Obviously, as I mentioned earlier, there's a direct impact of dollars--for example, that's how we did the sensitivities on the price of oil--and that directly translates into higher value of output and investment. Keep in mind that the economic injection into any economy, regardless of where it comes from, will result in a multiple of itself, so that's very sensitive.
The input-output model has limitations that are used; there's no question about that. Interactions among different sectors that create these ripple effects of economic stimulus are based on established relationships of how much input each industry provides to the others. We have what we call coefficients that tell us, for instance, that one-tenth of the output of crude oil comes from manufacturing, just to throw that out. Those coefficients change over time, so the input-output analysis is more accurate in a shorter range of time, since those coefficients can be assumed to be stable and unchanging, but to the extent there's technological change over time, those relationships among different sectors will change, and therefore distributional impacts will also change. In other words, they're static coefficients, not dynamic ones; they don't take into account that the farther you go into the future, the less accurate it's going to be.