Your question cannot be answered precisely, because no one knows exactly what "collapse" is or how it works mechanically. Some interpretations of QM deny that something labeled collapse even occurs (i.e. they say collapse is not physical, as @BillAslept mentions). However, this much is known - assuming there *is* something called collapse: 1. Time of collapse has no bearing whatsoever. If you want to say there is collapse at one place in spacetime X that occurs *before* it occurs in another spacetime location Y, that's fine. But strictly speaking, there is no standard theoretical notion that something at X "causes" something to happen at Y (in the normal sense of the word "causes"). 2. It is equally satisfactory to say Y's collapse causes X's earlier collapse, as strange as that sounds. In fact, particles can be entangled after they are measured. If that doesn't throw traditional notions of causality into question, I don't know what does. That process is called Delayed Choice Entanglement Swapping. See for example: [Experimental delayed-choice entanglement swapping][1] 3. Keep in mind that no experimental test of entanglement has ever demonstrated anything relating to ordering of observation (collapse) as a factor. If there is such an element, it is completely hidden to testing. Note that the theoretical treatment of entanglement predictions does not give any preference to ordering, as only the final measurement context is a factor. 4. Consequently, in consideration of the above: It should be clear that reference frame plays absolutely no known role in the experimental predictions or outcomes. Any statement otherwise is simply an assumption with no observational or theoretical basis. 5. And finally: no conservation laws appear to be violated in entanglement tests. There are some interpretations in which that might occur, and there have been proposals to transmit energy using entanglement. However, nothing concrete is evident at this point giving any indication that conservation of anything might be violated. I hope this helps. [1]: https://arxiv.org/abs/1203.4834