Tuesday, January 10, 2017

A puzzle for the risk manager

The last two posts were essentially about picking a value-stock portfolio and managing the risk. And they were lessons that I thought I could implement.

This is stuff I find harder. So I am looking for your input.


This is the portfolio of a fairly well known value investor in March 2008. I have taken the name off simply because it doesn't help but there was roughly $4 billion invested this way.

To put it mildly this portfolio was very difficult over the next twelve months.

Sector allocation Positions
Banks - Europe 24% Fortis, ING, Lloyds, RBS
Banks - Japan 14% Millea, MUFJ, Mizuho, Nomura
Banks - USA 8% Bank of America, JP Morgan
Technology - PC & Software 18% Linear Technology, Maxim Integrated products, Oracle
Semiconductor equipment 14% Applied Materials, KLA Tenecor, Novellus Systems
Beer 20% Asahi, Budweiser, Group Mondelo, Heinekin, InBev
Media 15% Comcast, News Corp, Nippon Television
Other 14% eBay, Home Depot, Lifetime Fitness, William Hill
Net effective exposure 127%
Shorts -16%
Net exposure 111%
Cash -11%

The PE ratios mostly looked reasonable and all of these positions could be found in quantity in the portfolios of other good investors. Its just the combination turned out more difficult than average.

Your job however is to risk-assess the portfolio.

Even with the considerable benefit of hindsight how would you analyse this portfolio?

What would you say as risk manager that made the portfolio manager aware of what risks he was taking?

What would you say if you were a third party analyst trying to assess this manager?

What is the tell?

Remember the portfolio manager here has a really good record and the "aura" around them. They are smarter than you.

And yet with the restrospectascope up there is stuff that is truly bad.

They had four European banks making up a quarter of the value of the portfolio. Most European banks went through the crisis hurt but not permanently crippled. Permanently crippled came later with the Euro crisis.

The four European banks here (Lloyds, RBS, Fortis, ING) however received capital injections so large that they were effectively nationalised. If you had thrown darts at European banks it might have taken hundreds of rounds to pick four that bad... They could not have been picked this bad by chance - they had to have systematic errors here.

There is something really wonky about this portfolio - and it is not by chance - so there was something faulty about the way the portfolio was constructed.


PS. It is fair to say some of the portfolio (News Corp for example) was awful in the crisis and came back stronger than ever. And some that I would have thought ex-ante high risk (such as the semi-conductor capital equipment makers) turned out okay - having "ordinary" draw downs in the crisis and recovering them since.

PPS. I kept the document this came from because at the time I thought the portfolio was absurd and would end in tears. But some of my thoughts then were wrong too - especially re the semiconductor capital equipment stocks.

Wednesday, January 4, 2017

When do you average down?

The last post explained why I think a full valuation is not a necessary part of the investment process. A decent stock note is 15 pages on the business, one page on the management, one paragraph or even one sentence on valuation.

Valuation might normally be a set of questions along the lines of "what do I need to believe" to get/not get my money back.

But I would prefer a simple modification to this process. This is a modification we have not done well at Bronte (at least formally) and we should do better. And that is the question of averaging down.


Warren Buffett is famously fond of "averaging down". If you liked it at $10 you should love it at $6. If it goes down "just buy more". And in the value investing canon you will not find that much objection to that view.

But averaging down has been the destroyer of many a value investor. Indeed averaging down is the iconic way in which value investors destroy themselves (and their clients).

After all if you loved something at $40 and you were wrong, you might love it more at $25 and you almost as likely to be wrong, and like it more still at $12 and could equally be wrong.

And before you know it you have doubled down three times, turning a 7 percent position into a 18 percent loss.

Do that on a few stocks and you can be down 50 percent. And in a bad market that 50 percent can be 80 percent.

And if you do not believe me this has a name: Bill Miller. Bill Miller assembled a startling record beating the S&P ever year for fifteen straight years and then blew it up.

Miller had a (false) reputation as one of the greatest value investors of all time: In reality he is one of the biggest stock market losers of all time and a model of how not to behave in markets.

How not to behave is be a false value investor, buying stocks on which you are wrong, and recklessly and repeatedly average down.


At the other end traders who (correctly) think that people who average down die. The most famous exposition of this is a photo of Paul Tudor Jones - with a piece of paper glued to his wall stating that "losers average losers".

And yet Warren Buffett and a few of his acolytes have averaged down many times and successfully. And frankly sometimes I have averaged down to great success.

At least sometimes - the Bill Miller slogan is correct: "lowest average cost wins". Paul Tudor Jones may be a great trader - but he is not a patch on Warren Buffett.


I would love it if I had an encyclopaedic knowledge of every mid-cap in Europe and could buy the odd startlingly good business when tiny and cheap. But the task is too large. The world is complicated and I can't cover everything.

But when I look at tasks that can be achieved by a four-analyst shop I have one very high on my list of things we can do and should do: We should get the average down decision right more often.

So I have thought about this a lot. (The implementation leaves a little to be desired.)


At a very big picture: averaging down when you are right is very sweet, averaging down when you are wrong is a disaster.

At the first pick the question then is "when are you wrong?", but this is a silly question. If you knew you were wrong you would never have bought the position in the first place.

So the question becomes is not "are you wrong". That is not going to add anything analytically.

Instead the question is "under what circumstances are you wrong" and "how would you tell"?


When you put it that way it becomes obvious that you must not average down (much) on highly levered business models. And looking at Buffett he is very good at that. He bought half a billion dollars worth of Irish Banks as they collapsed. They went approximately to zero. But he did not double down. He liked them down 90 percent, he did not like them more down 95 percent.

By contrast these are the stocks that Bill Miller blew up on: American International Group, Wachovia, Washington Mutual, Freddie Mac, Countrywide Financial and Citigroup. They were all levered business models.

By contrast you can probably safely average down on Coca Cola: indeed Buffett did. It is really hard to work out a realistic circumstance in which Coca Cola is a zero. And if it is still growing there is going to be a price at which you are right - so averaging down is going to go some way to obtaining an average cost near or below that price. 

Of course even Coca Cola is not entirely safe. You could imagine a world where the underlying problem was litigation - where some secret ingredient is found to be a carcinogen and where the company faces an uncertain future of lawsuits. It is not likely - and if it happens you are going to get at least some warning that this is a circumstance on which you could be wrong. Whatever, outside that circumstance on which you might be warned, Coca Cola is not a leveraged business model subject to bankruptcy and is almost entirely unlikely to halve four times in a row. You can average that one down.

Operationally levered business models

Not every business model is as as safe as Coca Cola. Indeed almost every business model is more dangerous than Coca Cola. A not financially levered mining stock can halve five or six times. If you have a mining company that mines coal at $40 per tonne, has no debt and the price is $60 a tonne it is going to be really profitable. But prices below $40 (highly possible) will take profits negative. Add in some environmental clean up and some closing costs and it is entirely possible that a stock loses 95 percent of its value. Averaging down when down 40%, some more when it halves, and then halves again and it will still lose two thirds of its value. The difference between averaging stuff like that down and doing what Bill Miller did is only one of degree.

It is still a disaster. And you will have proven Paul Tudor Jones adage: losers average losers.


There is another iconic way that value investors lose money - and that is technical obsolescence. Kodak was made obsolescent and was a value stock all the way down to bankruptcy. The circumstances on which you might be wrong (digital photography going to 95 percent of the market) could have been stated pretty clearly in 1999.

You might thing it was worth owning Kodak as a "cigar butt stock" - plenty of cash flow and deal with the future later. There was a reasonable buy case for Kodak the whole way down. But technical obsolescence is always a way you should be wrong. When the threat is obsolescence you are not allowed to average down.

Bill Miller averaged Kodak down. Ugh.


If I could improve our formal stock notes in any way I would like an ex-ante description of what circumstances we are allowed to average down a particular stock, and how much.

We have a default at Bronte - and the default at Bronte is that we have a maximum percentage for a stock (typically say 9 percent but often as low as 3 percent depending on how we assess the risk of the stock) and as the fund manager I am allowed to spend that whenever I want but I am not allowed to overspend it. If we have a 6 percent position with a 9 percent loss limit and it halves I am allowed to add three percentage points more to the exposure. But that is it. Simon, being the risk manager, isn't particularly fussed if add the extra when the stock is down 30 percent of 50 percent, but I can't add it twice. If it is a position on which we agree we are allowed to risk 9 percent then I am allowed to risk 9 percent.

We will not fall for the value investor trap of losing 18 percent on a 7 percent position.

We have made a modification of this over time. And that is every six to nine months I get another percentage point to add. That is at Simon's discretion - but the idea is that the easiest way to find out whether you are wrong is to wait. After a year or two the underlying problem will usually become public. If time has not revealed new information then we are allowed to risk more.

But we can and should do better with ex-ante descriptions under the circumstances in which we are prepared to add and circumstances where we are not. The problem is that you can wind up in a mindset where you always where you want to add, where you think the world is against you and you are right and you will just be proven to be right.

Clear ex-ante descriptions of the issue (which require competent business analysis) might help with that problem.


The bad case of averaging down

The iconic bad situation to average down is a levered business model involving fraud. It is surprisingly common because people who run highly levered business models have very strong incentives to lie or to cover it up when things turn to custard. I can think of two recent examples: Valeant and Sun Edison.

Much to my shame I added to my (small) position in Sun Edison as it fell. Ugh. But also this was a highly levered business model and thus by definition the sort of place where losers average losers. I should not have done it - and I won't in future.

But the highly levered business models apply fairly generally. When Bill Ackman rang Michael Pearson and asked if there was any fraud at Valeant he already had the wrong mindset. Then he added to a large holding in a company with over 30 billion dollars in junk-rated debt. Losers average losers.

Incidentally our six month rule (before you were allowed to add) would have saved Mr Ackman a lot of extra losses. Time has revealed plenty about Valeant. And it would have saved me at Sun Edison too.


Whilst I think that someone asking me (as per the last blog post) for a valuation on every stock is absurd, I think it is entirely reasonable for them to ask "under what circumstances would you average down". If you can't answer that you probably should not own the stock. I should insist on it with every long investment.


Tuesday, January 3, 2017

Valuation and investment analysis

I just had a chat with someone who wondered why I did not have a valuation for everything in my portfolio - a buy and a sell price.

My reaction: such (false) precision was silly and ultimately counter-productive.

To demonstrate I will give you a set of accounts for a consumer staples company.

Annual Standardised in Millions of U.S. Dollars
Net Sales7,6586,9777,212
Total Revenue7,6586,9777,212
Cost of Revenue, Total3,6333,4543,860
Cost of Revenue3,6333,4543,860
Gross Profit4,0253,5233,352
Selling/General/Admin. Expenses, Total2,6652,4462,368
Selling/General/Administrative Expense2,6652,4462,368
Labor & Related Expense------
Advertising Expense------
Interest/Investment Income - Operating------
Investment Income - Operating------
Interest Exp.(Inc.),Net-Operating, Total------
Unusual Expense (Income)36(195)0
Restructuring Charge361800
Impairment-Assets Held for Use------
Loss(Gain) on Sale of Assets - Operating0(375)0
Other Unusual Expense (Income)------
Other Operating Expenses, Total------
Other Operating Expense------
Other, Net------
Total Operating Expense6,3345,7056,228
Operating Income1,3241,272984
Interest Expense, Net Non-Operating(297)(208)(168)
Interest Expense - Non-Operating(297)(208)(168)
Interest Capitalized - Non-Operating------
Interest/Invest Income - Non-Operating336306151
Interest Income - Non-Operating232154151
Investment Income - Non-Operating1041520
Interest Inc.(Exp.),Net-Non-Op., Total3998(17)
Gain (Loss) on Sale of Assets------
Other, Net13597
Other Non-Operating Income (Expense)13597
Net Income Before Taxes1,3641,4051,064
Provision for Income Taxes496471386
Net Income After Taxes868934678

As you can see - it has net income after taxes of just under $900 million.

I am not even going to bother inserting a balance sheet. The company has some debt (as seen by the interest expense) but there is little doubt the debt can be paid - and you can give me a valuation before debt if you want.

There are some substantial (foreign) cash balances as well as well as some investments. The debt and the cash balances and investments are roughly a wash - so you can safely ignore them.

The company has a long record of slow but steady growth - but it has grown a bit faster than that for the past few years. The CEO has been a vast improvement on other CEOs and has done some optimisation.

There is no doubt about the validity of the business. I guarantee you that you have consumed the product.

Also it is a highly stable product and hence should be very amenable to valuation. Volume growth is unlikely to exceed 5% in any year. A volume decline of 5% would be an unlikely disaster. However the last year did have volume growth above 5%.

Before you read any further I want you to write down a range of valuations. Just a lower bound (where on this information you would be falling over yourself to buy it) and an upper bound (where you would be falling over yourself to sell it).

Go on - write it down.

The trick is 40 lines further down - so write down your numbers before you scroll further...

Yes further down.

Further down still.

A little further down.

Okay - I have changed the dates. The real dates for this are 1987, 1986, and 1985 respectively.

And the company in question is Coca Cola.

These are the accounts Warren Buffett bought his stake on.

The market cap is now $178 billion.

I do not think any of you would have come up with a number anywhere near that high. Even if you had bought the stock at the high range for plausible values (say 30 times earnings) the return from then to now was (highly) acceptable. The stock was trading at about 12 times earnings then.

Net income is now over $7 billion and the multiple has expanded a lot.


I do not need to say it - but a valuation was not important in the buy case and would have detracted from the buy case a great deal.

The valuation as such was pretty trivial. Was it realistic to assume that the company over a reasonable time frame could return $12-15 billion to shareholders. The answer to that was a resounding yes.

Was there a margin of safety around that?

Again a resounding yes.

So the stock was easily able to be owned.


The questions that mattered (and still matter) is "can the product be taken to the world", and will the next generation think of it in the positive light the last generation thought of it.

The answers are less obvious now than they were then. Young people it seems drink Red Bull rather than Coke in surprising numbers. They are your future.


This is a general quality of investment analysis. Proper valuations are far more art than science. DCF valuations - especially of something growing near or above the discount rate are famously sensitive to assumptions. The right comparison is to the Hubble Telescope: move direction a fraction of a degree and you wind up in another galaxy.


By contrast there are some things for which a proper valuation should be done and can be done.

If you own a regulated utility what you really own is a regulated series of cash flows with regulatory risk around them.

An accurate valuation is part-and-parcel of the analysis - because it delineates what you own.


The battle here is to work out what the salient details are. Sometimes they are whether young people will continue drinking Red Bull. Sometimes they are working out a technological change.

In rare cases they are working out valuation.

Mostly valuation is simply about bounding a margin of safety. And most of that involves understanding the business anyway.


PS. If you work for a shop that requires a valuation for everything quit now. The pretence will either kill you or your performance.

PPS. I do not think there is a margin of safety around Coca Cola any more. Not enough to make me interested anyway.


Later post-script:
This is in the comments so frequently that if  you look at Coke's appreciation (and compare to the S&P) Buffett has not done that well. Some even say "if you ignore dividends". But that misses the point.

Here is an extract from Berkshire's last annual report:

 Berkshire has a round 400 million Coke shares at a cost base of $1299 million. The dividend is $1.40 per share - or $560 million per year.

That is a 43 percent yield in dividends on his cost base.

If you wish to ignore the dividends (as my commentators do) may you please give them to me.


General disclaimer

The content contained in this blog represents the opinions of Mr. Hempton. You should assume Mr. Hempton and his affiliates have positions in the securities discussed in this blog, and such beneficial ownership can create a conflict of interest regarding the objectivity of this blog. Statements in the blog are not guarantees of future performance and are subject to certain risks, uncertainties and other factors. Certain information in this blog concerning economic trends and performance is based on or derived from information provided by third-party sources. Mr. Hempton does not guarantee the accuracy of such information and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based. Such information may change after it is posted and Mr. Hempton is not obligated to, and may not, update it. The commentary in this blog in no way constitutes a solicitation of business, an offer of a security or a solicitation to purchase a security, or investment advice. In fact, it should not be relied upon in making investment decisions, ever. It is intended solely for the entertainment of the reader, and the author. In particular this blog is not directed for investment purposes at US Persons.