Does timing based on simple moving averages (SMA) work for U.S. Real Estate Investment Trust (REIT) indexes? If so, which moving average is best? In his March 2012 paper entitled “The Market Timing Power of Moving Averages: Evidence from US REIT Indexes”, Paskalis Glabadanidis tests the effectiveness of SMAs for timing ten value-weighted and ten similar equal-weighted U.S. REIT indexes. A monthly close above (below) its SMA signals investment in the REIT index (cash, estimated as the 30-day U.S. Treasury bill yield) the next month. He focuses on a 24-month SMA, but includes robustness tests based on 6-month, 12-month, 36-month, 48-month and 60-month SMAs. He applies baseline one-way trading frictions of 0.5% for entering and exiting a REIT index. Using monthly value-weighted and equal-weighted levels of ten U.S. REIT indexes during 1980 through 2010 (31 years), he finds that:
- Over the entire sample period, a 24-month SMA timing rule dominates corresponding buy-and-hold returns across the 20 REIT indexes, with uniformly higher average net returns and lower return variabilities (and therefore higher net Sharpe ratios). Advantages of timing based on average annualized net returns range from 2.1% to 11.2%.
- Associated four-factor (market, size, book-to-market, momentum) annualized net alphas range from 6.1% to 17.4%. These alphas are:
- Generally robust in the first and second halves of the sample period.
- Mostly negative during bull markets.
- Generally robust to adjustment for investor sentiment and equity market liquidity, but not to adjustment for the bond default spread (difference between Moody’s BAA and AAA corporate bond yields).
- For one-way trading frictions of 0.5% over the entire sample period, SMA rule net alpha generally weakens as the measurement interval increases from six to 60 months.
- For a 24-month measurement interval over the entire sample period, breakeven one-way trading frictions relative to buy-and-hold range from 3.4% to 20.2% across the 20 REIT indexes. Breakeven trading frictions are also generally large across different SMA measurement intervals.
In summary, evidence indicates that simple moving average timing rules work well for REIT indexes based on monthly data, with a short measurement interval (six months) working better than longer ones.
Cautions regarding findings include:
- The analysis does not address the costs (trading frictions for acquiring and rebalancing individual REITs plus management fees) of implementing REIT indexes as tradable assets. Incorporating such costs would materially reduce reported returns.
- Sample and subsample sizes are not large in terms of number of independent SMA measurement intervals and trading signals.
- Trading frictions are not constant over the sample period (see “Trading Frictions Over the Long Run”), so the assumption of a constant 0.5% may mislead.
- Testing of SMAs with different measurement intervals on different REIT indexes introduces data snooping bias, such that the best (worst) combination likely overstates (understates) reasonable out-of-sample expectations.
