First, a couple of assumptions.
1, the output from the "fence power supply" is a pulse rather than a continuous voltage.
2, the meter in the power supply uses some sampling technique to read the fence voltage.
Second, a bit of theory.
When the power supply pulses the fence wire, the pulse travels along the fence at (nearly) the speed of light, 300,000,000 meters per second.
When the pulse encounters an impedance discontinuity in the wire, some of the pulse will be reflected back to the source.
The type of impedance discontinuity will determine the polarity of the reflection, if the discontinuity has a higher impedance than the surge impedance of the fence wire then the reflected pulse will have the same polarity as the incident pulse.
But if the discontinuity has a lower impedance than the surge impedance of the fence wire the reflected pules will be of the opposite polarity to the incident pulse.
Third, what I think could be happening here.
The flourescent bulb is a load and will have a lower surge impedance than the fence line
The reflected pulse will return to the power supply, the round trip distance is 2000ft so the reflected pulse will arrive back at the power supply after 0.5µs.
If the voltmeter samples the reflected pulse, the voltage will read low.
If some of this seems a bit far fetched, google Time Domain Reflectometer for more information.
As the man said in the film " That is all I have to say about that"
JimB