Comment by blippage | Hacker News Reader

Comment by blippage | Hacker News Reader

Parent Full thread

Warren Buffett bought IBM shares in 2011 at $170, and sold them in 2018 at $145. Oops.

When he bought, he marvelled at how the CEO was targetting 20% returns on capital. When I heard this, I realised that Buffett had forgotten one of his lessons that he himself espoused: that companies that "target" returns on capital are should be avoided. IIRC, it was one of his reasons for bailing out of Fannie Mae (?). His timing was impeccible on that one, as the whole thing exploded not long thereafter in the 2008 financial crisis.

Also at the time of his purchase, it was known that the company was beginning to offshore. It struck me at the time that this could cheapen the whole thing and turn into a disaster.

So there were at least two facts at play here that should have forewarned Buffett that it might not be a good purchase. I was actually pretty astonished at the time that a guy as savvy as Buffett didn't seem to anticipate the problems ahead.

And now we come to the $64,000 question: how does a company add value to shareholders? It's a simple question, but I bet most won't know the answer to it. The answer is provided by Prof. Damodaran Aswath: a company adds value by targetting returns on capital in excess of the cost of capital, risk-adjusted.

So actually, doing stuff like maximising returns on capital is a sub-optimal goal. Not that you should invest in something marginal, either.

Another problem with these corporate-type guys is that they do a lot of "management by numbers". It's all about "KPIs" (Keypoint Indicators) and other "metrics".

Back when Motley Fool wasn't a shitpile that it is today, there was a guy in the UK that put together a portfolio of "family companies", i.e. quoted companies with a substantial management and/or family stake.

How did the companies fair? Very well, actually. The portfolio did great relative to the indices. Why? Well, my conclusion is that family-concentrated businesses were run conservatively as proper operating businesses, rather than a bunch mathematical formulae to be tweaked and fiddled with.

"KPIs" (Keypoint Indicators)" --> I believe KPI stands for Key Performance Indicator. Source: https://www.klipfolio.com/resources/articles/what-is-a-key-p...

"Another problem with these corporate-type guys is that they do a lot of "management by numbers"" --> What is the alternative to this? Management by intuition? Coordinating a company with ~350,000 employees around the world is definitely going to require some structure and metrics to measure how things are going. There are certainly ways that this can be abused, but I don't think the answer is to have no measures of how the business is performing.

> So actually, doing stuff like maximising returns on capital is a sub-optimal goal.

Could you explain why this is sub-optimal? It's sub-optimal because of the added risk?

How does this work for very big companies that have to engage in multiple markets simultaneously (due to couplings between markets and other competitors - eg. FB bought Instagram, but FB's main market was not buying other social media companies, but it fit into their strategy to get more active users)?

> Could you explain why this is sub-optimal? It's sub-optimal because of the added risk?

What usually happens is that they make it a quarterly or annual target, so they do things that expensive in the long term to hit the target for the period.

For example, I can increase my profit for the quarter in an airline by deferring maintenance, so the plane flies more. But if the plane ends up being out of service for longer next year because of that decision, that is sub-optimal.

Another thing they can do is shift revenue around or offer large future discounts for upfront orders or shift project work depending on when end of year is. All things that are suboptimal, but work for the numbers.

As a non financial example, one company I worked for had quarterly goals around test coverage. But they didn't really budget time for it. It was just expected that it would somehow be done.

What happened was that worthless tests that called as much code as possible without testing anything were written to hit the coverage number. So even when time was available to write tests, the perception was that we already had test coverage so they never got fixed.

> Could you explain why this is sub-optimal? It's sub-optimal because of the added risk?

Suppose my cost of capital is 5%. My current business (selling widget A) is returning 20%. Great, my ROCE is 20%. Now let's say I can't expand my current product line any further. I can only sell so many widget A's.

But supposing I invent a new widget, widget B. Its return on capital is, say 10%. Should I manufacture widget B? The answer is: yes. My cost of capital is still 5%, but I'm earning a 10% return on it.

So the value of my business has increased. My overall ROCE will go down, though, because they'll now be a mix of sales at 20% ROCE and some at 10% ROCE. But even though my ROCE has decreased, it is still a good idea for me to produce widget B.

It may "look" bad to investors that my ROCE is going down, but I'm actually producing value for them. That's why I say that ROCE maximisation is sup-optimal. If I scrap widget B, them I'm actually doing investors a disservice.

I'm assuming "all things being equal", like risk profile.

In practise, they of course won't be. Companies could play a short-term game, like IBM seems to have done, by outsourcing to India, or whatnot. Sure, in the short term, ROCE has increased, but in the long term people will get cheesed off by their poor service.

It can work they other way, too. Take, for example, Kraft vs Nestle. Let's call it the "American" model vs the "European" model ;). Kraft took over UK chocolate manufacturer Cadbury's a few years ago now. They cut costs and produced something that consumers didn't like so much. In some ways I guess they had to, considering the amount of dough they blew on the acquisition.

Nestle, OTOH, tried to sell their stuff in Africa. That required a lot of money spread over a long period of time, because Africa has poor infrastructure. But what it meant was, when things got going, they were in a position to sell products to Africa when others weren't.

It's a long game vs a short game. Of course, just because it's a long game doesn't necessarily imply it's better.

But you can see that "hitting next quarter's targets" is apt to produce disappointing results over the longer term.

As regards FB buying Instagram, I have no sage advice. I could never see what the fuss was about with FB anyway, but they're making plenty of money, so I make plenty of mistakes.

Big companies often trot out phrases like "synergy" and "strategic acquisitions". They might work, they might not. Statistically speaking, odds are against acquisitions. Companies generally overpay for what they believe is exciting stuff, only to have the whole thing be a huge financial fiasco. Just ask Yahoo. They did alright with Alibaba, though. As another example, take Google's acquisition of YouTube. As far as I'm aware, they never made a profit out of it.

Thanks for explaining!

So isn't "returns on capital" measured in dollars, not percentages? I mean investors/shareholders will look at dividends and stock price appreciation, not ROCE.

> It's a long game vs a short game. Of course, just because it's a long game doesn't necessarily imply it's better.

Yes, exactly. Maximizing returns is not specific enough, "maximizing over X timeframe" is. And this should be decided by shareholders, right? (And so they elect a board that then picks a strategy.)

> YouTube

YouTube brought in revenue of $6.01 billion in advertising revenue during the quarter. Sure a lot of that goes to content creators, royalties, etc. And running the site costs a lot of money too, but if YT can't make a profit on this much money then I'd be very surprised.

Reply on news.ycombinator.com