Housing Tenure And Poverty: A Note

Note written for circulation in March 2024

This note explores housing tenure in the part of the distribution where the poverty line is, defining the line by the SNZ material hardship indicator. The note does not explore the AHC income-expenditure measure,[1] partly because there are insufficient observations but mostly because, as explained in the appendix, the measure is not very satisfactory. The numbers in the notes are only for children and do not include adults.

The statistical analysis treats the material-hardship index as continuous. The current index is not – it simply adds together responses to 17 questions about hardship weighted equally.[2] A more satisfactory index could be constructed but has not been. It would be nearer continuous. We are assuming the existing index mimics the better index reasonably well. The analysis uses linear interpolation which usually works reasonably well over the limited ranges explored here. (Critics are welcome to propose non-linear models.)

The data source is from the Table E3 ‘The material wellbeing of children in selected household contexts (6 groupings using MWI scores): HES 2018-19’ in Child Poverty in New Zealand.[3] Note that price, incomes, interest rates and rents have changed since then. Probably so has hardship.

The following table shows some key distributional statistics by tenure.

Table: Proportions in Material Hardship Categories by Housing Tenure.

<>             2018-19 Percent of Children in Material Hardship by Tenure             Tenure Total Below 10% 10-20% Owned with mortgage 48.4 14.9 29.5 Owned no mortgage 10.5   3.3  2.5 Private rental (no assistance) 17.8 15.0 18.6 Private rental (assistance) 16.0 41.8 33.2 Social Rental   7.3 25.0 16.2 Total 100 100 100

Source: Table E3, Child Poverty in New Zealand

The Table reports that 48.4% of children live in owned households with mortgages and another 10.5% of children in owned households without mortgages so almost three in five children live in homes owned by their parents (or guardians). The other two in five live in private rental accommodation with or without assistance (33.8%) and publicly provided rental accommodation (7.3%).[4]

However, the proportions are very different for those at the bottom of the hardship scale, with under one in five children (18.3%) in the bottom decile living in owned homes. That proportion increases to almost one in three (33.0%) in the second bottom decile with the mortgages proportion increasing faster than the non-mortgage proportion. The converse is close to three-fifths living in private rentals in the bottom decile (56.8%) and a quarter living in public rentals (25.0%). The proportions in rentals in the second to bottom decile fall to just over a half in private rentals (51.8%) and a sixth in public ones (16.2%).

For an illustrative exercise, suppose the poverty line was equivalent to 15% of the material hardship index. There would be 164,000 children living below this line. Of them, 17.8% are in homes with mortgages, 3.1% in homes without mortgages, 65.9% in private rentals and 23.2% in public rentals.

Suppose the 15% poverty line was reduced to 14%. There would now be 11,000 fewer children on this new poverty measure. Some 35% of them would be living in homes with mortgages and another 2% would be in mortgage-free homes. The proportions living in private rentals would be 50% and another 13% in the public rental. The proportions living near any poverty line can be very different from the total proportions living below that poverty line.

(This result is not so much a paradox but illustrates the caution that while poverty lines are a way of measuring poverty. They are clumsy. For instance, one can envisage a policy which reduced the number below the poverty line by providing extra income to those near it, funded by reducing the funds for the poorest. There are more sophisticated measures of assessing a distribution for poverty but they are neither used, nor explored, in New Zealand.)

This has been a static comparison looking at the (hardship) distribution in a given year (2018-19). The lesson for comparisons through time is that the proportions in poverty will change if the tenure groups have different experiences – in particular in the current instance, if the price of home ownership including mortgage interest rates changes relative to the price of private renting and/or the price of public renting. While this note has explored in terms of material hardship, the conclusion is equally true for income-based poverty lines.

Appendix: The After Housing Cost Measure

The MSD report also provides a similar tabulation using a measured called AHC of Equivalised Income after housing costs are deducted (Table B2). The same analysis as for the Material Hardship Index could be carried out, although it would be even further limited because fewer points on the scale are reported. The text mentions in an endnote, that the measure is a curious one consisting of subtracting an income from an expenditure – applied mathematicians warn against mixing concepts in this way.

However, there is an even more serious weakness of the AHC measure when it is used to evaluate poverty. The income less expenditure measure is adjusted for household composition by dividing by an equivalence scale. This makes the comparison between a one- and two-person household, say, more meaningful. The equivalence scales assume there are economies of scale for larger households, so they are not per capita ones (and they treat children as being less than an adult). The most important source of economies of scale is housing. Indeed, it is likely that in all expenditure other than housing the economies of scale are negligible and that a per capita scale (with a lower amount for children) would provide a better comparison after housing expenditures are deducted. However, the AHC measure uses the same strong economics of scale equivalencing factor. So I try to avoid using the AHC measure since it is obviously wrong, and in particular underestimates effective incomes (and hence poverty) in larger households which are, most commonly, ones with two adults and children.

 I have been well aware of the problem for decades. The very first equivalence scale I ever used was based on a New York needs list. When I converted it to New Zealand prices, the scale changed markedly because of the different housing prices between the two locations. I have since done a lot of work on equivalence scales (as have others) but the HRC funding agency refused to extend funding when I had been making good progress.[5] Poverty research was not its priority.

The difficulty with the BHC measures are that they do not allow for differences arising from housing costs (e.g. they treat a household with a mortgage the same as one without). The research cited in the previous endnote was exploring this issue.

Endnotes

[1] Because it is an income measure with an expenditure measure subtracted, it is not clear what it means conceptually. Which adds to the economist’s discomfort with it.

[2] It gives, for instance, the same weighting to ‘going without’ as ‘economising’.

]3] by Bryan Perry, published by the Ministry of Social Development, in June 2021 (p.69-70).

[4] The Household Survey omits those living in emergency accommodation – slightly less than 4% of all children.

[5] Carson, S. & B.H. Easton (2002) Household Equivalence Scales           

Easton, B. H. (2004) The Econometrics of Household Equivalence Scales