The Cost of Financial Flexibility: Information Opacity, Agency Conflicts and REIT at-the-Market (ATM) Equity Offerings

In contrast to traditionally underwritten security offerings, where a predetermined number of shares are sold at a fixed price at a single point in time, the establishment of an ATM program grants managers the option to sporadically issue equity over an extended period at prevailing market prices. See Hartzell et al. (2019) for further details on the ATM issuance process.

It is important to note that prior research has focused on the relatively low direct costs associated with ATM share issuance as a benefit of this form of financial flexibility. For example, Hartzell et al. (2019) and Howton et al. (2018) document broker fees of approximately 2%, on average, of gross proceeds for ATM issuances.

While we recognize taxable income and cash flow are not perfectly equivalent metrics, and acknowledge many REITs routinely pay and sustain dividend payout ratios well in excess of 100% of reported net income, these mandates nevertheless provide some constraints on the ability of firms within this industry to retain cash flows.

An extensive prior literature discusses the use of cash holdings (Denis and Sibilkov 2010; Almeida et al. 2004; Dittmar and Marhart-Smith 2007; Kalcheva and Lins 2007; and Riddick and Whited 2009), low leverage policies (Marchica and Mura 2010; Ang and Smedema 2011), access to lines of credit (Ooi et al. 2012; Lins et al. 2010; and Campello et al. 2010; Sufi 2009; Hardin and Hill 2011; Harrison, Luchtenberg and Seiler, 2011), the use of secured debt financing (Stulz and Johnson 1985), and the covenant structure of unsecured debt (Riddiough and Steiner 2020) as various mechanisms for firms to achieve financial flexibility.

See Hartzell et al. (2019) for a more detailed description of the process used to identify ATM firms. Our sample period begins in 2006 as this is the first year following significant regulatory reform on eligibility requirements for ATM program access.

As a point of comparison, Lee and Masulis (2009) find the direct underwriting costs of SEOs for industrial firms average slightly more than 5%, while Gokkaya et al. (2013) document direct costs of 4.64%, on average, amongst REITs.

The use of book values in the residual income approach may understate the equity value of real estate for property portfolios that experience a significant amount of price appreciation over time. Conversely, the use of book values would overstate the equity value of real estate for property portfolios that experience significant capital depreciation over time. Thus, as real estate values were generally increasing over our sample observation interval, we recognize and acknowledge the residual income approach may, by construction, marginally understate the true cost of capital in our sample.

Lee et al. (1999) conclude using a longer period for an estimation window does not significantly affect estimates. We note that the requirement of three years of FFO data limits the window over which we can perform our analysis.

Book value of equity is forecasted using the clean surplus relation (B_t = B_t-1 + NI_t –DIV_t), in which dividends are assumed to be 85% of FFO. While REITs are required to pay out at least 90% of taxable income as dividends, REITs have paid approximately 71% of FFO, on average, over the past 20 years according to NAREIT. Thus, our 85% assumption is a conservative estimate of dividend payout.

It is important to note that neither of these two approaches rely on book value estimates of the real estate property portfolio as was the case with measures derived from the residual income approach. Thus, any concerns with REIT book values systematically understating true asset values due to the lack of mark-to-market accounting should be mitigated for these two alternative metrics. Reassuringly, in the empirical results which follow, all five cost of capital metrics employed provide qualitatively similarly findings and implications.

Botosan and Plumlee (2005) utilize long term earnings forecasts (e.g., Year 5 and Year 4) in place of short-run earnings forecasts to mitigate this issue. However, the low frequency of FFO forecasts available at these horizons in the IBES database precludes the use of this modified specification in our analysis.

Principal component analysis has been frequently employed in the prior literature to reduce the idiosyncratic measurement error of individual implied cost of capital measures (e.g., Hail and Leuz 2009; Chen et al. 2011).

To ease economic interpretations across cost of capital measures, we constrain the distribution of our principal component measure to have the average mean and standard deviation of our other four proxies. It is important to note that relaxing this constraint does not impact the statistical significance of our findings.

This mitigates concern that our main results may be driven by a fundamental change in the level of the cost of capital in the post-crisis period. More formally, we ensure our main results are robust to the exclusion of the crisis-period in a subsequent robustness check.

An et al. (2011) utilize a similar framework to examine the relation between REIT transparency and firm growth. However, they utilize (1-R²) to measure firm transparency rather than opacity. Consistent with An et al. (2011), we also construct this variable using lower frequency data (e.g., weekly data) to reduce any noise inherent in its construction and obtain robust results upon its inclusion in each of our empirical specifications.

As in Riddiough and Wu (2009), we eliminate observations that do not fit within the following bounds: 0 ≤ CASH ≤ 0.5.

Similarly, as firms mature, the market may gain valuable insight into the stability and consistency of a firm’s continuing operations. Such information may well reduce investor uncertainty, thereby further lowering required returns.

For further insight into the importance of staggered boards on firm level operations and outcomes, see Jiraporn and Liu (2008), Jiraporn and Chintrakarn (2009), and Jiraporn et al. (2012), among many others.

Cashman et al. (2016) identify clawback provisions as a value-relevant governance mechanism for equity REITs.

Institutional holdings may also enhance price discovery through increasing the available supply of shares to short, thereby facilitating the rapid incorporation of negative information into security prices.

To mitigate concern that our main ATM finding reflects a mechanical relation with our cost of capital measures that is driven by the negative stock price reaction which often accompanies equity share issuance (e.g., Jegadeesh 2000), we include quarterly lagged appreciation returns (APPRET) as an additional explanatory variable in each of our specifications and confirm our results are robust to controlling for the influence of past stock price changes. It is also important to note that Hartzell et al. (2019) find significantly less negative price reactions around the announcement of an ATM relative to a comparable SEO.

Hartzell et al. (2019) provide evidence that firm characteristics associated with a firm’s level of financial constraint, potential growth opportunities, and external monitoring impact the likelihood of establishing an ATM program. Since managerial decisions to establish an ATM program are likely to be non-random, we follow this literature and utilize a Heckman (1979) two-stage estimation to control for sample selection effects in a subsequent robustness check. We utilize average underwriter ranking scores for past seasoned equity issuances (UWRANK) to meet the exclusion requirement of the first stage regression since prior certification may influence the firm’s choice of whether to establish an ATM program. The results from the second stage regression are reported in Table IA.A1 of the Internet Appendix. The insignificant inverse mills lambda confirms that sample selection is not of prime importance in our examination of the cost of capital.

We obtain qualitatively similar results when using our other four proxies for the cost of capital, though statistical significance is reduced due to the smaller sample sizes.

It is important to note that the difference in the cost of capital prior to the establishment of an ATM program for firms that close their programs versus those that announce follow-on programs is not statistically different from zero in our univariate analysis.

Though unreported, we also find preliminary evidence that the cost of capital effect is mitigated for firms with high ATM program completion rates.

For brevity, we only report coefficient estimates for our main variables of interest in Table 5. We report coefficient estimates for each of our control variables in Table IA.A2 of the Internet Appendix.

In their comparison of ex-ante firm characteristics for ATM and SEO users outside of the REIT industry, Billett et al. (2019) identify ATM firms as being smaller, less levered, and more opaque. These differences highlight the importance of an additional line of research that examines the implications of these financing mechanisms specifically within the REIT market.

Consistent with the previous literature, we exclude t = 0 from these regressions since it is difficult to assess which quarter, before or after the event announcement, it should belong to. Examples of similar methodological approaches include Gertner et al. (2002), Brisker et al. (2013), and Howton et al. (2018), amongst others.

To mitigate concern that our cross-sectional findings are driven by our choice of OPACITY as a measure of the firm’s information environment, we construct an additional set of robustness checks utilizing two alternate measures of firm transparency: (1) the number of analysts following the firm, and (2) an indicator variable based on whether the firm is followed by Green Street Advisors, a prominent buy-side research company focused specifically on analyzing REITs. Since these information measures are increasing in the level of firm transparency (and therefore negatively correlated with our main OPACITY measure), an increase in firm transparency should help mitigate the adverse impact of ATM program access on the firm as increased transparency reduces the uncertainty associated with managerial decisions. Our results from these additional robustness checks are consistent with those using OPACITY as our measure of the firm’s information environment. A more in-depth description of these tests and associated results are reported in the Internet Appendix (Section IA.B1 and Table IA.A3, respectively).

Audit fee data were obtained from the AuditAnalytics database.

For brevity, we only report coefficient estimates for our main variables of interest in Table 9. We report coefficient estimates for each of our control variables in Table IA.A4 of the Internet Appendix.

Though unreported, multivariate tests that include the full set of firm characteristics, property type and time (quarter) fixed effects confirm the results of each of our univariate tests.

We obtain book value of equity data from S&P Global Market Intelligence.

Howton et al. (2018), amongst others, utilize a similar strategy to identify a REIT’s relative level of financial constraint.