Simulating Heterogeneous Farmer Behaviors
Table 3 . Mixed-effects multinomial logit model to assign agent behavioral types
Table 4 . Deviations from target production levels
In this case , we run a random-effects OLS model for each policy and information segment of data with standard errors clustered at the individual level .
For adoption decisions , we calculate the probability that an agent deviates from its target adoption decision , which means the probability that an agent changes its adoption decision away from the theoretical prediction . The metric is defined as the absolute difference of the actual adoption decision and the target adoption decision :
Since the variable AdoptChange is binary , we run a random effects logit model for each policy and information treatment segments with individual clustered standard errors . The result of the model is shown below :
Based on the above regressions , we parameterize the agent ’ s actual production
178