To minimize climate risks of agricultural price hikes during meteorological disasters, this study aims to demonstrate the feasibility of forecasting agricultural prices several months ahead and evaluates the reproducibility of annual price fluctuations. We use the crop yields forecasted by the crop model and apply the quasi-dynamic large-scale global computable general equilibrium (CGE) model with 88 countries/regions to evaluate as many countries as possible worldwide. From the simulation results, the model’s accuracy to trace actual crop price fluctuations, which was measured by the regional average correlation coefficient data during 1995-2015, ranged from 0.13 to 0.26 per crop, and almost 30% of targeted countries marked statistically significant traceability. Such accuracy was higher in the developed liberal countries. Regarding forecasting 3-6 months ahead of the harvest, in approximately 20% of the targeted countries, the CGE model can reproduce actual price fluctuations, and a 3-month extension of the forecast period reduces the reproducibility by 16.7% for the correlation coefficient on average of four crops. Thus, the reproducibility of the model was not high, but in countries with statistically significant reproducibility, forecasting price fluctuations several months ahead can be used to prepare for meteorological disasters.