Public versus Private Market Arbitrage: International Evidence for Listed Property Companies

Abstract

This paper examines the performance of real estate firms that issue seasoned equity with the stated purpose of investing in private market assets. Prior literature documents that (i) firms, in general, underperform following a season equity offering and (ii) growth firms underperform value firms. We propose a stylized model where firms may arbitrage a public market premium relative to the private market by investing seasoned equity proceeds in the latter market. We hypothesize and test this “public versus private market arbitrage” hypothesis for an international sample of 531 listed property companies spanning 12 countries. Consistent with the predictions of our model, we find that growth firms, those with relatively higher public market values, outperform value firms only under the condition where the stated use of proceeds is for investment purposes as opposed to all other uses, i.e., not investment-related. Our empirical evidence is based on buy-and-hold abnormal returns, time-series portfolio regressions, and firm-level, cross-sectional analysis. Overall, our results are consistent with a value-added strategy of public versus private market arbitrage and highlight the key consideration in the related capital allocation decision.

Keywords:

• Public versus private market arbitrage
• net asset value
• value versus growth
• seasoned equity offerings

Acknowledgments

Our work has benefited from discussions with many of our colleagues. We specifically thank Peter Albrecht for valuable suggestions regarding our theoretical model. Moreover, we thank Joseph Ooi, Mariya Letdin, Gianluca Marcato, Calvin Schnure (NAREIT), participants at the University of Reading 2018 Research Seminar, and past ARES, AREUEA International and ERES conference participants. We are particularly indebted to Bill Hardin (Editor), and two anonymous referees for their detailed comments. All authors acknowledge the support provided by the IREBS Foundation. Downs gratefully acknowledges the support of The Kornblau Institute at Virginia Commonwealth University.

Notes

1 In general, the academic literature on “multiple arbitrage” is relatively sparse, although the term has been associated with a type of value-creation strategy for private equity firms. See, for example, the citation of a McKinsey Study by Matthews et al. (2009).

2 We gratefully acknowledge the suggestions of two anonymous referees for motivating this additional analysis and the third prong of our empirical approach. A cross-sectional analysis allows us to control for heterogeneity across countries that cannot be accounted for in our time-series, portfolio analysis. Furthermore, it is important to note that there are considerable institutional differences across international listed-property markets. For instance, and as stated by a referee, in Hong Kong, the real estate markets are dominated by large cap real estate operating firms (mostly developers), and the REIT market in Hong Kong is relatively small and fragmented; whereas in the US, the REOCs are small in size compared to REITs, and operate in technically very different environments across the world. Our firm-level analysis, which is described in the empirical strategy section, along with the corresponding results, account for these concerns.

3 In other recent studies, Feng and Wu (2021) find that firm-level local economic growth is positively related to the equity value of REITs that allocate more assets in high growth areas. Huerta-Sanchez, Ngo and Pyles (2020) find a net positive effect in REIT operating performance for asset acquisitions relative to equity acquisitions. These studies address REIT growth and complement our work on public versus private market arbitrage.

4 Riddiough and Wu (2009) investigate REIT investment from a different perspective. Although their study includes equity offerings as a control, the primary focus is on the interplay between investment and liquidity management.

5 Yavas and Yildirim (2011) find the correlation as well as the lead-lag relationship of returns in public versus private real estate markets is dynamic, and that it can change across real estate sectors and within individual firms. In the context of our analysis, Yavas and Yildirim’s results suggest that CEOs of listed property companies have only a limited window to benefit from potential arbitrage opportunities. We incorporate this implication in our empirical design.

6 This is not a strong assumption. Goodwin (2013) compares the SEO share price with the prior day closing price and finds an average discount of only 1.77%.

7 We thank an anonymous referee for drawing our attention to this potential selection issue.

8 Our choice of the 11 IFRS countries included in our sample follows Woltering et al. (2018), who require a minimum of five listed property companies at each point in time.

9 We, again, gratefully acknowledge two anonymous referees for suggesting the firm-level, cross-sectional analysis.

10 Due to the exclusion of overlapping SEOs, the number of SEOs used in this section is generally lower than shown in the descriptive statistics (Table 1) and also lower than in the subsequent “portfolio-approach” analysis, which includes overlapping SEOs. Removing overlapping SEOs also has the advantage of reducing the possibility that our results are driven by a few high-frequency SEO firms.

11 Gokkaya et al. (2013) provide evidence of lower direct issuance costs for REITs versus comparable industrial firms, which may explain the relatively better performance of SEOs for the listed property companies in our sample. Moreover, Devos et al. (2019) document that REIT analysts tend to be highly optimistic around IPOs, and not optimistic surrounding SEOs. The corporate finance literature, generally, associates capital issuance optimism with low future returns.

12 Following the suggestion of an anonymous referee, we also compute BHARs using an alternative performance benchmark, where the control group (i.e., the benchmark return) is limited to property firms that had no SEOs over the same periods. Again, growth firms with investment SEOs perform noticeably better than their value counterparts, although the difference (4.89%) is not statistically different from zero. The results are available from the authors upon request.

13 In unreported results we also examine the abnormal returns of investment SEOs for the 12 months preceding the SEO. We note a comparable underperformance in the 12 months preceding an investment SEO for both value stocks (-5.0%) and growth stocks (-6.0%). Although, most of the underperformance can be attributed to the period at least eight months prior to the SEO, which suggests that a relationship between the announcement effect and other events directly linked to the SEO event is less likely.

14 Our empirical approach using BHARs and calendar-time portfolio regressions precludes using firm-fixed effects.

15 We follow the recent SEO literature which reports mean BHARs and calendar-time regressions using equally-weighted portfolio returns. Autore et al. (2009), Cline et al. (2014), Chan and Walter (2014), Silva and Bilinski (2015), and Kim et al. (2018) are all examples that follow Loughran and Ritter (2000). The latter stresses that value-weighted portfolio returns can be driven by individual, large firms. This concern is particularly acute for our sample as it is limited by the number of firms in the property sector and the fact that firms in some countries are systematically larger than in other countries. In the next section, we specifically address the potential influence that country- and regional-specific differences may have on our results. Moreover, Fisher et al. (2012) contend the upward bias of equally-weighted return indices is less pronounced for REITs relative to non-REIT stocks and that the bias has diminished over time. Any remaining bias in our sample should not impact the conclusions of our analysis as both our portfolio return and benchmark return are equally-weighted.

16 The monthly SMB, HML and WML factors are obtained from Kenneth French’s website (http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/datalibrary.html). French’s data library provides regional factors for Asia Pacific ex Japan, Europe, Japan, and North America, so we use the regional factors for the respective countries.

17 Note that the number of observations for growth stocks with Investment-SEOs is slightly reduced as there are some periods where none of the firms satisfy both criteria. The relatively small number of observations for this particular and critical aspect of our analysis prohibits a further split of the sample (e.g., by countries or regions).