The Information Content
of REIT Credit Rating Actions and Transparency
Alan Tidwell1
Alan Ziobrowski2
Paul Gallimore3
and
SeungHan Ro4
1
Department of Accounting and Finance, Turner College of Business, Columbus State
University, 4225 University Avenue; Columbus Georgia 31907, Phone (706) 507-8160, Fax
(706) 568-2184, Email [email protected]
2
Department of Real Estate, J. Mack Robinson College of Business, Georgia State University,
35 Broad Street, Suite 1405; Atlanta, GA 30303, Phone (404) 413-7726, Fax (404) 413-7736,
Email [email protected]
3
School of Real Estate & Planning, Henley Business School, University of Reading,
Whiteknights, Reading, RG6 6UD, UK, Phone: 44 (0) 7926 890530, Fax: 44 (0) 118 378 8172,
Email [email protected]
4
Department of Real Estate, Konkuk University, 1 Hwayang-dong, Gwangjin-gu, Seoul, Korea
143-701, Eamil [email protected]
Abstract
We examine the short-run and long-run price reaction of equity REIT shares following credit
rating actions, testing the transparency of the REIT structure. Generally, the economic effect on
the stock price is subdued for both upgrades and downgrades compared to prior literature
examining the broader U.S. equity market. An examination of trading volume revealed a
significant increase in trading in reaction to downgrade credit rating changes, with a more
subdued response to upgrades. The findings support the notion that REITs are more publicly
forthcoming about the expectation of positive news in comparison to negative news.
Keywords: REITs, Transparency, Credit Rating Actions, Information Content
1. Introduction
REITs primarily rely on external sources of funding for the financing of capital projects and
asset acquisitions, as the REIT corporate structure is unique in that they have a relative inability
to retain earnings due to the regulatory provision requiring a minimum 90% distribution of
taxable income (Ooi, Ong, and Li 2010). Therefore REITs initially finance growth with
unsecured credit, utilizing a “bridge financing strategy” and then refinance the short term debt
with a combination of public long-term bond debt or seasoned equity (Elayan, Meyer, and Li
(2004)). Credit lines also often act as a substitute for cash (Hardin, Highfield, Hill and Kelly,
2009) capable of funding up to 17% of assets (Hardin and Hill, 2011) and offering a form of
insurance to REIT firms (Ooi, Wong, and Ong, 2011). As a result, REITs subject themselves to
frequent monitoring in the capital markets (Gosh, Chinmoy, Nag and Sirmans (1997) and Hardin
and Hill (2008)). This reliance on external financing has motivated many studies examining
REIT capital structure, transparency, and financing decisions.
The literature points to credit ratings as a factor that broadly impacts external financing decisions
and capital structure. REITs target debt levels to obtain credit ratings just above the investment
grade cutoff point where clear differences in financing cost and length to maturity can be
observed (Brown and Riddiough (2003) and Highfield, Roskelley and Zhao (2007)), and REITs
with banking relationships are more likely to have credit ratings (Hardin and Wu (2010)).
Additionally, Campbell, Dodd, Hill and Kelly (2012 working paper) offer evidence that credit
ratings are inversely related to dividend volatility and measures of financial constraints.
The credit rating literature of non-REIT firms supports the notion that rating agencies are
information specialists. The research suggests that changes in ratings or rating announcements
1
disseminate new information to shareholders, and have an effect both in the short-term and longterm on security prices and the firm’s financing decisions.1 Although the REIT literature is rich
with studies documenting the significance of credit ratings in relation to financing decisions, to
our knowledge the REIT literature is silent on studies examining the information content of
credit ratings and performance.
Given the competing views of REIT transparency, our study contributes to the literature by
empirically testing the pricing and volume reaction of REITs to changes in their long term or
unsecured credit ratings, and measuring the shareholder wealth effects of these changes. We
examine REIT returns before the rating action, at the time of the rating action, and following the
rating action, and test the null hypothesis of no price reaction. Abnormal trading volume is also
examined, as previous literature supports the notion that differential results can occur across
returns and volume, and in some cases trading volume might be more informative.
We compare our results to previous studies that examine market reactions to credit rating
changes across the broader U.S. equity market. Given the importance of credit ratings to the
REIT sector, if REITs have informational transparency, then security pricing should adjust in the
weeks or months prior to the announcement in anticipation of the event. As such, REITs should
not exhibit a substantial stock price reaction to the announcement of credit rating changes. This
study tests the null hypothesis of no price reaction surrounding a REIT’s credit rating change.
We then test for systematic determinants impacting the magnitude of CARs produced as a result
of these credit rating changes.
Collectively, the findings suggest that credit rating changes appear to be anticipated, however not
fully. Particularly in the case of credit rating downgrades, credit rating changes seem to
2
disseminate some new information to market participants. While news of credit rating upgrades
appears to be more transparent. Additionally, an analysis of the magnitude of REIT abnormal
returns on possible explanatory variables points towards dividends as being a significant
moderating variable.
From a trading volume perspective, consistent with the return analysis, credit rating downgrades
result in higher trading activity compared to credit rating upgrades. This asymmetric response
could be caused by a lack of pricing consensus among individual investors at the time of a credit
downgrade announcement, a potential result of firm management concealing negative news and
their willingness to release positive news leading up to rating actions as suggested by Kothari,
Shu and Wysocki (2009).
The implications of our findings generally support the contention that REITs are relatively
informationally transparent. In circumstances where the indications of REIT transparency are
less strong, e.g., credit rating downgrades, they are still stronger than has been found in prior
studies of this condition in non-REIT stocks subsequent Regulation Fair Disclosure.
2. Background Literature
2.1 Stock Reaction
Research on the broader US equity market shows that corporate credit rating downgrades and
negative Credit Watch announcements affect stock prices negatively. Holthausen and Leftwich
(1986) examine the common stock price reaction to bond rating changes and Credit Watch
announcements and find evidence of a stock price response to all events except an actual rating
upgrade. Their findings are similar to Griffin and Sanvincente (1982) in that bond downgrades
result in significant price reactions while bond upgrades do not produce a significant reaction.
3
Jorian, Liu and Shi (2005) offer evidence that the time period studied may have an impact on this
differential. The authors study the change in the information content of ratings announcements
pre- and post-Regulation Fair Disclosure (Reg FD). They contend that because post-Reg FD,
rating agencies have access to inside information that cannot be disclosed to other analysts, the
information content (i.e. stock price effect) will be more profound post-Reg FD. Indeed, they
find that the stock prices react in greater magnitude to downgrades following Reg FD than
previously reported, and, now rating upgrades have a positive and statistically significant impact
on stock prices.
Dichev and Piotroski (2001) examine the long-run stock returns following bond ratings changes,
using Moody’s bond rating changes from the 1970 to 1997 time period. They find limited
abnormal performance for upgrades, but statistically significant negative abnormal returns of 10
to 14% in the year subsequent to a downgrade. Thus, they conclude the market does not fully
anticipate the negative implications of downgrades for future profitability. The authors suggest
that information processing bias like optimism can lead to erroneous conclusions when applied
to downgrades resulting in subsequent adjustments in stock prices. It is unclear why investors
react asymmetrically, with downgrades generally producing a stronger reaction than upgrades in
non-REIT firms. Ederington and Goh (1998) suggest that the observed asymmetric market
response is because firms voluntarily release good news to the public but rely on credit rating
agencies to release negative or unfavorable information. Kothari, Shu and Wysocki (2009), in a
study corporate dividend changes and management earnings forecasts, suggest that a greater
amount of news is impounded in stock prices prior to good news announcements relative to bad
news, and therefore the authors conclude that good news tends to be leaked to the market. These
findings are consistent with the survey evidence collected in Graham, Harvey, and Rajgopal
4
(2005), as managers have strong incentives to withhold bad news and chance that the firm’s
economic condition improves prior to the required news disclosure.
2.2 Trading Volume Reaction
Trading volume reflects investors’ activity by summing all investors’ trades, while security
returns reflect an aggregation of investors’ activity. Therefore, returns may be less sensitive than
trading volume to the information content of credit events. Increased volume reflects a lack of
consensus regarding the price, induced by information discovery. This lack of consensus
regarding the price leads to increased trading activity that may be counterbalancing from a price
perspective (Beaver 1968). Bamber and Cheon (1995) find that although price and volume
reactions are on average positively related to earnings announcements, the relationship is weaker
than expected. Their findings generally support those of Kim and Verrecchia (1991) that when
an announcement generates a differential belief among investors, trading volume is often high.
This differential belief is caused by varying degrees of precision in private information.
2.3 REIT transparency
The notion of transparency in the REIT literature is a source of debate. Damodaran, John, and
Liu (1997) and Hardin and Hill (2008)) attribute REIT transparency to the regulatory provision
requiring REIT’s to distribute 90% of taxable income. This provision limits REIT management’s
ability to access discretionary income, and requires REITs to fund growth through accessing
external capital markets. Gentry Kemsley and Meyer (2003) argue that the value of a REIT is
simply the aggregate fair market value of its assets, making the valuation of REITs relatively
transparent. Similarly, Hartzell, Kallberg and Liu (2005) and Hartzell, Kallberg and Liu (2008)
suggest that equity REITs are relatively simple to value because they hold portfolios of tangible
assets and have a transparent structure. Higgins, Ott and Van Ness (2006) find evidence that the
5
1999 Funds From Operation (FFO) accounting change reduced the information asymmetry that
existed between REIT shareholders, REIT management and market participants. And, Hartzell
and Starks (2003) suggest that the relatively high institutional ownership in REIT firms leads to a
more active monitoring of REIT management, potentially reducing agency conflicts. However,
Feng, Ghosh, He and Sirmans (2010) contend that the monitoring role of institutional owners in
the REIT sector is inconclusive.
Campbell, Ghosh and Sirmans. (2001) suggest the limits imposed on sources of REIT income
limit managers’ experience and employment potential resulting in entrenchment as a defensive
mechanism. In a similar fashion, Ghosh and Sirmans (2003) suggest that legal provisions
restricting ownership prevents large blockholders from acquiring an ownership interest, thus
reducing the threat of takeovers and contributing to management complacency. Due to this
limitation, Anglin, Edelstein, Gao, and Tsang (2011) examine the quality of REIT corporate
governance and find that information asymmetries are reduced in REITs with high quality
corporate governance. These provisions collectively may contribute to management
entrenchment and insulate REIT managers from takeovers by unsatisfied shareholders who are
limited in their ability to acquire large ownership stakes.
3. Data
We use the SNL database containing historical credit ratings of US real estate firms rated by
Moody’s, Standard and Poor’s, Fitch and Dominion Bond Rating Service (DBRS) during the
January 2000 to December 2009 time period. To estimate the cumulative abnormal return (CAR)
and calendar time regression parameters, the daily and monthly returns from January 2000 to
December 2011 were used as appropriate given the estimation period or holding period. It is
6
noted, that in November 2000 Regulation Fair Disclosure came into effect allowing firms to
share information with credit rating analysts without making it publicly available. (Jorion, Liu,
and Shi 2005).
Our dataset consists of firm-level credit ratings, from which we use the ratings for long-term and
senior unsecured debt. Firm-level debt ratings were considered for this study because these
ratings presumably provide a proxy for a firm’s ability to meet its debt obligations, as opposed to
secured debt ratings which are unique to the real estate attached as collateral. During our sample
period long-term and senior unsecured debt ratings account for 1156 credit rating actions, with
yearly totals increasing progressively in the first part of the decade, peaking in 2006 with 206.
Rating actions can include: ‘initiate’, ‘affirm’, ‘upgrade’, ‘downgrade’ or ‘remove’. The largest
number (833) was ‘affirm’. Downgrades outnumber upgrades (140 to 101), a juxtaposition
echoing credit rating research on non-REITs (Hand et al. 1992). Figure 1 plots the yearly
numbers of upgrades and downgrades over the period, showing that these generally move in
tandem for the two types of debt notes. Table 1 presents these yearly numbers by rating agency.
Standard and Poor’s and Moody’s are the most active agencies with a combined 64% share of all
rating actions and 71% share of all rating changes
Please Insert Figure 1 and Table 1
Our sample of 241 U.S. REIT rating changes (upgrades/downgrades) is then reduced to filter out
redundant actions, actions whose effect is contaminated by other announcements, and cases
where we do not have all necessary daily returns.2 Similarly to Dichev and Piotroski (2001) we
only retain the event that occurs first if the rating actions for the same firm’s long-term and
senior unsecured debt are announced within a four-day window. We use an additional filter
7
where the information content of credit rating announcements may be contaminated by other
firm-specific news (i.e. earnings announcements, dividend distributions, mergers and
acquisitions activity, debt retirements). We consider an announcement contaminated where any
firm-specific substantial price-relevant news event is detected by Lexis-Nexis within a three-day
window surrounding the day of a rating action. Additionally, our analysis requires the
availability of daily price returns, to calculate abnormal returns during our estimation window.
The sample of US REIT credit rating upgrades and downgrades during January 2000 to
December 2009 is accordingly reduced from 241 to 108. The filtering process resulted in
approximately 55% data loss.
The remaining sample consists of 61 credit rating upgrades (60 pre-estimation time period) and
48 credit rating downgrades, totaling 109 U.S. equity REIT rating changes. Tables 2 and 3
describe the sample rating changes by year and agency, and magnitude. Consistent with
Highfield, Roskelley, and Zhao (2007) and Campbell, Dodd, Hill, and Kelly (WP 2012), the
BBB- to BBB+ rating level appears to be the most active containing 34% of the changes. A
rating level change (e.g., BB- to B+) occurred in 37% of the observation.
4. Shareholder wealth effects
In this Section we explain the methods employed and present the results of the tests measuring
abnormal returns during ‘windows’ that (1) are contemporaneous with the credit rating
announcement date, (2) precede the credit event, and (3) follow the credit event. We also explain
the method used in the analysis of cumulative abnormal returns (CARS), as we check for
systematic firm-specific explanatory variables
4.1 Abnormal returns in contemporaneous credit rating announcement windows
8
We begin by examining the impact of credit rating actions as reflected in REIT stock price
movements surrounding a credit announcement event: specifically, the extent of any abnormal
return (AR) during an announcement window(s). We adopt the market model event study
methodology as established by Brown and Warner (1980, 1985), to estimate a firms abnormal
return on each day of the announcement event window. We then aggregate the daily abnormal
returns to produce the cumulative abnormal return (CAR) for that window. We do this for four
separate windows treating the day of the announcement as listed by SNL as Day 0, these
windows are: (1) Day 0 and Day +1; (2) Day 0 to Day +2; (3) Day -1 and Day 0; (4) Day -1 to
Day +1.3
The abnormal returns for event j are thus calculated as:
∑
where d represents each day of the event window, Rjt is the actual daily return for the firm and
Rjmt is the market return over the event’s estimation period. The market return is estimated using
the CRSP value-weighted and the CRSP Ziman REIT value-weighted indices as the market
proxy. We estimate the model parameters αj and βj using an estimation window beginning on
Day +60 with a maximum estimation length of 315 trading days. 4 In setting our estimation
window to follow the rating change, we take into account previous research that has shown
downgrades tend to occur after other bad news and when the firm’s stock price has performed
poorly (Ederington and Goh 1998). 5 As a robustness measure, we also examine a pre-event
estimation window beginning on Day -375 with an estimation length of 315 days. After
calculating event CARs, we then calculate and report the cumulative average abnormal return
(CAAR). Where, the CAAR is the arithmetic average of all sample event CARs.
9
In addition to the equally-weighted CAAR, we also calculate the precision-weighted CAAR
(PWCAAR). The PWCAAR weights each event’s CAR in inverse proportion to the variability in
their prediction errors. The Patell (1976) test is used when the PWCAAR is reported and
represents a standardized abnormal return test that estimates a separate standard error for each
credit event. The t-statistic used when calculating the significance of CAARs equally weights the
observations, limiting the CAAR model to assuming constant error variance across event CARs.
The PWCAAR accounts for the possibility of a non-constant error variance across different
event CARs. However, the t-statistic used in CAAR is robust to cross-sectional dependence,
which may be problematic for the Patell test. Both the Patell and t-statistic test the same
hypothesis that the population parameter mean is equal to zero; however, the weights applied to
the observations may differ.
The nonparametric generalized sign test was also conducted using the normal approximation to
the binomial distribution as described by Cowan (1992). The generalized sign test compares the
fraction of positive abnormal returns during the event period with the fraction obtained during
the estimation period. The null hypothesis is that these fractions are the same.
The results for the CAAR, PWCAAR and generalized sign test using both the CRSP valueweighted index and the CRSP Ziman value-weighted are shown in Tables 4 and 5. In this study,
we generally do not observe a significant stock market reaction to upgrades across the
contemporaneous event windows when we use the CRSP value-weighted index as the market
proxy. For upgrades, the two-day (0, +1) CAAR is not statistically significant at -0.05% and
0.07% respectively for the post and pre estimation event time periods. The PWCAAR is also not
statistically significant having parameter estimates of -0.19% and -0.07%. Furthermore, the
10
nonparametric generalized sign test did not detect an abnormal number of positive market
adjusted returns, based on the respective estimation periods.
When we substitute the CRSP value-weighted index with the sector specific CRSP Ziman REIT
value-weighted index we find that the parameter estimates in the two-day (0, +1) CAAR remain
insignificant, however the four-day (-1, +1) CAAR window becomes statistically significant. The
PWCAARs also remain insignificant in the two-day (0, +1) window, but four-day (-1, +1)
CAAR window becomes significant in the pre-estimation analysis.
Please insert Figure 3 and Table 4
For downgrades, when examining CAAR’s using the CRSP value-weighted market index as the
market proxy in the post estimation period analysis, we only find a significant market reaction in
one of the four event windows: the (0, +2) CAAR. An analysis of the PWCAARs and the
nonparametric generalized sign revealed no statistically significant market reactions to credit
rating downgrades in the post estimation period analysis. When the pre-estimation model
specification is considered, credit rating downgrades generally produce negative and significant
CAARs ranging from 0.18% to -2.03%.
Please insert Table 5 and Figure 3
From a REIT pricing perspective when using a broad market index, credit rating announcements
generally do not appear to disseminate new information to the market suggesting that REITs
appear to be relatively transparent in this regard. After adjusting the model to include a REIT
sector market proxy, credit announcements appear to disseminate some new information to
market participants, particularly credit downgrades. These results suggest that the CAAR and
11
PWCAAR parameter estimates are moderately sensitive to model specifications with regards to
the market proxy, and timing of the estimation period. Models calculating αj and βj based on the
pre-estimation window and those that include the REIT Ziman index as a market proxy tended to
produce larger and more significant coefficients. The economic effect of this information content
remains marginal, the average CAAR and PWCAAR coefficient produced across our upgrade
sample is 0.23% while the downgrade sample produced and average coefficient of -0.81% across
the 32 CAAR and PWCAAR windows.
4.2 Abnormal returns in windows preceding the credit rating announcement
The credit ratings of REITs have a real effect on the cost of capital; therefore if REITs are
transparent then we would expect to find a stock price reaction prior to the credit change
announcement. We examine windows for 3 months, 6 months and 1 year prior to the
announcement by using calendar-time regression to estimate the abnormal returns. For each
window, we construct daily portfolios composed of firms that have an announcement event that
follows within the specified window period (3 months, 6 months, 1 year). We calculate the return
series for the portfolios and then regress the excess of this return over the risk-free rate against
the Fama-French (1993) risk factors, using the CRSP value-weighted market index return Rmt,,
‘small-minus-big’ market capitalization factor SMB,‘high-minus-low’ market-to-book ratio
factor HML, and Carhart’s (1997) momentum factor UMB. We also substitute the CRSP valueweighted market index with the CRSP Ziman REIT value-weighted index as the market proxy.
The model is:
(
)
The intercept, αi represents the average monthly abnormal return.
12
We estimate the model by both an equal weighted OLS model and a weighted least squares
model (WLS), where each trading month is weighted by the number of observations in the
calendar-time portfolio corresponding to the time period. Both the OLS and WLS models test the
null hypothesis that α = 0.
We find significant positive abnormal returns for upgrades in all three time periods, with the
magnitude greatest three and six months prior to the announcement. When examining
downgrades, the coefficients are generally negative with the one-year preannouncements periods
producing significant abnormal returns. These results provide evidence supportive of the
contention that investors anticipate credit rating changes and trade accordingly.
Please insert Table 6
4.3 Abnormal returns in windows following the credit rating announcement
Research suggests that focusing on only a short return window may be incomplete as stock
prices may under-react to firm-specific announcements, adjusting to new market information
over an extended time horizon.6 We therefore test the long-run performance of REITs following
rating changes by again estimating WLS and OLS calendar-time regression models, using six
months, one year and two years as the periods for construction of the calendar-time portfolios.
The WLS and OLS calendar-time regression model results reveal evidence of abnormal returns
of REITs following upgrades, but not downgrades. The economic magnitude of abnormal returns
following upgrades range from 0.50% to 0.90% a month across the time horizons, with the one
year time period producing the largest impact. The abnormal returns for downgrades are
statistically insignificant across the three holding periods, and are robust to the methodology
employed. These results when taken in combination with the previous short-run performance
13
findings suggest that although credit events are anticipated, market participants still seem to
under-react to the long-run benefits of positive credit rating changes, possibly due to
conservatism (Griffin and Tversky 1992; Edwards 1968). Results are presented in Table 7.
Please insert Table 7
4.4 Summary of Shareholder Wealth Effects of Credit Rating Changes
In summary, we find positive abnormal performance preceding a credit upgrade announcement
and negative abnormal performance preceding a credit downgrade announcement. In the window
covering the day the credit rating change was announced and the day immediately subsequent we
generally do not find a significant reaction for upgrades, but do find statistically significant
negative reactions to downgrades. For the 12-month time period following the announcement we
find statistically significant positive abnormal returns for upgrades. Downgrades generally
produce negative coefficients, but they lack statistical significance at conventional levels.
Section 5. Determinants of CARS
We employ an OLS regression model to examine the relationship between abnormal returns and
a set of potential firm-specific variables. The selection of control variables is based on prior
literature. We obtain accounting data including market capitalization, total assets, total debt, net
income, and operating expenses from COMPUSTAT, 10Q and 10K SEC filings. Dividends per
share were collected from the CRSP/Compustat Merged Database Fundamentals Quarterly data
set. The number of analysts following a REIT is acquired from Institutional Brokers’ Estimate
System (IBES), while institutional holdings are obtained from the CDA/Spectrum 13(f)
Institutional Holding data set, provided by Thomson Reuters. The model is formally expressed as:
14
where, the selection of firm specific variables employed in the CAR analysis is based on prior
empirical findings. Shleifer and Vishny (1997) argue that larger firms’ managers are less
bounded by shareholders’ discipline due to less governing power of shareholders in larger
companies, therefore we control for firm size (lnSIZE). Debt ratio (DebtR) is controlled for to
examine Jensen’s (1986) free cash flow theory. This theory contends that managers make better
investment decisions when the firm has more debt, since less free cash flow makes management
less likely to waste resources. Campbell, Petrova and Sirmans (2006) contend that operating
performance may affect firm value; we control for operating performance prior to the
announcement of a credit rating change from the income (NIR) and expense (ExpR) perspectives.
Similarly, Bauer, Eicholtz, and Kok (2010) find that shareholders of firms with high payout
ratios are less likely to benefit from greater corporate governance, we control for this through the
inclusion of dividends scaled by shares (DivPS). While, Hartzell, Sun and Titman (2006) find
that institutional investors play a significant monitoring role for REIT shareholders, and Chung
and Jo (1996) find that the number of security analysts following a firm is positively related to
the market value of the firm. Therefore, REIT institutional holdings are controlled for using INST
and the number of analysts that forecast the performance for a REIT is controlled for using
ANALY.
Table 8 shows the descriptive statistics of the event CARs and potential explanatory variables
used in the analysis. As we might expect, net income ratios, institutional investors, analyst
coverage, and dividends per share for upgrade events in Panel A are relatively higher than those
for downgrades in Panel B. Also, upgrade firms seem to maintain lower debt ratios. Panel D
15
exhibits correlation coefficients among the explanatory variables. The correlation among firm
size, institutional ownership, number of analysts, and dividends per share is relatively high
indicating that larger REITs are more attractive to institutions, have greater analysts interest, and
pay out more in dividends.
Please insert Table 8
Table 9 presents the results of regressions. The first column combines samples of upgrades and
downgrades, where the CAR is the two-day (0, 1) absolute value of the cumulative abnormal
return, and shows that the amount of dividends per share subdues the magnitude of the stock
price reaction to credit rating changes. This finding suggests that credit rating changes have
lower impact on firms with relatively higher dividends, a finding complementary to prior
literature (Bauer, Eicholtz, and Kok (2010)). Surprisingly, we find a positive and statistically
significant coefficient between abnormal returns and institutional ownership. When we separate
upgrade and downgrade credit events, it becomes clear that institutional ownership significantly
increases the magnitude of CARs resulting from downgrade credit rating changes, but not
upgrades. This finding is consistent with the contentions of Kothari, Shu and Wysocki (2009)
and Graham, Harvey, and Rajgopal (2005) that firm management may be incentivized to conceal
bad news from the market until disclosure is required. The remaining parameter estimates related
to operating performance generally have the expected sign, but do not significantly explain REIT
CARs in our sample.
Please insert Table 9
6. REIT Trading Volume
16
We examine the impact of credit rating actions as reflected in trading volume movements
surrounding a credit announcement event by estimating the log-transformed volume of each
security in our sample, following the methodology of Campbell and Wasley (1993). This is
carried out in a similar fashion to the abnormal return model, except the log-transformed volume
replaces the security total return. Trading volume models have been shown to be powerful tests
of the information content contained in earnings announcements.
The trading volume models are estimated using the percentage of outstanding shares traded on
the CRSP value-weighted market index. We calculate the cumulative abnormal trading volume
(CAV) and the precision weighted abnormal trading volume (PWCAV) and their arithmetic
averages (CAAV and PWCAAV) across the event observations. Abnormal trading volume is
calculated by subtracting the expected trading volume, based on Vjmt and derived from the
estimated market model, from the daily trading volume as the natural log of the percentage of
outstanding shares traded Vjt for the firm.
∑
∑
We apply these to the same set of four event windows (Day 0 and Day +1, Day 0 to Day +2, Day
-1 and Day 0, and Day -1 to Day +1) as for our abnormal return analysis, and use the same pre
(Day -375; Estimation Length 315 Days) and post (Day +60; Estimation Length 315 Days)
estimation windows.
17
The generalized sign test compares the fraction of abnormal positive trading volume during the
event period with the fraction obtained during the estimation period. The null hypothesis is that
these fractions are the same.
These results are reported in Table 10. We observe a significant positive trading volume reaction
to credit downgrades. This is likely the result of a lack of precision in preannouncement
information, as firms prefer to guard negative information contributing to a lack of pricing
consensus, Kim and Verrecchia (1991). We find credit upgrades exhibit a weaker influence on
investors’ trades. This suggests that some investors may treat credit upgrades as a confirming
event and thus are less inclined to adjust their portfolio.
Please Insert Table 10
7. Conclusion
This study provides additional insight into the information content of REIT credit ratings, and
the degree of REIT transparency. Principally, this study examines the price reaction of REIT
shares during ‘windows’ that are contemporaneous with the credit rating announcement date,
precede the credit rating change, and subsequent the credit rating change. We also search for
informational asymmetry in the REIT industry by examining the relationship between the
magnitude of the price reaction resulting from a credit rating change and a set of potential firmspecific variables.
In the short-run, for upgrades, we find no substantial evidence of abnormal returns immediately
following credit rating changes when using the CRSP value-weighted index as the market proxy,
consistent with the notion of REIT transparency. When the sector specific CRSP Ziman REIT
index is substituted as the market proxy we find limited evidence of abnormal returns
18
immediately surrounding credit rating changes, however economically this reaction is
substantially less than the reaction found in prior studies examining the broader U.S. equity
market (e.g., Jorian Liu and Shi (2005)). From both an abnormal return and trading volume
perspective, downgrades generally produce a stronger reaction than upgrades both statistically
and economically.
The analysis of CARs surrounding credit rating changes reveals that the amount of dividends per
share attenuates the stock price reaction to credit rating changes, complementing evidence found
in prior research. This finding suggest that credit rating changes have a lower impact on firms
paying high dividends, at least at the time of the announcement.
When we examine the time period preceding a rating change we find superior abnormal returns
in all three time horizons (three-months, six-months and one-year) for upgrades suggesting
positive information is being absorbed into equity pricing. When downgrades are examined we
find evidence of negative abnormal returns in the six-month and one-year time periods preceding
a rating downgrade. From a pricing perspective, a substantial amount of information content
contained in credit rating changes appears to disseminate to market participants prior to the
official credit rating change news release, as the stock price adjust in the months leading up to
the event.
We also examine the long-run performance of REITs subsequent to rating changes and find
significant monthly abnormal returns for firms with upgrades for up to two years. We find no
significant abnormal performance for downgrades. These findings when taken in combination
with the short-run performance results indicate that although anticipated, investors tend to
undervalue the effect of a positive credit rating change, perhaps due to conservatism (Griffin and
19
Tversky 1992; Edwards 1968), and consequently the stock price continues to adjust in the tested
post announcement time periods.
The results observed in our study suggest managerial transparency might be contingent on the
type of information being disseminated. Information from upgrades (i.e. positive news) seems to
enjoy greater transparency, less uncertainty and translates into greater preannouncement
information precision. In contrast, a credit downgrade (i.e. negative news) tends to provoke
substantial positive trading activity surrounding a credit event announcement suggesting a lack
of transparency in preannouncement information. These factors in our interpretation in the form
of differences in response to upgrades versus downgrades, and in trading volume versus pricing,
serve as pointers for further research direction in this field.
20
Table 1 Descriptive statistics for REIT senior unsecured and long-term debt rating actions by year
and agency.
Agency
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Rating Downgrades
Moody's
2
1
0
0
2
0
2
0
8
5
S&P
8
6
7
5
0
3
4
5
28
20
Fitch
1
2
2
5
0
1
2
1
7
13
DBRS
0
0
0
0
0
0
0
0
0
0
Subtotal
11
9
9
10
2
4
8
6
43
38
Rating Upgrades
Moody's
0
1
0
2
2
3
3
6
1
1
S&P
2
2
3
7
4
8
12
4
5
0
Fitch
2
0
0
0
3
6
6
6
1
9
DBRS
1
0
0
0
0
0
0
1
0
0
Subtotal
5
3
3
9
9
17
21
17
7
10
Affirm
Moody's
4
6
6
15
13
13
19
5
10
9
S&P
6
23
42
51
38
49
75
52
47
35
Fitch
3
8
18
16
29
34
66
45
38
24
DBRS
0
1
2
2
7
6
4
4
8
0
Subtotal
13
38
68
84
87
102
164
106
103
68
Initiate
Moody's
0
2
1
0
1
2
3
0
1
0
S&P
0
2
2
1
4
2
3
1
2
1
Fitch
1
0
2
3
3
1
5
0
1
5
DBRS
0
1
0
0
3
0
0
0
0
0
Subtotal
1
5
5
4
11
5
11
1
4
6
Remove
Moody's
0
0
1
1
3
0
0
2
0
1
S&P
0
1
1
0
1
0
2
2
1
0
Fitch
0
0
0
0
1
0
0
2
6
2
DBRS
0
0
0
0
0
0
0
0
0
2
Subtotal
0
1
2
1
5
0
2
6
7
5
Total Rating Actions
30
56
87
108
114
128
206
136
164
127
Notes: The rating action changes were collected from the U.S. Real Estate SNL database from 2000
through 2009. All rating actions are for the senior unsecured and long-term debt credit ratings of U.S.
public REITs.
21
Table 2 Sample REIT rating changes by year and agency.
Standard and Poor's
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Sum
Upgrades
2
2
2
4
3
4
9
2
2
0
30
Moody's
Downgrades
4
0
3
2
0
2
2
3
4
8
28
Upgrades
0
1
0
1
0
3
2
5
1
1
14
Fitch
Downgrades
2
0
0
0
1
0
1
0
2
3
9
Upgrades
1
0
0
0
2
2
4
4
0
2
15
DBRS
Downgrades
0
2
0
4
0
0
1
0
1
4
12
Total
Upgrades Downgrades Upgrades Downgrades
5
1
0
4
2
0
0
3
3
0
0
2
6
0
0
5
1
0
0
5
2
0
0
9
4
0
0
15
3
1
0
12
7
0
0
3
15
0
0
3
2
0
61
48
Table 3 Sample REIT rating changes by magnitude.
Revised Rating
A
A
ABBB+
BBB
BBBBB+
BB
BBB+
B
BCCC+
A2
2
BBB+
BBB
7
Previous Rating
BB+
BB
BB-
B+
B
B-
4
1
1
CCC+
CCC
1
6
6
7
1
BBB-
3
6
13
11
3
4
2
1
6
2
4
1
1
7
4
2
1
22
Table 4 Market model mean abnormal return and precision weighted cumulative average abnormal
return (CAAR) for credit rating upgrade changes.
Percent of
Precision Generalized
Window
CDA
CARs
CAAR
Weighted Sign
Patell Z
(Days)
t-stat.
Positive
CAAR
Panel 1 Credit Rating Upgrades CRSP index (+60, +315) estimation period
(0,+1)
57%
-0.05%
-0.19%
0.99
- 0.14
-0.74
(0,+2)
53%
0.27%
0.20%
0.47
0.64
0.64
(-1, 0)
58%
0.31%
0.24%
1.24
0.90
0.96
(-1,+1)
62%**
0.26%
0.07%
1.75
0.61
0.21
Panel 2 Credit Rating Upgrades CRSP index ( -375, -60) estimation period
(0,+1)
53%
0.07%
-0.07%
0.60
0.22
-0.32
(0,+2)
53%
0.29%
0.06%
0.60
0.69
0.22
(-1, 0)
55%
0.22%
0.22%
0.85
0.65
0.96
(-1,+1)
58%*
0.36%
0.19%
1.37
0.87
0.66
Panel 3 Credit Rating Upgrades Ziman REIT index (+60, +315) estimation period
(0,+1)
48%
0.04%
-0.06%
-0.30
0.20
-0.35
(0,+2)
53%
0.31%
0.22%
0.47
1.19
1.00
(-1, 0)
60%*
0.34%*
0.22%
1.49
1.58
1.21
(-1,+1)
58%
0.44%**
0.21%
1.24
1.66
0.96
Panel 4 Credit Rating Upgrades Ziman REIT index ( -375, -60) estimation period
50%
0.32%
0.16%
0.25
1.06
0.89
(0,+1)
(0,+2)
60%*
0.48%*
0.25%
1.80
1.29
1.18
(-1, 0)
65%**
0.43%*
0.31%** 2.58
1.43
1.79
(-1,+1)
53%
0.74%**
0.45%** 0.77
2.00
2.11
Notes: This table presents the market model cumulative average abnormal return for a sample of
REIT rating actions from 2000 through 2009. The CDA t-statistic is calculated based on the crosssectional standard error. Parameters for the market model are estimated using CRSP daily REIT
prices, and the estimation period associated with the event. Either the CRSP value-weighted index or
CRSP Ziman REIT value-weighted index was included as the market proxy.
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, for a two-tailed test.
23
Table 5 Market model mean abnormal return and precision weighted cumulative average abnormal
return (CAAR) for credit rating downgrade changes.
Percent of
Precision
Generalized
Window
CDA
CARs
CAAR
Weighted
Sign
Patell Z
(Days)
t-stat.
Positive
CAAR
Panel 1 Credit Rating Downgrades CRSP index (+60, +315) estimation period
(0,+1)
48%
-0.56%
-0.39%
-0.04
-0.99
-1.10
(0,+2)
52%
1.08%*
0.53%
0.53
1.55
1.21
(-1, 0)
50%
-0.65%
-0.44%
0.25
-1.14
-1.23
(-1,+1)
42%
-0.40%
-0.44%
-0.91
-0.57
-1.01
Panel 2 Credit Rating Downgrades CRSP index ( -375, -60) estimation period
(0,+1)
46%
-1.31%***
-1.53%*** -0.47
-5.96
-4.43
(0,+2)
52%
0.18%
0.17%
0.40
0.51
0.51
(-1, 0)
40%*
-1.39%***
-1.64%*** -1.33
-6.38
-4.70
(-1,+1)
40%*
-1.60%***
-2.03%*** -1.33
-6.47
-4.43
Panel 3 Credit Rating Downgrades Ziman REIT index (+60, +315) estimation period
(0,+1)
42%
-1.13%**
-0.72%*** -0.96
-2.22
-2.58
(0,+2)
48%
-0.49%
-0.41%
-0.10
-0.79
-1.22
(-1, 0)
40%
-0.69%*
-0.57%**
-1.25
-1.35
-2.04
(-1,+1)
44%
-0.66%
-0.55%*
-0.67
-1.06
-1.61
Panel 4 Credit Rating Downgrades Ziman REIT index ( -375, -60) estimation period
(0,+1)
48%
-1.57%***
-1.82%*** -0.25
-6.26
-9.35
(0,+2)
44%
-0.96%***
-0.79%*** -0.82
-3.12
-3.32
(-1, 0)
40%*
-1.13%***
-1.33%*** -1.40
-4.49
-6.83
(-1,+1)
44%
-1.27%***
-1.54%*** -0.82
-4.14
-6.48
Notes: This table presents the market model cumulative average abnormal return for a sample of REIT
rating actions from 2000 through 2009. The CDA t-statistic is calculated based on the cross-sectional
standard error. Parameters for the market model are estimated using CRSP daily REIT prices, and the
estimation period associated with the event. Either the CRSP value-weighted index or CRSP Ziman
REIT value-weighted index was included as the market proxy.
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, for a two-tailed test.
24
Table 6 Fama-French with momentum 4-Factor Calendar-Time Portfolio Regressions prior to
REIT credit rating changes.
Prior 3 month
returns
Prior 6 month
returns
Prior 12 month
returns
Panel 1a Returns Prior to Credit Rating Upgrades CRSP index
Calendar-Time (OLS)
0.013**
0.013**
0.009*
(1.96)
(2.17)
(1.58)
Calendar-Time (WLS)
0.009*
0.011**
0.009**
(1.56)
(2.27)
(1.77)
Panel 1b Returns Prior to Credit Rating Upgrades Ziman REIT index
Calendar-Time (OLS)
0.011**
0.013***
0.010**
(2.18)
(2.87)
(2.13)
Calendar-Time (WLS)
0.009**
0.011***
0.008**
(2.20)
(3.01)
(2.20)
Panel 2a Returns Prior to Credit Rating Downgrades CRSP index
Calendar-Time (OLS)
-0.006
-0.008
-0.010*
(-0.55)
(-0.85)
(-1.39)
Calendar-Time (WLS)
-0.014
-0.021**
-0.017**
(-0.92)
(-1.71)
(-1.90)
Panel 2b Returns Prior to Credit Rating Downgrades Ziman REIT index
Calendar-Time (OLS)
-0.002
-0.006
-0.007*
(-0.29)
(-0.87)
(-1.47)
Calendar-Time (WLS)
0.002
-0.007
-0.010**
(0.24)
(-0.91)
(-1.75)
Notes: This table presents the performance of REITs preceding rating changes by estimating a
weighted least squares model (WLS), where each trading month is weighted by the number of
observations in the calendar-time portfolio corresponding to the time period and an equally
weighted OLS model. The models are calculated controlling for the Fama-French (1993) risk
factors and momentum using the CRSP value weighted market index return and the CRSP Ziman
REIT value-weighted index return (Rmt) ,small-minus-big market capitalization factor SMB, highminus-low market-to-book ratio factor HML, and momentum UMB.
The excess returns
for portfolio p are calculated as:
(
)
The excess returns are presented in the table with the associated test-statistic in parenthesis.
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, for a two-tailed test.
25
Table 7 Fama-French with momentum 4-Factor Calendar-Time Portfolio Regressions following
REIT credit rating changes.
6-month returns
1-year returns
2-year returns
Panel 1a Upgrades CRSP index
Calendar-Time (OLS)
0.006*
0.009**
0.007**
(1.36)
(1.91)
(1.91)
Calendar-Time (WLS)
0.006*
0.006*
0.004
(1.50)
(1.41)
(1.04)
Panel 1b Upgrades CRSP Ziman REIT index
Calendar-Time (OLS)
0.007**
0.008***
0.006***
(2.29)
(3.27)
(3.22)
Calendar-Time (WLS)
0.007***
0.006***
0.005***
(2.69)
(2.62)
(2.84)
Panel 2a Downgrades CRSP index
Calendar-Time (OLS)
-0.008
-0.003
0.002
(-0.93)
(-0.42)
(0.27)
Calendar-Time (WLS)
-0.010
0.001
0.006
(-0.98)
(0.10)
(0.86)
Panel 2b Downgrades CRSP Ziman REIT index
Calendar-Time (OLS)
-0.003
-0.003
0.000
(-0.43)
(-0.66)
(0.00)
Calendar-Time (WLS)
-0.004
-0.001
0.002
(-0.49)
(-0.13)
(0.34)
Notes: This table presents the long-run performance of REITs subsequent rating changes by
estimating a weighted least squares model (WLS), where each trading month is weighted by the
number of observations in the calendar-time portfolio corresponding to the time period and an
equally weighted OLS model. The models are calculated controlling for the Fama-French (1993)
risk factors and momentum using the CRSP value weighted market index return and the CRSP
Ziman REIT value-weighted index return (Rmt) ,small-minus-big market capitalization factor
SMB, high-minus-low market-to-book ratio factor HML, and momentum UMB.. The excess
returns for portfolio p are calculated as:
(
)
The excess returns are presented in the table with the associated test-statistic in parenthesis.
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, for a two-tailed test.
26
Table 8 Descriptive statistics of continuous variables in CAR regression analysis.
Summary Statistics
(1999 - 2010)
Continuous Variable
CAR
lnSIZE
DebtR
NIR
ExpR
DivPS
INST
ANALY
-0.005
21.217
0.582
0.016
0.036
0.416
0.646
8.528
Panel A: Upgrades
Mean
Standard Deviation
0.035
1.220
0.169
0.083
0.036
0.285
0.291
5.606
Minimum
-0.141
17.785
0.006
-0.091
0.007
0.000
0.003
0.000
Maximum
0.091
23.833
0.938
0.839
0.222
2.080
1.000
19
-0.016
20.782
0.651
0.004
0.039
0.397
0.604
7.958
Panel B: Downgrades
Mean
Standard Deviation
0.047
1.147
0.130
0.021
0.041
0.340
0.298
5.149
Minimum
-0.141
17.785
0.342
-0.047
0.007
0.000
0.003
0.000
Maximum
0.091
23.003
0.938
0.099
0.222
2.080
1.000
18
21.217
0.582
0.016
0.036
0.416
0.646
8.528
Panel C: Upgrades and Downgrades
Mean
-0.005
Standard Deviation
0.035
1.220
0.169
0.083
0.036
0.285
0.291
5.606
Minimum
-0.141
17.785
0.006
-0.091
0.007
0.000
0.003
0
Maximum
0.091
23.833
0.938
0.839
0.222
2.080
1.000
19
Panel D: Correlation Among Continuous Variable
lnSIZE
1
DebtR
0.075
1
NIR
-0.108
-0.349
1
ExpR
-0.075
-0.044
0.311
1
DivPS
0.319
-0.012
-0.074
-0.280
1
INST
0.404
0.326
-0.084
-0.121
0.231
1
ANALY
0.600
0.178
-0.098
-0.056
0.307
0.582
1
Notes: CAR is two-day (days 0, +1) Cumulative Abnormal Returns (CARs) for an equally
balanced portfolio around the event announcement; lnSIZE is the natural logarithm of the firm’s
total assets at the end of the last quarter prior to the announcement; DebtR is the debt ratio of the
total debt divided by total assets of the firm at the end of the last quarter prior to the
announcement; NIR is the firm’s net income divided by the total asset at the end of the last
quarter prior to the announcement; ExpR is the operating expense divided by the total asset at the
end of the last quarter prior to the announcement; DivPS is the dividends per share; INST is the
number of shares owned by institutional investors divided by the total number of outstanding
shares at the end of the last quarter prior to the announcement; ANALY is the number of analysts
to forecasts of FFO for REITs at the end of the last quarter prior to the announcement.
27
Table 9 Regressions to examine residual CARs on possible explanatory variables.
Up & Dn (Abs. CAR)
Coefficient
Constant
0.098
Upgrades
Downgrades
t-stat
Coefficient
t-stat
Coefficient
*
(1.82)
-0.007
(-0.10)
0.159
(1.10)
*
(-0.61)
(-1.36)
(-1.46)
(-1.00)
(-0.13)
(-1.69)
0.001
0.001
0.000
0.043
-0.013
(0.31)
(0.08)
(-0.01)
(0.47)
(-1.08)
-0.008
0.037
0.327
-0.198
0.018
(-1.19)
(0.70)
(0.81)
(-1.13)
(0.83)
-0.009
-0.001
(-0.82)
(-0.96)
-0.065
0.002
Downgrades
lnSIZE
DebtR
NIR
ExpR
DivPS
-0.013
-0.004
-0.031
-0.036
-0.015
-0.016
INST
ANALY
IntDnDebtR
IntDnExpR
0.035
-0.001
0.052
0.068
Observation
108
60
48
0.220
4.015
0.025
1.215
0.051
1.360
(0.00)
(0.31)
(0.25)
Adjusted R
F:
2
***
(3.14)
(-1.22)
(1.52)
(0.45)
**
t-stat
(-2.20)
(0.89)
Notes: CAR is two-day (days 0, +1) Cumulative Abnormal Returns (CARs) for an equally balanced
portfolio around the event announcement; Downgrades is an indicator variable equal to 1 if an event is a
downgrade and 0 otherwise; lnSIZE is the natural logarithm of the firm’s total assets at the end of the last
quarter prior to the announcement; DebtR is the debt ratio of the total debt divided by total assets of the
firm at the end of the last quarter prior to the announcement; NIR is the firm’s net income divided by the
total asset at the end of the last quarter prior to the announcement; ExpR is the operating expense divided
by the total asset at the end of the last quarter prior to the announcement; DivPS is the dividends per share;
INST is the number of shares owned by institutional investors divided by the total number of outstanding
shares at the end of the last quarter prior to the announcement; ANALY is the number of analysts to
forecasts of FFO for REITs at the end of the last quarter prior to the announcement; IntDnDebtR is the
interaction between Downgrades and DebtR; IntDnExpR is the interaction between Downgrades and
ExpR.
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, for a two-tailed test.
28
Table 10 Market model, log-transformed Value-Weighted Volume Index for credit rating changes.
% Cumulative
Generalized
Precision
Window
Abnormal
Mean Cumulative
Sign Z Test CDA
Patell
Weighted
(Days)
Volume
Abnormal Volume
t stat.
Z
CAAV
Positive
Panel 1a Upgrades (+60, +315) estimation period
43%
-12.46%
-13.55%*
-0.55
-1.09
-1.65
(0,+1)
38%*
-22.30%*
-23.15%**
-1.33
-1.59
-2.29
(0,+2)
37%*
-13.60%
-13.11%*
-1.59
-1.19
-1.59
(-1, 0)
42%
-12.81%
-14.21%*
-0.81
-0.91
-1.41
(-1,+1)
Panel 1b Upgrades (-375, -60) estimation period
(0,+1)
53%
19.16%**
24.60%***
0.95
1.85
2.59
(0,+2)
58%**
25.63%**
33.55%***
1.73
2.02
2.89
(-1, 0)
55%
18.35%**
25.76%***
1.21
1.77
2.72
(-1,+1)
60%**
34.89%***
43.69%***
1.99
2.75
3.76
Panel 2a Downgrades (+60, +315) estimation period
(0,+1)
63%**
43.76%***
49.54%***
2.18
3.74
5.05
(0,+2)
56%
58.18%***
65.40%***
1.31
4.06
5.45
(-1, 0)
58%
37.64%***
44.21%***
1.60
3.22
4.51
(-1,+1)
56%
57.27%***
65.91%***
1.31
4.00
5.49
Panel 2b Downgrades(-375, -60) estimation period
83%***
109.62%***
109.18%***
5.04
7.36
10.65
(0,+1)
81%***
154.78%***
155.05%***
4.75
8.49
12.34
(0,+2)
81%***
100.68%***
101.02%***
4.75
6.76
9.87
(-1, 0)
81%***
154.35%***
154.42%***
4.75
8.46
12.31
(-1,+1)
Notes: This table presents the market model cumulative average abnormal trading volume for a sample of
REIT rating actions from Nov. 2000 through Dec. 2009. Parameters for the market model are estimated
using the CRSP value-weighted index for the estimation period associated with the event.
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, for a two-tailed test.
29
Figure 1 Number of credit rating changes (e.g., upgrades and downgrades) of real estate firms (2000 –
2009). Source: SNL
Rating Changes by Year
25
20
15
10
5
0
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Long-Term Upgrades
Senior Maturity Unsecured Upgrades
Long-Term Downgrades
Senior Maturity Unsecured Downgrades
30
References
Anglin, P., R. Edelstein, Y. Gao, D. Tsang, How Does Corporate Governance Affect the Quality
of Investor Information? The Curious Case of REITs, Journal of Real Estate Research, 2011, 33,
1-23.
Beaver, W., The Information Content of Annual Earnings Announcements, Journal of
Accounting Research, 1968, 24, 40 – 56.
Bamber, L. and Y. Cheon, Differential Price and Volume Reactions to Accounting Earnings
Announcements, Accounting Review, 1995, 70, 417 – 441.
Bauer, R., P. Eichholtz, and N. Kok, Corporate Governance and Performance: The REIT Effect,
Real Estate Economics, 2010, 38, 1-29.
Brown, S. J. and J. B. Warner, Measuring Security Price Performance, Journal of Financial
Economics, 1980, 8, 205 – 258.
Brown, S. J. and J. B. Warner, Using daily stock returns: The Case of event studies, Journal of
Financial Economics, 1985, 14, 3 – 31.
Campbell, C.J., C. E. Wasley, Measuring Security Price Performance Using Daily NASDAQ
Returns, Journal of Financial Economics, 1993, 33, 73 – 92.
31
Campbell, R., C. Ghosh and C. F. Sirmans, The information content of method of payment in
mergers: evidence from Real Estate Investment Trusts (REITs), Real Estate Economics 2001, 29,
361 – 387.
Campbell, R. D., M. Petrova and C. F. Sirmans, Value Creation in REIT Property Sell-Offs, Real
Estate Economics, 2006, 34, 329 – 342.
Campbell, R., C. Dodd, M. Hill, and W. Kelly, Determinants of REITs Credit Ratings, Working
Paper, Mississippi State University, Troy University, University of Mississippi, and University of
Southern Mississippi. – Presented at 2012 ARES Meeting.
Carhart, M., On Persistence in Mutual Fund Performance, Journal of Finance, 1997, 25, 57 – 82.
Chandy, P R, L. P. Hsueh and Y. A. Liu, Effects of preferred stock re-rating on common stock
prices: Further evidence, The Financial Review, 1993, 28, 449 – 467.
Chung, K. H. and H. Jo, The Impact of Security Analysts’ Monitoring and Marketing Functions
on the market Value of Firms, Journal of Financial and Quantitative Analysis, 1996, 31, 493 –
512.
Cowan, A. Nonparametric Event Study Tests, Review of Quantitative Finance and Accounting,
1992, 2, 343 - 358.
32
Damodaran, A., K. John and C. H. Liu, The determinants of organizational form changes:
evidence and implications from real estate, Journal of Financial Economics, 1997, 45, 169 – 192.
Dichev, I. D., and J. D. Piotroski, The long-Run Stock Returns Following Bond Ratings Changes,
Journal of Finance, 2001, 56, 173 – 203.
Ederington, L. H., J. C. Goh, Bond Rating Agencies and Stock Analysts: Who Knows What
When? Journal of Financial and Quantitative Analysis, 1998, 33, 569 – 585.
Edwards, W., Conservatism in human information processing, In B. Kleinmutz, (ed.), Formal
Representation of Human Judgment, Ney York: John Wiley, 1968.
Elayan, A., T. Meyer, J. Li, Creditworthiness or Management Signal? An Empirical
Investigation of Loan Commitments Obtained by Non-Taxable Firms, Journal of Real Estate
Finance and Economics, 2004, 28, 59-79.
Fama, E.F., K. R. French, Common risk factors in the returns on stocks and bonds, Journal of
Financial Ecomomics,1993, 33, 3 – 56.
Feng, Z., C. Ghosh, F. He and C. F. Sirmans, Institutional Monitoring and REIT CEO
Compensation, Journal of Real Estate Finance and Economics, 2010, 40, 446 – 479.
33
Gentry, W. M., D. Kemsley and C. J. Meyer, Dividend taxes and share prices: evidence from real
estate investment trusts, Journal of Finance, 2003, 58: 261 – 282.
Ghosh, C. and C. F. Sirmans, Board independence, ownership structure and performance in real
estate investment trusts, Journal of Real Estate Finance and Economics, 2003, 26, 287 – 318.
Ghosh, Chinmoy, Nag and Sirmans, Financing Choice by Equity REITs in the 1990s, Real
Estate Finance, 1997, 14, 41 ‐ 50.
Graham, J., C. Harvey; and S. Rajgopal, The Economic Implications of Corporate Financial
Reporting, Journal of Accounting & Economics, 2005, 40, 3–73.
Griffin, P. and A. Sanvicente, Common stock returns and rating changes: A methodological
comparison, Journal of Finance, 1982, 37, 103 – 119.
Griffin, D. and A. Tversky, The weighing of evidence and the determinants of confidence,
Cognitive Psychology, 1992, 24, 411 – 35.
Hand, J. R. M., R. W. Holthausen and W. Leftwich, The Effect of Bond Rating Agency
Announcement on Bond and Stock Prices, Journal of Finance, 1992, 47, 733 – 752.
Hardin, W. and M. Hill, Credit Line Availability and Utilization in REITs, Journal of Real
Estate Research, 2011, 33, 507-530.
34
Hardin, W. and Z. Wu, Banking Relationships and REIT Capital Structure, Real Estate
Economics, 2010, 38, 257-284.
Hardin, W. and M. Hill, REIT Dividend Determinants: Excess Dividends and Capital Markets,
Real Estate Economics, 2008, 36, 349-369.
Hartzell, J. C., J. G. Kallberg and C. H Liu, The role of underlying real asset market in REIT
IPOs, Real Estate Economics, 2005, 33, 27 – 50.
Hartzell, J. C., J. G. Kallberg and C. H Liu, The role of corporate governance in initial public
offerings: evidence from real estate investment trusts, Journal of Law and Economics, 2008, 51,
539 – 562.
Hartzell, J. C. and L. T. Starks, Institutional investors and executive compensation, The Journal
of Finance, 2003, 58, 2351 – 2374.
Hartzell, J. C., L. Sun and S. Titman, The Effect of Corporate Governance on Investment:
Evidence from Real Estate Investment Trusts, Real Estate Economics, 2006, 34, 343 – 376.
Higgins, E.J., R.L. Ott, R.A. Van Ness, The Information Content of the 1999 Announcement of
Funds from Operations Changes for Real Estate Investment Trusts, Journal of Real Estate
Research, 2006, 28, 241-256.
35
Highfield, M., K. Roskelley, and F. Zhao, The Determinants of the Debt Maturity Decision for
Real Estate Investment Trusts, Journal of Real Estate Research, 2007, 29, 173-199.
Holthausen, R. W. and R. W. Leftwich, The Effect of Bond Rating Changes on Common Stock
Prices, Journal of Financial Economics, 1986, 17, 57 – 89.
Ikenberry, D. L., J. Lakonishok, and T. Vermaelen, Market under reaction to open market share
repurchases, Journal of Financial Economics, 1995, 39, 181-208.
Jensen, M. C., Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers, American
Economic Review, 1986, 76, 323 – 330.
Jorion, P., G. Liu and C. Shi, Informational effects of regulation FD: evidence from rating
agencies, Journal of Financial Economics, 2005, 76, 309 – 330.
Kim, O. and R. Verrecchia, Trading volume and price reactions to public announcements,
Journal of Accounting Research, 1991, 29, 302 – 321.
Kothari, S. P., Shu, S. and P.D. Wysocki, Do Managers Withhold Bad News? Journal of
Accounting Research, 2009, 47, 241 – 276.
Loughran, T., and J. Ritter, The new issues puzzle, Journal of Finance, 1995, 50, 23 - 51.
36
Ooi, J., S. Ong and L. Li, An Analysis of the Financing Decisions of REITs: The Role of Market
Timing and Target Leverage, The Journal of Real Estate Finance and Economics, 2010, 40, 130
– 160.
Patell, J., Corporate Forecasts of Earnings Per Share and Stock Price Behavior: Empirical Test,
Journal of Accounting Research, 1976, 14, 246 – 276.
Shleifer, A. and R. W. Vishny, A Survey of Corporate Governance, Journal of Finance, 1997, 52,
737 – 783.
Spiess, K. and J.A. Afleck-Grave, Underperformance in Long-Run Stock Returns
Following Seasoned Equity Offerings, Journal of Financial Economics 1995, 38, 243-267.
Sufi, A., The real effects of debt certification: evidence from the introduction of bank loan
ratings, Review of Financial Studies, 2009, 22, 1659 – 1691.
Tang, T., Information asymmetry and firms’ credit market access: Evidence from Moody’s credit
rating format refinement, Journal of Financial Economics, 2009, 93, 325 – 351.
37
Acknowledgements: We would like to thank participants at the 2010 ARES and 2010 mid-year
AREUEA meetings, and the journal reviewers for valuable comments.
38
1
See, Holthausen and Leftwich (1986), Hand, Holthausen and Leftwich (1992), Dichev and Piotroski (2001),
Chandy, Hsueh and Liu (2005), Sufi (2009), and Tang (2009), for example.
2
We began with 272 public U.S. real estate company rating changes (upgrades/downgrades) and removed real estate
companies not having the REIT corporate structure (i.e. real estate operating companies), leaving 241 REIT rating
changes.
3
Extending windows beyond the event date is to ensure that the window includes the press-release of the
announcement date. We find almost all of these announcements are detected in Lexis Nexis, usually appearing on
the listed announcement date or the subsequent day.
4
We recognize the timing of the events might not allow for 315 trading days in all estimation windows, therefore we
establish 50 days as a minimum criteria for firm events to be included in the analysis.
5
Holthausen and Leftwich (1986) suggest results over short intervals such as the windows around an announcement
date are not sensitive to the specification method.
6
See, for example, Loughran and Ritter (1995), Spiess and Affleck-Graves (1995) and Ikenberry, Lakonishok, and
Vermaelen (1995).
39