It seems quite obvious that the United States will have to increase energy taxes, first because energy prices need to better reflect the negative externalities they exert on the economy (pollution, wars, congestion, etc.), and second because the government will need to raise revenue from somewhere. So it is of particular interest to verify what consequence this would have, for example on labor markets. If the cost of an input increases, and it is cheaper elsewhere (or if not, the gap with elsewhere is reduced), one should expect labor demand to be reduced. But how much?
Olivier Deschenes does such an empirical exercise by exploiting the cross-state variation in energy prices and finds a price elasticity of 0.15. Given that the electricity prices are supposed to rise by 4%, a reduction in employment of 0.6% or 460,000 people is inferred. Unfortunately, these estimates cannot be relied on for several reasons. First, estimates are based on employment differences across states due to energy price differences. Imagine that electricity prices increase in Michigan, and this drives some jobs to Wisconsin. This is what this elasticity measures. But when electricity prices increases in all states, cross-state movements of jobs should not happen. They may move abroad, but this is not measured with this procedure, which certainly overestimates the correct elasticity.
Second, these are reduced form estimates that give us very little understanding how various agents in the economy would react to those price changes, especially if they are larger that the one observed in the data. One needs here some structural model to understand what would happen, a model that would, for example, include the industrial structure of each state. Computable general equilibrium (CGE) models have been used extensively for similar exercises, and should be much more reliable than this reduced form (Lucas Critique anyone?). Third, if you do not increase energy prices, it is quite obvious that some other tax will have to increase. An obvious candidate is labor income tax. Now that is certainly going to reduce employment as well, and possibly more than the 0.6% from above. In other words, the energy tax can very well be a lesser evil (on top of reducing pollution).
Some press will pick up this paper and rail against energy taxes. That is most unfortunate.