Accounting research has usually approached studies on cost stickiness through models with costs and output variations in logarithms of relative changes. But this approach is unable to detect cost stickiness properly in sectors characterized by small business units with scarce influence and chances to depart from sector trends, such as agriculture. The well-established economic tradition of research on price stickiness express changes in prices as differences in prices of a given period with respect to previous periods. This study fi nds empirical evidence, with a sample of farms, that models expressing changes in costs and outputs as differences are more precise in detecting stickiness than those expressing variables as logarithms in relative terms. The latter only detected it for the biggest farms, while the former did it for the whole spectrum of farm sizes. Results are also robust after controlling for changes in transactions, time, type of farming and location.