Introductory briefing to the Minister of Statistics: Measuring Child Poverty Date: 2 November 2017 Priority: Medium Security level: In confidence File number: MM1736 Contact details Name Position Telephone First contact Grace McLean Diane Ramsay Key messages Private Secretary to the Minister of Statistics Senior Manager, Labour Market & Household Statistics [redacted under s9(2)(a)] [redacted under s9(2)(a)] [redacted under s9(2)(a)] [redacted under s9(2)(a)] NZ does not have an agreed approach to the conceptualisation and definition of child poverty. A range of child poverty measures are currently produced by MSD using data from Stats NZ s Household Economic Survey. The data currently produced has some limitations for the ongoing measurement of poverty reduction targets, including small sample sizes that mean data cannot be meaningfully broken down for smaller sub-populations. International experience, including from the UN, recommends that a dashboard of indicators should be used in the measurement of poverty, including both monetary and non-monetary measures. Officials would be happy to discuss how measures can align with government aims. Establishing the measures and thresholds for measuring and monitoring poverty requires making value judgements. To ensure judgements are well informed, there is a strong argument for Stats NZ to be an active player in the development of measures of child poverty, particularly with regard to the methodologies and data quality. As well as working with others to develop child poverty measures and targets, Stats NZ should lead or co-lead the reporting against these targets. This would make use of Stats NZ s expertise and reputation for independent and robust advice. We are already working with Treasury, DPMC, Oranga Tamariki, and the Ministry for Social Development on a range of options for better measuring and tracking child poverty. This working group expects to advise the Prime Minister on the development of draft legislation and the medium-to-longer term options attached to this briefing, by the end of this week. 1
As Stats NZ is involved in the working group and developing advice for the Prime Minister, you may wish to seek greater involvement in the ongoing development of child poverty reduction measures, targets, and reporting. Eleisha Hawkins Director, Office of the Government Statistician and Chief Executive Hon James Shaw Minister of Statistics Date: Purpose 1. The government has committed to reducing child poverty in New Zealand. Stats NZ can play a lead role in the design and implementation of better ways of measuring and reporting on child poverty. Background There are a number of different approaches to measuring child poverty in New Zealand 2. A suite of child poverty measures is currently published in the Ministry of Social Development s annual report on Household Incomes in NZ. This same suite of measures is published in the Commissioner for Children s Child Poverty Monitor. The measures include: Fixed line income measures (where the poverty line is set in a given year, and inflation adjusted for other years) Moving line income measures (where the poverty line each year is set as a proportion of median income) Material deprivation measures (where living conditions are severely affected by a lack of resources) Severe poverty (intersection of income poverty and material deprivation) Poverty depth (looking at how far the poor are from the poverty line) 3. The number and proportion of children living in poverty varies across these measures and the threshold used. The overlap in the income and material deprivation levels is not high. 4. The measures are based on data from the Household Economic Survey (HES), which provides data on income, expenditure, deprivation/hardship and net worth: The income measures are based on household disposable income before and after housing costs, using 50% and 60% of household median income. The deprivation measures are derived from the Material Wellbeing Index (developed by MSD) that gives direct information on the day-to-day living conditions that household s experience. Households going without 7 and 9 items (out of 13) are used as thresholds for identifying those children experiencing hardship. 5. In addition to the above suite of measures, there is interest in measuring persistent child poverty, for which there is currently no adequate data source. 2
6. As well as being an issue of importance to New Zealanders, the eradication of poverty is a Sustainable Development Goal, and there is an expectation that we will monitor our progress towards this goal, including amongst children. NZ signed up to the SDGs in 2015. We can learn about measuring child poverty from international practice 7. Child poverty is a complex multi-dimensional phenomenon. There is a lack of consensus about what poverty means in developed nations. There is no internationally agreed standard for measuring poverty. The way that it is measured varies from country to country. 8. The range of approaches used in the measurement of poverty cover both monetary and non-monetary approaches, relative and objective, and include the following types of measure: Income Deprivation/material hardship Consumption/expenditure Subjective (feeling you don t have enough) Multi-dimensional 9. However, measuring child poverty is also sensitive to decisions about how income and deprivation, and elements that make them up, are defined for measurement. These decisions can result in different poverty levels being reported at a given time and to differences in the reported composition of those identified as poor. 10. A United Nations Economic Commission for Europe Taskforce on Poverty recently recommended that a dashboard of indicators should be used in the measurement of poverty, including both monetary and non-monetary measures. There are a number of key issues to be addressed in the measurement of child poverty in NZ 11. NZ does not have an agreed approach to the conceptualisation and definition of child poverty to underpin its measurement, and there is limited capability and expertise in NZ in the measurement and analysis of poverty. 12. The small sample size of the main current data source (the HES sample of 3,500 to 5,000 households - only a third of which contain dependent children) constrains the amount of monitoring and analysis of child poverty that is possible, particularly for small groups such as Māori and Pacific populations, different age groups of children, beneficiary groups etc. 13. Annual income data relies on respondents recall. This can be subject to error due to the respondent omitting some income sources or being unaware of income details. 14. The HES is run continuously over a year from July to June with respondents being asked about their income over the previous year. Therefore data can refer to a two year period and may not pick up impact of policy changes for some time. 15. Disposable income estimates are modelled via Treasury s Tax and Welfare Analysis model and take some time to produce. 16. Administrative data such as tax records in Stats NZ s Integrated Data Infrastructure (IDI) are at an individual level and it is difficult to identify families and households. Further, administrative data does not currently cover deprivation measures or housing costs. 3
17. Measuring persistent poverty is problematic as we do not currently have a survey that tracks a cohort of people over time. Changes in household composition in administrative data are hard to track. 18. Establishing thresholds for measuring and monitoring poverty requires making value judgements. Next steps Officials will brief the Prime Minister this week 19. We are working with Treasury, DPMC, Oranga Tamariki, and the Ministry for Social Development on measures of child poverty. This working group expects to provide written advice to the Prime Minister on the measurement options outlined in the appendix, by the end of this week. 20. As well as laying out the options, the paper will cover how legislation will be drafted to give effect to the child poverty targets and measures. Stats NZ is well-positioned to play a lead role in the development and implementation of measures 21. The attached options lay out how Stats NZ and other agencies may collect data to measure and track child poverty. Whichever approach is chosen, it is likely that additional resources will need to be allocated to this work. 22. Stats NZ will remain actively involved in any discussions related to the measurement of child poverty, and participate in any inter-agency groups: We will continue to provide advice on the quality of existing data sources, and on the strengths and limitations of existing data sources (both survey and administrative) for the measurement of child poverty. We will guide decisions on which of the existing suite of indicators should be used in setting targets, from a data and methodological perspective. 23. As well as working with others to develop child poverty measures and targets, it is our view that Stats NZ should lead or co-lead the reporting against these targets. This would make use of Stats NZ s expertise and reputation for independent and robust advice. 24. Such a role may be similar to the one we play in the production and sharing of environmental reports, under the Environmental Reporting Act 2015. Our role under the Act has been instrumental in ensuring that the focus of discussion is on the issues and implications identified, rather than the underlying data. As Minister of Statistics, you may want to seek to play a key role in Ministerial discussions and meetings with officials 25. As the Minister of Statistics you have a key role in promoting the use of data in understanding New Zealand. To reflect this, you may wish to be more deeply involved in the ongoing development of child poverty reduction measures, targets, and reporting. For example, you could seek to be included in the working group s meetings with the Prime Minister, and receive copies of written advice provided. 26. Through this involvement, you could recommend Stats NZ taking a lead role where it fits with our expertise and reputation. 4
Appendix officials have identified a range of options to explore further 1a. Increase the sample size of HES to improve the robustness of data on child poverty 1. A much larger sample of households (of around 14,000 households about 5,000 with children) would be needed to produce information of the quality needed for monitoring a child poverty target. 2. Pros 3. Cons Would allow the production of more robust estimates of child poverty, including of sub-groups within the child population eg Maori, different age groups of children etc. Would produce data for income and deprivation measures, and subjective poverty measures if required Would improve the quality of poverty data for the whole population not just children Could be implemented relatively quickly if funding was available, ideally for the 18/19 year (first data available in late 2019). Would increase respondent burden, in what is already a burdensome survey, at a time when we are seeking to move away from surveying to use of data from administrative sources Would require further funding of around $2-3 million per year. 1b. Change HES from an annual survey to a two-yearly survey, and use the savings to increase the survey s sample size 4. Pros 5. Cons As for 1a, but would not require additional funding. Would take longer to implement than option 1a Would make the monitoring of child poverty more difficult through having fewer data points: could not report annually on targets Sample size increase not as large as in option 1a Would affect the frequency of collection of expenditure and net worth data (currently collected every three years) Would cause operational issues in Stats NZ through uneven distribution of work across years. 2. Make greater use of administrative data 6. We could improve the quality of administrative data on child poverty, particularly through the addition of household and demographic details (from linking census and/or survey records to admin data). 7. Pros: Consistent with Stats NZ and public service plans to reuse existing data Would enhance the value of administrative data more broadly not just data on child poverty Would not increase respondent burden 5
Would make the measurement of persistent income poverty possible, in addition to other income poverty measures (if we could measure changes in households) Would be more cost-effective than options 1a and 1b. 8. Cons: Would take longer to implement than the HES expansion option Production of non-monetary indicators would not be possible Could require further funding, which is hard to quantify Time lags for reliable information on certain sources of income such as selfemployed can be up to two years from the end of the tax year. 3. HES-Administrative data hybrid option 9. A combination of increasing the sample size of HES and replacing direct collection of income, by linking to tax data in the IDI. 10. Pros 11. Cons Better quality income data than that collected from respondents Reduced respondent burden compared with direct collection of income data Would be able to produce longitudinal data to measure poverty persistence Reduced burden on respondents may allow us to collect additional data. Could be some timing delays particularly with self-employment data Likely to take some time to implement Would require additional funding, as per option 1a, with indicative additional funding required of $2-3 million per year. 4. Develop a dedicated Child Poverty Survey 12. A dedicated child poverty survey could include a longitudinal component to measure poverty persistence. 13. Pros: 14. Cons: Would be able to be specifically designed to produce the required data on child poverty Would be possible to produce a rich set of data for the analysis of child poverty Could be integrated with other initiatives to collect data on children or youth. Would need to be repeated annually to monitor the child poverty targets Would increase respondent burden, whereas we are generally seeking to move away from surveying to more use of data from administrative sources Would be more costly than the previous two options, with indicative additional funding required of $3- $4 million per year. 6