Health Reform Impact on Healthcare Equity Real Estate
Investment Trusts
Opportunities and Threats of the of the Patient Protection and
Affordable Care Act for US Healthcare Investments
Master Thesis
UNIVERSITY of AMSTERDAM
Amsterdam Business School
Faculty of Business and Economics
MSc. Finance – Real Estate Finance
Supervised by Dr. Gianluca Marcato
July 1
st, 2017
submitted by
Jonathan Buderer
Student ID: 11373326
Statement of Originality
This master thesis is the work of Jonathan Buderer. Hereby, I declare to take full responsibility for the contents of this document. I hereby confirm that I have written the accompanying thesis by myself, without contributions from any sources other than those cited in the text. This applies also to all figures and tables included in the thesis.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Empirical findings of existing literature showed that health reforms have an impact on the prices of healthcare stocks. Despite the increased importance, so far no study extended this body of literature by investigating on the reform impact on healthcare Equity Real Estate Investment Trusts (EREITs). President Trump’s campaign promise to repeal and replace PPACA raises the question to what extent the (healthcare) EREIT sector will be affected by such plans. This study provides insights on sector impacts, as well on the relationship between operator (healthcare stock) and the underlying real estate asset (EREIT) by means of an event study model. The findings reveal high sensitivity of the entire EREIT sector, and significant impacts on the EREIT subsectors Healthcare, Industrial Office, Retail, Lodging and Residential in the event of health reform. Obtained coefficients suggest that the reform had a negative impact on the EREIT sector. The comparison of price movements of HC operators and HC EREITs of the same industry reveals inconsistency for the majority of industries identified. By defining and investigating four separate sub-vents of PPACA, the study reveals that the market is inconsistent in assessing the health reform between the sub-events. This holds for both the operator and REIT sector. Assuming a repeal of PPACA will reverse the impact observed on its introduction, investors might benefit from President Trump’s decision to repeal PPACA. However, the challenge lies within picking the right subsectors and the right sub-event as the impacts differ between both.
Table of Contents
List of Figures ... V List of Tables ... VI List of Abbreviations ... VII
1. Introduction ... 1
2. Background Information and Related Literature ... 4
2.1 The Patient Protection and Affordable Care Act 2010 ... 4 2.2 The Impact of Health Reforms ... 4 2.3 Hypothesis Development ... 7 3. Methodology ... 10 3.1 Three Steps of an Event Study ... 10 3.2 Normal and Abnormal Returns ... 11 3.3 Event Study Specifications and Interpretation ... 15 4. Data ... 18 4. Results ... 20 4.1 Impact of Health Reform on HC Operators and REITs ... 20 4.2 Impact of Health Reform on REIT subsectors ... 23 4.3 HC Operators and HC REITs in comparison ... 26 5. Conclusion ... 33 References ... 37 Appendix ... 40
List of Figures
Figure 1 Popilation Over 65 Years Old ... 1 Figure 2 Proportions of Total Market Capitalitzaion, by Asseet Type ... 2 Figure 3 Time Line around the Event ... 12
List of Tables
Table 1 Sub-Events of PPACA and Coefficient Interpretation ... 16
Table 2-A Subsector Composition EREITs ... 18
Table 2-B Subsector Composition of HC Operators ... 19
Table 3 CAARs of HC Operators and EREITs ... 21
Table 4 CAARs of REIT Subsectors during Short Event Window ... 24
Table 5 CAARs of 14 HC Subgroups (Operators and EREITs) during Short Event Window ... 27
Table 6 CARs of 10 HC EREITs on Event 3 ... 30
Table 4-B CAARs of EREIT Subsectors during Extended Event Window ... 40
Table 5-B CAARs of 14 HC Subsectors (Operators and EREITS during Extended Event Window ... 41
List of Abbreviations
AR Abnormal Return
BBA Balanced Budget Act
BIPA Benefits Improvement and Protection Act
CAAR Cumulative Average Abnormal Return
CAR Cumulative Abnormal Return
EREIT Equity Real Estate Investment Trust
HC Healthcare
HIPAA Health Insurance Portability and Accountability Act
MMA Medicare Prescription Drug, Improvement and Modernization Program
PPACA Protection Patient Protection and Affordable Care Act
1. Introduction
Demographic change is one of the mega trends of the 21st century, with significant
implications on the real estate sector. In 2010, the U.S. population of seniors aged 65 and older comprised 13.1 percent. By 2060, this age group will account for 23.6 percent (Figure 1). In terms of absolute numbers, that implies more than doubling from 40.1 million to 98.2 million people (NGKF, 2016). Given that fact the frequency of doctor visits increases with age, demand of people in need of medical care services is expected to increase significantly in the future. CBRE (2016) experts point out that due to this high demand, as well as limited asset supply in traditional commercial real estate, this will lead to an increasing interest of international investors in the healthcare (HC) real estate sector. Increasing transaction volumes and promising risk-adjusted returns already attract investors to benefit form such developments.
Figure 1
Population Over 65 Years Old
Source: NGKF, 2016, U.S. Census Bureau, 2017
Equity Real Estate Investment Trusts (EREITs) specialized in health assets offer an easy accessible investment vehicle for those investors. Specialized EREITs own and manage a variety of HC related properties such as senior housing, hospitals, medical office buildings and skilled nursing facilities (NAREIT, 2017). In December 1999, 9 HC EREITs, accounting
for 4.6 billion US dollar, were included in the Equity REIT Index. Today, the number of healthcare EREITs included in the FTSE NAREIT US Real Estate Index Series, which covers all US REITs and publicly-traded real estate companies, increased to 20 with a total capitalization of more than 103 billion US dollar (NAREIT, 2000; FTSE, 2017). With a capitalization weight of 10.83 percent, HC REITs account for the second largest REIT subsector. Only the residential REIT sector, accounting 11.79 percent, is larger in terms of capitalization. This ever-increasing growth trend makes HC EREITs a relevant field for academic research. The relative growth to other subsectors is depicted in Figure 2. From 2000, no other subsector revealed such growth as the HC REIT sector (Feng et al., 2011).
Figure 2
Proportions of Total Market Capitalization, by Asset Type
Source: Feng et al., 2011
Despite the promising outlook, the past has shown that the development of the healthcare sector strongly relies on political reforms that lay down the economic foundation of this sector. The introduction of health reform Patient Protection and Affordable Care Act (PPACA) in 2010 represents an example of the recent past that influenced the returns of healthcare related companies significantly (Al-Ississ & Miller, 2013, Dong, 2014). In terms of returns, this event divided the HC sector into winners and losers. While the market seems to have adjusted to these changes, investors should expect further shocks to HC returns which relate to the recent surprise election of US President Trump in November 2016, who promised to “simultaneously repeal and replace” PPACA (Washington Post, 2017). The combination of
President Trump’s campaign promise, the growth of REITs as an attractive investment vehicle and the increasing demand of HC assets creates relevance to investigate on how the EREIT market is impacted in the event of the health reform. Understanding the implications of such a repeal on the HC real estate market will provide valuable insights for investors and policy makers alike. Furthermore, the relationship between HC operators (tenants) and HC EREITs will be in focus. So far, literature on the economic impact covers only the HC operator sector and leaves the EREIT sector disregarded. Moreover, existing literature on HC operators reveal ambiguous results: Al-Issis and Miller (2013) find that PPACA has a harmful impact on the sector, whereas Dong (2014) claims the opposite. Such ambiguity raises the question if sectors can be positioned as winners or looser of a reform at all, or if the impact may change during the progress of a reform. Looking at both sectors may reveal insights on which sector can provide more stability in the event of health reform. Summarizing, this thesis aims to provide answers to the following research questions:
1) How does the U.S. EREIT market assess the PPACA? 2) Is the market able to differentiate between EREITs?
3) Does the health reform impose a risk or a chance for EREITs and HC EREITs? 4) Is there consistency between HC EREITs and HC operators of the same industry in
terms of return movement?
This thesis applies the event study methodology to investigate on these questions. This measure allows using the prices of securities (returns) to measure the changes in value due to an event (Fama, 1991, MacKinlay, 1997). Based on the ambiguous findings on the HC operators, the event study will be conducted on both samples of HC operators as well as EREITs.
The thesis is structured as follows: Section 2 provides background information on the PPACA, as well as related literature of the study field. The section ends literature findings on which hypotheses are based on. Section 3 presents the methodology used to test the hypotheses. Further, model specification and interpretation are commented. Section 4 explains data sources and an presents the data used for the event study. In section 5 the main results for EREITs, HC EREITs and HC operators are presented, interpreted and compared to academic literature. The analysis will start form a broad sector level and will narrow down to individual firm level. Section 5 concludes the main results and mentions limitations of this work.
2. Background Information and Related Literature
This section will start by providing background information of the major goals of the PPACA and the scope of the reform provisions for the American society. Next, related literature will be presented. Based on literature findings the hypotheses will be formulated.
2.1 The Patient Protection and Affordable Care Act 2010
The PPACA, commonly referred to as Obamacare, represents one of the most significant legal interventions in the U.S. HC system since Medicare was enacted in 1965 (Huntington et. al, 2011, Reuters, 2012). After passing the House of Representatives on October 10, 2009, and being passed and agreed by the Senate on December 24, 2009 and March 22, 2010, respectively, the bill became law by President Obama’s signature on March 23, 2010 (Congress.gov, 2017). Achieving near-universal coverage through share responsibility among government, individuals and employers represented the central aims of the bill. Other major purposes of the act were the improvement of health insurance coverage in terms of quality and affordability, as well as increasing the efficiency of health insurance coverage by the reduction of wasteful spendings (Rosenbaum, 2011). The variety of measures necessary to achieve such aims makes the PPACA an extremely complex legislation. When fully implemented, the act should establish basic legal protections that until then have been absent: guarantee access to affordable health insurance coverage from birth through retirement with the intention to decrease the number of uninsured Americans by more than half. In numbers, the act should increase the insurance coverage to about 94%, which would reduce the uninsured Americans by 31 million.
The bill came with increased financing requirements which are achieved by direct tax penalties if individuals are not covered by a health plan, provided either by the employer or by individual purchase, indirect health penalties for individual plans and indirect expenses for employer-based plans with low coverage levels and/or diverted wages, “capped flexible spending accounts (FSAs) and health spending accounts (HSAs), and indirect expenses for all with increased costs for drugs and medical devices in order to cover new fees (Huntington et al., 2011).
2.2 The Impact of Health Reforms
A wide variety of studies offer non-empirical insights about the major health reforms and the underlying provisions implemented in the U.S. during the modern REIT era. Atchinson and
Fox (1997), Nichols and Blumberg (1996), Nunn and McGuire (2005) and Dwyer et al. (1996) provide detailed insights of the provisions of the Health Insurance Portability and Accountability Act (HIPAA) of 1996. Bazzoli and Lindrooth (2005), Philips et al. (2004) and Heiber-White (1997) examine the provisions of the Balanced Budget Act (BBA) one year later, 1997. Gold and Achman (2001) and Chaikind et al. (2001) focus on the health reforms introduced in 2000, the Benefits Improvement and Protection Act (BIPA), whereas Kirchhoff (2016), O’Sullivan et al. (2004) and Channick (2006) summarize major implications of the Medicare Prescription Drug, Improvement and Modernization Program (MMA) of 2003. Rosenbaum et al. (2011) examine the most recent health reform PPACA of 2010.
Whereas plenty of literature provides qualitative analyses of the provisions, another body of literature applies empirical approaches to assess the economic impact of health reforms. The common approach of doing so is by means of an event study.
Khansa et al. (2012) examine the impact of the HIPAA provisions in 1996 on the market value of HC firms, information security (IS) and other information technology (IT) firms. By analysing the individual provisions and referencing related literature, they derive three individual hypotheses for each firm sector. These hypotheses state the impact (positive or negative) expected for each sector in terms of stock price movement. A positive impact is associated with an increasing stock price, a negative impact with a declining stock price around the announcement date. They confirm the hypotheses as the find a detrimental impact on HC stocks and a positive impact of IT and IS firms.
Friedman (2009) investigates on the economic impact of the passage of MMA of 2003 on the profit of pharmaceutical firms using stock market returns. He finds that the Act significantly increased the profit for prescription drug producers with high Medicare demand. In a first step, he obtains a measure of the profitability of each brand-name drug from a stock market event study around FDA approval of that drug. Next, extending the event study methodology, he estimates the impact of the Act on profits by comparing the profitability of drugs with high and low Medicare market shares, before and after the passage of the law. He finds that the Act significantly increased the profit for prescription drug producers with high Medicare demand.
Dong (2014) studies the economic impact of PPACA (2010) on listed HC firms using financial statement data obtained from Compustat and stock return data from CRSP. Similar as Khansa et al. (2012), the objective is to evaluate which companies could benefit from the legislation and which could not. To do so, he subdivides the HC sector into 12 sub-categories and applies the event study methodology. He further runs cross-sectional regressions to
control for the heterogeneity among firms and industries by using variables that reflect the financial and operational characteristics on the individual firm level and the industry level. Obtained positive cumulative abnormal returns reveal that operators of hospitals, medical clinics, and spatially outpatient facility operators, reacted positively to the health reform. Moreover, interpreting the coefficients of the industry dummy variable, they find that the magnitude of the abnormal return is relatively larger and more positive for hospitals and specialty outpatient facility operators, but more negative for medical instrument manufacturers, wholesale drug distributers, medical laboratories, and home care providers. Finally, they conclude from their findings that investors seem to interpret the Act as being beneficial for a specific group, which indicates sub-categories where growth opportunities exist.
Similar as Dong (2014), Al-Ississ and Miller (2013) investigate the impact of the PPACA on HC stocks. In order to evaluate the market’s assessment of the impact of this health reform, they look at the surprise election of Republican Scott Brown to the US Senate in January 2010, as his victory over the Democrat candidate opponent was expected to significantly reduce the likelihood of passing the reform. By means of an event study they find that Brown’s unexpected victory was associated with abnormal returns of 2.1% in the overall HC sector and 6% for managed care firms. As the event reduced the probability of health reform passage, they conclude form the positive coefficients that the reform had a harmful impact on the overall HC sector, as well as on the subsector managed car. Moreover, they find that HC facilities such as hospitals experienced abnormal losses of 3.5%. Hence, the introduction of the heath reform was assumed to be beneficial for HC facilities as due to the high probability of a loss of Brown in the election the market’s opinion about rising prices was incorporated in the return of the REITs already.
Research as stated above provides insights of the specific reforms and their economic impact on listed equities. All studies follow a similar approach in which hypotheses are set up by first analysing the individual provisions and then deriving possible consequences for different sectors. Provisions with expected harmful implications for a specific sector are found to be harmful in terms of negative returns after the enactment of the reform; for provisions with potential positive implications, the opposite applies. Interestingly, event studies on the same event (Dong, 2014, Al-Issis & Miller, 2013) reveal ambiguous results. Many reasons, such as differences in methodology application, e.g. in defining event window, market model or sample size, could account for such differences. In this specific example however, differences may stem from the different events defined as event date. Even though
the provisions of the bill remained broadly similar between both events, there were significant differences in the bill (Al-Issis & Miller, 2013). In this context, ambiguous impacts for certain sectors are not surprising.
The main take-away besides the individual return behaviours of sectors to such exogenous shocks as represented by a health reform is that the HC operator market is significantly affected, and further, that markets (investors) are able to take account of the bills individual and capable to differentiate among firms even within the HC sector by selling and buying stocks accordingly (Al-Issis & Miller, 2013). In the context of the increasing importance of the HC EREITs and the gap in literature regarding this sector, this study aims to investigate on the impact on EREITs in general, and on HC EREITs in particular. Thereby, the key focus is on how EREITs performed in comparison to stocks instead of linking the bill’s provisions to observed return reactions. The ambiguous results of literature on HC stocks (operators) suggest that investigating on the consistency of subsectors reacting to each sub-event of a reform process may reveal more insights than the attempt to find plausible reasons for each sector’s reaction. Including four sub events to this study may help explain the controversial conclusions about the impact of PPACA. Justification to investigate on this topic requires, besides increased importance and literature gap, reasons why and how the sector is expected to be impacted.
2.3 Hypothesis Development
Several reasons support the assumption that the health reform will impact the HC EREIT sector in particular. First, the increased importance of investments in HC REITs, as presented in the introduction, suggests that investors spend more time to track and analyse their investments and hence, take actions when major decisions about reform passage is announced and enacted. Investors expecting a negative impact on their portfolio may sell shares exposed to the reform. REITs expected to benefit may attract investors to buy, respectively.
Second, the transition and maturation process of the REIT market during the 90s, which was characterized by heavy growth and increase analyst activities, gave investors the possibility to benefit from more widely distributed and more reliable information about REITs than ever before. Following such improved information flow about REITs made “REIT prices better reflect the performance of their underlying real estate assets” (Clayton & Mackinnon, 2003). As the value of the underlying asset is driven by the cash flow generated by the rent payments of tenants, the impact on the HC operator is expected to expand to the property level (Geltner et al, 2006). In the presence of revenue-based rent payments, reduced
revenues will of the operator due to a reform translating directly to the property. In line with Boudry et al. (2012), who find a strong relation between the REIT and direct real estate returns, the impact is expected also to be observable in the returns of the REIT.
Third, the common factor found by Terris and Myer (1995) and Myer and Webb (1994) driving returns of REITs and stocks within the same sector provides support as it lets assume that a close relationship between REITs and stocks also remains in times of HC reforms. Myer and Webb (1994) found a significant positive relationship between the index of retail REITs and three indices of retail stocks. A significant positive relationship could be found for two out of eight studied retail REITs. The work by Terris and Myer (1995) further investigates on this topic. For the period from 1985 to 1992, they apply a two-factor regression model to examine the relationship between returns on HC equity REITs and HC stocks. They find multiple positive contemporaneous relationships between six of the seven REITs studied and portfolios of other HC stocks. When they further specify their analyses by unbundling the portfolios by their property types, they find that four of the six REITs with multiple relationships with HC stock indices, show significant correlation between the classification of their portfolios and their SIC indices. For ‘Health Equity Properties’, which invests exclusively in equity holdings of long-term care facilities, they observe a significant positive relationship with the index composed of skilled nursing stocks. For ‘Universal Health Reality Income Trust’ they find significant relationships with the indices composing of hospital and outpatient stocks, respectively, as the REIT held his majority of equity investments in acute care, rehabilitation and psychiatric hospitals. Their findings confirm the argument of a common factor, or factors, affecting both the returns of HC equity REITs and stocks. They conclude that the homogenous nature of HC and the provision of percentage rents in HC REITs are reason for the closer link between financial success of HC facilities and HC providers than other categories of real estate (e.g. retail as found by Myer & Webb). Hence, as Al-Issis and Miller (2013) and Dong (2014) showed that the HC operators are affected by the PPACA, one can derive that EREITs should be affected as well.
As the PPACA represents the largest overhaul of the U.S. HC system with far-reaching impacts the question arises if the reform will, besides HC REITs, also impact other REIT subsectors. Several reasons support this view.
First, regarding the main reform goal of providing HC to additional 31 million, one can derive consequences for multiple sub-sectors. The impact on HC assets seems logical: e.g. hospital assets are expected to benefit through the direct link to the tenant’s performance
(percentage rents), as tenants increase revenues by limiting the amount of non-reimbursed medical treatments of uninsured (Al-Issis & Miller, 2013). However, such improved health conditions may also translate to other asset classes such as residential or retail. Based on the scope of the reform that directly affects every American citizen, either positively by receiving improved insurance coverage or natively by facing increased HC costs, it seems reasonable to expect an industry-wide impact on the short-term.
Secondly, it is the composition of REITs that support the view of the reform affecting multiple REIT subsectors. Geltner and Kluger (1998) point out that REITs rarely contain strictly one property type. Whereas for a diversified REIT this seems to be a necessary condition, it implies that also specialised REITs are also invested in other subsectors. According to this, assuming HC assets to be impacted by the health reform, a retail or residential REIT (primarily invested in on asset type) may also be impacted by the presence of HC assets in its underlying portfolio. For Young (2000), liquidity, risk aversion, and scale are possible reasons for investors and investment managers alike to hold equity REIT securities from a variety of property-type sectors. If investors are aware of such impure investment allocation of REITs that are marketed as a specialized REIT, the impact of other REITs in times of health reform seems to be justified.
Last, the study of Young (2000) provides convincing arguments that support the aforementioned view. Interested about the degree of integration within the EREIT market, he examines the correlations of rolling 60-month returns of six pairs of EREITs predominantly invested in either residential, industrial, office or retail properties. For the period of 1898 to 1998, he finds correlations between the pairs upward toward 1.0, which provides evidence that REITs primarily invested in a certain asset type (specialized REITs) became more integrated. Found integration may translate to times of health reforms, meaning that the impact on a certain sector may transfer inversely on other sectors. Hence, an impact on HC EREITs as a reaction on the health reform may cause other EREIT subsectors to be impacted simultaneously. By combining the expectations about the impact of the health reform on EREITs, and HC REEITs in particular, the following hypothesis can be derived:
H1: Health reform has a significant impact on HC EREITs and other EREIT subsectors
The findings of Terris and Myer (1995) and Myer and Webb (1994) provide also insights on how the EREIT sector reacts in comparison to the HC operator sector (in terms of price movements). The relationships found indicate that not only an impact can be expected; it
further allows to assume that the sign (positively or negatively) of HC EREITs and HC operators of the same industry (e.g. hospital, senior housing, outpatient) will be the same in the event of the health reform. That would imply that on a specific sub-event of PPACA, similar impacts (positive or negative) will be found for the same subsectors, e.g. hospital REITs and hospital operators. This assumption is strengthened by Geltner et al. (2006), Boudry et al. (2012) and Clayton & Mackinnon (2003), who confirm the linkage between the performance of operator (tenant) and the performance of the underling real estate asset. Based on these findings, the second hypothesis can be formulated as follows:
H2: HC operator subsectors and HC EREITs with similar industry exposure reveal similar
impacts
3. Methodology
This section introduces the methodology used to assess the hypotheses developed in the previous abstract. As the main interest is about the impact of health reforms on the returns of HC EREITs, event study is chosen as the appropriate methodology. Event studies in economic literature use the prices of securities (returns) to measure the changes in value due to an event (Fama, 1991, MacKinlay, 1997). In line with MacKinlay (1997), assuming rationality among investors, the effects of any health reform should be reflected immediately in the returns of EREITs. This assumption allows the researcher to construct a measure for the economic impact by using security prices of a relatively short time period. Applied for more than 30 years, the basic statistical format of event studies has not changed and is still based on the classic stock split event study of Fama et al. (1969) (Kothari and Warner, 2006). The key focus of an event study is to examine the return behaviour for a sample of firms, which are exposed to the same event (Kotari and Warner, 2006).
3.1 Three Steps of an Event Study
Goeij and De Jong (2011) identify three steps to conduct an event study:
(1) Event Definition: First of all, the event of interest needs to be defined. This study investigates on the PPACA by testing sub-events which are attributed to increased or decrease the probability of the reform being enacted as they are expected to reveal stock market reactions. The event window should include at least the event date and the day after, however, expanding this window by multiple days allows capturing price effects on a longer horizon. In
case in which it is likely that the market obtained information about the outcome of the event prior to the event, it is reasonable to assume that this information was immediately incorporated in the prices (MacKinley, 1997). Including days prior, so-called pre-event returns, allows investigating for such a case. This study applies, in line with similar studies (Al-Issis & Miller, 2013), two different event windows around each event date. The first event window will include one trading day prior and one after the event date, which accounts for 3 trading days in total. The latter is an extended event window of 12 trading days in total, including 1 trading day prior and 10 trading days after the event date. This window is long enough to allow markets to readjust in case of an initial overreaction and to account for additional information available the public during this time. Observations of the short window will be in focus in the part results. The results of the extended event window help to validate findings of the short window and hence, serve as robustness check.
(2) Model for the Normal Returns: To identify significant impacts on security returns (REITs and operators), a model is needed to calculate normal returns, which serve as a benchmark to identify abnormal returns. Even though a variety of models exist, they all suffer some degree of bias as they are “incomplete descriptions of systematic patterns in average returns” (Fama, 1998, p. 291). As a consequence one should be aware of that observed abnormal performance might be biased. In line with Dong (2014), the market model of Fama et al. (1996) is used to calculate normal returns.
(3) Abnormal Return Calculation and Analysis around the Event Date: When normal returns are calculated on the basis of the market model, obtained estimation coefficients are used to calculate abnormal returns during the event window. Abnormal return (AR), defined as the estimated error term of the normal return model can then be interpreted as a measure of the impact of the event (sub-events) on the price of the security. By aggregating ARs on the days of the event window, the impact of the overall event can be measured. For testing abnormal returns on statistical significance, a testing framework that defines the null hypothesis is applied.
3.2 Normal and Abnormal Returns
Applying the event study methodology requires defining windows (time periods) on which basis normal and ARs are calculated. The estimation window is a period before the event, which serves to estimate normal returns. In line with Al-Issis and Miller (2013), this study sets the estimation window from 300 to 30 trading days prior to the event of interest. In Figure 1, the estimation window is defined as [T1, T2]. Following the time line in Figure 1,
the next window is the event window, defined as [t1, t2]. The event window spans around the event date t=0. The event window is not included in the estimation window in order to prevent any influence of the event on the normal performance parameters (Ball & Brown, 1968, MacKinley 1997).
Figure 3
Time Line around the Event
Source: Goeij and De Jong, 2011
In line with Dong (2014), this research estimates normal returns and calculates ARs by means of the market model. As basic inputs, the model requires daily security returns Ri,t and
a market returns Rm,t. Security return Ri,t for firm i at time t is defined as:
𝑅!,! = ln Stock price !,! Stock price !,!!! Market return Rm,t is defined as:
𝑅!,! = ln
𝑃𝑟𝑖𝑐𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑛𝑑𝑒𝑥 !,! 𝑃𝑟𝑖𝑐𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑛𝑑𝑒𝑥 !,!!!
𝑅!,! = 𝛼!+ 𝛽! 𝑅!,!+ 𝜀!,!
The market model relates return of any given security to the return of the market portfolio Rm,t. αi and βi represent unknown parameters to be estimated for each firm i. This
model intends to remove the effects of economy wide factors from the return on firm i’s stock, which leaves a portion of the return attributable to the firm specific information. The latter is reflected by the error term in the equation, which holds the impact on the firm i’s stock due to the event. Hence, AR is given by:
(𝜀!,! = 𝐴𝑅!,!).
To identify whether the cross-sectional distribution of returns at the time of the event can be considered abnormal, the mean of the distribution of ARs is calculated as it is the most widely approach applied in academic literature and also the preferred model if one aims to investigate whether the event is associated with a change in security holder wealth and testing the sign (positive or negative) of impact (Kothari et al., 2007). In line with this approach, the null hypothesis is typically used to test whether the mean abnormal return at time is equal to zero. The mean abnormal return for any period t, for a sample of N firms, is given by:
𝐴𝑅! = 1
𝑁 𝑒!" !
!!!
It is likely to occur that any impact of an event (abnormal return) is not limited on the event date, but also are found around the event. In a case in which the event is partially anticipated, one would expect abnormal return behaviour prior to the event. Furthermore, to examine the speed of adjustment to new information (market efficiency) about the event taking place at the time of the event, returns past the event date should also be considered. The process of estimating such phenomena is called time-series aggregation. This study applies the cumulative average return method (CAR). This method sums up observed abnormal performance starting at time t1 through time t2:
𝐶𝐴𝑅!" = 𝐴𝑅!,!
!!
CAR test the null hypothesis whether the mean abnormal return performance is zero and reflect the security holders wealth changes around the event. Furthermore, market efficiency can be investigated by applying the model to post-event periods. Systematically nonzero abnormal returns following the event would highlight inconsistency with efficiency.
To investigate the impact on samples of firms, the average abnormal return (AAR) can be calculated across all firms i= 1, …, N:
𝐴𝐴𝑅!= 1
𝑁 𝐴𝑅!"
!
!!!
In a last step, the cumulative average abnormal return (CAAR) over the event window (t = t1, …, t2) can be computed for samples of firms by:
𝐶𝐴𝐴𝑅 = 1
𝑁 𝐴𝐴𝑅!
!!
!!!!
Finally, statistical tests are required to test whether obtained ARs are statistically different from zero at a 1, 5 or 10 percent significance level. Such framework tests the null hypothesis:
H0: E(ARit) = 0
This study applies t-tests that assume ARs to be independently and identically distributed and to follow a normal distribution with mean zero and variance σ² and AARt ∼ N(0, σ²/N). For AARs, the test statistic is defined as:
𝑇𝑆! = 𝑁 𝐴𝐴𝑅!
𝑠! ~𝑁(0, 1)
Similar tests can be applied to test the statistical significance of CAARs. Here, the null hypothesis to be tested is:
For which the corresponding t statistic is given by:
𝑇𝑆! = 𝑁 𝐶𝐴𝐴𝑅!
𝑠! ~𝑁(0, 1)
3.3 Event Study Specifications and Interpretation
In this study, securities of main interest are U.S. HC REITs. As the comparison to the corresponding operating sector of publicly traded U.S. HC operating companies (non-REITs) is necessary to evaluate hypothesis 2, ARs are calculated for both sectors. This study runs a separate event study on the HC operators instead of referring to the results of researchers because of two reasons: First, event study methodologies vary between referenced literature on health reform impact on HC operators and applied methodology of this study. Consequently, results of EREITs would be compared to results assembled with a different research design. Secondly, as literature presents ambiguous opinions on how subsectors perceived the health reform, it is not at the discretion of the researcher to decide which results to use as a reference.
Application of aforementioned methodology requires further specifications about the events of interest. As this study investigates on the PPACA, events increasing or decreasing the probability of the reform being enacted are expected to reveal stock market reactions. The studies by Al-Issis and Miller (2013) and Dong (2014) show that for a single reform different event dates can be applied. Concluding form those ambiguous results, it seems as the market interprets the impact of the reform differently on different stages of reform progress or that changing reform provisions may reveal varying impacts. The latter is in line with (Binder, 1998) as he associates an issue with the characteristic of regulatory change, as it usually occurs via a lengthy and uncertain process. To gain better understanding on this issue, four events are investigated in this study. Those are the dates on which the two bodies of the Congress, the U.S. Senate and the U.S. House of Representatives, agreeed on a bill. If the bill is agreed upon in one body of Congress, it is then brought to the other body, which also votes on the passage of the bill. In some cases on request an amendment, which then requires another approval by the other body. When both bodies agree on the bill, it is presented to the President, who can either approve the bill and sign it into law or veto the bill (USA.gov, 2017). Whereas a veto of the president is rather unlikely, any house vote represents an event of which the outcome is not predictable. Hence, those house votes represent an appropriate
event date. Regarding PPACA, 3 main house votes are applied in this study: The House Passing on October 8, 2009 (Event 1), the Senate Passing on December 24, 2009 (Event 2) and the Senate Agreement (Event 3) on March 22, 2010. Inspired by Al-Issis and Brown (2013), this study also applies the Brown election on the January 19, 2010, as a forth event (Event 3). The four events are attributed with different effects on the probability of the bill to be enacted. Events 1, 2 and 4 are expected to increase the probability as the events represent official governmental votes on which the majority of members agreed on the provisions of the bill. Event 3 represents the election of Republican Brown in the Massachusetts Special Election. As this win shifted the majority ratios in the senate in favour of the Republicans, tried to prevent the enactment of the health reform, this event is associated with a reduction in the probability of health reform passage. Table 1 presents the four events of interests and the underlying effects of event outcome on the probability of health reform enactment.
Table 1
Sub-Events of PPACA and Coefficient Interpretation
(1) (2) (3) (4)
10.8.2009 12.24.2009 1.19.2010 3.22.2010
House Passing Senate Passing Brown Election Senate Agreement
Event on reform probability Increase Increase Reduce Increase
Sign of Coefficient (ARs) Positive Positive Positive Positive
Impact Interpretation Beneficial Beneficial Harmful Beneficial
Sign of Coefficient (ARs) Negative Negative Negative Negative
Impact Interpretation Harmful Harmful Beneficial Harmful
Table 1 present 4 sub-events identified to impact the probability of the PPACA to be enacted. The first row categorizes each event into reducing or increasing the probability of the PPACA being enacted. Rows 2 and 3 describe how positive signs of the coefficient ARs (averaged and cumulated) should be interpreted. A positive sign indicates a beneficial impact on Event 1,2, and 4, and a harmful impact on Event 3. Rows 4 and 5 describe how negative signs of the coefficient should be interpreted.
As event 3 represents a reverse impact in terms of probability, it is important to note that coefficient interpretation has to take this into account. Other than on Event 1,2 and 4, on which a positive CAR or CAAR reveals over performance relative to the index, a positive coefficient on Event 3 would indicate over performance on an event that reduced probability of health reform passage. Consequently, a positive CAR or CAAR would reveal market’s assessment of the reform on the company/sector as harmful. For negative coefficients, all
events reverse in terms impact (beneficial/harmful) on the company/sector. Assuming broad similarity of the reform in terms of provisions (Al-Issis & Miller, 2013), company/sector reveals consistency in terms of impact if the event study reveals positive coefficients on Events 1, 2 and 4 and a negative on Event 3, indicating that on all 4 events the company/sector could benefit from the health reform. Vice versa, consistency is found if the event study reveals negative coefficients on Event 1, 2 and 4, and a positive on Event 3, indicating that on all 4 events the health reform has positive impact on the company/sector observed.
To estimate the normal return by means of the market model, this study uses two market indices as a reference. To investigate the impact within the equity universe, the CRSP value-weighted equity total return index, in the following named equity index, is used. This index contains daily returns, excluding all distributions, on a value-weighted market portfolio. To observe the impact of REIT subsectors relative to the overall U.S. REIT market, the CRSP Ziman REIT Value-Weighted Index Total Return, in the following named REIT index, is used. Furthermore, an estimation window to obtain normal returns, estimated by coefficient βi in the market model, has to be defined.
To test the first hypothesis, CAARs will be calculated for the entire REIT sector as well as for REIT-subsectors. Significant coefficient of the former will strengthen H1 as it indicates an impact on the entire REIT sector. For the latter, significant coefficients of multiple subsectors will validate H1 as it reveals impacts of the health reform on specific REIT subsectors. To test the second hypothesis, CAARs of HC operator subsectors are matched with the corresponding (same industry) CARs of individual HC REITs. H2 is supported if co-movements in terms of coefficient signs can be observed between both sectors. Vice versa, opposing signs weaken H2.
To test the results for robustness, all analyses are run for the short and the extended event window. Inconsistent signs between the same events of the same sector will indicate that the market may has overreacted and adjusted for this overreaction (Al-Issis and Miller, 2013). This should be taken into account when interpreting coefficients. Furthermore, the results can be confirmed if the results obtained by using two different indices (Equity and REIT index) reveal similar impacts on the firms or samples investigated. Hence, each analysis will apply both indices.
4. Data
To calculate the normal returns with the market model, equity index returns and daily security returns for the HC operators are obtained from the Compustat/CRSP database provided by Wharton Research Data Services (WRDS). REIT index data, as well as REIT and property type classifications, are obtained from the CRSP/Ziman Real Estate Database, also provided by WRDS. Additional information on asset allocation on the individual HC EREIT level, which is necessary to investigate H2, it obtained from the SNL Database. SNL database reports portfolio allocations on a six-monthly basis. Searching for key terms such as “hospitals” “senior” “care” allows categorizing the REIT by investment focus. As the main analysis of H2 focuses on event 3, data of the second half of 2009 is taken into account for subsector comparison.
Table 2-A
Subsector Composition of EREITs
CRSP REIT Type Classification Code
CRSP REIT Type Description
Mean Total Assets (in m €)
Total Assets (in m €)
Subsector Assets /
Sector Assets Number of Firms
1 Unclassified 1,959 13,711 3.4% 7 2 Diversified 4,195 41,947 10.3% 10 3 Healthcare 2,404 24,037 5.9% 10 4 Industrial / Office 4,593 114,826 28.2% 25 5 Lodging 3,276 36,041 8.8% 11 8 Residential 4,059 60,884 14.9% 15 9 Retail 3,712 100,229 24.6% 27 10 Self Storage 3,759 15,037 3.7% 4 Total 406,711 100% 109
Table 2-A presents the subsector composition of EREITs included in this study. Besides the CRSP REIT Type Classification Code, the industry description is provided. Information on subsector size is provided in columns 2 to 6 in terms of mean of total assets, total assets and the share of subsector assets of sector assets (total).
Sample selection for the EREIT sector includes the following steps. First, return information is downloaded for the time period January 2008 to June 2010, which covers the entire estimation window as well as the extended window of all four events investigated. Using the REIT type information provided by the CRSP database, non-equity REITs are excluded by keeping only firms with the REIT type classification 3 Equity REITs. Next, REITs with incomplete return information, indicating potential acquisition or bankruptcy of
the REIT during the event study period, are dropped from the sample. After running the steps, the sample of 109 U.S. equity REITs with compete return and index information (daily) is obtained. Using the REIT property classification provided by the CRSP database, the sample can be subdivided into 10 subsectors. Property type Unknown, Mortgage and Mortgage
Backed Securities are dropped for the subsector analysis as no or to little companies are
assigned to this property type label. Composition of the EREIT subsectors are provided in table 2-A. As seen in the table, the Retail subsector accounts for the largest sample in terms of absolute number of firms (27). Largest in terms of total assets is the Industrial Office sector, which account for 28.2 percent of the total assets of all 109 EREITs. The HC EREIT subsector counts 10 EREITs, which account for 5.9% percent of total assets.
Sample selection for the HC operator sector follows a similar approach. After return data is downloaded for the same time, companies with incomplete return information are dropped from the sample. HC operators are selected by keeping only firms for which the
three-digit Standard Industrial Classification-Code (SIC-Code) labels the company as an healthcare industry security. SIC definitions are obtained from the Occupational Safety and Health Administration. Left with 333 HC operators, 12 subgroups can be created on the basis of the same SIC-codes. Those subsectors serve as the reference to investigate on similar impacts between HC REITs and HC operators. Descriptive statistics of the subsectors are provided in table 2-B. As seen in the table, subsectors vary significantly in the number of firms as well as in absolute asset size. Subsectors 1 Medical Chemicals, Botanical Products & Pharmaceutical Preparations and 2 Medical Instruments & Apparatus are the largest
subsectors. Other subsectors reveal less variety in size. Table 2-B
Subsector Composition of HC Operators Operator
Subsector ID
SIC-Code Industry Description Mean Total
Assets (in m €) Total Assets (in m €) Subsector Assets / Sector Assets Number of Firms
1 283 Medical Chemicals, Botanical Products & Pharmaceutical
Preparations
6,044 888,483 41.4% 147
2 384 Medical Instruments & Apparatus 2,464 229,167 10.7% 93
3 385 Ophthalmic Goods 5,224 20,896 1.0% 4
4 512 Wholesale-Drugs, Proprietaries & Druggists' Sundries
5 591 Retail-Drug Stores and Proprietary Stores
16,544 99,264 4.6% 6
6 632 Accident/Health Insurance & Medical Service Plans
31,290 657,080 30.6% 21 7 801 Office & Clinics of Doctors of
Medicine
898 2,695 0.1% 3
8 805 Nursing & Personal Care Facilities 1,371 8,226 0.4% 6
9 806 Hospitals 3,752 41,273 1.9% 11
10 807 Medical Laboratories 1,920 17,279 0.8% 9
11 808 Home Health Care Service 6,308 50,465 2.3% 8 12 809 MISC Health, Allied Services &
Specialty Outpatient Facilities
4,531 63,434 3.0% 14
Total 2,148,221 100% 333
Table 2-B presents the subsector composition of HC operators included in this study. Subsectors are arranged based on SIC codes provided by the CRSP database. Information on subsector size is provided in columns 3 to 7 in terms of mean of total assets, total assets and the share of subsector assets of sector assets (total).
4. Results
This abstract presents the results of the event study. To provide a comprehensive structure, the presentations starts with the results on larger samples and narrows down to smaller samples and individuals firm level. First, findings on the impact of health reform on HC operators and REITs are presented. Next, results of EREIT subsectors are presented. Finally, the comparison of HC operators and HC EREITs is presented in the last part. Each part describes the main results and puts findings in the context of related literature and the hypotheses of this thesis. To improve the readability of this thesis, tables containing the results of the extended event window (robustness checks) are included in the Appendix section.
4.1 Impact of Health Reform on HC Operators and REITs
Table 3 presents the CAARs of two samples of 333 healthcare operators and 109 EREITs. To allow for comparison, CAARs for both sectors are calculated with the market model using the equity index. The results serve as a starting point to evaluate the validity of the first and
second hypothesis, that the events of the health reform cause impacts on the entire EREIT sector, and that similar impact patterns can be observed between sectors.
CAARs of HC operators are presented in the rows 1 and 2, for the short and the extended event window, respectively. In total, four significant impacts can be observed on the events of interest. For the short event window, a negative CAAR of -0.93 percent is found at the significance level of 1 percent on the Event 3, followed by a CAAR of 1.30 percent at the significance level of 1 percent on Event 4. For the extended event window CAARs of -5.44 percent and -4.89 percent, both significant at the 1 percent level, are found for Events 1 and 3, respectively. These findings of the short event window (row 1) confirm literature findings that the health reform had a significant impact on the HC operators in general and support the opinion of Dong (2014) who states that the health reform was beneficial for HC operators.
Table 3
CAARs of HC Operators and EREITs
Sector N Event
Window
(1) (2) (3) (4)
10.8.2009 12.24.2009 1.19.2010 3.22.2010 House Passing Senate Passing Brown Election Senate Agreement
Healthcare Operators 333 (-1, 1) - 0.0049 -0.0029 -0.0093*** 0.0130*** (-1.56) (-0.84) (-2.99) (3.94) Healthcare Operators 333 (-1, 10) -0.0544*** 0.0145 -0.0489*** -0.0077 (-7.85) (1.59) (-7.72) (1.02) Equity REITs 109 (-1, 1) -0.0154*** 0.0058*** 0.0254*** -0.0064 (-5.22) (2.87) (8.73) (-1.47) Equity REITs 109 (-1, 10) -0.0655*** -0.0169** 0.0433*** 0.0177** (9.88) (-2.35) (7.04) (2.55)
In this table CAARs are calculated across all companies of the REIT and stock sample, separately. The stock sample covers 333 U.S. HC publicly traded HC companies, the REIT sample 109 U.S. equity REITs. CAARs are calculated for the short and the extended event window, (-1, 1) and (-1, 10), respectively. The estimation window to estimate the normal return ranges from 300 to 40 days prior to the respective event date. T-statistics are stated in brackets. ***, ** and * indicate the statistical significance level of 1% , 5% and 10%, respectively.
The changing signs between Events 3 and 4 are consistent in terms of event probabilities: Event 3, which is attributed to reduce the probability of the health reform to be passed, reveals negative CAARs, whereas Event 4, which is attributed to increase the probability of the health reform to be passed, reveals a negative sign. However, the negative CAAR of -5.44
percent on event 1 weakens this view as the Event 4 is attributed to increase the probability of health reform passage. This view corresponds to Al-Issis and Miller’s (2013), who conclude an overall harmful impact of the reform on the sector. This finding of inconsistent impacts can be used to explain why two separate studies on the same health reform using different event dates derive ambiguous impacts. The market seems to evaluate the implications on HC operators differently at different stages of reform progress.
Consecutive rows 3 and 4 of Table 1 present the CAARs of the REIT sector. Within the short event window, three out of four events reveal significant CAARs. Significant at the 1 percent level, CAARs of -1.54 percent, 0.58 percent and 2.54 percent are found throughout events 1 to 3, respectively. Applying the extended event window, consistent findings regarding sign and significance are found for event 1 and 3. Based on consistent events 1 and 3, the health reform imposed a harmful impact on the entire REIT sector. The positive CAAR on event 2 in the short window, however, weakens this position and reveals inconsistency about how the sector perceived the reform. Even tough the sign of the CAAR on event 2 changes when applying the extended event window (-1.69 percent at 5 percent significance level), inconsistency remains with the CAAR of 1.77 percent on event 4, significant at the 5 percent level. Although summarizing the CAARs suggests that the health reform seems to impose a harmful impact on the entire REIT sector, ambiguous CAARs suggest that for the entire REIT market a unilateral impact cannot be derived.
Several key findings can be derived from Table 3: Event 3 reveals the best results in terms of significance and consistency between the windows, both for HC operators and EREITs. This is in line with the assumption about the Brown election to reflect the event for which the probability of unexpected outcome was the higher than for other reform events and implies that interpreting these results may provide more reliable insights than the other events (Al-Issis & Miller (2014).
Second, the fact that seven out of eight events caused significant impacts on the entire REIT market supports hypothesis 1 that the reform has far-reaching impacts on multiple sub-sectors. However, concluding validity of H1 would come at a too early stage of analysis, as it remains unclear if the impacts are caused by specific subsectors of EREITs or are spread randomly among all REITs. Specified literature and their findings suggest that the HC subsector contributes to the significant impact on the general EREIT sector.
Third, insights about event consistency are provided for both sectors. Observing all 4 events, both sectors show CAARs that revel contrary signs in terms of event impact (positive or negative). Compared to the operators, the REIT sector reveals even more ambiguity about
this impact (similar signs on Event 2 and 3). Testing multiple events shows that market’s opinion about the health reform and its provisions seem not to follow a consistent pattern. As this holds for both the operator sector as well the EREIT sector, commonality in terms of general reform assessment between both sectors becomes visible.
Lastly, findings of the REIT sector in the light of the study by Al-Issis and Miller (2013) should be commented. As stated, they ascribe the HC facility sector as a winner of the bill based on their observation of the sector loosing 3.54 percent on event 3. Table 3 suggests that the entire EREIT sector suffers from the reform. If their conclusion about HC facilities to be a winner is true, there should be several sub-sectors found on which the reform had a negative impact causing the overall EREIT impact to be negative. In order to investigate on this issue, the EREIT is split up into subsectors to gain better understating of which sectors caused the entire sector to be impacted.
4.2 Impact of Health Reform on REIT subsectors
Based on the analysis of Table 3, which reveals the entire EREIT sector to be highly sensitive to events regarding the probability of health reform enactment, the next abstract investigates if the impact can be attributed to specific subsectors, as formulated in hypothesis 1, or if instead the impact is distributed among all EREITs. For this, the sample of EREITs is split into 8 subsectors according to the property type labels provided by the CRSP database. Table 4 shows the results of the short event window (-1, 1). Here, columns 1 to 4 present the CAARs on the events of interest on the basis of the REIT index. Columns 5 to 8 present respective results but apply the equity index to enable insights on how the subsectors performed relative to two different indices. In the following results of each subsector are compared for both the REIT and equity index. Findings of the extended event window are provided in the Appendix (Table 4-B). Findings from this table are discussed if significant sign changes appear within the same index group on the same event and if the results help to clarify inconsistent findings observed in the short-event window.
The subsector of highest interest is the HC EREIT sample, which consists of 10 HC EREITs. Remarkably, a significant impact on the sector can only be observed on two of the four events. Using the REIT index, event 3 presents the CAAR of 0.99 percent, significant at the 5 percent level, which is consistent with the CAAR of 2.70 percent at the 1 percent significance level using the equity index, as both CAARs reveal the reform as being harmful to the sector. The impact is in line with the negative CAAR of -1.02 percent on Event 1 using the equity index. Furthermore, findings are confirmed by applying the extended event
window to test for robustness: a positive significant CAAR is found on Event 1, a negative significant CAAR on Event 3 (equity index).
Table 4
CAARs of EREIT Subsectors during Short Event Window
REIT type N (1) (2) (3) (4) (5) (6) (7) (8) 10.08.09 12.24.09 1.19.10 3.22.10 8.10.2009 12.24.09 1.19.10 3.22.10 House Passing Senate Passing Brown Election Senate Agreement House Passing Senate Passing Brown Election Senate Agreement Market Index CRSP REIT CRSP REIT CRSP REIT CRSP REIT CRSP Equity CRSP Equity CRSP Equity CRSP Equity Healthcare 10 0.0002 -0.0036 0.0099** 0.0051 -0.0102*** 0.0034 0.0270*** -0.0047 (0.04) (-1.13) (2.57) (0.97) (-2.67) (1.21) (6.49) (-0.84) Unclassified 7 0.0053 0.0065 -0.0014 -0.0032 -0.0040 0.0120** 0.0138 -0.0115 (0.55) (1.47) (-0.14) (-0.37) (-0.44) (2,30) (1.51) (-1.47) Diversified 10 -0.0040 -0.0056 -0.0080 -0.0113 -0.0157 0.0020 0.0111 -0.0218 (-0.33) (-0.95) (-0.62) (-0.43) (-1.20) (0.31) (0.83) (-0.86) Industrial / Office 25 -0.0065 -0.0048** 0.0122*** 0.0130 -0.0205*** 0.0043* 0.0359*** 0.0002 (-1.18) (-2.13) (2.59) (0.91) (-3.64) (1.81) (7.13) (0.01) Lodging 11 -0.0252** -0.0146 -0.0074 0.0128 -0.0415*** -0.0069 0.0198 -0.0020 (-1.97) (-1.10) (-0.48) (1,40) (-3.23) (-0.52) (1.19) (-0.24) Residential 15 0.0041 0.0024 -0.0035 0.0074* -0.0063 0.0101** 0.0159*** -0.0034 (0.59) (0.66) (-0.82) (1.78) (-0.88) (2.43) (3.24) (-0.72) Retail 27 0.0046 0.0001 0.0059 0.0052 -0.0089 0.0091** 0.0295*** -0.0075 (0.83) (0.02) (1.15) (0.97) (-1.62) (2.06) (5.43) (-1.30) Self Storage 4 -0.0122 0.0068 0.0108 -0.0101* -0.0243 0.0145*** 0.0323*** -0.0219*** (-0.74) (1.47) 81.40) (-1.70) (-1.38) (3.39) (3.76) (-3.46)
In this table CAARs are calculated for 8 REIT subsectors. In total, 109 REITs are grouped on the basis of the property type classification code provided by the CRSP database. All CAARs are calculated for the short event window (-1, 1). The estimation window to estimate the normal return ranges from 300 to 40 days prior to the respective event date. The first 4 columns contain the CAARs of the four events calculated on the basis of the REIT index. Columns 5 to 8 present the CAARs of the same events, however, calculated on the basis of the CRSP equity index. T-statistics are stated in brackets. ***, ** and * indicate the statistical significance level of 1% , 5% and 10% respectively.
The direct comparison of HC operators and HC EREITs reveals striking insights: Whereas in Table 3 significant CAARs indicating a positive impact dominate for HC operators, the positive signs for the comparable REIT sector on Event 3 indicate that the health reform imposes a harmful impact on the sector. This puts pressure on the validity of second hypothesis that for similar subsectors of operators and REITs similar effects are expected.
Besides the HC subsector, the Industrial Office subsector consisting of 25 EREITs is affected by the health reform. The CAARs based on the REIT index of -0.48 percent and 1.22 percent on the consecutive Events 2 and 3 are significant at the 5 and 1 percent level and are consistent in terms of the signs suggesting a harmful impact. Using the equity index, the signs are in line with the REIT index CAARs (event 1 and 3), only on event 2 a positive CAAR of 0.43 percent is not in line but instead suggests a positive impact. Checking the extended window supports the first opinion that the sector would suffer under the reform, as obtained significant CAARs on event 1 and 2 are positive, whereas the signs on event 3 are negative.
Next, the Lodging sector composed by 11 EREITs showed significant impacts. On Event 1, this sector experienced a CAAR of -2.52 percent at the 5 percent significance level, which consistent with obtained significant CAARs using the equity index and the extended event windows. The signs indicate a negative impact on the sector. However, the extended window reveals inconsistent significant impacts on the subsequent events: CAARs of 10.42, -7.11 and 12.25 percent on Events 2, 3 and 4 using the REIT index would suggest that the sector could benefit from the reform. Combining the findings does not allow a clear assessment of the event being positive or negative for the subsector.
The results of the Residential sector follow the inconsistent pattern of the previous subsector. In the short window, two significant CAARs are obtained that suggest a positive impact, whereas one suggests a negative impact. Using the REIT index, CAARs of 0.74 percent at the 10 percent significance level are confirmed in the long window by a CAAR of 2.82 percent at the 1 percent significance level. Regarding the equity index on Event 2 and 3, a CAAR of 1.01 percent and 1.59 percent. The latter is confirmed in the extended window by a CAAR of 5.09 significant at the 1 percent level.
Similar in terms of inconsistency are the results for the retail subsector. For the short window and the equity index, the CAAR of 0.91 percent at the significance level of 5% on Event 2 is inconsistent with the CAAR of 2.82 percent, significant at the 1 percent level, as they reflect a positive and negative impact, respectively. On equity index, event 3 reveals consistent CAARs of 2.95 and 4.96 percent significant at the 1 percent level for the short and extended event window. More consistent are the findings for the self-storage sector. Five out
of six CAARs indicate a negative impact of the reform on the sector. However, due to the small sample size of only 4 EREITs in the sample, the validity of the finding is limited.
In general, the latter analysis reveals two remarkable insights: as formulated in the first hypothesis and presumed from the significant impact of the entire EREIT sector (Table 3), the reform has far-reaching effects on the entire REIT market by affecting the subsectors of HC, Industrial Office, Lodging, Residential and Retail significantly. Applying the equity index reveals more significant CAARs of the subsectors, both in the short and extended event window. This is in line with the findings of Table 3, as the obtained CAARs of the entire EREIT sector revealed significant effects when using the same index. By using a wider index as a benchmark, this REIT-effect becomes even more observable, and hence, increases the validity of H1. The analysis based on the REIT index is a helpful analysis as it reveals the subsectors which over- or underperformed within the REIT universe. This allows the conclusion that the significant sectors observed provided an additional risk premium within the REIT universe compared to the equity market.
Besides the extent of impact, it is the direction of impact (positive or negative) that deserves more attention. Subsectors HC and Industrial Office are negatively hit during the majority of events. Furthermore, none of the subsectors reveal CAARs that suggest an overall positive impact on the REITs. Those findings are ambiguous in the context of the research by Al-Issis and Miller (2013), who ascribe the HC facility sector to benefit from the reform based on their observation of the sector loosing 3.54 percent on event 3. In the context of this analysis it cannot be validated that this positive impact applies for the entire HC subsector. However, the fact that the subsector as a whole reveals more negative impacts does not exclude the possibility of single HC EREITs or asset classes to benefit from the reform. Analysis on the individual REIT level will reveal more insights about this.
4.3 HC Operators and HC REITs in comparison
In order to test the second hypothesis, in-depth understanding about the HC operators and HC EREITs is required. For this, HC operators are split into subgroups and HC REITs into individual REITs. By matching the event study findings of these, conclusions if subsectors moved parallel in terms of sign. Each sector is benchmarked against the appropriate market index to ensure comparability between the sectors. That is for the EREITs the REIT index, and for the HC operators the equity index.
