The numbers say "yes," adding weight to last year's benchmarking survey.
Does productive efficiency help determine an electric utility's prospects in regulated or competitive markets? Is productive efficiency a better marker of real-world success than simple financial attributes, such as cash flow, dividend ratio or operating income?
In unregulated markets, higher productivity translates directly into relative declines in costs and prices, and by extension, greater ability to compete and prosper. In the regulated arena, it improves the utility's earning potential through more favorable regulatory treatment, especially under performance-based ratemaking regimes. Now, however, in financial markets too, there is ample evidence that productive efficiency can secure high marks for a utility.
Prompted by competitive pressures emanating from open access on the one hand and induced by the potential rewards of performance ratemaking on the other hand, in recent years electric utilities have begun to pay increased attention to the efficiency with which they conduct their business. Productive efficiency of U.S. utilities in the production and delivery of electric power was examined in two recent studies, the results of which were published in two issues of Public Utilities Fortnightly.fn1 Using different analytic procedures, these studies produced indices of productive efficiency for about 100 investor-owned utilities, and explored patterns underlying the observed variations in efficiency in terms of geography, utility size, resource composition and other relevant dimensions.
This article serves to extend those studies. It focuses on the implications of efficiency and productivity in terms of a utility's standing in financial markets.fn2 Using the results of operational efficiency rankings developed in the previous studies, in this paper we explore the relationship between productive efficiency and a utility's performance in financial markets. We show that quantitative measures of technical efficiency derived from formal econometric and optimization models are reasonable instruments for capturing technical and operational performance and, arguably, are very useful partial predictors of how a utility's stocks perform.
We begin with a working hypothesis that a utility's market performance depends mainly on a short list of objective characteristics: its demonstrated financial fundamentals, the regulatory and competitive settings in which it does business, resource flexibility and operating efficiency.fn3 In line with modern finance theory, we assume that capital markets work efficiently - that the market price of a utility's stock represents its true asset value. In turn, we assume that the market price reflects all relevant information and provides the necessary clues to what market participants think about corporate actions and their likely impacts on shareholder wealth. Informed either by their own perceptions or guided by the evaluations of rating agencies and other intermediaries, investors show their expectations of the enterprise's current and future levels of risks and returns by "bidding" for its stocks in financial markets.
A large number of factors, both objective and subjective, related to market opportunities, competitive position, perception of risks and operating environment are considered when investors and rating agencies evaluate the worth of a company's stocks. Standard & Poor's, for example, uses a complex set of criteria in its rating method that incorporates both quantitative metrics such as financial ratios and qualitative and subjective business indicators related to prospects for growth, stability, decline and regulatory environments. They also explicitly take into account how efficiently a utility manages and utilizes its resources.
Objective Measures and Analytical Method
There are two popular measures of market performance: price-earnings ratio (P/E) and market-to-book ratio (M/B). These ratios combine accounting and market data to provide an objective, though admittedly imprecise, picture of market performance.
For electric utilities, and all utilities in general, the M/B ratio is a relatively stable measure with small short-term variations. Compared to other industries, these ratios are also generally very low, averaging about 2.0. M/B ratio arguably is a better measure of market performance for utilities due to this stability and also because the P/E ratio largely is subject to rate-setting cycles and affected by the lag between allowance of earnings and the actual booking by the utility. We tested each of these ratios in our analysis. The market-to-book ratio produced better results with respect to primary statistical properties of the results.
To measure the unique effects of productive efficiency on the M/B ratio, it is important to account as fully as possible for the influence of other factors thought to influence that statistic. In the parlance of econometrics, this step is often called "controlling" the effects of other variables. Regression offers the appropriate and perhaps simplest technique for this purpose. The technique is used to describe the causal relationship between a dependent variable (in this case the M/B ratio) and a set of explanatory variables that "explain" the variations in the former.fn4
In our analysis, we used a linear regression equation to describe the relationship between M/B ratio and its hypothesized determinants. We took into account five sets of factors: (1) financial indicators, (2) regulatory "assets," (3) competitive settings, (4) generating capacity mix (fuel diversity) and (5) operational efficiency.fn5 Regulatory treatment, presumably a relevant factor, was not included in the analysis due to its highly subjective nature.
1. Financial indicators: Business and financial ratios in six conventional ratio categories of liquidity, activity, profitability, stock, dividend and cash flow were included in the analysis as follows:
long-term debt as a percentage of total assets (liquidity ratio);
total revenue minus fuel purchases/ total assets (activity ratio);
operating income as a percentage of operating revenues (profitability ratio);
earnings per share before taxes
common dividend as a percentage of operating revenue (dividend ratio); and
cash flow from operations/interest payments (cash flow ratio).
2. Regulatory "assets": These items occur primarily in the form of the utility's expenditures on social programs and conservation and load management activities. Since they are not equivalent to the utility's other physical and potentially "bondable" assets, they are expected to be viewed unfavorably by the financial community, and hence have a negative effect on market performance.
3. Competitive settings: Intensity of competition in the market(s) in which a utility operates is expected to have a direct impact on its perceived financial prospects. Depending on the utility's strategic plans, how well it is poised for competition, and how effectively it communicates these to the public, this perception may or may not be favorable. This variable is represented in this analysis in the form of an index of deregulation activity. The index ranges from one to four based on the following designation:
retail choice approved for all customers, index value = 4;
retail choice approved for some customers, index value = 3;
retail choice under consideration, index value = 2;
little or no regulatory action on retail choice, index value = 1.
For utilities with operations in more than one jurisdiction, we weighted the index values, based on total megawatt-hour sales. Given the uncertainty inherent in the dynamics of competitive markets, we expect deregulation and open access to affect adversely market value. Figure 1 shows the M/B ratio for the various NERC regions.
4. Capacity mix: Fuel diversity provides flexibility in a changing environment. Potential supply disruptions and price fluctuations can raise rates and ignite political and regulatory pressure that ultimately may lead to erosion of financial performance. We hypothesize that the ability to alter generating sources and take advantage of lower cost fuels would be viewed favorably by financial markets. Fuel diversity in this analysis is represented in terms of a modified measure of entropy, the "G" index with a range of zero to one.fn6 It measures how generation of power is distributed among hydro, steam, nuclear and other fuel sources. The less concentrated is the power output, the higher the value of "G." Thus, if all of a utility's power were generated from a single source, the value of "G" would take the value of zero. Conversely, if generation were distributed evenly among four fuels, the value would be unity.
5. Operational Efficiency: For this category we relied on the universe of utilities and their efficiency ranks used for The Fortnightly 100, as published in Sept. 1, 1998. The method there employed Data Envelopment Analysis. DEA uses a mathematical optimization to construct a convex production frontier, tracking the input/output ratios of the most efficient companies. Companies that form the production frontier are considered efficient and receive a score of 1.0. All other utilities receive an efficiency score between 0 and 1.0, based on distance from the production frontier.
Other Miscellaneous Factors. The operation of nuclear plants poses a special challenge for electric utilities. While a nuclear plant can offer significant opportunities in resource flexibility, the aging stock of nuclear plants is likely to increase financial exposure resulting from higher maintenance expenses and uncertain costs of decommissioning. More stringent environmental restrictions also can add significantly to operation and maintenance costs and in disposal fees for spent fuel. To test this hypothesis, we included nuclear plant ownership as a separate variable in our analysis.
By contrast, we ignored consideration of several other variables, known to have a bearing on perceptions of performance and value in the financial community, particularly management and prices. Surely a utility's management is of paramount importance to its market performance, since management's decisions affect all areas of the utility's operations. But management quality and effectiveness are subjective measures and are hard to quantify in a consistent way. Price of the product also has a bearing on how the utility is assessed, for it influences competitiveness probably more directly than any other factor. But it is important to keep in mind that effectiveness of management and product prices are strongly correlated with operating efficiency and their effects are already captured in the efficiency index. Their inclusion would therefore be redundant and would confound the statistical results of the regression analysis.
The Data. The historical data we used in estimating the parameters of the regression model covered the 1995-1997 period. The POWERDAT( Database, an electronic compilation of the information contained in Federal Energy Regulatory Commission Form 1 and other utility-specific financial data, was the primary source of information for this analysis.fn7 Additional financial data were obtained from the Bureau of Labor Statistics and FERC. As noted above, The Fortnightly 100 supplied the list of the 100 utilities analyzed and their efficiency ranks. However, our regression model in the present article was estimated using information on 77 utilities. Lack of complete and reliable information on financial indicators for some utilities prevented the inclusion of all 100 utilities in the sample.
Reading the Numbers:
What They Suggest
The statistical results of the estimated regression equation are summarized in Table 1. For each variable, figures in the table show the mean values for all variables included in the analysis during the 1995-1997 period, the estimated slope coefficient (beta), the t statistic and estimated elasticity at the mean.
The "mean values" show the average of the various explanatory variables, as well as that of the dependent (M/B ratio) variables. For each explanatory variable, the estimated coefficient measures the impact of a one-unit change in that variable on the M/B ratio. The t statistic is a measure of the statistical significance of the relationship between the explanatory variable and the M/B ratio. A value of greater than 2 in absolute terms generally indicates a statistically significant relationship. For each explanatory variable, the elasticity (ei) values are calculated at the mean as the product of the estimated coefficient and the ratio of the means of the dependent (y) and the explanatory (xi) variables. It measures the percentage change in the dependent variable that is likely to arise as the result of 1 percent change in the explanatory variable. Positive values of the elasticity indicate the particular explanatory variable has a positive impact on the M/B ratio (i.e., when the explanatory variable increases, so does the M/B ratio).
As can be seen in Table 1, all estimated parameters have the correct signs, supporting the hypothesized relationships between market-to-book ratio and its determinants. Judging by the values of the t statistics, these effects also seem to be significant from a statistical point of view. The three exceptions are the stock and liquidity ratios and fuel diversity. The estimation results also produce a relatively strong fit for the overall relationship with a multiple coefficient of determination (R2) of 0.64, showing that 64 percent of the observed variations in the M/B ratio are explained by the specified relationship.
The results support the hypothesized relationship between market-to-book and the financial ratios in all cases. With the exception of liquidity, these effects are statistically significant in all cases. For example, the positive sign on the coefficient associated with profitability shows that it has a positive effect on market performance and that this effect is statistically significant. In absolute terms, as measured by its "elasticity at the mean," 1 percent increase in profitability ratio would lead to one-tenth of 1 percent change in M/B ratio, on average.
Fuel diversity appears to affect favorably market-to-book ratio, but its contribution is small (elasticity at the mean = 0.007) and surprisingly insignificant from a statistical perspective (t-statistic = 0.23). The binary variable indicating nuclear plant ownership has a negative sign and is statistically significant at a 95 percent level of confidence. This result shows that indeed there is uncertainty among investors concerning the ability to keep these stations running smoothly and economically.
Based on our findings, open access and retail choice are also likely to produce an erosive effect on investor confidence. Further analysis of the relationship between our index of deregulation activity and average M/B ratios at the level of the 10 NERC regions (see Figure 1, M/B Ratio as a Function of Deregulation Activity) indicates a strong negative correlation between the two variables (correlation coefficient = -0.81). It appears that, at least at the aggregate level and in the short run, progress in deregulation of electricity markets is regarded with a certain caution and pessimism among investors.
Productive Efficiency: Why It Matters
In this analysis we set out to explore some of the determinants of market performance in the electric utility industry and, specifically, to measure the unique contributions made by productive efficiency. The "strength" of any of the explored relationships is illustrated by the size of the elasticity measures displayed in the last column of Table 1. Productive efficiency shows the largest elasticity of all the explanatory variables examined in this study. The results show that a 1 percent increase in productive efficiency is likely to lead to more than six-tenths of 1 percent increase in M/B ratio. (Compare that to the impact of "stock ratio," where a 1 percent increase leads to merely 0.017 percent increase in M/B ratio.)
The strong relationship between productive efficiency and market performance as measured by the M/B ratio is apparent in the two-variable scatter plot shown in Figure 2. Judging by the statistical test (t=2.19) of its estimated coefficient, one can also conclude that it is highly improbable that the observed impacts would be attributable to chance.
In the final analysis, it is apparent that productive efficiency matters a great deal. After all, greater efficiency and higher productivity is of paramount importance from the perspective of costs and quality of service. Sub-standard productivity can lead to political (customer relations), regulatory and competitive problems for a utility. But we also have seen that productive efficiency is a broad concept, encompassing myriad operating issues and conditions. It also is difficult to measure. Based on the results of this analysis, we propose that quantitative measures of relative efficiency derived from formal statistical and optimization models are good proxies for the more subjective indicators used by the financial markets in evaluating a utility's business and financial prospects.
Hossein Haeri, Ph.D. is director of energy information services at PG&E Energy Services, in Portland, Ore. Matei Perussi is a statistical analyst at PG&E Energy Services. The views presented in this article do not necessarily represent those of PG&E Energy Services. M. Sami Khawaja, Ph.D. is the president of Quantec, an economic consulting firm in Portland.
Editor's Note: Last fall, when Public Utilities Fortnightly published a study ranking electric utilities on the basis of productive efficiency, known as The Fortnightly 100, some readers took umbrage. They argued that any survey ranking electric companies for efficiency in the generation sector ignores a host of potential biases, such as differences in asset mix, geographical location, population density, regional trading patterns and, more fundamentally, the fact that certain restructuring policies involve a sell-off of generating assets.
Here, two authors from that article return with the same data set to explore whether operational efficiency - as defined and analyzed in last fall's article - can show any positive relationship with a utility's performance in financial markets. They are joined here by a third author who worked with them on a previous article, published in 1997, that also ranked utility operational efficiency.
In the final analysis, as the authors show so well, productive efficiency still matters. It matters a lot. - B.W.R.
1 See "The Fortnightly 100" by Janice Forrester, M. Sami Khawaja, Hossein Haeri and Michael Carter in Public Utilities Fortnightly, Sept. 1, 1998, p. 26. See also "Competitive Efficiency: A Ranking of U.S. Electric Utilities" by Hossein Haeri, M. Sami Khawaja and Matei Perussi in Public Utilities Fortnightly, June 15, 1997, p. 26.
2 Although we have used "productive efficiency" and "productivity" interchangeably, the two terms are not synonymous. "Efficiency" describes how well a production process is run; "productivity" measures its result in terms of utilization of resources.
3 This approach is also consistent with the Standard and Poor's general evaluation framework and rating criteria outlined in its Rating Methodology for industrials and utilities.
4 Analytically, the regression model is defined as:
Where y is the dependent variable, a is the intercept of the regression line with the vertical axis, xi represents the independent variables (i = 1¼ k), k is the number of independent variables, bi is the estimated coefficient measuring the effect on y resulting from a one-unit change in xi and e is the error term, i.e., the unexplained portion of the variations in y.
5 As mentioned above, operational efficiencies measure how well a utility uses capital, labor, O&M and fuel in generating electricity. As such, operational efficiencies simply measure how well inputs are converted to output, and the various other variables used in this study are not reflected in the calculation of efficiency.
6 This index is a variant of the entropy index (H), which is a useful measure of dispersion, or inversely, concentration. As applied here, it is represented analytically as follows:
Where MWh is the power produced by fuel i (i = 1 ¼ ), MWht^ denotes total generation, and "log" is the natural logarithm. In order to facilitate interpretation of the results, we use "relative entropy," the "G" index, defined as follows:
Note that when only one fuel or generation source is used, since G is indeterminate, its value is set to zero.
7 POWERDAT( is a registered trademark of Resource Data International Inc., Boulder, Colo.
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.