It is reasonable that a national institute is one step from the international reference, and a calibration company should then be able to be two steps from the reference. Your calibrated instruments could then be three steps away, and the reference in your test jig four steps away (unless you send in the jig for calibration).

Anyway - for most produced equipment, it is very easy to have voltage, resistance and time/frequency references way better than what the produced equipment needs. It takes a bit more work to get a good temperature reference, but it can be done. Light can be problematic. It isn't too common that the test jig needs to measure distances.

In the end, most critical parameters can be handled quite well unless the produced equipment is special measurement equipment. But when producing high-end measurement equipment, the solution is normally to calibrate these similarly to calibrating the production instruments. The test jig is just there to detect production quality issues, not to calibrate the instrument.