Stranded Costs: Is the Market Paying Attention? (A Look at Market-to-Book Ratios)

Fortnightly Magazine - May 15 1996

Investors are taking stock

of utility exposure to price competition.The utility trade press and even the general financial press have featured the views of regulators, utility executives, legislators, and various consumer advocates on the stranded-cost question. Stranded costs easily represent the most contentious issue facing the electric industry as it moves to an era of competition. The financial analysts have weighed in as well, from the fixed income as well as the equity side of the house.

But rarely does one find evidence of what utility shareholders are saying about stranded costs and competition. That's because they speak with their wallets and not with their mouths. Utility stock performance in 1995 provides evidence that equity investors are taking serious account of the exposure of utilities to stranded costs and competition. Those utilities most exposed saw their stocks trading at lower market-to-book ratios.

Just as electric producers compete for sales, analysts compete to develop methods to estimate exposure to stranded costs. In this case, we started with data on stranded investment contained in recent reports from Moody's and Standard and Poor's, and used that information to study the 1995 year-end, market-to-book (M/B) ratios of stock prices for some 69 utilities. We found that we could explain approximately 20 percent of the differences among the M/B ratios in terms of estimates of stranded costs. Moreover, when we considered differences across utilities in earned returns on equity, we could then explain fully half the variance among M/B ratios.

One added point might come as a surprise: We found we could not explain any additional variance in M/B ratios by considering incremental kilowatt-hour production costs, or the percent of power generating output sold to industrial customers.

The Market:

Timing is Everything

Currently, the power-supply component of retail rates lies significantly above new wholesale power prices, while embedded costs exceed marginal costs substantially. Contributing to these higher-than-market retail rates are a number of factors: large sunk investment in nuclear generating plants, decommissioning liabilities, costly purchased-power (QF) contracts, and various mandated social programs, such as demand-side management (DSM). Prices on new wholesale power contracts, on the other hand, have been falling due to increased competition, low fuel prices, and technological advances that make combined-cycle generation turbines competitive with larger baseload plants.

An end to the exclusivity of the utility franchise - permitting access to market-priced power - would tend to drive retail power rates down to a level that would create "stranded costs." Potential stranded costs may or may not be currently incorporated into rates. Stranded costs currently allowed in rates customarily include generation assets, long-term purchased power and fuel contracts, and mandated social programs. Stranded costs not currently in rates can include regulatory assets that have been deferred to mitigate short-term rate effects, or simply deferred costs of generation plant and DSM. Recovery of any costs above market may be jeopardized by the transition to a more competitive electric industry.

In addition to the big policy questions - whether some or all of these costs should be recovered, and what mechanisms should be used to recover them while minimizing anticompetitive effects - there is a major timing issue. How much time do we have - not only to answer these questions, but to implement solutions in an orderly way? Shareholders are already reacting to new market realities and discounting stocks with stranded-cost exposure. It remains to be seen how they might respond to specific stranded-cost phase-in plans in the states.

The Problem: Cost or Revenue?

Two recent reports by Moody's and S&P highlight the stranded-cost debate and present different methods of calculating the magnitude of the problem.

The Moody's report (released in August 1995) estimates potential stranded costs for 114 U.S. investor-owned utilities at $135 billion.1 The report adopted a method of calculating stranded costs that assumes a utility can be compensated in the market for both an energy and a capacity component. Thus, any margin earned from selling energy above variable cost is applied to the recovery of fixed costs. The report compares each company's capacity charge for covering fixed costs with the competitive market price for capacity, which is based on the marginal cost of the most expensive unit needed to satisfy expected peak demand in differing regions of the country. Stranded costs equal the difference between the margin over variable cost and the capacity charge for fixed costs.2

The S&P report (released in November 1995) focuses instead on a utility's potential revenue loss from retail competition, positing a severe as well as a reasonable case.3 The severe case assumes immediate direct access for all customers and no regulatory mechanism for the recovery of stranded costs. The reasonable case assumes immediate direct access for commercial and industrial users only and a 50-percent stranded-revenue recovery mechanism. S&P isolated revenues associated with generation and sales of electricity and the purchase and resale of power, excluding revenues and costs from transmission and distribution. For each customer class (i.e., residential, commercial, and industrial), S&P assumed a market-clearing price for generation sold, then compared that price to each utility's production costs, and then multiplied the difference by the 1992-94 average sales volume to determine each utility's potential lost revenue.

Our analysis covered a sample of the 69 utilities that were common to both studies. We used the Moody's estimates of stranded costs as a percent of equity for each utility, and S&P's reasonable and severe estimates of lost revenues as a percent of total revenues. In the select cases where a holding company owned more than one electric utility, we aggregated and standardized the estimates of stranded costs and lost revenues for the applicable utility subsidiaries.

Three Models: A Correlation

A simple, ordinary least-square (OLS) regression equation was used to determine whether stranded costs, as measured by the Moody's and the S&P reports, can explain variability in equity performance, as expressed by the M/B ratio.

We selected M/B ratios (a utility's stock price divided by its book value) as a useful variable to measure investor perceptions of the value of utility company's assets on December 31, 1995.4 Higher M/B values are associated, all else being equal, with greater expected value and profitability. Thus, we assumed in the study that a utility's M/B ratio should decline if shareholders doubt that regulators and/or the market will provide 100-percent recovery of stranded costs. (Conversely, utility-specific stranded costs should not affect the M/B ratio if investors perceive that regulators can guarantee 100-percent cost recovery.) A competitive environment that forces utility write-offs would strain earnings and reduce investor expectations of higher future dividend growth. This negative expectation should cause equity prices to decrease, thus causing a decrease in M/B ratios, all else equal. Therefore, investors who anticipate stranded costs in the future would expect the benefits of holding the stock to decrease.

Utility M/B ratios were regressed on three estimates of stranded cost. The first equation used Moody's estimates (standardized from the original report of stranded cost as a percentage of total book equity at risk). The second used S&P reasonable-case estimates (reported initially as lost revenue as a percentage of total revenues). The third used the S&P severe-case estimates (also expressed originally as lost revenue versus total revenues). Three separate equations were developed, one for each estimate of stranded costs.

The data indicated a statistically significant negative relationship between a utility's M/B ratio and the stranded-cost estimates.5,6 Regardless of which estimate of stranded costs was used, an increase in a utility's exposure to stranded costs was associated, on average, with a decrease in a utility's M/B ratio.

The first equation (Moody's estimates) revealed that a 10-percentage-point incremental increase in a utility's stranded cost as a percent of equity (e.g., from 50 to 60 percent) produced a 1.7-percentage-point decline in the M/B ratio. A 1-percentage-point incremental change (e.g., from 10 to 11 percent) in lost revenue exposure (S&P reasonable case) led to a 4.3-percentage-point decline. Finally, a 1-percentage-point change in revenues lost (S&P severe case) produced a 1.4-percentage-point decline in M/B ratio.

Interestingly, the three stranded-cost exposure estimates each explained approximately the same amount of M/B variability across the 69 utilities covered in the sample: 19.3, 19.1, and 20.6 percent, respectively (see Table 1).

Other Factors? Return on Equity

As seen above, stranded-cost estimates taken alone can explain one-fifth of the variation in M/B ratio among the 69 sample utilities. The next question becomes: What about the other 80 percent? Is it possible that other variables better explain the behavior of M/B ratios? In an attempt to capture some of the other 80 percent or to displace the stranded-cost explanation, we tested three other independent variables with each of the three stranded-cost estimates: 1) earned return on equity (ROE), 2) production costs, and 3) percent of load accounted for by industrial customers.7 Neither of the last two variables proved statistically helpful. However, when ROE was introduced alongside the stranded-cost estimates, we found we could then explain fully one-half of the variation in M/B ratio.

With these three new independent variables, added to each of the three model runs noted earlier, we assumed three hypotheses. First: ROE measures profitability so, all else being equal, higher ROE should be associated with a higher M/B ratio. Second: Production costs reflect a utility's ability to compete in a less-regulated market; thus, higher production costs, all else equal, should accompany a lower M/B ratio. Third: Percent of industrial load suggests how much of a utility's load may lie at risk in the early stages of competition; consequently, a lower percentage of industrial load, all else being equal, should attend a higher M/B ratio.

The model runs indicated that the three independent variables estimating stranded costs remained negatively associated with M/B ratios and remained statistically significant. Even after including the additional independent variables - ROE, production cost, and industrial load - the stranded-cost estimates remained

statistically significant.8 The only other variable besides the stranded-cost estimates in

the model to be statistically associated with M/B was ROE.

In the three multiple regression model runs, ROE was positively associated with M/B ratios and demonstrated high statistical significance. A 1-percentage-point incremental increase in ROE (from 10 to 11 percent, for example), led to approximately a 12-percentage-point increase in the M/B ratio for each of the three stranded-cost estimates. In combination with ROE, the Moody's estimate of stranded cost led to a 1-percentage-point downward change when there was a 10-percentage-point incremental increase in a utility's stranded cost as a percent of equity (such as from 50 to 60 percent). A 1-percentage-point change in lost revenue exposure in the S&P reasonable case (say, from 10 to 11 percent) led to a 3.5-percentage-point change in the M/B ratio. Finally, a 1-percentage-point change in the S&P severe case led to a 1-percentage-point change in the M/B ratio.

Running three multiple regressions substantially increased the ability to explain the variation in M/B ratios. As shown in Table 2, the multiple regression equations explain 50.7 (Moody's), 55.3 (S&P reasonable), and 54.8 percent (S&P severe) of the variation in M/B ratios among the sample utilities when ROE is introduced.

* * *

Increased utility exposure to stranded costs leads to a decrease in its M/B ratio. By year-end 1995, exposure to stranded costs had become a serious factor in investment decisions. The Moody's break-even approach and the S&P lost revenues approach provide equally powerful tools to explain equity investor preferences. t

Agustin Ros is an advisor to the chairman of the Illinois Commerce Commission and is currently working at the Federal Communications Commission on rules to implement the federal Telecommunications Act of 1996. Mr. Ros received his BA in economics from Rutgers University, and his MS and PhD in economics from the University of Illinois at Urbana-Champaign. John Domagalski is an associate of the utilities/energy

division of Coopers & Lybrand

Consulting/Palmer Bellevue. Mr. Domagalski has a BS in commerce from De Paul University. Philip O'Connor is a principal of Coopers & Lybrand Consulting/Palmer Bellevue. Prior to forming Palmer Bellevue in 1985, he served as chairman of the Illinois Commerce Commission. Mr. O'Connor earned an MA and PhD in political science from Northwestern University.

1 See, Moody's Investors Service, Stranded Cost Will Threaten Credit Quality of U.S. Electrics (August 1995).

2 Fixed production costs include nonfuel operating and maintenance expenses, fixed payments under long-term power contracts, interest on debt and property taxes, depreciation, and deferred assets. Costs of equity are not accounted for in this model.

3 See, Standard & Poor's, 1995 Utilities & Perspectives, Vol. 2, No. 48, Special Edition, Direct Access Threatens Electric Utility Revenues (November 27, 1995).

4 Market-to-book ratios were based on 1995 year-end stock prices and September 1995 book values, and taken from Regulatory Research Associates, Electric Utility Monthly-January 1996.

5 All "statistically significant" variables in this analysis are significant at either the .01 or .02 level - i.e., there is no more than a 2 out of 100 probability that the results of this analysis are random or occurred simply by chance.

6 Determinants of Market-to-Book, Linear Regression Estimates

Moody's S&P reas. S&P severe

Constant 1.662 1.739 1.720

Coefficient -.169 -4.260 -1.354

T-Stat -4.179* -4.156* -4.349*

Adjusted R2 .193 .191 .206

*Statistically significant at the .01 level.

7 Return on equity (1995 year-end average earned), production cost (1994 fuel and O&M expenses per Kwh generated inhouse), and percent industrial (1994 industrial Mwh sales as a percent of Mwh sales to ultimate customers) were taken from the Regulatory Research Associates Electric Utility Monthly-January 1996, Electric Utility Operating Cost Data (October 17, 1995), and Average Retail Price of Electricity: 1994 & Comparative Historical Data (May 22, 1995).

8 Determination of Market-to-Book, Multiple Regression Estimates

Moody's ROE Prod Cost %Ind. Load

Coefficient -.089 11.991 -.053 -.199

T-Stat -2.481** 7.063* -.841 -.859

Adjusted R2 .507

Constant .353

S&P reas. ROE Prod Cost %Ind. Load

Coefficient 13.540 12.079 .050 -.151

T-Stat -3.648* 7.583* .711 -.677

Adjusted R2 .553

Constant .215

S&P severe ROE Prod Cost %Ind. Load

Coefficient -1.038 11.971 .032 -.207

T-Stat -3.538* 7.449* .461 -.932

Adjusted R2 .548

Constant .256

*Satisfically significant at the .01 level.

**Satisfically significant at the .02 level.



Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.