Which matters most: Cost? Price? Sales? Regulation?
Many investors no longer think of electric utility stocks primarily as dividend-rich, income-oriented investments. Instead, they have begun to consider new criteria in evaluating utility stocks (em criteria that might help explain some of the variations in equity price performance now seen among various utility companies. Such criteria might include changes in state regulation, or company-specific data that track competitive indicators such as production costs, asset mix, price advantage, or customer profile.
How well do these new criteria predict utility stock performance, as measured by market-to-book ratio (M/B, or equity share price divided by book value)?
As we have found in our most recent study, at least three-fourths of the variance in M/B ratios observed in June 1996 across 73 utilities can now be explained by differences in regulation and certain competitive indicators: 1) return on equity (ROE), 2) industrial prices, 3) embedded cost of generation capacity, and 4) the relative progress achieved by state regulators toward industry restructuring. This level of explanation (75 percent) marks an improvement over our previous study, conducted last year and published in PUBLIC UTILITIES FORTNIGHTLY, %n1%n in which we found we could use ROE and stranded-cost estimates by bond credit rating agencies to explain 51-55 percent of the variance in year-end 1995 M/B ratios across 69 utilities.
Stranded Costs: Still Significant
In our previous study we relied on three different calculations of stranded costs in the electric generating sector, taken from reports by Moody's %n2%n and Standard & Poor's (S&P). %n3%n The Moody's report measured stranded generation plant investment (as a percent of equity); the S&P report looked at the potential utility revenue loss from retail competition (as a percent of total revenues) and gave two estimates, positing both "reasonable" and "severe" cases. An econometric analysis of M/B ratios for 69 utilities for which Moody's and S&P stranded cost estimates were both available revealed that a combination of ROE and one of the stranded-cost estimates explained 51-55 percent of total variance.
Here, however, we have chosen an expanded sample of 73 utilities. In the first phase we tested each of the three stranded-cost estimates from the earlier study in combination with ROE for year-end 1995, and then again for the end of June 1996, to see if there had been any significant change in the explanatory power of the stranded-cost estimates prepared by the rating agencies. Using the expanded sample of 73 companies, we found that the explanatory power of the stranded cost estimates has increased substantially by the end of June 1996, as compared to year-end 1995. Investors appear to have further factored stranded-cost exposure into their investment decisions (see Table 1). The explanatory power of the Moody's and the S&P "reasonable" case estimates appears to have increased by one-half during the past six months (em from 6 and 11 percent of variance to 9 and 17 percent, respectively. The variance explained by the S&P "severe" case has nearly doubled (em from 10 percent to 19 percent. At the same time, however, the explanatory power of ROE has increased only slightly (em from 45 percent of variance to 48 percent. Overall (combining the results for ROE and stranded-cost estimates), the Moody's estimates, the S&P "reasonable" case, and the S&P "severe" case estimates have each improved in their capability to explain variance in M/B ratios (em from 51 to 58 percent, 55 to 66 percent, and 55 to 67 percent, respectively.
(Compare these findings with the earlier study published last May, %n4%n in which we found that the three stranded-cost estimates were statistically significant and could explain roughly 20 percent of M/B variance.)
The differing methods for calculating stranded costs provide some insight into investor preferences. For example, the variance explained by S&P's estimates is almost double that of the Moody's estimates. Because the S&P calculations focus on potential revenue loss, as opposed to above-market investment in generation, the S&P correlation may indicate that investors are gravitating more toward cash flow in a competitive environment rather than any particular recovery of investment in plant prescribed by regulators.
Competitive Indicators: Some Useful, Some Not
Though helpful in understanding investor preferences and stock performance, the stranded-cost estimates calculated by Moody's and S&P are not readily updated or easily calculated. This drawback prompted us to look for standardized categories of data that might prove more easily available for testing as electric utility equity value drivers.
In addition to ROE, we tested five other financial variables, fourteen competitive or operational factors, and two variables that reflect regulatory policy.
Financial Variables: %n5%n
1) dividend payout ratio,
2) common equity ratio,
3) pre-tax interest coverage,
4) fixed-charge coverage, and
5) cash-flow dividend coverage.
Competitive/operational Factors: %n6%n
1) production costs, measured in dollars per kilowatt-hour ($/kWh),
2) non-production costs ($/kWh),
3) purchased power expenses ($/kWh),
4) total electric operating expenses ($/kWh),
5) load factor,
6) reserve margin,
7) average industrial price,
8) percent of load accounted for by industrial customers,
9) percent of fuel mix (in megawatt-hours, or MWh) accounted for by nuclear generation,
10) generation capacity cost, measured in dollars per kilowatt ($/kW),
11) generation assets as a percent of total assets;
12) five-year growth in total MWh sales;
13) capacity factor; and
14) customer density (customers per square mile).
1) Relative perceived regulatory risk of investing in utilities in a particular state, as ranked by state, %n7%n and
2) Relative progress of state toward industry restructuring, as ranked by state. %n8%n
Only three of these 21 possible new independent variables make a significant contribution in explaining the variance in M/B ratios. These three new variables account for a little over one-fourth of the total explained variance (with ROE explaining almost one-half). As to be expected, and as in the initial study of 1995 year-end M/B ratios, the key equity driver is a financial variable, return on equity, which explains 48.4 percent of total variance in M/B ratios. The new model, using more easily accessible data than the stranded-cost calculations of Moody's and S&P, finds nearly 23 percent of variance explained by competitive factors.
Industrial electric price levels account for 15 percent, while the capital cost of generation investment accounts for another 7.7 percent. The degree of regulatory restructuring in the state served by the utility was associated with an additional 3.7 percent of total variance (see Table 2). All four variables are statistically significant in their relationship to M/B ratios at the .01 level, meaning that the association would be likely to occur by chance only 1 time out of 100.
As expected, ROE, as a standard measure of profitability, is positively associated with M/B ratio, with a higher ROE indicating a higher M/B ratio.
Again, as expected, because industrial price is a reflection of a utility's ability to supply a commodity to its most vulnerable market segment in a less-regulated market, a higher industrial price has a negative association, indicating a lower M/B ratio. This variable captures some of the essence of the more complex-to-calculate S&P "reasonable" and "severe" cases and, therefore, should be considered as an easily available surrogate to the S&P estimates.
An easily accessible surrogate variable for the Moody's stranded cost calculation is generation capacity cost ($/kW installed less accumulated depreciation), reflecting a utility's embedded generation cost and to some considerable extent the fixed costs that would need to be covered by revenues generated from sales of electricity at market-clearing prices. As expected, the relationship is negative, with higher generation capacity costs indicating a lower M/B ratio.
We found it more difficult to attach an "expected" explanation to the relationship to M/B ratio indicated by the regulatory restructuring variable. As noted above, our data came from Regulatory Research Associates (RRA), which categorized states into five tiers based on their relative progress toward industry restructuring. Tier 1 indicated that a restructuring plan had been adopted, while Tier 5 indicated that no substantive restructuring initiative was underway. The relationship discovered by this analysis is positive, in the sense that the higher the number (em meaning the less that the state has done to restructure regulation (em the higher the M/B ratio. %n9%n At this point, this relationship raises more questions than it answers.
One might view the inverse relationship between restructuring progress and M/B ratio as "expected," but for different reasons. For instance, investors might see state restructuring efforts as leaving utilities less able to collect revenues under the traditional system and therefore less able to meet fixed costs and generate earnings. Or progress in state restructuring might indicate that market pressures in the state are so intense as to motivate regulators and legislators to meet consumer demands for choice. In that case, investor reaction might not be adverse to regulatory restructuring as such but, rather, to the indication that the market in the state is becoming more competitive.
On the other hand, one might see the inverse relationship as "unexpected," since investors should see restructuring as an indication that states are prepared to address such issues as stranded-cost collection. In any case, the modest relationship of regulation to M/B ratios deserves closer review, since we can expect more states to undertake restructuring initiatives in the near future.
Overall, this new model leaves one-fourth of total variance unexplained. Future research may take other variables into consideration, such as 1) the quality and vision of company management, 2) merger opportunities, 3) diversification activities, or 4) specific regulatory issues, such as plans for
performance-based ratemaking. t
John L. Domagalski is a senior associate of the utilities/ energy group of Coopers & Lybrand Consulting (C&L). Mr. Domagalski holds a BS in commerce from De Paul University. Agustin J. Ros, a senior analyst at the National Economic Research Associates, Inc., holds an MS and PhD in economics from the University of Illinois at Champaign-Urbana. He served previously at the Federal Communications Commission and as an advisor to the chairman of the Illinois Commerce Commission. Philip R. O'Connor is a principal of C&L, and previously served as chairman of the Illinois Commerce Commission.
Mr. O'Connor earned an MA and PhD in political science from Northwestern University.
1"Stranded Costs: Is the Market Paying Attention?" by Agustin J. Ros, John L. Domagalski, and Philip R. O'Connor, PUBLIC UTILITIES FORTNIGHTLY, May 15, 1996, p. 18.
2"Stranded Cost Will Threaten Credit Quality of U.S. Electrics" (Moody's Investor Service, August 1995).
3"Direct Access Threatens Electric Utility Revenues." 1995 Utilities and Perspectives, Vol. 2, No. 48, Special Edition (Standard & Poor's, Nov. 27, 1995).
4See note 1, supra.
5ROE and dividend payout ratio (1) were taken from Electric Utility Monthly (Regulatory Research Associates, July 1996). Common equity ratio (2), pre-tax coverage (3), fixed-charge coverage (4), and cash-flow dividend coverage (5) were taken from Electric Utility Quality Measures (Regulatory Research Associates, July 3, 1996).
6Production costs (1), non-production costs (2), purchased power expenses (3), and total electric operating expenses (4) were taken from the Electric Utility Operating Cost Data (Regulatory Research Associates, October 17, 1995). Load factor (5) and reserve margin (6) were taken from Electric Utility Capacity Data and Construction Data: 1994-1997 (Regulatory Research Associates, Nov. 9, 1995). Average industrial price (7) and percent of load accounted for by industrial customers (8) comes from Average Retail Price of Electricity: 1995 & Comparative Data (Regulatory Research Associates, May 23, 1996). $/kW generation capacity cost (10) and generation assets as a percent of total assets (11) were taken from The Grand Bargain (Salmon Brothers, Feb. 14, 1996). Five-year growth in sales (12) and capacity factor (13) were derived using the Resource Data International PowerDatÔ Database. Customer density (14) was taken from the Electric Utility Performance Profiles (Utility Data Institute, Feb. 1995).
7State Regulatory Evaluations, (Regulatory Research Associates, Apr. 15, 1996).
8Electric Industry Restructuring Update, (Regulatory Research Associates, Apr. 29, 1996).
9The RRA ranking of the states according to the degree of regulatory and industry restructuring involves the use of inherently ordinal data. An argument could be made that ordinal variables, for which distance between each successive value is not of the same magnitude, but only directional, are not customarily used in multiple regression analysis. In this case, however, the judgement was made to test the RRA rankings, along with our statistical assumption that the five ranking values involve reasonably similar distances between those values.
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.