Late cycle investment: How much does the timing of a recession matter?
UK pension plans concerned about how to invest in a volatile, late cycle environment may want to consider these two practices.
Stay the course: Rebalancing policies are crucial, particularly at times of market stress
We’ve long argued for investors to have a robust rebalancing regime in place if they are to ensure that their portfolios remain consistent with their strategic risk preferences.
Portfolios that are not rebalanced can drift substantially, potentially leading to unexpected risk exposures and greater volatility. Rebalancing is first and foremost a risk management tool to prevent such portfolio drift from happening.
However, even investors who routinely rebalance their portfolios may be tempted to suspend their rebalancing policies at times of extreme market stress. The unprecedented falls in global equities in response to the Covid-19 outbreak are a case in point. The speed and ferocity of the market downturn, and the accompanying tightening of liquidity has put pressure on rebalancing strategies, especially for pension funds or other investors with negative cash flows. This may potentially lead investors to delay any scheduled portfolio rebalancing.
Many investors may also simply want to avoid suffering higher transaction costs at a time of elevated market volatility. Others may be reassessing their asset class return, risk and correlation assumptions given growing concerns over the extent of the long- term economic damage caused by the virus.
The case for rebalancing
To consider the case for rebalancing, we looked at rolling 12-month returns over the last 30 years for an investment strategy that has been fully rebalanced every month to an equally weighted US benchmark (33% equities, 33% credit and 33% Treasuries), and then compared the performance of this rebalanced portfolio to a strategy that started out at equal weights but has been allowed to drift over time (EXHIBIT 1).
We’ve used US data, which provides fuller data over this longer time period, and performance is net of trading costs. Our key concern is to consider a portfolio that combines asset classes with different risk and return characteristics, and we later stress test for higher costs and different base allocations.
EXHIBIT 1: ROLLING 12-MONTH EXCESS RETURN OF FULL REBALANCING STRATEGY VERSUS ASSET ALLOCATION DRIFT STRATEGY
The performance of the two strategies has clearly been similar for much of the time, but differences have been material at times of market volatility – particularly around the dotcom bubble (1999-2000) and the global financial crisis (2008-2009). The signs are that performance has diverged again in the Covid- 19 crisis.
The periods prior to a crisis are often characterised by strong risk asset returns, when returns from defensive assets are low and equity markets may still be in relatively bullish mode. In these periods, a rebalancing strategy will underperform a drift strategy, as it routinely sells out of risk assets for risk management purposes – this underperformance in itself can give rise to questions about the merits of a rebalancing strategy, as it creates an apparent “leakage” of value from the portfolio.
However, in a drift strategy, portfolios may suffer a rising overweight to risk assets. And in periods of market volatility, an overweight to risk assets increases exposure to falling markets. A rebalancing strategy will avoid such an overweight occurring, and will therefore at least initially outperform a strategy where an overweight has been allowed to develop. As markets continue to fall, however, the rebalancing strategy will begin to underperform as the weight to risk assets in the drift strategy falls below that of the rebalancing policy.
Although rebalancing at this stage of market turbulence can feel like catching a falling knife – with investors continuing to buy equities even as asset prices fall further – the rationale for continuing to rebalance is to ensure that the portfolio is at full weight in risk assets at the beginning of any rebound and is able to “buy on the dips” as markets recover. In our analysis, underperformance of the rebalanced portfolio in downturns was more than paid back when markets recovered.
Overall, we find that rebalancing adds value on average, after allowing for transaction costs, with the rebalanced portfolio outperforming the drift strategy by 0.23% per annum over the full period of the analysis, resulting in a terminal value that is 6.49 percentage points higher, and having achieved that outcome with lower volatility of 5.9% per annum compared to 6.2%.
Our conclusion therefore is that, unless investors have a crystal ball that allows them to call the top or bottom of the market, suspending a rebalancing policy in response to market volatility – or due to late cycle “portfolio leakage” – is likely to be value- destroying.
Assessing the impact of starting points
This analysis is, of course, very dependent on the starting point – for example, a starting point early in the cycle allows more time for an overweight to risk assets to build up in the drift strategy compared to the rebalancing strategy. As EXHIBIT 2 shows, starting our drift strategy at initial equal weights at the beginning of 1990 allowed the allocation to the equities to drift up to 47% by the end of September 2000—on the eve of the bursting of the dotcom bubble.
EXHIBIT 2: ASSET ALLOCATION DRIFT OVER TIME
Of course, it’s unlikely that many investors would allow their asset allocation to drift quite so far before implementing some degree of rebalancing, which would be expected to reduce exposure to riskier assets going into a period of volatility.
To examine this further, we looked at performance in distinct market phases, as represented by equity markets. For each bull market or recovery phase, we considered the impact of initial underweights of various sizes to risk assets. Similarly, we looked at the impact of initial overweights to risk assets at the beginning of a market downturn (EXHIBIT 3).
EXHIBIT 3: FULL REBALANCING STRATEGY VERSUS ASSET ALLOCATION DRIFT STRATEGY WITH DIFFERENT INITIAL ASSET WEIGHTS OVER DISTINCT MARKET PHASES
For each market phase, we show the start weights to equities and we assume that at the outset the remainder of the portfolio is split equally between credit and Treasuries. We then show the end weight assuming a drift strategy, as well as the performance compared to a strategy that is regularly rebalanced to equal weights. Three main observations can be made:
Overall, the analysis serves to demonstrate that in a drift strategy, even with ad hoc rebalancing from time to time, a portfolio can “accidently” fall out of line at points of market inflection, and this can drive material divergence in performance relative to strategy. Rebalancing is a key risk control mechanism to avoid such “accidents” occurring.
Assessing the impact of base allocations and trading costs
Our results so far have been calculated for a strategic allocation that is equally weighted to equities, investment grade credit and Treasuries. To find out how the strategic allocation influences the results, we looked at a more equity-oriented portfolio (50% equities, 25% credit, 25% Treasuries), a credit-oriented portfolio (25% equities, 50% credit, 25% Treasuries) and Treasury-oriented portfolio (25% equities, 25% credit, 50% Treasuries).
The same patterns of relative performance were observed across our range of base portfolio allocations, suggesting that the rationale for regular rebalancing applies no matter which strategic allocation is chosen (EXHIBIT 4).
EXHIBIT 4: ASSET ALLOCATIONS ARE EQUITY / IG CREDIT / TREASURIES
On trading costs, we’ve used US data to provide deeper long- term analysis and also to give better visibility on trading costs in both normal and stressed conditions. However, the US is the most efficient market in the world and trading costs in other markets can be materially higher. To test whether the cost of rebalancing in in more expensive markets would outweigh the benefits, we applied cost loads to our US-based trading cost estimates and compared the terminal values, as shown in EXHIBIT 5.
EXHIBIT 5: INCREASE IN TERMINAL VALUE COMPARE TO DRIFT STRATEGY
The benefits of rebalancing, measured as how much bigger the terminal asset value is compared to a drift strategy, reduce in a fairly linear way as we increase the cost load. However, even when trading costs are twice as high as our base trading cost estimates, the terminal value of the rebalanced portfolio had still increased by 5.80 percentage points more than the drift strategy over the full period of our analysis (compared with 6.49 percentage points in our base cost case).
To negate the benefits of rebalancing – such that there is no increase in terminal value compared to the drift strategy – our analysis suggests that trading costs would need to run at over 16 times our base estimates. The results therefore suggest that trading costs should be no barrier to regular rebalancing.
Assessing the impact of market liquidity
An oft-cited obstacle to rebalancing in periods of volatility — particularly by pension funds with negative cash flows – is the lower levels of market liquidity.
Our analysis assumes zero cashflows into or out of pension funds. Funds with positive cashflows should generally find it easier to rebalance even in volatile markets, as there is less pressure on them to sell overweight assets to top up underweight positions. However, a lack of liquidity can be challenging for funds with negative cashflows that need to sell assets in volatile markets. And if these funds are also using derivative instruments for hedging purposes, the associated collateral movements can add further strain.
We don’t see cashflow concerns as an argument for or against rebalancing per se. Instead, pension funds should ensure that they have a clear liquidity and collateral management policy in place to create operational resilience around servicing cash outgo and collateral management. Funds should also reduce reliance on the sales of assets as the primary means of servicing predictable and regular cashflows.
Another concern for pension funds is the difficulties they can have including illiquid assets in rebalancing policies. A pragmatic approach to this problem is to allow wider ranges of deviation, and where possible to use vehicles that can deliver income or occasional liquidity, which can then be reinvested in the illiquid portfolio or elsewhere as befits the rebalancing objective.
Given the liquidity challenges faced by some funds in volatile markets, we looked at the effectiveness of rebalancing only partially as a more pragmatic approach compared to the theoretical ideal of full monthly rebalancing. We find that partial rebalancing of portfolios still brings benefits by ensuring that allocations don’t drift too far out of line. EXHIBIT 6 shows how the rolling 12-month excess returns (relative to drift) from a portfolio that only rebalances “halfway” each month differs from the fully rebalanced portfolio. The periods of divergence in performance are highlighted.
EXHIBIT 6: ROLLING 12-MONTH EXCESS RETURN OF FULL REBALANCING STRATEGY AND HALFWAY REBALANCING STRATEGY VERSUS ASSET ALLOCATION DRIFT STRATEGY
We can see that the halfway rebalanced portfolio works pretty much just as well as the full rebalancing strategy almost all of the time. The reason is that the halfway rebalanced portfolio will eventually rebalance fully back to its base allocation – it will just take a period of a few months, rather than a single month, to do so. Accordingly, portfolios do not drift meaningfully away from the base strategy.
We also look at the standard deviation of returns for full rebalancing versus the returns for partial rebalancing (EXHIBIT 7). The larger the standard deviation, the less the partial rebalancing strategy is capturing the benefits of full rebalancing. This analysis suggests that most of the benefits of full rebalancing are still captured by a 50% (halfway) rebalancing, but the margin of benefit is smaller for lower levels of partial rebalancing. A partial rebalancing approach can also mitigate the “catching a falling knife” effect in sharp market downturns.
EXHIBIT 7: TRACKING ERROR VERSUS FULL REBALANCE
Tactical asset allocation works alongside rebalancing
Having a rebalancing policy does not, of course, preclude or negate active asset allocation in any way. In fact, a robust rebalancing policy helps to make active asset allocation decisions clearer and more accountable.
Rebalancing provides transparency, showing what action would have been taken in the absence of a tactical asset allocation process, and therefore allows the size of the active decision – and its outcome – to be measured.
For example, simultaneous decisions to fully rebalance an underweight position to an asset class in a portfolio (via an established rebalancing policy) and also to implement an active tactical underweight position in that same asset class may actually look very similar to a drift strategy: combined these two decisions offset each other and may result in no change to the asset allocation.
However, the clear advantage of having a rebalancing policy in place is that it makes it much easier for pension funds to account for and measure the effectiveness of active asset allocation decisions, compared to allowing allocations to drift. A rebalancing policy reinforces the base strategy as a risk anchor for all active asset allocation decisions.
The Covid-19 outbreak abruptly brought the long equity bull market to an end. The road to economic recovery appears to be gradual and further market volatility is likely. In this environment, it may be tempting for pension funds and other investors to suspend portfolio rebalancing policies, particularly for funds facing liquidity constraints.
Our analysis shows that pension funds should resist the temptation. In all the scenarios that we tested, across several market cycles, regular rebalancing of portfolios led to superior results compared to allowing portfolio allocations to drift in terms of risk management.
The key thing is to have a rebalancing policy and stick to it, even through periods of market turmoil. If liquidity proves an obstacle, even a partial rebalancing provides a strategic risk management discipline in line with long-term investment goals. The appropriate rebalancing policy will depend upon individual asset allocation decisions, portfolio implementation arrangements and cashflow circumstances. The message, however, is clear – rebalance as much as you can, and don’t abandon your policy.
Description of data used and core methodology
We used historical benchmark data from January 1990 to represent monthly total returns for equity, credit and bond assets. For equity the S&P 500 Dollar Index is used; for credit the Bloomberg Barclays US Aggregate Corporate Index; and for bonds the Bloomberg Barclays US Treasury Index. US market data are used to allow a longer history for the backtest analysis.
Portfolio rebalancing is assumed to occur at the end of the month for simplicity, where the degree of drift is calculated as the month-end actual asset allocation compared to the strategic asset allocation target. The relevant transaction costs incurred to achieve the policy rebalance level are then deducted to arrive at the end of month post-rebalance portfolio value. Note: in our transaction cost assumptions we differentiate between ordinary and stress market conditions to explicitly model the higher transaction cost associated with volatile market periods.
Transaction cost assumptions are derived by conducting historical bid-ask spread analysis on liquid tradable markets, such as the S&P 500 Futures and US Treasury Futures, combined with expert judgement overlay from the relevant investment and trading desks. We divided the backtesting period into months that experienced market stress and ordinary trading periods in order to estimate transaction costs under normal and stress market conditions. We define stress market conditions with reference to the prevailing level of the VIX index, where the long-run 75th percentile level is set as the threshold.
APPENDIX: VIX INDEX - MONTHLY
Opinions, estimates, forecasts, projections and statements of financial market trends are based on market conditions at the date of the publication, constitute our judgment and are subject to change without notice. There can be no guarantee they will be met.