Property news usually goes one of two ways, up or down. But with so many unfulfilled prophecies of bubbles on the Internet, you have to wonder where they’re getting their information from – especially if you want to invest. The first thing journalists look at are house price indices (HPI), not debt, vacancy, rental or construction rates,
In essence, HPIs are online databases of single-family house prices that monitor national and local variations in home values. The principal is simple: information on the value of each property transaction is collected over certain period of time (e.g., one/three months or a year) and then crunched into an average or median price that buyers and industry experts can use to get an idea of the market.
These three ominous words are meant to spell out the “real” state of the real estate industry, which unfortunately, is far more complex than that. Basing a decision on a HPI is about as safe as buying a house at auction blindfolded depending what other people are bidding. In fact, Mark Twain was probably looking at a house price index when he uttered these famous words: “There are three kinds of lies. Lies, damned lies and statistics.”
HPI methodology, be it an official source like the Federal Housing Finance Agency (FHFA) or industry players like National Association of Realtors (NAR) or Zillow, is rife with complacency. So much so that is even uncharacteristic of statisticians, who publish thick handbooks on their method that read like an apology in disguise. They can get so absurd that I’m actually starting to enjoy reading them, especially when they write things like this:
“No data on the housing characteristics is required to calculate the median [house price]… The set of houses traded in a period is typically small and not necessarily representative of the total stock of housing.”
Eurostat, Handbook on Residential Property Prices Indices (RPPIs)
There are many reasons to get rattled by HPIs. For example, just last year property portals around the world were celebrating Spain’s long-awaited real estate recovery. However, in my profession, it doesn’t pay to jump on the bandwagon too soon so I did some research and discovered a darker truth: Spain’s real estate recovery hinges on five local markets that absorbed 70% of all transactions – not that anyone else reported on that angle. Without further ado, here are some of the big (and many) problems with house price indices.
The Repeat Sales method
This method, used by the FHFA, only includes houses that have been sold twice or more. Obviously, there are some gaping holes in a system that doesn’t take into account new home sales, especially in markets where there’s a lot of building going on. Luckily for them, no one ever reads the HPI manual except for some academics like Nagraja, Brown and Watcher from Fordham University and the University of Pennsylvania who are also sceptical. Here’s what they say:
Claim up to $26,000 per W2 Employee
- Billions of dollars in funding available
- Funds are available to U.S. Businesses NOW
- This is not a loan. These tax credits do not need to be repaid
“Previous research has shown that repeat sales homes are fundamentally different from single sales; in light of this work, it is difficult to argue that such indices can truly represent the housing market.”
“The indices are computed from a small subset of all home sales. Consequently, they may be unrepresentative of the housing market as a whole. We find that in our data, the sample size is reduced significantly if only repeat sales homes are included: between 33% (Columbia, SC) and 64% (San Francisco, CA) of the data are single sales homes.”
Repeat Sales House Price Index Methodology
Averages, medians and adjectives
“One major drawback of simple median based indices is that they provide very noisy estimates of price change.”
Eurostat, Handbook on Residential Property Prices Indices (RPPIs)
Persons tasked with calculating the index use averages and medians to determine a market-wide estimate. The (arithmetic) average simply adds all the house prices together, then divides the total by the number of houses. It’s not used because it can be deformed a few excessively high or low sale prices. A weighted average, as used by the FHFA, assigns more value to certain segments in an attempt to balance the results according to dominant price categories. Another option, the median, finds the central point above and below which 50% of the houses were sold respectively. This can also be skewed by excess around the edges of price distribution, but is still the preferred method for Trulia, Zillow and NAR.
Some indices also use the geometric average (like the British Land Registry) or the mode (the value that occurs most often). If you use all these methods on the same data set, you will come up with surprisingly inconsistent results, as demonstrated by the experts at Knight Frank, a global property giant:
“Average” house price in London (UK) by method, 2015 | |||
Arithmetic average | Geometric average | Median | Mode |
£530,239 | £416,302 | £385,000 | £250,000 |
Source: Knight Frank | |||
Asking prices, sale prices and cognitive bias
While most HPIs use sale prices to calculate market trends, it’s not unusual to find some that only reference asking prices, especially if it’s an online property platform like Trulia, which provides both. There are a few practical indicators like market conditions and competition that can cause these two reference points to diverge, making asking prices ill-fitted to evaluate the market. At the same time, we are victims of a cognitive bias that leads us to naturally overestimate the value of our belongings including the home we are putting on the market. So never take it for granted that a HPI is based on sale prices, make sure it is.
No house was created equal
No one has really figured out a way to factor in property characteristics like the size, number of rooms, etc. Say you want to buy a home in Austin (TX), Zillow will tell you that the median home value is $247,500 as of March 2016, which is 8.7% higher than last year. But the average American town has houses and condos of different sizes and shapes: one house has a private entrance, a swimming pool, three bedrooms and bathrooms – the other has one bathroom, a bedroom and a tiny garden right beside the highway. So realistically, an average or median price doesn’t mean much here. Furthermore, this value could focus on a densely developed area like New York where there are more condos than houses. In this case, if there are more condo sales, the value will reflect this and won’t mean much for people trying to buy an actual brownstone, will it?
Non-disclosure states
There are 12 non-disclosure states in the US, the laws of which basically prohibit the collection of transaction sale prices or public access to them. As a result, any company compiling house price estimates for the region have to find alternative sources of data like real estate agencies, which are going to be partial because no company has the monopoly of the market.
Non-disclosure states in the US | |
Alaska | Montana |
Idaho | New Mexico |
Kansas | North Dakota |
Louisiana | Texas |
Mississipi | Utah |
Missouri (some counties) | Wyoming |
Source: zillow | |
Should we just ditch HPIs?
The answer is “no”, even if they are very approximate measures of the market, because they are good indicators of housing trends. They reflect the state of demand in that area, but also how prices evolve, allowing us to chart the current stage of the real estate cycle. Knowing if it’s growing, stagnating or falling is fundamental to making a good investment and planning your exit strategy whether you are in it for the long haul or just a quick buck. There’s no easy way to overcome the methodological bias of HPIs but knowledge is power, so the more sources you can find and cross reference, the more relevant your conclusions will be.
