Having read the paper by Das, "The Surprice Element: Jumps in Interest Rates", I am concerned about the following; the paper prposes a mean-reverting jump diffusio process for the interest rates, and finds the characteristic function which is used to derive the closed-form solution for moments of the distribution of interest rates. However, in the diagnostics section the closed-form solutions are used and and compared against the moments of the changes in the intrest rates in the actual data. How is this possible since the closed form solutions correspond to the moments of the actual interest rate and not its change!E.g for the variance = (s^2+l E(J^2)) (1-exp(-2kt))/(2k), where k and s, l and E(J) are the speed of mena reversion, diffsuive volatility, jump intensity and expected jump size.I would realy apreciate your help!

Can someone give me some hint on this please? I would real appreciate it!