College savings —
Bucket balances vs. annual expenses
Cumulative contributions vs. cumulative expenses
Terminal disposition — where the surplus lands and what it's worth
Flexibility vs. tax efficiency — split comparison
Per-child cost summary
Year-by-year projection
Cost data & 529 reference
College Board — 2025-26 published averages
Education inflation — what the data actually shows
Recent nominal increases (2024-25 → 2025-26): public in-state +2.9%, public out-of-state +3.4%, private +4.0%. Against ~2.7% CPI that is roughly 0–1.4% real.
Last decade in real terms: average published tuition and fees declined in the public sectors and rose only ~2% cumulative at private four-years. The "college compounds at 5–6% forever" heuristic describes 1980s–2000s dynamics, not the current regime.
Suggested settings: 4% nominal education inflation (≈1% real at 3% CPI) central; 5% conservative stress; CPI-matching optimistic.
Sticker vs. net price
This model plans at sticker price — the conservative bound. Net tuition & fees after grant aid average ~$2,300 at public four-years and ~$16,910 at private four-years (2025 dollars, first-time full-time students including non-recipients). Merit-driven institutional discount rates at privates exceed 50% on average.
A sticker-price plan that exactly funds is overfunded in expectation. That asymmetry is the core argument for keeping part of the pool in the flexible bucket.
Financial aid parity
Parent-owned 529s and parent taxable brokerage are treated identically on the FAFSA — both parental assets assessed at a maximum 5.64% in the Student Aid Index. The 529-vs-brokerage split is therefore a pure tax/flexibility trade with no aid consequence. (Retirement accounts are excluded entirely; home equity is excluded on FAFSA but not CSS Profile schools.)
Colorado specifics (CollegeInvest)
Overfunding escape routes (in priority order)
AOTC collision with the conversion window
The American Opportunity Tax Credit ($2,500/yr per student, first 4 years) requires paying ≥$4,000 of tuition from non-529 dollars — but phases out at $160–180K MAGI (MFJ), a threshold frozen in law. College years that overlap a Roth-conversion window running MAGI to the 24% bracket ceiling forfeit AOTC entirely. If conversion-window MAGI is a given, there's no reason to reserve brokerage dollars for tuition to capture it.
Model conventions
Engine nominal; display toggle deflates by CPI. Both buckets grow at the nominal return; brokerage return is reduced by the annual tax-drag input (dividend/distribution taxes). Basis is tracked in both buckets: contributions add basis, withdrawals consume it pro-rata.
Spending: 529-first is the default — qualified 529 withdrawals are tax-free, and every 529 dollar spent on school is a flexible dollar preserved. Brokerage sales are grossed up for LTCG on the embedded-gain fraction at sale time.
CO rebate: modeled as a cash inflow of state-rate × 529 contributions, credited to the brokerage bucket each year (that's where a tax refund actually lands).
Solvers: required monthly holds the chosen split and finds the contribution such that combined balance never dips below zero. Not modeled: aid, scholarship probability, AOTC, per-plan fees.