ESTADO DEL ARTE / CULTURA

The United States took important steps to address the problem of inadequate accrual for NDTs in the late 1980s. Prior to that time, few utilities made any provision to accrue for decommissioning at all (MacKerron 1989: 107), and whatever funds had accrued could be held inside the company as bookkeeping entries even while the actual cash was spent to cover other expenses. Thus there was no guarantee that the money would actually be there when needed. In 1988, the NRC promul- gated rules that required such funds to be held in independent trusts. This rule-making addressed the most critical risk: to prevent decommissioning funds from being raided or lost in a bankruptcy. Comparable procedures have not been used in all countries. The United Kingdom, for example, accrued billions of pounds to decommission nuclear infrastructure. But poor financial controls resulted in much of the decommissioning funds being used to support operating expenses, and little remained for the intended purpose (Schneider et al. 2009: 83).

Although the use of external trusts solved one important risk of decommissioning funding, uncertainty remained as to whether reactor own- ers would have enough time to accrue the funds needed to finance the plant shutdown. Two issues

drove these concerns. First, there was little actual experience with reactor decommissioning to guide cost estimates, and financial models were generat- ing rapidly escalating values. Second, there was a well-founded concern, especially after deregulation, that the power produced by nuclear reactors was so expensive that many would need to shut down prior to the end of their license life, with accumu- lated trust funds falling well short of what would be needed to decommission them.96

Actual decommissioning costs for modern light-water reactors that have run for decades are still largely unknown. However, a number of other market shifts have reduced the risks of widespread shortfalls. First, stranded cost rules eliminated many of the capital recovery pressures for exist- ing reactors, allowing them to remain operating in competitive markets. This write-off effectively lengthened by decades the time period over which the reactors would be able to accrue the needed funds. Second, higher capacity factors at these reactors brought down costs per kWh. Coupled with rising energy prices, operating margins at the reactors improved, consequently eliminat- ing most discussions of premature closure. Third, NRC license extensions have been granted on a regular basis, extending the time frame—up to an additional 20 years—over which decommissioning funds can be accrued.97

Rule changes that allowed the trusts to invest in a diversified asset base, rather than just low- yielding credit instruments, also boosted the ability of portfolio growth to help fund decommissioning costs over the long term.98 _{According to Duff &} Phelps (an investment firm that invests on behalf of NDTs), equity constitutes the majority of NDT 96 A third potential risk comes from inadequate accruals by owners during the years in which a plant is operational. The GAO notes that federal licensees must only provide a “statement of intent” that decommissioning funds would be supplied when necessary (GAO 2003: 5). This does not seem to have been a problem, however, as real funding has taken place.

97 As of the end of June 2009, the NRC had granted license extensions for 54 of the 104 operating U.S. reactors.

98 At the inception of NDTs in 1984, allowable investments were restricted to treasuries, municipal bonds, and bank certificates of deposit. This policy was shifted to a “prudent investor” standard in 1993, likely to enable a higher rate of growth in assets. As of 2007, investor-owned utilities had a median equity allocation in their trust funds of 55 percent, while public power entities were normally restricted from investing in equities at all in their trust funds (Duff & Phelps 2008a: 4, 7). Although the recent market pullback has caused substantial reductions in the value of NDT assets, a broader mix of assets has historically boosted risk-adjusted portfolio growth rates over the long term.

holdings.99 _{While equities tend to be more volatile,} they have demonstrated a higher return over the long term than bonds or cash.

Much uncertainty remains, however. Short- term losses are one worry; NDTs lost more than 20 percent of their value between December 2007 and December 2008—a drop of more than $9 billion after new contributions (Duff & Phelps 2009b: 4). In June 2009, the NRC notified own- ers of 26 reactors (25 percent of the operating U.S. fleet) about decommissioning fund shortfalls. The deficits ranged from $12 million to $204 million (Burgdorfher 2009). Annual funding rates also were well down, at $562 million for 2008 versus $1 bil- lion or more for 2001 to 2004. However, this shift may be due in part to license extensions that length- en the period of accrual (Duff & Phelps 2009b: 4). An analysis performed a few years ago by the GAO illustrates the difficulty of trying to assess funding adequacy for a highly uncertain cost many years in the future. The agency’s survey of 222 individual utility trust funds and 99 utility own- ers (Williams 2007: 1052) was quite sensitive to asset performance assumptions. With pessimistic assumptions, 181 of 222 funds analyzed were below the benchmarked need.100 _{With the most} optimistic assumptions, only 18 were below the benchmark (Williams 2007: 1079).

The longer time frame for accrual of decom- missioning funds and the general long-term upward trends in markets suggest that investment performance will return to historical norms that may be sustained over time. Indeed, the stabiliza- tion of capital markets during 2009 and 2010 support such a conclusion. Of greater concern, however, is whether the ultimate costs of decom- missioning will be higher than current assump- tions, resulting in shortfalls at a time when there

is no opportunity to recover additional funds from ratepayers. There is much to suggest that there may be problems:

• Decommissioning-cost assumptions are vola- tile year to year. Data collected by Duff & Phelps show a worrying year-to-year volatility in the average estimated cost to decommission reactors as well as variation by type of owner. The average decommissioning cost per kWe for investor-owned reactors as of December 31, 2007 was $594, dropping to $550 a year later. For publicly owned reactors, the trend was the opposite, with average cost estimates rising from $470/kWe in 2007 to $510/kWe in 2008 (Duff & Phelps 2009a, 2007). While it is not possible to identify the causes of this variance, the shifts indicate that there is no consensus on costs and that other factors (e.g., available cash from operations) may influence assumptions about decommissioning.

• Decommissioning-cost estimates continue to escalate well above the rate of inflation. This trend has increased the target NDT accrual by more than $4.6 billion over the past two years alone (Gram and Bass 2009).

• Investment gains from delaying decommis- sioning are offset by the loss of specific knowl- edge. While a longer period to accrue funds for decommissioning—even in the post-closure period—has traditionally been considered a plus, Schneider notes that this assumption is not always valid. For example, “the benefit of radioactive decay in the case of delayed decom- missioning is offset by the knowledge loss” about the facility and the skills needed to prop- erly decommission it. Some countries, such as France, require immediate reactor decommis- sioning (Schneider 2009).

99 The composition of NDT funds at present can be approximated based on a 60 percent share in the Standard & Poor’s 500 and the remaining funds allocated similarly to Barclays Capital Aggregate Index (Duff and Phelps 2009b).

100 The number of trust funds can exceed the number of reactors for two reasons. First, many reactors are fractionally owned, and each owner requires its own segregated trust fund. Second, the original legislation allowed for either a qualifying or a nonqualifying trust, and over time, one of each might exist for a given ownership position. Recent changes in the law have made trust-fund consolidation easier, thus reducing this problem.

Even assuming that utilities accurately predict the cost to decommission a facility, if the growth in NDT assets is slower than the rate of decom- missioning-cost escalation, a shortfall will be likely. Between 1986 and 2008, the NRC’s minimum cost estimates for decommissioning grew at a com- pound annual rate averaging 7.8 percent (Duff & Phelps 2009a). Assuming that decommissioning assets were invested in a portfolio comprised of 60 percent Standard & Poor’s 500 and 40 percent Barclays Capital Aggregate Index (Duff & Phelps 2009b), NDT returns during that same period would have averaged roughly 9.5 percent for a net gain above cost escalation. However, while NDT earnings are taxed at a low rate, they are still taxed. Using a 20 percent federal rate and an incremen- tal 3 percent average state rate, after-tax returns for IOUs drop to 7.3 percent, about 0.5 percent lower than the escalation in expected decommis- sioning costs. IOU balances were $30 billion as of December 31, 2008 (Duff & Phelps 2009a), generating an annual deficit of $145 million. Because POUs are not taxed at all, investment returns keep up with need, generating a surplus over cost escalation of about $85 million per year, even assuming a lower equity share (and returns) for POU NDTs. The net shortfall between the two sectors is $60 million per year.

Two points are important here. First, even if industry-wide calculations suggest little net deficit, the shortfalls could be quite large for spe- cific plants. Second, these shortfalls could build greatly because of the lost compounding of invest- ment earnings year to year if returns lag cost escalation, and small changes in the yield gap that could result in much larger (or smaller) deficits over time.