Since home price movement is such a large component of residential mortgage asset valuation, mortgage market traders and investors can profit from a thorough understanding of the most commonly published real estate indices.  How are home price indices created?  How does each index computational method differ?  What is the sample size of the underlying real estate transactions to validate and support the specific real estate index values?  How do the differences in the underlying indices impact the published results of the indices?  And finally, how does a mortgage valuation expert utilize these various real estate indices in their mortgage loan valuation?

      The most prominent indices followed are the OFHEO (www.ofheo.gov), S&P Case-Shiller™ (www2.standardandpoors.com) and LoanPerformance HPI™ (www.loanperformance.com).  Although they all have the same basic foundation of identifying and utilizing sales pair data, there are a few major differences to point out and remember when analyzing the index results.  The most important distinction between the indices is that the OFHEO uses only loans purchased by Fannie Mae or Freddie Mac while the S&P Case-Shiller uses transaction data taken from municipal recording offices throughout the country.  The LoanPerformance HPI data is obtained from their large real estate database and has similar diversification to the S&P Case Shiller data.  The effect of the distinction in the data sources is that the OFHEO is calculated without consideration to non-conforming balance loans as well as any conforming balance loans funded under non-agency guidelines.  Another difference is that the S&P Case-Shiller factor the actual home sale price into their weighting calculation as opposed to just the change in price.  This causes the change in more expensive homes to impact the index greater than less expensive homes.  OFHEO’s index weights price trends equally for all properties and the LoanPerformance HPI creates sub-indices to breakout different home price levels around the area average.

      To calculate the indices, data are collected on transactions of all residential properties during the months in question which will be either a quarterly or monthly timeframe.  The main variable used for index calculation is the price change between two arms-length sales of the same single-family home.  For each home sale transaction, a search is conducted to find information regarding any previous sale for the same home.  If an earlier transaction is found, the two transactions are paired and are considered a “repeat sale.”  Sales pairs are designed to yield the price change for the same house, while holding the quality and size of each house constant.  It is important to hold property type consistent because the factors that determine the demand, supply, and value of housing are not the same across different property types.

      The indices are designed to reflect the average change in all home prices in a particular geographic market, however, individual home prices are used in these calculations and can fluctuate for a number of reasons.  In many of these cases, the change in value of the individual home does not reflect a change in the housing market of that area; it only reflects a change in that individual home.  To address these concerns the indices use a methodology known as weighting sales pairs.  The outlying transactions for which price points are given less weight are homes that show a high turnover frequency, homes with a longer than average interval between sales and homes with price points that are inconsistent with the surrounding area.

            With regards to geographical coverage, there are two standard measures – breadth and depth of the data.  With regards to breadth, OFHEO Indices are created separately for each of the nine U.S. Census Divisions, each of the 50 states and the District of Columbia, every MSA as recognized by the Office of Management and Budget, and one for the total U.S.  The S&P Case Shiller produces indices individually for 20 selected MSAs, a “Composite 10” which aggregates 10 broad MSAs, a “Composite 20” which aggregates all of the 20 MSAs that they track individually and a national version which is made up of the nine U.S. Census Divisions.  They also recently started breaking out 17 of their 20 MSAs by pricing tiers of low, medium and high for added granularity.  The LoanPerformance HPI offers a U.S. national, four U.S. splits for west, mid-west, northeast and south, the nine U.S. Census Divisions, state level, 150 MSAs, and further granularity of various counties and zip codes.  The OFHEO indices will utilize loan purpose type data from “all-transactions” but they do offer “purchase only” versions of the total U.S. and nine Census Divisions indices.  The S&P Case Shiller indices are only provided using “purchase only” data.  The S&P Case Shiller and LoanPerformance HPI indices are updated monthly with the exception of the S&P Case Shiller U.S. National which is updated quarterly.  The OFHEO indices have traditionally been updated quarterly but beginning in March 2008, the Census Divisions and U.S. National indexes will be updated and released each month.

            With regards to depth of the data, the OFHEO is typically thought to have the largest data set relative to its competitors, in reality, it depends on the index type.  OFHEO’s purchase only sales pair data is only about 15% of their aggregate pool which includes refinance transactions.  It is fair to say that the OFHEO “All-Transactions” index has substantially more depth to its data set than the competition, however, its depth of purchase only sales pairs is quite comparable to that of S&P Case Shiller.  The count of purchase only, sales pair transactions that have recently contributed to calculating the indices is roughly 75,000 for the monthly index.  Historical data of refinance transactions has shown a bias towards higher home price valuation.  The point is clear that when an appraiser is determining a home value without the influence of the price set between a buyer and seller, the home value tends to be higher.  Therefore, I believe a comparison of purchase only indices is most realistic and each index creator has comparable data depth in this space.

            Lets compare index results on purchase only transactions between the third and fourth quarter of 2007 and the annual change from 2006 Q4 and 2007 Q4.  The OFHEO report for the total U.S. shows that the index was down 1.3% from the prior quarter and down 0.3% for the year between 2006 Q4 and 2007 Q4.  The S&P Case Shiller reports that their national index was down 5.4% and 8.9% for the quarterly and annual change respectively.  The LoanPerformance HPI results are closer in line with S&P Case Shiller.  The fact that liquidity in the mortgage market has drastically migrated from non-agency to agency underwriting should explain a large part of the difference between the OFHEO and S&P Case Shiller results.  Clearly recent rate sheets would show mortgage financing is better priced and more easily obtainable in the agency space which will provide more bid support to those properties underlying loans in the OFHEO index over the portion of non-agency loans in the S&P Case Shiller.

            The graph below shows the OFHEO and S&P Case Shiller U.S. Indices dating back to 2000 1Q.  Each index has been given a starting value of 100 at 2000 1Q to better track the index differences from that point forward.  The S&P Case Shiller shows an approximate 90% increase before peaking in mid-2006 while the OFHEO “purchase-only” index peaks at an approximate 65% in mid-2007.  When following this historical data along with the recent quarterly and annual changes discussed above, it is important to identify the trend showing the S&P Case Shiller to produce more erratic results during the housing boom and bust.  The OFHEO’s more muted results should largely be related to the exclusion of the non-agency programs that fueled housing speculation and additionally less magnified by not using a loan size weighting method similar to that used by S&P Case Shiller.  A less impactful but still important reason is the fact that OFHEO utilizes more granularities in their indices by tracking data for every state and over 300 MSAs.  This should have the effect of smoothing out the peak to trough changes when incorporating data from areas that experienced less robust speculation that where outside of major cities and MSAs which are largely the data set used by the S&P Case Shiller.

      When modeling mortgage credit cashflows the severities of loss applied to loan foreclosure frequencies should reflect the exposure depictive of loan to value ratios adjusted for current home value.  Utilizing these indices are more cost and time efficient than obtaining updated appraisals on a loan level basis.  The process of applying index results where a current home price is revised based on the observed index price movement for a corresponding time period is a relatively easy process, the difficulty is in deciding which index to use.  Although there are major differences between the indices, each offer distinct, beneficial attributes which can be considered.  One analyst could view the best attribute of the S&P Case Shiller to be the pronounced movements of volatile MSAs such as New York and apply the more drastic pricing movement as further protection when modeling a currently delinquent, large balance Alt-A loan.  For instance, the following graph compares the change in the New York Area as reported by OFHEO and S&P Case Shiller since 2005.  You can see that S&P Case Shiller is reporting a greater level of depreciation than OFHEO.  Given that a substantial portion of loans in New York will be above the conforming loan limit, it makes sense to model losses with the protection based on adjusted LTVs as reported by S&P Case Shiller.

            Another smart trader who wants to show a strong bid for a premier credit pool with geographical concentration in a less volatile MSA could use the OFHEO index because the S&P Case Shiller might not track the individual MSA and modeled data from a broader MSA could produce a bid depictive of higher risk collateral.  Lastly, a prepayment analyst should utilize adjusted LTV information to impact rates of prepayment based on a borrower’s ability to refinance.  The higher LTVs will have fewer refinance options than the lower LTVs.

      When looking at recent index results the most alarming differences are in the dispersion of declining values among the volatile MSAs.  If you look at the following graph you can see that Phoenix and Tampa are examples of locations that were slower to trend upward but once did saw a much more dramatic “bubble” than New York and Chicago.  Phoenix experienced a housing peak of almost 60% greater than Chicago and looks to have a rate of decline that will surpass Chicago in six months.  The red flag here is the inherent risk in adjusted loan to value ratios on loans in the Phoenix area.  In contrast, although New York reached a peak that is comparable to Phoenix and Tampa, New York is not seeing as rapid a rate of decline and therefore has less inherent adjusted LTV risk.