Distress Risk and Stock Returns on Equity REITs

Delisting is defined as a removal of a listed firm from stock exchange. Default occurs when a firm is not able to meet obligations in interest or principal payment on time. Bankruptcy is restricted to the cases filed under Chapter 7 and Chapter 11 in this study.

We only identify a few papers on the bankruptcies in REITs. Asness and Smirlock (1991) explore the information transfer from the bankruptcy of a REIT to other REITs with similar asset portfolios. Liu and Liu (2013) investigate the stock market reactions of retail REITs to the tenants’ bankruptcies.

The predictive abilities on subsequent distress/bankruptcy may differ in these measures. Agarwal and Taffler (2008) find that the predictive powers between accounting-based measures (Z-score) and market-based measure (expected default probability from option pricing model) are similar. Yet Hillegeist et al. (2004) show that the market-based measure from option pricing model dominates the measures from accounting-based models (Z-score and O-score).

A partial list of asset pricing studies in REITs include: McIntosh et al. (1991), Peterson and Hsieh (1997), Redman et al. (1997), Allen et al. (2000), Friday et al. (2000), Chui et al. (2003), Chan et al. (2005), Ooi et al. (2009), Price et al. (2012), DeLisle et al. (2013), Ghosh et al. (2013), Cakici et al. (2014), Glascock and Lu-Andrews (2014), Giacomini et al. (2015, 2017), Anzinger et al. (2017), Bond and Xue (2017) and others.

Z-score is calculated as: 1.2(working capital / total assets) + 1.4(retained earnings / total assets) + 3.3(earnings before interest and tax / total assets) + 0.6(market value of equity / total liabilities) + 1.0(sales / total assets). The higher the value, the lower the probability of bankruptcy.

O-score is calculated as: −1.32 - 0.407log(total assets) + 6.03(total liabilities / total assets) - 1.43(working capital / total assets) + 0.0757(current liabilities / current assets) – 1.72(1 if total liabilities > total assets, 0 otherwise) – 2.37(net income / total assets) – 1.83(funds from operations / total liabilities) + 0.288(1 if a net loss for the last two years, 0 otherwise) – 0.521(net income this year – net income last year) / (absolute value of net income this year + absolute value of net income last year). We follow the Dichev (1998) to calculate the O-score. The higher O-score value indicates the larger probability of financial distress.

Failure probability is calculated following the 12-month logistic regression in Table IV of Campbell et al. (2008): −20.26 × NIMTAAVG +1.42 × TLMTA −7.13 × EXRETAVG +1.41 × SIGMA −0.045 × RSIZE −2.13 × CASHMTA +0.075 × MB − 0.058 × PRICE −9.16, where NIMTA is the quarterly net income divided by market value of the total assets, MTA is the market value of total assets equal to the sum of total liabilities and market capitalization, NIMTAAVG = (1 – γ³)/(1– γ¹²)[NIMTA j-1 + γ³ × NIMTA j-2 + γ⁶ × NIMTA j-3 + γ⁹ × NIMTA j-4 (γ = 2^–1/3; j refers to quarter j), TLMTA is the total liabilities divided by the market value of total assets, EXRET = log (1 + stock return) – log (1 + S&P return), EXRETAVG = (1 – γ³)/(1– γ¹²)[EXRET t-1 + γ × EXRET t-2 + γ² × EXRET t-3 + … + γ¹¹ × EXRET t-12] (t refers to month t), SIGMA is annualized standard deviation of the firm’s daily stock return over the previous 3 months, RSIZE = log(stock market cap/S&P500 total market value), CASHMTA is the ratio of cash and cash equivalents over market value of total assets, MB is the ratio of market value of total assets divided by adjusted book equity value, adjusted book equity value = book equity value +0.1× (market capitalization - book equity value), PRICE is the log stock price in previous month truncated at $15. The failure probability is calculated in each month.

EDF is calculated based on Merton’s model assuming that the equity of a firm is a call option on the total firm value with strike price equal to the total debt value. EDF is the implied probability of default in one-year horizon, estimated by the stock equity value, stock return, the total debt value, the volatility of firm value and risk-free rate. The details can be found in Bharath and Shumway (2008).

The Fama-French daily factors are extracted from Kenneth French’s website. To calculate the idiosyncratic volatility from the regression, it requires that a firm should have at least 15 daily returns in a month.

REITs are sorted by the distress risk variables constructed based on available accounting and financial information (see the footnotes 5–8). We find that more than 50% of distressed REITs are correctly assigned to the riskiest portfolios (Portfolio 5) in our sample and around 70% are in the riskiest and second riskiest portfolios (Portfolios 5 and 4). There are a few cases that distressed REITs are misclassified into the safest portfolios. However, Table 5 shows that overall the REITs in the Portfolio 5 are riskier than those in the Portfolio 1 but earn lower return. The profitability to long the safest REITs and short the riskiest REITs based on EDF or F-score is not affected by the potential misclassification error. We thank a referee’s suggestion to discuss the classification error in the portfolio construction.

Vassalou and Xing (2004) also construct the portfolios based on the expected default frequency from option pricing model in previous month and rebalance the portfolios monthly. They show that the return differentials between high default risk portfolio and low default risk portfolio are positive and significant in equally-weighted portfolios but insignificant in value-weighted portfolios.

We appreciate the suggestion of the robustness tests using the wild bootstrap method from a referee.

We also run OLS regressions of Equation (1) with clustered standard errors by firm and month. The results are similar to the main results. In addition, the results from estimators by the wild bootstrap method are also essentially the same as the main results. These results are not reported to conserve the space but available upon request. We thank the suggestions from a referee.

The major change in REIT industry was the creation of UPREIT structure, which allowed the investors to avoid capital gain tax and was firstly introduced in the Taubman’s IPO in 1993. The Revenue Reconciliation Act of 1993 modified the “five or fewer” rule that restricts the holdings of institutional investors on REITs. The institutional investors gradually dominate REIT market after 1993 (Chan et al. 2005). Also, in early 1990s, the organizational structure of REITs shifted from “externally advised” to “internally-advised” managerial form, and the latter can better align the interests between the advisors/managers and the REIT shareholders (Ambrose and Linneman 2001). The equity REIT market had experienced dramatic growth after these fundamental changes.

To conserve space, we only report the results from value-weighted portfolio returns. The results from equal-weighted portfolio returns remain similar and are available upon the request.

In unreported results, we run time series regressions of long-short portfolio returns on the Fama-French three factors and the VIX. The alphas become trivial in the regressions on equal-weighted and value-weighted portfolio returns sorted by EDF or F-score.