A More Encompassing Look at Housing Affordability: REALTORS® Affordability Distribution Curve and Score
February 16, 2017 | Danielle Hale, Director of Housing Statistics
Because housing is a big part of many Americans’ financial wellbeing the ability to access homeownership – housing affordability – is an important topic. The National Association of REALTORS® (NAR) Housing Affordability Index is a great way to simply understand how affordable the housing market is or isn’t for the typical home buyer and how that compares over time. However, it has a few shortcomings. First, not everyone is the typical home buyer. While looking at the median family is a good representation of the middle of the market, the situation for those with incomes above or below the median family income can look quite different. Second, the NAR Housing Affordability Index uses home sale prices to determine affordability, so it’s backward looking. We know what affordability was, but what about the affordability of housing inventory that is active on the market right now?
To answer these questions, the NAR Research Division and realtor.com partnered to do an analysis of affordability at different income percentiles for all active inventory on the market. First, we have a summary measure—the Realtors® Affordability Score—that summarizes what is going on for different income percentiles in a single measure for each state. Like the Housing Affordability Index, the Realtors® Affordability Score is useful for highlighting overall trends in affordability over time. Another result of this research is Realtors® Affordability Distribution Curve. With the distribution curve, we can zoom into certain income segments of the market and see how many houses are affordable to those in the particular income group.
Let’s start with describing a Realtors® Affordability Distribution Curve. First, we gather income data for households in our desired market. Then we construct a maximum affordable house price for the income level using a down payment percentage determined from recently locked mortgages from Optimal Blue. Once we have the maximum affordable house price for a given income percentile, we look at active listings on realtor.com to see what percent of homes on the market are priced less than that maximum affordable house price. In other words, we see what share of homes on the market in that area is affordable to that income percentile. When we graph income percentiles against the share of homes on the market that are affordable to those income percentiles, we get the Affordability Distribution Curve. As seen below, a median income household (50th percentile) can afford roughly 46 percent of houses currently on the market. For households below this, there are fewer affordable homes on the market; a household in the 35th percentile can afford roughly 28 percent of houses on the market. For households above this, there are a greater number of affordable homes on the market; a household in the 75th percentile can afford roughly 74 percent of homes on the market.
After we find the Affordability Distribution Curve, we can create the REALTORS® Affordability Score. The score is simply 2 times the area under the distribution curve. The score will vary between 0 and 2. A score of 0 will result when no household can afford any of the homes that are currently on the market. A score of 2 will result when all households can afford all of the homes that are currently on the market. A coefficient of 1 generally suggests a market close to equality, in other words, homes on the market are affordable to households in proportion to their income distribution. For example, a household in the 20th percentile of income can afford 20 percent of homes currently listed on the market given prevailing purchase terms. However, a coefficient of 1 could also suggest very high affordability for certain income segments and low affordability for other income segments. In general, higher coefficients suggest better affordability conditions for a broader scope of households, but a review of the full distribution curve would be needed to determine what income segments might have the best affordability conditions.
NAR Research and realtor.com computed distributions and scores for January 2016 through January 2017. In January 2017 the US had a Realtors® Affordability Score of 0.92, which generally means that households in many income percentiles can afford a smaller share of houses on the market than their income percentile. Looking at the distribution curve for January 2017, we see that the whole Realtors® Affordability Distribution Curve falls below the Equality Line, but the gap is generally smaller for upper income percentiles. The US Affordability Score decreased from 0.97 in January 2016, due to rising prices across the country and rising mortgage rates that occurred in the last three months of the year. However, North Dakota, Alaska and Wyoming saw increases of 0.03 on average in their scores in 2017.
Midwest states have some of the most accessible housing markets in the US. Indiana come out on top as the most accessible with a ratio of 1.23, meaning that it’s 23% beyond the equality line, and typically lower-income households can afford a greater share of currently listed homes. Out of the 50 states and DC, 28 have scores greater than the US score. And 70% of these states are located in the Midwest and South region. On the other side of the spectrum, Hawaii is the least equal state with a score of 0.52, followed by California, the District of Columbia, and Oregon. In many of these areas, rising home prices have almost priced the lower-income population out of the market.
While looking at active inventory is a valuable innovation of this index, there is also a potential drawback of using listing prices—they do not always relate to final sale prices in the same way. In very hot sales markets, homes can sell for more than the listing price. In slower sales markets, buyers may be able to purchase a home by offering slightly less than the asking price. For example, looking at the data from the Profile of Home Buyers and Sellers we can see that in the past decade the typical final price to asking price ratio ranged from 96 to 99 percent for recent buyers. While this is not a huge deviation over time, it’s something to keep in mind. Further, the REALTORS® Confidence Index shows that in any given month, price discounts can range significantly from property to property. In the most recent month of data, just over 2 percent of properties sold at a price eight percent or more above asking price, while just over 20 percent of properties sold at a similar-sized discount from the asking price.
Since we are using listing prices to determine what is affordable, we are assuming that the housing inventory on the market will all sell for the exact price at which it is listed. If homes are regularly selling above list price, then the market may be slightly more unaffordable than our Affordability Score suggests whereas if homes are regularly selling below the list price, the market may actually be more affordable than our Affordability Score suggests. Also, if certain parts of the market (upper or lower price tiers) are more prone to price discounts or bidding wars, that will also paint a different picture of affordability.
In spite of these limitations, the new Realtors® Affordability Distribution Curve and Score are valuable tools to assess the affordability of different markets to different population income groups on a relative basis and over time.
For more information, view the Realtors® Affordability Distribution Curve and Score data page here >
 Active inventory data comes from realtor.com and will include some but not all new homes as well as existing home inventory.
 This score was inspired by the Gini Coefficient which is used by economists to understand income and wealth distributions and inequality.
 Additional levels of geography may be produced in the future.
 These distributions are modeled somewhat after the Lorenz Curves used to estimate the Gini coefficient, but instead of considering a single variable as the Lorenz Curves do, we are interested in the comparison of income and house prices, namely, what percent of homes for sale are affordable to households at each income percentile.
 We use income distribution data from Nielsen. Nielsen data is provided as numbers of households within income brackets, so we can calculate the percentile within, above, or below any bracket. See detailed methodology here: http://www.tetrad.com/pub/documents/popfactsmeth
 The maximum affordable house price assumes that 30 percent of a purchaser’s income can go to pay for the financing, property tax, homeowner’s insurance costs, and a mortgage insurance premium if the down payment is less than 20 percent. We assume that homes are financed with a 30-year fixed-rate fully-amortizing mortgage at the prevailing mortgage rate. Mortgage rates are those advertised on realtor.com during the period analyzed.
 Because we allow the down payments to vary over time with actual market conditions, they can affect measured affordability in a direction that might not be intuitive. When purchasers are putting down higher (lower) down payments, measured affordability is higher (lower). This is because when the upfront sum of money used in the down payment is higher (lower), there is a lower (higher) remainder of the house purchase price left to finance. Thus with a higher (lower) down payment, the monthly payment is lower (higher).
 Several states have coefficients that equal 1 in the study period. They are Delaware (July 2016); Louisiana (February 2016); Maryland (May 2016, June 2016, December 2016); Minnesota (December 2016); Virginia (January 2017); and Wyoming (June, July, August, September). As can be seen from the respective distribution curves, some of these states are more uniformly close to the line of equality while others have segments with high affordability offset by segments with low affordability.