Refinance vs Forgiveness: 2,500 Student Loan Scenarios

The "refinance or stay federal?" question is the most-asked one in student-loan personal finance, and it has more wrong answers than right ones. We swept 2,500 realistic borrower profiles to find the exact line where each strategy wins.

Finding 1: Profession beats balance

Among the 896 PSLF-eligible profiles, IDR + PSLF was the cheapest strategy in 93.2% of cases. Among the 1604 non-PSLF profiles, private refinance won 41.1% of the time. The single biggest determinant of the right answer is not your balance, your interest rate, or your credit score — it's whether your job qualifies for PSLF.

Practical implication: if you work for a 501(c)(3), government, or qualifying public-service employer, the math almost always favors federal IDR over refinance — and refinancing forfeits PSLF eligibility forever.

Finding 2: The RAP tax bomb changes the non-PSLF math

The new RAP plan (PL 119-21, effective July 1, 2026) forgives the remaining balance after 30 years, but unlike PSLF, the forgiven amount is taxable as ordinary income in the year of forgiveness. We modeled this at a 24% effective rate.

The effect: for non-PSLF borrowers with large remaining balances at year 30, the tax bomb often erases the apparent forgiveness advantage. Among non-PSLF profiles with balance ≥ $80,000, RAP beat refinance in only 0.3% of cases. The high-balance non-PSLF borrower is the population RAP looks most "generous" for on paper and least beneficial for in practice.

Finding 3: Refi tier pricing is the hidden lever

Private refinance APRs are sharply tiered by income and credit profile. In our model: $150k+ salary gets 5.5%, $90–150k gets 6.5%, $60–90k gets 7.8%, sub-$60k gets 9.5%. At the bottom tier, refi rates are higher than the federal Direct Loan rate of 7.20% — making refinance an objectively worse trade.

Refi only wins when (a) your salary is high enough to qualify for the cheap tier AND (b) you don't have a PSLF path. Both gates have to be true.

Finding 4: The gap between best and worst strategy is large

Median difference between the cheapest and most expensive of the three strategies per profile: $49,958. 90th percentile: $241,071. This is the cost of picking wrong — a profile-specific decision that no one-size-fits-all calculator can capture without inputs about your profession and salary trajectory.

Sample borrower profiles (8 across the salary spectrum)

Methodology

• 2,500 profiles, deterministic via Mulberry32 seed 20260524.
• Balance: $10,000–$250,000, weighted toward smaller balances (squared distribution).
• Starting salary: uniform $35,000–$200,000.
• Salary growth: uniform 1.5%–5.0% annual nominal.
• PSLF eligibility: 35% of profiles. Approximates the share of the labor force in qualifying public-service / 501(c)(3) employment.
• Federal rate: 7.20% (weighted average of 2026–2027 Direct Loan rates — Undergrad 6.53%, Grad 8.08%).
• IDR formula: 8% of (AGI − $20,000 poverty floor) / 12, monthly. Approximates new RAP tiered 1–10% bands averaged across middle-income borrowers.
• PSLF: 120 qualifying monthly payments, forgiven balance tax-free.
• RAP: 360 qualifying monthly payments, forgiven balance taxed at 24% (proxy for federal marginal bracket).
• Refi rates: tiered by salary — $150k+ → 5.5%, $90–150k → 6.5%, $60–90k → 7.8%, sub-$60k → 9.5%. Reflects 2026 private refi market.

Limitations

• RAP rule details are still settling. The Department of Education is mid-rulemaking on the tiered IDR formula. Our 8% flat is a deliberate simplification.
• SAVE plan injunction is modeled as 0% interest / forbearance for existing borrowers — they do not appear in this dataset because their counts are paused.
• Tax bomb modeled at flat 24%. Real RAP forgiveness lands in a single tax year and can push the borrower into higher brackets temporarily; effective rate could be 22–32%.
• State income tax not modeled. Most states tax forgiven balance; some (CA, IN, MS, NC, WI) don't. Adjust your personal estimate accordingly.
• Variable-rate refi loans not modeled. All refi quotes assumed fixed.