Hello,I am computing the first moments of yields levels and changes.I am finding that skewness and kurtosis of yields levels are very different than for yield changes:1) yield changes are skewed to the left but yield levels are slightly skewed to the right2) yield changes have fat tails (kurtosis) while yield levels do notI am confused and I do not know whether to consider yields levels or changes to analyze the distributional properties of yields.I do not know what's more relevant: focusing on levels or changes an in which cases you focus on changes rather than levels.Can anybody help me?Thanks,

Time series of yields usually have a near unit root, so for most statistical purposes you want to work with changes (absolute or relative).

Ok. Thank you

Hi,what numbers did you get for the skewness and kurtosis of yield changes?Many papers deal with yield changes, or yield spread changes (perhaps best known: Collin-Dufresne Goldstein Martin 2001 Journal of Finance), but nobody ever gives you a basic distributional of their data.I'm investigating bond returns, and want to see how my values for skewness and kurtosis compare with the literature.