Invest Without Emotions
Indian bull vs. the US bull since 2012:
| SYMBOL | Cumulative Returns | Annualized Returns |
|---|---|---|
| SPY |
115.63
|
15.25
|
| EPI |
70.47
|
10.36
|
| INDY |
78.93
|
11.35
|
| SCIF |
74.11
|
10.79
|
| PIN |
50.53
|
7.85
|
| INP |
68.36
|
10.10
|
| INCO |
164.49
|
19.69
|
| INDL |
20.98
|
3.58
|
| INXX |
38.16
|
6.15
|
Investing in the Indian story has been a drag on returns for US$ ETF investors.
The code is on github.
A fund’s alpha – returns over a benchmark – is often quoted and widely misunderstood. The root of the misunderstanding comes from investors assuming that alpha is a constant – which it is not – and the funds using benchmarks that the investors cannot actually invest in. Even if an investor decides to go “passive,” there is still an active choice that needs to be made regarding the basket of ETFs he needs to invest in. Let’s answer the first question: What exactly is the active manager’s value add?
We select three ETFs, NIFTYBEES, JUNIORBEES and M100, since they are popular and span a fairly decent spectrum of traded stocks in the market. Then, we do a rolling (window of 200) linear regression of returns over 200 days of a few midcap funds (selected at random.) The intercept is the alpha of the fund vs. the ETFs. Here’s how the alpha varies over a period of time:
Two out of the three funds have negative alpha over the ETF basket right now. However, that doesn’t mean that they will stay there.
As an investor, you can use the betas obtained by the regression over the ETFs to “replicate” the fund at a point in time. For example, if you set the start date as the date at which each of the funds had peak alpha and just held onto the basket, here’s how the relative performances look:
In all cases the basket fixed at the peak performs at par or better than the fund. However, you never really know what the “peak” is when you are living through it. What if you fix the basket right at the beginning?
In two out of three cases, we see funds beat ETF baskets.
The NIFTYBEES ETF – an ETF that is indexed to the NIFTY – has less daily volume than RELIANCE. Median daily volumes of NIFTYBEES is around 31,000 whereas RELIANCE sees more than 3161,000.
NIFTYBEES:
RELIANCE:
The NSE introduced a Liquidity enhancement scheme (LES) for market making in equity exchange traded funds (ETFs) effective from December 15, 2014 till February 28, 2015. It was then extended to June 30, 2015 (see appendix). The results have been mixed.
BANKBEES:
JUNIORBEES:
M100:
Daily volumes of M100 went down during the program. Here is the full list, volumes in ‘000s:
The program might have resulted in tighter bid-ask spreads but there was no surge in volumes. Retail investors remain disinterested in ETFs.
ETFs (Exchange Traded Funds) offer investors a convenient way to gain exposure to a particular index. Since these funds are not actively managed, they are measured by their how cheap they are (in terms of asset management fees) and their tracking error. Before we begin, some key terms:
Kurtosis is a measure of “peakedness” of a distribution. For a normal distribution, Kurtosis is 3. Positive excess Kurtosis indicates fat tails while negative indicates peakedness.
Skewness is a measure of asymmetry of a distribution. Positive skew indicates a long right tail while a negative skew indicates a long left tail.
How effective have Indian ETFs been? Lets pull up some histograms and see for ourselves.
The Nifty ETF actually shows a -ve skew and a lower kurtosis compared to the index. This is how tracking error and fees manifests itself in daily returns. However, their impact on cumulative returns is minimal. The story for less liquid and higher-fee ETFs are different.
Notice the big difference in kurtosis and skewness? This is tracking error personified. The story for the Juniors’ are not that different.
Given that there really isn’t much of a push either from investors or from asset management firms on ETFs, the dynamics are unlikely to change in the short term.
The two holy grails of investing: dollar cost averaging and low-cost investing come together if you systematically invest in an index ETF. We took a look at returns on doing an SIP on JUNIORBEES, an ETF that tracks the Junior Nifty index, that was introduced in 2003.
| Start Year (Jan) | IRR |
|---|---|
| 2004 | 10.77% |
| 2005 | 9.48% |
| 2006 | 8.66% |
| 2007 | 8.47% |
| 2008 | 9.81% |
| 2009 | 8.70% |
| 2010 | 4.91% |
| 2011 | 7.40% |
| 2012 | 8.58% |
| 2013 | 5.33% |
The question we set out to answer was: What would typical returns be if you systematically invested in a low-cost index ETF over different periods of time?
So we assume that the investor buys Rs. 5,000 worth of JUNIORBEES at the closing price on the last day of each month. We accumulate the units, the cost basis and the P&L over different periods of time, starting at 2004 and moving forward in one-year increments.
The dollar cost averaging ensures that you buy more ETF units when the index goes down and less of it when it trades higher. And by tracking the IRR we ensure that we normalize returns for the investment period.
We expected nominal returns to be higher than what we observed. Between 2004 and 2014, inflation was often running in double digits. So even a 10% IRR would actually be negative real returns. Investors probably would have made better returns if they had kept the money in a bank fixed deposit instead. So from a purely returns perspective, an SIP on an index ETF doesn’t make sense.
Caveat: Just because the real returns were negative with this approach in the past, doesn’t mean that it will be so in the future.