Economics of the Greenseeder Hand Planter, Discrete Choice Modeling, and On-Farm Field Experimentation

Corn yields in developing countries are lower than in developed countries partly due to planting methods that involve hand dropping of multiple seeds per hill. Researchers from Oklahoma State University (OSU) recently developed the Greenseeder Hand Planter (GHP) to replace such methods. The first essay determines economic breakeven levels of seed and labor savings, increases in corn yield, and reduced loss of N through reduced ammonia volatilization. Results suggest a $50 GHP used to plant 3 hectares per year would need to increase corn yields on average by about 1.12%, use 12.19% fewer seeds, or reduce labor man-days by 38.66% to equal expected net returns from traditional methods. In the second essay, I conduct Monte Carlo experiments to measure bias in the conditional logit (CL) and independent availability logit (IAL) when there is no choice set formation and when choice sets are stochastically formed. I also compare the performance of the two models using empirical data on paddlefish angler preferences collected in Oklahoma. Both the CL and IAL work well when their own assumptions hold, but not under the alternative’s assumptions. However, the IAL produces unbiased and less efficient parameter estimates when individuals actually choose from the full set of alternatives. Empirical results suggest the IAL is able to predict the attribute-cutoff. To avoid limitations from small-scale agronomic trials, there has been a movement toward large-scale, on-farm field trials but questions remain as how best to conduct them and when it is most profitable to quit them. The third essay addresses these questions by using a fully Bayesian decision-theoretic approach. Data are from Monte Carlo simulations assuming a corn-input stochastic plateau production function. Results suggest the best way to conduct such experiments is to allocate to each of the 10% of the plots, 0 lb. of N, half of N*, and 150% of N* under a 30-plot experimental design. Results further indicate that it optimal to quit such trials in year 2. Sensitivity analysis confirms the optimal quit period but suggests such experiments are most profitable by allocating unalike N levels to all of the 30% of experimental plots.