Murphy by xiaoyounan


									                                  Modeling Cessation Lag:

           How quickly does risk fall once regulation is implemented?

                                 Tammy Barlow MURPHY, Ph.D.
             2004-2005 AAAS Science and Technology Policy Environmental Fellow, U.S. EPA
                Assistant Professor, Dept. of Economics, University of Massachusetts-Boston


ABSTRACT: This paper examines the implications of cessation lag for estimating the benefits
of regulation that reduces the risk of exposure to contaminants with adverse health effects.
Several approaches to modeling cessation lag and the implications for the present discounted
value of benefits are addressed.
    Recently the U.S. Office of Management and Budget (OMB) and the U.S. Environmental

Protection Agency’s Science Advisory Board (SAB) have recommended that EPA include a

model of cessation lag in its economic analysis of regulatory actions that reduce the risk of

exposure to contaminants with adverse health effects. The term “cessation lag” refers to the

amount of time it takes, once exposure to a risk is eliminated or reduced, for the risk level of

individuals that were previously exposed at a given level to decrease to the risk level associated

with a lifetime of exposure to the lower level. The cessation lag is not necessarily equal to

latency, which is the time period between the start of exposure and an increase in risk. For

example, the latency period for smoking and increased lung cancer risk is about 20 years, but

cessation lag is considerable shorter.

    The issue of the appropriate cessation lag model has direct implications for the time profile

of benefits resulting from a policy that lowers the population’s exposure to a particular

environmental risk, thereby affecting the present discounted value (PDV) of benefits from the

proposed policy. All else held constant, the longer the cessation lag is, the lower the PDV of the

benefits of a reduction in exposure to a contaminant. Thus, the manner in which cessation lag is

modeled may determine whether or not a proposed regulation passes the benefit-cost analysis


    The existing research on cessation lag modeling has overwhelmingly focused on the

relationship between smoking and cancer risk. EPA first began wrestling with cessation lag in

the context of the Clean Air Act (1999) and the Arsenic Rule (2001, 2003), using studies from

smoking cessation as the foundation of their analysis, even though smoking is not the

contaminant targeted by either of these two regulatory actions. Despite further debate in 2004 in
the context of reduction in exposure to particulate matter through CAA regulations, it remains

unclear how to model cessation lag in the absence of data on the specific disease processes

involved in many of EPA’s current proposed regulations (e.g. the relationship between

chlorination dis-infection by-products (DBPs) and bladder cancer risk in EPA’s proposed Stage 2

DBP Rule fro drinking water). A valid cessation lag model must recognize that for many

contaminants, risk reduction will have an effect on acute, medium-term and chronic health

conditions. The cessation lag modeling efforts of Chen and Gibb (2003) are commendable;

however, their approach requires considerable data (e.g. hazard functions). Such data is often

unavailable for environmental contaminants and their suspected adverse health effects. The

mode of action of a contaminant as a cancer promoter vs. cancer initiator can sometimes be used

to approximate cessation lag; however, for many contaminants (e.g. DBPs), we cannot be certain

whether the contaminants acts primarily as cancer promoters or initiators, or as both (e.g.

cigarette smoke).

   This paper will explore alternatives for modeling cessation lag in the face of considerable

uncertainty. EPA’s proposed Stage 2 Dis-infection By-Products Rule, which is slated to be final

in 2005, is used as a case study of a regulation where the maximum annual benefit possible is not

likely to be realized in the time period immediately following regulation implementation (i.e.,

where cessation lag is not expected to be zero). The paper also explores the possibility that

cessation lag may depend on exposure duration to the status quo level of the contaminant.

Sensitivity analysis of the regulation’s estimated economic benefits to cessation lag model choice

is conducted. The paper also offers recommendations for future data collection and empirical

research that will allow for improved estimation of cessation lag, and therefore, improved

estimation of the economic benefits of environmental health risk regulation.

To top