What better thing to do than play with the United Van Lines inbound/outbound shipment data (that I blogged about here)? I regressed net inbound percentage (inbound minus outbound percentage) against the 2004 RPI Economic Freedom Index (EFI) using simple OLS. I had complete data for the 48 continental US states. A higher EFI means less economic freedom. A negative net inbound percentage means a larger proportion of shipments were outbound rather than inbound.
Here are the summary stats.
The average net inbound percent was 0.01, or 1 percentage point difference. I expected a negative relationship. Here's what STATA spat out at me.
|Linear regression||Number of obs||= 48|
|F( 1, 46)||= 14.88|
|Prob > F||= 0.0004|
|Root MSE||= .11774|
|netinpctbound | Coef.||Std. Err.||t||P>|t|||[95% Conf.||Interval]|
|efi | -.0096332||.002497||-3.86||0.000||-.0146593||-.0046071|
|_cons | .2606362||.0664169||3.92||0.000||.126946||.3943264|
What to make of this? All this tells me is that a higher Economic Freedom Index (meaning a lower amount of economic freedom) is statistically associated with a lower net percentage of inbound shipments. Economically speaking, a one point increase in EFI is associated with a decrease in net inbound percentage by just under 1 percentage point. Some salt for y'all:
1. Correlation does not imply causation, although my sense is that people want to be where they have freedom. After all, the Berlin wall was not meant to keep people
2. Omitted variables bias may be at play. Migration, like anything, depends on marginal costs and marginal benefits. Economic freedom surely plays a roll, but so do other things (weather, travel costs, etc.) that I have not controlled for.
3. The data periods do not line up (although I think it's safe to assume that the 2007 EFI isn't going to be that different from the 2004 one).
****Oy! Thanks to Dan Hill in the comments for noting my faux pas.