Corporate Real Estate Holding and Stock Returns: Testing Alternative Theories with International Listed Firms

Abstract

This study examines the relationship between corporate real estate (CRE) holdings and stock returns before and after the Global Financial Crisis (GFC). We find that (1) the United States and the United Kingdom show a negative relationship before the GFC and positive after the GFC. (2) Firms that pay positive tax or have positive R&D investments are not systematically different from the full sample. This finding cannot support the "scarce capital" theory or the tax incentive explanation, but it is consistent with the “empire building” theory. After the GFC, financial constraints tightened, and both CRE holding and stock returns dropped. (3) European (excluding the United Kingdom) sample shows a positive relationship in the pre-crisis period. This finding is compatible with the "illiquidity premium" theory. However, the association becomes inconclusive in the post-crisis period. (3) The Japanese sample shows a negative association between CRE and stock returns in the pre-crisis period, like the United States and the United Kingdom. However, the relationship becomes statistically insignificant in the post-crisis period, consistent with the theory of financial constraint tightening after the GFC.

Appendices

Appendix A

This appendix mainly discusses two strands of the literature: the motives to own CRE and the relationship between CRE holdings and firm performance.

Motives to Own CRE

In the main text, we indicate that there are different motives to own CRE beyond production needs. Each purpose could result in another nexus between CRE holdings and returns. The first motivation for CRE holding is "empire building." Due to weak corporate governance, firms may over-invest in CRE and make less investment, and R & D. It leads to a negative correlation in CRE holding and stock return. Based on Real Estate Investment Trusts (REITs) data in the U.S., Sirmans (1999) hypothesizes that specific sets of corporate governance mechanisms are needed for firms with substantial real estate holding. Sing and Sirmans (2008) employ a sample of 228 stocks listed in Singapore and formally reject the hypothesis that corporate governance mechanisms are independent of a firm's real estate ownership. Thus, the result is consistent with Sirmans (1999). Coles et al. (2006) show a strong causal relationship between management incentives and firms' behavior on investment policy, debt policy, and risk-taking. Employing a sample of U.S. listed corporations, Du et al. (2014) find no evidence for a return-enhancing role for CRE holdings, suggesting that CRE holdings are a form of managerial "empire building." In firms with weak governance, over-investment in the CRE is more likely to occur, and higher CRE holdings are associated with lower returns to shareholders. Dong et al. (2012) employ the Listed Chinese firms and find that corporate governance, state ownership, and preferential tax policy explain the CRE holding.

The second motive is related to CRE's collateral channel effect, which will lead to a positive nexus between CRE holdings and returns. Firms use CRE as inputs of production and collaterals to raise debt for investment, and firms could benefit from the appreciation of CRE holdings (Bernanke & Gertler, 1989, 1990; Chaney et al., 2012; Gan, 2007a, 2007b; Kiyotaki & Moore, 1997). For instance, Ogawa et al. (1996) and Ogawa and Suzuki (1998) find that the land price fluctuations in Japan would affect corporate investment behaviors. Gan (2007a) finds that, during the early 1990s, the investment rate of an average firm in Japan drops by 0.8 percentage points resulting from a 10% drop in land value. Chaney et al. (2012) also find that firms' investments in the U.S. are substantially affected by the shocks to the value of real estate holdings. For example, during 1993–1997, a $1 increase in collateral value leads the representative U.S. corporation to raise its investment by $0.06.

Relationship between CRE Holdings and Firm Performance

After discussing the motives to own CRE, we review the literature on the relationship between CRE holdings and firm performance. The first strand of research employs the idiosyncratic return (Alpha) and systematic risk component (Beta) to measure firm performance. Table 11 provides a summary of their main findings. For example, in the case of the United States, Deng and Gyourko (1999) employ firm-level data for 717 companies from 57 different non-real estate industries in the U.S. in 1984–1993 and find that firms with high degrees of real estate concentration and high Beta experience lower returns. However, employing a similar sample period (1985–1994), Seiler et al. (2001) find no relationship between CRE holdings and systematic risk and excess return.

On the other hand, Tuzel (2010) finds that CRE holdings positively affect abnormal returns in non-real estate firms in the U.S. from 1963 to 2003. In the case of other economies, Brounen and Eichholtz (2005) explores international CRE effects using samples from 18 industries and nine countries in the year 1992, 1995, 1998, and 2000, and find a significantly negative relationship between CRE holdings and systematic risk, while no association between CRE holdings and idiosyncratic risk. Finally, Cheong and Kim (1997) find that a listed manufacturing firm's CRE holdings had no significant effect upon the return-on-investment in its stocks from 1987 to 1991 in Korea.

On the other hand, Liow and Ooi (2004) use entirely different measures of firm performance. They evaluate stock return by two value-based metrics: economic value added (EVA), and market value added (MVA). Based on the data of listed non-real estate firms in Singapore from 1997 to 2001, the authors find that CRE hurts non-real estate firms' EVA and MVA. Based on the data of listed non-real estate firms in Singapore from 1997 to 2001, the authors find that CRE hurts non-real estate firms' EVA and MVA.

Another strand of literature explores the impact of CRE holding on other aspects of a firm's operation. For instance, Zhao and Sing (2016) empirically test the relationship between CRE holdings and the production risk of firms, which is measured by the volatility of the output per unit of capital. The publicly listed U.S. firms' data from 1984 to 2011 prove that CRE holding is significantly and negatively correlated with a firm's productivity risks. As a result, firms with high productivity risk (more volatile firms) hold a relatively lower level of the CRE.

Table 11 Previous literature on the relationship between CRE holdings and returns

Appendix B

This section presents a simple model of a firm, which can engage in R&D investment and corporate real estate (CRE) investment.

There are two periods, $t=\mathrm{0,1}$. At time 0, a risk-neutral firm endowed with an amount of initial capital K and a linear technology to produce can choose to invest in R&D investment, which would boost productivity and invest in CRE, whose valuation in time 1 can be different. For simplicity, we assume that all these investment decisions are discrete. More specifically, the firm which invests $D$ units of capital, $0<D<K$ has a probability $p$ to be successful, $p\in [\mathrm{0,1}]$, and its productivity would increase from $A$ to $Ag$, $A>0,g>1$. If the firm fails, the productivity remains to be $A$. On the other hand, the firm can also acquire 1 unit of CRE, which costs ${P}_{h}$ units of capital in period $0$, ${P}_{h}>0$. In period 1, the valuation of the CRE would become ${P}_{h}\varepsilon$, where $\varepsilon$ represents an idiosyncratic valuation shock. The shock has finite and positive support, $\varepsilon \in \left[{\varepsilon }_{L},{\varepsilon }_{H}\right], 0<{\varepsilon }_{L}<{\varepsilon }_{H}<\infty$. We assume that the first moment is also finite, $0<E\left(\varepsilon \right)<\infty$. We assume that the valuation shock is independent of the risk involved in the R&D if R&D efforts are ever be made. Alternatively, the firm may rent CRE from the market at a rate ${R}_{h}$, ${0<R}_{h}<{P}_{h}$.¹⁶ And to produce in period 1, the firm needs to pre-install capital in period $0$. To simplify the analysis, we assume that $K-D-{P}_{h}>0$. We introduce two indicator functions to represent the firm's R&D and CRE investment decisions. Formally,

$$I^R=\begin{Bmatrix}1&firm\;invests\;in\;R\&D\\0&otherwise\end{Bmatrix},$$

$${I}^{H}=\left\{\begin{array}{cc}1& firm\ invests\ in\ CRE\\ 0& otherwise\end{array}\right\}.$$

Thus, the firm which maximizes the expected value of the profit is

$$max.E\left(\pi \right)$$

where $E\left[\pi \left({I}^{R},{I}^{H}\right)\right]=\left\{\left[pAg+\left(1-p\right)A\right]{I}^{R}+\left(1-{I}^{R}\right)A\right\}*\left[K-{I}^{R}D-{I}^{H}{P}_{h}-\left(1-{I}^{H}\right){R}_{h}\right]+{I}^{H}{P}_{h}\varepsilon$. This formula looks more complicated than it is. For instance, the profit for a firm engaging in both R&D and CRE investment is simply

$$E\left[\pi \left(\mathrm{1,1}\right)\right]=\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[K-D-{P}_{h}\right]+{P}_{h}E\left(\varepsilon \right).$$

Similarly, the profit for a firm engaging in R&D but not CRE investment is simply

$$E\left[\pi \left(\mathrm{1,0}\right)\right]=\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[K-D-{R}_{h}\right].$$

The profit for a firm engaging in CRE but not R&D investment is simply

$$E\left[\pi \left(\mathrm{0,1}\right)\right]=A*\left[K-{P}_{h}\right]+{P}_{h}E\left(\varepsilon \right).$$

The profit for a firm engaging in neither R&D nor CRE investment is simply

$$E\left[\pi \left(\mathrm{0,0}\right)\right]=A*\left[K-{R}_{h}\right].$$

Since the investment decisions are discrete, we simply compare different options pairwise.

Lemma 1. If $\frac{1}{g-1}*\left[\frac{E\left(\varepsilon \right)}{A\left(1-\frac{{R}_{h}}{{P}_{h}}\right)}-1\right]>0,$ $E\left[\pi \left(\mathrm{1,0}\right)\right]>E\left[\pi \left(\mathrm{1,1}\right)\right]$ if and only if $p$ is sufficiently large.

Proof. The proof is straightforward. Observe that

$$\begin{array}{l}E\left[\pi \left(\mathrm{1,0}\right)\right]-E\left[\pi \left(\mathrm{1,1}\right)\right]\\ =\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[K-D-{R}_{h}\right]-\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[K-D-{P}_{h}\right]-{P}_{h}E\left(\varepsilon \right)\\ =\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left({P}_{h}-{R}_{h}\right)-{P}_{h}E\left(\varepsilon \right)\end{array}$$

>$0$ if and only if $p>{p}_{1}^{*}$, where ${p}_{1}^{*}=\frac{1}{g-1}*\left[\frac{E\left(\varepsilon \right)}{A\left(1-\frac{{R}_{h}}{{P}_{h}}\right)}-1\right]$. Since $\frac{1}{g-1}*\left[\frac{E\left(\varepsilon \right)}{A\left(1-\frac{{R}_{h}}{{P}_{h}}\right)}-1\right]>0$, ${p}_{1}^{*}>0.$

Notice further that in practice, $\frac{{R}_{h}}{{P}_{h}}$ is very small. Hence, if $\left(\varepsilon \right)>A$, then it is likely that$\frac{1}{g-1}*\left[\frac{E\left(\varepsilon \right)}{A\left(1-\frac{{R}_{h}}{{P}_{h}}\right)}-1\right]>0$.

Notice further that if our condition is violated, for instance, $E\left(\varepsilon \right)<A\left(1-\frac{{R}_{h}}{{P}_{h}}\right)$. In that case, it means that every firm which satisfies the stated assumption would find it better to invest in R&D only, rather than both R&D and CRE investment.

Lemma 2. If $\left[\frac{A\left[D+{R}_{h}\right]-{P}_{h}\left(A-E\left(\varepsilon \right)\right)}{A\left(K-D-{R}_{h}\right)}\right]>0,$ $E\left[\pi \left(\mathrm{1,0}\right)\right]>E\left[\pi \left(\mathrm{0,1}\right)\right]$ if and only if $p$ is sufficiently large.

Proof. The proof is again straightforward. Observe that

$$\begin{array}{l}E\left[\pi \left(\mathrm{1,0}\right)\right]-E\left[\pi \left(\mathrm{0,1}\right)\right]=\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[K-D-{R}_{h}\right]-\left\{A*\left[K-{P}_{h}\right]+{P}_{h}E\left(\varepsilon \right)\right\}\\ =p\left(g-1\right)AK-\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[D+{R}_{h}\right]+{P}_{h}\left(A-E\left(\varepsilon \right)\right)\\ =p\left(g-1\right)A\left[K-D-{R}_{h}\right]-A\left[D+{R}_{h}\right]+{P}_{h}\left(A-E\left(\varepsilon \right)\right)\end{array}$$

$>0$ if and only if $p>{p}_{2}^{*}$, where ${p}_{2}^{*}=\frac{1}{\left(g-1\right)}*\left[\frac{A\left[D+{R}_{h}\right]-{P}_{h}\left(A-E\left(\varepsilon \right)\right)}{A\left(K-D-{R}_{h}\right)}\right]$. Since $\left[\frac{A\left[D+{R}_{h}\right]-{P}_{h}\left(A-E\left(\varepsilon \right)\right)}{A\left(K-D-{R}_{h}\right)}\right]>0$,${p}_{2}^{*}>0.$

First, notice that $A\left(K-D-{R}_{h}\right)>0,$ and $\left(g-1\right)>0$ by assumption. Hence, it suffices to study the term $A\left[D+{R}_{h}\right]-{P}_{h}\left(A-E\left(\varepsilon \right)\right)$. And $A\left[D+{R}_{h}\right]-{P}_{h}\left(A-E\left(\varepsilon \right)\right)>0$ iff ${P}_{h}E\left(\varepsilon \right)<A{P}_{h}-A{R}_{h}-AD$. Notice also that

P_hE(ε):

return from investing in CRE.

AP_h:

addition return from investing in R&D only.

AR_h:

loss from investing in R&D only.

AD:

return from investing in R&D.

Thus, the RHS $A{P}_{h}-A{R}_{h}-AD$ is the net return from R&D only, while the LHS ${P}_{h}E\left(\varepsilon \right)$ is net return from investing in CRE. If LHS < RHS, then the firm will invest in R&D only when the probability of success in R&D is sufficiently high.

Lemma 3. $E\left[\pi \left(\mathrm{1,0}\right)\right]>E\left[\pi \left(\mathrm{0,0}\right)\right]$ if and only if $p$ is sufficiently large.

Proof. The proof is again straightforward. Observe that

$$\begin{array}{l}E\left[\pi \left(\mathrm{1,0}\right)\right]-E\left[\pi \left(\mathrm{0,0}\right)\right]=\left\{\left[pAg+\left(1-p\right)A\right]\right\}*\left[K-D-{R}_{h}\right]-A*\left[K-{R}_{h}\right]\\ =p\left(g-1\right)A\left[K-{R}_{h}\right]-\left\{\left[pAg+\left(1-p\right)A\right]\right\}D\\ =p\left(g-1\right)A\left[K-D-{R}_{h}\right]-AD\end{array}$$

$>0$ if and only if $p>{p}_{3}^{*}$, where ${p}_{3}^{*}=\frac{1}{\left(g-1\right)}*\left[\frac{D}{\left(K-D-{R}_{h}\right)}\right]$. Notice that ${p}_{3}^{*}>0$. It means that some firms would find it optimal not to make any investment, should they inherit a probability of success low enough.

Based on the three lemmas, we can define a new quantity ${p}_{1}^{**}=max\left\{{{p}_{1}^{*},{p}_{2}^{*},p}_{3}^{*}\right\}.$ And for $p>{p}_{1}^{**}$, it is necessary that $E\left[\pi \left(\mathrm{1,0}\right)\right]=argmax\left\{E\left[\pi \left({I}^{R},{I}^{H}\right)\right]\right\}$. In other words, it means that investing in R&D but not in CRE is the best strategy if the probability of success in R&D is sufficiently high.

Appendix C

Tables 12, 13

Table 12 Different measures of CRE in the previous literature

Table 13 Panel regression with the corporate governance index