On equal weight and market-cap weight.

Apologies if I misrepresent mungofitch in my haste to get this out before the barman rings the bell for the current era. I’ve had this post in the back of my mind for many months.


I believe Jim has written in the past that an Equal Weight (EW) index beats the equivalent Market Cap weight (MC) index because:

(1) It literally just does. It’s there in the historical record (… for the markets & eras he studied…)

(2) You are less exposed to brutal losses affecting 1 or 2 very large ‘market favourite’ companies which get very overweighted in an MC index then randomly go SPLAT in some sense (failure, or catastrophic rapid decline). [A]

(3) You can imagine a simplified model of the market, 100 companies, you buy 1% of each, everything wiggles around according to it’s ‘true inherent value’ long term - so you are picking up more of the ones that are currently ‘cheap’ relative to their future price/inherent value, even without having any model of what cheap or value actually means.


BUT!

I want to point out that this model depends on a particular set of dynamics, particularly, ‘most stuff trundles along, cheap stuff outperforms’ for (3); or a dynamic in which companies rapidly go SPLAT for (2).


I propose, equal weight isn’t just about the buying, it’s also about the selling. Market cap doesn’t sell much, equal weight does - very regularly and systematically.


Here’s a different thought experiment:

A sector has two companies. Could be two actual companies, or it could symbolise ‘the group of long-term-will-be-good companies vs the group of long-term-will-be-bad companies’. Goodcorp and Badcorp.

Maybe it’s Google and Myspace and the year is 2007. We pick up 50% of each.

Now, suppose that one of the companies steadily loses value over a period of 10 years, and the other steadily gains in value, and prices wobble around but eventually tend to track the real value of the company. Eventually, perhaps Badcorp goes SPLAT suddenly.

I believe this is a not too ridiculous a model for quite a lot of history and quite a few sectors of the economy. What does the equal weight index do, repeatedly, over a period of years?

It steadily sells you out of the long-term-GoodCorp companies and into a bigger and bigger holding of BadCorps, pre-SPLAT.

EW wins when buying, if companies are very inefficiently priced, because you win every time a cheap thing later realigns with true value. And it wins when huge market cap companies go from GoodCorp to SPLAT overnight, or when tiny GoodCorp companies suddenly soar.

But it loses when selling, because it is more likely to sell whatever is doing well and put it into something that isn’t. Depending on whether you’re in an era or sector of ‘regression to mean’ for inherent value growth or ‘winners keep winning’ for inherent value growth, that makes a difference.

EW loses when a trend of failure continues long term due to the characteristics of the companies involved, you just steadily end up with a small amount of the good stuff and a (now equally small) long-term collection of steadily-bought failure.

In contrast, MC happily lets your portfolio ride a spaceship to the moon on GoodCorp.


Conclusion: I propose that that every index weighting approach has its season and probably has a ‘sector bias’ (I believe for MC vs EW it was 1/3 MC wins, 2/3 EW wins in Jim’s data, long term & across all sectors). And since countries have sector biases at different points in time (US → oil, then banks, then tech, for example, in the last 20 years), that means even a country-level MC or EW index might perform very differently than you would expect from the long term stats.

If you are in a time when steady failure over time is common, this factor may cause MC to win by avoiding the ‘rebalance into failure’ factor.

If you are in a time when the dominant sector is filled with companies that are all much the same (oil circa 2000? with Shell, BP, CVX, XOM, … much of a muchness), perhaps equal weight will be more likely to win.

If you are in a sector where winners and losers change place often (pharma? tech?), that may have an effect too, I think favouring EW, because often you’re trading against fashion and into the next winners.

A world where the largest companies go SPLAT easily? Better stick with EW.

If you are in a time when an MC-favouring sector is ascending (tech) and EW-favouring sector is in decline (oil? banks?), maybe it’s time to switch weighting style.

Particularly - the main point I want to make - if you are in an era with a dominant market sector where market leaders tend to hyper-dominate the entire sector and even the entire economy, you should see equal weight merrily distributing your gains into companies that will continue losing, over and over, for years. Any time your big winners make a gain, EW will ensure it gets thrown away on hopeless declining companies.

Since we seem to be entering a new era with higher rates perhaps for a long time, e.g. the precursor to a junk bond crisis, I think this stuff is worth thinking about.

I believe someone (WendyBG? [B]) shared a statistic about 12% of the companies in the Russell 2000 being effectively walking dead, sustained by low fed rates? What if now is not the time to be repeatedly trading out of winners into more and more of those zombies as they descend into the grave. Are the current market leading sectors (by cap weight) likely to be impacted by debt problems and rising rates compared to the sectors that would be favoured by equal weight?

Hmm.


I hope the discussion can continue on the new boards, if anyone is interested to talk about this.

LM


[A] Yes, banking and insurance sectors of 2008, I am looking at you. And today: probably Tesla, all Chinese stocks, and possibly Meta, maybe also NVIDIA/AMD/TSMC if something foolish happens.

[B] Can’t find it right now - was posted recently and along the lines of e.g. https://mishtalk.com/economics/zombie-corporations-10-of-com…

5 Likes

p.s. If anyone is looking to poke holes in my EW vs MC argument, here’s a starting point: while it’s obvious that 50% of companies are in ‘the bad half’ of any group and 50% of companies are in ‘the good half’, it isn’t at all reasonable to simply assume that all those in the bad half are declining in value long term, and those in the good half increasing [A]… this highlights a danger of thought experiments.


[A] Unfortunately, the numbers are even worse…

https://www.irishtimes.com/business/personal-finance/most-stocks-are-flops-just-1-of-stocks-account-for-all-market-gains-1.3973200

“However, Bessembinder’s latest study, Do Global Stocks Outperform US Treasury Bills? – which examines the performance of more than 61,000 global stocks between 1990 and 2018 – shows most stocks really are flops.”

“The median stock fell by 15 per cent over the period in question. Only 45 per cent made any gains.”