A Discrete Choice Mean Variance (EV) Cost Model to Measure Impact of Household Risk from Drinking Water Pipe Corrosion

In traditional investment decision making, one tool commonly used is the mean variance model, also known as an expected-value variance (EV) model, which evaluates the anticipated payout of different assets with respect to uncertainty where portfolios with higher risk demand higher expected returns from an individual. This thesis adapts this framework to a cost setting where decision makers are evaluating alternative physical assets that carry lifetime cost uncertainty for maintenance. Specifically, this paper examines homeowner choices for their home plumbing systems in the event of a pinhole leak, a tiny pin-sized hole that forms in copper, drinking-water pipes. These leaks can cause substantial damage and cost homeowners thousands of dollars in repairs. Since pinhole leaks are not related to the age of pipe material, a homeowner is subject to the risk of additional costs if a pinhole leak occurs again despite their repair efforts.

The EV cost model in this paper defines two discrete choices for the homeowner in the event of a leak; to apply a simple repair at lower cost and higher future cost uncertainty, or to replace their plumbing with new pipe material, usually made of plastic, at a higher upfront cost but lower likelihood of future expenses. The risk preference of homeowners are demonstrated by their repair strategy selection, as well as the level of cost they incur to reduce uncertainty. Risk neutral individuals will select the repair strategy with the lowest lifetime expected cost and high variance, while risk averse homeowners will prefer to replace their plumbing with higher cost but lower variance. Risk averse individuals are also exposed to indirect costs, which is an additional unobserved cost in the form of a risk premium the homeowner is willing to pay to remove all uncertainty of future pinhole leak expense.

Expected costs and variances are also higher for regions in the U.S. that experience elevated leak incident rates, known as hotspots. Using this mean variance cost framework, indirect cost can be quantified for homeowners in hotspot regions and compared to the rest of the U.S. to evaluate the magnitude of pinhole leak risk. The EV cost model estimates risk premiums on pinhole leaks to be $442 for homeowners in hotspots and $305 for those in the rest of the U.S. Finally, this paper examines the impact of pinhole leak cost uncertainty on the U.S. economy. Of an estimated $692 million in annual pinhole leak costs to homeowners, this study estimates a lower bound cost of $54 million per year (7.8% of estimated national annual cost) in risk premium that homeowners would be willing to pay to avoid pinhole leak cost uncertainty.

Information in this study on the role of risk in home plumbing decisions and indirect costs would be helpful to policymakers and water utility managers as they deal with infrastructure management decisions. Furthermore, the EV cost methodology established in this paper demonstrates an effective use of mean variance modeling under cost uncertainty.