active passive and smart beta part 2 active vs passive investing
This is Part 2 of the 5-part series on Active, Passive and Smart Beta strategies.
In Part 1, we did an introduction and saw how they fit into the overall investment strategy landscape. We also compared them along 3 dimensions – Discretion, Cost and Capacity.
In this blog, we will focus on Active and Passive style of management and do a detailed comparison. More specifically, we will try to answer the question – Should you invest in an actively-managed Mutual Fund or passive index funds? Or is there a middle ground?
We introduced the Capital Asset Pricing Model (CAPM) in Part 1. Since we will be using CAPM extensively in this analysis, let’s do a quick refresher.
CAPM states that expected excess return (excess of risk-free rate) on any security is proportional to excess return on the market (the equilibrium portfolio) and that proportion is the security’s exposure to the market and is called beta.
ER - Rf = b x (ERm - Rf)
Very simply put, CAPM states that exposure to market is your one and only source of rewarded return – market is the ONLY risk factor that is rewarded.
Therefore, as per CAPM, expected excess return (over and above beta-adjusted return) should be zero and hence active management cannot add any value.
However, the entire goal of active management is to deliver excess return over beta-adjusted return i.e., the “alpha”. By introducing “alpha” fits into the above equation we get:
ER - Rf = a + b x (ERm - Rf)
The role of active management is, in essence, to deliver positive alpha. Alpha is often misunderstood as excess returns over the benchmark and this is what is also used by the Mutual Fund industry – outperformance over the benchmark.
In reality, alpha is the excess return over benchmark, adjusted for beta. A large-cap manager outperforming Nifty 50 index with a beta of 1.2 may not actually be delivering alpha.
In this analysis, we will use the correct definition of alpha to analyse the performance of actively managed Mutual Funds.
Note 1: CAPM model is not perfect and has come for a lot of criticism over the years. At the same time, it still remains the cornerstone of quantitative finance and is widely used. And when used solely for the purpose of analysing actively managed funds, it’s a simple and brilliant model.
Note 2: Our analysis of Mutual Fund alpha is not new. It has been done exhaustively in the US markets. However, to the best of our knowledge, such an analysis for Indian Mutual Funds is not available.
Our aim is to discover alpha (if any) generated by actively managed Mutual Funds.
We rely on CAPM style linear regression (time series) of historical data for each fund to extract the alpha generated.
First step is to fix the universe of funds. For our analysis, we include all types of Equity Mutual Funds except Hybrid Funds. Since we are interested in beta-adjusted returns, the analysis can be applied to any type of fund (large caps, small caps, thematic, sectoral). Therefore, we do not exclude any category.
Next step is to define the benchmark or the market. Obvious choice is the Nifty 50 Index. However, we do not consider the returns of Nifty 50 TRI directly. Instead, we consider the lowest-cost investible proxy for Nifty 50 Index - Nippon India ETF Nifty BeES (NiftyBEES). A passive investor cannot directly hold the index and hence we use NiftyBEES as the benchmark.
We could have also used Nifty 500 index as our market but instead we use Nifty 50 due to availability of liquid and low-cost investible proxy.
Benchmark is same for all categories of funds since we are interested in beta-adjusted excess returns.
Next, we take last 10 years of data from 1st Jan 2012 to 31st Dec 2021 and split it into 2 equal buckets of 5 years each - Jan 2012 to Dec 2016 and Jan 2017 – Dec 2021.
For each 5-year bucket, we select only those funds that have at least 4 years of data. For example, in Jan 2017 – Dec 2021 bucket, a fund that started in August 2017 will be included but a fund that started in April 2018 will not be included.
For each bucket, and for each fund that qualifies in that bucket, we regress the excess monthly returns (over liquid returns) against excess monthly returns of NiftyBEES to get an estimate of alpha and beta.
We also combine the two buckets to create a long-term bucket of last 10 years (since 1st Jan 2012) and repeat the regression exercise.
The purpose of creating 2 5-year buckets is to assess the consistent of performance. The purpose of long-term bucket of last 10 years is to see if results change significantly if Mutual Funds are held for a long term.
Finally, many analysis of Mutual Fund performance suffer from something called the “survivorship bias”. These analysis only include the funds that exist today and does not include funds that were closed (likely due to underperformance). Research has shown that using only live funds can overstate returns by 1% to 2% compared to a universe that includes both live and dead funds (see On Persistence in Mutual Fund Performance, Mark M. Carhart).
We control for survivorship bias by including terminated funds in our analysis. However, due to data availability issues, our analysis may not have completely eliminated the survivorship bias.
Let’s first look at the table that summarizes our analysis
The table above shows the total number of funds analysed in each period, average and median alphas, percentage of funds with positive and negative alpha.
On the face of it, the results look mixed. While the period from Jan 2012 to Dec 2016 (henceforth called Period 1) looks amazing for Mutual Funds, an exact opposite happened in the next 5 years from Jan 2017 to Dec 2021 (henceforth called Period 2).
Percentage of funds delivering positive alpha in Period 1 was a staggering 89%. But it fell dramatically to 43% in Period 2. If we consider the last 10 years (henceforth called Period 3), roughly 80% of funds delivered positive alpha.
So is it okay to conclude that actively managed funds have outperformed the passive index?
Not really.
Most important data is the last column in the table above – Percentage of funds with “Statistically significant positive alpha”.
Statistical significance is used in statistics to refute a null hypothesis. In our case, the null hypothesis is that alpha is 0 and hence a statistically significant positive alpha allows us to infer that the fund indeed has a positive alpha. If alpha estimate is statistically insignificant, we cannot infer that the fund has a positive alpha.
Look at percentage of funds that have delivered statistically significant positive alpha. Even for Period 1, only 28% of funds delivered positive alpha.
Looks like we need to explore further.
The period of Jan 2012 to Dec 2016 needs special attention here. The scale of outperformance is huge with median alpha of 4.4%. This is diametrically opposite to what happened in the next 5 years (Period 2) where median alpha is negative 0.4%.
Let’s for the time being keep statistical significance aside as we will not be trying to infer anything in this section.
In fact, performance of last 10 years itself is dominated by huge outperformance in Period 1.
So what explains this? We will discuss a few possible explanations (not exhaustive).
One possible explanation could simply be skill. But 89% is a huge number.
While we are not refuting skill (in some managers), we cannot attribute the entire 89% to skill alone.
So what else?
A closer look at data shows that most of the outperformance in Period 1 (especially the ones that are statistically significant) is driven by small-cap funds followed by mid-cap funds.
This indicates the impact of “size” factor and/or market inefficiency in small and mid-cap space
To explore the size factor, we look at the chart of Nifty 50 vs small and mid-cap indices. We do see massive outperformance by small and mid-cap indices during this period.
In order to better understand this, we have regressed small and mid-cap returns with Nifty 50 returns during this period. As expected, both mid and small cap indices delivered positive alpha of 5.4% and 2.0% respectively (although both statistically insignificant).
Next, we control for the size factor, and check if alpha still remains. Surprisingly, it does.
So, exposure to the “size” factor does not really explain the alpha. That brings us to our last plausible explanation.
Indian markets are still not mature and far from being efficient vis-à-vis developed counterparts.
This inefficiency is even more pronounced in the small and mid-cap space where there are multiple opportunities to extract alpha.
To explore this, we carried out the same regression exercise, but just for the large cap funds. Average alpha fell from 5.1% to 2.7% and only 6 funds (out of 30 large cap funds) delivered statistically significant alpha.
Table below shows the list of 6 large-cap funds with statistically significant alpha during this period.
But this does not explain inefficiency entirely. A stronger argument in favour of inefficiency is the performance in the next 5-year period (Period 2) which also happens to be the most recent 5 years.
In our view, the best possible explanation is as follows:
First, the markets have become significantly more efficient in recent times. This has led to lesser alpha opportunities and hence a significant shrinkage in aggregate MF alpha. This is a trend that will continue to get stronger.
Second, most of the star performers of Period 1 (small and mid-cap funds) hit capacity constraints due to huge inflows based on (1) past performance and (2) phenomenal growth in Equity MF AUM in general.
As defined in Part 1 of this series, capacity is the size up to which a strategy can grow (asset under management) without impacting its performance. It is a well-documented fact that the performance of actively managed funds deteriorate as the fund size grows.
And we have seen number of small and mid-cap funds closing fresh inflows as they were overrun by capacity constraints.
Again, this trend is here to stay. In fact, this is the harsh reality of active management – your past outperformance leads to large inflows which in turn leads to future underperformance.
Here, we bring back the constraint of statistical significance.
We see if a fund delivered positive alpha (statistically significant) during a period, did it deliver a statistically significant alpha in the next period. Here are some findings:
Tables in Annexure show funds that have delivered statistically significant alpha in the last 10 years taken as a whole.
Active vs Passive debate has been raging on for years.
Our take is simple:
Active managers do add value. But not all of them. Empirical studies and our own research indicates that roughly 50% (or less) active managers will outperform the benchmark. William Sharpe has even tried to show this as a fundamental phenomenon in his work - The Arithmetic of Active Management.
Even this 50% number would be lower if you control for the underlying factor exposures (beyond the market exposure). Research has shown that most of the managers who outperform are doing so by implicit exposure to risk factors – effectively a smart beta portfolio.
Having said that, active management is a huge industry and it is unlikely to go anywhere in the near future. But the trend suggests that it will become increasingly difficult to outperform the markets going forward.
In that sense, the choice of active manager becomes paramount. And odds are not stacked in your favour. Randomly picking a Mutual Fund implies a less than 50% chance of outperforming the market.
Past performance is of no help either as we have seen in our analysis of Period 1 and Period 2. In fact, past performance could end up negatively impacting the future performance due to capacity limits.
Should you then just stick to passive index investing or is there a middle ground? Well, that’s the focus of Part 3 of this series. Keep reading!
Table: Funds that have delivered statistically significant alpha in the last 10 years (Jan 2012 – Dec 2021)
