Maybe there's a problem analogous to "clustering" in opinion polls. I presume every now and again econometrics people have to tweak the way they measure GDP.

Maybe there's a problem analogous to "clustering" in opinion polls. I presume every now and again econometrics people have to tweak the way they measure GDP. Maybe if someone proposes a plausible tweak and it results in a surprisingly high or low figure their idea gets ignored, while it might be taken seriously if it ended up moving the figure towards the historical average.