# Quantifying the Economic Impact of Extreme Shocks on Businesses using Human Mobility Data: a Bayesian Causal Inference Approach

**Authors:** Takahiro Yabe, Yunchang Zhang, Satish Ukkusuri  
**arXiv:** [2004.11121](https://arxiv.org/abs/2004.11121)  
**Date:** 2020-04-01  
**Summary page:** https://www.summarizepaper.com/en/paper/2004.11121v1/

## Key points

- Frequency and intensity of extreme shocks, such as natural disasters, have been increasing globally
- Accurately quantifying economic impact on local businesses post-disaster is crucial for effective planning
- Shift towards utilizing large-scale human mobility data for assessing damages inflicted on businesses by disasters
- Study by Takahiro Yabe, Yunchang Zhang, and Satish Ukkusuri leveraged mobile phone location data to estimate causal impact of hurricanes on business performance
- Bayesian structural time series model used to quantify causal impact of disasters on businesses post-hurricane Maria in Puerto Rico
- Hierarchical Bayesian models utilized to uncover how various business characteristics influence long-term resilience post-disaster
- Novel approach introduced to quantifying business resilience could aid policymakers in enhancing disaster preparedness strategies

## Summary

In recent years, the frequency and intensity of extreme shocks, such as natural disasters, have been on the rise globally. This has led to substantial economic losses for numerous cities. It is crucial to accurately quantify the economic impact on local businesses following such catastrophic events for effective post-disaster assessment and pre-disaster planning. Traditionally, surveys have been the primary method used to assess damages inflicted on businesses by disasters. However, surveys are often costly and time-consuming to implement. They also suffer from spatio-temporal data gaps and lack scalability. Recently, there has been a shift towards utilizing large-scale human mobility data, such as mobile phone GPS information. This allows for observing and analyzing human movement patterns with unprecedented granularity and scale. In this study conducted by Takahiro Yabe, Yunchang Zhang, and Satish Ukkusuri, location data collected from mobile phones was leveraged to estimate and analyze the causal impact of hurricanes on business performance. To quantify the causal impact of disasters on businesses post-hurricane Maria in Puerto Rico specifically, a Bayesian structural time series model was employed. This model predicted the hypothetical performances of affected businesses had the disaster not occurred by using performances of unaffected businesses outside the disaster areas as covariates. The methodology was tested on 635 businesses across 9 categories in Puerto Rico to assess their resilience in the aftermath of Hurricane Maria. Furthermore, hierarchical Bayesian models were utilized to uncover how various business characteristics such as location and category influenced the long-term resilience of businesses post-disaster. This study introduced a novel and more efficient approach to quantifying business resilience that could potentially aid policymakers in enhancing disaster preparedness strategies and relief efforts. By incorporating human mobility data into economic impact assessments following extreme shocks like natural disasters, this research provides valuable insights for improving disaster response mechanisms at both local and global levels.