends and opportunities from the research literature. Disaster Prevention and Management, 12 (2): 97 – 112.
Erikson, K.T. (1994) A New Species of Trouble: Explorations in Disaster, Trauma, and Community. New York: W.W. Norton & Co.
Gencer, E. A. (2013). Natural disasters, urban vulnerability, and risk management: a theoretical overview. In The Interplay between Urban Development, Vulnerability, and Risk Management (pp. 7-43). Springer: Berlin Heidelberg.
Leitner, M., and Guo, D. (2013). Analyzing the Impact of One Important Unplanned Exceptional Event, Hurricanes, on Crime in Louisiana, US Using a Visual Analytics Approach. In Jekel, T., Car, A., Strobl, J. and Griesebner (Eds.) GI Forum 2013: Creating the GISociety. Conference Proceedings. Austrian Academy of Sciences Press.
Quarantelli, E. L. (2005). Catastrophes are different from disasters: some implications for crisis planning and managing drawn from Katrina. Understanding Katrina: Perspectives from the social sciences.
World Bank (2010) Natural hazards, unnatural disasters: the economics of effective prevention. The International Bank for Reconstruction and Development: Washington.