Like everything else, 'high-impact grants' suffer from diminishing returns. The first few will be obvious and worthy causes: the next few will possibly provide a benefit; but eventually institutionalising handouts will have the same result it always does: subsidising ruthless opportunists who know how to pull strings, at the expense of those who really deserve it. Here's an idea: why don't we just let everybody give as much or as little as they want to whoever they think deserves it at the time? Admittedly, that might not yield million-dollar salaries for charity CEOs, like these:

http://www.charitywatch.org/hottopics/Top25.html

but it seems a lot more enlightened than your model.