## "Back-to-basics" strategies challenge enterprise-risk philosophies.

Nearly a year ago, cover story announced the rise of the chief risk officer (CRO). "Utility...

## Competitive Efficiency: A Ranking of U.S. Electric Utilities

(ai) in the above equation. That is, had all utilities used the same amount of each input, all differences in output levels would be represented in the intercept.

In estimating the efficiency level associated with each utility, the most efficient utility would be defined as the one with largest intercept. In other words, the most efficient utility represents 100-percent efficiency, and all other utilities are compared to it.

4The data, estimation results and summary statistical properties in SAS output format are available from the authors by request.

**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. **

50-75% 5 86.8%

over 75% 5 81.3%

Utility with gas sales Yes 47 90.7%

No 47 90.3%

Percent Industrial 0-20% 22 89.1%

20-40% 60 91.0%

over 40% 12 90.0%

Holding Company Yes 30 91.6%

No 64 90.0%

Hydro electric % of sales 0 39 90.0%

0-10% 47 91.0%

over 10% 8 90.0%

1Several econometric techniques have been developed for obtaining the measurement of each component. The computational procedures, however, are complex and inexact.

2One study employing this technique was published in PUBLIC UTILITIES FORTNIGHTLY. (See, "The Efficient Utility: Labor, Capital, & Profit," by D. Thomas Taylor and Russell G. Thompson, Sept. 1, 1995, p. 25.) That study used Data Envelopment Analysis, a mathematical programming technique, to estimate relative efficiencies of 13 investor-owned utilities. Some of that study's flaws and certain weaknesses of its methodology were later noted by Matthew Morey and L. Dean Hiebert. (See, "Measuring Utility Efficiency: A New Frontier" [letter to editor], PUBLIC UTILITIES FORTNIGHTLY, Jan. 1, 1996, p. 7.)

3The estimated equation was formulated as:

Ln(Yit) = Siai +SjbjLn(Xijt) + LFit + Îit where Ln(Yit) is the natural logarithm of total output in megawatt hours, Ln(Xijt) is natural logarithm of a set of j inputs (labor, capital, fuel and material), LF is the load factor, and T is a trend variable with values of 1 to 6 representing each year of data from 1990 to 1995. Index i refers to utilities, and index t refers to time periods.

Ît is an error term representing two elements: statistical noise (vit) and inefficiency (ui): Îit = vit + ui). The decomposition of the error term into its two components may be done in several ways. The fixed effects approach assumes differences in the efficiency of different utilities are captured in their respective intercepts by the term (ai) in the above equation. That is, had all utilities used the same amount of each input, all differences in output levels would be represented in the intercept.

In estimating the efficiency level associated with each utility, the most efficient utility would be defined as the one with largest intercept. In other words, the most efficient utility represents 100-percent efficiency, and all other utilities are compared to it.

4The data, estimation results and summary statistical properties in SAS output format are available from the authors by request.

**Articles found on this page are available to Internet subscribers only. **