Because Revised Romanisation (which I assume is the transcription you're using) is not a morphophonemic transcription nor a strict transliteration. Thus, results of predictable phonological processes like tensing of obstruents are shown explicitly.

In Yale, I would write /eps.e.yo/, which is a letter-for-letter transliteration of the corresponding Hankul, complete with division into camo. No need to note the tensing of the /s/ because it's completely predictable. However, in RR and McCune-Reischauer, this consonant would be written double.