Testing Granger Non-Causality in Panels with Cross-Sectional Dependencies
This addresses a methodological bottleneck in econometrics and statistics for researchers analyzing causal relationships in dependent panel data, representing a strong domain-specific improvement.
The paper tackles the problem of testing Granger non-causality in panel data with cross-sectional dependencies by proposing a new approach that aggregates p-values instead of statistics, showing it yields lower false discovery rates at the same power for large samples and dependencies, and successfully discovers the true causal relation between COVID-19 cases and deaths where state-of-the-art methods fail.
This paper proposes a new approach for testing Granger non-causality on panel data. Instead of aggregating panel member statistics, we aggregate their corresponding p-values and show that the resulting p-value approximately bounds the type I error by the chosen significance level even if the panel members are dependent. We compare our approach against the most widely used Granger causality algorithm on panel data and show that our approach yields lower FDR at the same power for large sample sizes and panels with cross-sectional dependencies. Finally, we examine COVID-19 data about confirmed cases and deaths measured in countries/regions worldwide and show that our approach is able to discover the true causal relation between confirmed cases and deaths while state-of-the-art approaches fail.