"Private" Provision of Publicly Useful Information: An Empirical Analysis of Public Notification Rules for Safe Drinking Water Act

Providing quality information on health-threatening contamination incidences can improve consumer welfare, because the consumers can, and often do, take self-protective actions given the information provided. Unfortunately, important contamination information often comes from private or quasi-private sources and the distinction between private and public information is also sometimes ambiguous. Thus, the regulatory authority may need to manage 'private' incentives of providing publicly useful information. There is no lack of literature arguing that supply of public goods is smaller than optimal when provision of those goods are in the hands of private agents seeking their own economic rents. The same argument seems to apply here --- private provision of publicly useful information is less or later than optimal. Though the argument makes an intuitive sense, there appears to be lack of empirical research investigating this issue. We provide one such study, examining a unique EPA dataset on the public notification enforcement and compliance for the Safe Drinking Water Act. The data contain 2,062 observations of maximum contaminant level violation incidences in Illinois from 1980 to 2006. The dataset allows us to estimate the determinants of the observed public notification timings. More importantly, we also estimate the impact of the new public notification rule on the coefficient for the contamination severity level. The new rule, issued May 2000 and fully enforced May 2002, (i) extended the PN compliance deadline (for tier 2 violations) of 14 days to 30 days and (ii) gives the primacy agency discretion to extend the PN deadline in some appropriate circumstances. Our results are disconcerting and indicate that the new rule has had welfare-decreasing effects. With linear models, one point increase in the severity level of a contamination incidence is significantly associated with an increase in PN delay of 8-9 days after the new rule but is not statistically significantly associated with PN delay before the new rule. Moreover, the new rule per se increases the PN timing by 21-24 days, which is approximately the simple effect of the extension rule. These results were robust to other specifications, in particular, to the Tobit specifications to account for censoring issues. Our empirical results seem to suggest that even in the case of mandated public notification, private or quasi-private managers of the public water systems still make private calculation of the benefits and costs of making information public --- the PN timing is significantly associated with various covariates, which one should not observe if the managers were truly welfare-maximizing agents. These findings should be taken seriously in lieu of guiding future information disclosure/public notification policies.