1 Correlates of Nuclear Energy: Back to the Future or Back to Basics?

On march 11, 2011, waves as high as 14 meters hit the nuclear power plants (NPPs) of Fukushima Daiichi in Japan, knocking out backup power and resulting in a loss-of-coolant accident when pumps went off line and temperatures rose high enough to damage fuel elements. The aftermath of the crisis left the utility operator of the ill-fated NPPs, Tokyo Electric Power Company or TEPCo, contemplating bankruptcy after suffering the largest loss ever for a Japanese firm outside the financial sector. Japan's Prime Minister, then Naoto Kan, subsequently called for upgrading safety and security inspections and announced that the country would not build new nuclear plants—a strict volte-face from pre-tsunami plans of a 30 percent growth in nuclear capacity over the coming decade (Economist Intelligence Unit 2011). The shutdown of almost two-thirds of the country's reactors (in response to both the accident and accelerated schedules of inspections) saddled utilities across the country with short- and long-term electricity shortages. Coupled with the formal scathing indictment of “human error” and poor safety planning, this stirred national debate over the costs and risks associated with future expansion of nuclear energy (FNAIIC 2012). Similarly, global public reflection on the costs and possible consequences of nuclear energy led many states to reexamine their nuclear industries, with divergent conclusions. For example, while the Germans committed to phasing out nuclear power and Italy scrapped plans to reconstitute nuclear capacity, other nuclear energy states (e.g., Russia and China), as well as aspirants (e.g., Vietnam and the United Arab Emirates [UAE]), elected to press ahead with existing plans for nuclear power development. As discussed below, Tokyo faces a confluence of factors that render nuclear a particularly appealing power source, and which resulted in the Japanese building nuclear plants when almost no other country was doing so. If Japan—given the powerful drivers motivating its pursuit of nuclear that include geographic isolation, resource scarcity, and status as a proponent of climate change mitigation—follows suit with Germany in abandoning this energy source, what can be expected of states with far less pressing motivations?

The issue takes on added significance in light of the mixed bag of nuclear energy states. As of 2012, thirty-one countries possess an operational commercial-scale nuclear energy infrastructure; and they are a diverse set of states, including Armenia, Belgium, China, Finland, Pakistan, the United Kingdom, Mexico, and Argentina. These states, as well as more than fifty others that have expressed interest in joining them, run the gamut in terms of standard of living, governance, location, population size, security environments, resource allocation, and levels of industrial development.

This raises the following questions: Are there characteristics shared by nuclear power–generating states that distinguish them from non-nuclear energy states? Do historical commonalities hold for contemporary nuclear energy aspirants? What are the implications for projecting the scope and scale of alternative nuclear energy futures? In short, how much do we know (or not know) about the correlates of nuclear power generation?

Answers to these questions are important for illuminating the contours of a prospective global expansion of nuclear energy. First, identification of common attributes can assist with distinguishing rhetoric from reality surrounding a much-heralded nuclear energy renaissance. Notwithstanding both the widespread resuscitation of interest in nuclear power before the Fukushima Daiichi disaster and the public shock waves that ensued, the lion's share of new reactor construction has occurred, to date, in states already in possession of a nuclear energy infrastructure, especially China, India, and Russia. There are notable cases of states that have articulated a strong commitment to nuclear power for decades but have achieved, at best, modest success, such as Indonesia and Turkey. Do these trends represent a new era or a temporary waypoint in the development of nuclear energy? Exploration of shared traits among existing and aspirant nuclear energy states can shed light on similarities and distinctions between historical and contemporary trajectories of nuclear energy stagnation, resurgence, or renaissance.

Second, systematic inquiry into shared characteristics of nuclear energy states can advance the nascent scholarly debate over the drivers of nuclear energy acquisition. Little is known about what motivates states to acquire nuclear power generation capacity beyond disparate statements by national leaders and governing bodies. Indeed, scholars are only beginning to probe the drivers of nuclear energy development using quantitative empirical analysis (Nelson and Sprecher 2008; Jewel 2011; Fuhrmann 2012). The extant literature typically focuses on explanations for either weapons capabilities or access to weapons-related technologies, or whether there is a link between acquisition of nuclear energy and weapons programs (latent or otherwise) (e.g., Meyer 1984; Sagan 1996/97, 2010; Singh and Way 2004; Jo and Gartzke 2007; Barnaby 2009; Fuhrmann 2009a, b; Kroenig 2009a, b; Findlay 2011). Most studies that address the acquisition of nuclear energy programs tend to give priority to idiosyncratic political motivations and decision-making processes, or analyze the political interests, national aspirations, or strategic consequences associated with the empowerment of domestic nuclear energy lobbies, rather than offering systematic analyses of general patterns and trends of commercial energy acquisitions across time, states, or regions (e.g., Hymans 2006; Solingen 2007; Fitzpatrick 2009; Malley and Ogilvie-White 2009; Rublee 2009). Although correlation cannot be equated with causation, by filling methodological gaps and refining metrics to test rival claims, a large-N study of common attributes among states with civilian nuclear power industries can build on recent scholarly insight by exposing important and systematic inflection points for alternative nuclear scenarios, as well as by identifying critical case studies for the future and rich analysis of drivers and processes associated with the diffusion of nuclear energy (Jewell 2011).

Finally, the character of the expansion of nuclear power has implications for policy. Will nuclear energy optimists or pessimists carry the day (see Busby, Chapter 5 in this volume)? Some people have noted that the aspiring nuclear energy states, considered in the aggregate, are decidedly weaker in governance and political stability than existing nuclear energy states, which accentuates concerns about accidents, terrorism, and weapons proliferation of a global nuclear energy revival. Given the historical correlation of political instability with states that possess nuclear weapons, this observation also draws attention to additional levels of national commitment required by newcomers to realize their nuclear energy ambitions (Miller and Sagan 2009; Jewell 2011). Yet without a systematic assessment of the range of potential drivers or an understanding of how current possessor states compare both over time and to non-nuclear energy states, it is difficult to comprehend the ramifications of these observations for managing future nuclear safety and security. A refined appreciation of distinguishing traits can provide purchase on whether the future nuclear energy club will be characterized by pariahs, such as Iran, or by engagers, such as the UAE—the former being the newest nascent state and the latter expected to be the next—with attendant implications for balancing international commercial and security policy responses.¹

This chapter synthesizes what we know and do not know about common characteristics of nuclear energy states. The first section reviews the diversity of the nuclear energy club and the challenges posed by extant explanations for nuclear power generation. The next section discusses the results of a statistical model that tests rival hypotheses concerning shared attributes among nuclear power–generating states from 1950 to 2001. Our findings confirm that nuclear energy states typically have large national incomes. There also is empirical evidence that energy insecurity, defined as reliance on foreign sources of energy, is positively correlated with construction of nuclear energy capacity. Our results also tentatively support the notion of a conspicuous drop-off in the pursuit of nuclear energy in the wake of major nuclear safety accidents, most notably following the Chernobyl disaster. In contrast, many popular explanations such as those rooted in economic growth, regime type, and strategic considerations do not appear to influence nuclear power development in a systematic manner. While it is clear that some nuclear energy programs grew from weapons programs and that some nuclear energy aspirants have been primarily interested in nuclear weapons, given that other states pursued nuclear energy wholeheartedly without working toward the bomb, the question of whether there is a systematic connection remains relevant.² The third part discusses the implications of these historical patterns for projecting the emerging nuclear energy landscape, highlighting both limitations of quantitative models and prospective inflection points for discerning alternative scenarios. The final section explores implications that will help determine whether contemporary nuclear energy aspirants are either doomed to stasis or poised to propel a global deepening and broadening of nuclear energy.

What is the profile of a nuclear power–generating state? This question does not lend itself to straightforward answers, given the checkered nuclear landscape. A cursory review of the thirty-one states that operate commercial NPPs, as well as those that have shut-down NPPs (i.e., Kazakhstan, Lithuania, and Italy), reveals tremendous variation along basic parameters. For example, while wealthy states are prevalent in the realm of nuclear energy, there are possessor states with low per capita national incomes (often considered a defining indicator of economic development). Historical inertia also resonates among the majority of the nuclear energy states. Most of the early entrants into the nuclear power domain were either North Atlantic high-income democratic countries with market economies or former Warsaw Pact members. States that meet one of these criteria still comprise nearly two-thirds of the countries that operate NPPs.³ The history of nuclear energy development is marked by countries that pursued nuclear energy after launching a bomb-related program (e.g., the United States and China), as well as by countries that embarked on extensive civilian nuclear energy programs without undertaking significant efforts to develop nuclear weapons (e.g., Japan and the Netherlands). There also are states—such as India, Brazil, Sweden, South Korea, Taiwan, Pakistan, Yugoslavia, and Iraq—that at one time or another displayed ambiguous motivations by pursuing parallel development of civilian and military nuclear programs (Bose 2005; Ollapally 2001). As well, NPP states have exhibited varying propensities to maintain a dedication to nuclear energy in the face of changing conditions, such as those concerning safety. As noted above, in the wake of the Fukushima Daiichi incident, Germany opted to scrap its plans to extend the life of its nuclear plants to 2036 in favor of phasing out domestic nuclear by 2022, while Italy and Switzerland canceled plans for new NPPs (Monitor's Editorial Board 2011). However, a recent report forecasted that of thirteen major nuclear energy states, only two (Japan and Germany) were not projected to increase nuclear installed capacity by 2020 (Economist Intelligence Unit 2011).⁴

At the same time, a heterogeneous group of more than fifty countries shows varying degrees of interest in developing national nuclear infrastructures—mostly power plants, but for a few this includes fuel cycle facilities downstream of existing mining or milling activities (WNA 2009; IAEA 2010b). Some of these aspirants have vacillated on the issue with changes in political administrations, economic conditions, or world events. Australia's varying policy stances offer prime examples of the effect of political change on nuclear power development (Coorey 2009). Egypt pursued nuclear power but changed course—allegedly in response to the Chernobyl accident—only to renew a favorable disposition toward nuclear energy (MIIS 2009). After protracted periods of disinterest, countries with large natural uranium reserves, such as Australia, Brazil, and Kazakhstan, now trumpet the advantages of moving up the value chain by constructing reactors and producing nuclear fuel (Falk, Green, and Mudd 2006; IAEA 2006; SCIR 2007).⁵ Some states, located in the Baltic, Persian Gulf, and Balkan regions, have been satisfied with arrangements that enable them to consume electricity from NPPs situated in other countries, such as the Slovenian/Croatian Krško power plant (WNA 2009). Nuclear power aspirants have ranged from Italy, a wealthy state that formerly generated nuclear power, to countries such as Ghana, Namibia, and Bangladesh that face acute limitations of resources and infrastructures. There have been a number of attempts in recent years to gauge which aspirant states are most likely to succeed. However, basing projections on the past may be problematic given that, until Iran's Bushehr NPP came on line, there had been no new entrants in decades (Jewell 2011).

There are conspicuous differences between aspirants and existing nuclear energy states. Although many developing states are interested in nuclear energy, including several in Africa and the Middle East, these types of states are not well represented among those with nuclear power. Most aspirants considered credible contenders are either large emerging-market countries (e.g., Indonesia or Turkey) or small but fast-growing nations (e.g., Vietnam).

On the surface, it seems obvious why many aspirants would covet nuclear power. Some (e.g., Bangladesh, the Philippines, and Indonesia) stand out as having low rates of access to electricity and negligible consumption per capita.⁶ Vietnam must build considerable capacity to sustain its rapid economic growth, and a fifth of the population has no access to electricity (UNDP 2007). Adding capacity on a large scale to meet rising demand while sustaining economic growth appears to be especially attractive for nations with similar profiles.

Notwithstanding generic motivations, the variation across existing and aspiring nuclear energy states presents challenges for systematic analysis. Part of the problem stems from ambiguity about basic requirements for building a nuclear power plant. Although some infrastructural and financial elements are straightforward—such as adequate water for cooling, an electrical grid of sufficient installed capacity to absorb the addition of a large-wattage nuclear plant, a highly specialized workforce (or the resources to attract such a workforce from abroad), and the ability to raise or borrow billions of U.S. dollars—there are less tangible requirements, including the political will, public support, and institutional capacity to shoulder the risks necessary to promote and sustain nuclear power development (Busby, Chapter 5 in this volume; Zhou 2010; Jewell 2011). Some aspirants more consistently display these requirements than others; and some of the requirements (such as grid limitations or public anxiety about nuclear energy) may only preclude aspirants that are otherwise long shots for NPP acquisition.

The diverse nuclear energy landscape poses particular challenges to individual case analysis. Although comparative studies generate critical insight into historical motivations and processes of nuclear energy acquisition for select states, they reveal little about patterns that can be generalized across cases (Poneman 1982; Jasper 1990; Zhou 2010). Similarly, the general attributes of nuclear energy states can be obscured by the idiosyncrasies of decision makers that may be crucial to historical cases but less readily transferable to contemporary aspirants.

Despite the challenges, and consistent with the global revival of interest in nuclear power, there has been a spate of research on the general patterns of nuclear energy acquisition that embrace alternative substantive and methodological approaches. Some studies identify how much nuclear energy production must expand to meet a specific goal—most notably climate change mitigation. One study, in particular, suggests that the sector must experience an almost threefold global increase by 2050 to serve as a wedge of a carbon-constraining strategy (Busby, Chapter 5 in this volume; MIT 2003, 2009). Even if attainable, this reveals little about historical trends, as only a few nations have imposed costs on carbon emissions, and they did so only recently.⁷

Other analyses are rooted in direct extrapolations from historic trends. These studies start from assumptions that issues that have traditionally plagued the sector will persist (Feiveson 2009; Squassoni 2009; Findlay 2010). Chief among these are financial costs (rising capital costs, cost overruns, and construction delays) and technological barriers (lack of market for small and medium-sized reactors or progress in more cost-effective designs). However, qualitative analyses of historic trends are problematic, because they neither systematically test the significance of specified drivers or impediments nor evaluate the idiosyncratic choices, characteristics, or technological advances associated with states that overcome inertia (Adamantiades and Kessides 2009).

Other studies examine prospective correlates of nuclear energy among states that possess a nuclear infrastructure. One study—which statistically tests, via a stepwise regression model, fourteen attributes across eighty-six states that either possess nuclear power or are considered to be candidates—concludes that factors such as size of coal reserves and fuel cycle capacity are negatively correlated with nuclear reliance; whereas international openness, democratic institutions, and energy insecurity are positively correlated (Nelson and Sprecher 2008; Nelson 2010). Although the study advances comparative analysis and offers insight into prospective common attributes across a range of states with nuclear intent, the methodological problems posed by the absence of a normally distributed dependent variable, a narrow data set that excludes states without nuclear capabilities or intentions, and neglect of large-scale temporal variation, give pause to its preliminary conclusions. Another paper makes benchmark comparisons between aspirant states and nuclear energy states at the time of adoption across a range of technical, economic, political, and energy security variables. As mentioned earlier, one problem with this approach (as the author readily admits) is that it is not clear that historical benchmarks are good indicators, given changes with respect to financing conditions, nuclear safety accidents, and developing technologies. This paper also did not consider strategic or security motivations; rather, it assumed the drivers to be in the realm of economics and energy security (i.e., driven by energy demand and diversification) for all aspirants (Jewell 2011). That said, both studies represent important advances at systematic analysis and suggest the promise and new directions for large-N statistical studies of the correlates of nuclear energy.

The dataset for our analysis, which is designed to capture variation in nuclear power development across time and between countries, includes 150 states for which consistent data was available for the period 1950 to 2001. Because this period exhibited distinct phases of NPP construction, we also analyze two time periods within it: 1950–1980 and 1981–2001.⁸ Figure 1.1 depicts the breakdown of NPP starts, defined as construction beginning on an individual plant, by decade. Roughly 94 percent of the NPP builders entered the field from 1950 to 1980, and nearly 75 percent of NPP starts occurred during that period. In contrast, the 1980s and 1990s were marked by a dearth of nuclear plant construction in all but a few countries. This is generally ascribed to a combination of long-wave construction cycles, reaction to nuclear accidents, and an escalation of capital costs (Farber 1991). Accordingly, two runs of the model were conducted with dummy variables capturing the before and after periods of Three Mile Island (TMI) and Chernobyl.

Following a standard scholarly practice, we used logistic regression analysis to estimate the relationships between independent variables and nuclear power development. Logistic regression is especially appropriate, given the structure of the dependent variable (discussed below) and the expected nonlinear relationship between the independent variables and NPP starts.

The dependent variable, referred to as NPP starts, consists of a dummy that is coded 1 if, in the country-year in question, the state initiated construction of one or more NPPs, and o otherwise. A dichotomous dependent variable is employed to discern common characteristics among states that decide to construct NPPs, as opposed to determining the intensity of reliance on nuclear power. The 1950–2001 period captures the earliest commercial NPP starts. The values are determined from data in the International Atomic Energy Agency's (IAEA's) Power Reactor Information System (PRIS), with the exception of a few NPPs that never came on line but are included in the analysis (Diaz-Balart 1990; IAEA 2010a; Nuclear Engineering International 2010). Data on research reactors or those used for purposes other than electricity generation for public consumption are not included, because the purpose here is to understand commercial nuclear energy development. To control for temporal dependence in construction start years, four control variables are employed that capture the time without a construction event and three cubic splines (Beck, Katz, and Tucker 1998).

The independent variables account for strategic, economic, and political factors that are believed to influence states' pursuit of nuclear energy. Below, we discuss the covariates in greater detail and review the empirical findings, which are displayed in Table 1.1.

The first set of independent variables relates to nuclear weapons hedging. The hypothesis is that states in high-conflict environments will be more likely to build NPPs as a means to a nuclear weapons breakout option. Although few states have developed nuclear weapons compared to those that have pursued nuclear energy, states nonetheless can be expected to develop capabilities that would allow them to balance against threats or adverse power alignments in short order.⁹ To build a nuclear weapon, a state must either secure foreign fissile material or produce it in indigenous fuel cycle facilities. Development of fuel cycle facilities sans a domestic power plant fleet can betray a state's offensive rather than peaceful motivations, thus accentuating perceived security dilemmas (Beardsley and Asal, Chapter 11 in this volume).¹⁰

Following the example of Singh and Way (2004), two variables capture the degree of conflict and rivalry exhibited by a country during the period. A five-year moving average of militarized interstate disputes captures the number of conflicts in which a country was engaged without inflating the significance of episodic bouts. The data comes from Version 3.0 of the Correlates of War project's Militarized Interstate Dispute (MID) dataset (Ghosen, Palmer, and Bremer 2004). We include only those conflicts rated two or higher on the hostility index to capture actual militarized disputes.¹¹ A second variable captures the effect of long-term rivalries that may escalate into “hot” conflicts (Klein, Goertz, and Diehl 2006).

The model indicates that there is no significant relationship between international conflict or rivalry and the construction of nuclear reactors. Neither our hypothesis nor the counterhypothesis that a conflict-prone country is not conducive to the economic development or stability required to construct a commercial nuclear power infrastructure was supported. Accordingly, states that build a nuclear energy program in response to a high-threat environment should be considered the exception not the rule. Notwithstanding apparent security concerns and the unfavorable economies of scale associated with its nuclear program, any strategic hedging motive behind Iran's pursuit of commercial nuclear energy, therefore, seems especially anomalous (Sciolino 2006; Cole 2009). Indeed, the behavior of other countries is inconsistent with the hedging motive. The UAE, which contracted with a South Korean consortium in December 2009 to construct NPPs, vowed not to develop sensitive fuel cycle facilities, as discussed by Macfarlane and Stulberg (Chapters 2 and 4 in this volume).

As illustrated in Table 1.1, however, enduring rivalries appeared significant at a 90 percent confidence interval during 1981–2001, the “dark ages” of nuclear power generation. This might signal that hedging was an important driver for those few states that continued building during the lull, but this conclusion should be treated with caution given the high national concentration of NPP starts. Of the five countries that accounted for 70 percent of the NPP starts during the period (Japan, South Korea, China, France, and Russia), only France was free of enduring rivalries. States that were both engaged in NPP starts and experienced enduring rivalries constituted only 6.3 percent of the enduring rivalry country-years from 1981 to 2001.

The second key independent variable we test is energy security. Although the concept has many definitions and dimensions, the primary focus in the literature is on a state's self-sufficiency or dependence on foreign supply (Sovacool and Brown 2009). We measure this variable in terms of the difference between total primary domestic energy production and total primary energy consumption per capita (Banks 2010; EIA 2010a). Although this captures the overall degree to which a state is import dependent, it does not differentiate deficits in the electricity sector that can be attenuated by nuclear production from those in the transportation sector that currently cannot be efficiently redressed by greater reliance on nuclear energy. However, while many states today (after numerous oil shocks) cannot substitute nuclear energy to reduce oil import dependency, that was not the case over much of the study period (Toth and Rogner 2006; Lee and Chiu 2011). Also, defining energy security strictly in terms of electricity generation overlooks the fact that some states can cheaply import power to balance domestic deficits, while others lack this capacity (due to strategic and geographical challenges). Analyzing only electricity rates, therefore, obfuscates the degrees to which states can substitute different forms or sources of energy (Jewell 2011).

Nuclear energy is expected to be most popular among states that depend on foreign energy supply. The rationale for this hypothesis derives from the characteristics of nuclear fuel that render its supply chain less vulnerable than in the hydrocarbon sector. First, most nuclear plants do not frequently refuel. A typical light water reactor trades out one-third of its fuel every eighteen months. Second, the high density of nuclear fuel makes it feasible to store multiple loads on site (Hore-Lacy 2006).¹² Whereas brief delays in fossil fuel deliveries can wreak economic havoc via the price of electricity as well as inflating strategic insecurities, nuclear fuel supply is significantly less vulnerable to short-term disruptions; and, because fuel costs are a small portion of nuclear-generated power costs, electricity prices are less affected by uranium price volatility than by fossil fuel price volatility.

Evidence from the model supports the hypothesis that states that are more dependent on foreign sources of energy are more likely to build NPPs. This is consistent for the overall model, as well across the 1950–1980 and 1981–2001 subsets. The energy security motive is exemplified by Japan and South Korea, two countries with limited indigenous energy supplies and intense energy demand that have relied on nuclear energy to redress deficits between national production and consumption. Tokyo's white paper on nuclear energy policy, written before the Fukushima Daiichi accident, states that: “The first priority in Japanese energy supply policy is securing the steady supply of energy necessary to support the lives of the people” (Atomic Energy Commission Japan 2000). However, some states, such as the United States, experienced rising import dependency during a period when they ceased building new reactors. This may reflect a more complex dynamic associated with energy surpluses or deficits, such as the proximity to or availability of alternative supply infrastructures.

The model reveals that the energy security variable is significant only at a 90 percent confidence interval during the first period, while significant at a 95 percent interval during the second period. This is not surprising, for two reasons. First, especially during the early period, there were prominent cases of large-scale energy surplus states (e.g., the USSR, Canada, and Iran¹³) that pursued nuclear energy. Second, during the later period, Japan and South Korea were building at a substantial rate while most others, including the highly import-dependent United States, had slowed down or stopped constructing NPPs. This supports the notion that energy dependence is not merely determined by the national aggregate energy deficit or surplus, and that physical constraints on pipeline and transmission line construction may indeed compound the significance of the national energy security equation.

There are reasons to believe that economics may drive the pursuit of nuclear energy. Most important, economic factors can directly shape the national demand for energy. Almost all nuclear energy states were among the largest economies in the world at the time they adopted nuclear power (e.g., although Armenia is a small economy, its nuclear program was initiated while it was part of the Soviet Union). Curiously, there is no similar conspicuous relationship tied to per capita gross domestic product (GDP). Even a cursory observation reveals that some of the highest per capita GDP states are small and do not display the magnitude of demand conducive to building nuclear plants, whereas a number of large emerging-market countries (with relatively small per capita GDPs) are increasingly major players in nuclear power development. This suggests that the scale of economic activity is more important than the standard of living in decisions to build NPPs. Because commercially available NPPs are large-scale plants (on the order of 1 gigawatt), there can be technical and economic barriers to small economies seeking to embrace nuclear energy. While there is a clustering of nuclear energy states among the highest GDP states, there are countries that developed nuclear energy which had relatively low GDPs at the time of adoption (Pakistan, Bulgaria, Finland, Hungary, and Romania), as well as aspirants that failed to develop nuclear power despite having GDPs of a similar magnitude to the nuclear energy states as a whole (Turkey, Poland, and Indonesia). However, even the poorest nuclear energy states were among the richest forty-five national economies when they adopted nuclear energy. In light of this situation, we include GDP as a measure of overall scale of economic activity within a country to capture the demand for energy, and expect that high GDP should correlate positively with NPP starts.

Economic growth is another common indicator of energy demand. While the specific causal connection remains contested, the correlation between energy consumption and growth remains inviolable (Ockwell 2008). It is expected that fast-growing economies should be more likely to construct NPPs in order to keep pace with rapidly rising demand. The East Asia nuclear energy states (Japan, South Korea, China, and Taiwan) are particularly emblematic of this argument. However, there are numerous examples of nuclear energy states with low growth rates, particularly in East-Central and Western Europe. There also are potential countervailing macroeconomic effects, in that wealthy states tend to experience slow growth rates relative to developing states.

We also include economic openness to test the assumption that a state that is more fully integrated into the global economy will have a higher probability of successfully constructing nuclear plants than will isolated states (Comin and Hobijin 2003). The historical record lends tentative credence to this logic, as only about a quarter of nuclear energy states developed their first plant indigenously (WNA 2010a). Almost none of the leading nuclear energy aspirants are capable of building plants without foreign assistance.

We use several variables to test the relationship between economic factors and NPP starts. The base GDP figure is a constant year dollar presented in purchasing power parity terms (i.e., Geary–Khamis international dollars) to avoid depressing the financial strength of emerging-market countries (Maddison 2009). The use of constant year dollars minimizes the confounding effects of countries with high rates of inflation or deflation. The raw GDP figure is transformed by taking the natural logarithm to mitigate the extreme skew of the variable. The economic growth rate is derived from the Geary–Khamis dollar GDP figures to measure national income. The standard measure of economic openness ([exports + imports]/GDP) is used in constant year dollar form (Heston, Summers, and Aten 2009; Banks 2010). However, we appreciate that the scale of some domestic economies, such as the United States and Japan, could make them appear to be closed when, in fact, those states are active traders.

The model demonstrates that the national income variable is significant and of the anticipated sign for both the entire 1950–2001 period and the two subset periods at a p value less than 0.001. This supports the hypothesis that the size of a country's economy is positively correlated to the likelihood of building nuclear power. However, neither economic growth nor openness is significant. It turns out that the average growth rate for builders of nuclear plants is 3.98 percent—almost identical to the overall average growth rate of 3.76 percent, given a standard deviation on the order of 6.00 percent. As mentioned, there are problems with using total trade over GDP as a measure of openness, because the very largest economies tend to look closed despite conducting sizable trade in absolute terms. Similarly, major transshipment points, such as Singapore, and small developing nations, such as Ghana, appear as open, suggesting that more research may be required to fully assess the significance of international openness.

Historically, nuclear energy states reflect varying governance types. One study noted a positive correlation between democracy and nuclear power reliance (Nelson and Sprecher 2008). Another study found that democracies are more likely to be responsive to nuclear accidents and, thus, less likely to build nuclear capacity in the wake of such events (Fuhrmann 2012). The relationship between regime type and preferences for nuclear energy, however, may vary over time. After all, the United States dominated the field in its early decades, but China is home to over one-third of the NPPs under construction. Still other studies disaggregate relevant governance issues beyond a straightforward autocratic-democratic index. Regime duration, for example, may capture the stability needed to promote nuclear energy. Similarly, one qualitative study compared the influence of U.S. and French regulatory systems on the development of nuclear power industries and found that the combination of a concentrated executive authority, a weak judiciary, and heavy reliance on bureaucratic expertise rendered France significantly more conducive to the development of a national nuclear sector (Delmas and Heiman 2001). Still another study defined institutional capacity in terms of stability and reliability of regulatory procedures that bear directly on the confidence in government policies to attract private investment (Jewell 2011).

We use data on both regime type and duration from the Polity IV dataset of Marshall and Jaggers (2009). The polity measure, consisting of a twenty-one-point scale that ranges from +10 for highly democratic states to −10 for highly autocratic states, forms the basis of the governance variable. This scale is derived from five subcomponent scores dealing with executive recruitment competitiveness, executive recruitment openness, constraints on the executive, regulation of political participation, and competitiveness of political participation. Regime duration begins with o during a transition year and increases by 1 for each year thereafter.

Neither regime type nor stability of governance is significant in either the overall model or the temporal subsets. This time-series–cross-section approach offers no support for a hypothesized link between democratic regimes and NPP development. Also, while most nuclear energy states appeared to be stable over the period they were building NPPs, there were several examples of nuclear energy states that built plants during periods of intense political transition, including Argentina, Brazil, Mexico, Pakistan, and South Korea. This also is generally consistent with other large-N studies that identified a diversity of regime types associated with past and present nuclear energy states (Jewell 2011).

In order to control for the effects of major nuclear disasters, two runs were conducted with an additional variable each. These crude analyses consist of a dummy coded o for every country-year before the event (TMI in the first run and Chernobyl in the second) and 1 for country-years thereafter. Three Mile Island (1979) and Chernobyl (1986) were the most notable accidents during the study period, and there has been much discussion of the impact of each of these events on the nuclear industry. In both runs, this variable was significant. However, the limited information content of the variable, contending alternative explanations (e.g., rising costs that may not be linked to these disasters), and the fact that construction start has a varying and sometimes substantial lag from the decision to build an NPP are suggestive of caveats to these conclusions.

Although the statistical model reveals the historical significance of economic size, energy insecurity, and the effect of accidents as correlates of nuclear power development, as well as suggesting a lack of support for other hypothesized correlates, it provides only limited insight into the causal logic of nuclear energy decision making. The basic challenges of this large-N study are compounded by the high concentration of NPP construction starts (almost 25 percent of countries included in the study have NPP starts during the study years, but over 70 percent of events are observed in just ten states) and the infrequency of events (NPP starts occurred in less than 5 percent of country-year observations).

More conspicuous, the model obfuscates the divergent patterns of NPP construction (depicted in Figure 1.2) that affected historical waves of acquisition among archetypal nuclear energy states—the United States, France, and Japan—and the implications these hold for aspirants. The United States, for example, engaged in a massive buildup of nuclear plants in the 1960s and 1970s, and then entered a period of protracted dormancy. France displayed a similar pattern during the period but with different long-run implications. France receives 75 percent of its electricity from the nuclear sector, which cannot meet all electricity needs owing to technological constraints; and, therefore, having reached its maximum nuclear capacity, France experienced a lull in new construction (WNA 2010b). Accordingly, while France's pattern of growth and stagnation in NPP construction is seemingly tied to the life cycle of nuclear plants, the similar pattern exhibited by the United States can more readily be attributed to the nuclear economics of increasing capital costs influenced by accidents, poor returns on investments, market uncertainties, and escalating delays and cost overruns (Ellis and Zimmerman 1983; Sommers 1980).¹⁴ In contrast to both of these cases, Japan displays a steady pattern of long-term NPP construction that has only recently wavered in the wake of the 2011 earthquake and tsunami.

The model offers only partial insight into several possible inflection points for a nuclear energy revival. First, while it supports the findings of other studies about the significance of energy insecurity as a motivation for nuclear power generation (Jewell 2011; Fuhrmann 2012), it suggests the need for delving deeper into understanding how states make trade-offs among options for reducing energy deficits, such as by imports via regional pipelines or transmission infrastructures. It is plausible that the unique technical and geographic vulnerabilities of Japan and South Korea shaped their respective patterns of NPP construction when virtually all other states had stopped new development.

Second, the model highlights the limitations for understanding NPP starts associated with traditional measures of autocratic and democratic regime types (see Montgomery, Chapter 7 in this volume; Hymans 2008). Study of the data reveals cases in which similarly rated states produce significantly divergent outcomes. For example, the United States and Japan, both highly rated democracies, have embraced different trajectories for nuclear energy growth from the 1980s until the 2010s. This suggests either the relative insignificance of the variable or that governance is not completely captured by the model. The latter may be the case, as we know that relevant decision-making and regulatory systems pertaining to nuclear power vary greatly across democratic regimes. In some cases, the state is the sole relevant decision maker, while in others private firms or public-private corporations are major players. Contrasting French (majority publicly-owned utility) and German (private utility market) approaches to nuclear power suggests that this may be a promising line of inquiry. Yet while distinction between public and private ownership may be a necessary component of understanding the role of governance, it alone is not likely to be sufficient to account for variation, since both the United States and Japan have maintained primarily private utilities. As some scholars suggest, China's centralized government and state-driven economy may be well-positioned to transcend otherwise paralyzing political interests and public ambivalence to spearhead a massive national buildup of nuclear energy (Zhou 2010).

Rather, what may be left out of the model and other conventional political indicators is how specific governance factors relate to risk-taking propensities, especially with respect to overcoming public anxiety and financing nuclear plant construction. Financing NPPs involves more capital than most private utilities hold as cash or can collateralize, or for which a single bank can diversify against a default. As one study indicates, the elimination of the excess risk premium for nuclear energy over coal or natural gas alone would make the sector economically competitive (MIT 2009). This suggests that different national risk perceptions and related mitigation strategies for nuclear energy are potentially important inflection points. Although the United States has undertaken initiatives to redress financial risks via loan guarantees and liability caps, these measures have not spurred significant growth in the nuclear sector. However, China has used loan guarantees, financial aid to the state-owned enterprises involved in the nuclear sector, and an insurance pool to successfully spur a buildup of nuclear energy at a rate not seen since the early 1970s (Zhou 2010). The majority of nuclear energy states have financed NPP construction directly with public funds so as to broadly distribute financial risks. Few states, such as South Korea, rely primarily on private equity or debt to initiate and sustain commercial nuclear power development, though the United States and Germany employed this method when they were building (IAEA 2008).¹⁵ Accordingly, part of the difficulty of capturing risk in the model is that both the definition and policy responses are subjectively determined. As reviewed by Macfarlane (Chapter 2 in this volume), the risk perception of policy makers, experts, the general public, and subnational actors can be radically different with direct implications for policy. Identifying how and why specific national institutional factors may accentuate or mitigate such risk perceptions, therefore, may provide crucial addenda to the literature on governance types and nuclear energy correlates (Hamalainen 1991; Sjoberg and Drottz-Sjoberg 2008).

The model also does not capture several emerging variables. Excluded from the analysis is the impact of climate change, owing to the relatively new focus on the issue as a tipping point for nuclear energy (Busby, Chapter 5 in this volume). China, for example, has set a goal to have carbon dioxide emissions 40–45 percent below 2005 levels by 2020, and plans to do this without curbing its impressive growth (Zhou 2010). There are other pollutants that raised concerns over part of the study period, including sulfur dioxide and nitrogen oxide, and regulation or taxation of them may have strengthened the relative appeal of nuclear power. One study, for example, showed support for the hypothesis that the Clean Air Act increased the attractiveness of nuclear power for the United States during the 1970s (Ellis and Zimmerman 1983). However, the lack of consistent international time-series–cross-section data for these pollutants makes it difficult to include them in a statistical model.

Similarly, the model does not capture the prestige value of nuclear energy, to the extent it exists. This variable is particularly difficult to include in a quantitative model because of its qualitative nature and the lack of a readily apparent proxy. The use of alliance portfolios as a surrogate measure, under the assumption that prestige-deficient powers likely would be dissatisfied with the status quo and would choose their alliances accordingly, provided mixed results for discerning the correlates of nuclear weapons pursuit and acquisition (Singh and Way 2004). Another study noted that there were statements by members of the leadership of six countries (the United States, the United Kingdom, France, the USSR, Germany, and Canada) that hyped the prestige associated with a national nuclear energy research and development program. This study also acknowledged that the elusiveness of the subject made it difficult to discern the relative impact of prestige on development of such programs (DeLeon 1980). Thus, without deeper qualitative and case analysis of the national prominence of prestige, it will be difficult to uncover its relative importance via statistical study of the correlates of nuclear power generation.

As one can see in Figure 1.1 and, less broadly, in Figure 1.2, nuclear energy experienced a clear tipping point in the 1980s. The nuclear accidents at Three Mile Island and Chernobyl are often cited as key causal factors in the collapse of the global nuclear energy industry. Although this conclusion may seem reasonable—based on the timing of the collapse, the literature on the economic effects of these events (e.g., Kalra, Henderson, and Raines 1993), and apparent sea changes in countries such as Egypt—the fortunate rarity of large-scale nuclear accidents and the evidence that nuclear energy was losing appeal before these events should encourage caution about simplistically concluding that the accidents were the dominant cause of the collapse.¹⁶ There is more to learn about the effects of nuclear accidents, and undoubtedly future studies addressing the Fukushima Daiichi disaster will provide further illumination of the significance of this relationship.

Forecasting nuclear energy scenarios is challenging under the best of circumstances, owing to the complexity of the landscape. It is especially problematic in the absence of sound deductive reasoning about decision making and critical testing of alternative scenarios, which are premature given the state of knowledge about the drivers of nuclear energy (Bueno de Mesquita 1984). However, both the statistical model and respective caveats together identify important conditions and safe bets for discerning characteristics of alternative trajectories for global nuclear energy growth.

First, with projections of rapidly increasing electricity consumption rates, the nuclear energy sector must grow significantly if only to maintain aggregate shares in global and national energy portfolios. Before the tsunami in Japan, the Energy Information Administration projected a 70 percent increase in worldwide use of nuclear between 2012 and 2035 to meet an approximately 50 percent increase in global energy consumption over the period (EIA 2010b). Our analysis suggests that signs of stagnation versus an expansion of nuclear power should be explored primarily in the behavior of the current nuclear energy states. Those with large economies and mounting gaps between domestic energy production and consumption—particularly those with few geographic or infrastructural alternatives for cheap imports—will be most capable of driving an expansion. For this reason, any evidence of stagnation or decline in key states, such as China, the United States, South Korea, and France, would likely be ominous for the future of a nuclear energy expansion. Japan will prove to be a key test, given its recent talk of discontinuing nuclear power and the temporary shutdown of almost all of its NPPs. China is engaged in a major nuclear power building boom, with over 25 gigawatts-electric (GWe) worth of NPP construction in the works, that looks similar to the U.S. position in the 1960s. Until the accident at Fukushima Daiichi, Japan and South Korea seemed poised to keep building nuclear power plants until they reach the technically feasible limits of nuclear capacity, to replace aging plants, and to make nuclear part of their strategy for meeting growing energy needs. Given France's enthusiasm for nuclear energy, it, too, is likely to maintain as much nuclear capacity as is technically feasible. However, the United States, which alone currently comprises over 25 percent of global installed nuclear capacity, presents a more ambiguous picture. Because the average operational U.S. NPP is over thirty years old, the United States will have to engage in substantial NPP starts just to replace plants that are nearing the end of their life cycles. To date, progress has been slow.¹⁷

The model leads us to believe that states with substantial and rising energy demands and energy security concerns will provide indicators of the extent to which states will be able to deepen their reliance on nuclear power. Accordingly, the first place to look for a nuclear resurgence will be in the decisions taken by states such as China, India, Russia, South Korea, and the United States. The important role of the large emerging-market countries in the expansion of nuclear power seems to be playing out. The BRIC (Brazil, Russia, India, and China) countries account for over 68 percent of the nuclear plants under construction, and adding South Korea raises that figure to almost 75 percent. However, some of these countries appear to more intensely participate in a deepening than others. India, for example, with 4.8 GWe of nuclear installed capacity in the pipeline, has had to add almost 7.0 GWe per year over the past few years to meet its energy needs (Busby, Chapter 5 in this volume; EIA 2010a). Russia provides an interesting case as a major net exporter of energy that is actively engaged in developing nuclear energy. With over 8 GWe in the works, Russia is the second largest NPP-building state behind China; though, unlike China, it has a reactor fleet almost as old, on average, as that of the United States.

The model also illuminates prospective attributes associated with a potential broadening of the nuclear energy field. Italy, for example, would have been considered the most likely to join (once again) the nuclear energy club before a June 2011 referendum that succeeded in blocking development of nuclear power. Italy is an advanced industrialized country with a history of operating nuclear power plants, thus avoiding the first-of-kind challenges that hound nuclear energy aspirants. Italy's decision not to pursue nuclear energy in the wake of the Japanese accident could be said to yield important insight into national risk assessment for nuclear power development. Other aspirants with an economic mass on par with nuclear energy states include Turkey, Indonesia, and Saudi Arabia. These states warrant close examination to discern the prospects for a broadening nuclear energy expansion.¹⁸

While the model predicts that states with an energy surplus are less likely to develop nuclear energy, several energy-producer states should not be overlooked. From the group noted above, only Turkey is not a significant net energy producer. Historically, a number of large-scale energy producers have built NPPs, including the Soviet Union/Russia, Canada, Mexico, and most recently Iran. Furthermore, Qatar, Kuwait, Saudi Arabia, and the UAE are among the countries with the highest per capita petroleum consumption in the world (significantly above the United States and Canada), and they are all Kyoto Protocol members. These countries still rely on petroleum-based power generation for which they can substitute nuclear energy in order both to facilitate meeting respective emission reduction targets and to free up hydrocarbons for global export. This factor could account for the UAE's progress toward nuclear power and would suggest that Kuwait and Saudi Arabia may not be far behind.

The specific contours of a global expansion undoubtedly will be influenced by factors that are uncertain and idiosyncratic and will not necessarily reflect historical trends. While there are problems of both collective action and scale inherent in reducing greenhouse gas emissions (Busby, Chapter 5 in this volume), this may likely be a motivator for some states. However, reducing these emissions will require some form of cost on carbon output. Furthermore and notwithstanding successive and recent events in Japan, nuclear power has had an impressive safety record, and the IAEA points to safety as the single most important factor for improving the viability of NPP financing worldwide (IAEA 2009). This speaks to the possibility of technical solutions, such as smaller reactors that may be more cost-effective (especially for smaller aspirants) and inspire greater confidence. Irrespective of how these unknowns unfold, it is clear that before we can forecast the precise trajectory of commercial nuclear energy, we need a more systematic and richer understanding of both past attributes and drivers of the sector.