Do photons emitted from a LED show bunching? If photons are emitted from a thermal source, we get photon bunching. For coherent radiation, the detection probability doesn't change after detecting a photon. For single photon sources, we get anti-bunching.
A LED isn't a thermal source but it isn't coherent either. Does this mean that we still have some amount of bunching?
I didn't find anything in the literature because everybody is all over single photon LEDs.
 A: One thing that you can be sure of is, for a large enough LED, you will get poisson statistics to a very good approximation. Neither bunching nor anti-bunching. The reason is simple: One photon comes from a certain part of the LED, the next photon is likely to come from a totally different part of the LED and head in a totally different direction. There's no way that either of these photons can influence the other.
The question is, what do I mean by "large enough LED"? 100 microns is definitely large enough. 100 nanometers is probably not. In between those, I don't know. I hope someone else will give a better answer!! :-D
(I'm referring to the size of the active area on the chip, not the size of the package.)
A: An LED is a collection of discrete atoms. When a current flows, an electron excites one every so often. The excited atom decays after a while and emits a photon. Assuming the current is uniform, variations in intensity come from randomness in arrival times and excited state lifetimes. Some time slices have many photons, some fewer. This distribution is called Shot noise. 
An LED with polished ends is a laser. Light is generated by stimulated emission. An excited population of atoms is maintained. The LED is filled with photons all in the same state. Every so often, one will stimulate an excited atom to emit a photon in the same state. If everything else is uniform, this will be subject to the same randomness as an ordinary LED. So again, you get Shot noise. 
