Journal on Policy & Complex Systems Vol. 2, Issue 2, Fall 2015 | Page 38

Assessing Values-based Sourcing Strategies in Regional Food Supply Networks : An Agent-based Approach
market channels , such as farmers ’ markets ) may vary over time , according to how well the producer ’ s products have sold through the food hub in previous distribution cycles . This update defines the producer ’ s degree of participation with the food hub and depends on the producer ’ s upper threshold for unsold inventory ratios . The unsold inventory ratio is simply the amount of inventory ( in food units ) that a producer has left at the end of a cycle , divided by the total number of units that he / she offered to consumers through the food hub at the beginning of the cycle . If this ratio is equal to zero ( i . e ., he / she sold the entire available inventory ), then in the next cycle , the producer will increase his / her offerings by 10 % ( up to its capacity ). If this ratio is greater than the producer ’ s upper threshold for unsold inventory , he / she will calculate a weighted average of the ratio of number of items sold to capacity , over the three most recent cycles and will change the percentage of capacity that he / she offers through the food hub in the next time step to that average value . If the ratio is greater than zero but less than the upper threshold value , the number of units that the producer offers through the food hub in the next time step will remain unchanged . It is assumed that if a producer ’ s participation drops to less than 10 % of his / her capacity at any point in time , that producer will no longer participate with the food hub for the duration of the simulation run ( i . e ., its status will become “ canceled member ”).
Food hub membership update . At the end of each distribution cycle , new producer and consumer agents are generated , and they are randomly assigned parameter values based on probabilities that were determined from the empirical food hub data . It is assumed that new consumer agents are created at a constant rate of two consumers per cycle . A new producer agent is created in every other cycle , on average . These rates approximate the actual rates at which new producer and consumer members joined the real-life food hub , based on food hub historical data .
Producer Selection Policies : Experimental Method

To assess the value of implementing various different producer selection

policies , the food hub manager agent may choose to intervene during the
“ Food Hub Membership Update ” submodel execution . To execute a given selection policy , when a producer agent is created and attempts to join the food hub , the manager will determine the producer ’ s attributes , assess how well these attributes fit the needs of the food hub and its consumers , and based on this assessment , decide whether or not to allow the producer to join the food hub .
Five different producer selection policies were tested to assess the impact of having the food hub manager intervene in the selection of producers :
1 . No centralized management of supplier selection : This policy represents the status quo — any producers who wish to join the food hub are allowed to join .
35