Published works

There are two main considerations in the first step of our methodology:

• identification of disaster-affected areas, the treatment group, which may be measured in different statistical units, such as Statistical Area-Level 2 (SA2) or Local Government Area (LGA); and
• determination of the comparison SA2s/LGAs (i.e., areas that can provide the counterfactual incomes had there been no disaster, the control group).

It is preferable to identify the disaster-affected areas at finer units because this approach will yield more precise estimates for the disaster effects.

Generally speaking, SA2s [1] provide finer units for analysing the disasters that struck regional areas because LGAs are relatively large in regional areas. By contrast, when analysing disasters that hit metropolitan/urban areas, LGAs provide the finer geographic variation while also capturing the jurisdictional differences that might be relevant to disaster impacts.

Before embarking on each case study, we carried out detailed ArcGIS work using Geographic Information System (GIS) tools. In identifying the disaster-hit SA2s or LGAs, we first overlaid the disaster zones on statistical and administrative maps. In particular, we used disaster maps to apply location-based analysis, that is, to vectorise and transform our raster data to map coordinates by using the ESRI shapefile formats provided by the Australian Bureau of Statistics (ABS). Here, it is important to achieve an almost perfect overlap between the raster and our target data in terms of the location, shape, and attributes of geographic features for different statistical units. In the following step, we determined the disaster-hit SA2s or LGAs.

In determining the comparison groups, again it is important to consider the nature of the disaster-hit areas. For regional disasters, it is more appropriate to choose neighbouring areas that are not hit by the disaster but share similar economic, geographic and topographic characteristics with the disaster-hit areas. For disasters that strike metropolitan areas, comparability is likely to be obtained from other metropolitan areas. In Australia, typically, capital cities mimic each other in terms of their economic, demographic and geographic characteristics. For example, Brisbane, Melbourne, Sydney, Perth and Adelaide are located on river banks and have agriculture-based hinterlands.

Below we describe how we specifically determined disaster-hit areas for each case study.