Obviously it'd be even better if he accepted transactions instead of ignoring them, but he's not actually preventing transactions from being confirmed.
We can't even know that for certain.
If he had transactions in the block he was working on, it is entirely possible that he wouldn't have found a nonce solution before someone else (since he'd be searching for a nonce in an entirely different block with an entirely different SHA-256 hash).
So, it is entirely possible that his creation of a block without any transactions has increased the speed of confirmations. How can we predict whether the network would have found a block sooner than they have if they were all working on a different set of blocks?