Igor, I think that CT is counting the Japanese spaces in the Statistics... because it doesn't recognize them as spaces.
Here is a Japanese space: « »
I think it is U+3000
Can you confirm?
(I'm working on a Japanese Power Point source file that seems to contain hundred of spaces, as CT counts about 6800 characters insteaf of 6100, as verified afer exporting the text to MsWord.)
The Statistics feature count characters without and with spaces (U+3000 taken into account). See the two separate cells.