Problems with Character Encoding Using Haskells Text.Pandoc - haskell

Question

I want to parse a LaTeX-File using Pandoc and output the text, like this:
import qualified Text.Pandoc as P
import Text.Pandoc.Error (handleError)
tex2Str = do
file <- readFile "test.tex"
let p = handleError $ P.readLaTeX P.def file
writeFile "A.txt" $ P.writePlain P.def p
writeFile "B.txt" $ file
While the encoding in file B.txt seems to be "right" (i.e. uft-8), the encoding in file A.txt is not correct.
Here the respective extracts of the files:
A.txt:
...
Der _Crawler_ läuft hierbei über die Dokumentenbasis
...
B.txt:
...
\usepackage[utf8]{inputenc}
...
Der \emph{Crawler} läuft hierbei über die Dokumentenbasis
...
Anyone knows how to fix this? Why does Pandoc use the wrong encoding (I thought, it uses utf-8 by default)?
Update:
I got a (partial) solution: Using the readFile and writeFile-Functions from Text.Pandoc.UTF8 seems to fix some of the problems, i.e.
import qualified Text.Pandoc as P
import Text.Pandoc.Error (handleError)
import qualified Text.Pandoc.UTF8 as UTF (readFile, writeFile)
tex2Str = do
file <- UTF.readFile "test.tex"
let p = handleError $ P.readLaTeX P.def file
UTF.writeFile "A.txt" $ P.writePlain P.def p
UTF.writeFile "B.txt" $ file
However, I still didnt get the clue what the actual problem was, since both Prelude.readFile and Prelude.writeFile seem to work uft8-aware...