Title:Renewable Electricity Generation
Hardcover Dimensions:6" x 9"
- 80 Hardcover pages
- Buy the Book
- Wind and solar power prove themselves uncompetitive even with large subsidies
- Rationales that fuel support for renewable energy are superficially appealing but deeply problematic
- Employment in renewable sectors created by renewable policies would be an economic cost rather than benefit for the economy as a whole
Download PDF Public policy support for renewable electricity—wind and solar power in particular—is substantial, taking the form of large subsidies both direct and indirect. A number of rationales usually are offered in support of those public policies; whatever their surface plausibility, they are deeply problematic both conceptually and in terms of the available data. In short, they are wholly unpersuasive and provide a weak basis for policy formulation. This second in a three-part Outlook series discusses these rationales. [Read part 1] [Read part 3]
Key points in this Outlook:
- The public subsidies that support wind and solar power in America are fueled by a set of rationales that are superficially appealing but deeply problematic.
- Supporters say renewables need subsidies because they are an “infant industry” or because they need to “level the playing field” with conventional sources. But the data show that scale and learning efficiencies are very unlikely and that solar and wind already receive the highest levels of taxpayer assistance.
- Unreliable wind and solar sources require backup power generation—a dual setup that imposes costs higher than the adverse environmental effects of conventional sources alone.
- The hoped-for creation of “green jobs” confuses benefits for special interests with costs for the whole economy. The net effect would be a reduction in overall employment.
As illustrated in the first in this series of Outlooks (No. 1, January 2012), wind and solar power are proving themselves uncompetitive even with large subsidies, both direct and indirect, at the state and federal levels. This policy support has yielded only small increases in the supply of electric power, at a very high cost. And so, a fortiori, the preservation and, perhaps, expansion of that policy support somehow must be justified. The central arguments for this prominent, long-standing support are numerous and varied but generally fall into the following categories:
• renewable energy as an “infant industry”;
• offsets for the subsidies enjoyed by conventional generation;
• the adverse environmental effects of conventional generation;
• resource depletion, or “sustainability”; and
• renewable electricity as a source of expanded “green” employment.
The Infant Industry Argument
This argument begins with the assumption that new technologies often cannot compete with established ones because the initial available market is too small for important scale economies to be exploited and because the downward shifts in costs that might result from a learning process, again, cannot be achieved without substantial expansion in market share. Accordingly, policy support for expansion of the newcomers’ share of the market is justified by supporters as a tool with which to allow the achievement of both scale and learning efficiencies.
One obvious problem with this argument is that the market for electric power already has several competing technologies, each of which began with a small market share, as is the case with all new technology. More generally, many industries employing competing technologies are characterized by the presence of scale economies, learning efficiencies, or both, but market forces operating through domestic and international capital markets provide investment capital in anticipation of future cost savings and higher economic returns. Accordingly, the infant industry argument is a non sequitur: the market can foresee the potential for scale and learning efficiencies and invest accordingly. No efficiency rationale for subsidies or other policy support follows from this argument.
In any event, the narrower issue is whether important learning or scale efficiencies remain available to be exploited for cost reductions for wind or solar generation. The pattern of average costs over time, controlling for the size of projects, should yield inferences about the remaining importance of learning efficiencies; if the infant industry argument is correct, we should observe in the data over the last decade or two declining costs for renewable electricity. For wind generation, the US Department of Energy (DOE) reports data on average project cost per megawatt (MW) over time, beginning in the early 1980s.
These data, while somewhat crude, show a rough pattern of declining average costs from the 1980s through about 2001 and then rising average costs through 2009: from about $4,800 per MW in 1984 to about $1,300 per MW in 2001, rising to about $2,100 in 2009, all in constant 2009 dollars. Because these data are weighted by capacity, the rising average costs per wind MW after 2000–01 suggest that further learning efficiencies no longer are available to be exploited unless, perhaps, future technological advances are made.
Other DOE data are available on average costs by project size for wind projects installed in 2007–09. The short period reduces the likely impact of learning efficiencies, yielding important information about the availability of scale economies. The data show that scale economies are important for only small wind projects (about $2,700 per MW for projects smaller than 5 MW) and that constant or slightly increasing average costs (about $1,800-$2,000 per MW) characterize projects larger than about 20 MW.
Reliable time-series data on costs for photovoltaic and thermal solar systems are more difficult to find; perhaps the only consistent series is provided by the US Energy Information Administration (EIA) for 2000–09.1 These data show a decline in costs per MW for both photovoltaic and thermal systems early in the decade, suggesting the exploitation of learning efficiencies and, perhaps, the use of more suitable (that is, lower-cost) sites. The data show also an increase in costs per MW after 2002; this suggests that no further learning efficiencies are available to be exploited, that the problem of rising site costs is significant, or both.2 On the other hand, a different DOE data analysis for photovoltaics only shows a decline in the capacity-weighted average installed cost between 1998 and 2008, from $10.80 per watt (2008 dollars) to $7.50 per watt. In short, the data are mixed in the case of solar generation systems. The infant industry assumption of significant learning and scale economies as a barrier to adoption of renewable technologies at best is far from obviously correct.
Leveling the Playing Field
The second central argument made in favor of policy support for renewables is essentially a level-playing-field premise: because conventional generation benefits from important tax preferences and other policy support, renewables cannot compete without similar treatment. A recent EIA analysis presents data that we can use to compare federal subsidies and support per megawatt hour (mWh) produced by different technologies.3 These data are presented in table 1.4
These data show that federal solar and wind subsidies in fiscal year 2010 were far higher—by two or three orders of magnitude—than those enjoyed by fossil fuels, nuclear, or hydroelectric generation. Accordingly, solar and wind technologies clearly are not at a competitive disadvantage because of subsidies enjoyed by conventional generation; quite the reverse is true.
Adverse External Effects of Conventional Generation
A negative “externality” is an adverse effect of economic activity, the full costs of which are not borne by the parties engaging directly in the activity yielding the adverse effect. A simple example is the emission of effluents into the air as a byproduct of such industrial processes as power generation. Clearly, power generation with fossil fuels imposes adverse environmental effects in the form of sulfur dioxide, nitrogen oxides, mercury, particulates, and other effluents. Accordingly, the EPA and individual states have established detailed programs for defining emission standards and for implementing attendant investment and enforcement programs.
If the negative externalities yielded by conventional generation are not internalized fully by current environmental policies—that is, if buyers and producers are not confronted with the full environmental costs that they impose on others—then the costs of conventional generation as perceived by the market would be (artificially) lower than the true social costs. At the same time, the unreliable nature of wind and solar generation imposes a requirement for costly backup capacity, as discussed in the first part of this Outlook series. So the question to be addressed is: given the magnitude of those externalities as estimated in the technical literature, are the additional (or marginal) costs of backup capacity imposed by renewable generation sufficient to offset any artificial cost advantage enjoyed by conventional generation?
A number of analyses of the externality costs of US electricity generation were conducted during the 1980s and 1990s. These studies differ somewhat in terms of methodology and focus but offer a range of estimates useful in terms of the question addressed here. The estimated externality costs for coal range from 0.1 cents per kilowatt hour (kWh) to 26.5 cents per kWh. For gas generation, the range is 0.1–10.2 cents per kWh. For oil, nuclear, and hydro generation, the respective ranges are 0.4–16.5 cents per kWh, 0–4.9 cents per kWh, and 0–2.1 cents per kWh.
The highest estimated figure for coal generation is 26.5 cents per kWh, or $265 per mWh. As presented in my first Outlook, a conservative estimate of the cost of backup capacity for existing wind and solar generation is about $368 per mWh, or roughly 37 cents per kWh. Accordingly, if all conventional generation were coal fired, existing wind and solar capacity would impose a backup cost externality about 39 percent higher than the environmental externality costs of conventional generation, under the implausible assumption that none of the conventional externalities have been internalized under current environmental policy.
But, in fact, coal generation is a bit less than 45 percent of total US generation, gas generation is about
24 percent, nuclear generation is about 20 percent, hydroelectric generation is about 6 percent, and renewables and other miscellaneous technologies make up the rest.5 If we use those figures and the highest estimates by fuel type noted above to compute a weighted-average externality cost for nonrenewable generation, the externality cost per conventional kWh is about 15.5 cents, or $155 per mWh. If we use instead the midpoints of the externality ranges listed above, the weighted average externality cost is 7.8 cents per kWh, or $78 per mWh.
Relative to the backup cost externality ($368 per mWh) imposed by wind and solar investments alone, those figures are sufficiently low to cast substantial doubt on the externality argument for renewables subsidies; current environmental regulation must internalize some substantial part of conventional externalities, and federal and state subsidies, both explicit and implicit, and requirements for minimum market shares for renewables also offset any artificial cost advantage enjoyed by conventional generation as a result of uninternalized externalities.
Note that, in terms of economic efficiency, subsidies for renewables intended to offset the (assumed) uninternalized external costs of conventional generation are a second-best policy at best. Such subsidies would reduce the (inefficient) competitive advantage of conventional generation yielded by the presence of some social costs not reflected in prices, but they would not improve the efficiency of costs or prices for conventional generation. And by biasing the perceived costs and prices of renewable generation downward, the subsidies would result in a total electricity market that would be too large. In short, the externality argument in favor of policy support for renewable electricity generation is exceedingly weak, far more so than commonly assumed.
The Resource Depletion, or “Sustainability,” Argument
“Renewable” energy has no uniform definition, but the (assumed) finite physical quantity of conventional energy sources is the essential characteristic differentiating the two in most discussions. In a word, conventional energy sources are depletable. In contrast, sunlight and wind flows replenish themselves, a central component of “sustainability,” perhaps a broader concept than natural replenishment; sustainability has been defined by the EPA as “the satisfaction of basic economic, social, and security needs now and in the future without undermining the natural resource base and environmental quality on which life depends.”6
As an aside, the energy content of sunlight and wind is finite, regardless of self-replenishment. These sources contain only so much convertible energy, and they are not always available. Moreover, the same is true for the other resources—materials, land, and so forth—upon which the conversion of renewable energy into electricity depends. In any event, the basic sustainability concept seems to be that without policy intervention, market forces will result in the depletion (or exhaustion) of a finite resource. Accordingly, subsidies and other support for renewable power generation are justified as tools with which to slow such depletion and hasten the development of technologies that would provide alternatives for future generations.
That argument is deeply problematic. Putting aside the issue of whether government as an institution actually has incentives to adopt a time horizon longer than that relevant for the private sector, the profit motive provides incentives for the market to consider the long-run effects of current decisions. The market rate of interest is a price that links the interests of present and future generations. If a resource is being depleted, its expected future price will rise, other things held constant. If that rate of price increase is greater than the market interest rate, then owners of the resource have incentives to reduce production today; by doing so they can sell the resource in the future and in effect earn a rate of return higher than the market rate of interest, thus raising prices today and reducing expected future prices. In equilibrium—again, other factors held constant—expected prices should rise at the market rate of interest.7 Under market institutions, the market rate of interest ties the interests of the current and future generations by making it profitable currently to conserve some considerable volume of exhaustible resources for future consumption.8 Because of the market rate of interest, market forces will never allow the depletion of a given resource.
Accordingly, the market has powerful incentives to conserve—that is, to shift the consumption of some resources into future periods. That is why, for example, not all crude oil was used up decades ago even though the market price of crude oil always was greater than zero, which is to say that using it would have yielded value. In short, the sustainability argument for policy support for renewable electricity depends crucially on the dual assumptions that the market conserves too little and that government has incentives to improve the allocation of exhaustible resources over time. The underlying rationale of that dual premise is weak, and little persuasive evidence has been presented to support it.
Renewable Power as a Source of Expanded Employment
A common argument for expanded renewable power posits that policies supporting that goal will yield important benefits in the form of employment growth in renewables sectors and stronger demand in the labor market in the aggregate. Both of those premises are almost certainly incorrect.
The employment in renewables sectors created by renewables policies actually would be an economic cost rather than a benefit for the economy as a whole. Suppose that policy support for renewables (or any other sector) had the effect of increasing the demand for high-quality steel. That clearly would benefit steel producers, or more broadly, owners of inputs in steel production, including steel workers. But for the economy as a whole, the need for additional high-quality steel in an expanding renewable power sector would be an economic cost, as that steel (or the resources used to produce it) would not be available for use in other sectors. Similarly, the creation of “green jobs” as a side effect of renewables policies benefits the workers hired (or those whose wages rise with increased market competition for their services). But for the economy as whole, that use of scarce labor would be a cost because those workers no longer would be available for productive activity elsewhere.9
Moreover, an expansion of the renewable electricity sector must mean a decline in some other sector(s), with an attendant reduction in resource use there; after all, resources in the aggregate are finite. If there is substantial unemployment, and if labor demand in renewables is not highly specialized, a short-run increase in total employment might result. But in the long run—not necessarily a long period of time—such industrial policies cannot “create” employment; they can only shift it among economic sectors.
In short, an expanding renewables sector must be accompanied by a decline in other sectors, whether relative or absolute, and creation of green jobs must be accompanied by destruction of jobs elsewhere. Even if an expanding renewables sector is more labor-intensive (per unit of output) than the sectors that would decline as a result, the employment expansion would still be a cost for the economy as a whole, and the aggregate result would be an economy smaller than otherwise would be the case.10 No particular reason exists to believe that the employment gained as a result of the (hypothetically) greater labor intensiveness of renewables systematically would be greater than the employment lost because of the decline of other sectors combined with the adverse employment effect of the smaller economy in the aggregate. There is, in addition, the adverse employment effect of the explicit or implicit taxes that must be imposed to finance the expansion of renewable power.
Because renewable electricity generation is more costly than conventional generation, policies driving a shift toward heavier reliance on the former would increase aggregate electricity costs, thus reducing electricity use below levels that would prevail otherwise. The 2007 EIA projection of total US electricity consumption in 2030 was about 5.17 million gigawatt hours (gWh). The latest EIA projection for 2030 is about 4.31 million gWh, a decline of about 16.6 percent.11 The change presumably reflects some combination of assumptions about structural economic shifts, increased conservation, substitution of renewables for some conventional generation, and a price increase from about 8.8 cents per kWh to 9.0 cents (in 2009 dollars). It would be surprising if that reduction failed to have some effect on employment.
Figure 1 displays data on percent changes in real gross domestic product (GDP), electricity consumption, and employment for 1970–2009. It is obvious from the aggregate trends that electricity use and labor employment are complements rather than substitutes; the simple correlation between the percent changes for the two is 0.61, meaning, crudely, that a percent change in one tends to be observed with a 0.61 percent change in the other in the same direction. The simple GDP/electricity and GDP/employment correlations are 0.67 and 0.85, respectively.
The correlations by themselves are not evidence of causation, the determination or refutation of which requires application and statistical testing of a conceptual model. But the data displayed in figure 1 make it reasonable to hypothesize that the higher costs and reduced electricity consumption attendant upon expansion of renewable generation would reduce employment, and they certainly provide grounds to question the common assertion that policies in support of expanded renewable electricity generation would yield increases in aggregate employment.
It certainly is possible that the historical relationship between employment and electricity consumption will change. Technological advances are certain to occur, but the prospective nature and effects of those shifts are difficult to predict.12 The US economy may evolve over time in ways yielding important changes in the relative sizes of industries and sectors, but again, the direction of the attendant shifts in employment and electricity use is ambiguous.
But there exists no evidence with which to predict that a reduction in electricity consumption would yield an increase in employment. Like all geographic entities, the United States has certain long-term characteristics—climate, available resources, geographic location, trading partners, ad infinitum—that play a substantial role in determining the long-run comparative advantages of the economy in terms of economic activities and specialization. Figure 2 presents the historical paths of the electricity intensity of US GDP (kWh per dollar of output) and of the labor intensity of US electricity consumption (employment per kWh).
From 1970 to 2009, the electricity intensity of GDP increased and declined at various points, but for the whole period it declined slightly at a compound annual rate of about 0.3 percent. The labor intensity of US electricity consumption—essentially, the employment “supported” by each increment of electricity
consumption—has declined more-or-less monotonically over the entire period, at an annual compound rate of about 1.05 percent. This may be the result largely of changes in the composition of GDP (toward services) and perhaps the substantial increase in US labor productivity in manufacturing.
But these data do not suggest that a reduction in electricity consumption would yield an increase in aggregate employment; if anything, they suggest the reverse. In short, while the employment/electricity relationship may have declined over time, there is no evidence that it is unimportant in an absolute sense, and it is far from inverse.
The central rationales usually offered in support of subsidies and other policies favoring renewable electricity are far weaker than the attention paid to them might suggest, both in terms of conceptual rigor and underlying support in terms of the available data.
Regardless of the nature and magnitude of public policy support for renewable electricity, it must compete with conventional technologies in terms of investment choices and market shares. One crucial dimension of such competition is the price of conventional fuels. The third part of this Outlook series (No. 3, January 2012) summarizes the direct and indirect subsidies enjoyed by wind and solar power and discusses the implications of ongoing developments in the natural gas market.
Benjamin Zycher ([email protected]) is a visiting scholar at AEI. This second in a series of three Outlooks is based on his new book, Renewable Electricity Generation: Economic Analysis and Outlook (AEI Press, 2011). The author thanks Kenneth Green for his suggestions on an earlier draft.
1. See the Electricity Market Module discussions contained within the “Assumptions” chapters of each year’s “Annual Energy Outlook,” www.eia.doe.gov/oiaf/archive.html (accessed December 19, 2011).
2. For photovoltaic systems, capacity costs fell from $5,386 per MW in 2000 to $4,744 in 2002 and then increased steadily to $6,239 in 2009. For thermal systems, the figures were $3,679 in 2000, $3,194 in 2002, and $5,237 in 2009.
3. See US Energy Information Administration, Direct Federal Financial Interventions and Subsidies in Energy in Fiscal Year 2010 (Washington, DC: US Department of Energy, July 2011), http://docs.wind-watch.org/US-subsidy-2010.pdf (accessed December 8, 2011).
4. Other things held constant, subsidies that affect the marginal (or incremental) cost of generation or the per-unit prices received by particular technologies are likely to affect market prices, even under standard rate-of-return regulation, and so might create a competitive disadvantage for other technologies not receiving equivalent treatment. An example is the per-unit production tax credit for renewable power. Other credits for renewables—for example, investment tax credits—might improve profitability without affecting marginal costs or prices directly. Such tax credits would attract additional investment into the industry over time, perhaps affecting market prices, but that price effect would be felt by all producers regardless of which ones actually received the subsidy. At the same time, even such subsidies would serve to reduce or eliminate whatever competitive disadvantages confront renewables as a result of policies in support of conventional generation.
5. See EIA, “Net Generation by Energy Source Data,” 2009, www.eia.gov/electricity/data.cfm#generation (accessed December 20, 2011).
6. See the brief sustainability discussion from US Environmental Protection Agency, “What Is Sustainability?,” February 2011, http://epa.gov/sustainability/basicinfo.htm (accessed December 19, 2011).
7. In reality, the long-run prices of most exhaustible natural resources have declined (after adjusting for inflation), in large part because of technological advances in discovery, production, and use.
8. Strictly speaking, it is not the price of the resource that should rise at the market rate of interest; instead, the total economic return to holding the resource for future use should equal the market rate of interest. That economic return includes expected price changes and capital gains, expected cost savings, and the like.
9. Considerable employment would be created if policies encouraged ditch digging with shovels (or, in Milton Friedman’s famous example, spoons) rather than heavy equipment. Such employment obviously would be laughable—that is, an obvious economic burden. There is no analytic difference between this example and the green jobs rationale for renewables subsidies.
10. Many advocates of renewables subsidies assert that solar and wind power are more labor intensive than conventional generation. The assumption of greater labor intensity for renewable power production is dubious: the operation of solar or wind facilities does not employ large amounts of labor, and it is far from clear that construction of solar or wind facilities is more labor intensive than construction of conventional generation facilities.
11. For 2007, see table 2, EIA, “Annual Energy Outlook 2007 with Projections to 2030,” February 2007, www.eia.doe
.gov/oiaf/archive/aeo07/aeoref_tab.html (accessed December 19, 2011). The British thermal unit (BTU)/kWh conversion was made by the author at a rate of 3,412 BTU per kWh. For the later projection, see table 55, EIA, “Annual Energy Outlook 2011: Electric Power,” 2011, www.eia.gov/forecasts/aeo
/sector_electric_power.cfm (accessed December 20, 2011).
12. Note that greater energy “efficiency” in any given activity can yield an increase in actual energy consumption if the elasticity of energy consumption with respect to the marginal energy cost of appliance use is greater than one. If, for example, air conditioning were to become sufficiently efficient in terms of energy consumption per degree of cooling, it is possible that air conditioners would be run so much that total energy consumption in space cooling would increase. A tax, on the other hand, whether explicit or implicit, increases the price of energy use, unambiguously reducing energy consumption.