Bayesian updating for binary variables: an application in the UK water industry

Abstract

Summary. In accordance with the regulation of the privatized English and Welsh water industry, water companies are required to submit an asset management plan every few years. A requirement of South West Water's asset management planning is to state the probability that some of its assets will need refurbishment or replacement within the following 5 years. Probability assessments are elicited for the company's assets based on information of varying quality. For the formulation and updating of beliefs we propose and employ the threshold model, a Bayesian updating procedure. The model's main aspect is Jeffrey's conditionalization. This is an updating rule based on a simple conditional independence assumption. According to this rule, it is not necessary to construct a fully specified joint density for the quantities of interest and the probability assessments.