Luna Yue Huang

Economist / Data Scientist

Resume | LinkedIn | GitHub | Scholar

bio | research | impact | teaching

aerial photo | mapping poverty | COVID19 | air pollution | RCT

The Effect of Large-scale Anti-contagion Policies on the COVID-19 Pandemic

(Paper | GitHub | Press Release | Project Page)

Large-scale anti-contagion policies such as travel restrictions, business and school closures, and shelter-in-place orders averted roughly 62 million COVID-19 confirmed cases in the US, China, France, Italy, South Korea and Iran (as of March/April), according to our recent Nature paper.

Figure: Actual no. of confirmed cases (left: Confirmed Cases), and the projected no. of confirmed cases, had no anti-contagion policies been enacted (right: No Policy Scenario)1.

United States
Confirmed Cases
No. of Cases: 364,726
No Policy Scenario
No. of Cases: 5,149,354

The economic costs of large-scale anti-contagion policies are enormous and highly visible. Their public health benefits, however, are harder to see because it is difficult to know the counterfactuals. We produced the first peer-reviewed analysis to quantify these benefits, informing policymakers with timely and reliable information. Our study has been covered in more than 300 news stories, including CNN, Washington Post, New York Times, NPR, and Reuters, and reached policymakers in the White House and the CDC.

Collaborating with a fantastic team of 15 researchers in the Global Policy Lab, I worked around the clock to deliver statistical analysis that is as rigorous and comprehensive as possible within a tight timeline - we released our first draft on MedRxiv 10 days after the project started, while most economic papers take several years to write. In this project, I led the China team, identified and validated epidemiological data from native Chinese sources, which are of higher quality than commonly used COVID-19 data from John Hopkins University, and coordinated the compilation of national and local anti-contagion policy deployment in China. Concerned about how systematic trends in testing capacity may bias our estimates, I replicated epidemiological models of COVID-19 underreporting rates and estimated the upper bounds of the induced biases. Concerned about the discrepancy between our results and the prior literature, I dug deeper and found clear irregularities in the data that the prior literature relied on.

Our analyses painted a grim picture of what would have happened, had it not been for concerted policy effort across the globe, both at the local and national level. Scroll down to learn more.

Confirmed Cases
No. of Cases: 74,473
No Policy Scenario
No. of Cases: 36,395,576
Confirmed Cases
No. of Cases: 24,920
No Policy Scenario
No. of Cases: 304,093
Confirmed Cases
No. of Cases: 129,235
No Policy Scenario
No. of Cases: 2,248,041
South Korea
Confirmed Cases
No. of Cases: 9,924
No Policy Scenario
No. of Cases: 11,557,091
Confirmed Cases
No. of Cases: 21,683
No Policy Scenario
No. of Cases: 4,921,398
  1. The area of the circles is proportional to no. of confirmed cases. The sizes of the circles are comparable across countries. Estimates shown above are on March 5 (for China), March 22 (for Iran), March 25 (for France), or April 6 (for South Korea, Italy, US). We do not present projections (or confirmed cases) for over 100 cities in China. We restrict our projections as such for reasons stated in the Appendix