The effect of climate change is expected to have a significant impact on production of agriculture products. This has led to renewed interest in estimating agricultural production that includes climate factors as well as the more traditional approach including economic and agronomic factors.

This study estimates grain (composite of wheat, barley, oats and rye) production, yield and area response to economic, rotational and climate factors for 15 cropping districts in Saskatchewan from 1976-2017. The theoretical model is a multi-product supply function. Prices included in the model include grains and oil seeds price indexes, a pesticide price index and fertilizer price index. Climate factors studied include monthly growing season (May-September) precipitation and average temperature for weather stations in each of the cropping districts. Agronomic factors include spring moisture, and two-year lags of grains, oil seeds and summer fallow area reflecting a three year crop rotation common in the area. All other variables are from the Statistics Canada database CANSIM.

The econometric model generalizes the fixed effects least squares dummy variable panel model by including contemporaneous correlations among errors of cropping districts. We do this because our data have a large time series dimension (41 years) compared to a cross-sectional dimension (15 cropping districts). Using the production identity production=yield*area, we estimate production yield and area as a simultaneous system rather than as individual components. We estimate a double log (Cobb-Douglas) form of the model supply function. This is more efficient than individual estimation commonly undertaken because it exploits full information. Climate variables capture monthly growing season first and second moment effects, so that mean and variance and co-variance effects of temperature and precipitation are captured. Monthly contemporaneous and dynamic interaction effects are included. This results in 45 parameter estimates capturing climate effects on grain production in Canada.

We compute marginal monthly temperature and precipitation elasticities evaluated at the monthly geometric mean of the data. Our results support the notion that the effect of climate on production, yield and area of grains in Saskatchewan depends on highly complicated growing season monthly impacts of mean temperature and monthly precipitation. No general predictions concerning the outcome of climate change effects can be determined without more detailed information.

Precipitation had a larger impact on production than temperature, indicating climate change forecasts that include precipitation and temperature changes that are critical to understanding how climate change could impact grain production in Saskatchewan.