Energy payback is not the same as carbon neutrality. The former only accounts for the economic value within a regional market and ignores externalities.
Just compare the price of a solar panel with the price of coal you'd have to pay for in China. That's the uppermost possible limit of what you could expect for CO₂ emissions if you didn't have to pay for anything else. Just conveniently assume all your workers and other raw materials are free. ;) You can probably buy two tonnes of coal in China for the price of a ~250W solar panel. That gives you about 4000 kWh of electricity. At a 15% capacity factor in China, the same panel can generate at least 6500 kWh in 25 years, even including quite substantial degradation. So if used to offset electricity generation within China, it's neutral after 15 years.
In reality, it's most likely much sooner because you don't need 4000 kWh of coal electricity to manufacture a 250W panel; you have to pay your workers and raw material suppliers instead.
All energy is not created equal. In order to make this calculation in a useful way, one would need to know the amount of carbon emissions per kWh for both the panel production site and ones local grid. And it's actually even harder than that, as one also needs to know if one is lowering base-load generation needs or surge-capacity. Base- and surge-generation are frequently not the same as carbon emissions go.
Consider for instance that you live near the Hoover dam (desert, lots of sun, prime area for solar); your panels might never be carbon-neutral.
Sure, it's complex. Take your example. The Hoover dam sells power into California. So installing solar panels in Boulder City is not that different than installing them in Los Angeles (there would be some extra grid losses in LA). If the Hoover Dam had the capacity to supply the entire southwest it would matter, but it doesn't have that capacity.
There's also the question of whether buying Chinese solar panels makes China more or less likely to convert to solar itself.
Even in a 100%-coal grid, I don't think it's likely that >1000 kg CO₂/kWp is necessary. German manufacturers can achieve 400 kg CO₂/kWp in the still-rather-mediocre German grid.
If energy is the only input to making a solar cell (the worst case) and the cell costs x Wh then the market price can't go below x * ($/Wh), so it pays back when you produced x Wh (thus not buying x Wh from the coal burning power plant) and any energy you produce is both profit for you, and also less total demand for coal energy.
Don't have the details of the study, it was from an informal evaluation from Jean-Marc Jancovici, a well known (in France) climate/energy specialist. A similar argument is made here:
Related quote: "If the photovoltaic panels made in China were installed in China, the high carbon intensity of the energy used and that of the energy saved would cancel each other out, and the time needed to counterbalance greenhouse-gas emissions during manufacture would be the same as the energy-payback time. But that’s not what’s been happening lately. The manufacturing is mostly located in China, and the panels are often installed in Europe or the United States. At double the carbon intensity, it takes twice as long to compensate for the greenhouse-gas emissions as it does to pay back the energy investments."
It's not that similar an argument, it suggests a carbon payback time of 3-10 years, not 30 or never (10 years is unfortunate but workable (especially if you expect it to improve, which is basically guaranteed), 3 years is fine, never would be unacceptable).
True, that depends on the country where the PV panel is installed. The original argument was probably made for France where electricity is very low on carbon.
http://www.nrel.gov/docs/fy04osti/35489.pdf
http://www.clca.columbia.edu/236_PE_Magazine_Fthenakis_2_10_...
It doesn't make any sense that China would be subsidizing their panel manufacturers to the extent you are implying.