Si entendí correctamente, usa la subconsulta de correlación para encontrar filas con más de un exceptn_criteria_rtrn_val distinto valor. Si no me equivoco será mejor utilizar la función analítica:
select vw.registration_id as REGISTRATION_ID, vw.batch_id as BATCH_ID
from
(select ss.registration_id as REGISTRATION_ID, ss.batch_id as BATCH_ID
,count(distinct sub_query.exceptn_criteria_rtrn_val) over(partition by ss.registration_id, ss.batch_id) as cnt
from submtd_srvc PARTITION(SUBMTD_SRVC_821370) ss,
(select a.exceptn_criteria_val, a.exceptn_criteria_rtrn_val
from EXCEPTN_CRITERIA a, EXCEPTN_EXPRESSION b
where a.EXCEPTN_EXPRESSION_ID = b.EXCEPTN_EXPRESSION_ID
and b.EXCEPTN_EXPRESSION_NAME = 'NC_CUSTOM_REV_CAT'
and b.IS_CURRENT_INDCTR = 1
and a.IS_CURRENT_INDCTR = 1) sub_query
where ss.REVENUE_CD = sub_query.exceptn_criteria_val
and ss.batch_id = 821370) vw
where cnt > 1
Order By vw.Registration_Id, vw.batch_id;