Thoughts on Auto Industry

The automobile industry is facing potentially massive disruption with the advent of new technologies. Electric vehicles, autonomous driving and ride sharing have the potential to alter industry value chains that have been in place over the last several decades. Thus, many established automobile companies could see their profitability erode as new competitors capture more of the value created.

There are no moats in the auto industry as it currently exists. Thus technological changes are inherently bad to the industry, because companies have to spend money to keep up with the changes and any benefits get passed on to the consumers. Any advantage gained from being a first mover are easily copied over.

1.      Electric Vehicles

The key technology to develop is to be able to produce batteries both in capacity and at low cost. Only Tesla is investing in battery production at this time but it should be possible for other companies to catch up as it might only take 2-3 years to bring a new factory online. There does not seem to be much of a barrier that would give one company a lasting competitive advantage over others in this area.

A charging network is another area that needs to be built up to sell electric vehicles to the general public. Here again there does not seem to be much of a barrier as current incumbents can subsidize existing gas stations to provide such facilities. A payment of $250K per gas station seems to be a reasonable price to pay which means that a network of 20,000 stations can be built at about $5 billion.

One potential risk is if Tesla cars turn out be like the iPhone and comes to dominate the auto industry, capturing the vast majority of the profits. There are a few factors that would mitigate this risk:

  • It is likely that Tesla would not be able to meet the entire demand all by itself – in fact the company acknowledged this when it released its patents. It is more likely to take up a share of the more profitable premium/luxury segment. Thus there would be a place for other auto companies.
  • The majority of the profits of the three Detroit auto companies is from pickup trucks. The owners of pickup trucks seem to have a high level of brand loyalty. These owners are mainly in the sparsely populated interior of U.S. rather than on the densely populated coasts where Tesla has built up a brand following. Still even if Tesla does not capture a vast majority of the market, it can kill industry profitability in this segment if its product gains significant acceptance.
  • Looking at the history of the automobile industry, there had been several improvements in the vehicles themselves but it has never made a meaningful difference in terms of changes to market shares of the companies involved. These improvements have always been copied by the other companies and the company that introduced the innovation did not enjoy any sustained increase in profits or market share.

Assuming a very rapid adoption of electric vehicles, in the scenario where people en masse decide they want to drive only EV’s say by beginning of 2018. Then many companies would rush to build out Gigafactories and this likely would grow exponentially. Starting with say 4 factories in 2018, then 8 in 2019, 16 in 2020 and peaking at 32 in 2021. It is likely that there would be many constraints even to reach a capacity of 32 factories in an year. Assuming very aggressively that companies can build a factory within one year, there would be 60 factories in operation at the beginning of 2022 meeting the needs of about 30 million cars or about 30% of the worldwide production during that year. This is very unlikely, but that is the maximum possible adoption rate for EV’s.

It would also be wrong to assume iPhone like adoption rate in the automobile market. iPhone reached a market share of about 35% of unit sales in 2016 since its introduction in 2007. Tesla and the whole BEV had reached a market share of about 0.4% in 2016 after it was introduced in 2008. Electric vehicles need a whole new infrastructure to be built up to support them whereas iPhone is leveraging existing infrastructure. Thus EV’s adoption is more likely a gradual process rather than a sudden spike.

The threat of electric vehicles is likely to show up 5 to 10 years from now when the technology would have established itself with electric vehicles making up 2-5% of the total automobile fleet in United States. It is less of a threat in the next 5 years as there is no existing or planned production capacity for electric vehicles to be a significant portion of the annual sales until then.

The real risk with electric vehicles to the existing automobile companies lies in improvements to the production processes. Companies that have sustained competitive advantages in this industry have relied on improvements to production processes like Ford’s assembly line and Toyota’s Toyota Production System. Incumbents have vast experience with internal combustion engines (ICE) but electric vehicles are much simpler and there might be better ways to assemble them. Tesla might gain such advantage if it is able to come up with such improvements based on it being the first mover.

2.      Autonomous Driving

Many companies are investing in autonomous driving including Google. It is likely that all the companies would have the technology that is on par with each other as they cannot be deployed until it works nearly flawlessly. Alternately, optimal solution in this case would be for the best autonomous software to be widely adopted by all the market participants. It would be in the best interests of the industry that all vehicles have the best possible software. In any given road, an autonomous vehicle would be just as safe as the least safe vehicle in that particular area. Thus, even if one company had a superior autonomous driving technology, it would not be a big competitive advantage by itself if it is the only one using it. There are further advantages to using autonomous driving software from a single provider as it would be simpler and probably more safer to coordinate and manage interactions between vehicles.  The company whose technology was adopted would end up with a royalty stream that all companies would have to pay for using that software.

So most likely scenario is that one or two companies would have a dominant share of on autonomous driving software but that company would not be able to monopolize the profits as much as the market share would indicate. All the automobile companies would have to cooperate and this would be more akin to Visa/MasterCard when they are owned by the banks.

An annual subscription to a mapping and navigation system that is accurate and up to date would also be required. Here again pooling of resources makes sense as having as much data as possible would make the system more accurate. This situation again points to a solution similar to autonomous driving software with Visa/MasterCard type provider owned by the automobile companies.

It would take another 5 to 10 years for the technology to mature itself to a stage where level 4 automation gains market acceptance as the technology needs to be validated in actual usage and regulations adopted for them. Since any failure can be fatal to both occupants and others on the road, the key system components would need to meet high manufacturing, testing and maintenance standards that are closer to the aviation industry.

This technology might have significant impact on the relative position of the various companies. For example, the enthusiasm for fast acceleration is absent for anyone other than the driver – passengers tend to be more sensitive to acceleration than drivers. When there is no driver, there is little advantage to buying a sports car or any premium features related performance. Factors such as interior design might take on more importance but where there is little chance of any sort of competitive advantage.

The advantage then would reside with the low cost mass producer based on scale. Autonomous driving would commoditize automobiles even more than it is now and industry is likely to consolidate into a few global companies.

An additional risk to this technology is if a technology company that relies on other sources of revenue (search or advertising) and which can derive additional revenue from data generated from autonomous software, subsidizes its vehicles, it could destroy industry profitability for the legacy automobile companies.

Even Tesla would lose some of its allure to customers when level 4 autonomous driving becomes fully operational, assuming autonomous driving software itself would either get monopolized by another company or commoditized without much differentiation. People would care less what car they are riding when they are just a passenger and are engaged in some activity and not focused on driving. A few of the advantages of the electric vehicle are partially mitigated as well – vehicle maintenance or refueling is less of a concern if you can ask the car to go to a dealer or an gas station by itself.

3.      Ride Sharing

This is likely to have only a marginal impact as many people would still prefer to own a vehicle – because they keep their tools handy in their pickup trucks, or drive a lot for those who are using pickup trucks as their work vehicles, families with young children who have a lot of things to carry and prefer to leave them in their vehicle, etc. This is more likely to reduce buyers only in dense urban areas but some of that compensated by higher usage from people who now take public transportation.

The risk is more when autonomous driving and ride sharing both take off. Then utilization per vehicle can be increased very significantly on the order of 10x from 15,000 miles per year to 150,000 – 200,000 miles per year. Even if this doubles the number of passenger miles used, this has the potential to dramatically cut down on the number of automobiles needed each year perhaps to half the current levels.

Since this would require level 4 autonomous driving to be operational, it would take another 5-10 years before it starts impacting the auto companies.


A risk with all these new technologies is that it would raise consumer expectations of what is needed in an automobile. If Tesla comes up with new features, the other automobile manufacturers need to match them and this could consume most of the free cash flow currently generated by these companies. Or it could reduce the prices the companies charge on their products. Thus overall profitability of the industry would decrease. Similar to Amazon’s impact on retailers. This is all the more likely because the companies that are driving these technologies have different motivations – Tesla seem to be driven more by environmental concerns and less profit minded; Google probably plans to use the data generated in self driving vehicles to improve its search/advertising business;

These risks are however at least 5 years away and more likely closer to 10 years. The risks are far away enough that one need not be completely scared away from currently investing in automobile companies, but they are not that far off that the risks could be ignored. Thus position size in this industry should be limited to a small percentage of the portfolio.


At what rate can Fairfax compound?

Fairfax Financial is a property casualty company based in Canada. It is led by an owner operator, Prem Watsa, who started the business in 1985. It is one of my first investments and it had been a major contributor to my portfolio around the time of the financial crisis. I had sold it in 2011 to deploy cash into beaten down financials. Now that many of these financials have rallied, I have reduced my allocation to them and wanted to take a fresh look at Fairfax.

An insurance company should be evaluated based on their underwriting and investing skills. There are a few insurers that are good at underwriting but it is rare to find insurers that are good at investing. Fairfax is that rare combination of an insurer that seems to do an acceptable job at underwriting and absolutely amazing job on the investment side. They have demonstrated repeatedly an ability to protect shareholders from adverse market conditions and to profit from market dislocations.

Analyzing Historical Investment Returns

The results of Fairfax Financial are going to the determined in large part by what the company earns on its investments. Hence, a critical part of valuation is estimating the company’s expected returns on its investment. So we need to analyze the drivers of past returns to get a sense of what returns can be expected in future given current bond and stock market valuation levels.

Fairfax breaks out the 15 year returns of its stock and bond portfolios and compares them to benchmarks.

FFH Returns

The above clearly demonstrates the outstanding investment results generated by Fairfax. However, we need to look at the entire investment portfolio since a significant part of the portfolio is invested in cash and short term investments and there are changes to allocation between stocks and bonds. By looking at total portfolio we incorporate the effect of market timing moves made by Fairfax.

In the table below I have used the aggregate bond index and Russell 2000 index to analyze the returns generated by the total investment portfolio. These are more appropriate benchmarks given the type of investments Fairfax actually makes.

Fairfax Investment Returns

Note: Fairfax Financial 2014 annual report, shows the 29 year return as 8.9% which is a simple average of all the years returns. An investor with a multi-year investment horizon would actually get a geometric average return and is thus a more appropriate measure of performance. For example, an investor who makes 20% in one year and loses 20% in the second year, would have a simple average return of 0%, but geometric average annual return of -2% over the two year period.

Investment Returns by DecadeBenchmark Returns

In the first decade, there were very little excess returns, but it is understandable that a deep value investor would have a tough time keeping up in a roaring bull market with no major pullbacks. They did take advantage of the attractive valuations and invested heavily, to the tune of 45% of the portfolio into stocks.

In the second decade, Fairfax was very cautious, dramatically reducing the stock allocation to just 13% on average. Fairfax also took advantage of the extreme divergence in valuations of new economy and old economy stocks, with its stock portfolio vastly outperforming the indexes (187% vs. -21% for Russell 2000). Stock selection played a key role in generating excess of 2.54% during this period, with changes to allocation being a drag to their performance. Even after correctly identifying that the market is extremely overvalued, by trying to take advantage of that with changing allocation hurt their performance. The extent of their stock portfolio outperformance, hid the massive opportunity cost of this decision.

Over the last nine years, investment returns were lower than the first two decades but with higher excess returns. However, about half the excess returns came from CDS gains during the financial crisis. These could arguably be considered as one-time gains and excluding them, the excess returns are about 1.38%.

The use of stock market hedges in recent years seems to be in response to their experience in 1999 – 2003, where they outperformed the market by 210%, but did not get much benefit due to their allocation of only 9% to stocks. Currently, despite their macro concerns, they have a close to normal allocation but with protection from hedges.

Since performance is influenced by starting and end points, it would be helpful to use a slightly different timescale. If we look at the performance over the last 15 years (2000 to 2014), Fairfax investment return is 8.17% compared to the equivalent benchmark index performance of 6.66%. Thus Fairfax outperformed the benchmark by 1.51% over the last 15 years.

Over the full 29 year period, excess returns were about 1.69%, including CDS gains. The excess returns are 1.25% excluding CDS gains. The outperformance of the entire investment portfolio is thus much more modest than when looking at just the performance of their stock and bond portfolios.

Key takeaways from the analysis of Fairfax investment returns:

  • Like most good value investors, they generate the most excess returns when markets are very volatile.
  • Portfolio size does not seem to be causing any drag on their investment performance.
  • Their stock selection has been the primary driver of performance while allocation changes based on macro have been a net overall drag.
  • Returns are going to be heavily influenced by yields given their propensity to hold nearly a quarter of the portfolio in cash and at least half of the portfolio in bonds under almost all conditions.

Expected Investment Returns

It seems reasonable to expect about 1.5% to 2.5% of excess returns over the relevant benchmark portfolio in line with what has been achieved in the past by Fairfax.

If cash was to return 1%, bonds 3% and stocks 5% over the next decade, the benchmark portfolio would return about 2.94%. With excess returns of 1.5% to 2.5%, total investment return for Fairfax portfolio would thus range from 4.5% to 5.5%.

Expected returns of 4.5% to 5.5% seem very low compared to Fairfax historical track record of 8.7%, but it just reflects the current investment climate of low yields. We can also approach this from a completely different angle. Fairfax had an average investment yield of about 4.8% over its 29 year history. Currently it is less than 1.6% (see Note 1 below). So the dividend yield is 3.2% lower than what Fairfax was getting in the past. In addition, rising valuations which caused lower yields, accounted for about 1% of the return (see Note 2 below) in the bond portfolio. These two factors accounting for about 4.2% of Fairfax portfolio return are not going to reoccur going forward. If Fairfax investment team performs just like it had in the past with the only difference being lower yields, then the investment returns should be about 4.5% (8.7% historical return – 4.2% from lower yield and capital return from falling yields).

Note 1: Fairfax portfolio yields 2.6% but actual yield is only about 1.6% because its hedges via total return swaps require it to pay dividends on its short book which would reduce the total yield on its portfolio. If hedges were closed, the drag from lower yields would drop by 1% to 3.2% and Fairfax investment portfolio would have expected return of 5.5%.

Note 2: We can calculate this by separating out the capital return and income return on the benchmark portfolio each year to estimate the contribution over the full period.

Thus, a return of 4.5% to 5.5% is a reasonable expectation from Fairfax investment portfolio over the next several years. This assumes cash returns 1% and bonds returns about 3%, but they currently yield only 0% and 2.5% respectively. Using current rates, an investment return of 4% is more probable in the next couple of years.

Normalized Earning Power

At an expected investment return of 4% to 5.5%, we can estimate the likely return on equity (ROE) over the next several years or until there is a significant change in market valuations.

Using a baseline investment return of 5% and breakeven underwriting, we can estimate the normalized earnings power as below. Fairfax has shown improved underwriting in the last few years and acquisitions have been focused on companies with good underwriting results. Thus, it might be reasonable to expect Fairfax to breakeven in underwriting, even though it historically lost money.

Normalized Earnings

Using a few plausible combinations of investment and underwriting results we can come up with a range of earning estimates. Since Fairfax has a small dividend and reinvests most of its earnings, book value growth is going to closely approximate ROE.

  1. Optimistic Scenario (Investment Return 5%; Combined Ratio 100%): ROE of about 9%.
  2. Conservative Scenario (Investment Return 4%; Combined Ratio 102%): ROE of about 5%.
  3. Baseline Scenario (Investment Return 4%; Combined Ratio 100%): ROE of about 7%.
  4. Highly Optimistic Scenario (Investment Return 5.5%; Combined Ratio 97%): ROE of about 12%.

Thus Fairfax is likely to compound book value anywhere from 5% to 12% over the long term with the lower end of the range being more likely.

There are lots of scenarios that could cause book value growth to deviate significantly from above, but this provides a baseline from which to look for deviations.

Especially, management target of 15% book value growth requires the deflation thesis to come through or a stock market correction of similar magnitude to the financial crisis to come to pass.


| Posted in Valuation | No Comments »

Is your investing style closer to Ben Graham or Warren Buffett?

Some of the mistakes I made in investing can be traced back to taking a Ben Graham approach to a Warren Buffett type of business. Ben Graham approach is primarily based on mean reversion in business, not just in stock prices, but in the fortunes of the businesses. Warren Buffett on the other hand, specialized in identifying the tiny sliver of businesses that defy mean reversion.

Knowing if an investment falls closer to the Graham approach or the Buffett approach would go a long way in determining a number of key portfolio considerations like position sizing, holding period, selling price and required margin of safety. Satisfactory results can be obtained with either approach, but what does not work is applying a Buffett approach to a Graham type business and vice versa.

Graham writes in Security Analysis that “the most striking characteristic of large American businesses is their repeated pendulum swings from better to poorer results and back” and that “a remarkably small percentage of these enterprises actually go out of business through voluntary liquidation or sheriff’s auction“. These two points are really the foundation on which a vast majority of Graham’s investment strategies are based on. He confesses this “to be of greatest importance in the determination of a sophisticated investment policy“.

Graham basically wants to take advantage of the phenomenon of mean reversion in businesses. So all he needs is some way to make sure that he is buying businesses below their mean value and wait for the businesses to recover. Graham never really tried to estimate the intrinsic value of a business but instead he just wants to make sure that he is paying a price far below what can be readily justified. His famous “Net Nets” strategy is the most dramatic illustration of this approach.

Graham felt that it would be difficult for investors to determine either the quality of management or the future growth rate of earnings. To him management quality should already be reflected in current earnings and to give credit separately to management would be double counting. Graham scarred from The Great Depression, thought it was not prudent to rely on earnings growth to justify the value of a business.

Buffett, having successfully applied many of Graham’s strategies, however pushed the value investing frontier to include high quality businesses rather than just broken down nags that are Graham’s bread and butter.

Buffett believed that there exists a very small subset of businesses, which due to strong competitive advantages are likely to earn high rates of return on capital and grow their earnings at an attractive rate for very long periods of time. Management quality, which to Buffett essentially means good capital allocation skills, can be accessed from past behavior.

Graham’s approach is primarily quantitative. He does not have a strong opinion on the intrinsic value of the business, management, returns on capital or growth. He spreads his bets on a large number of holdings to reduce the risk of any one individual business failure. As he relied on mean reversion, he was actively selling shares when they rose closer to intrinsic value or if he found something that is a better value.

Buffett’s approach on the other hand combines both quantitative and qualitative factors. As he is trying to find the rare business that has strong competitive advantages that are unlikely to deteriorate for decades, his search would generate only a handful of ideas over a lifetime. This approach calls for an intimate understanding of the business along with a firm conviction on the intrinsic value which allows for taking concentrated positions. As the businesses grow intrinsic value at an attractive rate, these can be held even if their price reaches or even slightly exceeds intrinsic value and thus turnover is very little.

Most investors who have studied them are well aware of these differences between the two. However, as Buffett evolved from primarily being a Graham type investor into his own style, his writings and actions have evolved as well in line with this change. To followers of Buffett, who are seeking to learn from him, his writings and his investments have to be viewed in context of the investment approach that he was following.

My mistake in this area is buying a high quality company that passed all of Buffett’s criteria and selling it at around 90% of my estimate of its intrinsic value. My thinking was dominated by Graham’s mean reversion warnings and suggestion to sell at close to intrinsic value. It was an expensive mistake.

Table: Summary of Ben Graham’s and Warren Buffett’s approach to investing

Ben Graham

Warren Buffett

Search Strategy Periodic screening on quantitative criteria Keep a target list of businesses
Number of Holdings Diversified (30+) Concentrated (5 to 6)
Holding Period 1 to 3 years or close to intrinsic value Very long term (20+ years) & Do not sell even if it reaches intrinsic value
Confidence in Intrinsic Value Estimate Low Firm conviction (even though it is a range of values)
Buying Price Two thirds of intrinsic value or less Slight discount to intrinsic value or  buy close to intrinsic value if there are additional kickers like untapped pricing power or buybacks
Source of Margin of Safety Discount to intrinsic value based on past earnings or current asset value Growth in intrinsic value based on quality of business and competitive advantages
Importance of Management Ignored Look at passion for business & capital allocation skill

Most investments do not come neatly packaged under Graham or Buffett categories, but knowing where in the spectrum they fall under, provides a useful set of portfolio tactics that can be applied for that situation.