Agricultural Producer Price Expectations

Producer price expectations underlie much of agricultural supply analysis. While producer price expectations would ideally be discovered experimentally, this is too costly. Instead, producer price expectations are usually represented in agricultural supply analysis by easily obtained hypothesized expectation formulations. In most cases, the hypothesized expectation formulations are functions of past prices. However, other formulations are sometimes used, such as current cash and futures prices, or initial payments in the case of grains marketed by the Canadian Wheat Board. This paper compares actual producer price expectations with a variety of hypothesized expectation formulations for wheat and canola in Saskatchewan. A test developed by Granger is used to determine the proxy models that are significantly dominant. The model that dominates as a proxy in the case of wheat price expectations is the two-year declining-weight moving average. The two models that dominate as a proxy in the case of canola price expectations are the first-order autoregressive and, as well, the two-year declining-weight moving average. There is no significant difference between the two models. Somewhat surprising is the performance of formulations based on futures prices. These formulations perform very poorly in representing producers' price expectations, even though they are found to be among the most accurate predictors of actual commodity prices. An even more interesting observation is the performance of the futures price model in the canola market. Even though the November contract in January explains very little of the variation in the actual commodity prices for that year, its error in predicting canola prices is not significantly greater than that of the best performing, the four-year declining-weight moving average, based upon the root mean squared error criterion.