Follow Insights Blog

CoreLogic

CoreLogic Econ

LATEST CORELOGIC ECON TWEETS

Quantifying the Riskiness of an Individual Property

Bin He    |    Property Valuation

 

A previous CoreLogic Insights blog – Measuring Real Estate Portfolio Risk Relative to the Housing Market– applied the Capital Asset Pricing Model (CAPM) to a real estate market using the CoreLogic Home Price Index (HPI) to approximate a portfolio’s beta. Beta measures the risk of a property relative to the benchmark market, and the higher the beta, the greater the risk. This blog includes additional uses for the ‘Real Estate Beta’ using updated analysis.

The unique value drivers and infrequent transactions for individual residential properties, or in financial terms their idiosyncratic characteristics, present a challenge for applying the CAPM methodology to this asset class. Unlike stocks, it is difficult to calculate periodic returns on a residential property because of the reduced transparency and price discovery in real estate assets. This problem is largely associated with the long holding period and infrequent sales transactions for individual properties. As previously discussed, the value of a property can be estimated using an AVM or hedonic model, which makes it possible to approximate the return and the beta on a property (the price variation of an individual property relative to general market movements for all properties).

Traditionally, beta is calculated using either a Least Squares regression or the covariance of the individual return and the market return, assuming a normal distribution for price changes. Since home price changes do not typically demonstrate a normal distribution, we use a nonparametric approach, called bootstrapping, to determine beta1. CoreLogic analysis shows that, on average, the beta calculated by bootstrapping is greater than the beta calculated by Least Squares, suggesting that a Least Squares calculation would underestimate the risk of a property2.

It is worth noting that beta only measures the systematic, or market, risk of an asset. Under the CAPM framework, risk can be decomposed into systematic risk and unsystematic (or idiosyncratic) risk. Systematic risk measures the uncertainty relative to the entire market and it cannot be reduced through diversification. Unsystematic risk is the uncertainty that comes with a specific property and can be reduced through diversification.

Once we find a more accurate way to calculate beta, we can use it for numerous applications. One of these is constructing a portfolio that tracks national HPI. This is particularly useful for small investors because they don’t usually have the capital to purchase a representative mix of properties across regional markets. To construct such a portfolio, we first calculate the beta for individual properties relative to the return of the national HPI. Next, we construct a portfolio that has a beta of 1 (or close to 1) so when the national HPI increases or decreases by 1 percent, this portfolio can be expected to do the same.

Figure 1 shows a hypothetical portfolio with 160 properties in California, Georgia, Nevada, Texas and Washington, valued at approximately $63 million.

Figure 1: Hypothetical Portfolio

State Number of Properties Value
California 40 $23,329,772
Georgia 40 $10,697,313

Nevada

30 $6,545,781
Texas 20 $5,217,551
Washington 30 $17,517,724
Total 160 $63,308,141

This portfolio reflects a beta of 1.016. Figure 2 compares the performance of this portfolio relative to the national HPI by plotting their respective monthly returns for the fifteen years between 2000 and 2015. As we can see, both portfolios appreciate at similar rates except for the period between 2008 and 2013 (the difference is more obvious in Figure 3). Figure 3 shows the cumulative growth rate for this portfolio and the national HPI since January 2000. Again, they follow each other very well, except between 2008 and 2013. The difference between the performance of the model portfolio and national HPI can be the result of two things. First, it is likely that the limited number of properties and markets in the portfolio (160 and 5, respectively) is insufficient to fully diversify idiosyncratic risk. Second, beta will likely change as a result of improvement or deterioration in the homes or neighborhoods that make up the portfolio. For example, if the schools in a neighborhood improve over time, then property values in this neighborhood could rise on a relative basis to the market by a greater factor than estimated in the original beta measure. As a result, portfolio betas should be updated from time to time to ensure they reflect current market and portfolio conditions3.

In addition to constructing a portfolio that mimics the national HPI movement, beta can be used in many other areas such as risk-based loan origination pricing and portfolio management. Lenders may want to charge a premium for high beta properties because of the associated risk, or if a portfolio manager is a risk-taker, she can add high-beta properties to the portfolio in order to capitalize the market appreciation.

[1] C.C. Frangos, P.J. Arsenos V, K.C. Fragkos, Capital Asset Pricing Model: Nonparametric Estimation of Beta Parameter by the Bootstrap Statistical Methodology, Archives of Econom History, Special Edition, pp. 305-316, 2007
[2] Comparison between bootstrap beta and Least Square beta is available upon request.
[3] Additional examples are available upon request.