Even the most dogmatic passive investor faces a series of active choices, from asset allocation to which precise index funds to hold.
One important active decision that attracts less attention, however, is rebalancing. A purely passive, hands-off investor will eventually see their initially diversified portfolio skew over time towards the winners.
Should investors accept that as the efficient operation of the market, or should they rebalance to the weightings of their desired asset allocation? And if they take the latter route, how often should they rebalance?
The results of a modelling experiment published at the end of May by Joost Driessen and Ivo Kuiper – of Tilburg University and Kempen Capital Management respectively – offer some insights into these questions.
For this research, Driessen and Kuiper imagined a portfolio consisting of a risk-free asset and two risk assets, namely an equity index and a government bond index, with an investment horizon of 20 years. They then modelled the performance of such a portfolio under a range of return and rebalancing conditions.
‘Our main finding is that, although continuous rebalancing is optimal, the loss of a suboptimal rebalancing strategy is limited to up to 30 basis points in terms of the initial wealth of the passive investor when market returns are unpredictable and transaction costs can be ignored,’ they reported.
Makes me say oh my....
So while frequent rebalancing – here defined as monthly – yielded the best performance, rebalancing less often did no great harm to the portfolio.
In this scenario, trading costs were excluded to focus solely on the effect of rebalancing and returns were presumed to be random, or independent and identically distributed in statistical terms, with the investor making no attempt to time or forecast the market.
To make their analysis more practical, Driessen and Kuiper also ran their numbers to include the impact of the trading expenses when rebalancing, presuming one-way trading costs of 0.1% for both the equity and government bond index.
Here, their conclusion was intuitive. ‘Our results suggest that reducing transaction costs by less frequent rebalancing clearly outweighs optimising the portfolio more frequently,’ they confirmed.
‘Our calculations show that most of the marginal gain is captured when limiting the rebalancing frequency to once every year when taking transaction costs into account.’
The trend is your friend
But what about a model in which returns are not random, but predictable to some degree?
This need not imply perfect hindsight, but rather that the investor believes in observable market trends such as the momentum factor or mean reversion. Driessen and Kuiper investigated portfolio performance in this paradigm too.
Using what is known as a vector autoregression model, they estimated that the cost of rebalancing every month compared with rebalancing once every year in these conditions was about 1% in terms of the investor’s starting wealth.
‘Hence by not rebalancing, the investor profits from the trending behaviour of market returns,’ Driessen and Kuiper summarised.
Turn down the frequency
Allowing for transaction costs again suggested that limiting rebalancing to once a year was an appropriate frequency.
In short, then, rebalancing a portfolio too often is detrimental to returns.
In a simple world where markets are genuinely random, the expense of trading erodes any benefits from monthly reweighting.
And in a more complex environment where returns adhere to some regular patterns, excessive rebalancing negates the momentum effect.
Passive investors, in other words, should remain passive for at least 364 days a year.