momentum in housing - uw shown to contribute to stock market momentum are ... on housing msa...

Download Momentum in Housing - UW  shown to contribute to stock market momentum are ... on housing MSA indices ... and contrarian strategies in the stock market,

Post on 24-Mar-2018




1 download

Embed Size (px)


  • Momentum in Residential Real Estate

    Eli Beracha and Hilla Skiba

    Abstract: This paper examines whether there is return momentum in residential real

    estate in the U.S. Case and Shiller (1989) document evidence of positive return

    correlation in four U.S. cities. Similar to Jegadeesh and Titmans (1993) stock market

    momentum paper, we construct long-short zero cost investment portfolios from more

    than 380 metropolitan areas based on their lagged returns. Our results show that

    momentum of returns in the U.S. residential housing is statistically significant and

    economically meaningful during our 1983 to 2008 sample period. On average, zero cost

    investment portfolios that buy past winning housing markets and short sell past losing

    markets earn up to 8.92% annually. Our results are robust to different sub-periods and

    more pronounced in the Northeast and West regions. While zero cost portfolios of

    residential real estate indices is not a tradable strategy, the implications of our results can

    be useful for builders, potential home owners, mortgage originators and traders of real

    estate options.

    Keywords: Momentum, Residential Real Estate, Predictable Returns, Zero Cost Portfolios

    Eli Beracha: Department of Finance, College of Business, Bate 3129, East Carolina University, Greenville, NC

    27858; Email:; Tel: 252-328-5824.

    Hilla Skiba (corresponding author): Department of Economics and Finance 3985, 1000 E. University Ave.,

    University of Wyoming, Laramie, WY 82071; Email:; Tel: 307-766-4199.

  • 1

    1. Introduction

    Since the early days of the finance literature on market efficiency, there has been

    evidence suggesting that buying past winners may generate positive abnormal returns (Levy

    (1967)). Following these findings, many finance practitioners have taken advantage of buying and

    selling stocks based on their momentum or their relative lagged strengths. As a response to a vast

    academic literature on contrarian investment strategies from the 1980s, Jegadeesh and Titman

    (1993) show that zero cost portfolio strategies that buy stocks that performed well and short sell

    stocks that performed poorly in prior periods generate significant positive returns over 3- to 12-

    month holding periods, independent of the systematic risk of the portfolios.

    Momentum in stock returns is related to many factors. Jegadeesh and Titman (2001) find

    that illiquid stocks in the U.S. market generate higher momentum profits. Zhang (2006)

    documents a stronger momentum in stocks for which information asymmetry is higher and

    returns are more volatile. A significant relationship between turnover and momentum is

    documented by Lee and Swaminathan (2000). Ali and Trombley (2006) show that momentum is

    positively related to short sale constraints. In a behavioral based study, Daniel, Hirschleifer, and

    Subrahmanyam (1998) show that investors overconfidence and self-attribution biases are

    positively related to momentum in stocks. Similarly, Chui, Titman and Wei (2008) document a

    positive relationship between cross-country momentum and country specific overconfidence

    proxied by cultural individualism. In their cross-country study, Chui, Titman, and Wei also find

    that in addition to overconfidence, factors that are significantly and positively related to

    momentum include uncertainty and transaction cost.

    The literature on the efficiency of the housing market dates back less than 25 years to

    studies by Hamilton and Schwab (1985) and Linneman (1986). However, many factors that have

    been shown to contribute to stock market momentum are characteristics of the housing markets as

    well. Several real estate papers that study the degree of real estate market efficiency point out

  • 2

    imperfections that are present in the housing markets. Specifically, these characteristics include:

    Information asymmetry due to high information cost, high transaction cost, absence of short

    selling, and infrequent trading (see for example Gau (1984), (1987), Atteberry and Rutherford

    (1993), Fu and Ng (2001)). Case and Shiller (1989) point out that the dominance of individuals

    with consumption rather than investment view in regard to housing and the lack of professional

    traders in the market contribute to its inefficiency. In the absence of large sophisticated investors

    and in the presence of large transactions costs, it is possible for housing prices to deviate from

    fundamentals throughout time. Case and Shiller (1989) are also the first ones to document

    momentum and predictability in housing returns. Using a sample of four U.S. metropolitan areas

    from 1973-1986, they confirm that last periods returns predict future price movements in

    housing prices.

    In this paper we use more than 380 metropolitan statistical areas (MSA) in the U.S. to

    examine whether momentum in residential real estate exists during the 1983 to 2008 sample

    period. Our results show long and statistically significant momentum effect in the U.S. housing

    markets. Using an autoregressive (AR) model we find that area-specific real estate return at time t

    is related to returns earned in previous quarters. Particularly, the models suggest that quarter t-1

    has mostly a slight mean reverting effect on the return real estate market experiences during

    quarter t, while the return during the four quarters spanning t-5 to t-2 positively correlates with

    current return. This means that MSAs experiencing returns above or below the U.S. housing

    average during quarters t-5 through t-2 are likely to earn returns above or below the U.S. average

    during quarter t, respectively. The positive effect of quarters t-5 to t-2 is especially strong in the

    early and late part of our sample. As a robustness check, we also employ a dynamic panel

    generalized method of moments estimation to the sample and its sub-periods, and confirm the

    initial finding.

    In order to gain more insight to the positive correlation between current and lagged

    returns and to the economic significance of momentum in real estate, we employ long-short

  • 3

    portfolio strategies. Similar to Jegadeesh and Titmans (1993) analysis on momentum of stocks,

    we examine the extent to which long-short portfolio strategies on housing MSA indices produce

    positive abnormal returns. Specifically, we construct zero cost portfolios that buy housing indices

    of MSAs that performed better than average in the past quarter(s) and short sell housing indices

    of MSAs that performed worse than average in the past quarter(s). Our results show that zero

    cost portfolios that are based on one to four quarters of MSA housing performance and held for

    one to four quarters before rebalancing, earn up to 8.92% on an annual basis during the 25-year

    sample period. The magnitude of the returns on the zero cost portfolios is especially impressive

    given that a traditional buy-and-hold strategy of the comprehensive U.S. housing index returns

    only 4.69% on an annual basis during the same time period.

    As robustness checks, we test the same long-short portfolio strategy on five separate sub-

    periods and on four broad geographic regions. Overall, we find that the momentum effect is

    robust to different time and region specifications. However, the momentum effect is especially

    pronounced in the West region and during the 2004 to 2008 period.

    Obviously, constructing long-short portfolios of houses is not necessarily a tradable

    strategy in itself. Nevertheless, our results provide timing insight to potential home buyers,

    builders1, and mortgage lenders

    2, as well as to traders of housing derivative contracts that are now

    available on the Chicago Mercantile Exchange (CME). Specifically, the results presented in this

    paper shed light on the magnitude of the momentum effect in the U.S. housing markets. Our

    results show that greater magnitude of momentum is associated with more extreme previous

    periods winners and losers and when momentum is based on a longer period of past


    1 For example, a builder may alter the decision of when and where to build based on area-specific

    momentum information and the projected delivery time of the structure. Similarly, potential home buyers

    may choose to delay their purchase if they have sufficient information that negative momentum exists in

    their area. 2 Mortgage lenders can use momentum information on housing to better estimate the future value of their


  • 4

    The rest of the paper is organized in the following way. Section 2 develops hypotheses

    while reviewing literature on housing and stock market momentum. Section 3 shows the data and

    methodology. Section 4 reviews the main results and Section 5 concludes.

    2. Momentum and Testable Hypotheses

    2.1 Momentum in Stock Prices

    Momentum in stock prices is a well-established phenomenon since seminal work by Levy

    (1967), who documents profitability of buying stocks with relative strength. In a more recent

    paper, Jegadeesh and Titman (1993) show that a portfolio strategy that buys stocks that have

    performed well and shorts stocks that have performed poorly in the past period(s) genera


View more >