Census 2023 - A Complete Failure or an Absolute Disaster?

The 2023 Census – a complete failure or an absolute disaster?

It was a census that Statistics Minister Deborah Russell said she would ‘absolutely’ stake her job on if there wasn’t a 90% response rate or higher. But as the collection process officially came to an end, and the raw figure putted out at just over 89%, not only did Russell keep her job, but she even had the nerve to call it all a ‘success.’

Maybe ‘success’ is in the eye of the beholder, but how anyone can look at this year’s edition and call it anything other than a disaster, frankly, is unfathomable.

For starters, the cost has been enormous. Initially, the budget was $210 million. That was already a significant jump from the funding used in 2018, but Stats NZ advised that it would only barely be enough to provide data in line with statutory requirements. As a result, the Government injected the Census with extra funding to combat growing operational costs and threw even more at it to mitigate the effects of Cyclone Gabrielle. Somehow, all up, the total figure ballooned to well over $300 million.

With the vast amount of funding available to Stats NZ, you’d think they would comfortably have enough to deliver a Census with excellent data and an efficient delivery process. As it turns out, apparently not…

The raw individual response rate ended at a meagre 89% meaning it falls under the 90% target set by Stats NZ through their KPIs. That figure could be even lower too given that the raw rate is calculated using duplicate responses and outdated population statistics. For context, the 2018 figure ended at an embarrassing response rate of 82%, a truly pathetic outcome given the cost. This year’s census has only provided marginally better results on a budget of nearly $200 million more.

It looks worse still when you compare these results to the censuses prior to 2018. In 2013 the national response rate was 92.9%, done on a third of the budget, and in 2006, the response rate was 94.8%. Consider that Stats NZ’s most expensive proposal for the 2023 Census of over $280 million aimed for an individual response rate similar to 2006 figures. With a budget of almost $40 million more than that, the 89% response rate we got isn’t just underwhelming, it’s appalling.

And not only did this census fail to deliver for New Zealand as a whole, it also disproportionally failed to appropriately count Māori and Pasifika for the third time running. In 2013 the response rate for Māori and Pasifika was under 90%. That was concerning enough, but nowhere near as bad as it was in 2018 where response rates fell to below 70% for both groups respectively. Stats NZ prioritised shifting that disproportion for the 2023 edition, yet the response rates in these communities have barely improved.

One might claim that the collection process was hampered by Cyclone Gabrielle, a disaster which tore right through Hawke’s Bay and Gisborne. But whilst that may excuse the disappointing outcomes through those regions, it fails to explain the lack of response elsewhere. Northland and Waikato were notably poor, and almost the whole North Island had responses below 90% going into the final week.

Despite these clear failings, Russell will pretend the KPIs don’t exist and continue to compare the results solely against the mess that was 2018, but this couldn’t be more misleading.

For context, the 2018 census was an experiment gone disastrous. It implemented a completely new design from its predecessors aimed at reducing costs and expediting the transition towards digitalization. Delivery was almost entirely structured through online access codes and distributed and collected using a heavily reduced ground crew. Not only did the team fail to keep costs down, but the online approach didn’t work…at all. Almost a fifth of New Zealanders failed to complete the census and nearly a third of all Māori and Pasifika

As a result, the independent review of the Census was scathing of the process. It was so damning that even Stats NZ’s Chief Executive, Liz MacPherson, couldn’t defend it and resigned. Core to its message was how the model undervalued the importance of paper in the delivery process and focused too much on the online-first approach, but it also acknowledged how the “aggressive reduction in the field workforce meant Statistics NZ had a reduced capacity to respond when the response rate began to fall below acceptable tolerance levels.” The review recommended that both points were addressed for future versions of the Census.

Consequently, changes were made to this year’s model. The number of paper forms initially deployed was increased and the field staff capacity boosted. Still though, 56% of households only received online access codes and the expansion of ground crew was well short of what was required. In the end, only 3500 workers were deployed making it barely half the size of the number in 2013.

Material and operational costs of delivering a traditional census have increased over the years, but the importance of quality data has never been more paramount. Given that the budget was well over twice the price of 2018 and 3 times more than 2013, why didn’t the team go back to basics? In all five of the initial proposals, Stats NZ had virtually the exact same delivery model. Given that resourcing ended up being practically infinite, Stats NZ could have sent paper forms to all households by default and possibly even doubled their ground crew.

Regardless of poor response rates and an inadequate delivery model, it was how Stats NZ incentivized people to complete their forms which was arguably the most concerning.

In May, Stats NZ partnered up with Warriors NZ using $150,000 of taxpayers’ money to provide free tickets and food to those who attended one of the games and filled out their Census form. It gave those stragglers a chance to complete their census and earn a prize and it was rightly denounced as incredibly unfair.

As it turns out, though, the Warriors Campaign wasn’t the only form of bribery used by Stats NZ. The New Zealand Taxpayers’ Union asked for costings on all incentives used to promote the Census. According to our OIA, both food and fuel vouchers were given out at up to $100 a head with a total budget of $1 million to support communities nationally and another $1 million for households in the Auckland region.

On the one hand, rewarding those who were too lazy to complete their forms on time is an insult to those who had already done so. More concerningly, though, it makes a mockery of the importance of completing the census. If people think that the only consequence for failing to fill in their forms is free food, why would anyone take it seriously.

Enforcing hefty fines on those who reject the Census is paramount for maintaining its national significance. Throwing millions of dollars at bribing those who have failed to complete what is legally required of them will only reinforce the apathy some New Zealanders already have towards completing their forms.

If providing an excellent Census wasn’t pivotal enough this year, then 2028 will absolutely be make or break. We cannot have our most important source of statistics be derived from poor data once again.

Russell will continue to hang her hat on the improvement from five years ago. Frankly though, for $317m, New Zealanders deserve much better.