Are you sure you have a problem? I tried generating a few values based on your method and a functionally equivalent alternative:
(defn argbval
[a r g b]
(bit-or (bit-shift-left a 24)
(bit-or (bit-shift-left r 16)
(bit-or (bit-shift-left g 8) b))))
(defn altargbval
[a r g b]
(+ (* 16777216 a) (* 65536 r) (* 256 g) b))
(defn -main
([& args]
(println "(argbval 25 21 23 29): " (argbval 25 21 23 29))
(println "(altargbval 25 21 23 29): " (altargbval 25 21 23 29))
(println "(argbval 0 0 0 0): " (argbval 0 0 0 0))
(println "(altargbval 0 0 0 0): " (altargbval 0 0 0 0))
(println "(argbval 255 255 255 255): " (argbval 255 255 255 255))
(println "(altargbval 255 255 255 255): " (altargbval 255 255 255 255))))
and got identical values.
(argbval 25 21 23 29): 420812573
(altargbval 25 21 23 29): 420812573
(argbval 0 0 0 0): 0
(altargbval 0 0 0 0): 0
(argbval 255 255 255 255): 4294967295
(altargbval 255 255 255 255): 4294967295
I got the same numbers on my pocket caculator.
Are there specific arguments that you have seen to produce incorrect results?
Perhaps it isn't a fault in this logic but in writing/reading the values to the BufferedImage.