There are a seemingly endless number of websites, blogs, and social media posts that tout how much some measure of the ratio of "average" home prices to "average" incomes has risen over the last few decades. While home price affordability is a real problem, I sense that the ratios reflect growing income inequality (not just rising home prices), and that that message gets muddied when presenting information as averages of averages, particularly when median is used instead of average.

As a hypothetical illustration (with inputs created to highlight my points), the table below shows a two-sector society (or state or areas of the country) with two income groups. Members of Type 1, the majority of the population, all have income of $60,000 and for those in the group that buy houses, pay 4x income or $240,000. The Type 2 population all have incomes of $150,000 and those who buy houses pay 6x income or $900,000. (Note, the notion that those with higher incomes are willing to pay a higher multiple for homes is discussion for another blog, but well supported in home price research. For example, those with higher incomes have other assets that can be used to pay for a house, so they are not constrained by income, and those in higher tax brackets benefit more having rent "prepaid" in owning a home, rather than paying rent with more heavily taxed income. Again, more to follow.)

Note that since the majority of the population is composed of Type 1, the median income of the overall population is that of the Type 1 population, or $60,000. On the other hand (in this example), the Type population has a higher tendency to own homes (85% v 50% for Type 1's) and therefore comprises the majority of homeowners (34% v 30% of the general population) for an overall homeownership rate of 64%. As such, the median home price is $900,000

A comparison of median home price to median income (far right column below) would result in a ratio of 15.0, even though no individual spent 15x income for a home. Net, be very wary of reports showing median vs median as the groups measured by income and those measured by home price (i.e. the overall population's income versus a subset of the population that buys homes) is not the same.

Similarly, even if average home price is compared to median income (column in yellow) the result of 6.3 is higher than the ratio of even Type 2's, as the population buying homes does not overlap 1:1 with median income. Expanding on that, a Type 1 resident switching from rental to homeowner status will (per this illustration) pay 4x income, not 6.3, as the lower-income home buyer is not looking to buy the median but more likely a starter home at a 4x multiple.

On the other hand, overall average home price vs average income generates a ratio of 3.9, below that of even the Type 1 population.

Net, for this example, averages need to be compared within groups/Types or weighted in a way that in not customarily presented.

To further drive home the point, I've assumed in Period 2 (so a few years later) growing income inequality (such as measured by many for 1990-today) with the Type 1 population seeing a 10% income gain, while the Type 2 population has a 30% income gain. I've left both groups tendency to pay some multiple of income the same (even though the higher income group might be in even higher tax brackets and have an increased incentive to "pre-pay" rent by buying a home). Since the majority of the population is Type 1, median incomes barely rise, while, since the majority of homeowners are Type 2, the median, and average home price rise dramatically.

Despite the fact that each group stills pays the same ratio for homes (within their group), the average home price divided by median income of the overall population has risen from 6.3 to 7.2 (a gain of 15%), beyond the ratio that the Type 2 population, even though no individual had to increase their home price to income ratio!

Net, growth in income inequality, has the potential to raise certain home price to income ratios, even if each underlying population doesn't change it's home price to income preferences.

Be careful of averages of averages!

As an additional practical matter, this hypothetical illustration assumed two separate population groups. A challenge for Type 1's is when the Type 2's can cross boundaries and move to their areas (think Boise, Nashville, Phoenix, much of Texas during Covid). The flood of higher income "immigrants" (at least as viewed by the earlier Type 1 residents) who can sell their high-priced homes from Type 2 land (e.g. California, New York) to buy more house in Type 1 areas, combined with the willingness of higher income Type 2's to pay a higher home/income ratio, had the tendency to price-out the prior Type 1 residents. (Follow AEI for more on the California arbitrage). Combine this with the Type 2 desire to move immediately, while new construction takes time, and you have the recipe for both a pop in home prices in the "invaded" Type 1 areas, and friction between those who had lived there before (when houses cleared at 4x income) as they realize that they now can't afford to buy a home in their new higher-priced region with their Type 1 income.

The production of new supply (to include rentals) over a 2-3 year planning and construction phase, as well as a reduced Covid impetus to move, has caused a leveling off, and even drop in home price in areas that saw massive inflows (e.g. Phoenix, Austin).

While the short-term impacts of population moves from Covid may have slowed, the notion of growing income inequality still exists, so be cautious when reading headlines comparing averages on averages. While policies need to be addressed to make more housing available (particularly where people are moving) the core issue of income inequality can't be forgotten.

While this blog differs from my typical work on how one can hedge, I think that my reaction to averages of averages should contribute to discussions about home prices and affordability.

Thanks, John