Co-Published

Janus suggests approach to limit tracking error without constraining excess returns

In a co-published feature, Janus Capital Group addresses the challenge of maintaining excess return potential while having reasonable expectations of modest tracking error.

A moderate tracking error is an important criterion for some pension fund sponsors, but how one constrains that tracking error is crucial.

It’s far more effective for fundamental active managers to control tracking error as the outcome of an investment process rather than as an input into the process.

Janus Capital Group suggests a way to construct a portfolio that allows fundamental, active managers to take moderate risk and focus on their stock-picking skills.

A solution
Managers who are indifferent to the benchmark and tracking error can potentially provide strong risk-adjusted returns. The challenge is to maintain excess return* potential while having reasonable expectations of modest tracking error.

Janus’s analysis showed that managers can improve the excess return potential of a portfolio by increasing the active portion of the portfolio, but this typically leads to a greater tracking error – an uncomfortable outcome for fund sponsors seeking the middle ground. One way to solve this dilemma is to build the portfolio in such a way that there is a low correlation of excess returns between each active sleeve and the benchmark portfolio.

One proven way to accomplish this is to divide the universe by sectors. Janus has found that many managers picking a few stocks in segregated universes may provide higher returns with less tracking error than a few managers picking many stocks from anywhere.

The most consistent sources of tracking error are sector/industry exposures, beta and, to a lesser extent, country weights relative to the index. Other sources are the size of the active portfolio and the correlations of excess returns among the underlying active exposures.

Maintaining country/sector neutrality relative to the benchmark and a beta of 1 will reduce two primary sources of tracking error. Both can be fairly easily addressed by a manager’s portfolio design. The more subtle step – yet one quite important to maximise risk-adjusted returns – is increasing the weight of the active portfolio, and decreasing the correlation of expected returns.

Janus believes an effective solution is a team-led approach. To ensure low correlations of excess returns, the portfolio would be divided among dedicated analyst teams, each constrained to one sector.

Each team or manager, however, should be unconstrained by the index holdings within that sector and by tracking error to the sector index. They should remain fully invested, with each segment reflecting the top ideas of the experts within an investment firm.

While stock-specific risk might be predominant, the total diversified portfolio should produce higher expected excess returns, a better risk-reward trade-off and a similar tracking error to a portfolio with explicit (and inefficient) constraints on tracking error. Tracking error is constrained, not manager skill.

Janus offers a representative team-led research portfolio as an example. The typical research portfolio has about 100 holdings, with each position representing an active stock-pick. The goal is to maximise the skill function by taking the top ideas of each analyst.

Each sleeve of the portfolio is sector-neutral relative to its benchmark. The sector teams do not consider index weights when building their sleeves, or sector portfolios. Their average sleeve has about 15 holdings, and the largest positions can at times exceed 15% of the sector portfolio.

When the sleeves are integrated into the portfolio, the expected excess returns generated by the individual teams are not diminished and the result is sector neutral. Further, as a result of the segregated universes, the correlation of excess returns of each sector portfolio to one another is low.

Exhibits 2 and 3 show the correlation matrix of returns and excess returns, respectively, of each sector portfolio. Because a correlation analysis of overall returns does not strip out the systemic risk of the market, looking at the correlations of excess returns may be more powerful.

Here the systemic risk is filtered and one looks only at the active portfolio, where the non-index investments exist. One sees the expected and desired low correlations of excess returns that can lead to a lower tracking error for the overall portfolio. In most cases, these correlations are close to zero.

In this structure there are no constraints on tracking error for each sector portfolio, and each sector portfolio’s tracking error is relatively high compared with its sector benchmark. However, as the sleeves are combined into the portfolio, the overall tracking error falls because of the low correlation of excess returns.

A simulation of several concentrated portfolios with randomly generated returns confirms this approach. The results, summarised in exhibit 5, show that over multiple time periods, tracking error as these portfolios are combined is generally lower than those for its component portfolios.

Conclusion
For pension fund sponsors seeking moderate tracking error, Janus has tried to show that an approach with specialised sector teams independently managing portions of a portfolio is an optimal solution. Its model demonstrates how a portfolio that maximises stock-picking skill while constraining tracking error may provide an effective solution for pension fund sponsors seeking middle ground.

* Excess returns: the difference between performance of a fund or investment strategy and its benchmark. A positive figure does not necessarily mean the fund or investment strategy has positive performance during the period.  

Making the Most of the Middle Ground (Updated white paper February 2010) is a collaborative paper produced by Janus Capital Group’s investment team.

¬ Haymarket Media Limited. All rights reserved.


Quick Poll

How will you alter your portfolio after the US election result?





« View results

Previous polls ››

Underweight US
  19%
 
Overweight US
  34%
 
Underweight EMs
  6%
 
Raise cash levels
  27%
 
No change
  13%
TOTAL VOTES: 62
Comment on this

« Back

Previous polls ››

November 2016 Magazine
AsianInvestor Magazine