Do Boards Shape REIT Performance? Evidence from the South African REIT Sector

We examine whether board activity (B_ACTIV), board size (B_SIZE), board independence (BIND), and board tenure (BOARD_TEN) are associated with the performance of South African real estate investment trusts (REITs) over the period 2013 to 2025. The REIT framework provides a rigorous setting to evaluate corporate governance theory, as statutory distribution mandates constrain payout discretion and contracted-income business models limit managerial opportunism, suggesting that governance effects concentrate within specific performance channels. We estimate dynamic panel models using a two-step system GMM framework with collapsed instruments, year fixed effects, Windmeijer-corrected standard errors, and firm-level controls for firm size (SIZE), leverage (LEV), and asset growth (GROWTH) to address endogeneity, unobserved heterogeneity, and performance persistence. We evaluate robustness through an endogenous-regressor specification, a bootstrap bias-corrected LSDVC estimator, and outlier-adjusted estimations. The sample comprises 30 JSE-listed REITs. We evaluate performance across funds from operations per share (FFO_PS), dividend yield (DIV_YIELD), return on assets (ROA), return on equity (ROE), return on invested capital (ROIC), and earnings per share (EPS). Our findings reveal that B_SIZE exhibits a statistically significant negative association with accounting profitability, where each additional director corresponds to a 1.0 percentage point reduction in ROE and a 0.32 percentage point reduction in ROA. The ROE effect remains robust across every identification strategy, including specifications treating board composition as endogenous and estimations winsorizing the dependent variables. Because firm SIZE remains statistically insignificant while LEV and GROWTH display their expected theoretical signs, the B_SIZE effect is isolated from firm scale. BIND demonstrates a directionally positive but specification-sensitive association with returns and payouts, whereas BOARD_TEN shows no robust association with any performance metric, and B_ACTIV effects attenuate once endogeneity is addressed. Overall, governance effects concentrate in operating efficiency and payout measures while remaining absent from per-share metrics, reflecting the precise channels through which boards exercise authority. Our findings caution against board expansion in this sector, highlight board scale as a transparent governance screen for investors, and demonstrate that meeting frequency and tenure benchmarks offer no reliable performance signal.