Multivariate analysis for dummies
Posted by Christie Malry on July 3, 2011 at 8:20 pm
Ritchie excels himself, with the following article about whether higher personal taxes will drive people out of the UK:
The latest ONS data on pop[ulation growth revealed the extraordinary fact that now we have a 50% tax rate in the UK fewer people emigrated.
Why is that?
Certainly it doesn’t suggest the rich are leaving in greater numbers.
Maybe it does suggest people like a more equal society.
And people came here in record numbers.
The myth that tax drives people away is, well, just a myth.
So’ like the rumour of fairies at the bottom of the garden shall we treat it as a load of old baloney - because that is what it is?
When challenged by his readers, he claimed that he was just being 'ironic' when in fact it's quite clear that he's being an idiot.
The factors which influence the decision whether or not to emigrate are complex. And it's obvious that tax will be one of these. If you're an entrepreneur and personal taxes in your country are 99%, then you will clearly want to move to a country with less punitive rates of taxation. But other factors are at work, such as the rates of tax in other countries, how easy it is to emigrate to another country, the languages they speak there, the system of laws there, what the weather is like there (yes, really), how easy it is to educate children there, and what the transport links are like.
Because there are lots of factors at work, it means that you can't meaningfully take simple observations and draw simple conclusions. You have to undertake a much more complex form of analysis, known as multivariate analysis. And even then you may only be able to draw broad inferences as to the direction of correlation.
Ritchie has shown time and time again that he is absolutely hopeless at multivariate analysis. He loves just to look at his desired outcome and a single variable and draw crass conclusions from random movements, while ignoring a whole host of other data included in the other variables. Two isolated observations, ignoring the other factors at play, are never going to be enough to work out what's really going on.
On the direction of correlation between taxation and emigration, which would ex ante appear to be positive, we can get some information from the team choices of NBA superstars (HT VeryBritishDude and Worstall). The article concludes that, in an environment where top pay is capped, NBA's top players choose to maximise their own personal take-home pay by choosing teams in low-tax states. This neatly eliminates a number of the possible variables that we identified above and leaves just a few, including the impact of personal tax rates.
What does Ritchie say to all this?
But you are showing weakness in multivariable (sic) analysis if you think tax the only cause
I don't. But because I really understand multivariate analysis, I know that a lower rate of emigration between two points is insufficient to conclude that higher tax rates do not correlate positively with greater emigration.
Leave a Reply