A concern I have about extrapolating from historical trends is the sensitivity to the form of the function used to extrapolate. For example, if a linear function was used here, then a strictly convex function (e.g. quadratic, exponential) might give a higher estimate for the trend without corporate campaigns, and we’d have to adjust the counterfactual impact of these campaigns downward. It seems plausible that the percentage cage-free would increase at an increasing rate, given how relatively flat it was from 2005 to 2010, and the possibility of some kind of network effects (e.g. in concern for or knowledge of animal welfare) that lead to some kind of compounding.
A concern I have about extrapolating from historical trends is the sensitivity to the form of the function used to extrapolate. For example, if a linear function was used here, then a strictly convex function (e.g. quadratic, exponential) might give a higher estimate for the trend without corporate campaigns, and we’d have to adjust the counterfactual impact of these campaigns downward. It seems plausible that the percentage cage-free would increase at an increasing rate, given how relatively flat it was from 2005 to 2010, and the possibility of some kind of network effects (e.g. in concern for or knowledge of animal welfare) that lead to some kind of compounding.