This paper studies how government debt variables impact estimates of the classic and new UIP puzzles for quarterly data between 2000 and 2020 of 6 developed countries in relation to the United States. I estimate country-pair VECMs to model cointegration relations between debt variables, price differences, interest rates differences and nominal exchange rate. I compare this framework with one without debt variables following Engel (2016) using quarterly data between 1979 and 2020. In the framework without debt, I don’t find the new UIP puzzle while in the framework with debt, I do find it. Government debt variables are significant and alter the sign of co-movements between difference in interest rates and far-ahead ex-post and ex-ante excess currency returns. The magnitude of the effect is economically relevant. Government debts coefficients cannot be uniquely associated with convenience yield story.