Evan Soltas
May 14, 2016

What Would a Carbon Tax Do?

For the last few months, I've working with Ishan Nath, an economics graduate student at University of Chicago and the Climate Impact Lab with Michael Greenstone, to understand the impacts of carbon taxation. Our project, specifically, concerns RGGI — the Regional Greenhouse Gas Initiative, a cap-and-trade scheme in the Northeastern United States — and I wanted to post some initial results about which I am quite excited. All of the work below I did with EPA data available here.

First, let's just look at how carbon-dioxide emissions at the average power plant in RGGI states and non-RGGI ("control") states. What we see is a sharp decline in carbon emissions that begins in the first quarter of 2009, when RGGI went into effect.

While this suggests that RGGI had an effect on carbon emissions, it's not conclusive. After all, 2009 was a big year — the deepest of the recession — and it might well be that an energy-demand shock reduced electricity production, and thus emissions, in RGGI states right around the timing of RGGI.

Implausible? Maybe a bit: The effect is quite sharp after the start of RGGI, and we don't see any cyclical dynamics in the graph. There's no decline in emissions in 2008, for instance, nor a clear recovery in 2010 and afterwards. But I think it would be eminently fair to question how well this graph identifies the causal effect of RGGI on carbon emissions.

What we need to do, in particular, is find some comparison — not just RGGI plant versus non-RGGI plant — that fixes this obvious problem posed by potential shocks to energy demand. So what we can do is compare the most-efficient and least-efficient plants across RGGI and non-RGGI states. The intuition here is that RGGI doesn't really affect you if you're a highly efficient plant in a RGGI state. First, because you're efficient, the additional marginal cost of the carbon tax isn't much, and therefore you respond weakly. Second, like most power plants, you're already near your capacity constraint, so it isn't possible to adjust production upwards in response to RGGI.

What RGGI should do, in other words, should be concentrated on the least-efficient power plants in RGGI states. And that's exactly what we see. In this graph, I've sorted every power plant in the United States into ten equal bins, ranked by their carbon-emissions efficiency, which I measure as the logarithm of the rate of CO2 emissions (in kilotons) per megawatt-hour of electricity, averaged over the 12 quarters prior to RGGI.

The plants in the lowest-efficiency decile in RGGI states saw average emissions fall 95 percent immediately after the implementation of RGGI. What RGGI seemed to do, simply put, is knock the dirtiest power plants off the grid.

This is actually the focus on my work with Ishan: What are the margins by which the electricity sector adjusts to carbon pricing? One option is that existing plants become more efficient. Another is that the worst plants shut down, and new plants come online that are better. A third option is that, in response to pricier electricity, electricity use falls.

Economists don't know the correct answer to that question, which is why many support market solutions that allow the private sector to determine the least costly way to reduce carbon emissions. Yet the United States, at least at the moment, seems to be pursuing a more "command-and-control" regulation approach to carbon-emissions reduction. I hope our work might give economists some sense of how close, or how far away, a regulatory "second-best" solution would be away from the "first-best" world of carbon pricing. It's both exciting and troubling to me that I have no idea what the answer to that question is.