Yearly Grain Output Predictions in China 1980-2004

China has a population of 1.3 billion and grain accordingly plays a crucial role in the Chinese economy. In this paper we suggest predicting grain output mainly by factor inputs and asset holding, and present a Systematic Integrated Prediction Approach (SIPA). The key elements of SIPA are an extended input-output model with assets, nonlinear variable coefficient forecasting equations, and using the minimum sum of the absolute values. Since 1980 we have used the approach to predict the yearly national grain output of China. The prediction lead time is more than half a year. The bumper, average, and poor harvests are accurately predicted every year. The average error rate over the period 1980-2004 is 1.9%.