The answer is 3 mW.

Any version of EZNEC can be used to do this calculation. The demo
program will yield slightly less accurate results because of the limited
number of segments(*).

I modeled two vertical wires, 1 meter high and 1 mm diameter, spaced 1
km apart, at 20 MHz, over perfect ground. The reported feedpoint
impedance varies with segmentation, from 1.988 - j952.3 ohms at 10
segments/wire to 1.72 - j882 ohms at 100 segments/wire. Accuracy is
likely to degrade with a larger number of segments, since even 100
results in segment length/diameter ratio less than NEC recommendations.
I used 100 segments/wire for the test.

One of the choices in EZNEC of far field strength reporting is in V/m at
1 kW input and 1 km distance. For this antenna, EZNEC reports 300.8 V/m
(RMS) at ground level.

EZNEC also permits setting a fixed power input, so this was set to 1 kW.
The resulting source voltage and current are 21270 V. and 24.12 A.
respectively.

A load of 1.72 + j882 ohms was placed at the base of the second
vertical. EZNEC reports a power of 3.234 mW being dissipated in this load.

Care has to be used when analyzing the current induced in one antenna by
another which is distant using numerical calculations. Errors can occur
due to truncation and other causes when the ratio of distances between
the two antennas is great relative to the segment lengths or to segment
distances within one of the antennas. However, EZNEC gets virtually
identical results when using mixed and double precision NEC-2
calculating engines, which indicates that the limit hasn't been reached
and that numerical problems aren't occurring. (Another check which can
be done is to reduce the distance between antennas by a factor of two.
The power in the load resistance should increase by a factor of four.)

Another critical matter is the setting of the load reactance. The
reactance is many times larger than the resistance, so a slight error in
setting its value will result in a large difference in load current and
therefore load dissipation. For example, if the segmentation is changed
from 100 to 50 segments/wire and no other change is made to the model,
the reported load power becomes 0.3917 watts. The reason is that the
reported source impedance is now 1.756 - j891.4 ohms, while the load is
still 1.72 + j882 ohms. Changing the load to the proper conjugately
matched value of 1.756 + 891.4 ohms returns the load power to the
correct value of 3.24 mW.

All given, I'd trust the reported load power to be easily within 10% of
the theoretically correct value.

(*) Results for 10 segments/wire are 1.988 - j952.3 ohms for the source
impedance, 300.71 V/m field strength at 1 km for 1 kW, and 3.24 mW in a
conjugately matched load impedance in the distant vertical.

Roy Lewallen, W7EL