Deregulation of electric utility firms : an assessment of the cost effects of complete deregulation vs. deregulation of generation only

UNIVERSITY OF
A BOOKSTACKS

Digitized by the Internet Archive

in 2011 with funding from

University of Illinois Urbana-Champaign

http://www.archive.org/details/deregulationofel905prim

!'■-■ L-

FACULTY WORKING

PAPER NO. 905

HIE LIBRARY QEItie

DEC 2 81982

^E^TYOF.LL.NO.S

Deregulation of E!ectric Utility Firms: An Assessment
of the Cost Effects of Complete Deregulation vs.
Deregulation of Generation Only

Walter J: Primeaux, Jr.

College of Commerce and Business Administration
Bureau of Economic and Business Research
University of Illinois, Urbana-Champaign

FACULTY WORKING PAPER No. 905
College of Commerce and Business Administration
University of Illinois at Urbana-Champaign
September 1982

Deregulation of Electric Utility Firms:
An Assessment of the Cost Effects of Complete
Deregulation vs. Deregulation of Generation Only

Walter J. Primeaux, Jr., Professor
Department of Business Administration

Abstract

During recent years, concerns about the real effectiveness of
electric utility regulation have led to a number of policy prescriptions,
including total deregulation and deregulation of only the generating
function. Proponents of deregulation of only generation function argue
that distribution is a natural monopoly and should not be subjected to
competition; they allege, without statistical data, that direct competi-
tion would result in higher distribution costs.

This paper presents an examination of the question of whether
competition in distribution would actually lead to higher costs. Data
from real markets, where direct electric utility competition actually
exists, show that competition in distribution does not lead to higher
costs. The main conclusion is that policy prescriptions, which only
recommend deregulation of generation only, are too limited in scope.

Deregulation of Electric Utility Firms:

An Assessment of the Cost Effects of Complete

Deregulation vs. Deregulation of Generation Only

by Walter J. Primeaux, Jr.
University of Illinois - Urbana-Champaign

INTRODUCTION

Concerns about real effectiveness of electric utility regulation
are not unique to recent times; indeed, the literature reveals that
through time, a number of authors have expressed serious criticism of
the existing regulatory process (e.g., Behling, 1938, Stigler &
Friedland, 1962, Moore, 1970).

Stigler and Friedland (1962) did not find any significance effects
of regulation of electric utilities. Moore's (1970) conclusions
included the observation that regulation has not reduced electricity
prices more than 5 percent and probably less than that amount.

C. Moore (1975) found that regulation had a perverse effect
because it actually caused higher consumer prices. Jackson (1969)
found that regulation did not succeed in reducing residential rates in
1940 and 1950, but was effective in 1960. One unusual empirical study
supporting the public interest theory of regulation, where regulators
maximize social welfare, is by Nelson (1982). The overall results of
these studies show little support for regulation as an effective insti-
tution and most authors express concern about economic performance in
an environment of commission regulation.

Another series of early studies examined, in a rigorous way, the
structure of the electric utility industry. These studies examined
the profit and price effects of combination utilities (those selling

-2-

both gas and electricity) compared with straight utilities (those
selling only gas or electricity) . These studies are typified by the
work of Mann (1970) and Collins (1973), although a number of other eco-
nomists have also investigated these situations. Results of the indi-
vidual studies in this group differed to some extent in their
assessments and conclusions concerning the effects of the accumulation
of very strong monopoly power in the case of combination utilities.
Yet, as a group, they developed rigorous statistical analyses and
raised the first serious questions concerning problems with the present
structure of the electric utility industry. The studies within this
group individually made policy recommendations regarding the value of
maintaining the existing structure compared with benefits of rivalry to
be gained from splitting up combination companies into competing
straight gas and electric companies.

In the 1970' s, a number of empirical studies examined the
Averch- Johnson overcapitalization effect under a regulatory constraint.
A number of authors engaged in these investigations but these studies
are typified by the work of Petersen (1976) and Needy (1976). These
kinds of studies generally showed that regulation caused firms to
employ excessive amounts of capital stock.

The above studies all raise some serious questions about the effect
of direct electric utility regulation and they demonstrate serious con-
cern about the outcome achieved under the existing regulatory process.

In 1968, I began research dealing with cities with two competing
electric utilitv firms. This research is similar in spirit to that

-3-

examining the competitive effects of combination gas-electric utili-
ties. The essential difference is that those combination studies
attempted to determine whether the beneficial effects of competition
were surpressed if a single monopoly sold both electricity and natural
gas within a given city, compared with situations where these two ser-
vices were provided by two separate companies. In contrast, my com-
petitive research has examined the effect of direct head-to-head com-
petition where two electric companies operate in a given city and
customers have a choice of being served by one company or the other.
Bellamy (1981) wrote a case study of such competition in Lubbock, Texas
and Primeaux (1974b) presented a case study of Sikeston, Missouri.

Through the years my research has examined a number of aspects of
the business to determine the effects of the direct competition between
firms alleged to be natural monopolists. This research has extended to
average costs levels, capacity utilization, price rigidity, case studies
of cities in which competition has existed over a long period of time,
and an examination of circumstances which have recently led to a decline
of direct competition in a few cities (Primeaux, 1974a, 1974b, 1975a,
1975, 1982). The lower average costs from direct competition were later
confirmed by Hollas and Herring (1982), using a slightly modified
sample.

My research was the first to examine the natural monopoly theory
under competitive conditions, using rigorous statistical methods.
Hellman's (1972) work was in process but unpublished when my direct
electric competition research program was begun, although this was
unknown to me. Even so, Hellman's research is largely composed of case

-4-

studies and examinations of the institutional arrangements surrounding
the direct competition where it existed or has existed In history. His
research, in any case, is interesting but It Is totally void of sta-
tistical analyses.

Moore (1976) and Ramsey (1976) later discussed some favorable
aspects of direct electric utility competition. These particular stu-
dies largely rely on previously published work and do not develop or
present any useful new data for examination; yet, they are of interest
to students of regulation. An earlier study by Seidel (1969),
discussed the favorable benefits of fringe area rivalry, where com-
petition exists only at and around the borders of service areas.
Seidel' s work is interested in assessing competitive benefits but it is
confined to limited competition.

More recently, Schuler and Hobbs have written a number of papers
using computer simulation to assess the effects of direct utility com-
petition. Their results generally show beneficial effects from the
direct competition. Some of this important work is presented in
Schuler and Hobbs (1981) and Hobbs and Schuler (1981, 1982). One
important difference between the Hobbs and Schuler research and the
research undertaken by the previous Primeaux studies is that their work
is based on simulation of engineering data, using analyses which do not
allow for any X-efficiency as found by Primeaux (1977). Their research
constitutes another approach to assessing the impact of direct com-
petition; yet, the realism of differential efficiency is not captured
by their analyses.

-5-

Seidel (1981) explains that the Primeaux research is the only work
actually using direct operating data (under competitive conditions) to
examine performance of electric utility firms. The point is that the
effects of the competition changes the position of the cost curves
through X-efficiency (Primeaux 1977), and these effects would possibly
impact upon other performance measures, such as consumer prices, for
example.

Plummer seems to be calling for additional research of the direct

electric utility competition question in the following quote concerning

my research.

If it could be demonstrated that such direct competi-
tion did not lead to major inefficiencies, then the
whole argument for treating distribution as a natural
monopoly could come tumbling down... (see Primeaux,
1975) (from: Plummer, 1981).

This study examines the effect of direct competition on the costs

of firms which generate and distribute power compared with firms which

only distribute power but do not generate. The results show that the

direct competition does not lead to major inefficiencies. It is not

necessary, therefore, that electricity distribution be treated as a

natural monopoly.

PURPOSES OF THE STUDY
The above discussion has shown that there has been much concern
expressed concerning the existing regulatory process as well as some
questions raised about the present structure of the electric utility
industry. These concerns have now reached the stage where policy makers
are seriously considering various alternatives and options which could

-6-

lead to drastic changes in structure and regulation of the industry
(Seidel 1981).

The Edison Electric Institute (1982) and various papers in Shaker
and Steffy (1976) explain that one possible regulatory reform is the
deregulation of generation only with regulatory control maintained over
distribution. This, however, is not clearly the best policy choice
(Seidel, 1981, Plummer, 1981). The superiority of this choice compared
with total deregulation has not been previously established.

Actually, the Primeaux (1975a) study contained both firms which
bought their power requirements for sale to consumers as well as firms
which generated power for sale to consumers. Consequently, the sample
lends itself to an important new analysis of the difference between
firms which only distribute power compared with those which generate
and distribute. An analysis of these results permit a determination of
the differential cost effects of each phase of operation. These results
provide the assessment of any inefficiency, mentioned earlier, which
may be caused by the direct competition which Plummer (1981) considers
to be very important. Moreover, an evaluation of these results provide
some insight into the comparative benefit of deregulating the genera-
tion function only vs. the benefit to be gained from deregulating both
the distribution and generation functions. These results are useful
because the data are from competitive markets, so competitive effects
are actually reflected in the data. Since competition will be a result
of deregulation, these data tend to show what would happen if deregula-
tion of each function took place and direct competition ensued.

-7-

METHOD

Nature of the Sample

The procedure followed for selecting the sample for the statistical
analyses was very similar to that used in Primeaux (1975a, 1977, 1981).
The directly competitive situations actually consist of cases where a
publicly-owned (municipally-owned firm operating in a single city)
competes with a privately-owned firm (operating in several cities).
The privately-owned firms do not allocate operating data to the indivi-
dual cities they serve; moreover, their competitive area constitutes
only a relatively small part of their overall operation, so the effect
of the direct competition upon these multi city firms is not very
significant. Consequently, the data of privately-owned firms is of no
value to this study, even if they were available. This complication is
elaborated upon in detail in Primeaux (1975a, 1977, 1978).

Because of the problem mentioned above, the sample was limited to
the municipally-owned firms actually facing direct competition and they
were compared with a sub sample of municipally-owned firms which are
monopolists. The cities included in the sample are presented in Tables
1A and IB in the appendix. Municipally-owned firms do not suffer from
the data problems mentioned above. The sample of matched firms was
selected from those presented in Primeaux (1975a, 1977, 1978) so as to
avoid statistical problems mentioned in those studies.

The sample consisted of firm data from 1964-1968, composed of five
years of pooled cross section-time series data, as in Primeaux (1975a).
The appropriate Chow tests to confirm the acceptability of the pooling
process are presented in the appendix. More recent data were not used

-8-

for two reasons. First, as Primeaux (1932) indicates, the sample of
available firms has declined in more recent years, so the choice was
made to use more older data instead of less more current data. Second,
more recent data would have been affected, to some extent, by the change
in energy supply characteristics which took place in the 1970's. To
some extent, those changes could have distorted operating results with
effects which would not be indicative of meaningful differences upon
which to make policy judgments. In addition to the previous justifica-
tion for using the older data, a previous study has explained that the
period around 1967 represented rather settled conditions for making
comparisons (Mann and Mikesell, 1971). All in all, extreme care was
taken in the sample selecting procedure and process to preserve the
integrity of the statistical results.

STATISTICAL ANALYSIS

The statistical analysis, model, and variables used in this study
follow closely those in Primeaux (1975a).

The statistical procedure was ordinary least squares multiple
regression analysis with equations in the form

Y = A + B.X. + B.X. + B.X, X

11 2 2 3 3 n

where:

Y is the estimated average cost of the firm

X. sales of electricity, in million of kilowatt-hours

X„ generating capacity utilization rate

X steam electric fuel cost

-9-

X, hydroelectric fuel cost

Xs consumption per commercial and industrial customer

X, consumption per residential customer
o

X7 cost of purchased power, per kwh.

XQ market density factor

o

X internal combustion-generation dummy

X Alabama dummy

X Indiana dummy

X Iowa dummy

X Maryland dummy

X... Missouri dummy

X _ Ohio dummy

X , Oregon dummy
lo

X 7 South Carolina dummy

X 0 South Dakota dummy
lo

X q Texas dummy

X„„ Nebraska dummy

X„ Alaska dummy

X „ competition dummy variable

X9„ interaction variable (X„„ with X.. )
An explanation of data sources and variable specification are provided
in the appendix.

The only difference between the above variables and those used in
Primeaux (1975a) is that in the earlier study the purchased power
variable was specified differently. As indicated in the appendix, the
purchased power variable in this study is the actual cost of purchased

-10-

power per KWH purchased; Che 1975 study used the proportion of purchased
kilowatt-hours of power to total kilowatt-hour sales. To assess the
impact of this change, the 1975 equation was specified with this new
purchased power variable. When the equation was run with the changed
purchased power variable, the competition dummy variable was changed
from -1.5155 mills per million kwh in the 1975 study to -1.353 mills
per million KWH. So the results are more conservative with the modified
specification. No other important differences occurred. Since purchased
power is quite important in the following analysis, the decision was
made to use the variable reflecting actual purchased power costs, instead
of the 1975 specification.

Although the remaining variables in this study are identical to
those in the 1975 study, there is a significant change in the procedure
used to develop the cost equations. As in the earlier study, Michigan
dummy variables are omitted to avoid the statistical problems which
occur when all dummy variables are included in equations. Primeaux
(1975a) combined firms which generated and distributed power with firms
which purchased their energy requirements and only distributed power to
consumers. The cost equations presented in that study, therefore, were
quite composite in nature and any differential impact of competition
upon the individual functions of the electric utility business was not
assessed.

The procedure used here represents a refinement, which should make
the results more useful for public policy purposes and should provide
some additional information toward answering the questions mentioned
earlier which were raised by Plummer (1981).

The sample for this study was divided into two types of firms; that
is one sub set consisted of those firms which generated and distributed

-11-

power. The second consisted of firms which only purchased their require-
ments and did not generate any power. The division was made because it
was thought that this approach would permit a better assessment of the
effects of duplicate facilities to engage in direct electric utility
competition. The cost impact of duplication of facilities caused by
direct competition probably does not fall equally upon firms which
generate and distribute and those which do not generate and purchase
all of their power requirements. Indeed, strong presumptions about the
nature of the differential cost impact are behind the arguments presented
by those who advocate deregulating only the generating function and
requiring the distribution function to remain as a regulated monopoly.

The following analysis permits an assessment of the cost effects of
complete deregulation of electric utilities after direct competition
ensued.

Table 1 presents the average cost equation for firms in the sample
which generate and distribute power. Many of these firms did buy some
power but they were not solely dependent on purchased power. Indeed,
the industry data show that many firms, both publicly and privately
owned, purchase power for resale, even though they also generate with
their own facilities.

Table 2 presents the average cost equation for non-generating firms.
The following discussion relates to the coefficients presented in both
Tables 1 and 2.

The coefficient of the sales variable (X ) reveals that economies
of scale accrued to the generating and distributing companies; however,
diseconomies of scale accrued to the non-generating companies, indicating

■12-

TABLE 1

POOLED REGRESSION
FIRMS GENERATING AND DISTRIBUTING

VARIABLE

X.

10
11

12
13
14
15

16

17

18

19
c20

21
^22

23

Sales of Electricity

(millions of kilowatt-hours)

Generating Capacity Utilization

Steam-Electric Fuel Cost

Hydroelectric Fuel Cost

Consumption per Commercial and

Industrial Customer

Consumption per Residential Customer

Cost of Purchased Power

Market Density Factor

Internal Combustion Generation Dummy

Alabama Dummy

Indiana Dummy

Iowa Dummy

Maryland Dummy

Missouri Dummy

Ohio Dummy

Oregon Dummy

South Carolina Dummy

South Dakota Dummy

Texas Dummy

Nebraska Dummy

Alaska Dummy

Competition Dummy

X * CD Interaction Variable

PARTIAL
REGRESSION STANDARD
COEFFICIENT ERROR

-.002

.000*

-.059

.014*

-.045

.020**

-.002

.002

-.032

.004*

-.213

.067*

-.013

.004*

-.190

.269

-.744

.468

2.030

.566*

2.039

.561*

2.731

.796*

2.129

.355*

.994

.468**

4.595

1.055*

4.636

.762*

3.688

.546*

-.952

.774

.977

.683

1.471

.337*

.007

.001*

Summary Statistic

N (degrees of freedom plus number of variables) 172

R2 .8151

Constant 23.287 (mills)

Standard error of estimate 1.3476 (mills)

Source: Derived from pooled data for the competitive and noncompetitive
utilities in Table 1A of Appendix.

*Significant at 1 percent level.

**Signif icant at 5 percent level.

***Signif icant at 10 percent level.

-13-

TABLE 2

POOLED REGRESSION
NON-GENERATING FIRMS ONLY

VARIABLE

c10

'16

:22

^23

Sales of Electricity

(millions of kilowatt-hours)

Consumption per Commercial and

Industrial Customer

Consumption per Residential Customer

Cost of Purchased Power

Market Density Factor

Alabama Dummy

Maryland Dummy

Missouri Dummy

Ohio Dummy

Oregon Dummy

South Carolina Dummy

Competition Dummy

X * CD Interaction Variable

PARTIAL
REGRESSION STANDARD
COEFFICIENT ERROR

,012

.003*

-.014

.003*

-.244

.065*

.690

.363***

1.867

.489*

4.583

2.581***

3.008

2.268

-.761

.709

1.897

1.371

■3.302

3.645

-.644

1.233

1.065

.916

-.009

.007

Summary Statistic

N (degrees of freedom plus number of variables) 52

T~ .9908

Constant 13.166 (mills)

Standard error of estimate .4255 (mills)

Source: Derived from pooled data for the competitive and noncompetitive
utilities in Table 1A of Appendix.

*Significant at 1 percent level.

**Signif icant at 5 percent level.

***Signif icant at 10 percent level.

-14-

that their average costs increased as their sales levels increased.

The market density variable (XQ), indicates that average costs decreased

o

as the number of customers per square mile increased, reflecting some

tendency for adverse pressure to occur on average costs of non-generating

firms. This same pressure, however, did not seem to exist for generating

and distributing firms because the X variable is not statistically

o

significant at the 10 percent level.

Another interesting result shown in Tables 1 and 2 is the coefficient
on the purchased power variable (X7). While this coefficient is nega-
tive and statistically significant (at the one percent level) for firms
which generate and distribute, it is positive and statistically signi-
ficant at the ten percent level for firms which only distribute power.
This difference, of course, occurs because of the relative importance
of purchased power to firms of each type.

The mean cost of purchased power to the non-generating companies
was $5.96 per 1000 KWH while the average price paid by firms which
generated and distributed was $10. 41 per 1000 KWH. Part of the explana-
tion for the differences in the price of purchased power between the
two groups of firms has to be found in the fact that the companies which
buy all of their requirements are probably influenced to follow that
strategy, to some extent, by the fact that they are able to buy at lower
prices. If purchased power were unattractively priced to them, they
would begin to generate their requirements.

In addition to the differences mentioned above, the following dif-
ferences between the Primeaux (1975a) study and the equations presented
in Table 1 should be mentioned. The 1975 study presented positive and

-15-

statistically significant coefficients on the steam electric fuel cost
variable, the hydroelectric fuel cost variable, and the internal com-
bustion fuel cost variable. Equation 1 shows a negative and statisti-
cally significant sign on the steam electric fuel cost variable, and
negative but insignificant signs on the hydroelectric and internal com-
bustion fuel variables.

Table 1 shows that for firms generating and distributing power the
competition dummy variable is negative and shows that the average cost
curve is shifted downward by 1.471 mills per million KWH, because of
the direct rivalry. This variable is significant at the one percent
level. The interaction variable of the competitive dummy with the
variable X shows the slope of the cost curve does change with direct
competition. It, too, is significant at the one percent level.

The essential fact is that the magnitude of the downward shift in
Table 1 was quite similar to that presented in the 1975 equation
(Primeaux 1975). That equation presented a downward shift effect of
1.5155 mills per million kilowatt-hours sales. That equation, too,
presented a positive interaction variable with sales, indicating that
the slope of the average cost function became steeper, as sales
increased. Together, the results presented here indicate that direct
competition does cause lower average costs for firms which generate and
distribute power after controlling for a large number of cost and demand
variables; however, after reaching 210 million kilowatt annual sales,
monopoly firms which generate and distribute power operate at lower
costs than their competitive counterparts. These results are con-
sistent and similar with those presented in Primeaux (1975). Although
the equation presented above is the appropriate specification, Table 2C
in the appendix presents this equation after excluding the interaction

-lb-

variable with Che variable X (X *CD). The results are consistent with
those presented above.

Perhaps the most interest result from Table 2 is also the com-
petition dummy variable and its interactioin variable with sales volume.
This variable shows that when scale, density effects, purchased power
as well as a large number of other key demand and operating charac-
teristics and state differences are all controlled for, that competitive
non-generating firms did not have higher costs than their monopoly
counterparts. The competitive dummy variable coefficient is, indeed,
positive, indicating that there is upward pressure on average costs
when direct competition exists; yet, the difference is not statistically
significant at the ten percent level. So the effect is unimportant.
Moreover, the interaction variable is also statistically insignificant,
so direct competition does not cause these distribution firms to operate
at higher average costs. The equation discussed here is the appropriate
specification. Table 2D in the appendix presents the same equation after
excluding X *CD. The results are consistent with those presented above.

The reader is reminded that the above discussion does not present
data or results for firms which only generate power and do not perform
the distribution function. It is not possible to develop analyses to
examine the effect of competition on the generation function only
because there are no firms facing direct competition which do not
distribute power. Nevertheless, the extension of these results to the
deregulation scenarios discussed above does not seem to be unreasonable.

CONCLUSIONS
The above results mean that firms which only distribute power do
not have higher costs under competition than under monopoly; conse-
quently, concern expressed by those who advocate deregulating the

-17-

generating function but not the distribution function, because of
important losses in economies of scale, seem to be unfounded. Even
though monopoly distribution firms may have the capability of operating
at lower costs than competitive distribution firms, that result was not
achieved. It is the X-inef f iciency which sets in, in a monopoly market
structure, which offsets the technical losses caused by the direct com-
petition. This kind of situation is discussed in detail in Primeaux
(1977).

Consistent with the Primeaux (1975a) study, these results show that
firms which distribute and generate, do enjoy substantial performance
improvement when subjected to direct competition.

Since this analysis employs real data from real markets, where
competition already exists (or does not exist in the case of the
monopolists) , the findings are useful for public policy consideration.
The results show that complete deregulation seems to be practical
because firms only distributing power, in a competitive market struc-
ture, did not incur higher average costs than their monopoly counter-
parts.

The advantage of following complete deregulation for firms which
generate and distribute is that there are cost economies through
improved X-eff iciency (Primeaux 1975a). The advantage of deregulating
firms which only distribute is because they incur no higher costs than
monopolists, when they face competition. If both groups are completely
deregulated it would then be possible to have price competition in
residential, commercial and industrial service, without concern for the
arbitrary regulatory process of rate of return regulation or rate making,

-18-

The market mechanism can automatically perform the regulatory function,
as it does in most other businesses.

The expected outcome from direct competition would be that the
rivalry would force the firms to become more efficient and operate at
lower costs; this fact, along with the competitive price rivalry, would
provide consumers with lower prices. The lower prices would be possible
because of the elimination of inefficiency and any economic profits
which may exist under the present arrangement.

-19-

REFERENCES

B. N. Behling, 1938. Competition and Monopoly in Public Utility
Industries. The University of Illinois, Urbana.

William H. Collins, 1973. "Comparative Performance of Combinations
and Separately Managed Electric Utilities," The Southern Economic
Journal.

Edison Electric Institute, 1982. Alternative Models of Electric
Power Deregulation, Washington, Economics Division, May.

B. F. Hobbs and Richard E. Schuler, 1981. "Estimating the Consequence
of Competition in the Distribution of Electricity — An Application
of Location Theory," presented to the Twenty-eighth Annual North
American Meeting of the Regional Science Association, Montreal,
Canada, November.

B. F. Hobbs and Richard E. Schuler, 1982. "A Spatial Linear Program-
ming Analysis of the Deregulation of Electricity Generation,"
presented at the Institute of Management Sciences Operations Re-
search Society of America, Detroit, April.

Daniel R. Hollas and Robert S. Herren, 1982. "An Estimation of the
Deadweight and X-Efficiency Losses in the Municipal Electric
Industry," Journal of Economics and Business, in press.

David H. Huettner and John H. Landon, 1978. "Electric Utilities:
Scale Economics and Diseconomies," Southern Economic Journal,
Vol. 44, No. 4, April.

Richard Hellman, 1972. Government Competition in the Electric Utility
Industry, New York, Praeger Publishers.

R. Jackson, 1969. "Regulation and Electric Utility Rate Levels,"
Land Economics, Vol. XLV, August.

William Kent. Mechanical Engineers' Handbook, vol. 1, Power ed. J. Kenneth
Salisbury, 12th ed. , John Wiley, 1950.

Patrick C. Mann, 1970. "The Impact of Competition in the Supply of

Electricity," Quarterly Review of Economics and Business, Vol. 10,
Winter.

Patrick C. Mann and J. L. Mikesell. "Tax Payments and Electric Utility
Prices," The Southern Economic Journal, vol. 38, July 1971.

Edward J. Mitchell, et. al. , 1975. Toward Economy in Electric Power,
Washington, American Enterprise Institute.

-20-

C. J. Moore, 1975. "Has Electricity Regulation Resulted In Higher
Prices? An Econometric Evaluation Using A Calibrated Regulation
Input Variable," Economic Inquiry, Vol. XIII, No. 2, June.

Thomas G. Moore, 1970. "The Effectiveness of Regulation of Electric
Utility Prices," Southern Economic Journal, Vol. XXXVI, April.

Thomas G. Moore, 197b. "Effects of Regulation on Electric Power

Companies," in Shaker and Steffy (eds.). Electric Power Reform:

The Alternative for Michigan, Ann Arbor. The University of Michigan.

Charles W. Needy, 1976. "Social Cost of the A-J-W Output Distortion,"
Southern Economic Journal, Vol. 42, January.

Randy A. Nelson, 1982. "An Empirical Test of the Ramsey Theory and
Stigler-Peltzman Theory of Public Utility Pricing," Economic
Inquiry, Vol. XX, April.

James L. Plumraer, 1981. "Scenarios for Deregulation of Electric
Utilities," Mioeo, November, EPRI.

H. Craig Petersen, 1976. "The Effect of 'Fair Value' Rate Base Valua-
tion in Electric Utility Regulation," Journal of Finance, Vol. 31,
December.

Walter J. Primeaux, Jr., 1974a. "A Reexamination of the Kinky

Oligopoly Demand Curve," Journal of Political Economy, Vol. 82,
No. 4, July-August.

Walter J. Priuieaux, Jr., 1974b. "A Duopoly in Electricity: Competi-
tion In A 'Natural Monopoly'," Quarterly Review of Economics and
Business, Vol. 14, No. 2, Summer.

Walter J. Primeaux, Jr., 1975a. "A Reexamination of the Monopoly
Market Structure for Electric Utilities," in Promoting Competi-
tion in Regulated Markets, Washington, The Brookings Institution.

Walter J. Primeaux, Jr., 1975b. "The Decline in Electric Utility
Competition," Land Economics, Vol. LI, No. 2, May.

Walter J. Primeaux, Jr., 1976. "Electric Utility Competition is
Feasible," in Shaker and Steffy (eds.), Electric Power Reform:
The Alternative For Michigan, Ann Arbor, The University of
Michigan.

Walter J. Primeaux, Jr., 1977. "An Assessment of X-Efftciency Gained
Through Competition," Review of Economics and Statistics, Vol. LIX,
No. 1, February.

Walter J. Primeaux, Jr., 1973. "The Effect of Competition on Capacity
Utilization in the Electric Utility Industry," Economic Inquiry,
Vol. XVI, No. 2, April.

-21-

Walter J. Primeaux, Jr., 1979. "Some Problems With Natural Monopoly,"
The Antitrust Bulletin: Journal of American and Foreign Antitrust
and Trade Regulation, Vol. XXIV, No. 1, Spring.

Walter J. Primeaux, Jr., 1981. "An Assessment of the Effect of Compe-
tition on Adversity-Intensity," Economic Inquiry, Vol. XIX,
October.

Walter J. Primeaux, Jr., 1982. An Examination of Direct Electric
Utility Competition: Some Perspectives on the Natural Monopoly
Myth (under review for publication by a University press) .

James B. Ramsey, et. al. , 1976. "Government Power, Regulation, and
Markets," in Shaker and Steffy (eds.), Electric Power Reform:
The Alternatives for Michigan, Ann Arbor, The University of
Michigan.

Marque Seidel, 1969. "The Margins of Spatial Monopoly," Journal of
Regional Science, Vol. 9, August.

Marque Seidel, 1981. Pennsylvania Electric Efficiency Scoping Study,
Harrisburgh, PA, mimeo.

R. E. Schuler and B. F. Hobbs, 1981a. "The Consequences of Alternative
Organizations of the Electric Utility Industry," presented at 1981
meeting of the American Economic Association, December.

William H. Shaker and Wilbert Steffy, 1976. Electric Power Reform:
The Alternative for Michigan, Ann Arbor, The University of
Michigan.

G. Stigler and Claire Friedland, 1962. "What Can Regulators Regulate?"
Journal of Law and Economics, Vol. 5.

Statistics of Publicly Owned Electric Utilities in the United States,
Washington, U.S. Department of Energy, various years.

Typical Electric Bills, Washington, U.S. Department of Energy, various
years.

Steam Electric Plant Factors, National Coal Association, various years.

Area Measurement Reports, U.S. Department of Commerce, various years.

D/112

APPENDIX

Table 1A

Cities in the Sample with Generation

Duopolies Years of Data

Anchorage, Alaska 1964-1968 5

Fort Wayne, Indiana 1964-1968 5

Maquoketa, Iowa 1965-1968 4

Hagerstown, Maryland 1964-1968 5

Allegan, Michigan 1964-1967 4

Dowagiac, Michigan 1964-1968 5

Ferrysburg, Michigan 1964-1968 5

Traverse City, Michigan 1964-1968 5

Zeeland, Michigan 1964-1968 5

Kennett, Missouri 1964-1968 5

Poplar Bluff, Missouri 1964-1968 5

Trenton, Missouri 1964-1968 5

Lincoln, Nebraska 1964-1965 2

Cleveland, Ohio 1964-1968 5

Columbus, Ohio 1964-1968 5

Piqua, Ohio 1964-1968 5

Sioux Falls, South Dakota 1964-1968 5

Garland, Texas 1964-1968 5

85

Monopolies Years of Data N

Richmond, Indiana 1964-1968 5

Algona, Iowa 1965-1968 4

Niles, Michigan 1964-1968 5

Wyandotte, Michigan 1964-1968 5

Hillsdale, Michigan 1964-1968 5

Lansing, Michigan 1964-1968 5

Sturgis, Michigan 1964-1968 5

Petosky, Michigan 1964-1963 5

Carthage, Missouri 1964-1968 5

Columbia, Missouri 1964-1968 5

Marshall, Missouri 1964-1968 5

Omaha, Nebraska 1964-1965 2

Springfield, Illinois 1964-1968 5

Anderson, Indiana* 1964 1

Logansport, Indiana 1964-1968 5

Eugene, Oregon 1964-1968 5

Watertown, South Dakota 1964-1968 5

Springfield, Missouri 1964-1968 5

San Antonio, Texas 1964-1968 _5

87

*Anderson ceased to generate in 1965.

Table 2B

Cities in the Sample Without Generation

Duopolies

Bessemer, Alabama
Tarrant City, Alabama
Bay City, Michigan
Springfield, Oregon
Greer, South Carolina

Years of Data

1964-1968
1964-1968
1964-1968
1964-1968
1966-1968

N

5
5
5
5
_3_
23

Monopolies

Florence, Alabama
Scottsboro, Alabama
Bristol, Virginia*
Rolla, Missouri
Greenwood, South Carolina
Anderson, Indiana**

Years of Data

1964-1968
1964-1968
1964-1968
1964-1968
1964-1968
1965-1963

N

5
5
5
5
5
4
29

*This is a matched firm for Maryland; so in the data, it is considered
to be a Maryland firm.

**Anderson generated in 1964 and ceased generation in 1965.

TABLE 2C

FOOLED REGRESSION

FIRMS GENERATING AND DISTRIBUTING

(X*CD interaction variable excluded)

VARIABLE

PARTIAL

REGRESSION

COEFFICIENT

STANDARD
ERROR

>-,
■■n
7

X8
X9

Xll

X12

x 3

X14
X1S

x16

Jl8

19

X20

x21

X22

Sales of Electricity

(millions of kilowatt-hours)
Generating Capacity Utilization
Steam-Electric Fuel Cost
Hydroelectric Fuel Cost
Consumption per Commercial and

Industrial Customer
Consumption per Residential Customer
Cost of Purchased Power
Market Density Factor
Internal Combustion Generation Dummy
Indiana Dummy
Iowa Dummy
Maryland Dummy
Missouri Dummy
Ohio Dummy
Oregon Dummy
South Dakota Dummy
Texas Dummy
Nebraska Dummy
Alaska Dummy
Competition Dummy

-.0U2

.000*

-.050

.014*

-.023

.021

-.022

-.022

-.033

.004*

-.178

.071*

-.016

.004*

-.422

.282

-.493

.496

1.387

.585*

2.159

.597*

2.565

.848*

2.109

.378*

-.107

.456

4.209

1.121*

4.591

.812*

3.170

.571*

-.721

.824

1.513

.718**

-.890

.326*

Summary Statistic

N (degrees of freedom plus number of variables) 172

T2 .7896

Constant 22.657 (mills)

Standard error of estimate 1.4373 (mills)

Source: Derived from pooled data for the competitive and noncompetitive
utilities in Table 1A of Appendix.

*Significant at 1 percent level

**Signif icant at 5 percent level

***Signif icant at 10 percent level

TABLE 2D

POOLED REGRESSION

NON GENERATING FIRMS ONLY

(X *CD interaction variable excluded)

VARIABLE

X

10
13
14

16
17

"22

Sales of Electricity
Consumption per Commercial and

Industrial Customer
Consumption per Residential Customer
Cost of Purchased Power
Market Density Factor
Alabama Dummy
Maryland Dummy
Missouri Dummy
Ohio Dummy
Oregon Dummy
South Carolina Dummy
Competition Dummy

PARTIAL

REGRESSION

STANDARD

COEFFICIENT

ERROR

.011

.003*

-.012

.002*

-.228

.064*

.589

.357***

-1.700

.474*

-5.348

2.531**

-3.968

2.157***

-.914

.705

1.227

1.277

-5.123

3.381

-.970

1.216

.373

.744

Summary Statistic

N (degrees of freedom plus number of variables) 52

R"2 .9907

Constant 13.976 (mills)

Standard error of estimate .4289 (mills)

Source: Derived from pooled data for the competitive and noncompetitive
utilities in Table 1A of Appendix.

*Significant at 1 percent level

**Signif icant at 5 percent level

***Signif icant at 10 percent level

CHOW TESTS

To ascertain whether it was statistically justified to pool the
time series data, it was necessary to determine whether the parameters
had shifted during the five-year time period covered by the data. The
statistical procedure involved computing a separate regression for each
of the five years and then applying an analysis of variance test (Chow
test). This procedure was followed for two different operations.
First, regression equations including all firms in the sample were run;
that is both the firms which did not generate and those which did
generate power were combined in the same equations. Second, the test
was run only for firms which generated power. These tests are presented
in the following two tables.

The tables show that the hypothesis of unshifted parameters cannot
be rejected, since the calculated F value, in each table, is less than
the appropriate table value. These results reveal that each year can
be treated as a separate observation.

It was not possible to perform the same test for firms which did
not generate power, because there were insufficient degrees of freedom
to run the necessary series of equations. Nevertheless, this step does
not seem to be necessary since these firms were included in the first
test and excluded in the second and no shift in parameters was indicated
in either case.

CHOW TEST

INCLUDES FIRMS WHICH ONLY DISTRIBUTE POWER
AS WELL AS FIRMS WHICH GENERATE AND DISTRIBUTE

Source of Regression

Std. Error
of Estimate

D.F.

K

DF+K

MSE1

Statistics

SSR2

Pooled Regression

1.5957

200

24

224

2.546

509.2

1964 Regression

1.9756

21

23

44

3.903

81.963

1965 Regression

1.9083

22

24

46

3.642

80.124

1966 Regression

1.8752

22

23

45

3.516

77.352

1967 Regression

2.3681

22

23

45

5.608

123.376

1968 Regression

2.0067

21

23

44

4.087

84.567

1 2

(Std. Error of Estimate)

"D.F. x MSE

Fc =

(509.2 - 81.963 - 80.124 - 77.352 - 123.376 - 84.567)

24 ~~ ~~

(81.963 + 80.124 + 77.352 + 123.376 + 84.567)
(224-21-22-22-22-21)

61.818

24 = 2.57575

447.382 3.8567414

116

,66786

Fc = .66786 < F^6 (.01) - 1.96.

CHOW TEST
INCLUDES ONLY FIRMS WHICH GENERATE POWER

Source of Regression
Statistics

Pooled Regression

1964 Regression

1965 Regression

1966 Regression

1967 Regression

1968 Regression

Std. Error

S.S.

of Estimate

D.F.

K

DF+K

MSE

Residuals

1.3476

150

22

17 2

1.816

272.4

1.6960

14

21

35

2.876

40.264

1.6415

14

22

36

2.695

37.73

1.5872

13

21

34

2.519

32.747

2.0104

13

21

34

4.042

52.546

1.6608

12

21

33

2.758

33.096

(272.4 - 40.264 - 37.73 - 32.747 - 52.546 - 33.096)

22
FC ~ (40.264 + 37.73 t- 32.747 + 52.546 + 33.096)

(172-14-14-13-13-12)

76.017

22 3.45532

196.383 1.85267

10 6

.18650

Fc = 1.8650 < F?n, (.01) - 2.00.

THE VARIABLES
The dependent variable is average costs for the firm. Total cost
for the firm, excluding taxes and tax equivalents, were divided by
annual sales in thousands of kilowatt hours.

X Sales of Electricity to All Customer Classifications. In millions
of kilowatt-hours. Larger sales levels would be expected to reduce
average costs, if economies of scale exists. (From Statistics of
Publicly Owned Electricity in the United States, various years.)

X Capacity Utilization. Total generating capacity for each firm was
multiplied by 8,760 (the number of hours in a 365-day year); the
product is the potential number of kilowatt-hours that each firm
could have provided during a year if capacity had been fully utilized,
without down time for repairs or maintenance. The potential capa-
city was divided into the number of kilowatts actually generated.
A higher rate of capacity utilization would be expected to reduce
average total costs. (Data from Statistics of Publicly Owned
Utilities in the United States, various years.)

X Steam Electric Fuel Cost. Composite fuel costs for all firms within
a given state were computed. These figures were adjusted for
burning efficiency by applying factors from Kent (1950). The pro-
ducts were then weighted by the proportionate utilization of steam-
electric generation and total generation. (Fuel costs are from
Steam Electric Plant Factors, various years.)

X, Hydroelectric Fuel Costs. This variable was constructed by weighting
the total hydroelectric production investment per kilowatt of hydro-
electric generating capacity by the proportion of total generation
accounted for by hydroelectric generation (data from Statistics of
Publicly Owned Electric Utilities in the U.S.).

' 5 Consumption per Commercial and Industrial Customer. The actual
average annual consumption of commercial and industrial customers
of each utility (data from Statistics of Publicly owned Electric
Utilities in the United States).

X Consumption per Residential Customer. The actual average annual

consumption of power per residential customer. (Data from Statistics
of Publicly Owned Electric Utilities in the United States.)

X-, Cost of Purchased Power. Constructed by taking total expenditures

for purchased power divided by the number of KWH purchased. (Data
from Statistics of Publicly Owned Utilities in the United States).

X Market: Density Factor. This variable was constructed by dividing
the number of thousands of square miles in each city into the
number of customers of all classes. (Customers from Statistics of

Publicly Owned Electric Utilities in the United States. Land area

-

from U.S. Department of Commerce, Area Measurement Reports, various
years.)

XQ Internal Combustion Generation Dummy. This variable indicated

whether a firm produced any amount of electricity by internal

combustion generation. Its value was one for firms that did, zero
otherwise.

X -X Scate Dummy Variables. Indicating the state in which the firm
was located.

X Competition Dummy Variable. This variable was used to indicate

whether a firm faced competition; it took a value of one if competi-
tion existed and zero if not.

Public Finance. There is no variable in the model to take into account
the cost and benefits of public finance, although it could well be an
important factor, especially if comparisons were to be made between
privately owned and municipally owned utility firms. The comparison
here is between municipally owned firms, not privately owned firms.
Municipal governments do not all have the same tax rate, and some muni-
cipally owned firms pay no taxes or tax equivalent. The problem this
created in cost comparisons was overcome by eliminating all tax and
tax-equivalent charges from the cost data. Municipally owned firms may
also enjoy lower capital costs than privately owned firms because of
lower external interest costs and capital contributions from the muni-
cipality. The first benefit was of no consequence to the analysis
since only municipally owned firms were included. The impact of the
second element is difficult to assess. Municipally owned utilities
will be disinclined to rely on capital contributions from the city if
they seriously wish to tie costs to the users of their services.

Furthermore, there is no reason to believe that the benefit of such
capital contributions accrued to the competitive subset of firms more
than to the other. It was therefore assumed that the effect, if any,
was distributed randomly among competitive and noncompetitive firms.

1ECKMAN

IINDERY INC.

JUN95

, "• „. f, N. MANCHESTER.
und.-ro-Pleasy ,ND|ANA 46962