The types of homes sold changed in July, which also lowered the median sale price
House price appreciation fell in July for the first time since December 2018, ending a 40-month streak of growth. But depending on the referenced statistic, this drop could be considered an extreme or minor correction.
One of the most-cited industry measures for home price changes is the median sale price, which determines trends based on the midpoint of all homes sold in a given market. In contrast, repeat sales indices, such as CoreLogic House Price Index (HPI) and the CoreLogic S&P Case-Shiller Indexmeasures appreciation based on the difference between a home’s current price and its previous sale.
Repeat sales stats more accurately measure appreciation, and data from July shows how using the median sale price can skew comments and assumptions about real estate market trends.
Figure 1: Evolution of HPI MOM and median price
Figure 1 shows the month-over-month changes in the HPI and median selling price since 2018, and both measures show that July’s decline was exceptional. Although the HPI drop was much smaller than the median, it was the only such drop in the month of July since 2010. The drop in the median selling price was the largest since January 2016, when it had decreased by 3%.
Still, the large discrepancy between the two measures demonstrates an important point: the slowdown in the housing market is more due to buyers adjusting in the types of homes purchased (or buyers exiting the market entirely) than to the decline in sellers’ prices, because the median sale price does not accurately account for the types of homes being sold.
The broader median sale price lens includes movements in all price levels of homes sold and does not isolate movements within a individual the price of the house. CoreLogic’s HPI uses a repeat sales methodology that avoids this by comparing a single-family home’s most recent sale price with its previous sale price..
Figure 1 illustrates the extreme seasonality of median price changes, with the winter months seeing declines of 1% to 2% each year. The HPI is also subject to seasonal movements but shows only minor declines each year.
The median sale price generally overestimates seasonal trends in real estate markets, as the sales mix changes dramatically during the winter months, with transactions shifting from more expensive suburban homes to smaller urban homes. Although demand decreases during these months, the actual effect of seasonality is small as more sellers may wait for buyers to return in summer instead of cutting prices to sell immediately. December 2021 underscores this point: the median shows that prices have fallen by 1.2%, but the HPI, which again records the price of more (or less) of a house sold compared to its last sale, shows that prices actually increased by 1%.
Going back to the July 2022 numbers, the reason the median sale price has fallen more than the HPI is that the profile of homes being purchased has changed, with a shift to smaller properties in more affordable areas, which reflects shrinking buyer budgets.
Figure 2 shows how much the average square footage of a home purchase was above or below the average square footage of a home purchase compared to the typical purchase since 2018. The sharp drop in recent months is clear. People buy smaller houses because of Rapid rise in mortgage rates in 2022, causing the average monthly payment to skyrocket. In July, the average home purchased was 30 square feet less than what typically sold in July in the same county.
Figure 2: Average square footage of homes purchased and deviation from monthly average
Figure 3 shows the deviation from monthly averages for buy shares by price level. Sales are calculated by zip code, with the most expensive 25% in a given Metropolitan Statistical Area (MSA) classified as “4” and the lowest 25% classified as “1”. It’s clear from this data that sales are transitioning to more affordable parts of the country. The most expensive areas have particularly fallen, dropping from 20% in mid-2020 to 18% today.
Figure 3: Share of purchases by postcode price level (seasonally adjusted)
That’s not to say home price declines aren’t on the horizon, and they’ve already arrived in many parts of the United States. Seattle; and San Jose, California. The CoreLogic HPI Market Risk Indicator predicts home prices in Boise have a very high probability of a decline of 10% or more over the next 12 months. It’s important to remember that when the market changes rapidly, the composition of sales also changes, leading to stats like median selling price often missing the mark. Indeed, median prices in Boise, Seattle and San Jose all fell more than 5% in July, suggesting the price drops were about double what they actually were.
 A simple example illustrates this concept. Assume that in June, 55% of home sales occurred in a state where all homes cost $300,000 and 45% in a state where all homes cost $600,000. The median returns a value of $300,000. Then in July, the prices are unchanged but the ratios are reversed so that 45% of sales are in the first state and 55% in the second state. The median is now $600,000, suggesting prices have doubled. A repeat sales index, however, would only compare homes to their last sale and would not correctly record any appreciation.
 Specifically, the chart shows how much/under square footage purchases compared to the average purchase in the same county and month from 2015 to 2022.