Suppose the locked-in contract will return an extra e after 5 years, and that the buy/sell spread (difference between buy and sell prices) for both the normal index fund and the locked-in contract is s.
Then I can short the locked-in contract and buy the normal contract at price s, and in 5 years I can unwind that getting back e-s, for a total risk-free return of e-2s.
So the maximum premium e for this product is 2s, no matter what the term. For a liquid index, this is tiny.
EDIT: I realised afterwards that I didn't take borrow costs into account in that calculation. Apart from the spread in the borrow costs contributing to the cost of the arbitrage transaction above, maybe the illiquid variant actually has an intrinsic difference? For example the custodian could lend out the asset with a promise that it wouldn't be recalled for a period.