Hi, I'm to do a multivariate regression analysis using TSCS data in R, and I'm a bit lost as to where I should start. My research question is how ALMP's have affected unemployment in four European countries over a 20 year time period, using expenditure data on subprograms from OECD. The resources I have from class don't show how to do a TSCS analysis so I'm wondering if people have tips on where I should start and what to do.
Well, you might have a look at this
Article Time-series - Cross-section data: What have we learned in th...
https://methods.sagepub.com/book/regression-analysis-and-causal-inference/n17.xml
I would try to look at the literature on the most convenient model and then to look how to estimate it in R. The most convenient model might not be trivial as ALMP spending is likely to be endogenous.
For instance
http://seekinginference.com/stats/TSCS-Education.html
I guess that the number of countries (N) is not large, which might pose some challenges due to how to deal with serial correlation...
https://stackoverflow.com/questions/14039257/set-time-series-for-cross-sectional-data-in-r
José-Ignacio Antón thank you! The second link was especially helpful. I chose Denmark, Sweden, Germany, and France as my countries, and my prof said it was fine even though it's a small number. He also suggested I code Denmark and Sweden into dummy variables in fixed effects in order to catch a consequently lower unemployment rate in the Nordic countries. At what point during the analysis should I do this?
Sunniva Smith ,
Unfortunately, I am more familiar with micro-panels, that is, panels with a large N and small T than TCSS (small N and large T), so I am not sure if my knowledge applies here. That is, my research is more on applied microeconometrics than applied macroeconometrics. You should check papers doing so. In the sort of micro-econometric research I usually do, we would enter the country fixed effects since the beginning or almost the beginning and, in any case, we would only use the specification without the fixed effects as a first model, but it would not be our preferred specification.
Best,
José-Ignacio
Chapter 8 of Croissant and Millo (2019), Panel Data Econometrics with R is a good survey of this area. The book contains a lot of information on the implementation of panel data with R and has lots of example programs in R which you may be able to .adapt to your needs.
(It might be better not to use abbreviations such as TSCS and ALMP which might not be understood immediately by the reader)
Hi again, I've made some serious progress in my regression. However, with a small N large T dataset, I have found out I have serial autocorrelation. How is it suggested I fix this? As well, I'm unsure about whether I should use a lagged variable for my variable Total (which is the total amount spent in budget on ALMP's measured in % of GDP). How do I figure out if this is needed? As well, I am unsure if I should run a Fixed Effects dummy regression. Any help?
Unemployment is the dependent variable and ALMP's are the independent variables... Try to execute a simple regression model on R then Interpret how the independent variable affects the dependent variable.
Best!!
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R