AFAICS, a supply doesn't have a power factor. It is determined entirely by the load downstream.
In the simplest terms, the generator is responsible for the alternating voltage, but your load is responsible for drawing the current. Given the available voltage at the point of connection, your load is the sole determinant of the magnitude of the current that it will draw, and its phase angle relative to the applied voltage. Therefore the PF is solely a property of your load.
I see your point, but let's build this up one step at a time.
The simplest case is a single generator and a single load, connected by a loss-free line. As the consumer, you measure the power factor of whatever load you choose to apply. Meanwhile, the supplier measures the power factor of the load that *it* sees, back at the generator output. In this very simplest case, the consumer's and the supplier's measurements of PF are obviously the same.
Now add another consumer and another load. Each individual consumer measures a PF that is determined by the current flowing into his own individual load. The other consumer's load may affect your supply voltage, but it does *not* affect the PF of your load (unless the load happens to be non-linear - but that is still entirely a property of your local load, not the supply).
Meanwhile, back at the generator, the supplier can measures the PF of
*its* load, as seen at the generator output terminals. This will depend on the individual PFs of the two separate consumers' loads, line lengths etc; but it does not depend on the generator - the generator itself doesn't *have* a PF.The complications arise when we have a distributed network with several generators, a very large number of loads, propagation delays and line losses. As you know, this is a hugely complex situation which includes the possibility of generators not being synchronised, and not necessarily contributing power to the network. This blurs the boundaries between "generator" and "load", as the network has some characteristics of both.
But even then, the same basic principles apply. Each individual generator has only one set of output terminals, and sees the rest of the network as its load. So it is still the *load* that has a PF - not the generator. I'm not sure how the "network PF" is defined... but it still doesn't affect you.
When you connect a load somewhere on the network, all the complexities drop away - we're right back to basics with a single connection point. Your load is the sole determinant of the current that it draws, and the phase angle relative to the applied voltage is still determined entirely by the characteristics of your load.
Whatever may be happening in the network upstream of your connection point, it does not concern you. When you connect a load, it will affect the voltage/current relationships somewhere back upstream in the supply network (at least to some small extent) but it still doesn't affect the PF of your load, as measured by you at your single point of connection.
[*] Another important detail is that the other load(s) must not affect the waveform of the alternating voltage available to you. That is why there are restrictions on the harmonic and non-sinusoidal content that loads may throw back into the supply. However, I don't believe that is where the original question was coming from.Hope this makes sense, because the coffee mug's empty now.