Photosynthesis only uses a subset of the visible spectrum. PV panels make power using a different (larger? certainly non-identical) subset of the visible spectrum, albeit at a maximum of maybe 30% efficiency in practice. At first glance, if solar panels convert enough unused parts of the spectrum into LED output that can drive photosynthesis, there might be a net gain on crop yield per m² versus traditional farming. I'm sure someone's run this experiment by now. If so, I'd be curious to see the results.
Incidentally, PV panels discard IR light which makes up 55% of all received solar energy. Concentrated solar power systems seem theoretically more efficient per m² since they capture all light as heat, but I imagine maintenance, energy storage, and heat engine/dynamo inefficiencies combine to bring their total efficiency down.
In short, back of the napkin calculations for this stuff are harder than they may seem! I'd love to see the results of some real-world experiments.