COVID-19 is forcing empirical researchers in international development to find new and innovative ways of collecting and using data.
Recent research conducted by Professor Simon Feeny, Dr Trong-Anh Trinh and Associate Professor Ashton de Silva has explored the extent to which monthly satellite nightlight data can be used to detect the impacts and recoveries from natural disasters in South-East Asia.
“Since disasters diminish production, consumption and public good provision, as well as disrupt power supplies directly, they can be detected via the loss of luminosity in nightlight data from satellites” reports Simon.
The Centre for Research on the Epidemiology of Disasters found that almost 12,000 people lost their lives to natural disasters in 2018.
A further 68.5 million people were affected and the damages from these disasters amounted to over US$130 billion.
Climate change is associated with an increasing number of climatic shocks such as floods and violent storms as well as droughts.
It is therefore becoming increasingly important that we understand the nature of these events, how regions typically recover and how best the international community can respond.
The research found that one-third of recorded disasters in the provinces of South-East Asian countries were detectable from space.
“To determine detection, there must be a significant departure in monthly nightlight luminosity data from its trend” argues Ashton.
Excessive cloud cover or disasters not resulting in a large fall of economic activity can prevent disaster detection using nightlight luminosity.
Not surprisingly, the intensity of a disaster is positively correlated to the likelihood that it can be detected.
Interestingly, the higher the level of economic development of the province in which the disaster occurs, the lower the probability of detection from space.
“This could be explained by more developed provinces having the resources to be able to quickly respond and offset any disaster-related reduction in economic activity” notes Trong-Anh.
The researchers also used the satellite nightlight data to trace a province’s recovery path following a disaster.
They found that when a disaster leads to nightlight data deviating from its trend, luminosity commonly returns to its trend following an initial contraction.
Moreover, while luminosity data for some recoveries are consistent with a new, permanently lower, growth path it is more common for luminosity to return to a higher growth path.
This is consistent with the principle of ‘creative destruction’ whereby the replacement capital stock associated with disaster recovery is more efficient than that which it replaces.
The finding that higher levels of disaster aid are associated with less chance that luminosity falls following a disaster lends further support to this notion.
“The effectiveness of disaster aid is notoriously difficult to assess and the findings from this research are therefore encouraging” commented Simon.
Centre for International Development - School of Economics, Finance and Marketing
