The thing about proof-of-work is that it is a proof. Assuming that sha256d is unpredictable, you will on average have use a certain amount of energy to find a low hash value.
With proof-of-retrieval, the name itself is lying. There is no way to prove that a piece of data went from one person to another. It should be called invoking-trust-in-retrieval-and-proof-of-having-the-file.
Along with that, because hashes have the property of being irreversible, blocks can be hashed and the winner cannot change the block without doing work again.
Because of these properties, filecoin must use PoW (which their whitepaper states they do).
According to their whitepaper, you must first construct the proof of ownership in the file, then you perform the PoW with that seed.
This leads to huge economies of scale because the startup cost of mining at equilibrium will include storing an abundance of files (probably all of them).