3 ms·
I need no convincing that decisions get made based on tax rates. However, I call "correlation is not causation" on this particular article. It is highly possib
by roenxi 9y ago
I need no convincing that decisions get made based on tax rates.
However, I call "correlation is not causation" on this particular article. It is highly possible that tax increases and migration are both linked to some third variable ('local conditions') and the /Testing the validity of the results/ section of the article does not convince me that this possibility has been ruled out. Establishing why there is a link here requires more information.
I have observed that governments hate raising taxes because it upsets voters. In practice, resistance to raising debt takes a back seat to upsetting voters. If a government is raising taxes, in my experience, the economy is struggling and making up the government budget with debt isn't an option (for whatever reason). It is possible that rate of tax follows a similar principle where prosperous regions need a lower rate to achieve a better result.
- twoodfin 9y agoThe authors note that shifts in migration patterns follow, rather than precede, shifts in top-rate (but not middle-rate) taxation.
- roenxi 9y agoYeah, but that isn't actually very strong causative evidence. If there is a confounding variable, maybe upending a scientists life and moving states takes longer than than the legislature raising the tax rate. Maybe the tax change compounds the impact of another variable and is followed by their relocation (ie, the tax change contributes to the decision, but isn't the root cause). The point is, "A causes B" is a much stronger statement than "A and B are related". There is solid evidence of a relationship, because the core of their study is a statistical test of correlation. The causative elements here are of a speculative nature, because they didn't gather any data on cause. Cause is being extrapolated from a correlation, which is dangerous thinking. The conclusion here is a good one to be circumspect about.