To benchmark or not to benchmark?

November 9, 2020by Atanuu Agarrwal0

Recently, there have been some speculative media reports that AMFI (Association of Mutual Funds in India) has written to SEBI (Securities and Exchange Board of India) to launch a new index or cap single stock exposure in an index. AMFI has denied these reports.

In any case, many fund managers have blamed their recent under-performance against the benchmark on the 10% single stock exposure cap mandated on mutual funds, while there is no such restriction on indexes. RIL (Reliance Industries Limited) is the stock causing this chagrin. RIL’s recent rally has resulted in it now constituting ~15% of the Nifty 50 index. Fund managers complain they were shackled by the artificial 10% cap and were unable to participate in its rally fully. They allege that this constraint is the main driver behind their under-performance vis-à-vis the Nifty 50 benchmark.

Smells like sour grapes

If you scratch beneath the surface you will see that most funds’ RIL holding is not even close to the 10% cap (see image below). The truth is that the ‘blame it on the umpire’ tactic doesn’t work here – the batsman just didn’t read the match situation and picked the wrong shot.

Overall, this whole episode points to a larger issue with benchmarks and its effect on fund managers.

RIL holding of mutual fund managers

A kludge in a very complex system

George Latham and Mike Clark explain the utility and possible of origin of benchmarks well here:

“ A starting point is to try to understand why benchmarks exist. What should their purpose be? It is widely agreed there exists an asymmetry of information between fund managers (as agents) and asset owners (as principals). The principals are often pension fund trustees on whom rests the responsibility of fiduciary duty, which is designed to ensure that decisions are made in the best interests of the trustees’ beneficiaries.   Over a long period of time, those interests have become progressively less well aligned with those of the fund managers that they employ.

So how are trustees supposed to hold fund managers to account, to make sure that they can demonstrate to their beneficiaries that they have done the right thing and discharged their fiduciary duty? Financial markets are complex systems. Trustees have limited resources at their disposal, certainly fewer resources than the fund managers they are supposed to be holding accountable.

A benchmark makes this uncomfortable responsibility more manageable. Give the fund manager a benchmark and tell them to beat it. Then ask them to regularly report on how they are doing and the answer is simple and straight forward. Ahead of the benchmark? “Well-done, we look forward to our next meeting.” A bit behind? “We need to keep an eye on you…” Behind again? “You’re fired!!”

Goodhart’s law explains well the knock-on effects of this phenomenon at a system level.

Goodhart’s law

In a paper published in 1997, Anthropologist Marilyn Strathern generalized Goodhart’s law beyond statistics, and made it more accessible to all of us through this phrase:

“When a measure becomes a target, it ceases to be a good measure.”

In simple terms, it implies that once any metric becomes a lead indicator for a system (like real GDP growth rate is for countries), the players in that system will start to game it.

To make this contextual, it is now very common for fund managers to start with a benchmark and try to deviate from it only in ways that will be palatable to investors. Many Indian fund managers would have invariably been asked at some point in 2020 – “Why is RIL not in your portfolio?” or “You exited your RIL position?! Explain yourself!”. It doesn’t matter that there are several stocks that have returned way more than RIL.

Benchmark ≠ Market

It is common to refer to major indices (Sensex, Nifty 50, Nifty 100 and Nifty 500) as the “market”. In the last couple of years, a lot of you would have look at your portfolios despairingly as they under-performed against this “market”.

This is because these benchmarks can easily deviate from the reality of the general market – in the last 3 years, the Nifty 100 (weighted by market cap; a common benchmark) has gone UP by 11.8% but the Nifty 100 EW (equal weighted) has gone DOWN by 8.8%! The reason for this differential is that the Indian market is very top heavy – just the top 30 stocks (1.25% of the total number of listed companies on the NSE) by market capitalization contribute 49% of the entire market!

A handful of stocks drove the out-performance of the Nifty 100 vs. Nifty 100 EW. Chances are that if you didn’t have these 5-6 stocks in your portfolio, it would have under-performed the “benchmark”. Hence, creating the toxic incentive structure for fund managers to not deviate too much from these benchmarks.


Risk is now seen as deviation from benchmark, rather than absolute risk. This approach fails to account for risk in the benchmark itself. Benchmarking short term returns for active managers against indexes disincentivizes them to deviate too much from the index in terms of their portfolio mix. This leads to lower incentives for research and analysis, which is a disservice to the investor.

Thankfully, machines don’t have to be governed by these human constraints. For instance, Upside AI’s algorithms don’t use indexes as inputs; they focus on maximizing absolute returns. We believe that this is an important consideration to deliver sustained alpha in the long-term over those same indexes.

Atanuu Agarrwal

Leave a Reply

Your email address will not be published. Required fields are marked *