Coloring Graphs to Estimate Causal Effects: A Diagonal Ramsey Approach

Development Economics X Paper Model Twenty-One


In this paper, we apply diagonal Ramsey numbers, which measure how large a graph has to be to guarantee that it contains a monochromatic subgraph of a given size, to the problem of estimating causal effects. We argue that graphs can be used as models of data or systems, and that diagonal Ramsey numbers can be used as estimates of treatment effects. We also extend our approach to multilevel data, where the assignment to treatment or the outcome may depend on various sources across network levels. We show how to classify units into homogeneous groups with regard to class-specific selection and outcome models, and how to account for both the nested structure of the data and potential heterogeneity in selection processes and treatment effects. We illustrate our approach with empirical examples. We propose the use of power calculations for improved rigorous treatment effects.

Opoku-Agyemang, Kweku (2023). "Coloring Graphs to Estimate Causal Effects: A Diagonal Ramsey Approach." Development Economics Paper Model Twenty-One. 

© 2023 Development Economics X Corporation. All Rights Reserved.