Gowrisankaran, Langer and Reguant have an excellent paper, Energy Transitions in Regulated Markets (WP), in the latest AER.

The basic idea is that regulation designed to prevent utilities from building useless power plants can induce them to keep obsolete power plants. Some background. We regulated electric utilities under the theory that they were natural monopolies and therefore we would do better by pushing their prices down. What’s a reasonable price? Hard to say, so regulated utilities were allowed to recoup their operating costs plus a fair return on their “rate base”—their capital stock. Makes sense, but once profits depended on the size of the capital stock, utilities had an incentive to build too much—the classic Averch–Johnson effect. Regulators responded with “prudence” requirements and the rule that capital must be “used and useful.” In a stable world, that rule is a check, albeit an imperfect check, on so-called gold-plating.

But now consider what happens in a time of technological change, such as a rapid decrease in the cost of generating electricity with natural gas (driven by fracking and improvements in combined-cycle natural-gas (CCNG) technology). In a free market, large decreases in costs would cause firms to abandon coal and move to natural gas—some would do this to make profits, others to avoid losses. In short, the market forces sunk investments to be abandoned when not profitable.

But there is another possibility under regulation. Tell the regulator that your plants are still viable. Well, telling is cheap talk so you keep burning coal to prove that the plant remains useful. If you can keep your base operating that’s better than abandoning it and to signal how valuable your coal plant still is, it may even be worth while to burn coal when the cost exceeds the price of electricity! The authors have some nice data on exactly this point.

Figure 3 takes a little work to understand, but the pattern is clear. Each point represents a state. In panel A, the vertical axis shows how much less likely a coal plant is to run when the cost of coal exceeds the price of electricity. Obviously, a strongly negative coefficient is the economically sensible response: when burning coal is more expensive than buying electricity, the plant should burn less.

The red points represent restructured states and the green points regulated states. In restructured states coal burning falls when prices fall, just as expected. Coal burning in regulated states responds much less. (I.e., the red points generally lie below the green points.) Indeed, the six states with the largest reductions in coal operation are all restructured states.

One objection to this analysis might be that utilities in general are just slow to respond to prices, so on the horizontal axis the authors plot how well utilities respond to a higher price of gas. Note that these coefficients are all negative and there is no obvious difference between regulated and restructured states. In both types of states, utilities respond well to the price of gas, but only in restructured states do utilities respond strongly to the price of coal. (Why coal and not gas? Because the used-and-useful standard binds on capital whose usefulness is in doubt—which, once gas got cheap, meant coal. In other words, the utilities have to defend coal to the regulators, not gas.)

Panel B on the right shows a slightly different way of presenting the same data. The vertical axis is again how much less likely a coal plant is to run when its cost exceeds the electricity price. The horizontal axis is the fraction of generation owned by electric utilities. Regulated states tend to be vertically integrated, while restructured states opened electricity generation to competition, so utility ownership and regulatory status are closely correlated. Regulated states generally have utility ownership above 60%, while all the restructured states but one are below 30%. The best-fit line slopes upward: in other words, the more generation a state’s utilities own, the less coal dispatch responds to price. A different perspective on the same story.

That is the direct empirical evidence. The authors then construct a more ambitious structural model. In theory, regulation could produce either too much or too little investment in the new technology; their estimates imply too much. Much, too much. Not only do regulated utilities retain too much coal, they also build too much gas capacity. In short, they accumulate both too much old capital and too much new capital. Averch–Johnson on steroids.

The bottom line is that regulation under dynamic conditions is much more difficult than under static conditions. My view is that it may not even be worth the candle.