Is the relationship between energy and agricultural commodities an important factor in the increasing price variability of food commodities? Findings from the literature appear to be mixed and highly influenced by the data frequency used in those analysis. A recurrent task in time series applied work is to match up data at different frequencies, while macroeconomic variables are often found at monthly or quarterly observations, financial variables are sampled daily or even at higher frequencies. In order to match up time series at different frequencies a common procedure is to aggregate the higher frequency to fit in the low frequency, this has the potential of losing valuable information, and generating misspecification. We study whether the use of mixed frequency estimations with data for the 2006-2011 period helps to improve the out of sample performance of a model that explains grain prices as a function of energy prices, macroeconomic variables such as exchange rate, interest rate, and inflation. Preliminary results suggest that an improvement is feasible, however it is tenuous beyond two months horizons.