Specialists play an important role in many fields. When it comes to investing, however, Tom Hosking believes generalists could have the advantage.
Reading time: 20 minutes
This article was featured in our Spring Journal which can be found here.
Over tens of thousands of years, we have increasingly become a society of specialists. Early hunter gatherer societies were likely characterised by generalists who were adept at a variety of tasks needed for survival. A key part of Homo sapiens’s subsequent success was its ability to form larger and more stable groups. The larger the group, the more scope for specialisation. Teamwork resulted in being able to hunt large prey such as woolly mammoths. But where flint spearheads could be made by an individual person, today’s guided missiles require the cooperation of thousands of specialists, each of whom would be unable to manufacture it on their own.
Adam Smith, writing in the middle of the Industrial Revolution, argued that the sole reason for economic growth was specialisation. Workers would be finely divided into roles enabling them to become more productive, and once the entire workforce was specialised economic growth would cease. Whilst Smith may have been wrong about it being the only determinant of growth, he was right to predict that society would become ever more specialised. Several people have claims to being the last person to know everything, most famously the polymath Thomas Young, but they all lived at least two centuries ago. As human knowledge advanced beyond the capability of even the most talented individual, our modern education system responded by encouraging students to focus on increasingly narrow specialties. However, there is a growing body of evidence that specialism is not always the best thing to do, as documented by David Epstein in his book, Range. Reading it motivated me to think about how this applies to fund management.
The field of investment is a good example of a subject where it may be better to be a generalist. Broadly speaking, there are two reasons for this. The first is rather obvious: specialist investors often miss the wood for the trees. Sector specialists can only assess how attractive an investment is relative to other investments in the same sector. For example, if you are a specialist banking analyst, you may be quite good at identifying which banks are good investments, but crucially only relative to other banks. It may be the case that no bank is a good investment, or equally that they all are. That can only be gleaned by looking at a range of sectors.
The second advantage is more subtle and far more interesting. A generalist’s analysis of a single investment alone may be more insightful than that of the specialist who knows everything there is to know about it, as counterintuitive as that may sound. At first glance, hyperspecializing is the most efficient way to develop a skill. However, in truth this is limited to narrowly constructed ‘kind learning environments’. In his research into the nature of expertise, Robin Hogarth characterises kind learning environments as fields where patterns repeat frequently and where feedback on performance is accurate and quick. As a result, learners improve simply by engaging in deliberate practice of the activity. Examples would include chess, golf or heart surgery.
However, investment is an example of a wicked learning environment. Patterns do not reliably repeat and are not obvious. Feedback on performance is often delayed and/or inaccurate. It may take years for an investment thesis to play out and for the patient investor to be proved right. This makes investment not just complicated but complex. John Kay contrasts finance to rocket science. Firing a rocket to Venus is incredibly complicated but it could be done again with exactly the same technique. The laws of planetary motion do not change and there is no reflexivity between what we think about Venus’s motion and how it actually behaves.
In contrast, an investment strategy that worked last year may not necessarily work this year. This is because, in investment, the underlying determinants of success are not completely known. They change over time, and they are affected by our beliefs about it. In this regard, the investor’s job is more complicated than the seemingly incredibly complicated job of firing a rocket into space.
When patterns don’t reliably repeat, experience alone doesn’t improve performance so developing effective habits of mind is more important. Hogarth argues that one must draw on outside experiences and analogies as a ‘circuit breaker’ to interrupt one’s leaning toward a previous solution that may not work anymore.
Creative thinking through analogies offers a number of advantages. The first is that it helps us to reason through a problem that we have never seen before. This is particularly important when we think about new technologies or business models. For example, estimating the future market size for a product or service that does not currently exist is a difficult problem. Creative thinking can help break down this puzzle into manageable chunks so that one can use what little one knows to solve what one doesn’t. What proportion of the population might buy it? How fast has the adoption rate been for tangentially similar products? The problem can be approached from many angles and the individual estimates don’t have to be accurate for the answer to be approximately of the right magnitude. We often ask these types of ‘Fermi problems’ in graduate interviews at Majedie to help us learn how a candidate thinks.
The second advantage to analogical thinking is that it helps us to look at a familiar problem in a new context. To succeed in investment, one needs non-consensus opinions and those non-consensus opinions need to be right some of the time (it’s actually possible to outperform the market while being wrong most of the time, if investments are scaled correctly). To form non-consensus opinions, it helps to be creative. And many studies have shown that the best ideas often come from combining insights from different fields. In a recent Global team meeting, when discussing the investment prospects for Equifax, a US credit ratings company, another member of the team suggested thinking about online retailers, such as Latin America’s MercadoLibre, who were using their data on customers to lend based on their own proprietary credit scores. Drawing on knowledge from a different sector and different geography improved our analysis of a particular investment opportunity. One can also draw analogies from history. In another recent Global team meeting, when discussing regulation of US big tech companies such as Facebook and Google, we drew from research on the antitrust cases of Standard Oil and other monopolies of the late nineteenth century.
The number and quality of analogies can be improved by working in teams of people with different academic backgrounds. Evidence from biology laboratories suggests that teams of scientists from a range of disciplines more readily solved problems than teams which consisted solely of a specific specialism (Dunbar et al. 1995). The researchers found that scientists from different backgrounds offered more analogies from distant domains which enabled them to solve more problems. This is one reason why it is helpful that the Investment team at Majedie is made up of people from a diverse variety of backgrounds and life experiences.
Studies like this may lead one to believe that the optimal team consists of specialists, each bringing narrow expertise to the table to help solve problems. However, it is teams of generalists that tend to perform better for number of reasons. Firstly, in the case of investment, it helps if a fund manager doesn’t feel compelled to own a stock in a certain sector solely due to the presence of a specialist analyst on the team. This aspect is increasingly important when managing concentrated, high conviction portfolios.
Secondly, specialist experts have a tendency to become overconfident in their predictions. Research from the CIA found that once the minimum information necessary to make an informed judgement has been obtained, additional information does not improve the accuracy of estimates, but it does lead one to become more confident in the judgement. This is a dangerous recipe for an investor: overconfidence can result in excessively rosy forecasts and bloated investment positions.
In a 2012 study, researchers asked private equity investors to forecast the return on investment (ROI) of the project they were working on (Lovallo et al. 2012). Afterwards they were asked about other investment projects ‘with a broad conceptual similarity to theirs’. On average, investors estimated the ROI of their own project to be c. 50% higher than the outside projects, leading participants to subsequently cut their initial prediction. The investors were falling victim to the ‘inside view’, where making judgements based narrowly on the details of a particular project leads one to form more extreme judgements.
Teams are better at overcoming the inside view than individuals. In his now famous work on ‘superforecasting’, Phillip Tetlock found that the best forecasters were also good collaborators. When he stuck them in teams, he found that they became 50% more accurate on average than in their individual predictions. This is because team members were able to share information and, in particular, information that was contrary to existing hypotheses. While it is people’s strong instinct not to look for reasons why they might be wrong, it is much easier for teammates to do so.
The final reason why teams of generalists are better than those of specialists is that being a generalist changes the way we think. It has been widely found that creative thinking is correlated with having a wide range of interests. Even in the hyper-specialised world of advanced science, researchers found that a scientist was 22x more likely to come up with a Nobel Prize-winning discovery if they had an artistic hobby such as painting or playing a musical instrument (Root-Bernstein et al. 2008). For the generalist, examining more sectors and recognising conceptual similarities between them helps the brain to create abstract models. We can then use those abstract models to think creatively about potential investments in new sectors.
Due to neuroplasticity, the structure of our brains is not constant. As certain neural pathways begin to be ‘exercised’ more regularly, the connections between different parts of the brain grow. For instance, some research suggests that learning a language can make you more creative. It is one of the reasons that I have been trying to learn German for the past year! Some research suggests that bilingual speakers are better at resolving conflicting information. Often different languages convey exactly the same piece of information using sentence structures that could barely be more different. So it is perhaps unsurprising that the ability of the bilingual brain to absorb an entirely different way of expressing and of understanding the world in a linguistic context should also apply to a more general context.
Not only is it important to be general in your knowledge but also in your philosophy. Different investment styles are appropriate for different macroeconomic environments. For example, it is well known that growth investing has outperformed value investing for the last decade, in stark contrast to the decade before that. This means that the best or most durable investors are flexible investors – those who are willing to change their approach to suit the existing climate. If you are less wedded to a particular investment philosophy, then you are more able to switch to another when facts change or your own thinking changes.
Research on aviation accidents suggests that ‘a common pattern was the crew’s decision to continue with their original plan … in the face of ambiguous or dynamically changing conditions’ (Orasanu & Martin, 1998). Similarly in investing, some of the worst returns occur when an investor sticks rigidly to an original investment thesis despite changing facts. Investment strategies should be situation-specific and not held onto at all costs.
Indeed, in a world where simple investment rules such as buying stocks with a low P/E ratio, high revenue growth or high return on capital are easily replicated by an algorithm in a smart beta ETF, why would an investor pay an active manager to follow such rules? The rapid growth of indices is changing what one should desire from an active fund manager. Increasingly what is valuable is the idiosyncratic return profile of a generalist.
Specialism in investment philosophy should be left to computers and complex analysis to humans; each to their respective strengths. Such a distinction has already been found in the world of modern chess where the optimal team was found to be a combination of computers and humans. In ‘advanced chess’ or ‘Centaur chess’, players are armed with a computer. The computer is able to analyse the optimal short-term moves given the situation on the board (tactics), leaving the human to focus on overarching strategy. Garry Kasparov, maybe the greatest human chess player ever, argues that ‘human creativity was even more paramount under these conditions, not less’. AI systems work well in narrowly defined worlds where relationships are stable. However, the bigger the picture, the more valuable the potential human contribution.
And that picture is growing rapidly. There is an accelerating divergence between the aggregate quantity of recorded knowledge and the limited human capacity to absorb it all. Perversely, this could result in a golden age for generalists due to the fall in communication costs that has accelerated in the last 30 years. Not so long ago, making connections between disparate pieces of research would have required access to a particularly well stocked and expensive to compile physical library. Today, one now has access to the accumulated knowledge of mankind in one’s pocket in the form of a smartphone! The fall in the cost of communicating information means that it is now much easier and more fruitful to be a generalist. And ironically the more information that specialists create, the more opportunities there will be for generalists to make new connections. At Majedie, we spend a material amount of our time, after reading research and companies’ results, thinking about applying what we’ve learned to other sectors and considering the wider implications on the portfolio. It is impossible today to be broad and deep in knowledge so we must begin by being broad and shallow, going deep only when required. We aim to synthesise information from different domains to help us form non-consensus opinions about the world, and invest wisely.
- The Bilingual Brain by Albert Costa (2020)
- Range by David Epstein (2019)
- Superforecasting by Dan Gardner and Philip Tetlock (2015)
- Sapiens by Yuval Noah Harari (2011)
- The Worldly Philosophers by Robert Heilbroner (1953)
- Thinking Fast and Slow by Daniel Kahneman (2011)
- Center for the Study of Intelligence, CIA (2005): ‘Do You Really Need More Information’
- Dunbar (1995): ‘How Scientists Really Reason’
- Kahneman & Klein (2009): ‘Conditions for Intuitive Expertise’
- Lovallo et al. (2012): ‘Robust Analogizing and the Outside View’
- Orasanu & Martin (1998): ‘Errors in Aviation Decision-Making’
- Root-Bernstein et al. (2008): ‘Arts Foster Scientific Success’
- Swanson (2001): ‘On the Fragmentation of Knowledge’