The null of the DW test is no first-order autocorrelation. It says nothing about higher orders of autocorrelation.
Thanks John C Frain you are absolutely right, I am asking about basically the value of DW. As a rule of thumb the DW=2 is being considered ideal, interpreting as no Auto-correlation. So may question is that, in panel data especially the case of Fixed Effect Modeling is there te role of auto-correlation is significant. If it is, then up to what value such DW>1.5 or >1 are acceptable?
Muhammad Kazim Jafri
The critical points of the panel DW test depend on the size of the time dimension, the panel dimension, and the number of variables in the regression. The critical values can vary a lot depending on these inputs. You will find critical values in
Serial Correlation and the Fixed Effects Model Author(s): A. Bhargava, L. Franzini and W. Narendranathan: The Review of Economic Studies, Vol. 49, No. 4 (Oct., 1982), pp. 533-549
This panel DW test is based on the within residuals. You should check your software manual as they may give more details of the actual DW statistic that they are calculating and the source of critical values.
Have you included period dummies?
- Badges
- Science topic
- Similar topics
- Thesis Research