The hash index type can be useful in cases where you are only doing equality (not range) searching on an index, and you don't allow NULL values in it. However, it is easy for hash indexes to become corrupt after a database crash, and therefore ineffective for queries until manually rebuilt. The advantages to using a hash index instead of a B-tree are small compared to this risk. You normally shouldn't ever use the hash index type. But if you are willing to spend a considerable amount of time analyzing its usefulness and ensuring the index will be rebuilt if it becomes corrupt, it is possible you can find a purpose for it.