Epidemiological and economic modelling of classical swine fever: application to the 1997/1998 Dutch epidemic

This paper describes a modelling system developed to simulate the epidemiological and economic effects of a classical swine fever (CSF) epidemic in the Netherlands. The system consists of four interlinked models plus a spreadsheet. The models are characterised by different levels of spatial and temporal aggregation. A spatial, dynamic, stochastic epidemiological model simulates the daily spread of the disease between individual pig farms and incorporates a variety of possible control measures. A second model aggregates daily production figures over farms and days to obtain weekly output of the sector, adjusting for output reductions due to depopulation of farms and quarantine restrictions. A sector-level, partial equilibrium, market and trade model simulates the weekly clearing of the Dutch pig market, allowing for market interactions and trade in piglets, slaughter pigs and carcasses. A fourth model calculates control costs arising from the various control measures adopted, and the losses of producers subject to these measures. A spreadsheet calculates the welfare changes of producers outside quarantine areas and other affected stakeholders, and aggregates them over the duration of the epidemic. The net welfare impact on the Dutch economy is also calculated. The system is illustrated by an ex post simulation of the 1997–1998 Dutch CSF epidemic. The distribution of welfare losses between pig producers and downstream stakeholders was shown to depend on the size of the epidemic and extent to which export demand for live animals would switch to carcass demand when live animal exports are stopped. Public expenditure on control measures and compensation was strongly related to the size of the epidemic, and increased more than proportionately with the duration of the epidemic and the number of detected farms. The modelling system can be used by policy makers for both ex post evaluation of epidemic control measures, and ex ante simulations of contingency plans. It could also contribute to identifying of a set of indicators that could enable policy makers to predict the likely size and cost of an epidemic while it is at a relatively early stage