Do investors holding an asset in the midst of a price bubble all come to share similar beliefs, even if they did not do so previously?  Asset price bubbles are unfailingly supported by a captivating narrative, which is ‘validated’ by price movements, draws in investors and allows for traditional valuation considerations to be discarded.  It is therefore plausible to assume that an asset price bubble involves the acceptance of new norms or a shift in regime, by those that choose to participate.

This notion is consistent with certain theories on collective behaviour – where a crowd or herd act in concert it is because they have developed shared ideas, which drive their actions. However, work by sociologist Mark Granovetter takes a different perspective; contending that concentrating on the motivations or attitudes of individuals was insufficient to describe group behaviour.  His focus, instead, was on the participant’s ‘threshold’ – using the example of a riot, Granovetter describes the concept as such:

“A person’s threshold for joining a riot is…the proportion of the group he would have to see join before he would do so” (1978, 1422)

There are three crucial insights provided by Granovetter in his work on collective behaviour:

– A group does not have to share (or come to share) values and preferences for them to engage in similar behaviour.

– The behaviour of others within a group has a material impact on the choices made by other individuals in that collective.

– The distribution of ‘thresholds’ within a group matters greatly for its overall behaviour.

In Granovetter’s simple model, individuals are faced with a binary decision – to join the riot or not – and their threshold is dictated by their assessment of the benefits and costs of action or inaction.   A ‘radical’ will have a low threshold, whereas a ‘conservative’ will have a high threshold.  In essence, Granovetter is assuming that an individual’s threshold is dictated by their attempt to maximise utility.  Whilst we may question the underlying assumption of rationality, the core tenets of Granovetter’s approach remain powerful.

Although not directly related, Granovetter’s insights have profound implications for how the composition of a financial market (the activity and preferences of its participants) impacts the development of asset price bubbles.  It allows us to move away from the notion that a bubble involves the sweeping acceptance of a new narrative / norms, or an alignment of exalted expectations, and instead consider the importance of the threshold for participation across different types of investor.

We can apply the threshold model of collective behaviour to the incidence asset price bubbles.  As with the riot example, individuals are faced with a binary decision – whether or not to purchase the asset /security in question – a choice over which their threshold for participation will have a material influence.  By creating heavily stylised and simplified groupings of market participants we can speculate as to how this threshold varies:

Investor Type	Threshold for Participation
Speculative	Very Low
Growth	Low
Momentum	Low / Moderate
Market Cap Passive (Soft Momentum)	Moderate
Value	High

A speculative investor is akin to the ‘radical’ in a riot, they are content to be initiators often participating in the absence of others.  They are likely to have latched onto (or helped create) a beguiling narrative, and perceive the potential benefits (vast monetary profits) to outweigh the costs of failure. By contrast the value investor adopts the role of ‘conservative’; their threshold for involvement is high as, by definition, owning an asset priced at a level significantly above intrinsic or ‘fair’ value, would threaten their reputation, process and beliefs.

The threshold for pure momentum strategies is ingrained, with their participation in an asset class / security directly driven by the behaviour of other participants.  The situation for market cap passive is somewhat different; whilst the approach can be considered a form of soft momentum there is no tipping point for its involvement, rather it will serve to reflect and potentially exacerbate the aggregate behaviour of other participants.

Working within this framework, the composition of a market is crucial in understanding the incidence of asset price bubbles.  For example, in a market with a significant proportion of speculative and momentum investors (with low thresholds for participation), the frequency of bubbles is likely to be higher and momentum can rapidly build.  Contrastingly, in a market with a preponderance of value (or at least valuation sensitive) investors, price bubbles should be less common and their development protracted, as their threshold for involvement is far higher.

Of course, many valuation sensitive investors do eventually succumb to an asset price bubble, when the weight of participation becomes overwhelming and their threshold is reached.  Using Granovetter’s theory, this will occur when the costs of not participating outweigh the benefits – as an asset price bubble grows, so do the costs to the value sensitive investor – underperformance / career risk / stigma / regret. The impact of such factors will depend on the individual and their environment, thus thresholds will differ markedly.

Asset price bubbles are an incredibly complex phenomenon with narratives, behavioural biases and information dissemination all vital components in their formation and persistence.  Using Granovetter’s threshold model of collective behaviour provides us not with overarching theory of price bubbles, but offers some crucial insights, most notably:

–  The development of asset price bubbles is heavily dependent on who participates in a particular market / security.

–  Asset price bubbles are about how individuals react to the behaviour of others, as much as the acceptance of the underpinning narrative.

– Market participants will have different thresholds for participation in an asset price bubble.

The most fertile ground for a price bubble is where an asset has heavy speculative interest (very low threshold) and is impossible to value by any traditional means;  this combination of features can lead to dramatic and (for a time) unchecked positive price momentum.  Examples of such situations spring readily to mind.

—

Key Reading:

Granovetter, M. (1978). Threshold models of collective behavior. American Journal of Sociology, 83(6), 1420-1443

Granovetter, M., & Soong, R. (1983). Threshold models of diffusion and collective behavior. Journal of Mathematical Sociology, 9(3), 165-179.