A comment compares estimated evasion and recovered sums as if they were the same indicator. I propose to verify definitions, period and perimeter of each number.
In the example, an estimate concerns unfulfilled obligations while a control result describes another activity. The comparison can only be informative if it maintains these differences.
Before you attribute merit to a technology, you need to read the method. An economic fact should not be used as evidence merely because it appears large or favorable to the thesis.