THE DETERMINANTS OF REIT CASH HOLDINGS

THE DETERMINANTS OF REIT CASH HOLDINGS William G. Hardin, III † Florida International University Michael J. Highfield †† Mississippi State University Matthew D. Hill ‡ University of Arkansas – Fort Smith G. Wayne Kelly ††† Mississippi State University First Draft: August 28, 2005 Current Draft: January 15, 2007 JEL Classification: G12; G14; G24 Keywords: REIT, Cash, Liquidity, Cash Flow, Working Capital William G. Hardin is an Associate Professor of Real Estate at Florida International University. Address: Department of Finance, College of Business Administration, Florida International University, 11200 SW 8th Street, Miami, Florida, 33199-0001. Office phone: 305.348.2680. Fax: 305.348.2751. Electronic mail: hardinw@fiu.edu. †† Michael J. Highfield, CFA is an Assistant Professor of Finance at Mississippi State University. Address: Department of Finance and Economics, College of Business and Industry, Post Office Box 9580, Mississippi State University, Mississippi State, Mississippi, 39762-9580. Office phone: 662.325.1984. Fax: 662-325-1977. Electronic mail: mhighfield@cobilan.msstate.edu. ‡ Contact Author: Matthew D. Hill, CTP is an Assistant Professor of Finance at the University of Arkansas – Fort Smith. Address: Department of Finance and Economics, College of Business and Industry, Post Office Box 3649, Fort Smith, Arkansas 72913-3649. Office phone: 479.788.7764. Fax: 479.788.7818. Electronic mail: mhill@uafortsmith.edu. ††† G. Wayne Kelly is an Associate Professor of Finance at Mississippi State University. Address: Department of Finance and Economics, College of Business and Industry, Post Office Box 9580, Mississippi State University, Mississippi State, Mississippi, 39762-9580. Office phone: 662.325.2562. Fax: 662-325-1977. Electronic mail: wkelly@cobilan.msstate.edu. This paper is derived in part from the Ph.D. dissertation of M. Hill at Mississippi State University. The authors thank Chuck Beauchamp, Randy Campbell, Greg Nagel and seminar participants at the 2006 American Real Estate Society (ARES) Annual Meeting Doctoral Consortium for helpful comments and suggestions. Financial support and research assistance for this project was provided by the College of Business and Industry and the Warren Chair of Real Estate at Mississippi State University. All errors remain ours. † THE DETERMINANTS OF REIT CASH HOLDINGS Abstract We examine the factors influencing the cash holdings of REITs with the view that the REIT industry should yield new information regarding the drivers of corporate cash policy due to the unique operating conditions faced by REITs. Our results show that REIT cash holdings are inversely related to excess cash flow, excess dividends, capital market access, leverage, operating performance, and the use of credit lines and are directly related to asymmetric information and growth. Overall, it appears as though REIT managers elect to hold little cash to increase transparency, thereby reducing the cost of external capital. Furthermore, our results suggest that externally-advised REITs hold significantly more cash than their internally-advised peers. THE DETERMINANTS OF REIT CASH HOLDINGS 1. Introduction Damadoran (2006) reports that, on average, Real Estate Investment Trusts (REITS) carry cash and equivalents equal to 1.57 percent of total assets, considerably less than the 17 percent average reported by Opler, Pinkowitz, Stulz, and Williamson (1999) for a sample of non-REITs. 1 Despite this substantial difference, the liquidity policies of REITs have been ignored by the finance literature, likely because internal capital accumulation by REITs is limited exogenously by mandatory dividend payments. In an effort to fill this void in the literature, we examine the factors influencing the cash holdings of REITs with the view that the REIT industry should yield new information regarding the drivers of corporate cash policy due to the unique operating conditions faced by REITs. The relatively small cash positions held by REITs is at least partially a consequence of the legal impairment of their ability to retain capital internally via a dividend requirement. A REIT maintains federal tax-exempt status if it pays shareholders at least 90 percent of taxable income via dividends, among other restrictions. 2 The extent to which the mandatory distribution of earnings reduces the ability of REITs to accumulate capital internally is the extent to which it forces them to seek expansion funding from capital markets. However, this restriction may understate REITs’ actual ability to accumulate cash since the mandatory dividend is calculated as a portion of taxable income, from which depreciation is already deducted without a reduction in cash flow. Given that depreciation is an especially large non-cash deduction for real estate firms, most REITs pay out much more in dividends than the minimum ninety percent required by law. Wang, Erikson, and Gau (1993) and Bradley, Capozza, and Seguin (1998) analyze REIT dividend policy and show the REITs in their samples have average dividend payout ratios (dividends scaled by net income) of 165 and 190 percent, respectively. These results suggest that REIT managers have discretionary capital at their disposal, implying the possibility of a free cash flow problem. More REIT Cash Holdings 2 importantly, these findings imply that REIT managers choose not to accumulate cash, despite their ability to do so. In addition to the dividend requirement, the predictability and strength of REIT cash flows likely contribute to their low level of liquid assets. REIT cash flows are relatively stable by design; that is, property leases are typically structured as long-term, binding payment contracts. This increased cash flow stability reduces the need to hold cash for precautionary purposes. Furthermore, as stated previously, the real estate assets on which these leases are written provide inherently high depreciation expenses, reducing net income but not cash flow. Taken together, the strength and predictability of REIT cash flows mitigates the marginal benefit of holding liquid assets. REITs also present a unique and straightforward opportunity to test the impact of agency conflicts on cash holdings because many financial decisions for REITs are made by advisors. Thanks to tax reform in 1986, REITs now choose explicitly between internal or external advisement. External advisors hold no ownership stake in the REITs they advise, so their interests are not automatically aligned with shareholders, a condition inviting agency problems. External advisors may exploit the principal-agent relationship by excessive cash retention, similar to the problem described by Jensen (1986). Subsequently, advisement type may influence the level of cash held by REITs. The models employed in this study to predict REIT cash holdings include factors known to influence the cash policies of non-REITs: leverage, operating performance, and the cost of external finance. However, an advantage of focusing on the REIT industry is the avoidance of financial variables having different meanings, which frequently occurs in most inter-industry corporate studies. This relative homogeneity enables us to develop and test models with REIT-specific financial variables. Our study provides several contributions to the corporate cash holdings and REIT-related literatures. First, it is likely that the REIT operating environment has important implications for cash policy, but the motives driving REITs’ relatively weak demand for liquid assets have not been examined. Second, explicit testing for the impact of advisement on cash holdings is a direct way to examine the freecash flow problem, via REITs, and will contribute to the literatures of both corporate cash holdings and REIT Cash Holdings 3 agency theory. Third, we provide evidence of the impact line of credit use has on REIT cash holdings. Due to data constraints, the effect of credit lines on corporate cash holdings has not been examined. Last, we use an original approach to model the effects of REIT discretionary cash flow and dividends on liquidity policy. In summary, our results show that REIT cash holdings are inversely related to excess cash flow, excess dividends, capital market access, leverage, operating performance, and the use of credit lines and are directly related to asymmetric information and growth. Overall, it appears as though REIT managers elect to hold little cash to increase transparency, thereby reducing the cost of external capital. Furthermore, our results suggest that externally-advised REITs hold significantly more cash than their internally-advised peers. 2. 2.1. Literature Review Determinants of Corporate Cash Holdings Several studies have investigated the determinants of corporate cash holdings. Kim, Mauer, and Sherman (1998) develop a theoretical model for the determinants of cash and test it empirically for a sample of non-REITs. They find that cash varies positively with the degree of asymmetric information, cash flow volatility, and the expected return on future investment opportunities. They also find that cash varies inversely with cash flow, leverage, probability of bankruptcy, and operating performance. Overall, these findings support three motives for holding cash developed by Keynes: transactions, precautionary, and speculative motives. Opler, Pinkowitz, Stulz, and Williamson (1999) examine corporate liquidity and report findings generally complementing the results provided by Kim, Mauer, and Sherman (1998). The authors offer two theories to explain the corporate liquidity management decision: static tradeoff and financing hierarchy. The static tradeoff theory holds that the optimal cash position is determined by weighing the marginal costs of cash against the marginal benefits. Hence, cash is made more or less valuable by the operating and capital market access characteristics of the firm. Meanwhile, REIT Cash Holdings 4 the financing hierarchy theory posits that firms adhere to a pecking order of financing choices, where cash is the preferred method of financing, followed by debt and equity, respectively. Their results show that liquidity varies directly with growth opportunities, cash flow, cash flow variability, and research and development expenditures and inversely with size, net working capital, leverage, and dividends. The authors conclude that the evidence supports the static tradeoff theory relative to the financing hierarchy theory. Ozkan and Ozkan (2004) investigate the effect of corporate governance, via managerial ownership, board structure, and financial institution ownership, on cash holdings for a sample of industrial firms in the United Kingdom. The relations between cash and the non-governance independent variables echo the findings of Kim, Mauer, and Sherman (1998) and generally affirm those presented by Opler, Pinkowitz, Stulz, and Williamson (1999). Regarding the governance variables, Ozkan and Ozkan (2004) find that only managerial ownership has a statistically significant impact on cash holdings. Specifically, the authors find that managerial ownership and cash are non-monotonically related. These findings show the success of certain types of governance in reducing the agency problems associated with the accumulation of cash. In a similar vein of research, Yan (2006) explores the determinants and implications of mutual fund cash holdings via a theoretical static tradeoff model where managers weigh the opportunity cost of holding cash against the expected trading cost of liquidating stocks to meet investor redemptions. The evidence provided by Yan (2006) shows that cash is positively related to the volatility of the fund cash flows and inversely related to the size of the fund. However, the results do not support the hypothesis that funds with more successful managers, those with better stock picking skills, hold less cash. 2.2. Agency Costs of Excess Liquidity The primary benefit of corporate liquidity is its use as internal funding for value-increasing projects. This benefit is not the only possible outcome. Erosion of shareholder wealth via agency costs associated with increasing managerial discretion over the excess funds is also possible. With the increased REIT Cash Holdings 5 discretion provided by excess liquidity, managers are increasingly able to accept unnecessary diversifying acquisitions or projects that deepen managerial entrenchment and/or provide additional perquisites. These help diversify the manager’s portfolio and/or expand their wealth at a cost to owners. Of course, this problem is not new. Jensen and Meckling (1976) state that the desires of management and the principals of the firm will undoubtedly diverge, giving principals an incentive to incur costs of monitoring or levering the firm, to reduce the costs of agency conflict. These agency costs represent a residual loss to shareholders. Building on Jensen and Meckling (1976), Jensen (1986) applies agency theory to firms with free cash flow. Jensen’s free cash flow theory states that internally generated funds in excess of those needed to fund positive net present value projects (free cash flow) erodes shareholder wealth through the agency costs of free cash flow. Jensen suggests two remedies for the free cash flow problem: debt contracts, where repayment serves as a disciplining force, and the market for corporate control, where cash held is viewed as a war chest. 2.3. REIT-Related Literature: Advisement Type Congress created REITs as passive investment vehicles, allowing individuals to invest in real estate similar to mutual funds. Initially, outside advisors were mandated to manage day-to-day REIT operations. This external advisement created agency problem since the external advisors could have no ownership stake in the REIT they advised. Fortunately, Congress changed tax codes pertaining to REITs in 1986 allowing for internal advisement. By definition, internal advisors have an ownership stake in the REIT they advise, which more closely aligns their interests with those of shareholders. Capozza and Seguin (2000) compare the performance of externally-advised and internallyadvised REITs. They find that externally-advised REITs underperform their internally-advised counterparts based on stock returns and Tobin’s Q. Interestingly, they show that property level cash flows are similar for both management structures, but interest expense consumes a much greater proportion of cash flow for externally-advised REITs. This finding suggests that externally-advised REITs use not only more debt, but they also use more expensive debt. Cappoza and Seguin (2000) posit that this result is likely caused by external advisor compensation schemes. The compensation of external advisors is based REIT Cash Holdings 6 on either the size of the asset base under management or the level of property cash flows. Hence, external advisors have an incentive to issue debt at rates exceeding the firm’s weighted average cost of capital. Building on the results of Capozza and Seguin (2000), Ambrose and Linneman (2001) hypothesize that externally-advised REITs will respond to market pressures and transition to the internal advisement structure because the interests of shareholders and managers are better aligned with the internal advisor structure. In fact, they find that the majority of the REITs sampled already use internal advisors. Also, the total equity market capitalization of externally-advised REITs grew by an annual compound rate of 22 percent, but the total equity market capitalization of internally-advised REITs grew by an annual compound rate of 71 percent during the sample period. The authors attribute the differences in growth to a superior ability for internally-advised REITs to raise capital, an increase in the number of internally-advised REIT IPOs, and the conversion of many externally-advised REITs to the internal advisement structure. In closing, Ambrose and Linneman (2001) suggest that external advisors mimic the financial policies of internally-advised REITs, citing increasingly conservative debt policies of externally-advised REITs. Additionally, the authors find that internally-advised REITs have significantly lower dividend payout ratios (dividends-to-funds from operations), supporting the notion that internally-advised REITs are similar to non-REIT operating companies pursuing growth strategies. 3. 3.1 Hypothesis Development Excess Cash Flow The likelihood of underinvesting is reduced as cash flow increases because managers are no longer forced to seek the approval of projects from external capital providers. Consequently, the benefit provided by cash holdings is diminished, leading us to expect an inverse relation between cash and cash flow. This hypothesis is consistent with the estimated inverse relation between cash and cash flow presented by Kim, Mauer, and Sherman (1998) and Ozkan and Ozkan (2004) for samples of non-REITs. Furthermore, for REITs, it is plausible that a negative relation between cash and cash flow exists because REIT managers optimally choose to not retain cash from cash flow to receive the benefits accruing from REIT Cash Holdings 7 increased monitoring provided by the capital markets. Since REITs rely on capital market funding to grow via acquisitions, transparency is a key concern for REITs. Meanwhile, accumulating cash by retaining cash flow would tend to reduce the degree of transparency regarding a REIT’s projects. So, a negative relation is consistent with managers of REITs reducing the agency costs associated with the free cash flow problem, at least regarding the accumulation of cash but not necessarily regarding unnecessary acquisitions, etc. Prior REIT literature generally measures the impact of cash flow by using funds from operations (FFO). However, it should be noted that FFO is not the best measure to estimate the impact of cash flow on REIT cash holdings because a portion of cash flow is already committed through the dividend restriction. Thus, overall cash flow may not have bearing on the level of liquid assets held. That is, only cash flow in excess of the dividend payment, discretionary cash flow, is available to reduce the likelihood of underinvestment. Thus, in our study excess cash flow is modeled by subtracting the mandatory dividend, 90 percent of net income from FFO. We follow the SNL definition of FFO identified as net income excluding gains or losses from sales of property, plus depreciation and amortization, and after adjustments for unconsolidated partnerships and joint ventures, scaled by total assets. 3 Regardless, we expect to find the same negative relation between cash and cash flow documented in the in the non-REIT studies discussed above. 3.2 Excess Dividends Opler, Pinkowitz, Stulz, and Williamson (1999) provide evidence that the payment of a dividend significantly reduces cash holdings. After all, since dividends reduce discretionary cash flow, increased dividends reduce the ability to accumulate cash. Dividends can also be used to provide financial flexibility when internal funds are limited and/or external financing is costly as corporate managers can elect to reduce dividend payments. However, as stated previously, the dividend policy of REITs is unlike that of non-REITs due to the 90 percent payout requirement mandated by law to maintain tax exempt status. While this appears to limit the flexibility of REITs, Wang, Erickson, and Gau (1993) and Bradley, Capozza, and Seguin (1998) show that equity REITs routinely pay out much more in dividends than REIT Cash Holdings 8 necessitated by the dividend restriction. Thus, dividends paid in excess of that required by the payout requirement could have been retained by the firm and used as a source of liquidity. Accordingly, cash is expected to be inversely related to excess dividends, defined as dividends paid minus the mandatory dividend payment and scaled by total assets. 3.3 Asymmetric Information Myers and Majluf (1984) show that asymmetric information increases the benefit provided by cash holdings because external financing is more costly for firms with high informational asymmetries. Increased costs of external finance raise the likelihood of foregoing profitable investment opportunities, while liquid assets allow firms to take advantage of profitable opportunities when presented. In fact, Kim, Mauer, and Sherman (1998), Opler, Pinkowitz, Stulz, and Williamson (1999), and Ozkan and Ozkan (2004) find that cash varies directly with the degree of asymmetric information. Although unrelated to the topic of cash management, Lang and Litzenberger (1989) and Wang, Erickson, and Gau (1993) use Tobin’s Q to measure REIT informational asymmetries. In most REIT studies Tobin’s Q is defined as the sum of market value of equity, short-term and long-term debt, and preferred equity divided by total assets. Thus, building on the evidence provided by non-REIT studies, we adopt Wang, Erickson, and Gau’s (1993) use of Tobin’s Q to test the hypothesis of a direct relation between cash and the degree of asymmetric information., 3.4 Capital Market Access Fazzari and Petersen (1993) show that larger firms are less likely to face borrowing constraints than smaller firms because they have better capital market access. This privilege allows them to borrow via bond offerings, commercial paper issues, or lines of credit. The likelihood of underinvesting decreases as capital market access improves, and, accordingly, the benefit of liquid assets decreases. Interestingly, Opler, Pinkowitz, Stulz, and Williamson (1999) find a significant inverse relation between cash and size, but Kim, Mauer, and Sherman (1998) find that cash and size are not significantly related. Kim, Mauer, and Sherman (1998), Opler, Pinkowitz, Stulz, and Williamson (1999), and Ozkan and Ozkan (2004), use total assets to proxy size. We argue that revenue is a superior measure to represent REIT Cash Holdings 9 capital market access for REITs since the market value of real estate typically exceeds the book value of real estate. Revenues are inflated to 2005 dollars using the consumer price index. Since cash and size are expected to be non-linearly related, firms must hold some minimum level of cash, the natural logarithm of revenue is used. Following the logic of Fazzari and Petersen (1993) and previous non-REIT evidence presented by Opler, Pinkowitz, Stulz, and Williamson (1999) , we expect an inverse relation between cash holdings and firm size as measured by total revenues. 3.5 Growth High growth combined with costly external financing can cause underinvestment. Unfortunately, firms with high growth rates are frequently subjected to the scrutiny of capital markets as they need resources to grow. When capital is issued, the cost of capital is often high due to uncertainty. Thus, managers may opt to accumulate cash holdings to fund future growth, thereby avoiding costly external finance. Using the annual asset growth rate as the proxy for growth, cash is expected to be directly related to growth. 3.6 Leverage Kim, Mauer, and Sherman (1998), Opler, Pinkowitz, Stulz, and Williamson (1999), and Ozkan and Ozkan (2004) estimate an inverse relation between cash and leverage. After all, when internal funds are scarce, firms can issue debt to generate cash. Leverage can also be used as a bonding mechanism to reduce the agency costs caused by the free cash flow problem. Accordingly, using the total debt-to-total assets ratio to measure leverage we follow the evidence presented by non-REIT studies cited above and hypothesize an inverse relation between cash and leverage. 3.7 Operating Performance REITs and other firms with superior operating performance have a legitimate case for retaining funds, assuming that past successful projects and efficient asset utilization can be expected in the future. For relatively transparent REITs, increased ROA allows greater retention of cash flows and provides for the provision of additional capital as the firm evidences property acquisition and management skills. Therefore, increased operating performance may send a positive signal to investors regarding the REIT Cash Holdings 10 capabilities of management, and, subsequently, investors may allow managers with superior operating performance to hold more cash. Accordingly, a direct relation between cash holdings and operating performance is expected. Operating performance is measured using return on assets, defined as net income available to shareholders scaled by total assets. 3.8 Governance REIT advisors select the properties to purchase, broker transactions, and seek financing for acquisitions. Each function has implications for cash policy. Since 1986, REITs choose between internal and external advisement. External advisors have no ownership in the REIT, blunting the alignment of interests between REIT management and owners. Additionally, external advisors may work for institutions, real estate firms, or REITs other than the REIT which they advise. The various agency problems caused by the use of external advisors are documented by Sagalyn (1996). Accordingly, external advisors may use cash holdings to entrench themselves, exploiting the principal-agent relationship by holding more cash than necessitated by the scale of operations. Therefore, an inverse relation between cash and internal advisement is expected. Ambrose and Linneman (2001) examine differences in financial characteristics of externally and internally-advised REITs. They find that the internal advisement structure dominates the REIT industry, as many externally-advised REITs have responded to capital market pressure by switching to the internal advisement structure. Additionally, they find that their sample of externally-advised REITs is smaller than the internally-advised sample. This suggests that externally-advised REITs are more financially constrained than internally-advised REITs, which, in turn, implies that cash is more valuable to externally-advised REITs. As with the agency explanation, the financial constraint explanation also predicts an inverse relation between cash and internal advisement. The structure and length of the REIT property leases, which vary by property type, affect underlying cash flows. Accordingly, property focus is used to control for cash flow volatility, and it is expected that cash holdings vary directly with cash flow volatility. Similarly, Bradley, Capozza, and Seguin (1998) measure REIT cash flow volatility via property-focus Herfindahl indices. REIT Cash Holdings 11 3.9 Credit Lines The relation between liquid assets and credit lines has not been examined in the corporate finance literature because of data limitations concerning corporate credit lines. However, credit line use may be a preferred method of financing, relative to corporate cash holdings, for a couple of reasons. First, the explicit cost of borrowing from a credit line is generally cheaper than long-term debt and may also be lower than the opportunity costs of holding cash. Next, lines of credit can be used for unexpected contingencies to smooth operational cash flow, similar to cash holdings, except that credit line drawdowns would only occur if the contingency actually occurred. Last, transparency is a key concern for REITs, as they rely heavily on capital markets to fund capital acquisitions. Because of monitoring provided by short-term debt lenders, REITs can reduce the degree of asymmetric information regarding their projects, thereby reducing the cost of external capital. Alternatively, cash holdings increase agency costs as they decrease monitoring. Campbell, DeVos, and Spieler (2006) show that REIT credit lines are generally used by less financially constrained firms with improved capital market access. Due to these reasons, we expect that REIT cash holdings will be negatively related to the use of credit lines. 4. 4.1 Data and Methods Data Description and Source The initial sample includes all REITs covered by the SNL REIT Datasource database over the period 1993 to 2005 (3,170 firm-years). The SNL database covers financial and property data for equity, mortgage, and hybrid U.S. REITs. Accordingly, non-equity REITs are deleted (578 observations) from the sample, and missing accounting (1,967) data further restrict the sample size. The final sample consists of 625 firm-year observations for 129 equity REITs over the 1998 to 2005 period. Thus, the dataset is an unbalanced panel with maximum, minimum, and mean number of individual REIT appearances in the dataset equal to 8 panels, 1 panel, and 4.8 panels, respectively. REITs which were acquired, merged, or delisted remain in the sample up to the time of delisting, reducing the likelihood of survivorship bias. Also, observations are allowed to enter, leave, and then re-enter the sample if the data requirements discussed above are met. REIT Cash Holdings 12 The descriptive statistics for the final sample of equity REITs are found in Table 1. The distribution of excess dividends relative to total assets is distributed symmetrically across the sample of REITs with the sample mean and median equal to 0.91 percent and 0.90 percent respectively. The average REIT in the sample pays out $15 million in excess dividends per year, illustrating that the payment of excess dividends is a non-trivial issue. 4 We note that dividend paying REITs with negative earnings are problematic within the context of our definition of excess dividends because these REITs are not required to pay dividends during the year in which earnings are negative. However, depreciation expense affords REITs with negative earnings the ability to pay dividends. Using our definition of excess dividends, REITs with negative earnings would have inflated levels of excess dividends. Simply deleting these observations would not be appropriate, as this might bias the sample, and the inferences would only be applicable to REITs with positive earnings. This issue is resolved by setting earnings equal to zero for the year in which earnings are negative. 5 That is, all dividends paid by a REIT with negative earnings are considered to be excess dividends. Thus, our interpretation of excess dividends is maintained without creating a sample bias. The REITs in the sample generate a substantial amount of excess cash flow, an average of $56 million per year for each REIT, but this is still not enough capital to grow the firm’s asset portfolio in a meaningful way. The distribution of excess cash flow as a percentage of total assets is slightly leftskewed with a mean of 2.70 percent and a median of 2.97 percent. The average Tobin’s Q is 1.16, which is similar to the average Q reported by Campbell, DeVos, and Spieler (2006). The means of the cash-tototal assets, revenue, total debt-to-total asset, and return on total assets variables (2.19 percent, $314 million, 53 percent, and 3 percent, respectively) are very similar to the sample averages of the same variables reported by Ghosh and Sirmans (2006); however, the mean of asset growth rate, 17 percent, is much lower than that reported by Ghosh and Sirmans (2006). This difference is attributable to the temporal changes in the data series as the Ghosh and Sirmans data are for the 1999 to 2000 period and the data in this study are for the longer 1998 to 2005 period. REIT Cash Holdings 13 Panels A, B, and C of Table 2 show the distribution of the sample across time, property focus, and advisement type, respectively. The observations are distributed across property type in a pattern that is reflective of the values of the underlying properties as components of the asset class. Panel C of Table 2 shows the distribution of the sample by advisement type. Approximately 92 percent of the observations are from internally managed firms. The externally advised firms pay out more of their FFO than the internally advised REITs as would be expected considering their higher potential agency costs. The excess dividend payments as a percentage of total assets for internally managed REITs is greater than the same ratio for externally advised REITs. This indicates, when taken with the dividend to FFO ratio, that internally advised REITs generate higher performance and retain more of their FFO than externally advised REITs even while exhibiting excess dividend payments. Table 3 presents the correlation matrix for variables used in this study. The values reported are Pearson correlation coefficients, and significant correlations are marked in bold. The correlations show evidence supporting many of the expected relations. Cash holdings are inversely correlated with excess cash flow, excess dividends, leverage, and internal advisement. Meanwhile, cash holdings are positively correlated with increased informational asymmetries growth, and operating performance. The direct correlation between cash and operating performance lends support to the view that managers with better operating performance are allowed to hold more cash. 4.2 The Determinants of REIT Cash Holdings We use cash divided by total assets as the dependent variable in the estimation of the models predicting REIT cash holdings. Cash is scaled by total assets to establish a proportional balance sheet relationship and also to mitigate the potential for the largest firms to dominate the results. We estimate the following empirical model: CASH i ,t = β 0 + β 1 EXFFOi ,t + β 2 EXDIVi ,t + β 3 Qi ,t + β 4 LN ( SIZE ) i ,t + β 5 GROWTH i ,t + β 6 LEVERAGEi ,t + β 7 ROAi ,t + β j Control Variables j ,t + ε i ,t . (1) REIT Cash Holdings 14 In Equation (1) excess cash flow, EXFFO, is calculated as funds from operations minus the mandatory dividend and divided by total assets. Excess dividends, EXDIV, is dividends paid minus the mandatory dividend divided by total assets. Tobin’s Q, represented by the variable Q, is the market value of equity plus total debt and scaled by total asset. REIT size is measured using fiscal year revenues in millions and in 2005 dollars, SIZE. The annual growth rate in assets is represented by GROWTH, while LEVERAGE is defined as total debt divided by total assets. Operating performance, ROA, is net income available to common shareholders divided by total assets. Model (1) is augmented to examine the impact of governance on REIT cash holdings by including ADVTYPE, an indicator variable equal to one if the REIT is advised internally, otherwise zero. We also examine the impact of lines of credit on REIT cash holdings by adding LOC, defined as the amount drawn from credit lines divided by total credit lines available to the firm. 6 Our set of control variables includes indicator variables for property focus. We reduce the number of property categories listed in SNL to seven by combining those with similar leasing structures. REITs classified as Retail have property focus types of Mall, Retail, or Shopping. REITs with property focus types of Diversified, Manufactured Homes, or Other are lumped into one category named “Other.”. The seven property focus categories are: Hotel, Industrial, Multi-family, Office, Other, Retail, and Selfstorage. 7 Given the lease structure for apartments and the property type’s greater transparency, apartments are used as the base case. Also included in the set of control variables are annual dummy variables controlling for time-specific factors influencing cash holdings in the pooled OLS models. 4.3 Econometric Considerations Tables 4 through 7 present pooled OLS, fixed-effects, random-effects, and cross-sectional regression results for the determinants of REIT excess dividends, respectively. Each econometric methodology is used to estimate three variations of Model (1). When applicable, t-statistics, reported parenthetically, are calculated using Newey-West (1987) robust standard errors to correct for serial correlation and heteroskedasticity. REIT Cash Holdings 15 Undoubtedly, some of the variation in REIT cash policy is caused by unobservable firm-specific factors, which, if correlated with the independent variables, implies the pooled results suffer from heterogeneity bias. In an effort to eliminate this bias, the fixed and random-effects methodologies are used. The decision to determine which methodology is preferred depends primarily on the correlation between the unobserved effect and the independent variables; however, it is difficult to determine whether the unobserved heterogeneity is correlated with any of the independent variables ex ante. If the unobserved effect and any of the independent variables are correlated, the fixed-effects methodology is preferred as the random-effects results will be inconsistent. However, no correlation between the unobserved heterogeneity and independent variables means the fixed-effects results are inefficient. A drawback of the fixed-effects approach is that all time-invariant independent variables are eliminated from the model. This is problematic within the context of our study because prior research shows that cash flow volatility is a key determinant of corporate cash policy (Kim, Mauer, and Sherman (1998) and Opler, Pinkowitz, Stulz, and Williamson (1999)), and our proxy for cash flow volatility is REIT property type. Therefore, we report the results for both the fixed and random-effects models and calculate Hausman’s test to determine the significance of the correlation between the unobserved heterogeneity and independent variables. The results from the fixed and random-effects models are reported in Tables 5 and 6, respectively. The t-statistics for the fixed-effects results are calculated using White’s (1980) robust standard errors. We also estimate cross-sectional regressions by estimating the regression equation on the timeseries averages (between estimator) of the financial variables for each REIT. As shown in Table 7, the between estimator methodology mitigates problems arising from fluctuations in firm characteristics, outliers, and serial correlation of the error terms. White’s (1980) robust standard errors are used to calculate the t-statistics for the cross-sectional models. The sample consists of 625 firm-years for the pooled, fixed-effects, and random-effects models, while 129 unique REITs are used to estimate the cross-sectional models. Dummy variables for sample years are used to control for time-effects in the pooled, fixed-effects, and random-effects models. Three REIT Cash Holdings 16 models are applied using the four methods noted. Model 1 is the base model. Model 2 includes the advisor type variable. Model 3 includes the line of credit variable. 5. 5.1 Results Pooled OLS Results The pooled OLS results for Models 1 and 2 provided in Table 4 show that REIT cash holdings have a negative relation with excess cash flow. The t-statistics, reported parenthetically, are calculated using Newey-West (1987) robust standard errors to correct for serial correlation and heteroskedasticity. The excess cash flow coefficient is significant at the 5 and 10 percent levels for Models 1 and 2, respectively. This finding suggests that the benefit of holding cash is reduced when excess cash flow increases because the likelihood of underinvesting is reduced. However, the significance of the excess cash flow coefficient is not robust once line of credit is included in the model. The sample of REITs studied has discretionary cash flow, so the free cash flow problem does exist, to some degree, for the REITs in this sample. 8 However, the estimated negative coefficient for excess cash flow suggests that the free cash flow problem seems to be controlled within the context of cash accumulation. That is, these results show that REIT managers do not stockpile cash holdings despite that the firm will then be forced to seek funding from the capital markets. It is likely that seeking funding from the capital markets is exactly what REIT managers intend to do, and this appears to be an optimal response given the importance of transparency to REITs. This finding corresponds to the inverse relation between cash and cash flow estimated by Kim, Mauer, and Sherman (1998) and Ozkan and Ozkan (2004). In the first two models, excess dividends exert a significant negative impact on cash. This result suggests that when internal funds are limited, REIT managers can elect to reduce dividend payments in excess of the restriction, indicating that excess dividends are a potential source of liquidity. In short, the ability to cut excess dividends reduces the advantages of holding cash. This result also provides support for the theory regarding the estimated relation between cash and excess cash flow. Given that REIT managers have discretionary cash flow at their disposal, they choose to pay excess dividends instead of accumulating cash, alleviating the free cash flow problem and increasing transparency. REIT Cash Holdings 17 Tobin’s Q represents the degree of asymmetric information faced by REITs in capital markets. The estimated relation between cash and Tobin’s Q is positive and significant for each of the models. As predicted, it appears that REITs with higher informational asymmetries hold cash to avoid costly external financing. This finding parallels the positive relation between cash and asymmetric information for nonREITs reported by Kim, Mauer, and Sherman (1998), Opler, Pinkowitz, Stulz, and Williamson (1999), and Ozkan and Ozkan (2004). The pooled results show limited support for an inverse relation between cash and size, as the size coefficient is only marginally significant at the ten percent level for Model 1. As predicted, cash has a significant positive relation with growth. This supports the view that REITs plow back cash to fund future growth in assets. This result also lends credibility to the estimated relation between cash and informational asymmetries, as growth options and asymmetric information are likely linked. The relation between growth and cash is robust across all three models. Cash varies inversely with leverage in Models 1 and 2 with the estimated leverage coefficients having statistical significance at the five ten percent levels, respectively. This finding corroborates the views that leverage is a substitute for cash and that debt is successful in mitigating cash accumulation. This result reiterates the cash-leverage relation estimated by Kim, Mauer, and Sherman (1998), Opler, Pinkowitz, Stulz, and Williamson (1999), and Ozkan and Ozkan (2004). Not surprisingly, leverage is not significant once line of credit is added to the model. This is likely due to collinearity between the variables. Return on assets has a statistically significant negative impact on cash for Models 1 and 2. REITs with superior operating performance hold less cash. Meanwhile, we predicted that cash and return on assets would have a direct relation. It may be that return on assets is a proxy for the opportunity cost of holding cash; therefore, as the opportunity costs of cash increase, firms will optimally hold less cash. The data suggest that the opportunity cost effect outweighs the flexibility that cash provides firms with superior management. REIT Cash Holdings 18 Dummy variables categorizing property focus classifications are used to control for cash flow volatility. The model includes an intercept, so one category, Multi-family, is left out of the model. The results show that REITs categorized as Hotel and Other hold significantly more cash than firms categorized as Multi-family. However, the inferences are somewhat limited as neither is significant for model two. The financial constraint and agency theories predict that externally-advised REITs hold more cash than those using internal advisors. Table 2c presents the mean of cash holdings after splitting the sample according to advisement type. The majority of the sample, 92 percent, is comprised of internallyadvised REITs, echoing the finding by Ambrose and Linneman (2001) that the incidence of external advisement usage has decreased. The null hypothesis of mean equality of the cash ratio by advisor type is rejected at the 5 percent level. This test result suggests that externally-advised REITs hold significantly more cash than internally-advised REITs, which is further corroborated by the correlation between cash and advisor type presented in Table 3. Model two presents the results after advisor type is added to the baseline model. The estimated negative relation between advisement type and cash is significant at the five percent level. Furthermore, the estimated coefficient for advisor type has economic interpretations as it shows that, on average, the sample of internally-advised REITs has a cash-to-total assets ratio that is about 3 percent less than externally-advised REITs. Given that the average asset base for the REITs in this sample is $1.85 billion, this implies that the sampled externally-advised REITs hold $55.5 million more in cash, on average, than their internally-advised counterparts. This is a non-trivial amount considering the opportunity cost of holding cash. The estimated impact of advisement type on cash supports the financial constraint and agency theories. This evidence supports the view that the incentives of internal advisors, as compared to those of external advisors, are better aligned with those of shareholders. Thus, external advisors are more likely to use cash to exploit the principal-agent relationship by hoarding costly cash in hopes of furthering entrenchment. Similarly, it is likely that externally-advised REITs are more financially constrained REIT Cash Holdings 19 (because of the aforementioned agency problems) than internally-advised REITs. Thus, externallyadvised REITs hold more cash as a reaction to the cost of external finance. Model three presents the results once credit line usage is added to the base model. As predicted, REIT cash holdings are negatively related to the use of credit lines. The relation between the variables is statistically significant at the five percent level. This finding supports the view that short-term debt financing serves as a substitute for cash, suggesting that financially constrained firms hold more cash than their financially unconstrained counterparts. Similarly, Almeida, Campello, and Weisbach (2004) find that financially constrained non-REITs hold more cash than financially unconstrained non-REITs. However, their proxies for financial constraint did not include the availability/use of credit lines. The importance of the availability of short-term debt on cash holdings is evident by assessing the statistical significance of other independent variables in Model three. Once the line of credit variable is included, the statistical significance disappears for excess cash flow, excess dividends, asymmetric information, leverage, and operating performance. It seems that access to and use of credit lines is the primary determinant of REIT cash policy. This result makes sense for a couple of reasons. First, the advantages of investing in cash are reduced when the firm has access to low cost short-term debt. Second, the REIT is monitored when the line of credit is accessed, which increases transparency and reduces the cost of external capital. 5.2 Fixed-Effects and Random-Effects Results The fixed-effects and random-effects regression results appear in Tables 5 and 6, respectively. First, we test to determine the statistical significance of the correlation between the unobserved heterogeneity and the independent variables using Hausman’s test (1978) for the base model. A comparison of the results from the base model estimated using the fixed and random-effects methodologies generates a Hausman test statistic of 50.37 and a p-value of less than 0.00. This suggests that the null hypothesis of no correlation between the unobserved firm effect and the independent variables should be rejected at any conventional level of significance, implying that the fixed-effects methodology should be used for this data. However, cash flow volatility should, theoretically, be an REIT Cash Holdings 20 important determinant of cash holdings and since our proxy for cash flow volatility is a dummy variable, this variable will be eliminated from the fixed-effects model due to the lack of variation across time. Hence, we also present the random-effects results. A secondary reason we present random-effects results is that we would like to generalize the results to all equity REITs, not just those REITs for which we have data. In Table 5, the fixed-effects results are presented using the full dataset. After accounting for the unobserved heterogeneity, excess dividends, Tobin’s Q, growth, and operating performance are no longer statistically significant. However, the significant negative relation between cash and excess cash flow is robust, even after including credit lines. In fact, the significance of the excess cash flow coefficient increases from ten percent to one percent upon the addition of the line of credit variable. Surprisingly, leverage retains its statistical significance after accounting credit line use. Similar to the pooled results, the credit line variable is statistically significant at the one percent level. Overall, even after accounting for REIT-specific fixed-effects, our results indicate that REIT cash holdings are inversely related with excess cash flow, leverage, and usage of credit lines. The random-effects results are presented in Table 6. These results are consistent with the pooled results. Excess cash flow, excess dividends, Tobin’s Q, growth, leverage, operating performance, advisor type, and credit line usage are statistically significant at retain the same sign as for the pooled results. One difference is that size has a statistically significant coefficient for models one and two and is negatively signed, as expected. This result suggests that REIT managers respond rationally to improved capital market access by holdings less cash. The estimated sign of size can also be viewed as evidence of the important role of monitoring. Larger firms are monitored more than smaller firms because they more frequently tap the capital markets for financing. Increased monitoring should reduce the likelihood of unnecessarily accumulating cash holdings. Opler, Pinkowitz, Stulz, and Williamson (1999) report an inverse relation between cash holdings and size. Unlike the pooled results, none of the property focus categories are significantly different from the base case. REIT Cash Holdings 21 5.3 Cross-Sectional Results Finally, the cross-sectional results are presented in Table 7. A few differences between the pooled and cross-sectional results are worth mentioning. First, using the cross-sectional methodology, excess cash flow and leverage are no longer significant. Like the other results, the credit line variable appears to be an important factor influencing REIT cash policy. 6. Conclusion We estimate the determinants of REIT cash policies because these firms hold such little cash relative to non-REITs. While the mandatory dividend requirement seems to be the culprit at first glance, a deeper understanding of REIT expenditures shows that the dividend requirement does not eliminate the possibility of accumulating capital internally. This ability to generate substantial cash flow annually may explain the paucity of REIT cash holdings. In fact, we document that the dividend restriction does not eliminate the ability of REITs to accumulate cash: REITs have the capacity to retain cash but choose not to. It appears that REITs avoid accumulating cash to increase transparency and reduce the cost of capital. Furthermore, our results show that dividend policy is only one piece of the puzzle explaining REIT cash policy as the strength and predictability of REIT cash flows coupled with access to short-term and long-term debt and the use of internal advisors clearly contribute to the incentives for REIT managers to avoid accumulating large sums of cash. Our main findings are robust to model specification and will be useful to future academic research, as the models developed are the first to investigate the factors influencing the determinants of REIT liquidity. Overall, the agency conflict caused by holding too much cash is effectively controlled within the REIT industry. A question that should be resolved by future research is the market value of REIT cash holdings. Given the relative transparency of REITs, it seems that their cash holdings would be highly valued. REIT Cash Holdings 22 References Ambrose, B. and P. Linneman, 2001, REIT Organizational Structure and Operating Characteristics, Journal of Real Estate Research 21, 141-162. Almeida, H., M. Campello, and M. Weisbach, 2004, The Cash Flow Sensitivity of Cash, Journal of Finance 59, 1777-1804. Bradley, M., D. Capozza, and P. Seguin, 1998, Dividend Policy and Cash-Flow Uncertainty, Real Estate Economics 26, 556-580. Campbell, R., E. DeVos, and A. Spieler, 2006, Investment, Liquidity, and Private Debt: The Case of REIT Credit Facilities, Working Paper. Capozza, D. and P. Seguin, 2000, Debt, Agency and Management Contracts in REITs: The External Advisor Puzzle, Journal of Real Estate Finance and Economics 20, 91-116. Damodaran, A., 2005, The Value of Cash and Cross-Holdings, Working Paper. Fazzari, S., G. Hubbard, and B. Petersen, 1988, Financing Constraints and Corporate Investment, Brookings Paper on Economic Activity 19, 141-195. Hausman, J., 1978, Specification Tests in Econometrics, Econometrica 46, 1251-1257. Jensen, M., 1986, Agency Costs of Free Cash Flow, Corporate Finance and Takeovers, American Economic Review 76, 323-329. Kim, C., D. Mauer, and A. Sherman, 1998, The Determinants of Corporate Liquidity: Theory and Evidence, Journal of Financial Quantitative Analysis 33, 305-334. Myers, S., Majluf, N., 1984, Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have, Journal of Financial Economics 13, 187-221. Newey, W. and K. West, 1987, A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Matrix, Econometrica, 55, 703-708. Opler, T., L. Pinkowitz, R.Stulz, R.Williamson, 1999, The Determinants and Implications of Corporate Cash Holdings, Journal of Financial Economics 52, 3-46. Ozkan, A. and N. Ozkan, 2004, Corporate Cash Holdings: An Empirical Investigation of UK Companies, Journal of Banking and Finance 28, 2103-2134. Sagalyn, L. 1996, Conflicts of Interest in the Structure of REITs, Real Estate Finance 9, 34-51. Wang, K., J. Erickson, and G. Gau, 1993, Dividend Policies and Dividend Announcement Effects for Real Estate Investment Trusts, Journal of the American Real Estate and Urban Economics Association 21, 185-201. White, H., 1980, A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity, Econometrica 48, 817-838. REIT Cash Holdings 23 Yan, X., 2006, The Determinants and Implications of Mutual Fund Cash Holdings: Theory and Evidence, Financial Management 35, 67-91. REIT Cash Holdings 24 Endnotes 1 It should be noted that Opler, Pinkowitz, Stulz, and Williamson (1999) study the ratio of cash and equivalents to net assets, total assets minus cash and equivalents. A REIT must meet three other provisions, in addition to the dividend payout restriction, to maintain their taxexempt status. First, five-or-fewer shareholders may not hold more than 50 percent of the REIT’s stock. Second, at least 75 percent of the total assets of the REIT must consist of real estate, mortgages, cash, or federal government securities, and a minimum of 75 percent of the REIT’s gross annual income must be derived from the ownership of real estate properties. Last, REITs must derive their income from passive sources such as rents and mortgage interest and not from the short-term trading or sale of property assets. 3 2 Prior to 2001 the dividend restriction equaled 95 percent of net income, and our calculation of excess dividends reflects this. Summary statistics in terms of dollars are available upon request. 4 5 We use this methodology only for the calculation of excess cash flow and excess dividends. For instance, the true value of net income available to shareholders is used to calculate return on assets. 6 Cash and equivalents is taken from SNL keyfield 3727. Total assets is SNL keyfield 220. Funds from operations comes from SNL keyfield 6116. Dividends paid is taken from SNL keyfield 14126. Market value of equity is SNL keyfield 6111, while total debt is from SNL keyfield 845. Revenues is taken from SNL keyfield 823. Annual asset growth rate is from SNL keyfield 295. Net income available to common shareholders is SNL keyfield 4433. Advisor type comes from SNL keyfield 63, while property focus is from SNL keyfield 6270. The variable Multi-family is.used as the reference category for dummy variable comparisions. The mean value of funds from operations in excess of the mandatory dividend payment is $55.7 million. 7 8 REIT Cash Holdings 25 Table 1. Descriptive statistics Variables CASH EXDIV EXFFO Q REVENUE GROWTH LEVERAGE ROA LOC N 625 625 625 625 625 625 625 625 573 Mean 2.19 0.91 2.70 1.16 313.90 16.97 52.99 2.64 41.34 Minimum 0.00 −18.51 −48.10 0.31 0.52 −83.16 0.00 −20.70 0.00 Median 0.81 0.906 2.97 1.10 148.66 6.68 53.25 2.77 43.96 Maximum 41.22 12.56 11.05 2.97 4,023.19 520.56 100.18 38.92 100.00 Table 1 shows the sample characteristics of the 625 firm-years for 129 unique publicly traded equity REITs over the period 1998 to 2005. CASH, EXDIV, EXFFO, GROWTH, ROA, and LEVERAGE are reported in percentage form. CASH is the ratio of cash and equivalents to total assets. EXDIV is the ratio of excess dividends, calculated as dividends paid minus the mandatory dividend payment, divided by total assets. EXFFO is funds from operations, defined as net income excluding gains or losses from the sale of properties, plus depreciation and after adjustments for unconsolidated partnerships and joint ventures, minus the mandatory dividend payment scaled by total assets. Q is the ratio of the sum of market capitalization, total debt, and preferred equity to total assets. REVENUE is fiscal year revenues in 2005 dollars and is in millions. GROWTH is annual asset growth rate. LEVERAGE is the ratio of total debt to total assets. ROA is net income available to common shareholders divided by total assets. LOC is the amount drawn from lines of credit divided by total lines of credit available. REIT Cash Holdings 26 Table 2. Time and Property Focus Distribution of Sample of REITs Panel A: Time Distribution of Sample Year 1998 1999 2000 2001 2002 2003 2004 2005 Total Number of Observations 6 90 92 89 86 84 90 88 625 REIT-Years (129 Unique REITs) Mean Cash to Total Assets 1.71 1.66 1.18 2.48 1.68 2.67 3.01 2.70 Panel B: Property Focus Distribution of Sample Property Focus Hotel Industrial Multi-Family Office Other Retail Storage Total Number of Observations 50 41 110 85 150 164 25 625 REIT Years (129 Unique REITs) Mean Cash to Total Assets 3.05 1.58 1.01 1.44 2.88 2.52 2.81 Panel C: Advisement Type Distribution of Sample Advisement Type Self-Advised Externally-Advised Total Number of Observations 572 53 625 REIT Years (129 Unique REITs) Mean Cash to Total Assets 1.93 4.90 Panels A, B, and C of Table 2 present the distribution of the sample across time, property focus, and advisement type, respectively, for the sample of REITs used for the excess dividends analysis. The sample consists of 625 REIT year observations for 129 unique publicly traded equity REITs over the 1998 to 2005 period. REITs are categorized on the basis of property focus (taken from SNL), and seven categories are used: Hotel, Industrial, Multi-Family, Office, Other (Diversified, Health Care, Manufactured Homes, and Other), Retail (Retail, Regional Mall, Shopping Center), and Storage. Excess Dividends to Total Assets and Dividend Payout are reported in percent form. REIT Cash Holdings 27 Table 3. Pearson Correlation Coefficients Variables EXFFO EXDIV Q SIZE GROWTH LEVERAGE ROA ADVTYPE CASH −0.197 −0.213 0.081 −0.155 0.305 −0.092 -0.065 −0.196 0.475 0.003 0.206 −0.028 −0.139 −0.269 0.012 −0.029 0.088 0.011 0.018 −0.387 0.024 0.146 −0.032 0.111 0.405 −0.067 −0.151 −0.116 −0.026 0.070 −0.054 −0.058 0.042 −0.310 0.105 −0.128 EXFFO EXDIV Q SIZE GROWTH LEVERAGE ROA Table 3 shows the Pearson correlation coefficients for the sample of REITs used in the determinants of REIT cash holdings analysis. CASH is the ratio of cash and equivalents to total assets. EXDIV is the ratio of excess dividends, calculated as dividends paid minus the mandatory dividend payment, divided by total assets. EXFFO is funds from operations, defined as net income excluding gains or losses from the sale of properties, plus depreciation and after adjustments for unconsolidated partnerships and joint ventures, minus the mandatory dividend payment scaled by total assets. Q is the ratio of the sum of market capitalization, total debt, and preferred equity to total assets. SIZE is the natural logarithm of fiscal year revenues in 2005 dollars and in millions. GROWTH is annual asset growth rate. LEVERAGE is the ratio of total debt to total assets. ROA is net income available to common shareholders scaled by total assets. ADVTYPE is a dummy variable equaling one if the REIT is internallyadvised, otherwise zero. The significant correlations are marked in bold. REIT Cash Holdings 28 Table 4. Pooled OLS Results Independent Variables Intercept EXFFO EXDIV Q SIZE GROWTH LEVERAGE ROA RETAIL HOTEL OFFICE INDUSTRIAL STORAGE OTHER ADVTYPE LOC Adj. R-Square Observations Predicted Sign (+/ — ) (−) (−) (+) ( —) (+) (−) (+) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) ( —) ( —) 0.192 625 0.220 625 Model 1 0.060 −0.162** −0.449*** 0.021*** −0.003* 0.028** −0.039** −0.149** 0.007 0.014** −0.002 −0.006 0.004 0.009* Model 2 0.082** −0.157* −0.451*** 0.018** −0.003 0.030*** −0.035* −0.158** 0.005 0.010 −0.004 −0.005 0.006 0.005 −0.027** −0.030*** 0.201 573 Model 3 0.039 −0.066 −0.129 0.012 −0.002 0.026* −0.020 −0.102 0.005 0.011** 0.001 −0.002 0.009 0.011** Table 4 presents pooled OLS regressions predicting REIT cash holdings. The sample consists of 625 firm-years for 129 unique publicly traded equity REITs over the period 1998 to 2005. The dependent variable is CASH, defined as the ratio of cash and equivalents to total assets. EXFFO is the ratio of funds from operations, defined as net income excluding gains or losses from sales of property, plus depreciation and amortization, and after adjustments for unconsolidated partnerships and joint ventures, minus the mandatory dividend payment divided by total assets. EXDIV is calculated as common dividends paid minus the mandatory dividend payment, divided by total assets. Q is the ratio of the sum of market capitalization, total debt, and preferred equity to total assets. SIZE is the natural logarithm of revenues in 2005 dollars. GROWTH is the annual asset growth rate. LEVERAGE is the ratio of total debt to total assets. ROA is net income available to shareholder scaled by total assets. RETAIL, HOTEL, OFFICE, INDUSTRIAL, STORAGE, and OTHER represent the property focus of the REIT and are accounted for using dummy variables. Multi-family is the base case. Annual dummy variables are used to control for time effects. ADVTYPE is a dummy variable equaling 1 if internally advised, 0 otherwise. LOC is the amount drawn from lines of credit divided by total lines of credit available. The standard errors for the pooled OLS model are corrected for autocorrelation and heteroskedasticity using Newey-West (1987) consistent standard errors. ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent level, respectively. REIT Cash Holdings 29 Table 5. Fixed-Effects Results Independent Variables Intercept EXFFO EXDIV Q SIZE GROWTH LEVERAGE ROA RETAIL HOTEL OFFICE INDUSTRIAL STORAGE OTHER ADVTYPE LOC Adj. R-Square Observations Predicted Sign (+/ — ) (−) (−) (+) ( —) (+) (−) (+) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) ( —) ( —) 0.371 625 −0.016*** 0.325 573 −0.189* −0.095 −0.007 0.001 0.017 −0.135*** −0.013 −0.213*** 0.022 0.008 0.009 0.023* −0.086*** −0.069 Model 1 Model 2 Model 3 Table 5 presents fixed-effects regressions predicting REIT cash holdings. The sample consists of 625 firm-years for 129 unique publicly traded equity REITs over the period 1998 to 2005. The dependent variable is CASH, defined as the ratio of cash and equivalents to total assets. EXFFO is the ratio of funds from operations, defined as net income excluding gains or losses from sales of property, plus depreciation and amortization, and after adjustments for unconsolidated partnerships and joint ventures, minus the mandatory dividend payment divided by total assets. EXDIV is calculated as common dividends paid minus the mandatory dividend payment, divided by total assets. Q is the ratio of the sum of market capitalization, total debt, and preferred equity to total assets. SIZE is the natural logarithm of revenues in 2005 dollars. GROWTH is the annual asset growth rate. LEVERAGE is the ratio of total debt to total assets. ROA is net income available to shareholder scaled by total assets. RETAIL, HOTEL, OFFICE, INDUSTRIAL, STORAGE, and OTHER represent the property focus of the REIT and are accounted for using dummy variables. Multi-family is the base case. ADVTYPE is a dummy variable equaling 1 if internally advised, 0 otherwise. LOC is the amount drawn from lines of credit divided by total lines of credit available. Annual dummy variables are used to control for time effects. The t-statistics (not reported) are calculated using robust standard errors. ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent level, respectively. REIT Cash Holdings 30 Table 6. Random-Effects Results Independent Variables Intercept EXFFO EXDIV Q SIZE GROWTH LEVERAGE ROA RETAIL HOTEL OFFICE INDUSTRIAL STORAGE OTHER ADVTYPE LOC Adj. R-Square Observations Predicted Sign (+/ — ) (−) (−) (+) ( —) (+) (−) (+) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) ( —) ( —) 0.209 625 0.24 625 Model 1 0.082** −0.193*** −0.311*** 0.015* −0.003** 0.027*** −0.054*** −0.103* 0.006 0.015 −0.002 −0.006 0.004 0.008 Model 2 0.010*** 0.033 −0.183*** −0.121** −0.348*** −0.070 0.015** 0.0130* −0.003* −0.001 0.028*** 0.027*** −0.046*** −0.027** −0.119** −0.079 0.004 0.005 0.011 0.012 −0.004 0.002 −0.006 -0.002 0.005 0.009 0.005 0.010* −0.026*** −0.025*** 0.227 573 Model 3 Table 6 presents random-effects regressions predicting REIT cash holdings. The sample consists of 625 firm-years for 129 unique publicly traded equity REITs over the period 1998 to 2005. The dependent variable is CASH, defined as the ratio of cash and equivalents to total assets. EXFFO is the ratio of funds from operations, defined as net income excluding gains or losses from sales of property, plus depreciation and amortization, and after adjustments for unconsolidated partnerships and joint ventures, minus the mandatory dividend payment divided by total assets. EXDIV is calculated as common dividends paid minus the mandatory dividend payment, divided by total assets. Q is the ratio of the sum of market capitalization, total debt, and preferred equity to total assets. SIZE is the natural logarithm of revenues in 2005 dollars. GROWTH is the annual asset growth rate. LEVERAGE is the ratio of total debt to total assets. ROA is net income available to shareholder scaled by total assets. RETAIL, HOTEL, OFFICE, INDUSTRIAL, STORAGE, and OTHER represent the property focus of the REIT and are accounted for using dummy variables. Multi-family is the base case. ADVTYPE is a dummy variable equaling 1 if internally advised, 0 otherwise. LOC is the amount drawn from lines of credit divided by total lines of credit available. Annual dummy variables are used to control for time effects. ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent level, respectively. REIT Cash Holdings 31 Table 7. Cross-Sectional Results Independent Variables Intercept EXFFO EXDIV Q SIZE GROWTH LEVERAGE ROA RETAIL HOTEL OFFICE INDUSTRIAL STORAGE OTHER ADVTYPE LOC Adj. R-Square Observations Predicted Sign (+/ — ) (−) (−) (+) ( —) (+) (−) (+) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) (+/ — ) ( —) ( —) 0.592 129 0.64 129 Model 1 0.064 −0.036 −0.726*** 0.032*** −0.003 0.042*** −0.027 −0.259** 0.005 0.019*** 0.001 −0.008 0.016 0.014** Model 2 0.075* 0.049 −0.013 0.084 −0.751*** −0.401** 0.033*** 0.019** −0.003 −0.002 0.044*** 0.031* −0.021 −0.013 −0.281*** −0.168** 0.002 0.006 0.016** 0.015** −0.001 0.003 −0.008 −0.003 0.017 0.016 0.010* 0.012** −0.031*** −0.030** 0.348 121 Model 3 Table 7 presents cross-sectional regressions predicting REIT cash holdings. The initial sample consists of 625 firm-years for 129 unique publicly traded equity REITs over the period 1998 to 2005. The dependent variable is CASH, defined as the ratio of cash and equivalents to total assets. EXFFO is the ratio of funds from operations, defined as net income excluding gains or losses from sales of property, plus depreciation and amortization, and after adjustments for unconsolidated partnerships and joint ventures, minus the mandatory dividend payment divided by total assets. EXDIV is calculated as common dividends paid minus the mandatory dividend payment, divided by total assets. Q is the ratio of the sum of market capitalization, total debt, and preferred equity to total assets. SIZE is the natural logarithm of revenues in 2005 dollars. GROWTH is the annual asset growth rate. LEVERAGE is the ratio of total debt to total assets. ROA is net income available to shareholder scaled by total assets. RETAIL, HOTEL, OFFICE, INDUSTRIAL, STORAGE, and OTHER represent the property focus of the REIT and are accounted for using dummy variables. Multi-family is the base case. ADVTYPE is a dummy variable equaling 1 if internally advised, 0 otherwise. LOC is the amount drawn from lines of credit divided by total lines of credit available. Annual dummy variables are used to control for time effects. The cross-sectional model uses White’s (1980) heteroskedasticity consistent standard errors. ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent level, respectively.