One interesting feature of this proposal is the concept of cliffs.

In some sense, it's anti-cliff: no threshold, no rising balance => forgiveness.

But using graduated rates to calculate payments could create a new cliff effect, especially b/c rates are assessed on total rather than marginal AGI.

From the borrower perspective:

Pro: Balances are guaranteed to fall over time, as long as borrower pays at least $10/month

Con: Monthly payments are likely to be be higher

Eliminating the threshold has big impacts:
1. Low-income borrowers (i.e. those who would otherwise fall under the earnings thresholds) would have to make monthly payments.
2. Monthly payments are calculated as % of AGI rather than % of discretionary income (i.e. income above threshold).

The new House GOP student loan plan looks similar to the Australian model: payment % is graduated (1%-10% depending on income) and no forgiveness. Main differences:

1. No earnings threshold
2. No interest capitalization
3. Principal balances guaranteed to fall by at least $50/m

This "Pay For Success" model isn't a panacea: some programs may not be a good fit, and there are transactions costs relating to developing/running the model, which requires a contractual partnership across multiple organizations.

But these challenges can be overcome as the model matures.

Making government more efficient is tough.

@socialfinance.org proposes a common-sense (and bipartisan!) approach: what if government programs paid for outcomes rather than inputs?

Taxpayer dollars get higher ROI, and states & private sector get freedom to innovate new approaches.