I have some experience in setting up low-volume production of a product.
Quality doesn't have to cost much more. It depends on how you implement "quality".
If you define implementing better quality by keeping the same build procedure but having more QA guys inspect at the end and pull off bad units for rework, this will kill you financially. You will be reworking too many units.
If you define implementing better quality by changing how the item is built in the first place to drastically reduce rejects at the end of the line, then it won't cost as much.
The key is to reduce the "ad-lib" part of the assembly. Ideally the assembly consists of fixtures and jigs to make sure every wire is routed identically, each unit has the exact same screw placement, etc. Unfortunately the cost of fixturing has to be divided up among the production run, so there is a tradeoff depending on how many units will be built with the fixturing. Adding $200 per unit to pay for fixturing may be OK,, but in a low volume production run adding $1000 may not be ok.
Unless worker-to-worker and unit-to-unit variables are removed, it simply isn't possible to build consistent units.
I agree it it mostly management decisions that affect quality rather than "bad" workers.
When the US auto industry was putting out poor quality products in the 80s and 90s, the auto workers were still highly paid. The difference now is the manufacturing process. GM is building some of the best cars and trucks in the world now. Were their engineers not as smart before? I doubt it. Amazing what having your back to the wall (bankruptcy) will do to motivate you. BTW, their current well-designed and high quality vehicles are still at the same price point in the market as they were when poor-quality cars were being built.
Steve