*** Spoiler alert: The title of this post is somewhat hyperbolic ***
Demographia’s “9th Annual International Housing Affordability Survey” has just been released and is receiving a lot of attention in various media outlets, such as the NZHerald. Indeed, NZ ‘s connection to the report is relatively strong – it was co-authored by a kiwi and the foreword is written by our very own Minister of Finance.
For those not in the know, the primary objective of the Demographia report is to evaluate housing affordability across a selection of “anglo” countries, namely Australia, New Zealand, Canada, the U.S., the U.K. and Ireland. This is a very admirable objective; after all if policy makers can better understand the complex range of factors affecting housing affordability, then this can in turn support more informed debate and policy settings.
Demographia measure housing affordability using the so-called “median-multiple” indicator, which they define as follows:
Housing affordability = Median house price / Median household income.
This is a pretty simple indicator: Take the median house price and divide by the median household income and, voila, you have a multiple describing the price of housing relative to incomes. Demographia then collect a swathe of data on house prices and incomes for all cities with populations of 1.0 million or more in Australia, New Zealand, Ireland, U.K., Canada, and the United States. Their results over time are shown below.
Since 2004 the trend in the median-multiple measure has diverged between countries; it increased in Australia, New Zealand, and Canada; stayed broadly constant in the U.S.; but declined in the U.K. and Ireland. While these are fairly innocuous results, the Demographia report then concludes (p. 3):
Overwhelming economic evidence indicates that urban containment policies, especially urban growth boundaries raise the price of housing relative to income. This inevitably leads to a reduced standard of living and increases poverty rates, because the unnecessarily higher costs of housing leave households with less discretionary income to spend on other goods and services. The higher costs ripple into rental markets, tightening the budgets of lower income households, who already suffer from lower discretionary incomes. The principal problem is the failure to maintain a “competitive land supply.” Brookings Institution economist Anthony Downs describes the process, noting that more urban growth boundaries can convey monopolistic pricing power on sellers of land if sufficient supply is not available, which, all things being equal, is likely to raise the price of land and housing that is built on it.
I read that and thought “hmm, that’s fairly strong stuff.” So at this point I thought it was worth stepping back a little.
First let’s examine the two key arguments the Demographia report advances in support of the median-multiple measure of housing affordability (p. 6):
- It is used by other reputable organisations, such as the World Bank, the UN, and Harvard; and
- It is simpler than other measures, which are “often not well-understood outside of the financial sector.”
The second reason given is rather vacuous and, frankly, a little condescending to anyone who does not work in the financial sector. And believe me, many economists do not work in the financial sector; at least not anymore.
On the other hand the first reason offered as justification for using the indicator is more understandable: If the median-multiple indicator is used by a range of reputable international organisations then it likely has more merit as measure of global differences in housing affordability.
At that point I tried to follow the sources provided in the Demographia report. Doing raised some fairly important issues: The World Bank link, for example, takes you to a relatively obscure web-page that appears to date from 1992, while the Harvard link appears to be an on-line catalogue of the indicators used in the U.N. report, rather than an independent publication attesting to the merits of the median-multiple indicator.
That leaves us with one “independent” reference, namely the U.N., lending credibility to the use of the median-multiple indicator of housing affordability. Following that link, however, reveals that the median-multiple indicator is but one of a myriad of indicators and checklists identified by the U.N. And perhaps more importantly, the U.N. do not present the median-multiple indicator in isolation, but instead consider it as one of two possible “housing affordability” ratios, as illustrated below.
Righto, so the median-multiple measure can be calculated using either median house prices or house rents. At this stage I was perplexed: Why does the Demographia study not (from what I can tell) mention the ratio of house rents to income as as a possible alternative indicator?
So I did some more digging, and found this graph in the Productivity Commission’s final report on housing affordability in NZ, which considered median rent to household (disposable) income, as illustrated below. In many ways this indicator is more comprehensive than that originally identified in the U.N. report, because it considers after-tax income.
Based on this graph the Productivity Commission concludes (p. 4):
During the house price boom, rents increased at around the same rate as generalised inflation. Across territorial authorities, rents grew in a relatively tight range of 2.3% per year (in Dunedin City) to 8.2% per year (in Buller District). In all cases, rent increases were significantly less than real house price inflation and the ratio of house prices to rents increased markedly, a departure from the long-term broadly stable relationship.
This apparently benign aggregate situation disguises a more difficult position for renters on lower incomes. In particular, people in the lowest two income quintiles spend a much higher proportion of their income on rent than people on higher incomes (Figure 0.5). Even though the situation appears to have improved since the late 1990s, those in the two lower income quintiles still spend, on average, more than 30% of their disposable income on rent, after allowing for government assistance.
Oh dear Daisy: It seems that the Productivity Commission has – using the other housing affordability indicator recommended by the U.N. study referenced by the Demographia report – come to a different conclusion: That housing affordability in NZ has been improving since the late 1990s.
Now at this point I want to caution that these other indicators do not prove that Demographia is necessarily wrong, only that their use of indicators may be too limited. Usefully, the Productivity Commission includes another indicator of housing affordability, namely the “home affordability index” (compiled by Massey University). This index considers the relationship between the costs of servicing a mortgage on the median house and median household income:
This indicator suggests that home affordability has been declining for about the last 4 years. Perhaps more importantly, the 2008 peak in the home affordability index (indicating relatively unaffordable homes) does not seem to be significantly higher than earlier peaks in, for example, 1989 and 1996.
So where does this leave the Demographia report? Well on I’m afraid to say that a first investigation throws up very little corroborating evidence to support their key conclusions, namely that the current price of housing in New Zealand is “unaffordable” relative to historical norms. That’s not to say that they’re wrong, only that their conclusions are not fully supported by the available evidence.
Nor is this to suggest that more affordable housing is not a valid objective: I certainly think it is. And just because the current price of housing is comparable to historical trends, we should still be interested in making housing more affordable, because as Deomgraphia note it is perhaps the most basic of human needs. So while I question Demographia’s analysis and conclusions, the subject is nonetheless very important and worth considering in more detail.
I only wish that 1) they had the time/energy to analyse these issues in more detail within their existing research and 2) news organisations did a little more research before reporting the results of studies like this. While I have more to say on this issue (and hope to do so in future posts), it’s now time for me to up stumps and head for tea (i.e. bed). Until next time …