RIP timely mortgage approval statistics

The Reserve Bank has recently confirmed that it is going to kill one of the most useful statistical series it produces.

A decade or so ago, near the peak of the last house price boom, the Bank began collecting data on the number and value of housing mortgage loan approvals.  The data were never perfect, and the Bank’s statisticians always, slightly disdainfully, labelled them as “experimental”.

But they had a lot going for them nonetheless:

  • they were weekly data, whereas most Bank series are monthly or quarterly,
  • they were reported and released very quickly (data for last week is on the Bank’s website now), whereas most Bank series take the best part of a full month to collect and report,
  • they were about mortgage approvals, rather than drawdowns, so captured information at a materially earlier stage of the process (when one is close to agreeing to buy, rather than when settlement occurs), and
  • they told sensible stories that could be reconciled, then and later, with other things we knew about the housing and housing finance market.

Given that the Bank now puts much more weight on direct regulatory interventions in the housing finance market than it ever did before, you would think they would want all the data they could get –  and particularly very timely data.

But no.  The housing mortgage approvals series is to be discontinued next month.

The Bank would no doubt respond that they have, over the last few years, put in place a new set of data collections around housing finance.  They even have a new series of housing mortgage commitments, but (a) they have only three years of (monthly) data, and (b) the monthly data are available only with a considerable lag.    Check out the tables for the new series: today is 13 October, and the data for August are now there (and mid-August is now almost two months ago).  By contrast, as I noted above, last week’s new mortgage approvals are already on the website for the long-running “experimental” series.    It is significant step backwards, in terms of the public availability of timely data on what is, for the Bank, and rightly or wrongly, clearly a major market.

I made a brief submission on the proposal to discontinue this data, noting

The cumulative loss of information from the change could easily be 8 to 10 weeks of information (if, say, a mortgage approval is typically given perhaps a month before drawdown).  Even allowing for the fact that the mortgage approvals data is not a perfect predictor of actual drawdowns, the cycles in the approvals data have given good and consistently informative reads for a number of years now.  With the Bank varying LVR restrictions on average about once a year, losing 8 to 10 weeks of forward data seems cavalier –  even having regard to the inevitable compliance costs for banks (which must now be quite low for an established collection).

To which the Bank’s response last week was

We acknowledge that the housing approvals statistics provide more frequent and timely data, however, we do not believe that this a sufficient reason to continue with the collection.

I don’t think their heavy regulatory interventions, and repeated recalibrations of the restrictions, in the housing finance market are well-warranted –  in law or in economics.  But to intervene that heavily and then to simply abandon frequent timely data seems reckless.   Better more comprehensive later data can refine the insights from early takes –  as final GDP data several years on are better than the first cut estimates, but we don’t simply abandon publishing the first cuts – but in much of the business the Bank is in –  monetary policy and regulatory interventions –  preliminary insights are vital, and inform (heavily inadequately) the forecasts, and policy responses.  That is why, for example, the Bank does business visits, uses opinion surveys, and so on

Inside and outside the Bank, I’ve always found the mortgage approvals data useful and interesting.   Because it is weekly data, and not seasonally adjusted, I’ve taken to presenting the mortgage approvals numbers in a chart like this:

mortgage approvals oct 2016.png

The vertical access is the number (not value) of mortgage approvals per capita (using mid year population estimates).  The horizontal access is the week of the year.

There are data back to 2004.  The blue line is the average for the years 2004 to 2013, the decade prior to the use of LVR controls, and which encompasses both most of the last boom and the post-recession “bust”.   As you would expect, there is some seasonality apparent in mortgage approvals, but mostly around the Christmas/New Year period.

I’ve also shown the data for 2006 –  the year during the last boom when mortgage approvals per capita peaked –  and for 2015 and 2016.

There have been weaker years –  2010 and 2011 notably –  but the volume of mortgage approvals this year has simply not been very high.  It has averaged around the same as last year’s numbers –  but had looked as though it might be moving a bit higher before the Bank announced its latest set of LVR controls.  One of the great things about this high frequency timely data is that one can see approvals dropping off somewhat in the last couple of months, as one would expect given the new controls.

Taking a longer perspective, relative to the 2006 peak year, mortgage approvals per capita this year (and last) are only about two-thirds now of what they were a decade ago.   This is consistent with house sales per capita data, which are also running well below the peaks in the previous boom.

None of this is to suggest that high and rising house prices are not a problem. They are, and they are a disgraceful reflection on politicians and policymakers.  But it does help to illustrate that the issue is not primarily banks flooding the market with credit, or even a mania of buyers desperate for anything that moves.  The dollar value of credit is growing no faster (probably slower) than one would expect given the increases in prices –  higher-priced houses require more credit –  and the volume of mortgage approvals is actually quite low.

High and rapidly rising house prices are mostly a reflection of the severely distorted market in urban land –  distorted, that is, by central and local government –  exacerbated by policies that deliberately set out to rapidly boost New Zealand’s population, even when it is well-known that the housing and urban land supply markets simply can’t cope –  or at least not without seeing prices going sky-high, the market response to artificial, regulatorily-induced, shortages.

Direct interventions in the housing finance market are an inappropriate use of (too widely drawn) Reserve Bank powers.  But with their readiness to deploy those powers, and chop and change the rules, frequently, it is most unfortunate that they are dropping the most timely and frequent data on housing finance we have.