Well, since you don't tell us the range of Vbe (the dependent variable) you are looking at, what can I say?
The fact is, over a sufficiently small range, any well-behaved function will APPEAR essentially linear.
ezplot(@exp,[0,.1])
We know it is NOT truly linear. But as you would expect, over that range, the higher order terms in a Taylor series approximation are virtually irrelevant. Widen the range sufficiently, and no problem. It now looks as you might expect.
ezplot(@exp,[-1,1])
