## Monday 9 May 2022

### Tax thresholds

The income tax thresholds haven't been touched since 2011, except for the addition of a 39% rate from \$180,000 on 1 April 2021.

I was a bit curious what the thresholds would look like if you indexed them for growth in wages over the period.

So I took the HLFS private ordinary-time average hourly wage series from SNZ and used that as an inflator. Whenever accumulated inflation warranted a \$1000 step change in a threshold, I applied the step change. So the \$14,000 threshold advances to \$15,000 when it should be at \$14,501, but stays there until it should be \$15,499.

I set it so these would only ratchet on 1 April each year.

Anyway, it looks like this.

The tax thresholds for 1 April 2022 would be:

 Band Rate \$0 - \$20,000 10.5% \$20,001 - \$69,000 17.5% \$69,001 - \$101,000 30.0% \$101,001 - \$188,000 33% \$188,001 + 39%

Treasury's calculator says the government would lose about \$4.1b in revenue if it updated the tax bands to the ones above, but it warns that it's not reliable over such large changes - and the calculator doesn't include the 39% band.

This way of inflating is pretty crude. A better one would track wage percentiles.

If \$14k was the nth percentile of the income distribution in 2011, what would be the number at the nth percentile now?

If we held the percentiles constant, the tax bands would have been, for 2020/21:

 Band Rate Percentile at top of band \$0 - \$25,000 10.5% 28th \$25,001 - \$65,000 17.5% 68th \$65,001 - \$90,000 30.0% 85th \$90,001 - \$180,000 33.0% 98th \$180,001 + 39%

Or another way of thinking about it. Keep the bands constant, as successive governments have, what happens to the percentile?

 Band Percentile 2011/12 Percentile 2020/21 \$0 - \$14,000 28th 19th \$14,001 - \$48,000 68th 50th \$48,001 - \$70,000 85th 73rd

I dropped the higher bands on this one because there was no bracket higher than \$70k in 2011/12. But \$180k is the 98th percentile in 2020/21. The 98th percentile in 2011/12 was \$138,000.

Note: the first percentile table here is updated to correct a cut-and-paste error.

UPDATE: All of the percentiles here from the IRD data are percentiles among wage and salary earners. The IRD dataset I'm using is only on wage and salary earnings. Excluded are:
• NZ Super
• Taxable welfare benefits
• Student allowances
• Earnings-related ACC payments
• Shareholder employee salaries (since there was no PAYE deducted).
The main effect of indexing to earnings among wage and salary earners would be that those on benefits that do not ratchet upward with wage and salary earnings would see themselves on lower tax rates over time, if they earn enough in non-wage/salary benefits to cross a threshold. I don't know that that's a bad thing for benefits that aren't indexed.