Capital Returns is an excellent collection of investor letters and essays by Marathon Asset Management, edited by Edward Chancellor and originally intended to be distributed within the firm.

Although it looks like many of Marathon’s calls were prescient, there is definitely some selection bias involved in the picking of the essays (this is acknowledged by Chancellor).

Because I am a huge fan of the essays, I will be posting a large summary in three parts on the blog. We will finish with a more condensed ‘main takeaways’.


Capital Returns by Edward Chancellor

What is capital cycle analysis

Capital is attracted to high-return sectors of the economy and leaves when returns become dismal. This process is never static, always dynamic. From the economy’s point of view, this process is Schumpeterian, flushing the misallocated capital.

Capital cycle analysis is about how competitive advantages change over time from the investor’s perspective.

Professor Ken French writes that firm asset growth is a larger determinant of consecutive five year returns than value, momentum and size (Fama, French 2014).

Larger firm investment is followed by lower returns. This mean reversion is not only due to animal spirits, companies that earn more than their cost of capital expand their capital base, hence increase supply in their industry, while those that fail behave in an opposite fashion.

Corporate investment rates is also a negative predictor for corporate profitability in national aggregates. The US has had very low rates of reinvestment after the mal-investment of pre 2007 was revealed, and is subsequently earning very high profit margins.

Why does French’s ‘asset growth anomaly’ exist?

  • Overconfidence : Managerial and investor fetishism for asset growth. Future levels of demand are very hard to forecast.
  • Competition neglect: Corporate investment that looks like it will earn above its cost of capital looks rational when looking at it from the isolated point of view. However, competitors are probably making the same decisions at the same moment. The feedback lag because of investment lead times makes this problem worse.
    • TC comment: this is discussed in Buffett’s shareholder letters on Berkshire’s textile operations as well. Textile capex for efficiency gains looks great in isolation, but every competitor is investing in the same efficiency gains. I’ll call this “the red queen effect”

Now, here, you see, it takes all the running you can do, to keep in the same place. – The Red Queen in Alice in Wonderland (TC’s analogy to Berkshire’s textile capex in reality versus )

  • Insight view and neglect of the outside view: Management focusing too much on the specifics of their investment story versus the wider base rate of success given the competitive backdrop. This leads to overemphasis on “my case is different”.
  • Extrapolation: Linear forecasts are made because of anchoring bias, anchoring on recent data because of recency bias. This myopic element creates linear extrapolations ignoring that reality is cyclical. For example, trade cycle, credit cycle, liquidity cycle, real estate cycle, profit cycle, commodity cycle, industry capital cycle. Linear extrapolation is hard-wired deep inside us. Low multiple value stocks with depressed earnings exploit this tendency to extrapolate the near past. To discriminate between low multiple depressed and inflated earnings companies means investors have to dig deeper in the capital cycle.
  • Skewed incentives
    • Executive pay is often linked to company size regardless of profitability (ROIC) so even reinvesting below cost of capital can perversely be attractive
    • Companies with large investment projects tend to exhibit short term momentum on the same time frame that executives are rewarded
    • Investment managers’ pay is also often short-term based
    • Investment banks need to cheer
      • as corporate investment is often associated with fees for raising capital for another department in the bank
      • the brokerage fees don’t help not cheering for stocks
  • Prisoner’s dilemma: Maybe future demand can accommodate one player from expanding operations. If everybody refrains from expanding for the good of the industry but one player, the non-cooperative gets the gains. Retaliations to protect market share often follow.
  • Limits to arbitrage: Shorting can be expensive in terms of risk budget (nobody knows when the cycle will turn exactly). Also, companies that invest in new assets typically have higher multiples hence market caps so not investing in them makes investors deviate much from the benchmark.


Fundamentals of analyzing the cycle

  • Focus on supply: Supply prospects are much less uncertain. Look at how capex-to-depreciation is evolving over time.
  • Beware the investment banker (analyst): Investment banks are being paid for capital expansion projects. Hence they are paid to drive the capital cycle.
  • Selecting the right corporate managers

“After 10 years on the (CEO) job, a CEO that retains 10% of earnings will typically have been responsible for 60% of the capital invested in the business”. – Warren Buffett

  • Generalists make better capital cycle analysts: Industry insiders are prone to take the inside view, as opposed to Kahneman’s Outside view, not seeing the wood for the trees. Generalists adopt the outside view and do capital cycle comparisons between industries. There’s also an element of job protection at work.
  • Adopt a long-term approach: Short-term timing the cycle is impossible. Make general calls and sit them out.

The Story of Cod

For two centuries, the pinch-point between abundance of cod on sea and scarcity on land were the small ports that processed cod. When the industry scaled to transatlantic trade, that pinch-point became the large market of Boston. A triangular route emerged where ships transported New England cod to Europe, subsequently took African slaves to the Carribean sugar plantations, and finally brought Carribean liquor to Boston.

Boston flourished until technology catched up:

  1. steam-powered ships
  2. freezing food (1920s)
  3. sonar to locate fish

Overfishing became commonplace as the new owners had to bear high fixed costs to buy these new technologies. After ports, markets and processors of cod, and consumers, became main beneficiaries of the trade.

The questions for investors is which company will become the equivalent of the Boston fishing market?

The commodity super-cycle

  • On oil (Feb 2012): OPEC countries have high spending budgets, creating a high “break-even oil price” for them. This will limit their quota discipline if prices go down
  • On buying and selling Vestas (March 2014): Marathon took a position in the wind turbine company Vestas when one of the key indicators of the capital cycle was close to bottoming out (i.e. around 1.2X capex / depreciation for a leader in a sector that has long secular tailwinds). They sold most of their stake in 2009 when Vestas had outperformed MSCI Europe by a total factor of x4 and the capex-to-depreciation multiple was at a stellar number of 3-5X. They added a lot to their residual position in 2013 after the stock got slammed and capex-to-depreciation was depressed at 0.5x.
  • Why EPS growth has lagged GDP growth (Sept 2014): EPS growth has lagged GDP growth despite general corporate profits growing from 1960 from 6% to 10% of GDP.


  • Buybacks and dilutive equity raises at the exact wrong timing
  • Management share comp.
  • M&A, IPO is procyclical
  • Private companies becoming more profitable relatively because of less agency problems

If supply outpaces demand growth, profits suffer. Hence it is possible that sectors with high demand growth bring little or no benefit to investors as capital floods in (e.g. airlines, semiconductors, mobile phones). This is also true on the country level, with negative correlations of subsequent equity returns and country GDP growth. Public equity returns in China are a perfect example.

Rather, investors should focus on supply. The starting point of company analysis should be prospective supply, not demand.

Figuur 1 Value in depressed and growth industries

Value in growth

Warning labels (Sept 2002)

High moat “quality” companies that revert slower to mediocre ROIC’s than the market implies do so because competing capital inflow is somehow restricted.

Quality small companies are easier to find in Europe because there’s an excess of large institutional investors. Some sectors are overrepresented in this universe, e.g. industrials, while sectors like pharma are overrepresented (typically >10bn companies).

A “growth” label company typically gets a “value” label after excess capital flows into its sector.

Long game (March 2003)

Long term investing works because there is less competition for really valuable bits of information.

Low turnover reduces friction costs and fewer decisions leading to hopefully fewer mistakes. The real advantage to this approach, in our opinion, comes from asking more valuable questions. The short-term investor asks questions in the hope of clues to near-term outcomes. The value of this short-term oriented data is modest at best, and to build a long term significant track record, this investor needs to pull off this trick many thousands of times to exploit marginal opportunities sustainably. The competition is ferocious.

The info Marathon seeks is long term oriented, typically capital allocation skills.

An example is Colgate’s line extension advertising investment in the ‘80s. It created a huge amount of value eventually but the focus at the time was short-term earnings drag. Apparently even today no investor questions have been raised to Colgate’s management about its large marketing spend, except for Marathon’s.

Double agents (June 2004)

In his University of California talk, Munger asked students when demand goes up if you raise prices. Students responded with certain luxury goods.

What Munger was waiting for in fact were goods sold through friendly middlemen. Geberit (also covered in Quality Investing book, extensively uses plumbers in their push – pull strategy: plumbers pull their product to where it was pushed, i.e. wholesalers. Geberit invests a lot in free plumber training so that it becomes more time efficient for them to choose Geberit. Also, Geberit invests a lot in innovation and quality of the product, as the plumber gets a commission as a percentage of sales and hence does not care about the price (indeed, a plumber rather sells a high priced product). Rather, the plumber thinks about factors such as hassle-free installation, durability and quality to protect his own reputation. By investing heavily in training for plumbers, innovation, and quality, Geberit is able to raise prices disproportionately.

The agent’s interest is typically quality, safety, reliability, availability. Cost to the customer is ideally not a concern for the agent (to the contrary, if he earns a personal commission).

Other examples: dentists, doctors, opticians, electricians, architects.

Digital moats (August 2007)

Marathon has been a long term investor in Priceline and Amazon, and stresses how investors back in the 2003-2006 days were stressing short term margins instead of focusing on execution of the long term stated plans of the founders.

Quality time (August 2011)

Commentators are focusing on mean reversion of elevated aggregate corporate profitability. Marathon stresses that it is more productive for bottom-up investors not investing in indices to focus on specific sectors and firm capital cycles (or lack thereof, in case of sustainable competitive advantages).

TC Comment: See John Hussman’s newsletters as an example for aggregate profitability views.

Escaping the semis’ cycle (Feb 2013)

The Philadelphia semiconductor index has underperformed the Nasdaq since ’94 inception by 200 basis points p.a. with greater volatility.

Some firms, such as Analog Devices, have performed well over the long term. Analog semiconductors bridge the gap between the real world and the digital world, e.g. sensors in airbags, microphones in mobile phones, temperature sensors etc. According to Marathon, sustainable competitive advantages exist in this niche of the semi sectors, because capturing what is happening in the real world is harder.

The human capital component is hard to replicate because engineering talent deepens with experience. The design process is more trial and error. To become an expert in analog devices requires many years and the tenure of the average engineer at Analog devices is 20 years. Lastly, each analog company’s process technology is quite distinct. [..] Thus, it is difficult for an engineer to be poached by another analog company without his productivity being significantly impaired. Lastly supply of new engineers is more constrained for the analog niche as new graduates are more likely to pursue the digital semi route.

This is all compounded by the fact that analog devices have more diverse end markets. While the overall market is relatively fragmented, market segments is much more consolidated.

Pricing power is aided by the fact that analog chips play a very important role but represent only a small portion of the total cost of devices. Moreover, once an analog chip (e.g. sensor) has been designed specifically for a product, switching costs are high as the whole production process has to be revised. This makes revenues more recurring as well over the product life-cycle.

Finally, the analog production process is less standardized than most hardware, and thus less vulnerable to obsolescence from Moore’s law. More than a third of Analog Devices come from products which are more than ten years old.

To be continued,