Nothing, except that to recoup the spectacular costs of deploying OTA HD video broadcasting in each area they want to serve, the streams would need to be encrypted, and Netflix OTA users would require special hardware to decrypt them.
What you're essentially proposing is that Netflix should become DirecTV.
That entire model is outdated anyway. Everything is converging on the internet: voice calls (WiFi calling, Skype, etc), IM (WhatsApp, Facebook Messenger), video (YouTube, Netflix, Hulu), radio (Shoutcast, podcasts, etc), books (Kindle, Apple, Nook), etc.
And people want Netflix to step backwards into a "you'll get the content we give you" mode that everyone hated? Having TV schedules is a deficit of traditional broadcasting, not a perk, as even if you like being spoon-fed programming that could trivially be re-created via on-demand (e.g. smart-queues, your playlist(s), curated playlists (e.g. your favourite celebrity's choices), etc).
Plus the infrastructure waste (duplication?) and all for what gain? It would be more useful to take all that money and invest in making the Netflix apps on smart TVs able to stream continuously like a TV "channel." Then do what Google Play Music has gone and add curated "channels" of content (although it is music in this example, not video).
Let's also mention that a single OTA HD channel is about 18 mbit/sec of data. So that expensive transmitter would be able to deliver a specific show to 4 or 5 people, max.
It's called broadcast for a reason. Everyone watches the same show.
If your idea involves using just the spectrum and not the actual ATSC signal, then you're doing what ClearWire did with WiMax. You'll need to give each subscriber a special modem and build out a network of towers around each city. That's even more money than a TV transmitter.
What you're essentially proposing is that Netflix should become DirecTV.