Hot takes and inequality data

Me at The NBR on some ... oddities ... in the latest HES wealth data. A snippet (ungated):

The HES tables sort households into quintiles by net worth. The first has the 20% of households with the lowest household net worth; the fifth quintile has the top 20%. The tables list the assets and liabilities held by households in each quintile. 

Households with less than $39,500 in net assets make up the bottom quintile. Collectively, those households own owner-occupied houses worth $3.99 billion, or $219,000 for the median household that owns the house they live in. The figure feels a bit low: are there really that many houses out there that are worth less than $220,000? But part-ownership of those houses could help make sense of it if parents putting down the deposit on the kids’ house gives them a stake in the ownership.

More puzzling was that, against that $3.99 billion in housing assets owned by the least wealthy 20%, were $4.899 billion in loans against owner-occupied residences. The last time I checked, banks needed to comply with 80% LVR ratios, not 123% ones. But even if there were no LVR considerations, what bank would lend multiples of the value of a house to the people in the least wealthy cohort? And so it was time to check the footnotes. 

The notes tell us that assets held in a business or trust are not counted, unless they are mentioned. 

It seems implausible that the least wealthy 20% of the population have put their houses into trusts but it could be a consideration in some cases. 

If a family owns houses in a trust, and the 20-something living in it has just graduated from university and has taken over the mortgage, the mortgage liability could show up in the accounts but not the asset. But that story seems to fall apart when we look back to the median tables rather than the totals. Among those first quartile households reporting ownership of a house, the median house value is $219,000. But the median debt on owner-occupied housing, among those with mortgages in first quartile households is $240,000. 

It does not make sense that the debt on the median house owned with debt is greater than the value of the median owned house in that quartile, unless the median home recently purchased by those in the bottom quartile is of substantially better quality than those that have been long owned by those in the bottom quartile. More plausible are problems caused by out-of-date valuations. Statistics New Zealand tells us that the valuations used in the HES are from government valuations which can be up to three years old. GVs are well behind actual house prices. 

If you have just purchased a house, the liability ledger will accurately reflect the value of your mortgage but your house could be undervalued by 20-30% – or more if you bought in the right place in Auckland. 

These measurement issues then mean that wealth held by the bottom quartile is probably strongly understated – as are the housing assets held by all other quartiles. It matters a lot more in the bottom quartile – it winds up having about a billion dollars more in total household liabilities than it has in total household assets. Counting education loans on the liabilities side of the ledger while not counting the value of the human capital it embodies also makes for problems. These loans loom large on the liabilities side for households in the first quartile but are mostly held by younger people with strong future earnings potential and good future upward mobility – not the cohort typically worried about in discussions of inequality.