impact of HACCP on costs and product exit, The

Journal of Agricultural and Applied Economics, Apr 2002 by Hooker, Neal H, Nayga, Rodolfo M Jr, Siebert, John W

Models and Hypotheses to be Tested

The data are evaluated with three pairs of econometric models, as discussed below. Each pair of models is constructed using alternative measures of plant complexity (RANGE or MIX) and scale (EMP or LNSALES-the natural log of SALES transformed due to scale differences between the variables).

Models 1 and 2

Costs are expected to be positively related to process complexity (RANGE or MIX). Similarly, facility modification variables (either adding new buildings-ADDFAC, expanding or redesigning current facilities-MODFAC) are hypothesized to increase the cost of compliance. As the number of HACCP trained employees (HAEMP, NEWEMP) rises, the implementation cost is also expected to rise. The indicators of plant scale (EMP and LN-- SALES) are expected to have negative signs indicating economies of scale in implementing HACCP. Older plants (AGE10) are presumed to require additional or more complicated redesign leading to higher costs. The next two variables are interesting and difficult to sign. Ex ante it may be expected that very small plants and those that choose to become custom exempt forecast higher HACCP compliance costs (recall the omitted dummy variable category is small plants which were required to be HACCP compliant by 1999). Therefore, the parameters on the dummy variables for very small plants and custom exempt plants (H2000 and CUSTX respectively) may be expected to take a positive sign. However, no known previous study has tested this. Also it is unclear if, once plant size is controlled for (with the scale measures), any additional impact should remain. Finally, it is not immediately clear which sign is expected for the parameter associated with the customer inspection variable (CUSTIN). Prior customer requirements for tight process control (as indicated by the federal/state inspection requirement) may subsequently lead to lower implementation costs (i.e., "good" firms are selected/signaled by meeting this qualification and thus require limited changes to attain full compliance).

Models 3 and 4

These models were estimated using a probit regression due to 0-1 dummy dependent variable. Generally the hypothesized relationships discussed above for the implementation cost equation (models 1 and 2) are expected to have similar effects on the probability that we see partial exit. However, two of the employee variables (number of people trained and number of new hires) may differ. If a plant chooses to formally train a larger number of workers in HACCP or hire extra workers it may then be able to support the previous range of products, being better able to manage a larger number of HACCP plans. It is not clear if there should be economies of scale effects on the probability of product withdrawal. While it is presumed that smaller plants proportionately have more complex processes and more numerous products, our current scope variables (RANGE and MIX) may be sufficient to account for these differences. Hence the lack of clear theoretical signs on the scale variables (EMP and LNSALES).