Building The Three-Column Comparison Over A Real Holding Period

We have set out elsewhere why the ordinary break-even calculation is the wrong tool for a cash-out refinance, and what replaces it: a three-column comparison run over a real holding period. That argument is at the cash-out refinance break-even.

This article is the construction manual. The method fails in practice for two mundane reasons — people choose a horizon that flatters the answer, and they populate the cells from advertisements rather than documents. Both are fixable.

Choosing the horizon, before you know what it does

Fix the holding period first, in writing, before you compute anything. Once you have seen how the columns move, the horizon stops being an estimate and becomes a lever.

The horizon is the number of years you realistically expect to hold the property and this financing. Not the loan term. Not thirty years by default. Anchor it to facts you already know: how long you have owned properties in the past, whether a move is foreseeable for work or family reasons, whether the house suits a household that is changing size.

If you genuinely cannot name a single number, name a range — five years and twelve years, say — and run the table twice. A method that produces the same ranking at both ends of your range is telling you something robust. One that flips is telling you the decision depends on a fact you do not have, which is itself the most useful output the exercise can produce.

The three columns, defined precisely

Column A — do nothing. Your existing first mortgage, unchanged, for the horizon. No new debt, no cash raised. This is the baseline, and it belongs in the table even when you are certain you need the money, because it is the only column that prices the option of not borrowing.

Column B — cash-out refinance. A new first mortgage replacing the existing one. The new balance is old balance + cash taken + any costs you finance, at the newly quoted rate, on the newly quoted term. This column replaces the existing mortgage including its rate — that is the defining mechanical fact of the instrument and the reason the comparison exists at all.

Column C — keep the mortgage, add a second lien. Existing first mortgage untouched, plus an equity loan or HELOC sized to the cash requirement only, at its own quoted rate, term and costs. Two payments, two schedules.

The scoring rule

For each column, compute exactly two numbers at the horizon:

1. Total paid over the horizon — every scheduled payment on every loan in that column, from month 1 to the horizon month.
2. Ending balance at the horizon — what you still owe on everything in that column at that month.

Score each column as the sum: total paid + ending balance.

That sum is what makes the comparison honest. Total paid alone rewards any column that defers principal; ending balance alone rewards any column with a punishing payment. Together they capture the whole cost of arriving at the horizon in that structure. Monthly payment does not appear in the scoring at all — it is a constraint you check separately for affordability, never a measure of cost.

If Columns B and C raise different amounts of cash, they are not comparable. Size them to the same requirement — an amount fixed from your own itemised need, not from whatever maximum a lender is willing to approve.

The figures in those cells must come from Loan Estimates, not from rate tables, advertisements or verbal quotes. The Loan Estimate is a standardised federal form, and its whole design purpose is line-by-line comparison between offers — identical categories in identical places. Use it as intended. Its cost sections are where every origination, title, recording and prepaid charge must appear.

For Column A you need only your current statement and amortisation schedule, which your servicer can provide.

For Column C, if the second lien is a variable-rate line, the column has a range rather than a point. Run it at the current rate and at two adverse assumptions of your own choosing, and state the assumptions on the page. A variable column presented as a single number is not a comparison, it is a guess with a decimal point.

A worked structure, with stated assumptions

Illustrative. No rates. Substitute your own.

Assume an existing first-mortgage balance of $268,000, a cash requirement of $55,000, and a horizon of 7 years.

• Column A: 84 payments on the existing loan; ending balance from your current schedule at month 84.
• Column B: new balance = 268,000 + 55,000 + financed costs, at quoted rate and term; 84 payments; ending balance at month 84 from the new schedule.
• Column C: existing loan's 84 payments and month-84 balance, plus a $55,000 second lien's 84 payments and month-84 balance, plus its costs paid at closing.

Twelve cells. That is the entire model.

Two structural patterns in the result are worth knowing before you look, because they explain most outcomes.

If your existing rate is meaningfully below what is quoted today, Column C usually wins, because Column B re-prices your whole balance rather than only the new money. That specific trade is examined at surrendering a below-market first mortgage rate.

If Column B's total paid looks attractive but its ending balance is high, you are looking at re-amortisation — a lower payment coexisting with a higher lifetime cost. See the re-amortisation trap.

What we would do

Fix the horizon before computing. Build all three columns from Loan Estimates. Score on total paid plus ending balance. And note that every column except A puts additional debt against the home, which is the collateral. This is reporting on method, not personalised advice.