The refinance cashback math, explained.

Why banks offer cashback

Switching a mortgage from one bank to another is high-margin business for the new bank. They acquire an asset with predictable payments and a long life. The cashback is acquisition cost, the same way a phone company offers $200 off to switch network. For a $700,000 refinance, the new bank earns roughly $25,000 to $40,000 of margin over the first fixed term. A $5,250 cashback (at $750 per $100k) is a sensible acquisition cost against that lifetime value.

For you, the cashback is just money. Paid into your nominated account at settlement, usually within 5 to 10 business days of the refinance completing. No tax on it (the IRD treats it as a discount, not income). No conditions on what you spend it on.

The claw-back clause

This is where the math gets interesting. Almost every NZ bank that pays cashback requires you to stay with them for a defined period, usually 3 or 4 years. If you refinance away before then, the cashback is pro-rated back. Some claw-backs are linear (after 24 of 36 months, you owe 1/3 of the cashback). Some are stepped (full claw-back inside year 1, half claw-back year 2, zero claw-back after year 3).

The claw-back is the bank's downside protection. If you switch again next year for another bank's $6,000 cashback, the first bank wants its money back. The implication is that cashback is not a one-off windfall. It is paid in exchange for a forward commitment.

The formula we use

For any refinance-vs-stay conversation, the calculation is:

1. New bank cashback, minus
2. Break cost at the current bank (if there is one; see our break cost piece), minus
3. Legal and transition fees (usually $1,000 to $1,500), plus
4. Rate savings at the new bank over the claw-back period, expressed in dollars,
5. = total net benefit over the lock-in period.

If that number is meaningfully positive (we look for $10,000+ net benefit on a typical $700k refinance), the move is worth doing. If it's break-even or slightly positive, the move is probably not worth the disruption.

A worked example

Indicative numbers only.

Current: $720,000 mortgage at 6.49% with 12 months remaining on the fixed term.
New bank offer: 5.79% on a 2-year fix, $5,400 cashback ($750 per $100k), 4-year claw-back.
Break cost on current loan: $4,200 (quoted in writing).
Legal/transition fees: $1,200.

Rate savings: 0.70% × $720,000 × 2 years (new fixed period) = ~$10,080. Then for the 2 years after that, you're free to move again or stay; the claw-back period extends to year 4 but the meaningful rate-saving comparison is for the new fixed term.

Net benefit over the 2-year fix: $5,400 + $10,080 − $4,200 − $1,200 = $10,080.
That's worth doing. Comfortably.

When cashback is a trap

Three patterns we see where the cashback math looks good on paper and doesn't work in practice:

1. You're going to refinance again next year. If your circumstances are likely to change (buying an investment property, separating, selling the house), getting locked into a 4-year claw-back at one bank can cost you more on the next refinance than the cashback saved you.
2. The new bank's servicing test is tighter. If your borrowing capacity at the new bank is meaningfully lower, the cashback is irrelevant because you can't actually move the same loan amount across.
3. The cashback compensates for a bad rate. If the new bank's carded rate is 30 bp above the market and they're offering $1,000 per $100k cashback to compensate, you may pay back the cashback in higher interest before the lock-in is up.

What the broker actually does

We model the full picture with all three or four real cashback offers on the table at the same time. We get break-cost quotes from the current bank in writing. We compare the carded rates the new banks would actually offer you (which are not the same as their advertised cards once your specific income and LVR are loaded). And we look at the claw-back window against your likely 4-year story.

The answer is usually "move, take the cashback, lock in the lower rate, accept the claw-back". It is not always that. A 15-minute conversation tells us which.