Look at page viii on your link. "Hardware costs constitute about half the total price of our small-battery systems. The largest single hardware cost for these systems is the 6-kW battery-based inverter ($3,596), followed by the PV array ($3,584) and the lithium-ion battery ($3,000)."

PV only is the orange box at the bottom. The cheapest option is 3,584 for the panels and total cost of 15,581. The most expensive option is still just 3,584$ for panels and a total installed cost of 47,171$. Trying to drop that 3,584$ portion is not going to make much difference a this point. But, needing to install more panels would significantly increase other costs.

Agreed, and thanks. It's clear that dropping panel costs really wouldn't affect the installed bottom line much.

I just focused on installation costs, which in all cases seem far less than 50% of the overall system cost. This suggests that there isn't much economic leverage in reducing (or preventing growth of) installation cost, either. Only in the large battery cases is there a single component (the battery) that dominates other cost components.

Also notice that this is for solar with energy storage. If you want to tap into the grid then you need a permit and an electrician in addition to an emergency cut off for firemen. That ends up being $Ks which are related to installation (not panel/inverter/mounting) rather than a battery.

Check Project Sunroof for a quote: https://www.google.com/get/sunroof

It's not uncommon for installation costs to be over 30% (especially for smaller installations and depending on roof conditions), and when you only get the Fed subsidy when you pay taxes that number ends up looking bigger (even if it's the same % in the end).

Inverters are improving too. New semiconductors are being developed and improved.