Good question!
Is the CT of the Ammeter too close to the Ballast's Core, resulting in an incorrect reading?
Additionally, could the Branch Circuitry be Multiwire with a common Grounded Conductor?
Additionally, could the unbalanced fixture's ballast be leaking some current to the equipment grounding conductor? (like a grounded segment at the Reactor Coil, or an uncapped AutoTransformer Multivoltage Tap lead connecting to grounded metal equipment)
Maybe someone has tapped into that 14 amp line, and made an L-G connection somewhere - causing the unusual imbalance.
Scott