In DB2, we often use the predicate EXISTS to verify if a row is present 
in a table based on our required criteria.  We do not wish to get any 
data from the table being read on the right side of EXISTS, we just 
want to see if some row is present or not.

In this case, EXISTS should be coded as

SELECT A,B,C FROM TABLE1
WHERE
D EXISTS
(
SELECT 'X'
FROM
TABLE2
);

Note: X can also replace a column name from TABLE2

In the existing code, you might see EXISTS used as below, where the 
complete matching data (by using SELECT *) from TABLE2 is obtained, 
to check for its existence. We will have performance degradation here. 
We need to ensure to have the code as the above example, where we get 
only a constant or single column.

SELECT A,B,C FROM TABLE1
WHERE
D EXISTS
(
SELECT *
FROM
TABLE2
);