From NAR Newsletter Dated August 2010

Interpreting NAR's Existing-Home Sales Data Series

by Wannasiri Chompoopet, Manager, Existing Home Sales, and Danielle Hale, Research Economist

NAR’s existing-home sales data is the premiere measurement of the health of the housing market. The data – EHS – is released both monthly and quarterly, with the quarterly reports reflecting state-by-state resales. It garners a significant amount of media coverage – regardless of whether existing-home sales have increased or decreased. But did you ever wonder how that data is calculated? This brief article gives an overview of the EHS methodology as well as how to interpret the numbers.

Each month, NAR’s Research Division receives data on existing-home sales (single-family, condos and co-ops) from local REALTOR® associations/boards and multiple listing services (MLS) nationwide. The data received captures 40-50 percent of all existing-home sale transactions. The data reflect the total number of closed existing-home sales in each association/board/MLS and also total sales within a variety of home price categories ranging from less than $30,000 to more than $600,000.

Participants in the survey are situated in every region of the country and provide wide geographic coverage of the existing-home market. While almost all reporting associations/boards/MLSs are located in, or adjacent to, metropolitan statistical areas, comparisons of their sales with the American Housing Survey (Census Bureau) show that, as a group, their experience is representative of sales activity and prices that generally prevail in each region of the country.

Monthly statistics are published for the United States and for the four Census regions of the country. State resale volume is based on the entire survey of nearly 700 associations/boards/MLSs and is reported quarterly to ensure each state receives optimal representation.

The monthly EHS economic indicator is based on a representative sample of 200 Boards/MLSs. The home sales data (raw data) is divided into the four census regions: Northeast, South, Midwest and West. The raw sales volume from the participating Boards/MLSs is carefully evaluated by NAR economists to ensure accuracy. In addition, the data is examined to factor out any potential problems – such as changes in REALTOR® association/board/MLS jurisdiction, changes in MLS vendors and/or staff, lack of response by participating associations or MLSs, and erroneous data. Once the “problematic data” have been extricated from the sample, the aggregated raw volume figures are weighted to accurately represent sales activity for each region of the country. This is also called the non-seasonally-adjusted volume. The weights are benchmarked every 10 years to reflect shifts in regional demand. The non-seasonally adjusted volume is then converted into seasonally-adjusted annualized rates (see below).

Similar procedures are used in calculating the quarterly state volume estimates. The quarterly state volume report uses the entire data-set (nearly 700 associations/boards/MLS). Due to the significant differences in sample size for the two reports, the state estimates must be adjusted to conform with regional totals.

Seasonally Adjusted Annualized Rate (SAAR)

It is necessary to “annualize” and seasonally adjust the existing home sales data so that month-to-month and quarter-to-quarter comparisons can be observed without seasonal variances distorting the overall picture. The annual rate for a particular month represents what the total number of sales for a year would be if the relative pace for that month were maintained for 12 consecutive months. Seasonal adjustments, which are determined by using the X-12 Variant created by the Census Bureau, are then used to factor out seasonal variances in resale activity.

For example, home sales in the winter months are generally much slower than sales in the summer months, primarily because of differences in weather. These disparities tell us little about actual trends in the housing market because they are mostly a result of seasonal variances. Therefore, seasonal variances are removed and an annualized rate is calculated to gain a greater perspective on how the resale market is actually performing. Seasonally-adjusted annualized figures are reported for both monthly and quarterly EHS series.

You can see the difference between SAAR figures and those that are “not seasonally adjusted and annualized” in the table “Existing Home Sales”. The seasonally adjusted annual rate of existing-home sales in June of 2010 was 5.37 million units, while the unadjusted figure is 564,000 units.

Seasonal trends alone are also much more apparent. Unadjusted home sales in the core winter months – January and February – are significantly lower across all regions – compared with activity in other months.

Mean vs. Median

In addition to home sales, the EHS series includes figures on average (or mean) and median home prices. Unless you work with data on a regular basis, you may not be 100% clear on the meaning of different averages and why you might prefer one over another. Many economic indicators, such as home price data, are meant to sum up information on a very large population made up of very different individual measurements (i.e., many homes, in many areas, with a wide variety of prices). We summarize this information by expressing data in averages—a measure of the central tendency of the group.

The averages found most often in published economic data are means (sometimes called averages) and medians (also sometimes called an average though less frequently). While there are a few different types of means, we’ll discuss the arithmetic mean. The mean is found by adding up the values of the items we want to average (i.e., a group of home prices) and dividing by the number of items. If you have 5 homes that have recently sold for $100,000; $150,000; $170,000; $190,000, and $250,000, we sum them ($860,000) and divide by 5. Thus, our mean average home price is $172,000.

The median is another measure of the central tendency. We line the homes up by value (smallest to largest), and find the value of the middle item. If we only had four items in our set, our median would be the mean of the second and third items.

Oftentimes, the mean is higher than the median. The mean is generally higher in the case of home prices because some homes can have a VERY high price and that raises the average. The median is probably a better idea of what is “typical”. Consequently, NAR focuses more on the median price of homes. Now that you know the difference, you can use this concept to understand a wide variety of economic variables.

So when you read that “existing home sales registered 5.37 million units in June of 2010” and that the median sales price was $183,700, you can recognize that that many home sales did not actually occur during the month at that price. Also note, that many other measures of housing activity – new home sales, housing starts, housing permits – are also presented in seasonally adjusted annualized rates, means and medians.

For more information about NAR statistics and the methodology behind them, visit www.realtor.org/research.