Fat tails and climate change: the case for a new approach to major infrastructure appraisal. Issue 3 (2nd September 2019)

• Record Type: Journal Article
• Authors: Hurst, Martin
• Is Part Of: Journal of mega infrastructure & sustainable development. Volume 1:Issue 3(2019)
• Page Start: 218
• Page End: 236
• Publication Date: 2019-09-02
• Subjects: Normal distribution -- fat tailed distribution -- net present value -- climate change -- net zero -- adaptive planning; Infrastructure (Economics) -- Planning -- Periodicals; Public works -- Planning -- Periodicals; Public utilities -- Planning -- Periodicals; 363.60684
• DOI: Https://www.tandfonline.com/doi/10.1080/24724718.2021.1980272
• Languages: English
• ISSNs: 2472-4718

Abstract

Abstract: Conventional project appraisal techniques such as net present value and P50/90 have a strong assumption of symmetric 'bell curve' probability distributions and rely heavily on 'comparative static' approaches assuming past trends can be extrapolated into the future. For long term mega infrastructure, or portfolios of, projects, these are probably not acceptable assumptions. The case for abandoning normal distribution approaches is strengthened by the presence of 'disruptors' (e.g., COVID-19) and tipping points/non-linearities. Among these, climate change is typified by highly asymmetric probability distributions with great uncertainty about the probability of, for example, higher temperature change and sea level rise. Furthermore, the consequences of the events at the top end of the probability distribution are not only very high but also extremely uncertain in themselves. Recent UK guidance accepts that conventional approaches to project appraisal may be sub-optimal here. Moves to reduce carbon emissions impose an extra uncertainty on many infrastructure projects in terms of technological development and the related pathways to net zero. So, distributions even if symmetric may display 'fat tails' with higher-than-Normal probability of extreme outcomes. Probability distributions will also be 'skewed': with a much higher than conventionally assumed probability of adverse consequences. But the more extreme, 'regression to the tail' concept needs further exploration before it can be used in practice. While there remains a place for NPV approaches, practitioners need to rebalance towards scenario analysis, adaptive pathways/more continuous assurance and non-monetised approaches. More work is also required to identify how and how far it is reasonable to adapt NPVs to skewed probability distributions.