You can assume that the functions and the measure $\mu$ are strictly positive and the integrals are finite.
I am asking because I found this paper that establishes a result that is rather similar. However, I am not sure if the monotonicity I am looking for is obvious, or if it is a special case of their result.
Thank you
In the case that $f,g$ are continuous.
For fixed $k,$ and $x\ge k,$ let $F(x)=\int_k^xf(t)dt$ and $G(x)=\int_k^xg(t)dt.$
Then $F/G$ is increasing on $(k,\infty)$ iff $\;0\le (F/G)'=(Gf-Fg)/G^2\;$ iff $\;0\le Gf-Fg\;$ iff $$\frac {F(x)}{G(x)}\le \frac {f(x)}{g(x)}$$ for $x>k.$
By the Second Mean Value Theorem, there exists $y\in (k,x)$ with $$\frac {F(x)}{G(x)}=\frac {F(x)-F(k)}{G(x)-G(k)}=\frac {F'(y)}{G'(y)}=\frac {f(y)}{g(y)}\le \frac {f(x)}{g(x)}.$$