Drinking water contamination and avoidance behavior: do bottled water sales adequately measure the cost of this behavior?

ABSTRACT

This paper examines the impact of anticipated duration for drinking water contamination on avoidance behavior. Differences in anticipated duration are based upon the type of public information provided to water customers: short-term boil water advisories (BWAs) versus long-term public notifications based upon Safe Drinking Water Act (SWDA) health-based violations. Avoidance behavior is measured by weekly bottled water sales at the county level in Kentucky. Our results confirm that BWAs have a much larger impact on bottled water sales than SWDA notifications. Bottled water sales increase by an estimated 3.8% in response to a 10-percentage point increase in population affected by BWAs during the week. This change in bottled water sales, however, represents a fraction of total avoidance behavior costs. Thus, bottled water purchases should be considered a very lower bound estimate for avoidance behavior cost estimates from contamination episodes of drinking water.

KEYWORDS:

• Public notifications
• Boil Water Advisory
• bottled water sales
• avoidance behavior

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. For more information see https://www.americashealthrankings.org/explore/annual/measure/water_violation/state/ALL

2. Twenty-four, mostly rural, counties in Kentucky were dropped because data for bottled sales were either unavailable for the whole study period or there were no stores located in the county covered by the Nielsen Company Scanner Data.

3. Allaire et al. (2019) report a coefficient estimate of 0.141 for a proportional variable measuring the population exposed to contamination. Their variable ranged in values between 0 and 1.0. Since the Violations T1% Affected variable represents a percentage of population measured between 0 and 100, our coefficient estimate is 100 times smaller, but represents same magnitude impact based upon the log-level formula for interpretation of regression coefficient impacts. Zivin, Neidell, and Schlenker (2011) report a 22% increase in bottled water sales when 100% of the population is exposed to a water contamination violation, which is about one-third lower impact than our results.

4. Pape and Seo (2015) report an increase in annual average expenditures on bottled water per household exposed to water quality violations as $20.52. This estimate is adjusted for 52 weeks and an average household size of 3.15 in 2008.

Additional information

Funding

This work was supported by the Deanship of Scientific Research, Vice Presidency for Graduate Studies and Scientific Research, King Faisal University, Saudi Arabia [Project No. GRANT251; Deanship of Scientific Research, King Faisal UniversityDeanship of Scientific Research, King Faisal University [GRANT251];