1
The mitigating effect of REITs on the amplitude of Commercial
Property Construction Cycles and the institutional characteristics
influencing it
Rens Boing (10003429)
Master Thesis
MSc Business Economics, Finance & Real Estate Finance track University of Amsterdam
June 2016
2
This document is written by Rens Boing who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Inhoudsopgave
1. Introduction ... 5
2. Literature ... 7
2.1 Real Estate price dynamics ... 7
2.2 Influence REITs on property cycles ... 10
2.3 Characteristics ... 11
2.4 Multicollinearity ... 16
3. Data ... 18
3.1 Real Estate Investment Trusts in sample countries ... 19
3.1.1 Development and performance of the REIT markets ... 19
3.1.2 Characteristics ... 24
3.2 Variables tested ... 26
3.2.1 Dependent variable ... 26
3.2.2 REIT market characteristics ... 29
3.2.3 REIT characteristics ... 33 3.2.4 Governance variables ... 35 3.2.5 Control variables ... 37 4. Methodology ... 40 4.1 Lagged variables ... 41 4.2 Regression models ... 42 4.2.1 Baseline model ... 42
4.2.2 Regression models with included explanatory interaction variables ... 43
4.3 Regression methodology ... 44 4.3.1. Panel regression ... 45 4.3.2. Assumptions ... 45 4.3.3. Specification tests ... 46 5. Results ... 48 5.1 Baseline model ... 49
5.1.1 Results methodology tests ... 49
5.1.2 Results ... 50
5.2 Final regression model ... 53
5.2.1 Results methodology tests ... 53
5.2.2 Final regression results ... 53
4
5.4 Country specific models ... 61
5.4.1 US ... 61
5.4.2. The Netherlands ... 63
5.4.3 France ... 64
5.5 Robustness checks ... 64
5.5.1 Only REIT data ... 64
5.5.2 Germany and UK observations omitted ... 65
5.5.3 Belgium and Germany omitted ... 65
6. Conclusion ... 65
7. References ... 68
Appendices ... 70
A. Assumptions ... 70
B. Residential property cycles in the individual European countries... 70
B1 European countries with a boom bust cycle in the residential market (France, The Netherlands, UK) ... 70
B2 European countries with an increasing or flat Real Estate cycle in the residential market (Belgium, Germany) ... 71
C. Individual Country level tables and graphs ... 71
D. Results Heteroscedasticity tests ... 74
D1. Results Log likelihood tests... 74
D2. Results Breusch Pagan test ... 74
E. Results tests for serial correlation ... 74
F. Results test for panel serial correlation ... 74
G. Results multicollinearity test ... 75
H. Results robustness checks ... 76
H1 Serial correlation ... 76
H2 Robustness check only REIT ... 78
5
1. Introduction
The financial crisis of 2008 has shown how devastating a housing price bubble can be for the global economy. Most economies have only recovered slowly in recent years while others are still waiting to fully recover. Many reports have been written about how the housing price bubble could have been prevented. In contrast to the residential property markets the commercial property markets in the United States and most of the commercial property markets in Europe haven’t experienced such a boom and bust (Packer, Riddiough, & Shek, 2013). The question is what constitutes the difference between the residential and the commercial property market. Some researchers ( (Packer, Riddiough, & Shek, 2013), (MacKinnon, 2010) and (Zhu, 2012)) suggests that the presence of Real Estate Investment Trusts are the key to understanding these differences. Real Estate Investment Trusts are listed property investment companies which primarily invest in Real Estate (Brounen & de Koning, 2012). These investment vehicles don’t pay taxes at the entity level and in return have to apply to strict regulations. Most important REITs are forced to be transparent and in most countries have to pay out around 85 percent of their investment income (EPRA, 2015). Due to the fact that REITs are so transparent and having to return to the capital markets frequently (due to the high payout ratio) the information asymmetry is low and the prices of REIT shares should provide accurate signals about the state of the underlying properties held by these REITs. Assuming that investment managers of REITs are capable and make good investments other agents in the property market could use REIT share prices to get indications about the property markets. All decision makers in the entire commercial property market can then use this information to make better capital allocation decisions for themselves. The two hypotheses are thus as follows: First REIT share prices are a good indicator of the underlying property market and second agents in the property market use this information to make investment allocation decisions. If these hypotheses are true then signals from REIT share prices will cause agents to allocate less to overheated markets and more to cold markets, damping the property supply cycle. The goal of this research is to prove these hypotheses for every country. The main research question is: Does the REIT market exert a mitigating effect on Commercial Property
6
Construction Cycles and which institutional characteristics influence this. Packer et al. (2013) have found proof for this mitigating effect for the United States and Japan but not for other countries. This research tries to prove this hypothesis by using a different proxy for the mitigating effect of the REIT market and by identifying and adding institutional characteristics of the REIT markets that increase or decrease this mitigating effect.
This is done by using a dataset spanning from 1997Q3-2015Q2 which contains data on Belgium, The UK, The Netherlands, Germany, France and The United States. First a baseline model is tested to test the significance of our proxy for the mitigating effect of the REIT market: Analyst coverage interacted with a dummy variable for being in a strong or weak economic state. A second regression is done using the baseline model augmented with the institutional characteristics.
This research finds evidence that the REIT market exerts a mitigating influence on the commercial property cycle in the US and France using Analyst Coverage as proxy for investor attention to the REIT market. Furthermore, the REIT market characteristics Institutional Ownership, Turnover, The Actual Dividend Payout Ratio and the variance of the REIT index are identified as characteristics that influence the mitigating effect of REITs on the amplitude of the commercial property cycle. The contribution of this thesis to the existing literature is twofold. First significant results have been found for a different proxy for the mitigating effect of the REIT market which enables a comparison between countries over a longer time period (Analyst Coverage interacted with a dummy variable indicating a good or a bad economic state). Second this research identifies and statistically proves that certain other characteristics either enforce or decrease the mitigating influence of REITs on the amplitude of the Commercial Property Construction cycle.
This thesis proceeds as follows: In section 2 the literature regarding Real Estate supply dynamics, the mitigating influence exerted by REITs on the amplitude of the property construction cycle and the institutional characteristics which determine this mitigating effect is given. Section 3 provides a brief description on the development, performance and importance of the REIT markets in our sample, the
7
dataset and the variables used in this research. Section 4 describes the regression methodology, section 5 contains the results and the robustness checks and finally section 6 concludes.
2. Literature
This section describes the literature regarding the mitigating effect of REITs on the property cycles and which institutional characteristics could contribute to this effect. Section 2.1 discusses the literature about the mitigating influence of the REIT market on the amplitude of the supply cycle. Section 2.2 and 2.3 identify institutional characteristics that could explain the differences in the level that REITs mitigate the amplitude of the commercial property supply cycle. Since a lot of characteristics that could influence the mitigating effect of the REIT market on the property cycle are tested, section 2.4 describes literature that tests relationships between our potential explanatory variables in order to address multicollinearity problems.
2.1 Real Estate price dynamics
This chapter provides an overview of the Real Estate price dynamics and compares the commercial and residential Real Estate markets in our sample. The comparison is made because Packer et al. (2013) conclude in their description of the supply dynamics of the US Real Estate market that although the housing market and the commercial Real Estate market seem to co-move, significant differences are evident. Both markets have seen a significant downfall during the crisis, but the commercial Real Estate market has recovered quickly after the crisis while the residential Real Estate market was by then still in the woods and only has started to recover very slowly in the US. The main reason Packer et al. (2013) give for this is that the presence of REITs caused agents in the property market to see that the underlying property markets were getting overheated through pricing signals of these REITs and didn’t overinvest. Therefore, there was not as much overbuilding as in the residential Real Estate markets. Packer et al. (2013) also consider two other rationales. First the severe Real Estate bust in the late 1980s and early 1990s caused market participants to be hesitant to invest, effectively damping the supply response to the huge capital inflow during the booming years 2006-2008. The second rationale is that the housing crisis in 2008 stopped the overbuilding in the commercial property market
8
before it could begin. Packer et al. (2013) show that the commercial property market lags the housing market with one year. Therefore, it could be that the commercial property market was on the brink of getting overheated when the crisis hit the housing market and that this stopped overbuilding in the commercial property market.
Finally, Packer et al. (2013) conclude by rejecting the argument that agents could distill signals from the public residential Real Estate market just as in the commercial property market since the housing market depends on more factors. Therefore, it is harder for agents to distill signals about the underlying housing markets from the price of the public housing shares. Furthermore, information about the housing sector was less viable, the home builder index wasn’t widely followed and the agents in the market didn’t pay attention to the newly developed ABX-CDS market according to Packer et al. (2013).
Therefore, this first rationale is deemed the most plausible by Packer et al. (2013) and is investigated by them. Since my research tries to find the same causal relation between the REITs and the decreased amplitude of the commercial property cycle this chapter will first compare the Construction Cycles between Europe, the US and the individual European countries by comparing the residential and commercial property prices. This in order to assess whether the same dynamics between the European non-residential and residential market are visible before testing whether European REITs causes the differences between the residential and property markets described by Packer et al. (2013). Figure 1 describes the movement of the commercial and residential property prices in the United Sates and confirms the findings by Packer et al. (2013) that the commercial property market in the US experienced a downfall one year after the residential property prices and that the commercial property market recovered shortly thereafter. Figure 1 shows that the cycle of the European commercial property market follows the property market to a large extent. The commercial property market in Europe also experiences a considerable downfall in 2007 and also recovers quickly. This is where the similarities between the United States and Europe end. The European residential property prices didn’t experience a profound drop as was the case in the U.S but have been stable over time.
9
In 2008 the residential property prices dropped slightly but not as much as in the US. In contrast to the commercial property market in the US the European Commercial Property market experienced a much larger drop in prices than the residential market.
Figure 1 Commercial and Residential property prices (Source: Hiebert&Wredenborg (2012))
Therefore, the claims done by Packer et al. (2013) don’t seem to apply for Europe however, one should keep in mind that there are intra country differences in Europe. These figures are averages for Europe, but the Real Estate residential supply dynamics per country plotted in Appendix B show that the property price indexes have large inter country differences. Appendix B1 and B2 show that France, The Netherlands and the UK experienced a residential boom bust cycle similar of that of the in the United States while Belgium and Germany experienced stable property prices or slow continuous growth. Hartmann (2015) indicates that the differences in (over) valuation of residential properties before the crisis is caused by the large institutional differences between the countries, some of these institutional differences could also explain the differences in mitigating effect for the commercial property market. Packer et al. (2013) are not the only ones who study the Commercial Property Cycles and concluded that REITs exert a mitigating effect. Zhu (2002) looks at the US commercial property
10
cycle as well and also suggests that the commercial property cycles are slowly disappearing. One of the reasons given by him is the signaling through REIT share prices. Zhu (2002) also investigates the commercial property markets of Europe, Australia and Japan and finds the same results for the countries in our European sample. Even though evidence from Europe suggest that there could be other reasons why commercial property cycles seem to be less pronounced than residential cycles, Zhu (2002) also finds that the amplitude of the commercial property cycles in European countries have decreased. Therefore, REITs could still be an important factor in mitigating the property cycle in Europe even though the boom bust pattern in the residential Real Estate cycle is less pronounced than in the US residential Real Estate cycle.
2.2 Influence REITs on property cycles
Mueller (2002) is one of the first to link the presence of REITs to the smaller amplitude of the commercial property cycle. He states that due to increased monitoring by research teams for REITs it is difficult for these REITs to overinvest. The consequence of this more extensive monitoring according to Mueller (2002) is that the information feedback loop has shortened due to increased information efficiency since information about demand fundamentals takes months instead of years to reach investors in the Real Estate market. He states that this shorter information feedback loop could prevent future booms and busts due to oversupply in the property market. Although he does not investigate his theory with formal statistical tests he uses figures to show that there was little overbuilding in the commercial property sector during the 1990s cycle and during the dot com bubble in relation to earlier property market booms.
MacKinnon (2010) describes the changing nature of the Real Estate markets due to REITs and notes that the fundamentals have become less important in determining property values in favor of the capital markets. MacKinnon (2010) looks at the most recent boom period and also finds that there was little overbuilding in the commercial property sector. He states that he does not perform statistical tests because it is not possible to determine precisely how the Real Estate market has changed because the time period to investigate this is too short.
11
Packer et al. (2013) are the first to make an effort linking the presence of REITs and a decreasing amplitude of the property cycle using a regression. First they test their hypothesis on the US from 1991-2011 and find economically and statistically significant results. Second they try to expand their investigation to France, Australia and Japan, but they only find statistically significant results for Japan. They claim the differing results are due to the different characteristics of REITs in the countries they have investigated and that there is future research needed to test the differences in REIT characteristics. Packer et al. (2013) give some suggestions about which characteristics of REITs and the Real Estate market of a country could explain the differing results. First the REIT market should be large enough for agents who invest in properties to look at REIT share prices for signals about which underlying property market they should invest in. Second Packer et al. (2013) state that structural characteristics related to the REITs in a country have an influence on the results. He suggests structural characteristics like:
The required dividend payout ratio of REITs, since this determines how often REITs have to return to the capital market.
Ownership by institutions, since institutions are monitoring REITs more closely than private investors.
Whether REITs are internally or externally managed, because internally managed REITs usually outperform their externally managed counterparts.
The complexity of REIT structures, since complex REITs are less transparent for investors. The variance of REIT share prices
2.3 Characteristics
Despite the fact that there has not been further research done investigating the link between REITs and the commercial property cycle there has been research done on the relationship between REIT characteristics and REITs return, momentum, firm value and investment levels. Some of the proxies used in these researches are useful to include in this research since these variables also seem relevant in explaining intercountry differences in the mitigating effect of the REIT market.
Hong et al. (2000) have researched the relationship between the effectiveness of momentum strategies and firm characteristics. They argue that some firm characteristics like size and Analyst
12
Coverage may cause information about the firms to reach investors quicker. Although this research isn’t related to REITs it does provide a useful proxy for how much attention investors pay to a firm. He uses Analyst Coverage as a proxy for attention the market pays to a firm, which is estimated by the number of analysts that have given a price estimate for a firm in a certain period in I/B/E/S. Goebel et al. (2013), Chui et al. (2003) and Devos et al. (2007) also use this proxy for the attention by the investors paid to the REIT market. Packer et al. (2013) use the total office investments made by the REIT sector divided by the total investment stock of the office market as a proxy for investor attention. They state in their paper that further research needs to be done using a better proxy for investor attention to REITs. Therefore, Analyst Coverage is used in this research, instead of the proxy by Packer et al. (2013).
Analyst Coverage interacted with a strong or weak economy will be used to test the mitigating effect of the REIT market on the property cycle. If this interaction variable provides the same results as in Packer et al. (2013) characteristics can be added to find institutional characteristics that influence this effect. There are two ways REIT market characteristics can influence the mitigating effect. The characteristics that are added can influence the effect of the interaction variable (thus the mitigating effect of the REITs) in two ways. First through influencing the ability to distill signals from REIT share prices about the underlying property market. This determines whether agents in the Real Estate market are able to use the signals from REIT prices to determine which markets are underinvested in or over heated. Second through influencing whether REITs are able to identify good investments better than other Real Estate market participants. If REITs are not able to make better investments due to better information or management talent, agents in the Real Estate market can’t look at REIT share prices for signals about the underlying markets.
First some literature about characteristics that could influence the price signals is examined:
One characteristic that could be important is the liquidity of the REIT shares. If a REIT’s shares are less frequently traded the market could pay less attention to them. This will cause that either the pricing
13
signals of REITs are distorted due to the illiquidity premium or the information incorporated in the share price is less due to the low trading volume.
A less transparent underlying Real Estate market could also cause investors to be hesitant to invest in REITs. Danielsen et al. (2000) found evidence that stocks of more specialized REITs are more transparent and therefore, more liquid. If countries have more specialized REITs, their REIT sector could be more transparent. This would enable investors to obtain signals from REIT share prices more easily. This effect would also be captured by the liquidity measure.
Liu (2006) has developed a convenient liquidity measure that incorporates more aspects of illiquidity than the conventional liquidity measures. Trading speed, trading quantity and trading costs are also incorporated in this liquidity measure. Brounen (2005) performs a cross country analysis of the determinants and magnitude of share liquidity and finds that different measures of liquidity can alter the results greatly. He states that by using one liquidity measure the results could be biased. Therefore, two other liquidity measures, Turnover by Value and Turnover by Volume, will be used in this research.
Packer et al. (2013) also indicate that the variance of REIT share prices could influence the mitigating effect of the REIT market. A higher variance would make it more difficult to distill information about the underlying property market from REIT share prices because there is more uncertainty about the true value of the REIT share price.
The liquidity constraints of REITs could also have an influence on their investment activity. One would expect that the REITs which are more cash constrained have higher dividend payout requirements. This cash constraint could cause REITs to forego profitable investment opportunities.
Therefore, countries with lower required dividend payout ratios could be enabling REITs to exploit more investment opportunities. This in turn could cause a larger mitigating effect of REITs in a country’s Real Estate market.
14
Furthermore, The Dividend Payout Ratio could also affect the ability of REITs to make better investments than the average market participant. A higher Dividend Payout Ratio could result in REITs being liquidity constrained causing them to have to forego on profitable investment opportunities. Riddiough and Wu (2009) have investigated the effect of financial constraints on a REITs investment sensitivity to financial market frictions. They find that financially constrained firms substitute actual dividends paid for investments. Harmin and Hill (2008) state that the Dividend Payout Ratio isn’t a very good proxy for the amount of actual dividends paid, because firms can subtract depreciation from their EBIT while their assets only depreciate very slowly. They show that the dividend payout ratio of firms is often above 100 percent of a firm’s Funds From Operations (FFO). The Actual Dividend Payout Ratio could thus be a better proxy for the retained cash of REITs in a country. Bauer et al (2013) also find in their research that corporate governance is more important for the performance of REITs who have low Actual Dividend Payout Ratios.
Hartzell et al. (2006) have researched the influence of corporate governance within REITs on the ability of REITs to find good investment opportunities. These characteristics could be useful for my research because REITs that can better identify investment opportunities are more likely to invest less in overheated markets and are more capable to find good investments in cold markets. They find that REITs with stronger corporate governance and with a larger amount of shares owned by institutional investors react better to investment opportunities. Hartzell et al. (2006) state that this is due to the fact that institutional investors monitor managers better than normal shareholders decreasing the chances that managers will overinvest during good times. However there could be reverse causality since Devos et al. (2013) found that Institutional Ownership in REITs decreased steeply during the latest financial crisis. Therefore, it could be that because of the performance of REITs rises the Institutional Ownership rises instead of the other way around.
Lecomte et al. (2013) have tested the effect of corporate governance systems on the performance of REITs in Singapore. The rationale for using data on S-REITs is that REITs from the Asia Pacific region
15
are externally managed. Corporate governance might be more important in externally managed REITs since the risk of agency problems is higher in externally managed firms. He finds a significant effect of corporate governance on performance in S-REITs. However, there is a declining effect since the corporate governance has improved since the inception of the REIT market in Singapore. Still the difference between internally and externally managed firms could be more important in the comparison with REITs from the Asia pacific. All the REITs in our sample are all internally managed but perhaps differences in the transparency of the REIT structures could cause governance being more important in some countries in our sample.
Edelstein et al (2011) have investigated the role of country specific factors in determining REIT performance. They investigate the role of country specific corporate governance, the legal environment and a country’s accounting standards. They find evidence of a negative influence of the legal environment of a country and a country’s corporate governance on the performance of the country’s REIT sector.
Packer et al. (2014) have done a follow up study comparing the same REIT markets and name some important differences in REIT and REIT market characteristics between the countries. They name market characteristics age, IPO volume, variance and market capitalization, market betas and correlation with the stock market and returns as factors that could differ between the countries. The REIT characteristics management structure and Institutional Ownership could be characteristics that potentially play a role in explaining country level differences. They conclude that the exceptional long term performance in the US is probably caused by transparent governance mechanisms, low debt ratios and a high level of management talent in the sector.
Corporate focus could also play an important role in determining whether REITs mitigate the supply cycle. Eichholz (2007) examines the different REIT market characteristics in Europe and the United States in his research paper. He names corporate focus as one of the most important aspects in which the two differ. REITs in the US are more focused on one property type and diversified geographically
16
while European REITs diversify across property types. Eichholz (2007) states that this difference causes the US REITs to be more successful than their European counterparts. He also states that due to different regulations and tax regimes for REITs it is difficult for REITs to diversify geographically. Boer (2005) does a research at firm level of different REIT markets and also finds that corporate focus increases performance. One could expect that a specialized REIT is more capable of recognizing opportunities in a cold market and identifying overheated market. Furthermore, agents can get more accurate signals about the underlying property markets of these specialized REITs. Boer (2005) uses Herfindahl indexes and data from SNL financial for the United States which could be useful to measure corporate focus in this research.
2.4 Multicollinearity
Some results of previous researches show possible multicollinearity problems between some variables used in this research.
Hartzell et al. (2006) state that the corporate governance variables are correlated with each other. The exact influence of corporate governance of REITs on diminishing oversupply could therefore be better extrapolated if an index is used. There have been multiple indexes constructed by researchers to measure corporate governance. Aggarwal et al. (2009) have constructed an index for the 23 different countries using 44 governance attributes. The percentage of firms which meet the minimum requirements serve as an indicator of a country’s corporate governance score. Gompers et al. (2003) have constructed the G-index, which also measures the level of corporate governance but on the firm level. However, Bauer et al. (2010) state that the G-index isn’t comprehensive enough since it only uses two dimensions of corporate governance. Therefore, Bauer et al. (2010) use the CQG index instead since it entails eight different corporate governance measures.
Goebel et al. (2013) have focused on which characteristics of REITs influence REIT returns while controlling for momentum and interest rate changes. They show in their research that Institutional Ownership and Analyst Coverage are highly correlated within their sample of REITs (ρ=0.522).
17
Illiquidity and Analyst Coverage are also significantly correlated in their research (ρ= -0.383). The high correlations could cause problems when including all three variables in my regression. However, they are still able to use the three characteristics in their statistical tests and find significant results. In the research done by Hong et al. (2004) the authors encounter the same problem with Analyst Coverage and another variable of their interest, firm size. These two variables are highly correlated and in order to separately estimate the effect of these two variables they propose a method for circumventing this endogeneity problem. First they regress Analyst Coverage on size and some other variables and save the residuals of this regression. These residuals then represent a proxy of residual Analyst Coverage and can be used to estimate the effect of Analyst Coverage irrespective of size. The same method will be used if the correlation between Analyst Coverage and these two variables is too high in this sample. First Analyst Coverage will be regressed on Institutional Ownership and illiquidity and the residual Analyst Coverage will be used alongside illiquidity and Institutional Ownership in the regressions.
The variable lagged price could be correlated with the corporate governance variables. A large number of researches has been done after the effect of corporate governance on performance. Some of these researches find evidence that corporate governance enhances performance but the most recent researches find no statistical evidence. For example, Gosh and Sirmans (2003), find weak evidence that corporate governance variables enhance performance and Cannon and Vogt (1995) find that internally managed REITs outperform externally managed REITs. Hartzell (2006) does find evidence that corporate governance influences investment behavior of REITs but doesn’t find that corporate governance affect performance. Bianco et al (2007), Gompers et al (2003) and Bauer et al. (2010) don’t find any statistical evidence that corporate governance affects performance either. Lecomte (2013) finds evidence that corporate governance does affect performance in externally managed REITs However REITS in our sample are all internally managed and therefore, this doesn’t cause a risk when testing the model. No significant multicollinearity problems are to be expected between the variables REIT price and the corporate governance variables.
18
So this research will build further on the work of MacKinnon and Packer et al. (2013) in trying to find proof that an important REIT market exerts a mitigating influence on the supply cycle. This is done by first using Analyst Coverage as a proxy for REIT importance instead of market penetration and then testing the variables described above on the effect of the proxy for the mitigating influence on the Property Construction Cycle (Analyst Coverage interacted with a dummy variable indicating a strong or a weak economy). The variables that could influence the mitigating effect REITs have on new supply are:
Institutional Ownership
A governance index as described by Bauer et al. (2010) Age of the country’s REIT market
The Dividend Payout ratio The Actual Dividend Payout Ratio The Variance of the REIT index
The liquidity of REIT’s shares measured by the Turnover of the REIT’s shares, the trading volume of the REIT’s shares and the measure created by Liu (2006).
Due to data availability not all variables can be tested. The variables that will be tested are described in section 5.
3. Data
In section 2 potential institutional characteristics are identified which could enforce or weaken the mitigating effect the REIT market has on amplitude of the Commercial Property Construction Cycle. This section describes the characteristics that are tested and how the variables which are used as proxy for these characteristics are constructed. The time period for the total sample spans from 1997Q3-2015Q2 and contains data for the following six countries:
The United States Belgium
France
The Netherlands Germany
19
The five countries from Europe are chosen because these are the countries with the oldest REIT markets (EPRA, 2015). The characteristics of these REIT markets will be discussed in the next section (3.1). The United States is chosen because it has the oldest REIT market in the world, good data availability and because Packer et al. (2013) have found significant results for the US. A list of all the non-residential REITs in each country is made by filtering the SNL financial database for all the REITs except the REITs active in residential construction.
3.1 Real Estate Investment Trusts in sample countries
Now a short description of the history, the importance and the characteristics of Real Estate Investment Trusts in each of the sample countries is given to provide some context to the test results.
3.1.1 Development and performance of the REIT markets
Real Estate Investment Trusts (REITs) are listed property companies who invest almost primarily in Real Estate. They are exempt from paying taxes at the entity level but have to apply to strict regulations in return. These requirements differ per country but often include dividend payout requirements, having to primarily invest in Real Estate and Real Estate related activities and restrictions on REIT’s debt levels. A more specific description of the institutional characteristics of the REIT markets in the sample will be discussed later on.
The first REIT system is introduced in the United States in 1960 to make investing in Real Estate accessible for investors without large resources and as a new way for property owners and developers to access the capital market (NAREIT). Early adopters of the REIT system where also Australia (1961) and The Netherlands (1969). One of the reasons for the early adaptation of the REIT regime in The Netherlands was that the Dutch already had affinity with Real Estate and there was a large pressure from institutional pension funds to create a beneficial legal structure to hold Real Estate (Brounen, de Koning, 2013).
In the US the REIT market only grew very slowly after its inception in 1960 and after mortgage REITs contributed to the crisis in 1970s by providing easy and cheap construction finance the trust in the
20
REIT sector was very low in the years up until the 1990s. This caused that the REIT market in the US almost didn’t grow at all in these years.
Due to the bad reputation REITs had during the 1980s REITs didn’t have access to the large amounts of capital and therefore, didn’t experience the huge boom the direct investors in the commercial property market experienced in the 1980s. Therefore, the problems from the Savings and Loans crisis in the US didn’t have a large effect on the REIT markets. During the Savings and Loans crisis all private property owners experienced financial distress and couldn’t gain access to capital through the traditional channels like banks. There was a need to acquire capital in a different way and the REIT became an interesting investment vehicle again since Real Estate ownership could be securitized through these REITs. This REIT could then access the broader capital markets which were largely unaffected by the Savings and Loans crisis. The advantage of having access to capital markets when others didn’t was that REITs could purchase assets from financially distressed owners who didn’t have access to the capital market against huge discounts. The advantage of having access to liquidity when others didn’t, resulted in high growth potential and therefore, low cost of capital. This caused a large increase in the number of REITs and the market capitalization of these REITs, as can be seen from Figure 2.
Important to note is that since investors were still weary of REITs due to the boom and bust in the 1970s they only listed reliable companies with high value underlying assets. This caused that the REITs that were publicly traded in the US were trusted companies with good management and were the reason that there was a high trust in the US REIT market in the years that came. The concentration of the best management talent and high value assets on the balance sheet is an important characteristic of the US REIT market.
Figure 2 shows the development of the REIT office market share in the US to show the growth of the REIT market in each country. Figure 2 also shows the REIT price index to illustrate the performance of the REITs in the sample countries. Due to data availability it is only possible to show the office REIT
21
market penetration estimated by Packer et al. (2013) and not possible to show the market penetration for the total residential sector. Later on a different proxy for the development of the entire non-residential REIT market in a country is shown (Analyst Coverage) which is used as a proxy in this research.
As can be seen from Figure 2 US-REITs did indeed experience a large rise in market share from the 1990s up until the 2000s after which the market share has steadily grown. In the crisis years the market share dropped but quickly recovered to its initial level in the years thereafter. The performance of the US REITs follows the same pattern of steady growth until the crisis years, after which recovery sets in quickly. However, the REIT index is not yet back on its level before the crisis.
Figure 2 Development and performance U.S REIT market. (Source: Packer et al. (2013))
Figure 3 Development and performance Dutch REIT market. (Source: Packer et al. (2013))
As can be seen in figure 3 the market capitalization of REITs in The Netherlands also began to expand from the 1990s. The market penetration of the Dutch office REIT market has grown fast since its
22
inception but has declined since the mid-2000s after which it has shown slow growth towards its initial value. The Dutch REIT index has mainly decreased with exception of a short period of growth during the mid-2000s.
In 1995 the Belgian government also implemented a REIT like regime, the SICAFI regime, to boost the development of Belgian Real Estate. Collective investment was very popular in Belgium at the time but there was no tax regime to invest collectively in Real Estate, therefore the SICAFI was introduced. Another reason was to compete with similar regimes in Luxembourg and The Netherlands (EPRA, 2003). The Belgian REIT office market share experienced rapid growth in the beginning of 2000s and has also shown high growth with a small drop around the crisis years. The performance curve of the office REITs in Belgium looks similar to that of the market share until 2007. After 2007 performance drops and doesn’t return to its pre-crisis level but remains stable.
Figure 4 Development and performance Belgium REIT market. (Source: Packer et al. (2013))
The success of REITs in the US stimulated countries all over the world to develop a REIT market to boost their national Real Estate sector. In the early 2000s several Asian companies introduced the REIT to pump capital in their Real Estate markets after the financial crisis in 1997. Since this research does not focus on Asian countries due to data availability the Asian REITs will not be discussed extensively. However, the Asian REIT market became successful quickly and due to the success of the REITs in Asia the European countries started to implement REIT regimes as well.
Their motives were largely the same as those of the US. First to stimulate Real Estate investment in the national property markets by allowing small investors to invest in a diversified Real Estate
23
Portfolio, without the large transaction costs that come along with investing in direct Real Estate. Second the REIT structure causes the cost of capital to decline for Real Estate companies, makes these companies more competitive and stimulates an efficient allocation of capital (Brounen, de Koning, 2012).
The first European country to implement a REIT regime after the implementation of the SICAFI was France in 2003. France implemented the SIIC regime for three main reasons, first to increase their ability to compete with foreign Real Estate investment companies who did have tax benefits in their own country. Second the French government wanted to generate resources to help the French budget deficit through the “exit tax” levied on institutions that want to transform into REITs (EPRA, 2015). Finally, the property companies thought that changing into a tax transparent legal entity would cause the discount to net asset value the property companies suffered from to disappear (Brounen, de Koning, 2012). The market share of the French office REITs has shown steady growth since its inception. There is a small decline during the crisis but not as pronounced as in the United States. The performance curve of the French REIT follows the market share curve but peaks two years before the market share does after which a large drop during the crisis is visible. In 2011 another drop in prices is visible after which the REIT index is still not back at its pre-crisis level.
Figure 5 Development and performance French REIT market. (Source: Packer et al. (2013))
Four years after the French, the United Kingdom introduced the UK-REITs in 2007 (EPRA, 2015). The UK office market share shows a lot of volatility with a large peak shortly after its inception after which
24
a large drop appeared and thereafter a slowly declining market share until 2012. The UK office market has had a steep negative performance curve since its inception but has begun to recover slowly.
Figure 6 Development and performance UK-REIT market. (Source: Packer et al. (2013))
Germany also introduced the G-REIT in 2007 (EPRA, 2015).The German office REIT market has gained market share quickly after which it has remained steady until 2012. The German office REITs experienced a considerable downfall in performance shortly after their inception but also a quick recovery afterwards.
Figure 7 Development and performance German REIT market. (Source: Packer et al. (2013))
3.1.2 Characteristics
Table 1 indicates the characteristics of the REIT sector in our sample countries. The table shows the institutional characteristics of the REITs and how they differ between the countries. The Netherlands, Belgium and the United States are old REIT markets while France has had a REIT regime a bit longer and the other two markets, the United Kingdom and Germany are younger REIT markets. Thus for our European sample only 3 out of the 5 countries tested have had REITs for a long time. This could bias our results. The data used in this research also uses data on these companies from the time before
25
they became REITs to prevent the panel data set from thickening towards the end of the sample time period. A robustness check will be done to see whether the test results change when only observations with REITs are used and observations without Germany and the UK to see whether our results change.
Table 1 REIT Characteristics
Be UK NL GER FR US
Year first listed 1995 2007 1969 2007 2003 1961
Number of REITs 8 33 5 4 20 179
Sector Mkt Cap (millions) 8.054 201.959 28.145 38.010 113.125 446.8
%Global REIT index 0.55 6.95 3.05 0.17 1.85 54.8
Min. share capital (€ millions) 1.25 0.06 0 15 15
Leverage constraint 65% None 60% 45% None None
Pay out 80% 90% 100% 90% 95% 90%
Institutional holdings(2015) 5% 10% 12% 6% 9% 32%
Sources: EPRA(2013), DataStream
The United States is the largest REIT market by a landslide with a market cap which constitutes almost 55 percent of the global REIT market. The United Kingdom follows with nearly seven percent of the global REIT market. Third is the Dutch REIT market with three percent and fourth and fifth are France (1.85 percent) and Germany (0.17 percent). The United States also has a larger number of REITs with 179 REITs while the European countries have on average 54 REITs. More REITs could be an advantage since there are more pricing signals available about the state of the property markets where these REITs invest in.
There are also large differences in the minimum amount of share capital a REIT must have in order to be listed as REIT. In Germany and France, a considerable amount of share capital (15 million) is needed whereas in The Netherlands and The United States there is no minimum.
Belgium, The Netherlands and Germany also have leverage constraints ranging from 45 to 65 percent while the US, France and the UK don’t have leverage constraints. One could argue that higher leverage constraints causes REITs to underinvest due to liquidity shortage. However, REITs obtain their funding from the capital market enabling them to operate with lower leverage levels (Zhu, 2002). It could be the case that higher leverage works as a governance mechanism since managers of REITs with high
26
leverage levels know they will have more difficulty obtaining funding from the capital market when their debt levels are already high. It could be that this causes them to be more selective in their investments. The argument of leverage as a tax shield doesn’t apply either since REITs don’t pay taxes. Leverage could work as a governance mechanism since managers are constrained by interest payments, limiting their Free Cash Flow, preventing them to engage in wasteful investments. However, this is already the case due to high payout ratios REITs have to uphold.
All countries in our sample have high required payout ratios ranging from 80 to 100%. A high required Payout Ratio means that REITs have to return to the capital markets often. This works as a governance mechanism since the REITs have to obtain funding from the capital market each time they want to do large investments. If a manager underperforms a higher premium will be asked when he returns to the capital markets for his next investment.
Finally, the Institutional Ownership differs significantly ranging from 5-22 percent. The average Institutional Ownership in the United States is 22 percent while the European average is only 8.2 percent. A higher percentage Institutional Ownership could mean better governance as investigated by Bauer et al. (2010) and Hartzell et al. (2006) but it could also signal about the pricing performance of REITs at issuance (Brounen & de Koning, 2012).
3.2 Variables tested
Unfortunately, not all variables described in section 2.3 can be tested due to data availability. Tables 2 and 3 show the number of observations, the mean and the minima and maxima of the variables tested in this research.
3.2.1 Dependent variable
The dependent variable is the percentage change in construction volume. The dataset used to construct this variable differs from that of Packer et al. (2013) due to data availability. Data about the value of non-residential construction put in place in this research is taken from Eurostat for the European countries and from the census of The United States government for the United States. The
27
data from Eurostat as well as the data from the US government are seasonally adjusted ( (Eurostat, 2006) and (Census government, 2003)).
Table 2 Summary characteristics total sample
Variables
N Mean Min MaxTrading volume 549 698662 25 14040000 Price 583 891 115 8631 Volatility 509 32 1 473 ΔConstruction volume 472 0.10 -17.2 21.6 ΔConstruction cost 407 0.0022 -6 2 Market capitalization 609 43949 2 776889 Turnover 563 0 0 0.0204 Institutional Ownership 432 11 0 32 Gearing ratio 606 37 0 77 Payout ratio 508 1 0 800
Actual Payout ratio 591 1 0 200
Table 2 shows the characteristics for all the variables tested in this research for the total sample. Analyst Coverage: Total number of estimates for earnings per share in the next quarter or fiscal year per quarter. Trading volume: Number of shares traded per REIT per quarter. Price: DataStream REIT Price index per quarter. Volatility: Mean of the individual REIT variance per quarter. ΔConstruction volume: Change in completions compared to the previous quarter in percentages per quarter. ΔConstruction cost: Change in construction cost compared to previous quarter in percentages per quarter. Market capitalization: Mean of individual REIT market capitalization (price*common shares outstanding) per quarter. Turnover: Mean of the share turnover of individual REITs per quarter. Inst. Ownership: Mean of individual REIT ownership by institutions per quarter. Gearing ratio: Mean of the individual REITs total debts/total equity per quarter. Pay Out ratio: Mean of the individual REITs dividends per share/earnings per share per quarter. Act. Pay Out ratio: Mean of the individual REITs Common dividends/Funds from operations per quarter
Packer et al. (2013) use construction volume in their model and since one of the goals of this research is to compare whether our baseline model yields similar results value put in place needs to be adjusted to come to volume put in place. The European dataset needs no adjustments since the index measures value put in place at constant construction prices (Eurostat, 2006). Therefore, the changes in construction value put in place will reflect the changes in volume. However, the value of non-residential construction put in place for the US is calculated with current prices and therefore, also includes effects of changes in the construction cost (Census government, 2003). To adjust this index, the change in the turner construction cost index for commercial properties is subtracted from the change in value put in place to get the change in construction volume for the US.
28
Table 3 REIT characteristics by country
Variables Belgium United Kingdom Netherlands Germany France United States
Coverage 23 155 57 135 135 725 Trading volume 2,4 822 26 0.503 34 2,830 Price 141 3,596 319 191 334 633 Variance 3 137 13 8 12 20 ΔConstruction volume 0.09 0.38 -0.070 0.026 -0.12 0.39 ΔConstruction cost -0.016 0.086 -0.002 -0.002 0.077 -0.136 Market capitalization 4000 30000 11000 544 20000 192,500 Turnover 0.0006 0.0024 0.0024 0.0002 0.0008 0.0055 Inst. Ownership(%) 5 10 12 6 9 22 Gearing Ratio(%) 38 39 37 18 36 52 Payout Ratio(%) 71 47 89 119 62 101
Actual Payout Ratio(%) 69 78 76 69 63 51
N 61 61 61 61 72 70
29
Finally, the percentage change in completions is used instead of the log change in completions. Although the difference in log completions with small changes is nearly equal to the percentage change in completions the model has higher explanatory power, measured in a higher R-Squared, when the percentage change of completions is used. This is peculiar since Table 2 and 3 don’t show large changes in completions.
Looking at the percentage change of volume in construction illustrated in Figures 8-12 for Europe and the US it can be seen that the amplitude of the European commercial property construction line is smaller than that of the US construction. However, the change in construction in this figure is an average for Europe so it can be expected that this amplitude is lower. As can be seen from table 3 the inter country differences are quite large. The fact that the mean change in construction volume for The Netherlands and France is negative while the mean for the UK is large and positive shows these opposite effects. The figures for the individual countries in Appendix B show that our data on construction volume don’t show a large decrease in the amplitude of the construction cycle. This was found in the research of Packer et al. (2013). In the European sample only the change in construction in Germany and France seem to be decreasing but the UK and The Netherlands have experienced more change in completions since the introduction of the REIT market. Belgium did not experience any significant increase or decrease in the change in completions.
3.2.2 REIT market characteristics
One way in which this research differs from the research done by Packer et al. (2013) is that it uses a different proxy for the attention paid by agents in the Real Estate market to a country’s REIT market, Analyst Coverage. The total Coverage in each quarter is taken from the I/B/E/S historical summary file and is the total number of fiscal year 1 and 6 estimates indicated by analysts in that quarter for each individual stock of a country’s REITs. Fiscal year 1 estimates in I/B/E/S are the estimates done by analysts about a REIT’s earnings per share for the next fiscal year. Fiscal year 6 estimates are the estimates done by analysts about REITs earnings per share for the next fiscal quarter. This variable interacted with a dummy variable indicating whether the economy is in a good or bad economic state
30
is the variable that tests whether a more important REIT market has a mitigating influence on the amplitude of the property cycle. The hypothesized effect of Analyst Coverage is the same as the effect of REIT market penetration in Packer et al. (2013). A REIT market which is followed by more analysts will have a stronger mitigating influence on the amplitude of the property cycle. The reason is that Analyst Coverage proxies attention paid by the investors in a country’s Real Estate market to its REIT market. More attention means a higher chance that the agents in the Real Estate market will distill information about the underlying property markets from REIT share prices. The relationship between the number of estimates and the change in completions is illustrated in Figure 8. As can be seen from figure 8 the more analysts provide an estimate for the earnings per share of a REIT in the US the more the volatility of construction volume seems to decrease. In Europe the volatility is too low to infer a clear pattern from the figure but it does seem that since more analysts started to provide estimates about the earnings per share of the REIT market the volatility has decreased. However, just as was the case with the change in completions, the number of analyst estimates differ a lot per country as shown by Table 2 and 3. For example, Belgium has a mean analyst coverage of 22 analyst estimates per quarter while the US has an average of 725 estimates. Looking at the individual European countries shown in appendix B it can be seen that there does seem to be a pattern of increasing analyst coverage with decreasing change in construction for Germany and France but not for the other countries in our European sample.
31
Figure 8 Analyst coverage and change in completions
Sources: I/B/E/S, Eurostat, Census US Government. Note: Analyst Coverage: Sum of Fiscal year 1 and 6 estimates provided by Analyst Coverages per quarter. Change in construction volume: the change in construction volume compared to the previous year. Change in completions: Change in volume put in place compared to previous quarter.
Liquidity can be measured in lots of different ways and there is still discussion about what the appropriate proxy should be. Liquidity in our sample is proxied by two different methods. First by the trading volume of the shares of the REIT index in DataStream. Second the mean of the Turnover of the individual stocks is used. As stated in part two of this paper it is expected that a more liquid market for REIT shares will increase the mitigating effect of the REIT market since agents can more easily extrapolate information from REIT share prices. Table 3 shows that the two measures differ greatly in magnitude as well as variance between the countries. The measure for trading volume is large and differs a lot between countries while the measure for turnover is much smaller and also differs less between countries. Looking at the magnitude of the liquidity variables one would expect that the mitigating effect will be the largest in the UK and Germany since turnover as well as trading volume are the largest for these countries. The high turnover would enable agents in the property market to get the most information about the underlying real estate markets from REIT share prices since these prices incorporate more information about the true value of the share price.
Variance is measured as the quarterly mean of the daily variance in the REIT price index composed by DataStream. The expected effect of this variable on the mitigating influence of the REIT market on the property cycle is negative since higher volatility means higher uncertainty about the correct price of a
-6 -4 -2 0 2 4 6 0 400 800 1200 1600 2000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 % #
Analyst Coverage and change in completions in Europe and
the United States
Analyst coverage US Analyst Coverage Europe Change in completions europe Change in completions US
32
REIT share. This troubles the agent in the Real Estate market to extrapolate signs from the REIT share prices about the state of the underlying Real Estate assets. Table 3 indicates that the Belgian REIT index has the lowest variance and that the UK REIT index has a much higher variance than the other REIT markets. The UK has one of the highest turnovers as well and as discussed above turnover has the opposite effect therefore, the strongest effect will determine whether the mitigating effect is reinforced or weakened.
Market capitalization is the sum of the market capitalizations of the country’s individual REITs, where market capitalization is defined as a REIT’s share price times its common shares outstanding. The market capitalizations of the individual REITs are taken from DataStream and are shown in table 3. Higher market capitalizations mean a larger REIT market and it can be expected that a larger REIT market gets more attention from other agents in the Real Estate market. Therefore, it is expected that there is a positive relationship between market capitalization and the mitigating influence of the REIT market on the property cycle. Figure 9 does indeed show that when market capitalization in the US rises the volatility of the completions becomes less. However, for the European sample this is not the case. On an individual country level Germany, France, Belgium and the UK seem to have a less volatile construction volume when the market capitalization of the REIT market is high. The Netherlands seems show the opposite effect of a higher market capitalization combined with higher volatility in completions.
33
Figure 9 Market Capitalization and change in completions
Note: Market Capitalization: Sum of the market capitalizations of the country’s individual REITs, Change in completions: same as in figure 8. Sources: Datastream, Eurostat, Census US government
3.2.3 REIT characteristics
A country’s dividend payout ratio is tested with two different variables. The first measure is the Actual Dividend Payout Ratio which is calculated by dividing the Funds From Operations of each REIT by its common dividends paid and taking the quarterly mean of all the country’s REITs as dividend payout ratio. Quarterly data on the Funds From Operations and common dividends paid are taken from DataStream. Second the standard Payout Ratio is used which is the dividends per share divided by the earnings per share. Data on the earnings per share and the dividends per share are taken from DataStream as well. A higher Payout Ratio means that managers are more restricted to take part in wasteful investments due to a smaller cash flow. Since not being able to pay dividend is heavily punished by the market managers are stimulated to make good investment decisions. Assuming agents in the market know this a higher Dividend Payout Ratio would mean better investments and agents in the market that pay better attention to its REIT market for signals about the state of the underlying property market. Figure 10 and 11 show that, while the US has a higher Payout Ratio, the European countries have a higher Actual Payout Ratio. The Actual Payout Ratio seems to show a clearer pattern in combination with the decreased volatility for the US changes in construction volume. On the individual country level there is no clear pattern visible between the change in completions and the Actual Payout Ratio and the Payout Ratio for the European countries. As can be
-6 -4 -2 0 2 4 6 0 200000 400000 600000 800000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 % %
Market Capitalization and change in completions for Europe
and the US
Market Capitalization US Market Capitalization Europe change in completions US Change in completions Europe
34
seen from Table 3 the Payout Ratio and the Actual Payout Ratio differ a lot. For example, in the UK the Payout Ratio is twice as high as the Actual Payout Ratio. This is different than hypothesized by Harmon and Hill (2008) who found that REIT Actual Payout Ratio is higher than the Payout ratio. Both variables will be tested to see which characteristic better explains the change in completions.
Figure 10 Actual Payout Ratio and change in completions
Notes: Actual Pay Out Ratio: Quarterly Mean of Funds From operations of each REIT divided by its common dividends paid. Change in completions: same as in figure 8. Sources: Datastream, Eurostat, Census Government
Figure 11 Payout Ratio and change in completions
Note: Payout ratio: Quarterly mean of Dividends per share divided by the earnings per share. Change in completions: same as in figure 8.
Sources: DataStream, Eurostat, Census US government
The Gearing Ratio is the quarterly mean of a country’s individual REIT’s debt divided by their total equity. The data regarding total debt and equity for the individual REITs are taken from DataStream.
-6 -4 -2 0 2 4 6 0 50 100 150 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 % %
Actual Pay Out Ratio and the change in completions for
Europe and the US
Actual Payout Ratio US Actual Payout Ratio Europe Change in completions US Change in completions Europe
-6 -4 -2 0 2 4 6 0 50 100 150 200 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 % %
Payout Ratio and change in completions for Europe and the
US
Payout Ratio US Payout Ratio Europe
35
A higher Gearing Ratio means that the individual REITs have a higher average Debt Ratio. This means that managers have a smaller Free Cash Flow due to interest payments and are more restricted in taking part in wasteful investments due to the fear of not being able to pay the interest on the loans. Figure 12 shows that the Gearing Ratio has remained quite constant over the years in the United States. In Europe it can be seen that the gearing ratio has risen after the crisis. On the individual country level in Table 3 it can be seen hat most countries have a Gearing Ratio around 35 percent but that Germany has a low Gearing ratio of 18 percent and the US has a high gearing ratio of 52 percent. As explained in section 2.3 the effect of the Gearing ratio is ambiguous depending whether it works as a governance tool or liquidity constraint. From the individual country graphs no clear pattern can be seen of countries with a high or low gearing ratio having less variance in their change in completions.
Figure 12 Gearing Ratio versus completions
Note: Gearing Ratio: the quarterly mean of the country’s individual REIT’s debt divided by their total equity. Change in completions: same as in figure 8.
Sources: DataStream, Eurostat, Census US government
3.2.4 Governance variables
The governance variables used by Hartzell et al. (2006): Institutional Ownership, Block Ownership and Directors and Owners Ownership were the ones that would fit best for this research on the basis of earlier research. Due to data availability only Institutional Ownership will be examined. However, Institutional Ownership is the only governance variable suggested by Packer et al. (2013) that could
-6 -4 -2 0 2 4 6 0 20 40 60 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 % %
Gearing Ratio and the change in completions for Europe and
the US
Gearing Ratio US Gearing Ratio Europe Change in completions US Change in completions Europe
36
influence the mitigating effect of REITs on the amplitude of the property cycle. The data about Institutional Ownership is taken from Thomson One Share Ownership Database filtered for ownership by institutions. As already stated in the previous chapter Institutional Ownership in our sample is highest for the United States although the European REIT markets have seen a rise in Institutional Ownership as well. The lines plotted in Figure 13 show the change in completions for Europe and the United States. A rise in ownership by institutions for US-REITs coincides with small changes in completions where Institutional Ownership drops during troughs and peaks in the change in completions. This support the hypothesis that a higher percentage of Institutional Ownership of REITs increases the mitigating effect of these REITs on the amplitude of the commercial property cycle. Europe has experienced a stable rise in Institutional Ownership so it is hard to make inference about the effect of Institutional Ownership on completions. Looking at the individual country level in Appendix B it can be seen that Germany and France seem to experience lower levels of volatility when Institutional Ownership rises. However, the United Kingdom and The Netherlands seem to show the opposite effect, volatility increases when Institutional Ownership rises. Table 3 also shows the Institutional Ownership percentages per country and as said in section 3.1.2 Institutional Ownership is higher in the US compared to the European countries
Figure 13 Institutional Ownership and change in completions
Note. Institutional Ownership: Mean percentage ownership by institutions. Change in completions: same as in figure 8. Sources. Thomson Reuters Share ownership database. DataStream, Eurostat, Census US government
-6 -4 -2 0 2 4 6 0 10 20 30 40 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 % %
Institutional Ownership and the change in completions for
Europe and the US
Institutional Ownership US Insitutional Ownership Europe Change in completions US Change in completions Europe
