Estimating the effect of time-invariant characteristics in panel data: wheat adoption in Western Canada

Abstract

A growing global population and the challenges of climate change have made the need to develop new and improved wheat varieties increasingly important. Creating varieties that are more disease resistant and tolerant to changing environmental conditions such that they will be widely adopted requires breeders to understand the needs of producers. Existing literature suggests that time-invariant traits such as disease resistance and brand matter in crop adoption decisions; however, studies using common panel data approaches are unable to identify the individual effects of time-invariant variables, as they are captured by fixed effects dummies. This study uses a recently developed econometric approach—the fixed effects filter model—that can estimate the effects of time-variant, slowly changing, and time-invariant traits. Using wheat variety adoption in the Canadian Prairies as our empirical setting, we find that both time-variant traits, such as varietal adaptability, and time-invariant traits, such as resistance to stripe rust infection, are positively correlated with adoption. We also find that seed brand has a statistically significant effect on adoption. Prior panel data studies on crop adoption have not given much consideration to time-invariant variety characteristics like disease tolerance and seed brand, but a better understanding of their impact on adoption can assist breeders in responding to emerging problems, particularly new diseases, more quickly and more efficiently. This will also enhance the work by crop research institutions, prevent considerable economic loss to farmers, and improve crop production.