As noted by Jason Collins in his excellent behavioural economics blog, Daniel Kahneman’s next book is expected to focus on the concept of ‘noise’ and how it impacts our judgements. Although often conflated with behavioural biases, noise is a distinct phenomenon that relates to the random variability in our decision making. Whilst biases exhibit a consistency of effect (at least in direction, if not magnitude), noise is defined by the absence of consistency. A watch that loses time each day is biased; a watch that can either gain or lose time during any given day is noisy.
In an article in Harvard Business Review, Kahneman, alongside his collaborators, discussed how individual choice is “strongly influenced by irrelevant factors” and gave examples of how professionals are prone to contradict their own previous conclusions. The problem of attempting to grasp the idea of noise is that it is so amorphous – whilst we can at least develop a (limited) framework for defining and understanding biases, by definition noise is hard to isolate and anticipate. Noise can stem from entirely spurious factors – such as mood, weather or hunger – or variables that we perceive to be meaningful, but are in fact meaningless. It is certainly possible to test whether noise exists in any given scenario – by observing decision making consistency – but this is only the start of understanding the issue.
Noise has profound implications for investors, but is often ignored or, at least underappreciated. It can be difficult to accept that our judgements can be shaped by erroneous, often farcically minor, factors. Furthermore, we are often uncertain about the key variables that define any given problem.
In the realm of investment decision making, we can define two separate forms of noise:
- When given the same objective data and relevant variables we are unlikely to make the same decision. This is consistent with Kahneman et al.’s article – even if we hold the meaningful factors constant, other irrelevant issues will lead to inconsistent choices.
- We don’t know what the relevant information is and therefore make decisions based on what we perceive to be ‘signal’ but is in fact noise. This is such a major problem – one which the industry perpetuates – that it is difficult to know where to begin.
We could crudely define these as unconscious noise and conscious noise. In the first case there are many factors that impact our decision making over which we have no real awareness and we would be reticent to acknowledge had any influence over us. In the second case, the issue is uncertainty about what constitutes relevant information and what is superfluous noise – this will vary by context and discipline, but it is difficult to think of an industry with a greater ratio of noise to signal than asset management. Conscious noise is the oxygen on which the industry, in its current form, exists.
Second by second coverage of random market movements (with accompanying narratives), heroic forecasts (usually wrong), luck masquerading as skill, complex products and every decreasing time horizons are just a few of the factors that contribute to the maelstrom of noise that investors are forced to navigate. Of course, this is good for the industry – simplicity and inaction are not typically an aid for revenue generation – but it fosters a situation where decision consistency becomes close to impossible for most investors.
Kahneman et al. proceed to argue that a “radical solution” for the problem of noise is the replacement of human judgement with algorithms, or a structured set of decision rules. They also acknowledge, however, that such processes are less effective in environments where uncertainty is high or where consistency is difficult to attain.
Can algorithms be effective in muting the incessant noise in investment markets, and even exploit it, to improve decision making? To a certain degree. One effective and humble decision rule / heuristic, is portfolio rebalancing. A structured and consistent approach to rebalancing a portfolio back to target weights is proven to be effective and cancels out a great deal of market noise. It ensures both that your portfolio doesn’t stray too markedly from its desired allocation, and that you consistently sell assets that have become more expensive and reinvest in those that have become cheaper. Whilst this might seem a simple course of action, rebalancing into assets that have struggled (amidst the prevailing negative market narrative that will inevitably accompany the poor performance) can be difficult without a formal / systematic decision rule.
The oft-mooted remedy to the problem of noise and inconsistency in human-led investment decisions is the movement to full automation and the use of complex algorithms / machine learning. It deals, at least in part, with the aforementioned ‘unconscious noise’ angle as the feelings of the decision maker are no longer a direct issue, however, at some point human judgement will inevitably exert an influence – for example, in the decision to initially invest in a strategy or to redeem, thus it does not provide full immunity.
More importantly, algorithms do not necessarily resolve the issue over noise in regard to the use of irrelevant information. Many an ETF has been created based on factors with no empirical credibility; therefore although the decisions within the process can be dispassionate, the very existence of such strategies is reliant on the fact that there is noise in the market. Furthermore, even the most sophisticated systematic approaches are vulnerable to trading based on patterns than are simply a consequence of random market movements, with no structural, technical, economic or behavioural reason to exist or persist.
Even without full automation, there are means of dulling the noise in human-led investment decision making, such as checklists. Whilst checklists are particularly effective in areas such as aviation and surgery where many of the checks can be simple and objective – whether the correct leg is being operated on, for example – they can still improve discipline and focus around decisions with inherently more subjectivity. Formally reviewing a list of your key criteria prior to making an investment can serve to highlight noise driven deviations from your core process and also acts as a useful record for reviewing historic decisions. Of course, using checklists when answers are subjective means the potential for manipulation is ripe, so it pays to be as rigid as possible when defining questions; however, even if this is not possible, checklists remain a useful means of reaffirming your investment principles amidst the market noise.
Noise is an inescapable feature of human judgement. In random and uncertain investment markets its influence is profound. Although it is impossible to eradicate, acknowledging its presence and taking steps to simplify and systematise certain decisions can be an effective way of turning down the volume.
Kahneman, D., Rosenfield, A. M., Gandhi, L., & Blaser, T. (2016). Noise: How to overcome the high, hidden cost of inconsistent decision making. Harvard Business Review, 94(10), 38-46.
Please note all views expressed in this article are my own and are not necessarily shared by my employer.