> If a code refactoring can increase dev productivity and shipping speed, then that is certainly worth something

The value of that is being distributed in very small increments over the whole dev team in the span of years. It's impossible to measure.

> Infrastructure updates can directly reduce server costs, latency, or have other benefits.

Well, most of the work is in that "other benefits". Very difficult to measure.

> Employers hire employees to increase the profitability of their business, and code quality is not the top priority.

With your incentive model it has 0 priority and the project will die under the weight of technical debt, inability to add features in reasonable time and critical defects appearing out of the blue.