Hi, I have the following question about the panel data analysis in the STATA: Some of the data series I use for my macro panel analysis are stationary I (0) and some are nonstationary I (1). My question is whether or not there is any panel analysis method similar to the ARDL bound model in the time series in the sense that I (0) and I (1)series are evaluated together or should I take the first difference of the I(1) series and put them into regression? Thank you very much, Özgür,
Assuming that the stationarity results are robust and correct type of unit root tests had been employed, you can use panel ARDL methods, which are MG, PMG and DFE depending on the Hausman Test results. Theoretically, you must use all the variables in differenced form, whether they are I(0) or I(1).
In STATA, for the command xtpmg, the variables have to be entered as differences (for example d.VAR). eViews automatically operates using differenced variables. There are no bounds test approach and VIF test for panel ARDL as a system. Instead, individual groups may be tested for bounds and VIF.
There is a panel test for valid error-correction based on the autoregressive distributed lag model, however, the critical values assume that all of the variables are I(1). This can be implemented in Stata, see the references below:
Bruno, G. S. F. (2005a) ‘Approximating the Bias of the LSDV Estimator for Dynamic Unbalanced Panel Data Models’, Economics Letters, 87, 361–366.
Bruno, G. S. F. (2005b) ‘Estimation and Inference in Dynamic Unbalanced Panel-Data Models with a Small Number of Individuals’, The Stata Journal, 5, 473–500.
An paper that applies this method is:
Goda, T., Stewart, C. and Torres García, A. (2019) ‘Absolute Income Inequality and Rising House Prices’, forthcoming in Socio-Economic Review.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series