Like the Case-Shiller Index the Home Price Index (HPI) is another broad measure of the housing market. It is published by the Office of Federal Housing Enterprise Oversight (OFHEO), which was created in 1992. The HPI also utilizes the “repeat sales pricing technique” where it compares the current sales price of a home to its previous sales price to calculate price appreciation/depreciation within a given housing market. This method is used to create a broad measure of the movement of single-family house prices which can be used to analyze market trends and understand the current housing market from a broad historical perspective. And, along with other information about the housing market and U.S. economy, the HPI can be used as a basis for predictions about future market trends.
 
The Many Flavors of HPI
The HPI provides more geographic granularity of data than any other housing market index available today. The HPI is not one but many indices, which include housing data based on the country as a whole, 50 states plus the District of Columbia, and 9 US Market Regions:
New England
Mid Atlantic
South Atlantic
East North Central
West North Central
East South Central
West South Central
Mountain
Pacific

There are an additional 363 individual indices for Metropolitan Statistical Areas (MSA). This is simply a fancy term used by the US Census department for “cities” Further, 11 of these “cities”, such as New York and Boston, are large enough (meaning there are more than 2.5 million people living in them) to be further sub-divided into 29 metropolitan divisions. And there is an index published for each of these 29 divisions.

Thus the HPI is actually a set of over 400 separate indices based on various geographic granularity which can be downloaded for free at OFHEO.

The HPI data is based on mortgage transactions from Fannie Mae and Freddie Mac and it captures house sales or refinancing. Simply put, Fannie Mae and Freddie Mac are “government-sponsored enterprises” which support the market for conventional (meaning that the loan amount of the mortgage is less than $417,000) home loans. Approximately 80% of all mortgages in the US are conventional loans, although not all of the conventional home loans are backed by Fannie Mae and Freddie Mac. Fannie Mae and Freddie Mac account for approximately 50% of all single-family conventional loans in the US. Thus the HPI data is based on data that represents approximately 40% of all single-family mortgages (purchases or refinances) in the country.
 
Like the Case-Shiller Index, the HPI had historically risen roughly in line with inflation - but beginning in about 1998 started to rise much faster (see graph).  However the increase in the HPI has been slightly less dramatic than the increase in the Case-Shiller Index.  This is partly because the Case-Shiller index has more exposure to costal "bubble" markets where mortgages are often too large to be eligible for purchase by Fannie Mae or Freddie Mac, and hence are not covered by the HPI.        

With any data source, it is important to know what is included and what isn't.

Interpreting HPI Data
HPI provides a rich, highly geographically granular view of the US housing market. However, as with any index, the information is subject to interpretation based on the data that is used to create it.

• HPI data does not include home sales where the loan amount is greater than $417,000. In higher-priced areas (such as most of the larger cities in the country) a significant number of transactions won’t be captured due to this limit. Thus the index is likely less representative of the market in markets with an average selling price greater than ~$500,000.
• As with the Case-Shiller index, excluding condos and other multi-family homes can obscure market trends. For example, condos tend to appreciate more slowly than single-family homes. Additionally they tend to see greater price depreciation in a weak housing market. Thus by not including condos, the index may overstate increases in the market and understate decreases.
• Also, like the Case-Shiller index, new homes are not included as there is no previous sales price to compare them to. Thus many high-end new developments are not included in the index. Although they would likely be excluded due to the $417,000 price limit regardless.
• HPI does not include mortgages insured by FHA, VA or other federal agencies. These tend to be mortgages issued to lower income people, or people in very rural areas (i.e. living in the boonies). Thus these mortgages tend to be for less expensive homes. Between excluding these homes as well as condos and other multi-family homes, the HPI data ends up excluding much of the lower-priced segment of the housing market. Thus the index is really a measure of the change in value of moderately priced homes vs. the housing market as a whole.

Additionally, graphs comparing HPI with Case-Shiller indices show a significant amount of divergence. Although these differences can be largely attributed to the fact that these two indices are based on separate data sets and utilize slightly different modeling techniques.

Broadly speaking however, HPI provides yet another tool to better understand the general movement of housing prices for either broad or extremely narrow geographical markets.
 
 

Additional Resources:

HPI Data