One of the great challenges in econometrics, the statistical branch of economics that deals with estimating the relationships between different variables in a system, is finding factors that are exogenous, e.g. originate entirely outside the system you are trying to analyze. This is tough because often times a variable A does not only influence a variable B, but B also has a feedback effect back on A, possibly through its effect on some other variable. And in such cases of endogeneiety most of the commonly used estimation techniques are invalid.
There are two main ways out of this dilemma: either you use far more sophisticated estimation techniques that are robust to this phenomenon or you try very hard to find a variable, some variable for which you can make a good case that it is truly exogenous.
This, I assume, is the great appeal of using weather as a predictive variable in empirical models. After all, whatever happens in an economy, it’s pretty clear that none of it is going to change the amount of sunshine or rainfall the next day. A Freakonomics column from a while back showcases such research, which has produced some very intresting insights, like this one about all our favorite German tribe:
Consider 19th-century Bavaria. The problem there was rain — too much of it. As Halvor Mehlum, Edward Miguel and Ragnar Torvik explained in a recent paper, excessive rain damaged the rye crop by interfering with the planting and the harvest. Using a historical rainfall database from the United Nations, they found that the price of rye was significantly higher in rainy years, and since rye was a major staple of the Bavarian diet, food prices across the board were considerably higher in those years, too. This was a big problem, since a poor family at the time would have been likely to spend as much as 80 percent of its money on food. The economists went looking for other effects of this weather shock. It turns out that Bavaria kept remarkably comprehensive crime statistics — the most meticulous in all of Germany — and when laid out one atop the other, there was a startlingly robust correlation between the amount of rain, the price of rye and the rate of property crime: they rose and fell together in lockstep. Rain raised food prices, and those prices, in turn, led hungry families to steal in order to feed themselves.
But violent crime fell during the rainy years, at the same time property crimes were on the rise. Why should that be? Because, the economists contend, rye was also used to make beer. “Ten percent of Bavarian household income went to beer purchases alone,” they write. So as a price spike in rye led to a price spike in beer, there was less beer consumed — which in turn led to fewer assaults and murders.