Evaluating the effectiveness of support programs

We can quantify the effectiveness of disaster relief/support programs using methods like DiD. However, the practical application of DiD may face challenges in selecting comparable treatment and control groups, particularly when support is targeted at specific subgroups within the affected population.

Example: Post-disaster tax relief by the Australian government

The Australian Government provides tax relief/breaks after disasters, specifically for businesses in the affected regions. In the case of 2010-2011 Queensland floods, the support was limited to small businesses with fewer than 20 employees before the floods.

Forming a treatment group of supported businesses and a control group of remaining businesses is challenging, violating the parallel trend assumption. Larger, more resourceful firms recover faster, while smaller, resource-constrained businesses take longer. In this scenario, regression discontinuity, focusing on the threshold (e.g. 20 employees), proves more relevant than DiD for estimating program effectiveness.

Other methods

Another approach to overcome the aforementioned selection problem is using randomised control trial (RCT), wherein affected units are randomly chosen to receive the treatment, such as providing a monetary grant. An example of this approach is found in the work of De Mel, McKenzie, and Woodruff (2011), who assessed the impact of capital access on the recovery of Sri Lankan microenterprises following the December 2004 tsunami.

Watch the video

Watch UNICEF’s video on Randomised Controlled Trials (RCTs) to understand the RCT method better. Feel free to share your thoughts or any questions in the comments section.

References

_{De Mel S, McKenzie D, and Woodruff C (2012) ‘Enterprise recovery following natural disasters’, The Economic Journal 122(559): 64-91.}