It is normal that they are different, in particular the phase. This is because STACKRT is an analytical solution, and the phase is correct on the interface, whereas the FDTD is a discrete method, its excitation source and the monitors cannot be on the front and back surfaces. Therefore you first need to correct the propagation phases. Please refer this article https://optics.ansys.com/hc/en-us/articles/360042095873-Metamaterial-S-parameter-extraction
Please note that, due to discretization, the simulated phase has some errors compared to the theoretical values. therefore even with such compensation, the phase might still not be the same in simulation and STACK. But with finer mesh such error can be reduced. I would suggest strongly to compare the absolute values of the Fresnel's coefficients.