Economic value of management information systems in pig farming

The research described in this thesis focused on developing and testing methods to determine the profitability of management information systems (MIS) in livestock farming Methods were first applied to evaluating MIS in pig farming. Economic value of MIS arises from the fact that farmers have limited time, motivation or skills to decide consistently. Therefore, positive research approaches that derive MIS benefits from actual decision making of farmers, such as survey studies and economics experiments, have more potential for MIS evaluation than normative research approaches that assess MIS benefit with predetermined decision criteria, such as decision tree analysis and simulation studies Firstly, a survey study was conducted to evaluate MIS in pig farming. To sort out the effect of "better management" from the actual benefits of MIS, a panel data set was created by combining data of the survey study with data collected on the same farms in 1983. Adjusted for farm, trend, and learning effects, production on farms adopting MIS increase by 0.56 piglets per sow per year (p=0.09). Tests for autocorrelation, influential observations, and nonequivalent control time-series indicated that this outcome is robust Secondly, the impact of farm characteristics on MIS profitability was investigate comparing two conceptually different farm classification methods within the same research population: the sociological "style of farming" approach and the farm-economic "management level" approach. Farmers with high management levels got more added value from MIS. Finally, an individual decision-making experiment was conducted to yield insight into whether laboratory economics experiments can be used as an alternative to surveys for determining the profitability of MIS in sow farming. Instead of linking MIS us to farm results directly, the effect of different information levels on decision making was investigated under controlled laboratory conditions. Many of the farmers in the experiment also participated in the above-mentioned survey study. Although an overall effect of MI was found in both the experiment and the survey study, experimental and survey MI estimates were not significantly correlated. Possible explanations for these uncorrelated estimates are discussed in the thesis.