Picture a Rotorua family in December 2013, checking their grocery receipts. The local food price index sits at 15,317. Twelve months later, same family, same supermarket, roughly the same basket of goods. The index reads 7,683.

That's a 50% drop. But nobody's celebrating, because nobody's food bill actually halved. What happened instead tells you something uncomfortable about how we measure the cost of living in New Zealand.

Stats NZ changed how it calculates the index. Not because the old method was wrong, but because measurement systems evolve. Base years shift. Categories get redefined. Methodologies update. And when they do, the numbers can leap or plunge in ways that have nothing to do with what you're actually paying at the checkout.

This matters because every argument about inflation rests on these indexes. When politicians debate whether benefits should rise, they cite food price data. When unions negotiate wages, they point to cost-of-living figures. When the Reserve Bank sets interest rates, it's tracking these same numbers.

The Rotorua data is particularly stark. From 2010 to 2013, the index climbed steadily: 14,492 in 2010, 15,309 in 2011, holding around 15,300 through 2012 and 2013. Then the 2014 methodology shift, and suddenly we're at 7,683. You can't compare 2014 to 2013 anymore. The scale changed mid-measurement.

Stats NZ doesn't hide this. Metadata notes explain when methodologies shift. But most people don't read metadata. They see a number, assume it's comparable to last year's number, and draw conclusions. That's how bad policy gets made.

Here's what the Rotorua case teaches us: price indexes are tools, not truth. They're useful for tracking relative changes within a consistent methodology. But when the methodology changes, historical comparisons become meaningless. You're measuring apples in 2013 and oranges in 2014, even though both are labelled "fruit."

This isn't unique to Rotorua or food prices. It happens across datasets. Employment figures. Crime statistics. Health outcomes. Every time Stats NZ improves its methods, it creates a break in the time series. Progress in measurement creates problems for comparison.

The fix isn't to stop improving measurement. It's to treat historical data with more scepticism. When someone tells you prices today are X% higher than five years ago, ask whether the methodology stayed constant. When you see a dramatic drop or spike in official figures, check whether it's real change or statistical housekeeping.

Because somewhere in New Zealand right now, a politician is citing food price data to justify a policy decision. And they probably haven't checked whether the numbers they're comparing were measured the same way. That gap between data and understanding is costing us all more than higher grocery bills ever could.

(Source: Stats NZ, food-price-index-detailed)