Imagine the following light intensity $P_0$ - current $I$ characteristic of a sample (as could be acquired with a Mini Photoelectric Spectrometer)
Clearly, it shows both anodic (positive) and cathodic (negative) currents.
If IMPS was performed around blue points, the X component will be negative because photocurrent at low frequencies will decrease with increasing light intensity. Or $\frac{dI}{dP} < 0$
On the other hand, if IMPS was done around red points, the X component will be positive because $\frac{dI}{dP} > 0$.
Hence,
IMPS distinguishes between falling and rising slopes of I-P characteristics but not between cathodic and anodic currents
