Academic journal article Real Estate Economics

A Test of a Buying Rule for "Underpriced" Apartment Buildings

Academic journal article Real Estate Economics

A Test of a Buying Rule for "Underpriced" Apartment Buildings

Article excerpt

Much research in recent years has dealt with the issue of whether the market for real estate assets is efficient. One method for testing market efficiency is to develop a trading rule to attempt to earn excess returns in the market under consideration. Linneman (1986) identified an intuitively appealing trading rule for investing in real estate. He examined the single-family housing market in Philadelphia using homeowners' estimates of the value of their homes in 1975 and 1978 (some 1975 values are based on actual purchases). He developed a hedonic regression for the 1975 values and then looked at the properties that had negative residuals when the hedonic equation was used as a predictor of value. If the negative residual was the result of undervaluation of the property by the seller, there might be opportunities to earn excess profits on the undervalued properties, implying inefficiency in the market. He compared the returns on the negative-residual properties with the total sample and found them to be higher but not sufficiently so that they outweighed the transaction costs involved in the trading strategy.

An adjusted version of this methodology is used here to create portfolios of potentially underpriced properties in order to compare their return performance with that of the total sample of properties. Linneman recognizes that using owners' estimates of value, rather than actual transaction data or a third-party appraisal, adds uncertainty to the return measures. And, as he states himself, "Of course, a sample comprised of units which sold twice would be preferred" (Linneman 1986, p.146). This study has the advantage of the availability of actual transaction data for repeat sales of apartment buildings, allowing for alleviation of both of these concerns.

Gau used this same database to assess the weak (1984) and semi-strong (1985) forms of market efficiency, using a sample transaction from each month to estimate changes in the price per square foot of apartment buildings. This paper extends his analysis by calculating actual full appreciation returns for properties which sold more than once, to test Linneman's trading rule.

This paper also uses a technique to adjust the returns for risk - the Sharpe ratio (reward-to-variability ratio) - which has not been applied to real estate transaction data previously. [Brueggeman, Chen and Thibodeau (1984) calculated Sharpe ratios for REIT data as an aside to their main research. They did not calculate the significance of the difference between the ratios.] Since the Sharpe ratio does not require the estimation of [Beta] from the capital asset pricing model (CAPM), it is an effective solution to the problem of adjusting real estate returns for risk. Jobson and Korkie (1981) have developed a test of whether the Sharpe ratios for two portfolios are statistically different. The test is used here to determine whether Linneman's (1986) trading rule can be used to earn excess returns on a portfolio of real estate assets which sold for less than their predicted value. The paper finds that, once portfolio returns are adjusted for risk, returns on "undervalued" properties are not significantly different from those for the overall portfolio of properties.

The Trading Rule and the Sharpe Ratio

Rationale for the Trading Rule

The price a property should command on the market is a function of the value of its physical and neighborhood characteristics. In empirical work in the real estate field, property values are typically estimated using hedonic regressions, of the form

In [P.sub.i,t] = [a.sub.t] + [B.sub.I,t] ln [X.sub.1,i,t] + . . . + [B.sub.n,t] ln [X.sub.n,i,t] + [[Epsilon].sub.i,t] (1)

where [P.sub.i,t] is the price of property i at time t, [X.sub.j,i,t] is the value of the [] characteristic for the ith property at time t, the B's are the relevant regression coefficients, [a.sub.t] is the regression constant and [[Epsilon]. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.