Follow Insights Blog

CoreLogic

CoreLogic Econ

LATEST CORELOGIC ECON TWEETS

CoreLogic Home Price Index Can Be Used to Mark Down Distressed Properties

Use discount factors to improve your portfolio mark-to-market results

Bin He    |    Property Valuation

The industry often uses the CoreLogic Home Price Index (HPI)TM to estimate fair market values between sales transactions.  While this methodology can deliver reasonable valuation estimates for standard properties and transactions, distressed properties, such as real estate owned (REO) and short sales, typically sell at a discount to this estimated value.  CoreLogic has found that complementing the traditional HPI value estimation technique with regionally specific property discount factors can produce more robust valuation estimates for distressed properties.  The methodology is explained below.

Traditional HPI valuation methodologies typically estimate property values between sales transactions by combining the most recent sales price by a regional HPI growth rate.  This value is derived from the previous sale to the time of the value estimate by using Equation 1.

In Equation 1,  is the previous sales price,  is the HPI at the time , is the HPI at the time , and  is the estimated fair market value at time . Following the actual sale of the property, it is possible to assess the strength of the value estimate,  by comparing it to its actual sales price.

As noted above, this methodology typically delivers value estimates for distressed properties that differ significantly from the actual sales price. This is not surprising since HPI only measures average market price movements rather than property-specific factors that are often present in distressed properties.

The equation that estimates a geometric repeat-sale HPI is expressed in Equation 2.

In Equation 2, D is 1 at the second sale, -1 at the first sale and 0 in any other instance. Coefficient B can be used to calculate the HPI growth. Since distressed sales are generally sold at a discount to the market average, we need something in Equation 2 to account for this discount. Simply adding a dummy variable can compensate for the discount on a distressed sale1, 2 while still giving us an unbiased HPI estimator.  Equation 2 then becomes Equation 3.

In Equation 3, the dummy variable REO is 1 if the second sale is a distressed sale, -1 if the first sale is distressed, and 0 in any other instance. Coefficient X is the average effect of the discount on distressed properties at different time periods. Equation 3 indicates there is another layer of growth that comes from distressed transactions in addition to the market average growth. If the first sale is distressed but the second sale is a regular transaction, then the growth rate for this pair will be the market average growth rate plus some positive adjustment in order to compensate for the discounted price that occurred at the time of the first sale. If the first sale is a regular sale but the second sale is distressed, then the growth for this pair will be the market average growth minus some effect in order to discount the price at the time of the second sale.  Figure 1 illustrates how this works.

Discount Factor for SF County

Discount Factor for SF County

For pair 1, the first sale is a distressed transaction, and the growth that comes from the average market is 1.5, which is the HPI at t2 divided by the HPI at t13. However, since the first sale is a discounted price, using the market growth alone to value the property at t2 would underestimate the sale price at time t2. Similarly, if the second sale is distressed, the total growth for pair 2 is less than the market average growth. Using an HPI in conjunction with the discount factor can improve the valuation of distressed properties significantly.

Figure 2 shows the estimated discount factors between 2008 and 2012 for San Francisco County using the above methodology.

Illustration of applying discount

Illustration of applying discount

The discount factor is used together with the CoreLogic HPI – excluding distressed sales – to mark-to-market pairs when one of the transactions is a distressed sale. Figure 3 shows the percent prediction error (PPE)4 and median absolute error for the results that do and do not account for the discount factor. Overall, we see significant improvement in the mark-to-market results.

1Andrew Leventis, Removing Appraisal Bias from a Repeat-Transactions House Price Index: A Basic Approach, OEHEO Working Paper No. 06-1.

2William Doerner, Andrew Leventis, Distressed Sales and the FHFA House Price Index, Journal of Housing Research, 2015, Vol. 24, No. 2, pp. 127-146

3While REO sales were included in the estimation of the CoreLogic HPI, by including the REO variable in Equation 3 the constructed index is comparable to an HPI estimated by removing matched pairs that have a distressed transaction as one sale in the analytic dataset, if the adjusted matched pairs yield a similar growth rate as the market average that does not include those matched distressed pairs.

4PPE measures the accuracy of the estimated value. The higher the PPE number, the better. PPE10 measures the percent of model results that have an error within 10 percent, PPE20 is the percent of model results that have an error within 20 percent, and PPE30 is the percent of model results that have an error within 30 percent.  

© 2016 CoreLogic, Inc. All rights reserved.