The majority of asset allocation models are rebalanced on the last day of the calendar month, but is this the best approach to maximize returns? Portfolio manager Aneet Chachra and David Elms, Head of Diversified Alternatives, tested the validity of this traditional market orthodoxy, with some surprising results.
Starting as kids, we impart calendar periods with significance as we mark birthdays, school years, holidays, etc. This continues as we grow up – more people go to the gym on the first day of a new month while fewer go on the last day. And to most of us, this December feels extra special – the simultaneous end of the month, quarter, year, and decade.
We often carry this human love of round calendar intervals into financial markets. The majority of asset allocation models were developed using data from the last day of the month, quarter, etc., and are rebalanced at the same periodicities. This default is easy to understand and maintain, but might not be the best approach to maximize returns.
For example, U.S. stocks show evidence of mean reversion intra-month. Good returns during the first part of the month are often followed by weaker returns over the second part of the month (and vice versa). This negative divergence appears to be largest when measured around the monthly option expiration date (third Friday of each month).
Since 2006, if the S&P 500 Index® has fallen between the start of the month and option expiry, its subsequent return from option expiry through the end of the month has been strongly positive, averaging +1.0% (median of +1.4%). Conversely, if the S&P 500 has risen through option expiry, it has subsequently returned -0.1% (median of 0.0%) over the remaining month.
This is a surprising result. It runs counter to financial orthodoxy that would expect subsequent period returns to be uncorrelated. We calculate that the probability of the rebound (intra-month gains occurring more often following drops) being random is below 5%. Exhibit 1 shows the relationship between pre- and post-option expiry returns for each month.
Exhibit 1: S&P 500 pre/post-option expiry returns
Source: Bloomberg, Janus Henderson, January 2006 to November 2019.
Note: Data from S&P 500 Index total return series, excludes transaction costs.
Past performance is not a guide to future performance.
Why are intra-month returns negatively correlated? There is no way to determine definitively, but two explanations seem reasonable.
The first is the effect of option hedging flows. For example, envision a scenario where investors hold a large number of S&P 500 put and/or call options purchased from dealers and the market is falling. In order to hedge their exposure, dealers need to sell the index, which in turn creates further selling pressure. These hedging flows are inverted in a rising market. However, on option expiry, the dealers’ immediate index exposure ends removing this directional pressure and the market tends to reverse. Historical data indicates that the S&P 500 is more likely to switch direction on option expiry, which supports this explanation.
The second is the effect of rebalancing flows. After stocks have fallen, most asset allocation models will rebalance by buying stocks at month-end to return to target weight (and vice versa). These flows run counter to the recent market trend. Historical data indicates that the S&P 500 is more likely to rise (fall) into month-end after a large price decrease (increase) in the index, which supports this explanation.
It’s likely that both effects play a role. Their combined impact creates mean reversion intra-month that smooths monthly returns. We can observe this in S&P 500 volatility. Since 2006, if we measured the S&P 500 only at month-end, the index has an annualized volatility of 14.2%. However, if we instead measured the S&P 500 on option expiry, the index shows a significantly higher annualized volatility of 16.4%.
Changing the measurement day away from the last day of the month increases calculated volatility. The practical consequence is on choosing when to rebalance a portfolio. The benefit of rebalancing increases when done at higher volatility intervals.
To test this, we created multiple 60/40 U.S. Stock/Bond models that are identical except for the day of the month they are rebalanced. Exhibit 2 shows annualized returns for the models rebalanced on the first three and last three business days of the month as well as on option expiry.
Exhibit 2: U.S. 60/40 model annualized returns using different monthly rebalancing days
Source: Bloomberg, Janus Henderson, January 2006 to November 2019
Note: Data from S&P 500 and Barclays Aggregate index total return series, excludes transaction costs. Past performance is not a guide to future performance
Although the differences are small, the conventional default of rebalancing on the last day of the month had the lowest return. There is no expected benefit from rebalancing at the same time as everyone else, while the crowding of flows may detract from returns.
The data suggests that picking any other day instead of the last day of the month is likely better. Alternatively, another option is to diversify portfolio “rebalance risk” by implementing partial rebalances throughout the month. Rebalancing on the calendar end date is easy and popular, but does not appear to be the best choice.