GROUPING을 이용해서 하시는 것도 괜찮을 것 같네요.

http://www.gurubee.net/lecture/2679

sum이 아니라

금액을 단가로 나눈값. 추가되어야 합니다.

네 그렇습니다.

원하시는 데이터가 이게 맞나 모르겠네요.

WITH T AS
     (SELECT '201507' YM, '단가' COL1, 1000 COL2
        FROM DUAL
      UNION ALL
      SELECT '201507' YM, '금액' COL1, 10000 COL2
        FROM DUAL
      UNION ALL
      SELECT '201508' YM, '단가' COL1, 2000 COL2
        FROM DUAL
      UNION ALL
      SELECT '201508' YM, '금액' COL1, 20000 COL2
        FROM DUAL)
SELECT   YM, COL1, COL2
    FROM (SELECT YM, COL1, COL2
            FROM T
          UNION ALL
          SELECT YM, '금액/단가' COL1, P2 / P1 COL2
            FROM (SELECT   YM, MAX(DECODE(COL1, '단가', COL2)) P1
                         , MAX(DECODE(COL1, '금액', COL2)) P2
                      FROM T
                  GROUP BY YM))
ORDER BY YM, INSTR('단가,금액,금액/단가', COL1) ASC;

네 원하는 결과입니다. 감사합니다.

SELECT YM
     , DECODE(COL1, NULL, '금액/단가', COL1)
     , DECODE(GROUPING(COL1), 1
            , LAG(SUM(COL2), 1, 0) OVER (ORDER BY YM, GROUPING(COL1), COL1 DESC)
            / (SUM(COL2) - LAG(SUM(COL2), 1, 0) OVER (ORDER BY YM, GROUPING(COL1), COL1 DESC))
            ,  SUM(COL2)) COL2
FROM T
GROUP BY YM, ROLLUP(COL1)
ORDER BY YM, GROUPING(COL1), COL1 DESC
;

SELECT ym 
     , CASE 
           WHEN GROUPING_ID(col1) = 0 
           THEN col1 
           ELSE '금액/단가' 
       END col1 
     , CASE 
           WHEN GROUPING_ID(col1) = 0 
           THEN MIN(col2) 
           ELSE MIN(DECODE(col1, '금액', col2)) / MIN(DECODE(col1, '단가', col2)) 
       END col2 
  FROM T 
 GROUP BY ym, ROLLUP(col1)

WITH t AS
(
SELECT '201507' ym, '단가' col1, 1000 col2 FROM dual
UNION ALL SELECT '201507', '금액', 10000 FROM dual
UNION ALL SELECT '201508', '단가',  2000 FROM dual
UNION ALL SELECT '201508', '금액', 20000 FROM dual
)
SELECT *
  FROM t
 MODEL
 PARTITION BY (ym)
 DIMENSION BY (CAST(col1 AS VARCHAR2(20)) col1)
 MEASURES (col2)
 RULES (col2['금액/단가'] = col2['금액'] / col2['단가'])
 ORDER BY ym, DECODE(col1, '단가', 1, '금액', 2, 3)
;