Why Shankar Sharma calls Nifty 50 P/E a colossal idiocy
I decided to write this piece because recently I noticed that a number of fund managers and social media research folks getting their thongs in a twist (the visual thought of this is horrifying, I admit) because Zomato replaced JSW Steel and this raised the Index PE multiple, and this, upset the bull argument that the market had become a screaming buy. (But when was it not, at least, if you heard all fund managers, wealth managers brokers and TV anchors for past several months?)
The rationale for using the P/E or P/BV to indicate under-or-overvaluation for the market rests on the basic principle that over the long-term, certain P/E (or P/BV) ranges signify that markets have become very cheap. In order to do that, analysts will compare today’s P/E to the P/Es that have prevailed in 1910 or 1933 or 1945 or 1966 or 1974 or 1982 or frankly, just about any period.
So, the logic ( if it can be called that) will go something like this: “In `19XX, the P/E for the SENSEX, Dow or the S&P500 was X, and now it is 0.5 X, so today, the market is cheap compared to its long term range.” or some such fustian codswallop.
Let's dig deep into why this Index PE ratio thingy is as silly as expecting Indigo airline staff to behave well.
This so-called P/E ratio for the market is based on something called the EPS for a wide market index. Simply put, this measure aggregates the earnings for all index constituents and then you come up with the EPS for the Index. And then, divide the Index with this EPS and you get the P/E for the Index. Or you can simply aggregate all the earnings of an index and divide the aggregate market cap of the index by the aggregate earnings to get the market’s P/E.
Just looking at this abomination, one gets to understand why almost no sell-side strategist gets their index targets right consistently. For, if you base your buy or sell decisions on the market on such flapdoodle, you will get the equivalent of judging the popularity of TV shows in India based on meters installed in 600 homes in a slum in Mumbai - remember the TAM ratings scam?
Don't screw the skew
Let’s take a simplified example of this Index P/E thing: assume there are 5 companies in an Index. They earn $10 million each. Their combined market cap is $500 mln. Hence, their “P/E” is 10x, i.e., $500 million/sum of earnings: $50mln.
Let’s take the same example with different numbers: The basket of five companies earns the same $50 million, but now the distribution is different. Company A earns the entire $50 million, while the other four earn zero profit. Market cap of the basket is the same as before: $500 million. P/E for this sample: 10x.
And finally, let’s take a third variant: Company A earns $100 million, the rest of the pack loses $50 million in aggregate, earnings for the Index comes to the same $50 million, and assuming market cap of this Index is still $500 million, you still end up getting a P/E multiple of 10x for the Index.
Begin to see the absurdity of this “P/E multiple for the market” concept? In each of the above case, the P/E is identical, but the complexion of the market is vastly different. In Case 1, the distribution of earnings is less skewed, and the skew increases markedly in the balance two examples. But does anybody account for the skewness of the earnings composition for the market? Well, when you say that the SENSEX will earn X EPS this year, and hence, the SENSEX trades at Y P/E, I don’t see any mention of the degree of skewness of this EPS, ever.
And even more importantly: are you indifferent to the three scenarios outlined above, given that the P/E is identical in all three cases? Not unless you are on hallucinogens.
But I am not yet done with this thing. The moonshine gets even more vivid:
The Index PE ratio: Absurdity Squared
As we saw above, we start with one absurd EPS number, then derive an even more absurd P/E number, for the Index. And then, we compare this P/E with the P/Es across market history. And what we get is Absurdity-Squared.
The aggregate earnings of any index basket across time periods are simply not comparable.
For starters, it is simply because the index constituents change very frequently.
Just to illustrate, between 1995 and present day, of the original NIFTY 50, only 12 - yes, 12 - remain. Thirty eight have changed, i.e. , over 70% of the basket is different today.
Let me jog your memory: we had in the original NIFTY, gems like (hold your nose, and keep a bottle of Gelusil handy): Ballarpur, Arvind Mills, Bombay Dyeing, Essar Gujarat , GE Shipping....I think I should stop before I get to Videocon because it was there in the NIFTY too.
By 1999, we had a bunch of these losers thrown out, to be replaced with the 200x PE Formula 1 stocks like Infosys, Zee, and Satyam.
And, now get ready for this... over the years, Cairn India, that permanent no-hoper rubbish IDFC.
Ready for this? Indiabulls Housing was also admitted to this hallowed club.
And just before Nirav Modi did his SWIFT heist, PNB was too.
And so the giant wheel of this Index rigmarole continues to present day and forever it shall continue.
Are you nauseous enough?
Globally, too, between 2000 to present day, 36% of the S&P500 constituents have changed, for the Footsie, 46% have changed; 59% have changed in the NASDAQ 100, 20% and 30% have changed for the German DAX and French CAC, respectively.
And there isn’t even a like-by-like substitution. Thus a manufacturing company can get replaced by a bank or a pharmaceutical company by a real estate company. Zomato replaced JSW Steel, remember?
What then, is the comparability of the P/E of an Index of 5-10-15-20 years back, with the same Index today, but with 30-50% of the constituents, with different industries, different sector multiples, frankly, everything being different?
Zero.
Forget the PE, the "Index EPS" itself is composed differently across era, depending on which sector dominates the Profit pie
Additionally, what contributes to the Index’s earnings - i.e., EPS - will differ dramatically from era to era.
Sometimes Oil & Gas might contribute 40-50% to an index's profit pool.
Sometimes Financials dominate.
Sometimes, tech profits can overpower the Index profit pie chart.
Each of these businesses trade at vastly differing multiples. And hence, the PE multiples of the same Index in an era when Energy dominated the profit pie, vs a Tech heavy profit pie, simply cannot be compared by anybody with an IQ of 30 or above.
Calling these eras apples and oranges is simply a mild rebuke. After all, those are still both fruits.
Even the unchanged companies change dramatically, business -wise, over years.
But it gets worse: even the constituent companies that do remain the same across eras, may not be the same companies in characteristics, at all.
For example, a company may diversify into other businesses; may acquire other companies (or even divest businesses), thereby changing its business profile markedly; further, acquisitions will lead to absolute earnings themselves changing, completely rendering meaningless the comparability of earnings of years prior to the big acquisition and the ones thereafter; become more global, thereby making its growth profile simply non-comparable to the company in 30 years ago.
Reliance is an excellent case in point. How can one possibly compare the Reliance PE multiple of today given the high services component in its overall business to the smokestack refiner of the 90s?
Ditto for Tata Motors, which went from a pure play CV player in the 90s, to a global passenger car company today.
IBM under John Akers in 1985 was a totally different company from what it became under Lou Gerstner. Can you at all compare IBM’s P/E of 1985 to its P/E of 1995 to its P/E of 2009, given its shift away from hardware to services?
Truth is: almost no company remains the same over 30 -40 years. So how can you even compare their multiples across time frames and different business models?
Into the Rareified Absurdity-Cubed Territory
Let’s shred this some more: can there be a single unified metric like the P/E or P/BV applicable to a totally heterogenous group of companies, like in a broad market Index?
Can you at all draw a unified " PE multiple" line through an auto company, a bank, a broker, an oil producer, a restaurant chain, a cosmetic company, a pharmaceutical company, an insurer, a steel company, a defense supplier, a computer software maker, a massage parlor chain (ok, ok, we’ll drop that). Each of these businesses differs completely on all parameters of business.
It is hard enough to make intra-sector valuation comparisons, unless all the players within a sector are near identical and homogenous. But they rarely are. Even a pure commodity company (say, a steel producer) may well differ from another commodity company because of balance sheet risk, growth options, proximity to end markets, access to raw materials, etc., all of which affect valuation ratios.
It beggars belief that vast sums of capital are allocated daily based on such imbecility.
The Macros - Interest rates, inflation - of the era are a huge determinant of PEs in that era... and these change dramatically over time, too.
Let’s look at this another way, and the complexity of this whole exercise (that of valuing the market) will simply blow you away:
The mathematical formula for the P/E ratio is: (1-g/r)/(k-g), where g stands for growth, r stands for Return on Incremental Capital, and k stands for Cost of Equity Capital.
Simply put, when you have high growth (g), and the Rate of Return (r) vastly exceeds your Cost of Capital (k), you will be blessed with a high P/E ratio. It’s pretty much the same with the P/BV: this ratio can be broken down into P/E x E/BV= P/BV. So if your RoE (Earnings/Book Value) is high (or getting higher), you will sport a higher P/BV.
(However, if you are earning these high RoEs in generally high inflation periods, you will get a lower P/BV, because your P/E will suffer due to high inflation (due to the resultant increase in k, cost of capital)).
Begin to see where we are going with this? We have seen earlier how the P/BV or P/E of any Index in a certain era depends on the characteristics of the constituents of the Index in that era.
But it also depends on the RoE, growth expectations, inflation (and expectations), interest rates (and expectations), cost of capital (and expectations) of the basket! (In fact, let me complicate this even more: if in an era, financials dominate the earnings pie chart of an index, as they did for the S&P500 in the 2002-2005 period, then even if the RoE of the Index may be higher, you would be well advised to pay lower multiples for this higher RoE, simply because financials produce high RoEs largely from the use of high leverage, which, in turn, increases risk, which in turn, increases cost of capital, which in turn, lowers P/E ratios… anybody still awake?)
In fact, interest rates are the single biggest determinant of a country's PEs... and nobody in their right senses, except dimwits on the research side, will assume that interest rates remain constant over decades.
For example, US rates have gone from 16% in 1982 to 5% to 1% to 5% over 40 years. Indian rates have gone from mid-teens in the 90s to single digits now.
P/E is nothing but the reciprocal of the earnings yield, i.e., Earnings/ Price. And this yield is the outcome in no small part, of the general interest rates in the economy at that point in time. We simply can't compare PEs for the market across decades; interest rates themselves bounce all over the place over decades.
Delivering the final coup de grace through the heart of evil NIFTY PE Ratio
Apart from all the fallacies and downright conceptual vulgarities involved in such an exercise as detailed earlier, unless you can adjust (I haven’t the faintest idea how one can do it) the k, r, and g of that era, and normalize it with the k, r and g of this era, the two ratios can never be made even remotely comparable. (Also don’t forget that markets are essentially forward looking, so the g, r, and k have to be the forecast values. And just how can you get the forecast, expected value of these variables, going back in time?)
So there you have it: this “P/E of the Market” concept is absolute baloney. It fails miserably to stand up to any analytical and conceptual rigor. So why do such bright (ahem) people use this?
I’ll tell you why: it is nothing but the lazy man’s approach to investing. It is an attempt to make complex things so oversimplified that everybody can swallow it.
It is simply the dumbing down of what is a very complex science: to figure out what the true intrinsic value of any market, sector or stock is.
There is nothing constant about an Index: its constituents, its profit pie chart, its profit skew, inflation and interest rates in the economy, and even the business models of the long surviving constituents.
While simplification is good, oversimplification is a crime, in wealth and in health.
You can barely just about compare PEs for a homogeneous group of companies at the now, instant point in time. But never historically.
And never ever for the market as a whole, longitudinally over time.
And before I leave, let me also demolish this foreign brokers' "cross country comparison" flatulence, in order to justify which market is " cheap" or " expensive" relative to others.
So, somebody please tell me: how on earth can you compare PE multiples of Brazil ( commodity heavy), to India (heavy with God knows what all!) to China ( Tech and financials) to Taiwan ( manufacturing), and then, pompously pontificate on cheapness of X vs Y market ?!
So, the next time your broker or fund manager tells you about how the market is cheap based on the Index PE multiple based on historical " PE ranges" , you know exactly what you need to do: give him a piece of your mind , ie, *@#&+##$&@+$&&@())(+§×÷π√£%^, send him a copy of this article and hang up.
And maybe, write to the Regulator saying that this entity is publishing false, misleading and dangerous research.(Views are personal, and do not represent the stance of this publication.)