Wednesday, February 2, 2011
Market distance signs: a clear break
I used geographically-weighted regression to get location specific coefficients for market distance in the regression: Rent = wheat yield + market-town population + distance to market-town. As I have gone on and on about before, the theory is that there should be a negative sign for distance, because the further you are from your market, the more it costs you to truck your produce in on market day. The fact that this sign changes has been perplexing me...and there is something interesting going on here. Look at the map below. Can you see that the locations to the west of that thick red line have a negative sign, those to the right a positive sign? I did put a marking on the map but it is a little faint. Wow! Such a clear boundary. But why is it there and not a few kilometres to the west or east for that matter? What are factors that decide the location of this line? Interesting!