Value insights: Markets as complex adaptive systems

‘Value insights’ presents the Ninety One Value team’s distinctive perspective on global markets.

A favourite sport among active equity investors is making fun of the efficient market hypothesis (EMH). This is a sport we like: not only does the EMH make implausible assumptions about humans – essentially that we are always rational (‘utility-maximising’) and that we are all the same (‘homogeneous preferences’); also, its main claim – that markets are efficient – is easily rebutted. Historical examples of market inefficiency abound. Even just in the last three years we have witnessed such an extreme mania in meme stocks, SaaS (subscription-as-a-service), renewables and SPACs (special-purpose acquisition companies) that we feel the argument against EMH scarcely needs to be stated. Be that as it may, one inconvenient prediction of the EMH – that beating the market is impossible – remains uncomfortably close to the truth, which goes a long way to spoiling all the fun we were having taking pot-shots at it.

Reconciling the paradox of EMH being wrong but one of its predictions being apparently correct, is categorically above our pay-grade. But Michael Mauboussin, a US investment strategist, author and academic, attempted it back in 2005 and we are yet to find a more convincing explanation. Mauboussin posited that the market is a complex adaptive system (http://www.e-m-h.org/Maub02.pdf): that is, it behaves much like a swarm of bees or an ant colony. Its individual participants follow their own incentives (non-homogeneous ones) which are not necessarily rational (non-utility-maximising), and participants learn, which makes the system adaptive as opposed to static. Their collective behaviour creates a system that possesses ‘self-organized criticality’: it has a mind of its own and it swings from calm to instability at ‘critical points’, moments when sudden large changes are produced by the accumulation of small individual stimuli. These critical points are endogenous and largely unpredictable, and therefore seeking specific causes for big-scale effects is mostly futile. Not that it stops people trying.

A complex adaptive system exhibits the following properties:

• Aggregation: it is more complex than each of its component parts (Adam Smith’s ‘invisible hand’)
• Adaptive decision rules: its participants learn, explaining the short-lived existence of anomalies, as well as the presence of…
• Nonlinearity: cause and effect may not be simplistically linked, but may interact to produce exaggerated outcomes
• Feedback loops: the output of one iteration becomes an input in the next, leading to amplification (positive feedback, exploited in momentum investing) or dampening (negative feedback)

These properties explain the existence of fat tails, which EMH doesn’t allow for, and the (almost) random-walk nature of stock prices. They also allow for heterogeneous participant expectations, as opposed to homogeneous ones. The problem, for the more mathematically inclined, is that these properties make it hard to ‘do the maths’ on investments, and they make a mockery of most of the things our industry seems to value: a rigid, repeatable investment process, strictly quantitative risk management (whose calculations, incidentally, are derived from the modern portfolio theory that underpins EMH) and large, hierarchical investment teams. One might be tempted to ask: are these products of institutional orthodoxy useful in a system barely more tractable than organised chaos?

Besides, accepting the unpredictability of markets makes way for a certain mental clarity and some helpful general principles. Firstly, risk and reward are not necessarily linked (in contrast to the tight link they exhibit in the EMH). It should be possible to find – with some work! - superior investments no riskier than average, a possibility banished under EMH because it looks a like free lunch. Secondly, any individual actor’s view, be it a sell-side analyst, expert, TV pundit or portfolio manager, should be given low weight. However, the market’s overall message should be considered carefully. Thirdly, adaptive systems are most efficient at setting prices, when they benefit from a diversity of views and different decision rules. Under these conditions, participant errors will tend to be independent and feedback loops will be negative, with a dampening effect. Conversely, such systems will be least efficient when individual participant errors become correlated to each other. In other words, things become mispriced when narratives capture the collective imagination, participants trade for non-fundamental reasons, and investors mimic each other. Lastly, because of all the randomness, investors need a reliable anchor of value. We think the only viable option here is some form of discounted cashflow (DCF): only careful DCF analysis can give investors reasonable confidence in the long-term value of what they are buying.

Seeking a competitive advantage in a complex adaptive system is wholly different to seeking one in an efficient market. Whereas EMH says that there is no free lunch, and therefore no point in looking for one, a complex adaptive system allows for free lunches but makes you work exceedingly hard to find them. It will also probably beat you up on the way to the restaurant, mostly for fun. You therefore need to structure your affairs in a way that is resilient to unpredictability, which means you need to be flexible. Our team seeks to achieve this by following these guidelines:

1. Be humble. Despite the occasional tendency towards exaggeration, a complex adaptive system usually gets it right.
2. Stay curious. Encourage creativity to surface by keeping egos and summary judgments out of the conversation. If anomalies are short-lived, you must remain open-minded about where the next one might arise. Our team’s only guiding principle is to look for mispriced securities, and that’s about as much high-level guidance as is necessary.
3. Look for opinion extremes. Valuation matters, and mispricings are more likely to arise where there is a breakdown in thought diversity, or where the decision rules of market participants have become correlated.
4. Understand what you’re buying. Focus on businesses that have relevance in a changing world and whose intrinsic value you can estimate – via a conservative DCF – with reasonable confidence. Always do your own work and do not rely on recycled information.
5. Prepare for the unexpected. Do not try to predict the future, but think in terms of probabilities. Be prepared for extreme swings to become more extreme.
6. Ensure you can see it to the end. Keep the team structure flat, information free-flowing and barriers to dissent low. This helps raise the quality of debate, as well as the team’s collective sense of ownership, and ensures both compatible emotional tolerances and appropriate investment horizons for the strategy. Find clients who understand what they have invested in and can deal with short-term unpredictability. Work for a firm that is aligned with you for the long term.

It may well be that none of this is translatable into mathematics (although if it were, we wouldn’t be the ones to discover it), which is why modern portfolio theory still forms the backbone of the industry’s risk management systems: its maths are an approximation, but they’re the only approximation available. Despite the lack of opportunities to calculate, we think the theory of complex adaptive systems is a much better lens to understand markets through, and within this context we think the principles above set us up with three durable competitive advantages: a) behavioural (points 1-3); b) analytical (points 4-5); and c) structural (point 6). This combination is hard to put together and even harder to preserve. But then again, investing is hard, and we think it would be unreasonable to expect institutional orthodoxy to lead to success.