Hello all. I am researching various variables that impact bond yield spreads. I use annual data for a sample of 14 countries over 20 years. Due to the persistence of bond spreads and the inclusion of lagged variables as regressors, I am exploring both IV and GMM methods. I would appreciate some thoughts on the following:
- I use a set of lagged variables as instruments for the IV method. Has anyone done similar work, and if so, are there any instruments( for the lagged yield spreads)that they would suggest I explore further?
- For the GMM method, I am exploring both xtabond2 and Kripfganz’s method. However, I read that for a cross-section of 14 countries, the GMM method may not be suitable. Is there any guidance on the size of N or T that can be used for GMM? (Roodman suggests small T and large N, but what exactly is meant by small T and large N?
