On Causal Discovery from Time Series Data using FCI

We adapt the Fast Causal Inference (FCI) algorithm of Spirtes et al. (2000) to the problem of inferring causal relationships from time series data and evaluate our adaptation and the original FCI algorithm, comparing them to other methods including Granger causality. One advantage of FCI based approaches is the possibility of taking latent confounding variables into account, as opposed to methods based on Granger causality. From simulations we see, however, that while the FCI based approaches are in principle quite powerful for finding causal relationships in time series data, such methods are not very reliable for most practical sample sizes. We further apply the framework to microeconomic data on the dynamics of firm growth. By releasing the full computer code for the method we hope to facilitate the application of the procedure to other domains.