Reassessing Probability and Uncertainty in Economics: Insights from Kay and King

Reevaluating Probability in Economic Models

In "Radical Uncertainty," John Kay and Mervyn King engage with complex probabilistic concepts to critique the reliability of traditional economic forecasting. The authors introduce the notion of 'Radical Uncertainty,' attributing it to Knightian Uncertainty, distinguished by Frank Knight. Knight differentiated between risk (known probabilities) and uncertainty (unknown outcomes), proposing a nuanced understanding of our predictive limitations.

The book emphasizes that conventional models often fail because they don’t account for unforeseen events — a theme underscored by the catastrophic banking failures of 2008. Kay and King argue that probability calculations typically hinge on the assumption that all relevant scenarios are known, an assumption that frequently leads to errors. Their reflections resonate particularly given recent economic turbulence which highlights the insufficiency of traditional forecasting methods.

The Limits of Probability Models

One critical point raised is the nature of probability itself. The authors seem to advocate that true probabilities exist in a statistical model, which I find somewhat contentious. Each probability calculation operates within a specified model, meaning that while the models can be flawed, the resulting probabilities are not "meaningless." They simply reflect the limitations of our understanding of complex systems.

An interesting dimension emerges when considering uncertainty that could be labeled as Bayesian. If one thoroughly understands the set of possible outcomes (X) but struggles to assign accurate likelihoods (µ), this creates a different realm of uncertainty. This distinction highlights the flexibility required in developing predictive models while acknowledging that misjudgments can occur at various stages of the evaluation process.

Errors in Interpretation

Kays and King occasionally misinterpret probability applications in notable historical contexts. For instance, they critique Nate Silver's calculation of the probability surrounding the 9/11 attacks as devoid of value. Such a statement overlooks the conditional basis underlying probability, where any model can yield meaningful outcomes based on the stipulations of its construction. The strength of a model rests upon its context and the validity of the assumptions made.

Moreover, when discussing Keynes' Principle of Indifference, the authors appear to misstep by suggesting inconsistent truths arise simply from the probabilistic framework utilized. Knowledge must adapt and evolve; thus an update to models must acknowledge that new information changes previously held calculations.

Bayesian Statistics and Probability Models

While Kay and King make a compelling case regarding the limitations of traditional models, their familiarity with Bayesian approaches is somewhat unclear. Significant developments in Bayesian statistics incorporate probabilities of probative phenomena — effectively creating a hierarchy of understanding that accommodates complexities of the predictive landscape.

The authors propose a concept termed “reference narratives,” which is framed as an alternative to the rigidity of probabilistic modeling. This approach suggests that individuals select a perceived likely scenario and operate under the assumption that it will occur, albeit while remaining aware of potential deviations from that path. It is a surprisingly pragmatic method, allowing for flexibility in thought amidst uncertainty.

Verisimilitude Versus Probability

A fascinating proposition arises when considering the paper by Süskind, which distinguishes between probability and verisimilitude — the latter being more about the "realness" of outcomes than mere predictive accuracy. This perspective shifts the focus from how often predictions are correct to how well they reflect underlying truths about the world. For example, a model that inaccurately predicts swan colors still serves a purpose in its statistical frequency, which may overshadow its failing verisimilitude.

Economic Forecasting in the Face of Uncertainty

The book tackles the inherent challenges of economic forecasting under these principles of uncertainty, articulating that reliable economic forecasts cannot be achieved under conditions of radical uncertainty. The authors suggest that economics relies on coordination, requiring stakeholders to converge on a common understanding of future scenarios, even when the validity of those forecasts is deeply questionable.

Ultimately, this notion suggests that the function of economic predictions might be less about accuracy and more about providing a common narrative for stakeholders to engage with. It's a nod to humanity’s age-old reliance on forecasts, reminiscent of ancient divinatory practices, where belief and adherence to a narrative often trump factual accuracy.

Concluding Thoughts

The conversation initiated by Kay and King opens up avenues for further exploration, particularly in understanding how economic systems can adapt in the face of profound uncertainty. The reluctance to engage with realities of uncertainty leads to a cavalier attitude towards economic forecasting that must be unpicked. If we are to construct models that genuinely reflect the vast tapestry of possible outcomes, we must first acknowledge the underlying assumptions and boundaries within the frameworks we create.

Forward-looking Insight: By embracing uncertainty as a fixture rather than an aberration, we can cultivate more resilient economic strategies that not only predict potential futures but adapt dynamically to the realities that unfold.