I am currently in a situation where I am limited to using data on about eight companies. However, each company's data is quite extensive on a weekly basis and it's in the form of panel data. This means that while the sample size is small, there's a large number of observations for each company due to the lengthy time dynamics, ensuring that the data is sufficiently comprehensive. My advisor has suggested employing the First-difference model (they reason that by structuring the data using the differences between t and t-1, wouldn't this separate the post-shock fluctuations from the pre-shock fluctuations, thereby effectively increasing the number of data points?). Is this a viable approach?
- Badges
- Science topic
- Similar topics
- Psychology
More Lee Wonguk's questions See All
Similar questions and discussions