Why do we assume convex transaction costs in theoretic models? What is the intuition and the strict definition of this assumption?
Well, I have in mind the Glosten model in 1994. Neither he provides some details, nor it is obvious why he defines convex transaction costs. He only claims that the quasi-concavity of the utility function, ismplied by the concavity of some function $U(\cdot)$ and the convexity of the transaction costs function $TC(\cdot)$. This is a link of the Glosten 1994 article, you can see in page $1132$. I wonder does he assumes convexity because this comes from any micro foundation assumption, or it is just for solving the problem, because he can not solve it differently. Why not assuming that the function of transaction costs is a linear function which in fact is a convex and a concave function simultaneousely.