I am currently working on a panel error correction model, which I am trying to build in order to measure the fair value of 5 Yr CDS for 20 EM countries. All individual series contain trends. I also tested for co-integration. I have selected some fundamental variables such as the credit rating (a low frequency), inflation, debt-to-gdp and GDP growth (higher frequency). I have standardized the last variable to get a better sense regarding the difference in the growth cycle among the sampled countries. I also included de Vix as a risk appetite proxy. As expected the model is more sensitive to changes in credit ratings and risk appetite. Taking into account the diversity of the panel, I am quite comfortable with the final structure of the model. The variables are significant at 1% level and model residual’s looks fine as well. However, I get weird results in countries like Peru, Colombia and Turkey. I have checked the time series and I didn’t find any problem there. Any suggestion or ideas????I use market consensus forecast to forecast the fair values. According to this, Thailand, and Argentina are the cheapest countries are China, Mexico and Ukraine the most expensive. Sounds this reasonable?? I would appreciate any commentBestM