In the dataset for monthly deaths by registration date, I think the deaths probably peak in August 2022 and not July 2022 because of a registration delay. The deaths peak in July in a dataset for monthly deaths by date of occurrence at Infoshare: https://twitter.com/mongol_fi/status/1759489040255001054/photo/1. The dataset is here under "Population > Deaths (VSD) > Month and year of death (Monthly)": https://infoshare.stats.govt.nz/.
The reason why the OIA response has a lot of deaths missing in 2023 in ages 0-20 could be because younger age groups have a longer registration delay than elderly age groups. The ONS has published a dataset about the impact of registration delay on mortality statistics: https://ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/impactofregistrationdelaysonmortalitystatisticsinenglandandwales/2021. In sheet 10 the spreadsheet included with the dataset, for example for deaths in ages 1-14 registered in 2021, 15% of deaths had a registration delay of over a year and 12% had a registration delay of 6 months to one year. Sheet 2a also shows that deaths from external causes have a long registration delay on average, but deaths from external causes account for a large proportion of deaths in ages 0-20.
You wrote that "OIA deaths & NZ doses administered data include non-residents". But how do you know that the OIA includes non-residents?
I make the assumption based on the fact that early 2019 months are accurately matching with the non-resident deaths, and that the OIA dataset has more deaths than the dataset containing non-resident deaths. I doubt that overseas deaths + resident population could overcome resident + non resident deaths on the island with a mostly positive net immigration. I don't see what else could be counted - aside for potential double-countings in the OIA dataset. But you're right this assumption should be confirmed.
In the dataset for monthly deaths by registration date, I think the deaths probably peak in August 2022 and not July 2022 because of a registration delay. The deaths peak in July in a dataset for monthly deaths by date of occurrence at Infoshare: https://twitter.com/mongol_fi/status/1759489040255001054/photo/1. The dataset is here under "Population > Deaths (VSD) > Month and year of death (Monthly)": https://infoshare.stats.govt.nz/.
The reason why the OIA response has a lot of deaths missing in 2023 in ages 0-20 could be because younger age groups have a longer registration delay than elderly age groups. The ONS has published a dataset about the impact of registration delay on mortality statistics: https://ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/impactofregistrationdelaysonmortalitystatisticsinenglandandwales/2021. In sheet 10 the spreadsheet included with the dataset, for example for deaths in ages 1-14 registered in 2021, 15% of deaths had a registration delay of over a year and 12% had a registration delay of 6 months to one year. Sheet 2a also shows that deaths from external causes have a long registration delay on average, but deaths from external causes account for a large proportion of deaths in ages 0-20.
You wrote that "OIA deaths & NZ doses administered data include non-residents". But how do you know that the OIA includes non-residents?
I make the assumption based on the fact that early 2019 months are accurately matching with the non-resident deaths, and that the OIA dataset has more deaths than the dataset containing non-resident deaths. I doubt that overseas deaths + resident population could overcome resident + non resident deaths on the island with a mostly positive net immigration. I don't see what else could be counted - aside for potential double-countings in the OIA dataset. But you're right this assumption should be confirmed.
Thanks for the details on the rest.