Time series momentum in real estate investment trusts

Abstract

This thesis explores the existence of the time series momentum phenomenon within individual real estate investment trusts (REITs). By analyzing data from 1991 to 2023, I discover a significant time series momentum effect within U.S. REITs with a value-weighted dollar-neutral strategy yielding an average annualized return of 9.04%. Interestingly, these returns cannot be explained by the more broadly studied cross-sectional momentum effect, yet cross-sectional momentum can be explained by time series momentum. Combining these two forms of momentum into a dual momentum strategy demonstrated greater effectiveness than either strategy alone, achieving an average annualized return of 17.88%.

Despite these strong returns, a subperiod analysis reveals that the profitability of time series momentum strategies is primarily concentrated in the first half of the sample period with none of the studied strategies generating statistically significant returns post-2008. A key flaw with these strategies is their propensity to suffer significant drawdowns during overall market recoveries which may have contributed to the weaker returns observed in the post-2008 period.

In response, I propose an adaptive momentum strategy that adjusts positions based on market conditions. The most effective iteration of this strategy produced a highly impressive average annualized return of 25.05% during the sample period. The strategy was also able to generate statistically significant returns during the post-2008 period.

Overall, these findings advance the understanding of momentum in the REIT market and also provide practical insights for enhancing performance through adaptive strategies.