Purpose
Evidence of the impact of social interventions on the wellbeing of New Zealanders is at the heart of investing for social wellbeing. However, many of the wellbeing outcomes that social interventions target are difficult to measure, and have traditionally been considered only through qualitative evidence or limited proxy measures. Improving the evidence base on the broader wellbeing impacts of social interventions will involve using new analytical techniques and making the best possible use of existing data.
One area of current policy interest in New Zealand is the issue of social housing. While the provision of social housing is one of the larger social sector interventions by central government, the available information on the impact of social housing on the wellbeing of recipients is relatively limited. This informed the SIA’s 2016 work in developing a Social Housing Test Case that focused on the net fiscal impact of placement in social housing. However, the fiscal impact of social housing is not a full measure of the net benefit of the intervention as it captures only the costs and omits the impact on the wellbeing of recipients.
This paper complements the Social Housing Test Case (Social Investment Unit, 2017) by applying the SIA’s wellbeing measurement approach (outlined in the accompanying paper, Are we making a difference in the lives of New Zealanders – how will we know?) to the case of social housing in order to provide information on the wellbeing outcomes of social housing recipients. In particular, the project aims to provide four key pieces of information to assess the impact of social housing interventions.
Methodology
The method adopted in this paper aims to move beyond a simple descriptive approach, to identify the difference in wellbeing outcomes for people before and after being placed in social housing. This is still not as good an estimate of the true causal impact of social housing as a genuine experimental evaluation, but by providing a dynamic picture of the change in wellbeing outcomes associated with a social housing transition, it significantly enriches the available evidence base.
To obtain information on the changes in wellbeing associated with a social housing transition, two main datasets are used.
The first of these is the Housing New Zealand (HNZ) Social Housing Dataset in the IDI that captures register information on applications for and placement in HNZ social housing.
The second main source of data is the four waves of the NZGSS included in the IDI (2008, 2010, 2012, and 2014). This is the primary source of information on wellbeing outcomes for individuals.
In addition to the two main data sources, MSD data in the IDI is used, in conjunction with the address register, to identify households in receipt of the accommodation supplement. Customs data is also used to provide information on spells overseas. The analytical approach adopted here involves three stages.
Key Results
Social housing is a policy intervention that is aimed first and foremost at improving housing outcomes. While a core part of the rationale for doing so is that better housing outcomes will contribute to better outcomes in other areas of wellbeing, the centrality of housing quality to evaluating the impact of social housing cannot be avoided.
For this reason, the results reported here are broken into two main sections.
The first section focuses exclusively on the impact of social housing on the housing dimension of wellbeing. This covers the physical characteristics of the house, as well as crowding and overall satisfaction with housing.
Going beyond housing, the second part focuses on the wider wellbeing impact of being placed in social housing. This looks at how the wellbeing indicators identified in table 3 vary between the before and after groups.
After considering the differences between the before and after groups, the third part of the results section looks at adjustments for response bias. This part of the report investigates the degree to which the imbalance in composition between the before and after groups produces a significant bias in the main findings.