Was it something I ate? Implementation of the FDA seafood HACCP program.

Increased incidence of (sometimes fatal) food-borne illness due to bacterial contamination has made food safety a widespread concern. These concerns motivated a new regulatory to food safety based on the Hazard Analysis and Critical Control Points (HACCP) concept. HACCP involves identifying points where the production process compromises food safety, developing systems to monitor food safety and corrective actions that address food safety failures, and maintaining records of these activities. This preventive approach is widely considered the most effective means of improving safety in food processing operations (Mayes and Mortimore 2001; Huleback and Schlosser 2002; Unnevehr and Jensen 1996).

This article examines the implementation of regulations mandating the use of HACCP program in the seafood processing industry introduced in 1998. Most studies of HACCP to date have focused on the U.S. Department of Agriculture's programs in meat and poultry processing. The seafood HACCP program is administered by the Food and Drug Administration (FDA) and is far more complex, covering a larger number of products and a broader range of safety hazards (including pathogens, toxins and physical contaminants). In contrast to the existing empirical literature, which concentrates on implementation costs (Caswell and Hooker 1996; Unnevehr and Jensen 1996; Roberts, Buzby, and Ollinger 1996; Antle 2000) or impacts on industry structure (Muth et al. 2002) assuming full or almost full compliance, we examine how enforcement monitoring translates into compliance with HACCP-based regulations.

Implementation of those regulations in seafood processing has apparently been problematic. Evaluations by the General Accounting Office (GAO 2001) and the FDA (2001) identified a number of shortcomings in implementation. Disturbingly, the population incidence of foodborne illness from pathogens associated with seafood consumption has been rising more rapidly than per capita seafood consumption, even as the incidence of foodborne illness from pathogens associated with the consumption of meat, dairy products, and eggs has declined. In 2004, the number of cases per 100,000 people of Vibrio illness was 47% higher than the 1996-1998 baseline, while per capita seafood consumption had increased by only 12% and the incidence of illnesses associated with Campylobacter, E. Coli, Listeria, and Salmonella had fallen relative to that same baseline (Centers for Disease Control 2005; National Marine Fisheries Service 2005).

This article develops a theoretical model of imperfect enforcement in a regulatory environment like that created by HACCP, where regulatory standards are idiosyncratic to each firm and where determination of the degree of compliance is subject to error. Using this model, we derive predictions about the FDA's optimal monitoring strategy and regulated firms' optimal compliance efforts, which we test using data from the first four years of the seafood HACCP program.

FDA's Seafood HACCP Program

Seafood consumption accounts for a disproportionately large share of foodborne illnesses in the United States, as in other OECD countries (GAO 2001; Cato 1998; Putnam and Allshouse 1999). The Federal Food, Drug, and Cosmetic Act makes the FDA responsible for regulating food safety in seafood processing. Prior to 1997, the FDA addressed that responsibility by conducting periodic inspections focusing on hygiene and sanitation at seafood processing plants. To improve the efficacy of its oversight and food safety in the seafood sector, in 1995 the FDA published a final rule requiring seafood processors to develop and implement HACCP systems for their operations, effective December 1997. The FDA incorporated enforcement of the new HACCP standards into its existing sanitation inspection regimen.

HACCP is based on well-established principles of food safety engineering (see Cato 1998; McEachern et al. 2001; Huleback and Schlosser 2002). The FDA's HACCP regulation requires seafood-processing firms to have a written plan that applies these principles to its operation and to keep records documenting adherence to that plan. Each firm's HACCP plan must be approved by the FDA, which conducts periodic follow-up inspections. FDA inspectors check the firm's paperwork, plant, equipment, and production process in the plan approval process and in subsequent enforcement inspections.

At the time of its introduction, HACCP was seen by all involved as a win-win proposition. It was believed that consumers would benefit from increased seafood safety; the seafood processing industry would benefit from increased demand due to increased safety that would outweigh the modest costs of compliance; and the FDA would benefit from an ability to utilize its scarce enforcement resources more effectively (FDA 2001). This belief suggests that the FDA viewed HACCP monitoring and its preexisting sanitation monitoring as complements rather than substitutes. Our empirical study examines the extent to which this belief is true.

Criticisms raised by the GAO (2001) and internally led the FDA to identify a number of measures to be taken to improve its implementation of the HACCP regulation, including increasing the frequency of inspections (from once every four years prior to the HACCP program to annually since its implementation), improving guidance and training for both industry and inspectors, and intensifying inspection efforts aimed at products that posed the greatest threats of foodborne illness (FDA 2002). The latter include (1) scombroid fish, which can produce toxic histamine compounds as a result of partial bacterial spoilage, (2) smoked fish, and (3) cooked ready-to-eat fish products, in which control of pathogens can be problematic. Our empirical analysis examines, among other things, the extent to which this targeting seems to have been implemented.

A Model of HACCP Enforcement and Compliance

The preceding discussion of HACCP-based regulation suggests that it has six distinctive features. First, HACCP is primarily a process standard and hence depends on firms' inputs (although monitoring of performance outcomes like bacterial counts is included in some HACCP plans). Second, the fact that each firm has its own approved HACCP plan indicates that the process standard is idiosyncratic to each firm. Third, during the enforcement phase, the standard with which the firm must comply is predetermined, so that the agency's sole choice involves the precision with which it monitors the firm's precautionary effort. Fourth, heavy reliance on inspections of paperwork mean that determination of compliance is subject to error. Fifth, penalties for noncompliance are set legislatively and are thus predetermined relative to enforcement strategy. Sixth, the FDA is required to monitor all seafood processing plants at some point in time.

In this section, we formulate a theoretical model of a government agency charged with enforcing HACCP regulations and a firm subject to that regulation. The firm's HACCP plan consists of a set of preventive activities designed to achieve a level of food safety the agency considers adequate, e.g., an acceptably low level of health damage due to consumption of the firm's products. Let a represent the firm's level of precautionary effort and h the level of effort specified in the firm's HACCP plan. If the firm is found to be out of compliance with this standard, it is subject to a penalty s. Both the HACCP standard h and the penalty s are set during the initial development of the HACCP plan independently from enforcement considerations and are, thus, predetermined for enforcement purposes. After the initial plan development and implementation, the firm must choose the level of precautionary effort, a, it wishes to exert. Let C(a) denote the (convex) cost of that effort.

Personnel from the regulatory agency periodically inspect the firm's facility and observe the firm's records and other evidence of its precautionary activity. We denote this observed level of precautionary effort as y. This observation is unbiased but subject to error:

(1) y = a + (1 - m) x [epsilon]

where e is a zero-mean error term and 0 < m < 1 represents the precision with which the agency inspects the firm. The agency is obliged to inspect the firm periodically, hence m > 0. If the agency were able to monitor processing at all times and places it occurs at the firm, it would be able to observe the firm's precautionary effort without error (m = 1). However, it cannot, hence m < 1 and thus the firm's true level of precautionary effort remains its private information. A higher value of m indicates greater precision in inspection, to be achieved through more frequent visits and/or lengthier and more thorough inspection during any given visit. Let K(m) denote the convex cost of monitoring.

The firm is found out of compliance and fined an amount s whenever observed precautionary effort falls short of that specified in the HACCP plan, y < [bar.a], or

(2) [epsilon] < [bar.a] - a/(1 - m).

Assume that the firm's observed effort is nonnegative, which implies [epsilon] [greater than or equal to] -a/(1 - m). Let [PHI](x) denote the cumulative probability distribution of the observation error, [epsilon]. Then the probability that the firm is found out of compliance is

(3) F(a, [bar.a], m) = [PHI]([bar.a] - a/1 - m) - [PHI] (-a/1 - m)/1 - [PHI](-a/1 - m).