• No results found

Is investing in a REIT a pure real estate investment?

N/A
N/A
Protected

Academic year: 2021

Share "Is investing in a REIT a pure real estate investment?"

Copied!
25
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Institution: University of Amsterdam, Amsterdam Business School

Program: MSc Business Economics, Dual track: Finance and Real Estate Finance

Document type: Master Thesis

Title: Is investing in a REIT a pure real estate investment?

Name of author: John VanderLogt

Student number: 10001516

Date: July 2014

(2)

Is investing in a REIT a pure real estate

investment?

By John VanderLogt Amsterdam Business School

Abstract

Real Estate Investment Trust (REIT) market capitalization increased from USD 9 billion to USD 154 billion (Brady and Conlin, 2004), making them one of the most common real estate investment vehicles. One of the recent puzzles in research has been the discrepancy in returns between REITs and private real estate. Fundamentally these are the same. However, they have shown different return patterns. Existing literature has solely focused on the differences in property management. This paper seeks to find out whether there are other determinants of the differences in returns, carrying not only large implications for private real estate owners possibly looking to turn their business into a REIT, but for all business owners looking to go public. This paper finds that the stock market is a big driver of REIT returns and therefore adds volatility to real estate returns. However, during a real estate bust, this stronger link between REITs and the stock market will partially protect REITs. Thus, REITs are not pure real estate investment vehicles since they are heavily linked to the stock market, making REITs a more volatile investment in ‘normal’ times but a safer investment in times of crisis.

(3)

Content

1 Introduction 3

2 Related Literature 5

2.1 Attribution Analysis 6

2.2 The drivers of REIT returns 7

3 Empirical Model 9

3.1 Autoregressive Conditional Heteroskedasticity model 9

3.2 Regression 10

4 Data 12

5 Descriptive Statistics 13

6 Results 15

6.1 Autoregressive Conditional Heteroskedasticity 15

6.2 Regression 16

(4)

1. Introduction

Between 1990 and 2002, Real Estate Investment Trust (REIT) market capitalization increased from USD 9 billion to USD 154 billion (Brady and Conlin, 2004), making them one of the most common real estate investment vehicles. The rise has come with high returns. As described by Brady and Conlin (2004), REITs have long outperformed privately held real estate. This is remarkable, since REITs and private real estate are fundamentally the same.

REITs are essentially real estate portfolio managers and as a firm they are publicly traded. They carry some benefits with respect to private real estate holders, such as the fact that they are tax exempt as long as they pay out 90% of their income1. REITs buy their assets from individuals and partnerships and thus are fundamentally equal to private real estate. This paper tries to add to the existing literature by answering the question of why these returns have differed so significantly.

REITs were created to be investors in real estate. The main advantage for investors is that it creates an opportunity to invest in a diversified portfolio of real estate. Creating an investment vehicle like this allows the investment in a REIT to be much smaller than when an investor acquires an entire (real estate) property. REITs have a tax advantage compared to corporations: they are tax-exempt as long as they pay out 90% of their net income. To ensure that institutions would not register as a REIT for the sole purpose of dodging taxes, extensive regulation was proposed and passed starting through 1961 and 1962 (Taylor and Bailey, 1963). Regulation included, among others regulations, the obligation for REITs to obtain at least 90% of their income from real estate property rent to ensure the company is involved mainly in real estate. Also, there are rules with the purpose of dispersing ownership. For example, no more than 50% of the outstanding shares can be owned by five or fewer shareholders. Once this regulation was fully past, the number of submissions of companies wanting to convert their firm to a REITs began to rise. In the early 1990s, the market capitalization began to take off significantly.

The aim of this paper is to further identify the factors that can explain the discrepancy between REIT returns and private real estate returns. Since some of these factors might be unrelated to real estate directly, the research question is: “Is investing in a REIT a pure real estate investment?” As described in the methodology section, this paper will find whether the

(5)

stock market, liquidity and leverage determine the return of REITs additional to pure real estate factors.

Implications of this research exist for direct commercial real estate owners. If their returns increase directly due to an increase in liquidity or due to a direct link with the stock market – some of these factors are discussed further in the methodology section – , this might provide an extra incentive for private real estate owners to turn their business into a REIT. Implications also exist for firms who are considering an IPO. The variables in the

methodology of this paper distinguish the returns of private real estate from those of public real estate. There is no reason to believe that the distinction in return is limited to real estate assets. Lastly, it has implications for diversification opportunities. Higher correlation between REIT and stock market returns indicates lower diversification benefits from investing in REITs and other publicly traded assets, regardless of whether their fundamentals show diversification opportunities.

The results of this paper do confirm the outperformance of REITs, but this is only significant when the financial crisis is included in the sample period. During ‘normal’ times, REITs perform no better than private real estate. Therefore, REITs should be seen as an investment that is safer than private real estate while yielding similar returns in normal times. Moreover, results suggest that the stock market does have a great influence on the

performance of REITs and that it takes much more that private real estate returns to explain the performance of REITs. Therefore the answer to the central question whether investing in a REIT is a pure real estate investment is conclusively answered. Investing in a REIT is not a pure real estate investment because it is heavily influenced by the stock market.

The remainder of this paper is divided in six more sections. Section 2 will provide an overview of what the existing literature has so far been able to explain and where it leaves gaps. Section 3 will discuss the data used. Section 4 will describe the empirical model that this paper uses to come to its results. Section 5 will provide a short but important description of the statistics of the data and Section 6 will provide the results. Concluding will be Section 7.

(6)

2. Related Literature

As mentioned in the introduction to this paper, REITs have grown rapidly during the early 1990s. To draw a clear image of this rapid change, Ott et al.(2005) examine REIT returns and hereby distinguish between the so-called old-REIT era (1981-1992) and the new-REIT era (1992-1999). Since the early 1990s, REITs benefited from clear and fully developed

(financing) policies that allowed them to raise capital quickly. This capital was mainly in the form of equity as opposed to retained earnings. The surge in real estate prices in the 1990s combined with a REIT’s ability to raise more money than a private real estate owner was able to do, REITs reaped the profits during these years. Ever since, REITs are considered to be different investment vehicles, rendering different returns from private real estate. This premise is supported by Chan et al.(2005). They look at the difference in return means, specifically on Mondays. They show that REITs have performed significantly better after the alternative capital structure of REITs was established. Besides capital structure, the authors ascribe the rise in returns mainly to the institutional participation and institutional nature of REITs. This institutionalization is related to the ability for REITs to hold a better diversified portfolio, but also to the fact that investors consider REITs as an investment vehicle to be more similar to stocks than private real estate is.

This shows that not only the size of REIT returns has differed from real estate, also the return patterns seem to have behaved differently. Over the last decade, some research has been performed with the intention of finding out why REIT returns have showed a different path then (private) real estate returns.

The existing literature differs from this paper in two distinct ways. Firstly, most literature related to this topic concerns the analysis of portfolios of REITs with which the authors try to explain REIT performance (with respect to privately held real estate). This so called attribution analysis involves the practice of comparing the performance of portfolio constituents and the weights to those constituents as a means of determining whether a portfolio manager performed well because of better selection or because of asset allocation respectively. One main distinction from the proposed research is that, whereas the existing literature tries to explain outperformance of REITs by looking at portfolio management, this paper’s aim is to find whether this outperformance can be explained by the fact that REITs have a different set of determinants outside portfolio management, unrelated directly to real estate. This allows this paper to make use of easily obtainable and large datasets. Secondly, a

(7)

lot of research has been done as to the characteristics of REITs and the correlation of REITs to other asset classes. The prevailing opinion is that REITs are vehicles hovering between common stocks, private real estate and even bonds. Some institutional characteristics of REITs distinguish them from these assets, such as the payout structure. In the current literature, many factors have been subjected to research as to how they (cor)relate with or drive REITs. However, these factors have not been used to explain the difference between REIT returns and the returns of private real estate.

2.1 Attribution analysis

As stressed above, any significant outperformance explained by attribution analysis indicates a better performance due to good management. In this section, an outline will be provided on the literature that investigates attribution analysis. In section 2.2 it shows that there is literature that suggests other factors than management might be able to explain a different return pattern for REITs compared to private real estate.

Firstly, there are numerous papers that perform the abovementioned attribution analysis. Derwall et al.(2009) conclude that management is indeed a very important factor. They come to this conclusion after having investigated REIT momentum. They find that REIT momentum cannot be explained by a regular factor model. Therefore they suggest that it might be the management that plays a bigger role than some researchers believe. Brady and Conlin (2004) look at the reasons for outperformance of REITs with respect to privately held real estate. The authors can be said to perform an attribution analysis, only looking at the portfolios of REITs. They find that property selection is a cause of differences in

performance. In their sample, the REITs own mostly mid-scale and high-end hotels. Because of the outperformance of hotels with respect to other property types, REITs outperform non-REIT real estate owners. The same goes for Redman and Manakyan (1995), who explore the portfolio determinants of REIT performance between 1986 and 1990. They compare REIT performance to equity and mortgage REITs. According to their research, a higher Sharpe-ratio for REITs is determined by the following factors: location of the property (western United States outperforms other locations), type of property (health care properties outperform) and investments in securitized mortgages. Once again, the conclusion is that REITs outperform because they are better at selecting outperforming real estate. Even though the approach is slightly different by Pagliari et al. (2005), they come to the exact same

conclusion. In their research, they compare the NCREIF (National Council of Real Estate Fiduciaries) and the NAREIT (National Association of Real Estate Investment Trusts) index.

(8)

These indices are price indices of unlevered private real estate and of REITs respectively. In order to do this correctly, they control for some of the methodological differences between the indices. Firstly, they control for appraisal smoothing in the NCREIF index. Secondly, they control for leverage. The NCREIF index consists of properties that are unlevered, whereas the NAREIT index contains REITs that can lever up to 40% usually. Finally and most importantly, they control for property selection. They find that, in the long run, neither of the two indices outperforms the other. However, this does not mean that property selection is the only or main driver of the discrepancy between REIT and private real estate returns. The authors explain that investors might still value the types of investment vehicles

differently due to liquidity and many other factors. Also, these differences might create very different patterns of returns in the short term and can alter the realized return of investors greatly.

2.2 The drivers of REIT returns

However, some papers move away from portfolio analysis and look for other factors correlated with REIT returns. In general, some research has been done on the drivers of REIT returns. Some papers looked at the macroeconomic drivers of Real Estate Investment Trusts’ returns. Ewing and Payne (2005) look at the effect of unexpected macroeconomic shocks on REIT returns. They find that macroeconomic shocks affect REIT returns in the same manner they affect real estate returns. Monetary policy aimed at raising the real interest rate in the short run adversely affects real estate activity. REIT returns respond negatively with a lag. This response occurs in similar fashion to the response of real estate returns. Similarly, an unexpected drop in economic growth causes REIT returns to drop where real estate returns would do the same.

In assessing the correlation between REIT returns with the return of other assets, one popular econometric model is the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) process. Fei, Ding and Deng (2010) use an Asymmetric Dynamic Conditional Correlation (AG-DCC) GARCH to look at the dynamics between REITs, stocks and private real estate. They find that REIT returns and stock returns are correlated and that REIT returns can be explained by this correlation. Also, macroeconomic variables determine REIT

volatility. The authors claim that REITs carry lower diversification opportunities than private real estate, suggesting that REITs are influenced by the stock market rather than purely by private real estate. Additional to the claims of Fei, Ding and Deng (2010) that correlation of

(9)

REITs with the stock market is higher, Chong et al. (2009) find that this correlation has increased steadily over time. The period from which they use their data is 1990-2005,

marking a significantly long period that makes their results highly accurate. Case et al. (2012) use a DCC-GARCH to examine the correlation between the returns of REITs and (non-REIT) stocks. They manage to identify three distinct periods of correlations. The results show that the correlation increased once REITs were included in the major stock market indices. This suggests that, compared to private real estate, REITs have a return pattern more similar to the stock market and therefore would suggest that there are other factors than attribution analysis that cause the pattern of REIT returns to diverge from the return pattern of private real estate. Older paper use a different approach. Ling and Naranjo (2003) use asset pricing models to assess the similarities in return. They find the same results as more recent papers such as Fei, Ding and Deng (2010) but their methodology has a lower degree of validity and scientific background.

Other papers provide evidence of REIT returns behaving more like private real estate than like stocks. Oikarinen, Hoesli and Serrano (2011) find that neither the NAREIT nor the NCREIF (REIT and private real estate indices respectively) are cointegrated with the stock market and that both indices are driven by the same real estate factors. These results are achieved by using a simple vector autoregressive (VAR) model. They elaborate that only in the long run, investing in REITs is a better diversification strategy than investing in private real estate. This goes against the findings of Fei, Ding and Deng (2010). The authors do express that there might be a stock market related factor of noise that is unrelated to real estate fundamentals, but they do not investigate this premise.

One obvious determinant of the outperformance of REIT returns would be liquidity. Bertin et al. (2005) compare the liquidity of REITs to the liquidity of common stocks. Their results prevailingly suggest higher liquidity for common stocks than for REITs. However, the results differ with measurement.

Interestingly, no extensive research has yet been performed on the liquidity of REITs as opposed to private real estate. One of the papers that looks at REIT liquidity is Niskanen and Falkenbach (2012). They look at the difference in liquidity between REITs and REOCs (Real Estate Operating Companies)2. They find a higher liquidity for REITs, which opens up the question as to how this influences their returns.

2

These differ from REITs mainly in terms of taxation and the activities they are required to perform (Niskanen and Falkenbach, 2012).

(10)

3. Empirical Model

3.1 Autoregressive Conditional Heteroskedasticity (ARCH) model

Before running the main regression of this research, an ARCH model will be used to examine the relationship between residuals of two variables. The aim is to obtain the same results previous literature found for two reasons. Firstly, it is important to confirm previous results since these results are important in answering the research question of this paper. Secondly, by obtaining the same results, one can confirm that the other results in this paper are in line with the previous literature as well.

Thus, following up on the research performed by autors such as Fei, Ding and Deng (2010) and Oikarinen, Hoesli and Serrano (2011), an autoregressive conditional

heteroskedasticity (ARCH) model will test whether REIT returns, private real estate returns and the stock market are linked. More specifically, it will test whether there is a relationship between the variance of each other’s residuals. When this test is performed with e.g. the stock market and REIT returns, the output would be the following equation.

In this equation, u is the residual of the stock market and the variance of the residuals of the REIT is dependent upon the squared residuals of the S&P500. Fei, Ding and Deng (2010) use a more elaborate model to assess the relationship between the stock market and REIT returns. However, the results of both models can be similarly interpreted.

According to the paper of Fei, Ding and Deng (2010), the returns of REITs can be predicted by looking at the correlation between REITs and the stock market. Section 5.1 pointed out that correlations rose when the financial crisis hit. Moreover, the outperformance of REITs changed because of the financial crisis, thus changed according to the level of correlation.

To find out whether historical returns can be predicted by correlation as Fei, Ding and Deng (2010) claim, data of before 2007 (pre-crisis) and data of the entire period including the years of crisis are compared.

(11)

Two ARCH models will be run. The first will show the relationship between the residuals of REITs and the residuals of the stock market (S&P500). The second will display the relationship between the residuals of REITs and the residuals of the NCREIF index.

3.2 Regression

A regression will be performed using REIT returns as the dependent variable. The independent variables used are not directly related to private real estate returns unlike e.g. GDP and other macroeconomic factors. First, REIT returns will be regressed on real estate returns using the real estate return index as a proxy. Fundamentally, the dependent variable should be the same as the independent variable. Then, other variables are added to the regression. These are leverage, the stock market – the S&P500 – and illiquidity, such as the following equation displays.

Illiquidity is defined as . This is a measure of illiquidity in

which Days stands for the number of observations in time period t, R stands for the return of a REIT on a certain day and Vol stands for the volume of trading in millions of dollars on that day. The variable Days will represent the 30 days preceding the moment in time at which the returns are observed. This measure is introduced by Amihud (2002) and is not yet used to measure liquidity of REITs.

Leverage will be set against every individual REIT, as the subscript i indicates. Book-leverage is defined as by Baker and Wurgler (2002): it is the book debt over total assets, where book debt is total liabilities plus the value of preferred stock (redemption value if preferred stock is missing) minus the deferred taxes and convertible debt. With illiquidity, this is the set of panel data, the rest are time-series.

One of the factors that might change the returns of REITs is the stock market. Fei, Ding and Deng (2010) find that the correlation between REITs and the stock market can explain REIT returns. A theory that might be linked to this finding is the interdependence of

(12)

publicly traded assets which is so high that investors were unable to immunize their portfolio during the financial crisis starting in 2007/2008 (Baur, 2012).

The inclusion of leverage is explained by the following. First of all, Alcock et al. (2013) show that leverage is used by REITs to manipulate their performance measure. Since REITs are public institutions, they are under higher pressure to perform well. Thus, if higher leverage indicates higher returns, this might be explained by the fact that REITs are public institutions. Second, Janssen, Teuben and Hordijk (2006) explain an increase in part-sales in real estate by increasingly dynamic trading demand by shareholders. REITs, as opposed to private investors, need to hold higher leverage since dynamic trading will allow them to manipulate performance to a larger extent. Lastly, Cohen and Steers (2013) suggest that liquidity is a driver of returns but do not investigate this any further.

The former three theories result in the following hypothesis:

Hypothesis

The difference in REIT and private real estate returns can be explained by other factors than attribution analysis.

This hypothesis implies that REITs and private real estate does not just display different returns because portfolios are managed differently. Hence, confirmation of the first hypothesis will mean that investing in a REIT is not a pure real estate investment and will thus answer the question of this paper.

When the variables show a significant degree of explanation when added to the regression, it is indicated that direct commercial real estate owners could increase their returns by turning their entity into a REIT without the necessity of improving the property management and portfolio management expertise. In this case, the conclusion would be that investing in a REIT is not more favorable because of higher expertise, thus that investing in a REIT is not a pure real estate investment.

(13)

4. Data

Only US REITs will be considered since in the US, REITs are publicly traded and thus data are easily obtained. This is not always the case for many other countries. The period 1990-2012 is used. This period captures the period of growth of REITs (Brady and Conlin, 2004).Datastream will be used to retrieve returns of US REITs, data on the US stock market and data on liquidity. CompuStat (North America / Simplified Financial Statement Extract) is used to find data on leverage of North American REITs (a total of 5501 records for REITs). One reason for the discrepancy in returns is the use of appraisal-based real estate indices (Oikarinen, Hoesli and Serrano, 2011). The NCREIF (the NCREIF Property Index) is one of those and will be used in this paper. It considers private real estate held for investment purposes only. Appraisal based real estate indices such as the NCREIF index are imperfect because of – among other reasons – infrequency in real estate transactions. This causes appraisals to rely more on historic valuations of properties, creating a perseverance of values or an insufficient adjustment of prices. This paper resolves this issue by using an

unsmoothing technique.

Reverse-engineering technique

First, NCREIF returns (2013) provides quarterly returns. The geometric annual return is calculated to then subject it to the process of unsmoothing. Subsequently, to unsmooth the NCREIF index, the reverse engineering formula is used. This formula is the following:

The unsmoothed return is calculated by using the formula in which K is the historically average lag (in years) of the NCREIF index with respect to an ex post transaction index, and where is the geometric annual return. Haurin (2003) compares the NCREIF to many indices, among which an hedonic price model which is one very common form of an (ex post) transaction based index. He finds an average lag of about 2 years. Therefore, the value for K is taken to be 2.

(14)

5. Descriptive statistics

In order to offer a quick indication of the dataset, some summary statistics are provided in the table below. When starting at the bottom, one can see that the values for illiquidity are very low. This is expected since a quick reminder will tell you that illiquidity is defined as the price change induced by a trade with a volume of a million US dollars.

Leverage is on average 57.4%, which is very high. REITs are restricted in the amount of leverage they have. Usually, this limit lies around 30-40%, much lower than the actual amount of leverage REITs take on in reality. It seems that REITs do try to increase their leverage to boost their performance. Therefore, this might be an indication that the inclusion of leverage might render some interesting results.

Table 1 Summary statistics.

This table provides descriptive statistics for every variable in the regression. The dependent variable as well as all independent variables are described below by mean, median and variance.

Variable mean median variance nr. of obs.

REIT return 0.185 0.151 0.058 1531

NCREIF return 0.094 0.140 0.058 1657

S&P500 0.064 0.090 0.032 1655

Leverage 0.574 0.585 0.051 1655

Illiquidity 0.003 0.0001 0.0001 1445

The first indication of relative REIT performance can be obtained when comparing their performance directly to the performance of private real estate (NCREIF) and the performance of the stock market (S&P500). Figure 1 shows a comparison of this performance between 1990-2012. One observation that can be made is that the NCREIF index shows lower volatility before the financial crisis of 2007 compared to REITs and the stock market. The standard deviation of NCREIF returns before the crisis is 13%, compared to 19% for REITs and 17% for the S&P500. When including the financial crisis starting in 2007, the standard

(15)

deviations shift: the S&P500 shows a standard deviation of 18%; both REITs and the unsmoothed NCREIF index show the same volatility of 24%.

Additional to this result, it is important to notice that REITs follow a return pattern much more like the S&P500 than like the unsmoothed NCREIF index. Therefore, as expected, the unsmoothed NCREIF index shows a much lower correlation with the stock market than REIT returns do. Correlations are approximately 0.28 and 0.48 respectively. Interesting to notice is that when one excludes the data after 2006, correlations appear to be much lower. Whereas correlations were initially a rounded 0.28 and 0.48, correlations for data without including the financial crisis drop significantly to a rounded 0.12 and 0.18. Considering Baur (2012) his finding that all assets became highly correlated during the financial crisis, this result is expected.

Figure 1 Returns of REITs, unsmoothed NCREIF index and the S&P500.

In line with the literature, the data show an outperformance of REITs with respect to the NCREIF index of an average rounded 1.8 percentage points per year. The biggest

outperformance was experienced during the start of the financial crisis – 2007 and 2008 show levels of outperformance of respectively 13 and 29 percentage points. However, when not taking into account years 2007 and 2008, REITs and private real estate seem to show similar average returns, only a small outperformance of REITs of 0.08 percentage points per year on average. -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8

(16)

6. Results

6.1 ARCH model

Table 2 displays the results of the autoregressive conditional heteroskedasticity models. Panel A shows the results of the models when all data is used, including the years during which the financial crisis occurred. Section 5 pointed out that correlations rose when the financial crisis hit. Moreover, the outperformance of REITs changed because of the financial crisis, thus changed according to the level of correlation. Therefore, the same ARCH model is run again without the data after 2006 to see if the results indicate a difference when the financial crisis is disregarded. These results are displayed in Panel B.

Using the results from Table 2, the equations become as follows: All data used: REIT – S&P500

All data used: NCREIF – REIT Pre-crisis data: REIT – S&P500 Pre-crisis data: NCREIF – REIT

The following displays a summary of the coefficients from the ARCH models.

Coefficients Pre-Crisis 1990-2012

REIT-S&P500 -0.279 0.247

NCREIF-REIT 0.960 0.828

When coefficients are high, there is higher volatility clustering, which means it’s easier to predict future return patterns. Similarly, Fei, Ding and Deng (2010) find values of just over 0.2 for the different sub indices of the NAREIT they use. Thus, during the crisis, the correlation between REITs and the stock market increased, making it easier to predict the return of REITs by looking at the returns of the stock market. This is in line with Baur (2012) who found that correlations rose during the crisis. It is also in line with the descriptive

analysis in Section 5. Interestingly, the coefficient of the equation comparing NCREIF and REITs dropped. This indicates that it is harder to predict future return patterns of REITs looking at private real estate returns.

(17)

The main result is in line with the conclusion obtained from the descriptive statistics in Section 5. At the same time, the results are similar to the results in Fei, Ding and Deng (2010).

6.2 Regression

First, the REIT returns are regressed against the unsmoothed NCREIF index. The results are straightforward and as expected. Panel A of Table 3 shows that the coefficient of the

NCREIF index is low, only 21.1%. This means that REIT returns are only to a small extent affected by private real estate returns. The constant is positive with a certainty of 95%, meaning that other variables that determine REIT returns have been left out. Thus, this result is fully in line with the fact that REIT and private real estate returns have behaved differently.

The second regression – with the addition of the variables book leverage, stock market and liquidity – yield the results shown in Panel B of Table 3. Maybe the most

impressive result is the increase in the R squared: it rises from 0.119 to 0.680, indicating a big increase in the degree of explanation of REIT returns. Of the three additional variables, only the stock market shows significant results with a p-value of 0.000. The highly positive coefficient is the expected result and is in line with the findings of Fei, Ding and Deng (2010). These regressions are also run with only pre-crisis data, but that did not yield any significantly different results and are therefore not included in this paper.

(18)

Table 2 Results of autoregressive conditional heteroskedasticity models

Both panels contain output from the ARCH model run in Stata. The bottom two lines (under sub header “ARCH”) display the output that eventually ought to be used to formulate the equation introduced in Section 3.1.

Panel A: All data

Panel A1: REIT – S&P500

Variable Coefficient SE z p-value

S&P500 return -0.151 0.187 -0.810 0.419

constant 0.117 0.051 2.310 0.021

ARCH

S&P500 return lag 0.247 0.416 0.590 0.552

constant 0.038 0.024 1.590 0.111

Panel A2: NCREIF-REIT

Variable Coefficient SE z p-value

REIT return 0.115 0.081 1.420 0.156

constant 1.081 0.013 80.600 0.000

ARCH

REIT return lag 0.828 0.710 1.170 0.044

(19)

Panel B: Pre-crisis

Panel B1: REIT – S&P500

Variable Coefficient SE z p-value

S&P500 return -0.187 0.354 -0.530 0.598

constant 0.123 0.056 2.200 0.027

ARCH

S&P500 return lag -0.279 0.529 -0.530 0.598

constant 0.041 0.027 1.510 0.131

Panel B2: NCREIF-REIT

Variable Coefficient SE z p-value

REIT return 0.017 0.060 0.290 0.768

constant 1.102 0.0184 59.95 0.000

ARCH

REIT return lag 0.960 0.973 0.990 0.324

constant 0.001 0.002 0.500 0.614

(20)
(21)

7. Conclusions

This paper adds to the existing literature that tries to explain the difference between the returns of REITs and the returns of private real estate. Thus far, the literature has focused on the quality of portfolio management by REITs as compared to private owners of real estate. This paper investigates whether there are other factors that could explain the discrepancy in returns. Other papers have found what REIT returns correlate and co-move with, but these factors have never been used to explain the difference in performance of REITs as compared to private real estate.

This paper makes use of a simple regression in which it combines REIT returns and some of the factors REIT returns have been proven to correlate with. The results lead to the following findings.

Firstly, even though Alcock et al. (2013) find that REITs increase their leverage to manipulate performance, the results of this paper show that this phenomenon does lead to outperformance of REITs compared to private real estate, but only to a small extent. The results are significant at a mere 6.3% level.

Second, even though liquidity is expected to induce outperformance of REITs, the results are not significant. Further research will have to focus more on liquidity as a driver of REIT returns.

Thirdly, the stock market is a big determinant of REIT returns. This is a result confirmed by the descriptive statistics in Section 5 as well as the results from the ARCH model and the main regression. Also, the stock market is much more of a determinant for REIT returns than it is for private real estate. This implies that real estate owners who opted to utilize their current real estate holdings to create a REIT before the financial crisis that started in 2007 would have been partly protected against the high price drops of private real estate. In normal times, they will experience higher volatility due to the link to the stock market. However, during these normal times, REIT owners will not experience a much better performance. In the end, holding a REIT only seems to protect owners from large price drops during a real estate bust, making it a safer investment yielding the same returns in normal times.

(22)

It is important to view the results and consequent implications in the context of the research. The gap in the current literature allows for a more simplified methodology. When a certain research area has not yet been exploited, a simple methodology can already render interesting results. However, it should be explored what kinds of more sophisticated

methodologies would be more suitable for this research. More reliable results can be obtained with an empirically stronger model.

Also, it would be interesting to dig deeper into the effect of liquidity of REITs as a determinant of the outperformance of REITs with respect to private real estate. Once again, a more sophisticated model might render more significant results for this specific determinant and might give more insight into the channel through which liquidity might affect REIT outperformance.

Nonetheless, the research laid out in this paper provides some interesting results and is worth exploring further as mentioned above.

(23)

References

Amihud, Y. (2002). “Illiquidity and stock returns: Cross-section and time-series effects”. Journal of Financial Markets Vol. 5, pp.31-56.

Alcock, J., Glascock, J., Steiner, E. (2013). “Manipulation in U.S. REIT Investment Performance Evaluation: Empirical Evidence”. The Journal of Real Estate Finance and Economics,Vol.47(3), pp.434-465.

Baker, M., Wurgler, J. (2002). “Market timing and capital structure”. Journal of Finance, Vol. 57(1), pp.1-32.

Baur, D.G. (2012). “Financial Contagion and the Real Economy”. Journal of Banking and Finance 36, pp.2680-2692.

Bertin, W., Kofman, P, Michayluk, D. and Prather, L. (2005). “Intraday REIT liquidity”. Journal of Real Estate Research, 27(2), pp.155–176.

Brady, P.J., Conlin, M.E. (2004). “The performance of REIT-owned properties and the impact of REIT market power”. Journal Of Real Estate Finance And Economics, Vol.28(1), pp.81-95.

Case, B., Yang, Y., Yildirim, Y. (2012). “Dynamic Correlations Among Asset Classes: REIT and stock returns”. The Journal of Real Estate Finance and Economics, Vol. 44(3), pp. 298-318.

Chong, J., Miffre, J. and Stevenson, S. (2009). “Conditional correlations and real estate investment trusts”, Journal of Real Estate Portfolio Management, 15(2), pp.173–184.

Derwall, J., Huij, J., Brounen, D., Marquering, W. (2009). “REIT Momentum and the

Performance of Real Estate Mutual Funds.” Financial Analysts Journal, Vol.65(5), pp.24-34.

Ewing, B.T., Payne, J.E. (2005). “The response of real estate investment trust returns to macroeconomic shocks.” Journal of Business Research, Vol.58, pp.293-300.

(24)

Fei, P., Ding, L.T., Deng, Y.H. (2010). “Correlation and Volatility Dynamics in REIT Returns: Performance and Portfolio Considerations”. Journal Of Portfolio Management, Vol.36(2), pp.113-125.

Janssen, J.E., Teuben, B.J.J., Hordijk, A. (2006). “Part-sales as an investment strategy: Analysis of part-selling of residential units in the Netherlands”. ERES Conference 2006.

Haurin, D. (2003). “US commercial real estate indices: transaction-based and constant-liquidity indices”. BIS papers, No 21, pp. 232-242.

Ling, D., Naranjo, A. (2003). “The integration of Commercial Real Estate Markets and Stock Markets”. Real Estate Economics, Vol.27(3), pp.483-515.

NCREIF returns (2013). http://www.ncreif.org/property-index-returns.aspx. Visited March 20, 2014.

NCREIF (2013). http://www.ncreif.org/faqsproperty.aspx. Visited March 20, 2014.

Niskanen, J., Falkenbach, H. (2012). “Liquidity of European real estate equities: REITs and REOCs”. International Journal of Strategic Property Management Vol. 16(2), pp. 173–187.

Oikarinen, E., Hoesli, M., Serrano, C. (2011). “The long-run Dynamics between Direct and Securitized Real Estate”. Journal of Real Estate Research, Vol. 33 (11).

Ott, S.H., Riddiough, T.J., Yi, H.C. (2005). “Finance, Investment and Investment

Performance: Evidence from the REIT sector.” Real Estate Economics, Vol.33(1), pp.203-235.

Pagliari, J.L., Scherer, K.A., Monopoli, R.T. (2005). “Public versus private real estate equity: a more refined, long-term comparison.” Real Estate Economics, Vol. 33(1).

(25)

Redman, A.L., Manakyan, H. (1995). “A Multivariate-analysis of REIT Performance by Financial and Real Estate Portfolio Characteristics”. Journal Of Real Estate Finance And Economics,Vol.10(2), pp. 169-175.

Taylor, G.S., Bailey, E.N. (1963). “Real Estate Investment Trusts.” Business Horizons, Summer 1963, pp.71-80.

Referenties

GERELATEERDE DOCUMENTEN

Alice and Bob set their threshold to detect eavesdropping to a bit error rate of Q + σ, where Q is the averaged quantum bit error rate.. Basis

for transmission of Schistosomiasis in Kayonza District Rwanda medical. Standley CJ, Adriko M, Alinaitwe M, Kazibwe F, Kabatereine NB, Stothard JR. Intestinal schistosomiasis

9 Building upon the constellations of state and non-state actors to induce regulation at transnational level, Börzel and Risse (2005) distinguish four types of

Want we kunnen natuurlijk wel heel gemakkelijk zeggen, oke dit is de nieuwe policy, maar we hebben al een overall corporate policy die al heel erg gefocust is op

[r]

Agressieve belastingconcurrentie wordt tegengegaan door middel van de CCCTB, omdat de CCCTB zorgt voor een geharmoniseerde grondslag binnen de EU. Hierdoor zullen er geen

Laskar, “A fully integrated UHF-band CMOS receiver with Multi-Resolution Spectrum Sensing (MRSS) functionality for IEEE 802.22 cognitive radio applications,” IEEE J.

Africa’s quest for economic and political union can be traced to the founding of the organisation for African unity (OAU) in 1963. The initial goal of the early African leaders