Measuririg the Effects of Sales Below Cost Laws 
in Retail Gasoline Markets 
by 
Rod Wesley Anderson 
A thesis submitted in partial fulfillment 
of the requirements for the degree 
of 
Master of Science 
in 
Applied Economics 
MONTANA STATE UNIVERSITY-BOZEMAN 
Bozeman, Montana 
December 1995 
11 
APPROVAL 
of a thesis submitted by 
Rod Wesley Anderson 
This thesis has been read by each member of the thesis committee and has been 
found to be satisfactory regarding content, English usage, format, citations, bibliographic 
style, and consistency, and is ready for submission to the College of Graduate Studies. 
Ronald N. Johnson I), /t-( /9 5 
I I Date 
Approved for the Department of Agricultural Economics and Economics 
Douglas Young l2_- ~-1 ~ 
(Signature) Date 
Approved for the College of Graduate Studies 
Robert L. Brown 
(Signature) Date 
lll 
STATEMENT OF PERMISSION TO USE 
In presenting this thesis in partial fulfillment of the requirements for a master's 
degree at Montana State University-Bozeman, I agree that the Library shall make it 
available to borrowers under rules of the Library. 
If I have indicated my intention to copyright this thesis by including a copyright 
notice page, copying is allowable only for scholarly purposes, consistent with "fair use" 
as prescribed in the U.S. Copyright Law. Requests for permission for extended quotation 
from or reproduction of this thesis in whole or in parts may be granted only by the 
copyright holder. 
Date _ ___;r-=c..=-r\ _..4_,.[ ....... 7 ..... S: _______ _ 
iv 
ACKNOWLEDGMENTS 
Dr. Ronald Johnson has contributed an immeasurable amount of time and effort 
towards this thesis. His attention to detail and ability to discern the relevant problems 
and questions have been motivational from its beginning. I am grateful for his guidance 
and patience, which have allowed me to learn throughout the stages of this project. 
The contributions of Dr. David Buschena and Dr. Thomas Stratmann were also 
essential to the completion of this thesis. Their efforts and the time they have given are 
gratefully acknowledged. 
Finally, I would like to thank my parents and family, whose support and concern 
is always present. 
v 
TABLE OF CONTENTS 
Page 
APPROVAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii 
STATEMENT OF PERMISSION TO USE ................................... iii 
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv 
TABLE OF CONTENTS ................................................. v 
LIST OF TABLES ..............................•....................... vii 
LIST OF FIGURES .................................................... viii 
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 
1. INTRODUCTION ..................................................... 1 
2. PREDATORY PRICING THEORY AND IMPLICATIONS 
OF SALES BELOW COST LAWS .................................... 4 
Sales Below Cost Laws ~d Predatory Pricing ........................... 4 
Economics and Predatory Pricing ..................................... 6 
Applying the Areeda/Turner Rule to SBC Laws ............... · .......... 21 
Alternative Explanations for SBC Laws ............................... 25 
Conclusion ................ · ...................................... 26 
3. DATA DESCRIPTION AND LEGAL REVIEW ........................... 28 
Introduction .................................... · .................. 28 
An Overview of the Gasoline Distribution System . . . . . . . . . . . . . . . . . . . . . . . 29 
The Components of Gasoline Price .................................... 30 
Wholesale Price .......................................... ; ... 31 
Taxes ...................................................... 33 
Retail Margin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 5 
Variables that Affect the Retail Margin ................................ 36 
Sales Below Cost Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 
Alabama ... · ................................................. 45 
VI 
TABLE OF CONTENTS --- Continued 
Colorado ..................................................... 45 
Florida ..................................................... 46 
Massachusetts ................................................ 4 7 
Missouri .................................................... 4 7 · 
Montana .................................. ; ................. 48 
New Jersey .................................................. 49 
Tennessee ................................................... 49 
Utah .........................................•.............. 50 
The Effect of SBC Laws: A Preliminary Analysis ........................ 50 
Conclusion ....................................................... 57 
4. EMPIRICAL TESTS AND RESULTS ................................... 64 
Introduction ................................. : .................... 64 
Econometric .Considerations ................. ·. . . . . . . . . . . . . . . . . . . . . . . . 64 
Model Specification and Results .•.................................... 65 
Measuring the Effect on Retail Margins . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 
Measuring Response Differences to Increases in 
Wholesale Price ... ; ........................................ 71 
Conclusion ....................................................... 79 
5. CONCLUSION ...................................................... 81 
REFERENCES CITED .................................................. 84 
APPENDICES· ................................ : ........................ 89 
Appendix A--Interpolation of Weekly Quantity Data ..................... 90 
Appendix B--Descriptive Statistics .................................... 92 
Appendix C--Motor Fuels Taxes by City ............................... 95 
Appendix D--Sales Below Cost Laws by' State ........................... 98 
Appendix E--Regression Results of Equation 4.1 .......... ; .......... ~ .. 100 
vii 
LIST OF TABLES 
Table Page 
1. Calculation of Gasoline Sales Tax ................................... 36 
2. Comparison of Descriptive Statistics ofLaw and Non-LawStates .......... 55 
3. The Effect of SBC Laws on Retail Margins ............................ 66 
4. The Impact of Active Legal Enforcement .............................. 68 
5. The Effect of SBC Laws on Retail Price ............................... 70 
6. Estimation of the Lag Structure of Prices with a Lagged 
Dependent Variable .......... · .................................... 73 
7. Test of Asymmetrical Price Responses ................................ 77 
8. The Differences in Retail Price Response 
to Increases in Wholesale Prices ................................... 80 
9. Descriptive Statistics of Key Variables ................................ 93 
10. Motor Fuels Taxes and Computational Formulas ....................... 96 
11. Sales Below Cost Laws by State .................................... 99 
12. Full Regression Results of Equation 4.1 ............................. 101 
viii 
LIST OF FIGURES 
Figures Page 
1. The Cost Functions of a Typical Finn ................................ 12 
2. Price/Quantity Combinations Under the Average Variable Cost Rule ........ 17 
3. The Gasoline Distribution System .................................... 30 
4. Supply and Demand Illustration of Quantity Purged of Price Effects ........ 42 
5. Retail Margins and Minimum Markups in Billings, MT .................. 59 
6. Retail Margins and Minimum Markups in Bozeman; MT . . . . . . . . . . . . . . . . . 60 
7. Retail Margins and Minimum Markups in Great Falls, MT .......... : . . . . . 61 
8. Retail Margins and Minimum Markups in Helena, MT ................... 62 
9. Retail Margins and Minimum Markups in Missoula, MT . . . . . . . . . . . . . . . . . 63 
10. Time Path for the Cumulative Adjustment of Retail Prices 
to Wholesale Price Changes .............. · ................... · .... 77 
IX 
ABSTRACT 
The retail gasoline market in some states is regulated by laws that prohibit the sale 
of gasoline at prices below the retailer's cost. The stated purpose of these laws is to .offer 
protection from predatory pricing, a practice whereby one ftrm seeks to eliminate its 
competition through loss inducing prices. However, the assumption that predatory 
pricing is occurring in the retail gasoline business is questionable. If predatory pricing is 
not occurring, the laws may instead protect less efficient firms by establishing a price 
floor that results in higher prices for consumers. 
To test this hypothesis, retail margins in states with and without such laws are 
examined. The results suggest that retail margins tend to be slightly higher in states 
where sales below cost laws are effective. These results are not consistent with the idea 
that predatory pricing is a frequently occurring phenomenon in the retail gasoline sector. 
1 
CHAPTER 1 
INTRODUCTION 
Legislation that has the intent of protecting consumers is often predicated on the 
fear of market power and monopoly prices. Since predatory pricing may ultimately lead 
to higher prices, it is one of the strategies that laws supposedly guard against. In the 
United States, this concern dates to the conception of antitrust laws, most notably the 
Sherman Act (U.S.C., title 15, sec. 1-7) of 1890.1 A current example is found at the state 
level where laws in 11 states explicitly prohibit the sale of gasoline at prices below cost. 
It is argued that absent these laws, firms would temporarily lower prices with the intent of 
eliminating competition. The ultimate objective is monopoly profits after the victims 
have. departed the market. 
If predatory pricing is occurring in the retail gasoline business, such laws, if 
effective, could protect competition and consumers. But if the structure of the industry is 
not conducive to predatory pricing, the laws could instead lead to inefficiency by 
restricting competition and decreasing social welfare. In this case, it is likely that the 
laws are the result of special interest groups, motivated by a concern for their own well-
being. 
1Robert H. Bork, The Antitrust Paradox, (New York, NY: Basic Books, Inc., 
1978), 19. 
2 
These laws are referred to as Sales Below Cost (SBC) laws and the objective of 
this thesis is to measure the effects of gasoline specific SBC laws on prices paid by 
consumers. Chapter 2 of this thesis provides an understanding of the relationship 
between predatory pricing and SBC laws. The analysis presented offers an indication of 
the likelihood of predatory behavior in the retail gasoline business. By definition, 
predatory pricing requires a level of prices that are below some measure of cost. It is 
therefore necessary for SBC laws to defme this level in a measurable way. But if such a 
definition is restrictive in terms of eliminating the low prices of more efficient firms, 
retail margins could be higher. 2 In providing a definition, we appeal to economic theory 
and utilize a rule developed by Phillip Areeda and Donald F. Turner.3 By comparing 
Areeda and Turner's rule with the SBC laws' defmition of predatory pricing, a sense of 
whether these laws are pro or anti consumer is provided. 
Following this analysis, it is argued that these laws are unlikely to enhance 
competition. Instead, they create a binding price floor so that retail margins will be 
higher in states with gasoline specific SBC laws. It should be emphasized that this 
hypothesis is conditional on the premise that SBC laws do in fact constrain behavior. 
Thus, in addition to the above, it is hypothesized that if the laws generally constrain 
pricing behavior retail prices in states with SBC laws will respond more quickly to 
increases in wholesale prices. 
2The reason for using the retail margin is explained in Chapter 3. 
3Phillip Areeda and Donald F. Turner, "Predatory Pricing and Related Practices 
Under Section 2 of the Sherman Act," Harvard Law Review 88:697-733 (1975). 
3 
The data to empirically test the hypothesis is described in Chapter 3. In addition, 
. . 
evidence of the laws' effects are obtained by examining related legal cases. Further 
evidence is sought by comparing the descriptive statistics of states with and without SBC 
laws. A graphical analysis for Montana is also utilized. Given this preliminary 
examination, Chapter 4 offers empirical tests of the effects of the SBC laws. Models are 
developed to test the hypothesis and their results are presented. Conclusions are also 
drawn concerning predatory pricing in the retail gasoline business. Chapter 5 
summarizes the connection between the theory of predatory pricing, implications of SBC 
laws, and the results of the empirical tests. The results do not support the notion that 
SBC laws protect consumers. 
4 
CHAPTER2 
'PREDATORY PRICING THEORY AND IMPLICATIONS OF 
SALES BELOW COST LAWS 
Sales Below Cost Laws and Predatory Pricing 
Proponents of Sales Below Cost (SBC) laws have often argued that these laws are 
necessary because of a perceived threat of predatory pricing. State legislators indicate 
concern over a decrease in competition due to a decrease in the number of independent 
gasoline retailers.1 It is argued that oil companies, refiners, and other petroleum 
marketers have the capability and willingness to set their prices below cost for the 
purpose of eliminating their competition. 
These predatory pricing practices are generally alleged to follow a particular 
pattern. First, the retail price of a gallon of gasoline is set below a level that allows an 
individual retail outlet to recoup the costs incurred in selling a gallon of gasoline. 
Wholesale gasoline prices and taxes constitute the majority of these costs, but the 
additional costs of doing business, such as labor, are included. Second, marketers with 
multiple retail locations use profits from one location to subsidize below cost prices at 
1See, Montana. Montana Retail Motor Fuel Marketing Act, Code. Annotated 30-
14-802, and Alabama. Motor Fuel Marketing Act, Statutes. Annotated 8-22-2. 
5 
another retail location, or individual retail outlets use profits from non-motor fuels 
products to subsidize prices on motor fuels. The sale of cigarettes, for example, may be 
used to subsidize the price of motor fuels. Third, vertically integrated oil companies use 
profits from upstream operations to subsidize motor fuels prices at their retail stations. 
For example, Exxon would use profits from its refinery operations to subsidize prices at 
Exxon owned stations. 
These alleged practices describe a situation in which a company is willin~ to incur 
additional costs for a period of time so that its competitors can be eliminated. In the short 
run, prices are lowered below costs and are subsidized by other products, outlets, or the 
wealth of the predator. The intended victim must follow the lower prices, thus incurring 
losses. The predator likewise incurs losses during this period, but is capable of 
maintainirig these losses for a longer period of time than the victim. Eventually, the 
victim is unable to endure the costs of the predatory campaign and is driven out of 
business. In a simple market with only two competitors and no entry into the market, the 
predator supposedly has sufficient mar~et power to capture monopoly profits. The losses 
incurred during the predatory campaign are presumably recaptured by higher monopoly 
prices in the post predation period. 
The assumption that such predatory practices are occurring should be prefaced 
with several important questions.. First, is it rational for a firm to engage in a predatory 
campaign, and if so, under what conditions? Second, how can it be determined if the 
pricing policies of a particular company imply predatory intent? Answers to these 
questions require a measure of cost and a definition of when prices below that measure of 
6 
cost are predatory. 
The SBC laws relating to motor fuels products do not deal explicitly with the first 
ofthese questions. However, since the underlying assumption is· that predatory practices 
occur frequently enough to warrant legislative action, it could be assumed that the general 
belief is that predatory behavior is a rational strategy and that the conditions necessary for 
this practice to be successful· are present. The majority of the laws do, however, define 
cost and a make a price below this level predatory. 
Economics and Predatory Pricing 
Economic theory offers guidance for analyzing the issue of predatory pricing. A 
starting point is to ask the question - what are the conditions under which predatory 
pricing is likely to occur? Several necessary· conditions are generally accepted. 
First, the predator firm must have an advantage over the intended victim. If both 
firms are identical, in terms of their cost functions, the predator will incur greater losses 
during the period of predation than the victim. As the predator decreases price, the victim 
must follow or risk losing business. When this happens, however, a price taking firm will 
also decrease the quantity it is willing to sell. Since the quantity demanded at the lower 
price is greater and the victim is selling a lower quantity, the predator must increase 
sales.2 Thus, the predator's losses will be greater than those of the victim. If competing 
firms realize this, the threat of predatory pricing is not credible. Moreover, the predator 
21f the price elasticity within the industry is relatively low in the short run this 
effect may be weak. 
7 
must realize this as well. Therefore, when the firms involved are identical, the victim 
would be better off than the predator during and after the predatory campaign, making it 
highly unlikely that predatory behavior would occur. 
On the other hand, if the predator has a cost advantage over a competitor, she can 
conceivably set a pric~ that results in losses during the period of predation that are less 
than those of the victim. If the victim ?annot incur losses for as long as the predator, 
predation may be successful. The threat of predatory pricing then becomes credible, and 
may be a rational, wealth maximizing strategy. 
The cost advantage does not, however, guarantee that the victim of a predatory 
campaign will exit the market. Several factors may affect this decision. Most notably, 
predatory pricing is illegal under U.S. antitrust law.3 Set;:tion 2 of the Sherman Act 
(U.S.C., title 15, sec. 1-7) prohibits actions to monopolize a market. The Robinson-
Patman Act (U.S.C., title 15, sec. 13), also deals with sales at low prices for the intent of 
eliminating competition. Since firms found guilty of predatory behavior may be subject 
to treble damages in private suit, the antitrust statutes can act as a deterrent to predatory 
pricing. Second, the acquisition of outside financing is generally available. If the victim 
believed that the predatory prices could not be maintained for a long period of time, it 
may be in her best interest to obtain outside financing until prices returned to their normal 
level. Finally, the argument can be made that the victim may maintain a presence 
3The gasoline SBC laws are industry specific laws that go beyond the Sherman 
Act (U.S. C., title 15, sec. 1-7) and Robinson-Patman Act (U.S.C., title 15, sec. 13) to 
protect independent gasoline dealers. 
8 
because she realizes the lower prices as predatory and not a change in the market or the 
level of competition. McGee argues that in such a case, a victim of predatory pricing 
would certainly want to maintain her presence since prices will, at worst, revert back to 
their previous levels.4 
Besides the predator having a cost advantage, the second condition that must be 
met is that the future expected benefits of predation must outweigh the. costs. Predatory 
pricing strategies are costly, both to the predator and the victim. It would seem plausible, 
therefore, to expect that a firm considering a predatory pricing strategy would not do so if 
the costs were greater than the expected gain. The question then becomes - what are 
these costs and is their magnitude offset by the future gains? The higher these costs 
appear to be, the less likely is predatory behavior. The magnitude of the costs will largely 
depend on the structure of the industry and the length of the predatory period. However, 
it is possible to .obtain a general perspective of the potential costs facing a would-b~ 
predator. 
The most obvious of these costs is incurred through setting a price that yields less 
than a normal rate of return. It has already been pointed out that these costs will be 
heightened by the need. on the part of the predator to expand output. Of additional 
interest, however, is the discounted value of future profits. Correctly analyzing the . 
situation faced by the predator requires the realization that one dollar received in the 
· future does not equal one dollar today. In present value terms, and at an interest rate of 
4John S. McGee, "Predatory Pricing Revisited," The Journal of Law and 
Economics 23, no. 2 (Oct. 1980): 296. 
9 
10 percent, one dollar received one year from now is worth slightly less than 91 cents. If 
that same dollar is not received until 10 years from the present date, it is worth a little less 
than 39 cents. As you go further out in time long term predatory practices become 
increasingly costly and it becomes less likely that future benefits will outweigh the costs 
of predation. The U.S. Supreme Court's 1986 decision in Matsushita Electric Industrial 
Co .. Ltd. v.Zenith Radio Corporation et al. 106 S. Ct. 1348 (1986) exemplifies this point. 
U.S. firms charged that below cost pricing had been used by several Japanese firms with 
predatory intent. The allegation that this practice had been occurring for 20 years led the 
Court to rule in favor of Matsushita, stating that it was unlikely that any firm could carry 
out a predatory strategy over a period of 20 years. 5 
Additionally, if the predatory strategy is to be successful, the current competition 
must not only be driven out of business, but must be kept from re-entering the market at 
some point in the future. Likewise, other potential competitors must be discouraged from 
entering if the current competition is eliminated. Realizing that she must somehow limit 
future entry, the predator must make it difficult, or preferably impossible, for the 
competitor to return or a new competitor to move in. Even if the predation is successful, 
the assets of the firm will remain and offer an opportunity for a relatively easy return by 
the victim or someone else. A solution to this dilemma is for the predator to not only 
eliminate the competitor, but her assets as well. Unless the value of the assets have for 
some reason been substantially depreciated, purchasing them is likely to be a costly 
5Matsushita Electric Industrial Co., Ltd. v Zenith Radio Corporation et al., 106 
S.Ct. 1348 (1986). 
10 
endeavor. Yet, to protect the market power the predator fought for, she must either 
incur this cost, or somehow make the assets unavailable to others. 
It follows from the above that if the industry is characterized by barriers to entry, 
predation may be more attractive. Given substantial barriers to entry, a firm that gains 
monopoly power will find it easier to protect its position when prices are eventually 
elevated. The barriers to entry or advantages possessed by the incumbent firm have been 
classified in three ways by Joe Bain.6 Bain argues that entry can occur easily in the 
absence of 1) an absolute cost advantage, 2) advantages resulting from product 
differentiation, and 3) significant economies of large scale. As an industry moves away 
from a definition of easy entry, an incumbent firm will be characterized by at least one of 
these three advantages. Each of these advantages implies that a potential entrant will face 
a relatively higher level of costs than the incumbent currently incurs or has incurred. If 
this is true, the new entrant will have difficulty if the incumbent firm lowers price even to 
the level of its current average costs. A similar result may occur if the industry is 
characterized by high sunk costs. In the presence of a credible predatory threat, a 
potential entrant may be discouraged from entering the market if the firm faces the future 
probability of later being driven from the market and thus forfeiting the resources 
.associated with the sunk costs.7 
6Joe S. Bain, Barriers to New Competition (Cambridge: Harvard University 
Press, 1956), 12. 
7Dennis W. Carlton and Jeffrey M. Perloff, Modem Industrial Organization, 
2d ed. (New York: Harper Collins College Publishers, 1994), 387. 
11 
With these conditions and their caveats in mind, we now consider the relationship 
between costs and the determination of when a price is predatory. The difficulty of 
accurately measuring costs and deciding which costs to consider has led to considerable 
debate over this issue. At the heart of this debate is a concern over destroying legitimate 
competition through a definition of cost that fails to capture true predatory intent. 
Motivated by this concern, Phillip Areeda and Donald F. Turner sought to apply 
economic analysis to a workable definition of predatory pricing. 8 In doing so they hoped 
to offer a well reasoned means by which a predatory price could be distinguished from 
competitive pricing. Their analysis and the resulting rule they propose for the 
determination of a predatory price is worth examining as it has been used in practice by 
the courts.9 
The Areeda and Turner rule, as it has become known, is based on the relationship 
between price and the basic measures of cost used in economic theory. Following their 
lead, a graphical representation of average cost, average variable cost, and marginal cost 
will be utilized. Figure 1 shows the typical cost functions for a firm. In a perfectly 
competitive market, each firm takes the price as given, and as such, faces a 
horizontal demand function at the competitive price. Profit maximization requires that 
price equals margin~l cost. In the long run, all firms earn zero economic profits so that 
the competitive price will be dictated by the intersection of marginal cost and the local 
8Areeda and Turner, 697-733. 
9Carlton and Perl off, 3 89. 
12 
minimum of the average cost curve. Thus, in a perfectly competitive market, the price 
will be P c as indicated in Figure 1. The profit maximizing monopolist, on the other. 
hand, can affe_ct price by producing the quantity at which marginal cost equals marginal 
revenue. She may then receive a higher price associated with a lower quantity. Prices at 
these profit maximizing (or loss minimizing) levels should not be considered predatory. 
Figure I. The Cost Functions of a Typical Firm 
$/q AC 
AVC 
q 
On the other hand, prices below this level may indicate a voluntary sacrifice of 
short run profits. The incentives of such a firm may be legitimately called into question, 
and suggest that predatory pricing has occurred. Areeda and Turner argue the necessity 
of such a condition, but guard against its sufficiency. It is possible that a frrni could 
13 
legitimately choose to price below the profit maximizing level. An example given by 
Areeda and Turner are new firms utilizing low prices to establish themselves in a 
market. 10 Additional qualifications are therefore necessary. 
Their analysis initially focuses on prices that are below the profit maximizing · 
level and either at or above average cost. A price at average cost indicates that a finn's 
total revenues exactly offset its total costs. As before, the potential exists that firms will 
be eliminated by such prices, perhaps intentionally. These are less efficient firms than the 
predator finn and will suffer greater losses. But such a pricing scheme is also consistent 
with competition. Areeda and Turner argue that even in cases when a price above 
average cost is exclusionary, it should not be considered a violation of antitrust laws. 
Prices below average cost indicate that a finn is operating at a loss. Areeda and 
Turner are careful to point out that this condition does not necessarily imply predatory 
intent. In this case, a finn can be operating under conditions of loss minimization instead 
of profit maximization.11 It is, of course, difficult in practice to determine when a finn 
may be trying to minimize its losses versus attempting to eliminate its rivals. Since such 
prices may indicate the possibility of predatory intent, and since equally efficient rivals 
will suffer losses under these conditions, a more precise definition of cost is needed. 
To accomplish this, Areeda and Turner further delineate prices mto two 
10Areeda and Turner, 703. 
11The term loss minimization is used in place of profit maximization only because 
price is below average cost. As such, a finn is operating at a loss and the optimal policy 
is to minimize losses. Loss minimization and profit maximization both imply that a finn 
is producing its optimal output. 
14 
categories. The first includes prices at or above marginal cost. Referring to Figure 1, this 
would coincide with a 'price/quantity combination such as point A. Equally efficient 
firms will be operating at a loss due to the pricing practices of the predator. However, as 
has been previously noted, it is not likely that equally efficient firms will be driven out 
under this condition. A possible exception is. a firm without compar~ble financial 
resources. 12 The assumption that outside financing is not available, however, has limited 
support. Establishing a price floor above marginal cost would therefore encourage 
inefficiency and adversely affect competition. For this reason, Areeda and Turner argue 
that prices in this range should not be considered predatory.13 
The second case includes prices that are below marginal cost. It is possible that 
price is below marginal cost and yet above average cost. In this case, neither the firm, 
nor its equally efficient rival are operating at a loss, even though the social optimum 
is not being reached. As such, prices above average cost remain non-predatory in terms 
of anti-trust law. 
Prices that are below both average cost and marginal cost potentially imply a 
different story. This corresponds with a price/quantity combination such as point Bin 
Figure 1. As previously, this scenari<? could exclude less efficient rivals while firms that 
12Areeda and Turner, 710. 
13 Areeda and Turner make several additional arguments to support this position. 
Prices at marginal cost imply thafthe social optimum is being reached. Forcing a firm to 
charge a higher price would decrease the quantity sold and create a deadweight loss. 
Also, such a policy would require that a price floor be created that was no higher than the 
loss-minimizing level of the firm. Calculating this price would impose prohibitively high 
administrative costs. (Areeda and Turner, 711.) 
15 
are as efficient will not suffer greater losses. However, at these levels, the additional 
increment to total cost of the last unit produced will exceed its price. The incentives of a 
firm selling at this price level can increasingly be called into question since it could 
reduce these losses by selling a lower quantity. Also, when price is less than marginal 
cost the social optimum is not being reached and a price floor above this level would 
decrease the loss to society. These conclusions lead Areeda and Turner to consider such 
prices as predatory. 14 
A potential argument offered in defense of such a price level is that the accused 
firm is meeting the lawful price of its competition. Most of the SBC laws included in 
this study allow this defense. Areeda and Turner argue that such a defense should not be 
valid for a price below both average cost and marginal cost. A firm that desires to enter 
a market may legally utilize such prices to obtain a market share. However, an 
incumbent firm, following the low price in an attempt to thwart new entry, is threatening 
legitimate competition. 
This leaves the final, necessary point concerning the Areeda and Turner rule. 
While defining marginal cost in economic terms is not exceedingly difficult, measuring 
marginal cost in practical terms is. That is, determining the additional increment to cost 
of an additional unit of output is not readily obtainable from typical accounting records. 
As such, the application of a marginal cost rule to real world settings is problematic. 
Areeda and Turner suggest instead the more observable average variable cost be used as a 
14Areeda and Turner, 712. 
16 
proxy for marginal cost. An average variable cost rule will either be more prohibitive 
than, identical to, or less prohibitive than a marginal cost rule. Figure 2 shows the same 
cost functions as in Figure 1, but with three points (marked A, B, and C) to indicate price 
ranges for these three cases. Prices like C will conform with the marginal cost rule and 
using average variable cost as a proxy will have no effect. However, at point A, average 
variable cost is above marginal cost. The average variable cost rule will thus classify 
prices as predatory when it should perhaps not. A firm cannot be minimizing its losses, 
however, when price is less than average variable cost. Its optimal choice is to shut 
down. This would partially diminish the restrictiveness of an average variable cost rule. 
At the point marked B, the opposite case prevails. Marginal cost is now above 
average variable cost and the average variable cost rule will allow prices that may be 
predatory. Areeda and Turner point out that predatory pricing under this scenario is 
particularly unlikely. With the typical cost functions, as depicted in Figure 2, the output 
implied within the range where marginal cost is greater than average variable cost 
indicates that the finn is operating near, or beyond, the optimum for that firm. 
Additional output will push the finn beyond the optimal capacity and thus force it to 
utilize more costly resources. As indicated previously, a predatory scheme involving 
lower prices will force the predator firm to increase output. However, if it is realized that 
additional output is increasingly costly, the likelihood of predation decreases. For these 
reasons, Areeda and Turner argue that an average variable cost rule is unlikely to have 
adverse effects on competition. 
17 
Figure 2. Price/Quantity Combinations Under the Average Variable Cost Rule 
$/q AC 
AVC 
q 
The rule's foundational characteristic is that efficiency matters. Competition that 
excludes relatively inefficient firms can'be efficiency enhancing, maximizing social 
welfare. However, debate exists over Areeda and Turner's position and their analysis. 
The disagreements are based on two points. First, some writers have oojected to the 
concept that less efficient firms don't matter in terms of anti-trust law. Connectedly, it 
has been argued that pre-existing monopolies will be capable of deterring entry through 
strategic behavior. Second, there is concern that social welfare maximization needs to 
take a more preeminent focus in the definition of predatory pricing. 
18 
The ability to deter new entrants partly determines the probability of predatory 
pricing. If a finn has monopoly power, it must maintain a credible threat by convincing 
other firms that they will lose money by entering. To accomplish this, the monopolist 
may seek to construct plants and purchase equipment so that in the advent of entry, the 
quantity it sells will yield a price that is unacceptable to an entrant. 
Williamson emphasizes this strategic· consideration and modifies it to account for 
the existence of.rules such as Areeda and Turner's average variable cost rule. He, 
however, proposes three alternative rules. 15 The first is referred to as the "Output 
Restriction Rule". A monopolist would not be allowed to increase output above the level 
that existed before the probability of entry until after a 12-18 month period during which 
the entrant could become established. The second ruie would allow the inctimbent firm 
to increase output to the level at which price was greater than or equal to short run 
marginal cost. Finally, a monopolist would be allowed to increase quantity to the point at 
· which price was at or above short run average cost. 
The monopolist's strategy is to choose an initial level of output based oh the 
current rule. For example, under the output restriction rule, the dominant firm will elect 
to initially produce a quantity such that the residual demand curve of a potential entrant is 
just tangent to its average cost curve. Thus, the best the entrant could do is break even. 
Under the second or third rule, the monopolist's initial quantity can be less. If entry 
150liver E. Williamson, "Predatory Pricing: A Strategic and Welfare Analysis," 
The Yale Law Journal 87, no. 2 (1977): 295-301. 
19 
occurred, the firm could increase quantity such that the price remained legal, yet was 
sufficient to deter entry. This would result in a residual demand curve determined by the 
lowest allowable price. Williamson suggests that the appropriate rule is that which yields 
the greatest level of social welfare. 16 
Scherer, likewise expressed disagreement with the Areeda and Turner rule and in 
part focuses on the issue of entry deterrence. 17 He argues that if a monopolist has an 
advantage because of economies of scale, the entrant may possibly face a minimum level 
of output below which it cannot profitably enter. The monopolist realizes this and creates 
a credible threat to increase output in the presence of entry. Its total output plus the 
entrants minimum level of output would drive price below the minimum of average cost. 
The monopolist may thus price above average cost and yet such pricing can be 
exclusionary. 
Additionally, Scherer considers the situation when a firm is operating with excess 
capacity.18 If price is also below average cost, the firm is operating at a loss. Scherer 
again assumes a pre-existing monopoly and then searches for plausible reasons for a firm 
to structure its plant so as to have excess capacity. One reason is that the firm has 
strategically positioned itself for the purpose of entry deterrence. Moreover, he argues 
that maintaining such excess capacity during non-predatory periods would create 
16Ibid., 307. 
17F.M. Scherer, "Predatory Pricing and the Sherman Act: A Comment," Harvard 
Law Review 89 (1976):869-889. 
18Ibid., 875 
20 
unnecessary welfare losses. 
Scherer also disputes Areeda and Turner's reliance on short run marginal cost for 
determining the gains, or losses, in social welfare of different price levels. 19 The basis of 
this argument is that pricing policies should be judged, in part, by their divergence from 
socially optimal levels. Prices that are above average cost and below marginal cost, for 
example? should be judged on a "proper resource allocation" criteria rather than on the 
relative efficiency of fmns.2° The rule derived from this is based on current and long run 
welfare losses. As such, the discounted value of future losses (e.g., the deadweight loss 
to society of monopoly prices) must be added to any current losses resulting from 
deviations of price from marginal cost. 
The above challenges to the Areeda and Turner rule assume a pre-existing 
monopoly and, as such, focus on entry deterrence. It is unlikely, however, that such 
strategic behavior would be successful if both firms have identical cost functions since 
the incumbent firm cannot create a credible threat.21 Also, a rule to determine predatory 
pricing based. on long run social welfare is likely to be as, or more, difficult to determine 
in practice as a marginal cost rule. A rule based on social welfare, if enforceable, may 
also be excessively restrictive in cases where price exceeds average cost. 
The Areeda and Turner rule may, likewise, be difficult to define and enforce in 
19lbid., 885. 
20lbid.; 884. 
21Carlton and Perloff, 396. 
21 
practice. The analysis does, however, provide a generally accepted framework that 
provides an opportunity to identify predatory pricing. For this reason it is a useful tool to 
apply to the retail gasoline industry as well as to gasoline SBC laws. If Areeda and 
Turner are going to err, they choose to err on the side of not adversely affecting legitimate 
competition by adopting a rule that is unnecessarily restrictive. Some may argue that the 
average variable cost rule is therefore too lenient. In light of the weak evidence 
supporting a preponderance of actual predatory pricing cases, Areeda and Turner's 
analysis seems appropriate.22 
Applying the Areeda/Turner Rule to SBC Laws 
How then can this general theory of predatory pricing and the determination of a 
predatory price be applied to the gasoline SBC laws? First, it may be of use to look at the 
three necessary conditions for predatory pricing and their relevance to this market. This 
will not indicate whether predatory pricing is occurring within the industry. Rather, it 
will offer an idea of the probability that such actions are likely to occur and permit 
conclusions to be drawn about the effects of the SBC laws. 
22There are a limited number of cases in which firms have been found guilty of 
predatory pricing. See, for ·example, Frank H. Easterbrook, "Predatory Strategies and 
Counterstrategies," The University of Chicago Law Review 48, no. 2 (1981 ): 313,4. 
Studies of such cases often indicate that predation in terms of pricing below cost is ·a rare 
occurrenc~. A counter example, however, is given by Malcolm R. Burns, "Predatory 
Pricing and the Acquisition Cost of Competitors," Journal of Political Economy 94, no.2 
(April1986):266-296. He empirically examines the record of the American Tobacco 
Company's purchases of its rivals between 1891 and 1906 and finds evidence of 
predatory pricing. 
22 
It has been argued that a predator firm must have a cost advantage over its rival. 
Conceivably, such an environment may exist in the retail gasoline industry. It is the 
larger firms that are most often accused of predation on the smaller, independent 
firms and it is possible that these firms have lower costs than their rivals. Larger firms 
may be able to supply a network of their stations more efficiently than a single 
independent firm can obtain its supply. Volume discounts may also favor larger 
competitors. It is important to note, however, that a cost advantage does not imply the 
ability nor the desire to engage in predatory behavior. The indication is only that large 
firms may be more efficient under certain circumstances. 
To be successful at predation future expected benefits must exceed the costs of 
· predation. Several of these costs were described above and the importance of discounting 
values was noted. Appealing to the belief that firms and individuals act in a rational 
manner allows one to infer that a frrm will not choose to undertake a predatory strategy if 
the costs outweigh the expected benefits. Unless it is expected that the rival will be 
driven from business in a relatively short period of time, these costs are likely to be quite 
high relative to the expected gains. 
As was pointed out previously, the predator firm will have to increase the volume 
it is willing to sell to lower price. However, industry wide demand has been shown to be 
quite price inelastic in the short run.23 Thus, it could be argued that the increase in 
quantity demanded due to a decrease in price would be inconsequential. But, if the rival 
23Carol Dahl and Thomas Sterner, "Analyzing Gasoline Demand Elasticities: A 
Survey," Energy Economics July (1991): 203-210 .. 
23 
firm decreases the quantity it sells at the lower price, the predator firm's losses will be 
greater than that associated with its typical sales volume. 
The primary costs associated with predation in the retail gasoline industry are the 
returns relinquished because of the lower price. It is likely that these costs can be 
computed with a high degree of certainty. The future profits with which these costs are 
compared will not be as certain. One reason for this uncertainty is the difficulty of 
predicting the likelihood of deterring entry. Gasoline retail outlets contain rather 
specialized assets in the form of their pumps and storage tanks. As a result, even if the 
rival firm goes out of business, the most likely buyer of the property would be someone 
interested in operating the same type of business. The predator firm would then be forced 
into a position where it would have to consider either purchasing the property or risk 
allowing the potential of a new entrant. 
The likelihood of deterring entry in the retail gasoline business for any 
meaningful length of time appears small. The higher prices that will exist after predation 
will certainly attract potential competitors. Moreover, the costs of building a new retail 
gasoline outlet, or of purchasing an existing outlet, is not prohibitive. Thus, the retail 
gasoline business does not appear to be characterized by any significant entry barners that 
could substantially restrict the number of retail outlets in a given location. Advantages 
may exist for larger firms in certain geographic areas, but this is not indicative of 
predatory practices. Although the possibility of predatory practices within the retail 
gasoline industry cannot be ruled out, the probability that it will be successful appears 
remote. 
24 
The use of the Areeda and Turner rule to detect predatory pricing allows a 
generally accepted framework by which the SBC laws can be compared. It is common for 
SBC laws to defme cost by one of two methods. The first, and most common, is to list 
the relevant components of cost. This can differ between states, depending on the 
definition of "the cost of doing business". The second method is to state a minimum 
markup (typically 6 percent) as the legitimate cost of doing business.24 This amount is 
then added to the delivered cost of the fuel, which includes the wholesale price and taxes. 
These various definitions of cost can be compared with the average variable cost rule 
suggested by Areeda and Turner and two observations can be made. First, those states 
that mandate a minimum markup as part of the definition of cost have the potential of 
setting a price floor in the short run that is above average variable cost and perhaps above 
average cost. The arbitrary nature of a percentage markup seems to preclude certain 
companies that may, for various reasons, be capable of more efficient operation than 
others. This appears to be precisely contrary to the care taken by Areeda and Turner to 
not unnecessarily harm legitimate competition. Second, the explicit definition of cost 
relied upon by the majority of the laws includes items that would normally be considered 
fixed costs. 25 For example, rental value of the property is included as an expense in the · 
24Also, a cost survey within the relevant market area is allowed as sufficient 
evidence of cost in some states. These surveys are, however, subject to the components 
of cost as defmed within the particular state's SBC law. 
25Fixed Costs are defmed as costs that do not change with the level of output. In 
particular, Alabama and Massachusetts include depreciation in their determination of cost 
and most other states include the fair rental value of the land and buildings. 
25 
majority of the laws. But this should be considered a fixed cost, not a variable cost. 
These two observations seem to point to a common result. If Areeda and Turner's 
analysis is correct, the price levels established by the SBC laws may exclude legitimately 
competitive behavior on the part of some companies within the industry. The possibility 
of including fixed costs in an equation that determines a predatory price may create a 
price floor that is above average cost for some firms. In this case, the laws may have the 
effect of causing retail prices, and therefore retail margins, to be higher than they 
otherwise would be. 
Alternative Explanations for SBC Laws 
Although SBC laws were supposedly designed to promote efficient pricing by 
reducing the threat of predatory pricing, there are other, more plausible, rationales for 
their existence. If independent dealers are struggling to compete due to changing market 
conditions conducive to large scale ownership, the laws may instead protect their 
·interests. Analysis of the legislative history ofthe Montana SBC law, for example, 
indicates that independent petroleum marketers and retailers initiated and supported the 
bill through its passage. In Montana, refiner ownership of retail outlets is uncommon. 26 
Independents, instead, accuse chain retailers of attempts to monopolize local marketsP 
26Motor Fuel Pricing Problems, prepared by Paul E. Verdon, StaffResearcher, 
(Helena, MT: Montana Legislative Council, 1990), 11-15. 
27Mike Dennison, "Restriction on Gasoline Prices Widely Criticized," Great Falls 
Tribune, 25 Sept. 1995, 2(A). 
26 
At a national level, major oil companies are the accused predators.28 This poses the 
question of whether the independent's belief of predation is accurate or if SBC legislation 
is merely protecting less efficient firms. 
The key to discerning which explanation is correct is a comparison of retail 
margins in localities with SBC laws and those where such laws are abs.ent. If the 
occurrence of predatory pricing is highly unlikely, as argued here, and if firms are either 
constrained by these laws from pricing below an established level or are afraid to 
compete in terms of low price, the effect will be higher retail margins. 29 Again, this 
hypothesis is conditional on SBC laws creating a binding price floor. If the laws are 
binding so that firms believe they will be accused of illegal pricing, an additional 
expected result is that retail prices in law states will respond more quickly to increases in 
wholesale prices than in non-law states. Firms operating with relatively low margins 
would have an incentive to increase their retail price quickly so as to not be accused of 
below cost pricing. 
Conclusion 
The theoretical considerations presented in this chapter began as a general 
28Caleb Solomon, "Independent Gas Stations Cry Foul Over Price Wars," Wall 
Street Journal, 1 April1991, 7(B). 
290n the other hand, if predatory practices are occurring and are yielding the 
alleged results, retail margins should, on average, be higher in states without SBC laws. 
This alternative hypothesis assumes that by examining cross sectional data, as will be 
done in this study, the researcher obtains results that reflect long run adjustments. 
27 
framework in which the occurrence of predatory pricing could be determined. The 
analysis of predatory pricing by Areeda and Turner and their ensuing definition of a 
predatory price was presented as a generally accepted means of determining whether 
predatory pricing had occurred. The issue of predatory pricing in the retail gasoline 
industry was then considered. This· behavior was determined to be highly unlikely, 
suggesting that retail margins in states with SBC laws would likely be higher than states 
without such laws. 
28 
CHAPTER3 
DATA DESCRIPTION AND LEGAL REVIEW 
Introduction 
A price floor that is binding is expected to alter the pricing practices of firms in 
that industry. But econometric models are necessary to empirically test this hypothesis 
and determine the magnitude of the effect. Estimation requires data on retail gasoline 
prices and other variables that may affect price. Retail wages, for example, can vary 
across localities and their affect on retail price must be accounted for to accurately 
measure the effects of the SBC laws. 
The purpose of this chapter is, in part, to describe the sources and calculations of 
the data necessary to test the effects of the SBC laws. As described in the previous 
section, SBC laws may result in higher retail margins. The calculation of the retail 
margin from the price data is, therefore, critical and requires explanation. Likewise, the 
data on additional variables that affect price will be described. Of particular interest and 
relative complexity is the derivation of a variable that will act as a proxy for seasonal 
demand changes. In addition, specific SBC laws will be described. These laws have 
resulted in litigation_and the major cases are examined. This legal review will offer a 
better understanding of how the various laws define cost and how actively these laws are 
enforced. The analysis begins with a brieflook at the gasoline distribution system in the 
29 
United States. 
An Overview of the Gasoline Distribution System 
The gasoline distribution system starts at the level of the refinery, where crude oil 
is converted into a variety of products, one of which is gasoline. The majority of gasoline 
consumed in the United States is also refined within the U.S., but a small percentage is 
imported. 1 From these sources, the fuel is sent, typically by pipeline, to a network of 
marketing terminals from where it is transported via truck to retail outlets. Direct supply 
outlets purchase fuel delivered from the terminals by the refiner. Such stations may be 
owned and operated by the refiner or by independent dealers. Alternatively, the gasoline 
may be purchased at the terminal by an independent business for the purpose of supplying 
its own independent stations or for reselling to other dealers. · People that provide this 
service are typically referred to as "jobbers". The jobber may also be under contract to a 
particular refiner to distribute the refiner's brand to branded retail outlets. These outlets 
may be owned by the refiner, the jobber, or another independent dealt(r who sells that 
brand name. Conversely, the jobber may also purchase unbranded fuel for distribution to 
independent stations operating under a non-refmer brand name. Figure 3 offers a 
schematic representation of this system. 
1According to the Department of Energy, approximately 3.3 percent of the total, 
annual fmished motor gasoline supplied was imported. Department of Energy, Energy 
Information Administration, Petroleum Supply Annual1993 (Washington, D.C.: U.S. 
Government Printing Office, 1994), Table 2. 
30 
Figure 3. The Gasoline Distribution System 
I Domestic Refinery I 
..., I Imported Gasoline I 
-
l 
'If 
I Marketing Terminals I 
, ... ,if 
Direct Supply Stations I Jobbers I 
-Refiner Owned or Independent / ~ 
I Independent Branded Stations I I Independent Unbranded Stations 
'{ 
Refiner Owned Stations 
The Components of Gasoline Price 
The retail price of gasoline is composed of several elements. At a basic level, 
these components are: 1) the wholesale price, 2) taxes, and 3) the retail margin. The 
empirical analysis will examine differences in retail margins across cities in the United 
States. For the period examined, weekly wholesale and retail prices were obtained from 
31 
The Oil and Gas Journal.2 The retail prices are daily averages of self service, unleaded 
gasoline, reported on a weekly basis from March 23, 1992 to De.c. 27, 199~. The 
wholesale prices are an average of branded and unbranded rack prices for the day of the 
reported retail prices. Of the 42 cities represented in the survey, two had observations 
that appear to be full service, retail prices. These cities 'were dropped from the sample. 
The following sections provide a more thorough description of the three components. 
Wholesale Price 
The most precise measure of the wholesale price paid by a typical gasoline station 
would be the retailer's invoice cost of the fuel purchased. In the absence of such 
information, it is necessary to rely on the posted prices given at the city terminals. 
Wholesale prices are generally reported as one of three distinct types: 1) rack, 2) dealer 
tankwagon (DTW), or 3) spot. The terminal prices, reported and published daily, are 
either rack or DTW.3 
2Severin Borenstein and Andrea Shepard, "Dynamic Pricing in Retail Gasoline 
Markets," (National Bureau of Economic Research, Research Paper No. 1269, Sept., 
1993), and Severin Borenstein, A. Colin Cameron, and Richard Gilbert, "Do Gasoline 
Prices Respond Asymmetrically to Crude Oil Price Changes," (Program On Workable 
Energy Regulation, University of California Energy Institute, PWP-00 1 R, June, 1994), 
report that retail prices from the Oil and Gas Journal (OGJ) are estimated from wholesale 
prices. The OGJ confirms this for the period previous to March 23, 1992. From March 
23, 1992 to Dec. 28, 1993 (also the time period considered by this study), the OGJ's 
source for retail prices was the Computer Petroleum Corporation (CPC). The CPC has 
said that retail prices supplied to the OGJ during this period were surveyed prices from 
weekly telephone surveys of the given metropolitan areas. Of approximately 40 stations 
in each city that participate in the surveys, 15 - 20 are chosen randomly to be surveyed on 
a particular day. The wholesale prices were also collected by CPC during this period. 
3Some of the other recognized sources for wholesale price data are The Computer 
32 
The rack price is the most commonly reported and referenced wholesale gasoline 
price. It assumes pickup of the fuel at the terminal and is the price commonly paid by 
jobbers. The rack price is further delineated as either branded or unbranded to signify 
whether the fuel carries the name of the company responsible for its refining (e.g. Exxon). 
The difference between branded and imbranded rack prices is often minimal, with 
branded rack prices generally being higher than unbranded rack prices. The higher 
branded price reflects the reputation associated with a brand name versus a "generic" fuel. 
Dealers who lease stations from, or who are otherwise under a direct contract with 
a refmer, typically pay the DTW price. Unlike the rack price, the DTW price presumes 
delivery to the station from the terminal. Dealers paying the DTW price are typically 
under a long term contract with the refiner.4 
Spot prices are determined by an auction process at primary trading centers such 
as New York or Houston. 5 Large quantities of fuel are traded at prices that include 
minimal storage, transportation, and marketing costs. 6 
The different wholesale prices generally follow a predictable pattern whereby 
Petroleum Corporation, U.S. Oil Week's Price Monitor, Platt's Oilgram Price Report, and 
The Oil Daily. 
4Philip E. Sorenson et. al., "An Economic Analysis of the Distributor-Dealer 
Wholesale Gasoline Price Inversion of 1990: The· Effects of Different Contractual 
Relations," (Washington, D.C.: American Petroleum Institute, 1991), 7-9. 
5lbid., 7. 
6 American Petroleum Institute, Policy Analysis Department, "An Overview of 
Gasoline Prices and Their Determination," (Washington, D.C.: American Petroleum 
Institute, 1990), 8. 
33 
DTW prices are highest, followed by branded rack, unbranded rack, and spot prices 
respectively.7 The higher DTW prices not only reflect the cost of delivery to the station, 
but also a low level of investment risk by the dealer and a high level of contractual 
commitment from the refiner. 8 
In addition, it should be noted that reported prices do not precisely indicate the 
actual price paid by either a reseller or a dealer. The actual invoice cost is typically 
discounted from the posted wholesale price. A source within the industry has indicated 
that these discounts represent rebates for higher volume purchases, short term incentives 
to purchase additional product, or perhaps to compensate a branded retail outlet for 
improvements to the land or building. As noted above, the data used here are an average 
of branded and unbranded rack prices. 
An accurate calculation of each city's retail margin requires the careful 
determination of the total amount of gasoline. taxes in each city. Motor fuels taxes are 
collected by federal, state, and local governments and, as such, differ from state to state 
and potentially between cities within the same state. All states impose a per gallon tax. 
In addition, some states have instituted state sales taxes that apply to some or all motor 
fuels sales. County and local governments likewise may have sales or per gallon taxes. 
Additional motor fuels taxes may be imposed at the state level, but are not always 
7Ibid., 11. 
8Sorenson et. al., 9. 
34 
referred to as motor fuels taxes. The petroleum products gross receipts tax in New Jersey 
is one example. Environmental protection fees are also often imposed by states on sellers 
of motor fuels. However, because these fees are not always based on a per gallon basis, 
they are not included in our measure of taxes. 
For each city in the sample, it was necessary to determine the applicable taxes. 
Several published sources were consulted for this information. 9 In addition, state 
departments of revenue and taxation were consulted whenever sources did not coincide as 
well as to determine the correct calculation methods of any taxes applied as a percentage 
of sales (sales tax). 
Taxes imposed on a per gallon basis are easily handled, all that is necessary is that 
one subtract the amount of the tax from the retail price to arrive at a net of tax price. 
However, for those states and cities that iffipose a sales tax on gasoline, additional 
calculations are required. The problem arises because states and localities can differ as to 
the base upon which the sales tax is computed. Some states include the federal and state 
per gallon taxes when computing the state and/or local sales tax. Other states take a 
credit for the state and/or federal per gallon tax before computing the sales tax. 
Accordingly, it is first necessary to determine the basis upon which the sales tax is 
9Sources include the U.S. Department of Transportation, Federal Highway 
Administration, Monthly Motor Fuel Reported by States (Washington, D.C.: U.S. 
Government Printing Office, 1992, 93) various issues, Table MF-121T; The Road 
Information Program, 1993 State Highway Funding Methods (Washington D.C.: TRIP, 
1993), 13; American Petroleum Institute, State Governmental Relations Department, 
"State Gasoline Excise Tax Rankings- May 4, 1993" (Washington, D.C.: American 
Petroleum Institute, 1993). 
35 
computed. 
To illustrate how taxes were measured, consider the city of Atlanta on June 7, 
1993. The relevant price and tax data in cents per gallon and on a percentage basis are as 
follows: 
Wholesale Price= 55.3 
Retail Price= 96.2 
State Tax = 7.5 
Federal Tax= 14.1 
State Sales Tax = 4% 
Local Sales Tax= 1% 
Here, the sales tax on gasoline is calculated after taking a credit for the state 
tax of 7.5 cents. Also, the local sales tax is administered by the state, so that the 
taxes can be simultaneously computed as a 5 percent sales tax. The price per gallon net 
of tax includes the margin, which is unknown. Calculation of the sales tax without prior 
knowledge of the margin, therefore, requires one to work backwards from the retail price. 
Crediting the state tax from the retail price yields the sales tax base. It is then possible to 
determine the taxable selling price as shown in Table 1. The per gallon sales tax is the 
difference between the sales tax base and the taxable selling price. Appendix C details 
the motor fuels taxes for each city. 
Retail Margin . 
Subtracting the wholesale price and all applicable taxes from the retail price yields 
the retail niargin. This amount represents, on a per gallon basis, the additional costs of 
operating a retail station such as payroll, rent, or depreciation and any profits. 
Differences in costs between cities will therefore affect the retail margin. Thus, it is 
36 
Table 1. Calculation of Gasoline State Sales Tax 
Retail Price 96.20 
Tax Credits: 
Federal Tax 0.00 
State Tax 7.50 
Total Tax Credits (7.50) 
Sales Tax Base 88.70 
Taxable Selling Price 
(Adjusted Retail Price/1.05) (84.48) 
Sales Tax 4.22 
important to control for these costs if we are to measure the effect of SBC laws. Other 
studies of the retail gasoline business have also used retail margins as the dependent 
variable. Borenstein uses retail margins to test whether there is price discrimination, 
based on a customer's willingness to switch stations, in the retail gasoline market.10 He 
points to the retail margin as ~ppropriate because he is concerned with the dealer's costs 
per transaction. In addition, Borenstein and Shepard study "implicit collusion" in the 
retail gasoline market and use retail margins as their dependent variable.11 
Variables that Affect the Retail Margin 
In the long run, the retail margin will be determined by the costs of operating a 
10Severin Borenstein, "Selling Costs and Switching Costs: Explaining Retail 
Gasoline Margins," Rand Journal of Economics 22, no. 3 (Autumn 1991):354-369. 
11Borenstein and Shepard, (1993). 
37 
station, and these costs can differ between cities. 12 Labor costs are one such cost. As a, 
proxy, we have computed an average retail wage for each city. This information was 
obtained from the 1992 census of retail trade. 13 Data for the flrst quarter payroll for all 
retail establishments was divided by the number of paid employees for the corresponding 
pay period. It is expected that the retail margin will be positively affected by differences 
in the retail wage between cities. 
In the long run, the cost of building, buying, or leasing a retail outlet will also 
affect the margin. In the absence of data specific to the retail gasoline business, the 
median value of housing for the metropolitan statistical area has been used as a proxy. 
This property value estimate is from the 1990 census of population.14 It is expected that 
higher property values will result in higher retail margins. 
Margins may also be affected by the cost of transporting fuel. If higher 
·population densities are correlated with a higher density of retail outlets, flrms in such 
demographic areas may be capable of more efficient distribution. Margins will, as a 
result, be lower. Also, sources in the industry indicate that higher population areas are 
12V ariables that change over time as well as across cities would be preferable, but 
this information is not available. 
13Department of Commerce, Economics and Statistics Administration, 1m 
Census ofRetail Trade (Washington, D;C.: U.S. Government Printing Office, 1994), 
Table 5. The retail census also includes payroll information for gasoline service stations 
(SIC 554) by city. This data was also used in the regression equations with similar· 
results. 
14Department of Commerce, Economics and Statistics Administration, County and 
City DataBook: 1994 (Washington, D.C.: U.S. Government Printing Office, 1994). 
38 
more likely to have refiner owned stations. Since population and population density 
are highly correlated, the population density variable will take into account differences in 
types of ownership between cities and account for overestimation of margins due to the 
use of an average of branded and unbranded rack prices.15 Density is measured as 
population per square mile. The population estimates are for 1992, based on the 1990 
census of population. 16 The land area is also based on the 1990 census. 
Seasonal changes in the quantity of gasoline demanded may also affect the retail 
margin. Sources within the industry agree that margins tend to be low, on average, 
during the winter. Starting in June and peaking in late July and early August, the margins 
increase in response to the higher demand recorded during this period. Although typical 
for the industry, this pattern can vary between cities. This indicates a market in which 
firms increase price in response to temporary increases in demand. 17 Likewise, retail 
prices will decline in periods of slack demand. In the presence ofhigh fixed costs, these 
temporary changes in demand do not induce either exit or entry of firms. 
Several sources of data on the quantity of gasoline sold are available. Two of 
these are from the Federal Highway Administration (FHA). The publication, Monthly 
15The average wholesale price is better represented as some weighted average of 
rack and DTW prices, depending on the number of ~efmer-owned stations. Since DTW 
prices B!e generally higher than rack, the margin will thus be overestimated when using 
an average of branded and unbranded rack prices. 
16Department of Commerce, County and City Data Book: 1994. 
170ver the long run, however, prices would likely approximate average cost. See, 
for example, American Petroleum Institute, An Overview of Gasoline Prices and Their · 
Determination, 1. 
39 
Motor Fuel Reported by States, records the number of gallons sold each month and is 
based on wholesale distributors' state tax reports. 18 The reported quantities, however, 
reflect time lags of up to six weeks before actual consumption. In addition, during the 
time period under consideration, several of the reported quantities appear to be 
inaccurate.19 Hi2hway Statistics, another FHA publication, also reports monthly 
quantities in a table entitled "Highway Use of Gasoline by Months". This data is also 
compiled from state tax reports, but are estimates of highway use of gasoline derived by 
subtracting an estimate of non-highway use from total use. 20 
An additional source is the Energy Information Administration's (EIA) Petroleum 
Marketing Monthly.21 These sales volumes are derived from EIA surveys of prime 
suppliers' sales to local distributors and retailers for consumption in a particular -state. 
Prime suppliers include refmers, importers, and other firms that sell gasoline across state 
lines. As such, this data may also reflect a time lag between the reported date and actual 
consumption. The use of this data is also made difficult by the implementation of a 
18Department of Transportation, Federal Highway Administration, Monthly Motor 
Fuels Reported by States, various dates, Table MF-33GA. 
19These inaccuracies are the result of reporting methods in several states that 
combine part of one month's total with the following month and are not a result of 
reporting errors on the part of the Federal Highway Administration. 
20Department of Transportation, Federal Highway Administration, Highway 
Statistics (Washington, D.C.: U.S. Government Printing Office, 1993), Table MF-26. 
21Department of Energy, Energy Information Administration, Petroleum 
Marketing Monthly (Washington, D.C.: U.S. Government Printing Office, 1993), Table 
47. 
40 
different survey method at the beginning of 1993. 
These sources create two additional complications. First, city level data are not 
available. The data are totals for an entire state. Seasonal demand for gasoline within a 
particular city may differ from the statewide totals, yielding an inaccurate representation 
. of the seasonal pattern. Second, the data are monthly totals whereas the price data are 
daily averages reported on a weekly basis. As such, a method of matching prices and 
quantities must be formulated. To accomplish this, weekly values for quantity are 
interpolated in such a manner as to preserve the seasonal trend in each state.22 These 
quantities are then normalized about each state's mean quantity. The rationale for 
normalizing the quantity variable is two fold. First, the estimation model uses time series 
cross sectionally pooled data and we desire a comparable measure for each city in the 
sample. Second, the use of normalized quantities negates the problem of collinearity that 
would otherwise result from high correlation between population and thus, population 
density, and quantity consumed. 
Adding the quantity variable to the regression equation, however, creates the 
potential for simultaneity bias. In this case, the retail margin is a function of quantity and 
the other independent variables. But quantity is a function of the retail price and 
therefore of the retail margin as well. The usual procedure for handling this problem is to 
utilize an instrument for the quantity variable that is purged of any price effect. 
To illustrate the procedure used in this study to purge the quantity/seasonality 
22See appendix A for a detailed description of the interpolation of the weekly 
quantity values. 
41 
variable of any price effect, consider the supply and demand functions shoWJ1 in Figure 
4. Let P 0 and Q0 be the initial equilibrium condition. If the demand function were to 
shift out to Db the new equilibrium price and output would be P1 and Q1• Because the 
supply function is upward sloping, price has increased. If price did not rise, a situation 
comparable to having a perfectly elastic supply function, then the quantity demanded 
would have been Q 1• Thus, purging the price effect from this increase in demand 
amounts tp obtaining an estimate of the quantity Q 1" Likewise, if demand were to 
decrease, an estimate of Q2 would be called for. It is apparent from the figure, 
however, that if one is to construct a quantity/seasonality variable that is purged of price 
effects, information on the underlying demand function is required. 
Consider a demand function for gasoline of the functional form, 
(3.1) 
. where Q is the observed weekly quantity detailed in appendix A, P is the observed retail 
price, 11 is the price elasticity of demand, and k is a vector of exogenous variables 
affecting the quantity demanded. By normalizi,ng the retail prices ·about the ·mean price 
for each city, equation (3.1) can be written as: 
(3.2) 
-Conceptually, P; is equivalent to P0 in Figure 4. Taking the natural logs ofboth sides 
and transferring terms yields the following transformation: 
42 
Figure 4. Supply and Demand Illustration of Quantity Purged of Price Effects 
p 
s 
Q 
-lnQ 1 -111nP 1 +111nP 1 =Ink 1 (3.3) 
Note that the left side of this equation will be larger or smaller than the observed Q1, . 
depending on whether the retail price is above or below its mean. It can readily be seen 
that if the retail price equals the mean price, or if the price elasticity is zero, the left side 
equals the observed Q1• As such, the left side of equation (3.3) represents quantity 
corrected for any price-effect. Because k is assumed to be exogenous, the price effect 
43 
corrected quantity, Q1, can be represented in equation form as: 
(3.4) 
Using equation 3.4, values oflnQcan be computed, given an acceptable range of 
short run, price elasticities.23 Taking the inverse log oflnQ yields Q, which can then be 
divided by Q. This normalized value is iticluded in the regression equation as a 
measure of seasonality that is purged of price effects. 
Sales Below Cost Laws 
Sales Below Cost laws originated with California's "Unfair Trade Practices Act" 
of 1933 and have subsequently been enacted in various forms in 38 states.24 These laws 
frequently encompass all retail sales and thus apply to the retail sale of gasoline. 
SBC laws that relate directly to trtotor fuels are either independent from the 
general SBC laws or are included as specific provisions within the general laws. These 
gasoline specific laws are of special concern for this thesis. According to a 1991 study 
conducted by the American Petroleum Institute (API), a total of 9 states had such laws. 
23Elasticity values of -.1 and -.2 were used in computing Q. See chapter 4 for the 
empirical results. Borenstein and Shepard (1993) follow a similar approach and use a 
similar range of elasticity values. 
24Department of Energy, Minimum Markup Laws in Gasoline Marketing: Ari 
Economic Analysis and A Legal. Economic. and Legislative Review, prepared by 
Synergy, Inc., March 1985 (Washington, D.C.: U.S. Government Printing Office), 1. 
This report analyzes the effects of general sales below cost laws on gasoline marketing. 
Their analysis includes a review of each state law, using some of the same categories as 
this study. 
44 
Since that study, gasoline SBC laws in Georgia and Michigan have been repealed. 
Missouri and Montana have initiated new laws, while Colorado and Tennessee added 
sections to previously existing laws making illegal the sale of motor fuels at prices below 
cost. Currently, a total of 11 states have gasoline specific SBC laws. Eight of these 
states, and a total of nine cities, are represented within the sample used in this study.25 
The gasoline SBC law for each of these eight states and Montana will now be 
analyzed. The definition of cost is of primary concern, but additional factors that further 
define the relative strength of each law will also be examined. The first of these is the 
penalty incurred for a violation. As the penalties become more severe, retailers will be 
more sensitive to maintaining legal or non-threatening prices. Second, if a law allows for 
private litigation the law may be enforced more frequently. Third, most gasoline SBC 
laws allow a defendant to meet the equally low price of a competitor without being in 
violation. This would lessen the hypothesized effects of the law. Fourth, most SBC laws 
require evidence of a defendant's intent to injure competition. The legal debate 
surrounding this requirement is whether harming an individual competitor constitutes 
injuring competition. The existence of such a clause and the state's legal position on this 
issue will affect the ability of the authorities to achieve a conviction. We now tum to the 
analysis of each law. 
25Maine and North Carolina also have gasoline specific SBC \aws, but are not 
represented in the sample considered by this study. 
45 
Alabama 
Alabama's Motor Fuel Marketing Act was passed in 1984.26 Cost is defined as the 
lesser of the invoice or replacement co~t, plus all taxes, inspection fees, freight charges, 
and the cost of doing business.27 A survey within the relevant market area may also be 
used to establish cost. Penalties for infringement of the law are up to $1,000.00 for each 
day on which a violation occurs. Private individuals may also sue for injunctive relief 
and damages. The meeting competition defense is allowed and the defendant's intent to 
injure competition must be proved. 
Colorado 
Colorado's Unfair Practices Act is a general sales below cost law that was 
amended in 1993 to specifically include sales of motor fuels. 28 The amendment 
became effective on July 1, 1993. Cost is defined as the lesser of invoice or replacement 
cost, plus the cost of doing business. This is further defmed as including all costs 
incurred in the operation of the business such as labor, rent, executive salaries, and 
depreciation. The penalty for each violation of the law is a fine of between $100.00 
26Alabama. Motor Fuel Marketing Act, Statutes. Annotated (1984) 8-22. 
27Under the Alabama Code, the "cost of doing business" is defmed as including, 
but not limited to: labor, rent, interest on borrowed capital, depreciation, selling cost, 
maintenance of equipment, transportation or freight cost, losses due ·to breakage or 
damage; credit card fees; credit losses, licenses, taxes, insurance, and advertising. 
28Colorado. Unfair Practices Act, Revised Statutes. Annotated (1993) 6-2-
105(1b). 
46 
and $1,000.00, or imprisonment of up to 6 months, or both. In addition to civil action . 
brought by the state, private parties may sue for damages. The defendant is allowed, 
however, to meet the legal price of a competitor. Violation of the law also requires an 
intent by the defendant to injure his competitors or destroy competition. 
Florida 
Florida's Motor Fuel Marketing Practices Act was enacted in 1985.29 For a non-
refmer (retail station not owned by a refiner), cost consists of the invoice cost, all 
applicable taxes, inspection fees, freight charges, direct labor costs, and the reasonable 
rental value of the retail outlet, less credit card allowances, trade discounts and rebates 
received. Refiner cost is likewise defmed with the exception that the invoice cost is 
replaced with the posted terminal price. Penalties for violation of the law are up to 
$5,000.00, with each day on which an infraction occurs considered a separate offense. 
The aggregate of this penalty may not exceed $100,000. Violations of the Act are 
investigated by the Department of Agriculture and Consumer Services and the 
Department of Legal Affairs is authorized to bring civil action. Private individuals are 
also allowed to sue for treble damages. The meeting competition defense is allowed and · 
intent to injure competition must be present. Competition is defmed so that injury to an 
individual competitor would be sufficient evidence of such intent. 
29Florida. Motor Fuel Marketing Practices Act, Statutes. Annotated (1985) 
526.301-3135. 
47 
Massachusetts 
Massachusetts passed a general sales below cost law in 1932. A gasoline specific 
law prohibiting sales below cost was enacted in 1950.30 This law's definition of cost 
includes the cost of motor fuel plus the cost of doing business. A cost survey is also a 
permissible means of establishing cost. The state is authorized to bring civil action and 
collect penalties of up to $1,000.00. Private parties, on the other hand, are allowed to 
bring suit for alleged violations, but only for the purpose of obtaining an injunction. A 
firm is allowed to meet the equally low price of a competitor. Under the general SBC 
law, intent to injure competition is required for a violation to have occurred. The 
gasoline specific law does not have such a provision. 
Missouri 
The Missouri Motor Fuel Marketing Act came into effect on Aug. 28, 1993.31 
Cost is defined as the lowest invoice cost or transfer price, less trade discounts, 
plus taxes, freight charges, and the "cost of doing business". This last term is further 
defmed as including, rental v~lue, licenses, taxes, utilities, insurance, and non-managerial 
labor. The state Attorney General is responsible for investigating violations of this law, 
which may result in penalties of between $1,000.00 and $5,000.00 for each day of 
30Massachusetts. Unfair Sales Act for the Retail Sale of Motor Fuels, General 
Statutes. Annotated (1950) 295P-W. 
31Missouri. Missouri Motor Fuel Marketing Act, Code. Annotated (1993) 
416.600-640. 
48 
infraction. In addition, private individuals may bring civil action for the purpose of 
obtaining an injunction or to recover treble damages. A defendant is allowed to nieet the 
equally low price of a competitor. Also, a violation requires the seller's intent to injure 
competition. 
Montana 
The Montana Retail Motor Fuel Marketing Act became effective in 1991.32 The 
law forbids the selling of motor fuel at less than its delivered cost plus the cost of doing 
business. The delivered cost of motor fuel includes t:he lower of invoice or replacement 
cost, less all trade discounts, plus all transportation costs and all applicable taxes. For a 
refiner selling at retail, the invoice cost is replaced with the posted rack price. In the 
absence of proof of lesser cost, the cost of doing business is defined as 6 percent of the 
delivered cost of motor fuel for retail sales. Violations of this law may result in a civil 
penalty of up to $1,000.00 per day. The Department of Justice or a county attorney are 
authorized to bring civil action for any suspected violation. Private parties injured by a 
violation may file suit for injunctive relief and actual damages. A meeting competition 
defense is included. Also, the effect of injuring or destroying competition is required for 
the action to be illegal. Prima facie evidence of such intent is given by· advertisement to 
sell, or the sale of, motor fuels below cost. 
32Montana. Montana Retail Motor Fuel Marketing Act, Code, Annotated (1991) 
30-14-801-806. 
49 
New Jersey 
New Jersey's law regulating the sale of motor fuels originally took effect in 1938. 
In 1953, this act was amended and became known as the Unfair Motor Fuels Practices 
Act.33 Minor amendments were again added in 1981. Sales at prices below cost is one of 
several practices considered illegal under this law. Cost is defmed as the invoice cost 
plus the expense of selling motor fuel. The specific components of this expense are not 
specified. The State Tax Commissioner is solely responsible for initiating civil action. 
Penalties of between $50.00 and $200.00 per offense may be given and the retailer's 
license may be suspended for 5 to 30 days. In lieu of these penalties, an offender may be 
imprisoned for up to 30 days. This law does not provide a meeting competition defense 
nor does it require the intent to injure competition or a competitor. 
Tennessee 
Tennessee's Petroleum Trade Practices Act was enacted in 1977. A section within 
this act, making sales below cost illegal, was subsequently enacted in 1988.34 The 
defmition of cost comes from Tennessee's Unfair Sales Law, which applies to all retail 
sales.35 It is defmed as the lower of the purchase price or replacement cost, less trade 
discounts, plus a 6 percent markup in the absence of proof of lesser cost. The minimum 
20. 
33New Jersey. Unfair Motor Fuels Practice Act, Statutes. Annotated (1953) 56:6-
34Tennessee. Petroleum Trade Practices Act, Code. Annotated (1977) 47-25-611. 
35Tennessee. Unfair Sales Law, Code. Annotated (1937) 47-25-201. 
50 
markup is given as a rule in determining prima facie evidence of a sale below cost, 
although the defendant may still offer evidence to the contrary. Penalties of up to 
$1,000.00 for each day of violation may be incurred. Private parties are entitled to collect 
treble damages. The law is not binding, however, if a price is set to meet the equally low 
price of a competitor. Also, a violation requires the intent to injure competition. 
Utah's Motor Fuel Marketing Act was enacted in 1981 and amended in 1987.36 
Cost is defined as the lowest invoice cost (or the lowest transfer price for a refiner owned 
station) within 5 days of the alleged infraction, less trade discounts, plus freight charges, 
taxes, and the "reasonable cost of doing business".37 In the absence of proof of lesser 
cost, this is presumed to be 6 percent of the posted retail price. The civil penalty is up to 
$5,000.00 per violation. In addition, private parties injured by such violation may bring 
suit for injunctive relief and may be awarded treble damages. A firm is allowed to meet 
the equally low price of a competitor as an exemption from the law. Also, the intent to 
injure competition must exist for a violation to have occurred. 
The Effects of SBC Laws: A Preliminary Analysis 
Initial evidence of the impact of SBC laws on the pricing behavior of firms can be 
obtained by several means. If the number or legal cases is relatively large, it may indicate 
36Utah. Motor Fuel Marketing Act, Code. Annotated (1981) 13-16-1-6. 
37Ibid., 13-16-2. 
51 
the laws are actively enforced. Alternatively, if the number of cases is low, it may 
indicate that enforcement is lax or that selling below cost seldom occurs in regulated 
states. The latter may occur because of the threat of litigation. More importantly, the 
cases and their rulings are prima facie evidence of what is occurring in practice. A Lexis 
search, and additional research, indicates a total of 6 cases resulting from SBC laws since 
1985. A case in Montana is currently being prosecuted. Of the 6 cases, 4 have occurred 
in Alabama, perhaps indicating stronger enforcement in that state. Two of the cases have 
resulted in convictions. 
The earliest case during this period occurred in Alabama in 1986. In Star Service 
& Petroleum Company. Inc. v. State of Alabam§, the state accused Star Service of 
selling motor fuels at prices below cost with the intent of injuring competition. 38 The 
Circuit Court ruled in favor of the state and granted injunctive relief. Star Service 
appealed, disputing the accusation of below cost selling by claiming that the Alabama 
Motor Fuels Marketing Practices Act allowed pooling regular and unleaded gasoline in 
the .calculation of cost. The court ruled that a separate determination of cost was required 
for each grade of fuel. 
The State of Alabama also brought legal action against a motor fuel marketer in 
1987. State of Alabama v. Mapco Petroleum. Inc. was dismissed by the circuit court on 
grounds that the Motor Fuel Marketing Practices Act was unconstitutionat.39 Upon 
appeal by the State, the· Alabama supreme court reversed this decision, thus upholding the 
38Star Service and Petroleum Co., Inc. v. State, 518 So.2d 126 (Ala. Civ. App. 
1986). 
52 
constitutionality of the law. It determined that the law did not infringe on individual 
rights in such a manner as to preclude lawful competition. The only section of the law 
that was deemed unconstitutional was that which allowed a plaintiff to prove a prima 
facie case by showing the defendant's price had been below the plaintiffs costs. 
A second conviction was handed down under the Alabama Motor Fuel Marketing 
Practices Act in 1990. In Money Back. Inc. v. Charles Gray. d/b/a Hilltop Chevron Tire 
and Service Center, Money Back, Inc. was accused of selling motor fuel at below cost.40 
The dispute originated during a "price war" in which both the plaintiff and the defendant 
admit to pricing illegally. The defendant, however, also admitted to having the lowest 
price of any competitor in that market. The plaintiff sued for an injunction and for 
compensatory damages of$5,409.00. The circuit court ruled in favor of the plaintiff on 
the basis that the sales were made "with the intent of injuring competitors or destroying 
or substantially lessening competition".41 Treble damages of $16,227.00 and attorney's 
fees and court costs of an additional $8,415.84 were awarded. Upon appeal to the state 
Supreme Court, the ruling was upheld, but treble damages were lessened to $9, 786.00. 
This reflected the court's decision that damages awarded to reflect a speculative "upward 
trend" in the plaintiffs business were not allowable. Under this ruling, the court also 
determined that the plaintiff was not required to prove that the defendant's intent was 
39State ex. rei. Galanos v. Mapco Petroleum, 519 So.2d 1275 (Ala. 1987). 
40Money Back, Inc. v. Gray, 569 So.2d 325 (Ala. 1990). 
41Ibid. 
53 
unlawful. 
The most recent case under the Alabama law occurred in 1992. Several Alabama · 
jobbers sued Mapco, Inc., an independent, unbranded refmer and retailer operating in 
. Alabama.42 Mapco was acc~sed of selling gasoline below cost for a period of 596 days at 
one location and 821 days at another. A counterclaim was made by Mapco against the 
plaintiffs on the grounds of violations of the Sherman Act (U.S.C., title 15, sec. 1-7), 
Robinson-Patman Act (U.S.C., title 15, sec. 13) the Alabama Motor Fuel Marketing Act, 
and the Alabama Unfair Trade Practices Act. The case was then removed to federal court 
where the decisions were in favor of Mapco in the original case and McGuire in the 
counterclaim. This ruling also found that the Alabama law required only injury to a 
competitor as sufficient evidence of a violation. However, the lack of intent to injure a 
competitor was an allowable defense. Also, the meeting competition defense was not 
valid if the defendant priced below a competitor's price. 
The most recent case was in Tennessee and was decided on March, 15 1993. In 
Ohern. Inc. v. Mapco Petroleum, Inc., Ohern, Inc. accused Mapco Petroleum, Inc. of 
selling unleaded gasoline at prices below cost on several specific dates over an 8 month 
period.43 The Tennessee Supreme Court requested that several questions be certified by 
the U.S. Circuit Court of Appeals concerning determination of a violation of the 
Tennessee Petroleum Practices Act (T.C.A. Sec. 47-25-611). In conjunction with this 
42McGuire Oil Co. v. Mapco, Inc., 612 So.2d 417 (Ala. 1992). 
43Ghem, Inc. v. Mapco Petroleum, Inc., 850 S.W.2d 447 (Tenn. 1993). 
54 
request, the defendant was granted its motion for Summary Judgement on this case. 
Several key elements resulted from the ruling. Although the court found sufficient 
evidence that Mapco had made sales at prices below cost, it determined that Ghem had 
provided no evidence ofMapco's intent to injure or destroy competition within the 
relevant market. As such, damage to an individual competitor from prices below cost did 
not constitute or imply illegal activity by the defendant. Injury to competition would be 
indicated by an "actual or threatened net decrease" in the number of competitors.44 This 
ruling was based on the necessity of "anti-trust injury" as an element in determining 
illegal pricing. "Anti-trust injury" was defined to include injury that results from 
· anti-competitive acts that the law was intended to eliminate. Thus, for a firm to be 
convicted of predatory pricing, the plaintiff must show that any injury incurred was the 
result of an attempt to monopolize the relevant market. 
The constitutionality of Utah's Motor Fuel Marketing Act was also challenged in 
1990. The State of Utah v. Rio Vista OiL Ltd., began in June of 1986 with a civil action 
brought against Rio Vista Oil by the state on charges of below cost sales.45 The district 
court ruled that the law was unconstitutional in that it disallowed a·defendant "due 
process" under both the State of Utah and Federal constitutions. This arose from several 
' 
specific issues. First, the law prohibited all sales below cost without allowance for the 
deferi.dant to show lack of intent to injure competition. Second, the law did not allow for 
44Ibid. 
45State v. Rio Vista Oil, Ltd., 786 P.2d 1343 (Utah 1990). 
55 
a defendant to price below cost for the purpose of meeting the price of a competitor. 
Third, violations of the law were difficult to determine due to its ambiguous definition of 
cost. The state appealed this decision and the Utah Supreme Court reversed the lower 
court's decision, thus upholding the constitutionality of the law. It ruled that the district 
court erred in its interpretation of these three issues and that the defendant's due process 
had not been infringed upon. 
These cases indicate that the laws are utilized, but in a limited fashion. This 
amount oflitigation may, however, be sufficient to affect margins based on the threat of 
penalties. The effect may be observable in the descriptive statistics of the sample when it 
is split according to states with and without SBC laws. If SBC laws are binding, we 
would expect the minimum value of retail margins in states with SBC laws to be higher 
than in states without SBC laws. The statistics in Table 2 are consistent with this 
hypothesis as the minimum is -3.000 in non-law states and 4.000 in law states. 
Moreover, this is not the result of a single observation, 34 observations in 14 different 
cites are below the minimum margin of states with SBC laws. It is noteworthy that the 
difference between the two means is statistically significant. (t = 7A8) 
Non-Law 3100 15.516 -3.000 37.522 
Law 620 17.394 4.000 33.700 
Total Number of Observations= 3720 
56 
This demonstrates a difference in the pricing behavior of firms in these two 
groups and provides evidence of a binding constraint. Further evidence is provided by a 
detailed look at retail and wholesale price data for five cities in Montana.46 The Montana 
SBC law itllows for a minimum markup as an alternative method of defining cost. By 
comparing this definition of cost with the retail margin for each city, several inferences 
can be drawn. If margins are consistently below the minimum markup for extended 
periods of time, it is likely the laws have no effect. But, if the margins are generally 
above the minimum markup, there are two possible explanations. Either the laws are, 
again, ineffective or they may have effect due to the threat of litigation. 
The retail price data for this analysis is from the American Automobile 
Association {AAA). These prices were surveyed on either a biweekly or monthly basis. 
The wholesale price data were obtained from Exxon, Inc. and are monthly averages of 
branded and unbranded rack prices. Calculating a retail margin using average biweekly 
(or monthly) retail prices and monthly average wholesale prices is less than ideal. To 
match the two series, average monthly retail prices were computed. For months with 
more than one retail price observation, a mean price was computed of that month's 
observations. If only one survey occurred during the month, an average was calculated of 
the previous month's l~t observation, the current observation, and the following month's 
first observation. In cases where only one observation occurred in consecutive months, 
the average was computed with the previous month's observation, the current 
46Montana is not included in the data set used to empirically test the hypotheses in 
the next chapter. 
57 
. observation, and the following month's observation. During the one month in which the 
survey was not conducted, the mean of the adjoining month's observations is that 
month's average retail price. 
The minimum markup is calculated as a percentage of the delivered cost of fuel. 
In Montana, this is 1 percent for wholesalers and 6 percent for retailers. The retail 
markup is therefore 6 percent of the sum of the rack price, taxes and the 1 percent markup 
at the wholesale level. In equation form this is: 
Retail Markup=.06((Rack Price+Taxes)+.Ol(Rack Price+Taxes)) (3.5) 
Figures 6 through 10 below show the graphical comparisons of retail margins and 
minimum markups for each of the five cities. In each case, the margins are typically 
above the minimum markup. Margins infrequently go below this level and, if so, for 
limited periods of time. This may be attributable to the threat of litigation, in which case 
SBC laws would appear to be effective. 
Conclusion 
The small amount of litigation evident in most SBC law states would not seem to 
be indicative of a binding price floor. However, the fact that litigation is not an every day 
event does not by itself imply that the laws' effects on retail margins are inconsequential. 
The fear of penalties in SBC states may be sufficient to alter behavior. The descriptive 
statistics of law and non-~aw states and the analysis of the Montana data are consistent 
with this premise. However, to examine this further and obtain a measurement of the 
58 
effect of the SBC laws, it is necessary to control for the other variables that may also 
affect the retail margin and vary between cities. The empirical analysis presented in the 
next chapter controls for these other variables. 
59 
Figure s: Retail Margins and Minimum Markups for Billings, MT 
0.3 
0.25 
c: 
0 0.2 
co 
C> 
L.. 
~ 0.15 
~ 
co . 
0 0.1 
Cl 
0.05 
0 
~ 
i\ I 
\ I 
\ 
....... v 
I I'" ~ ,......., ...-1'-' 1--' 
1/92 4/92 7/ 32 
Billings 
) 
/ \ 
i/ \ 
\ I .,1\ 
\ v 1/ \/ 
\ 
..... 
v \V'~ 
!'-' f-o 
-Jjl...j 
09 1/ 93 4/ 93 7/93 09 1/ 94 4/ 
Date 
-111- MARGIN_..,_ MIN M/U 
r--. 
...... 
,_, ~ ...... 
94 '7! 94 
60 
Figure 6. Retail Margms and Mmimum Markups for Bozeman, MT 
0.3 
0.25 
c: 
0 0.2 ('Q 
(!) 
.... 
~ 0.15 
e! 
·m 
0 0.1 
0 
0.05 
0 
1\ 
\ 
\ 
r..... 
...... 
/ 
v 
.....,,._..; ~ 
1/~2 4/ ~2 7/ 
Bozeman 
lL' !'-"! ~ 
v !/ \ 
I \ I.--' 
ir-"' \ I i\ I 
if )~ I 
!\ I 
\V \I 
I'-' ...... 
~2 10 9. 1/ ~3 4/ ~3 7/ ~3 09 1/~4 
Date 
-a- MARGIN- MIN M/U 
~ 
1'-
\.. 
"". iH 
_,..,., 
""" 
4/ ~4 ._,, 94 
61 
Figure 7. Retail Margins and Minimum Markups for Great Falls, MT 
Great Falls 
0.25 
0.2 / 1\ 
\ 
c I 
.Q ~~ VI\ h\ a; (!) 0.15 I 
L.. \ ~s-t'' I \ i/ )H ~ (]) c. I J \ ~ 0.1 '\ _j 1!! 1\ v ) iH 0 ..... 0 :) ~ '""" 1'-' ~· ' j ,...... ~ ,....... .._I'-. ~ ,_.. ...... 0.05 ,.._., \ 
0 1 ...... ,~-~2-t-l-4-l/9-:2+-+7-+/3-l:2f-f-0+9-+,, -+-1/+-93+-+4-i/f-93+-+7-+/9~:31-+0+9~f-1/+-94-l-f4-,t-~4+-+7-l/ ~-f-1·4 
Date 
- MARGIN- MIN M/U 
62 
Figure 8. Retail Margins and Minimum Markups for Helena, MT 
0.25 
0.2 
c: 
..Q 
-m 
<.9 0.15 
.... 
(]) 
c. 
~ 0 
.!!! .1 
0 
0 
0.05 
0 
\ 
4 IJ 
1/~2 
/ 
I 
1/ 
...... 
~ 1 ~ 
....., 
''I 
4/32 7/92 
Helena 
..... 
1/ 1\ 
~~ I 
I 
I \ 
\ ...., 
I 
1\ ! 1/ 
f-' ...... J.r- L\. ..... r-.. !'-.... 1/ -,/ \ ...... 
' ~ I 'I 
09 1/ 33 4/:}3 7/~3 09 1/ 
Date 
-111- MARGIN-'~'- MIN M/U 
I\ 
\ 
...... 
~ ~ 1'-' 
94 4/ ~4 7/ ~4 
63 
Figure 9. Retail Margins and Minimum Markups for Missoula, MT 
0.35 
0.3 
§ 0.25 
m (!) 
..... 0.2 
Q) 
c. 
~ 0.15 
.!Q 
0 0 0.1 
0.05 
0 
\ 
\ 
1/~2 
/ 
v 
"~--~ '/ 
,..... 
4/ 92 7/ ~2 
Missoula 
\ 
\ 
1\ 
~v \ \ 
1/ 1\ \ 
~~ !\ II\ 
/ \ 1 ~~ 1/ I 1/ ~~ 
\ ..... ~ 
\ I 
r-' 
IJ 
rr 
0/9. 1/ 93 4/ 93 ..,, 93 0/9 1/ ~4 4/ 
Date 
---MARGIN-.- MIN M/U 
\ 
)~ 
'""'! 
,_,I-' 
94 ..,, ~4 
64 
CHAPTER4 
EMPIRICAL TESTS AND RESULTS 
Introduction 
The purpose of this chapter is to empirically test for the effect of the SBC laws. 
The analysis begins with the econometric specification of the models used in undertaking 
the various tests. Several models are constructed. The first of these include 92 weekly, 
time dummy variables to account for a seasonal pattern in the retail margins. The issues 
of asymmetric response to changes in wholesale prices and the lag structure of prices are 
also empirically examined. 
Econometric Considerations 
The use of pooled data presents several problems. For example, this is time series 
data and the error terms may be serially correlated. Indeed, diagnostic tests performed on 
margin data for each city indicated the presence of first order autocorrelation that varies 
across cities. Additionally, in cross sectional data there is the potential problem of 
heteroscedastic errors. Application of the Bruesch-Pagan test rejects the hypothesis of 
homoscedasticity. 2 Estimation of the models therefore requires correction for an error 
2T.S. Breusch and A.R. Pagan, "A Simple Test for Heteroscedasticity and 
Random Coefficient Variation," Econometrica 47 (1979): 1287-1294. 
term with the following specification; 
65 
E(e2 \=a2 
ir i 
e =p e +J.t it i i,t-l it 
with J.ttt-N(O,a!1 ) 
The estimation procedure uses the method described in Kmenta (1986) to correct for 
cross-sectional heteroscedasticity and first order autocorrelation with pooled data. 3 
Model Specification and Results 
Measuring the Effect on Retail Margins 
The model used to test the hypothesis that SBC laws may result in higher retail 
margins is 
Margin it= Po +P1Retwageit +P2Prvaltt +P3Popdentt + 
T-1 
P4NSSit+PsLAWtt+ L YuWu+ett • 
where Margin = the retail margin, 
Retwage = the retail wage, 
Prval = the property value, 
Popden =the population density, 
t=l 
NSS = a dummy variable that takes a value of 1 for cities within f?tates that 
prohibit the sale of self service gasoline and 0 otherwise,4 
(4.1) 
LAW = a dummy variable that takes a value of 1 for cities within states that have 
a gasoline specific SBC law and 0 otherwise, and 
Wt =a weekly dummy variable for timet (T=93). 
3Jan Kmenta, Elements of Econometrics, 2d ed., (New York: MacMillan 
Publishing Company, 1986), 618-622. 
4New Jersey and Oregon prohibit the sale of self service gasoline. In this sample, 
Newark and Portland are affected by this law. 
66 
Table 3 lists the regression results. The coefficient on LAW is positive and 
significant, indicating that gasoline SBC laws increase retail margins, on average, by 
approximately .8 cents. With the exception of the retail wage, all other variables have the 
expected signs and are statistically significant at the 5 percent level or greater. 5 Of 
additional interest are the substantially higher margins in states that prohibit the sale of 
self service gasoline. 
Table 3. The Effect of SBC Laws on the Retail 
Retwage 
Prval(,OOO) 
Popden(,OOO/sq mile) 
NSS 
LAW 
-0.007 
0.040 
-0.113 
5.810 
0.841 
-0.545 
6.053*** 
-2.148** 
5.732*** 
1.976** 
Adjusted R2 = .49 Denotes significance at: *** .= 1%, ** = 5% 
Number of Observations= 3720 
It is a reasonable assumption, however, that the effect on margins is greater if 
the laws are actively enforced. As was shown in Chapter 3, four of the six recent cases 
occurred in Alabama. Thus, if active enforcement increases the likelihood that SBC laws 
have effect, the margins in Alabama should be higher relative to other states with SBC 
5See Appendix E for the results of this regression, including the time dummy 
variables. · 
67 
laws. A model to test this is 
Margin 11 =Po +P1Retwage11 +P2Prval11 +P3Popden 11 + 
T-1 
P4NSSu+PsLAWu+P~Lu+ L YuW1,+eu • 
t=l 
(4.2) 
where AL = a dummy variable that takes a value of 1 if the city is in Alabama and zero 
otherwise. 6 
All other variables are as previously defined. Table 4 shows the regression results for this 
equation. The coefficient on the variable AL indicates that margins are 2.22 cents higher 
in Alabama relative to other states with SBC laws. The variable LAW remains positive, 
but is not statistically signif1cant. Thus, active enforcement, as in the case of Alabama, 
appears to be an important factor for SBC laws to lead to higher margins.7 
An additional test of the effect on margins can be made if a particular state's SBC 
law became effective during the time period of the sample. Missouri and Colorado fit 
this criteria. A simple model to test for higher margins after the effective date of the law 
for an individual city is 
Margin, = Po + P1LAW1 +TIME, +E1 
where LAW = a dummy variable that takes a value of 1 for the time period after the 
effective date of the SBC law and is 0 otherwise, and 
TIME = a time trend variable to account for inflationary effects. 
6Birmingham is the only city in Alabama represented in the sample. 
(4.3) 
'Tennessee and Utah have each had one case since 1985, indicating only a slight 
amount of enforcement. A dummy variable that includes these states in equation 4.2 is 
not statistically significant, supporting the notion that active enforcement is a primary 
determinant of the effect on margins. 
68 
Enforcement 
Retwage -0.013 -1.036 
Prval(,OOO) 0.043 6.423*** 
Popden(,OOO/sq mile) -0.098 -1.840* 
NSS 6.106 5.857*** 
LAW 0.359 0.743 
AL 2.220 2.564** 
Adjusted R2 = .49 Denotes significance at: *** = 1%, ** = 5%, *=10% 
Number of Observations = 3 720 
This model is run for each of three cities represented in Missouri and Colorado. In each 
case, the coefficient P1 is negative and not statistically significant, indicating that· 
introduction of the laws did not affect the margin. 
Equations 4.1 and 4.2, however, do not take into account seasonal demand 
changes that may vary between cities. To account for this, the quantity variable detailed 
in Chapter 3 was included. Data was obtained from the Federal Highway 
Administration's table "Highway Use of Gasoline by States". 8 Since these data are 
estimates distributed so as to preserve seasonal driving patterns, some error may be 
involved. Moreover, lags of up to 6 weeks before consumption make these questionable 
8Department of Transportation, Table MF-26. The obvious reporting errors in the 
alternative sources makes their use difficult and authenticity questionable. 
69 
proxies for current demand. 9 
If the observed quantities accurately measure demand, quantity and margin should 
be positively correlated. Recall the quantity/seasonality variable constructed in the 
previous chapter. Observed weekly quantities, Q, were interpolated from the source 
listed above. An instrument, yielding price-effect corrected quantities, Q, was 
constructed and these variables were then normalized around their mean values. When 
the price elasticity of demand, T}, is set to zero, the normalized Q is equal to Q. The 
inclusion of that variable into equation 4.1 yielded a coefficient that was not significantly 
different from zero, contrary to the hypothesized effect. Experimenting by including the 
price effect corrected quantity forT}= -.1, increased the coefficient value to 13.35, while 
forT} = -.2, the coefficient value became 30.77.10 The variable LAW, however, was not 
affected by the inclusion of either Q or Q. Since the results are highly sensitive to small 
changes in the price elasticity and there appear to be potentially serious errors in the data, 
we rely on the weekly time dummies to accommodate the seasonal demand patterns. 
Another but more complicated model in terms of the error structure, uses the retail 
price as the dependent variable. In this case, wholesale prices and taxes are included as 
explanatory variables since these vary over time and across cities. In equation form this 
IS 
9Borenstein and Shepard (1993) included a similar quantity measure in their 
paper. They also note that current demand may affect margins through peak-load pricing 
or as a result of increased demand elasticity during peak driving months. 
10Both coefficients for Q are statistically significant at the 1 percent level. 
70 
Retpriceit =Po +P1 Whlpricf?u +P2Taxit +P3Retwage11 +P4Prval11 + (4.4) 
T-1 
PsPopdentt +P6NSStt +P1LAW;;+ L ytwtt +ett ' 
where Retprice = the retail price, 
Whlprice = the wholesale price, and 
Tax= motor fuels taxes. 
t=l 
All other variables~ including time dummies, are as previously defined. The regression 
results are shown below in Table 5. Again, the coefficient on the variable LAW remains 
positive, but it is not statistically significant. However, we note that the coefficient on the 
Whlprice variable is only 0.557. This would seem to suggest that wholesale price 
changes are not fully passed on to the retail level. This issue is addressed in the 
following section. 
Table 5. The Effect ofSBC Laws on Retail Price 
Whlprice 
Tax 
Retwage 
Prval(,OOO) 
Popden(,OOO/sq mile) 
NSS 
LAW 
Adjusted R.2 = .51 
Number of Observations= 3720 
0.557 
1.073 
-0.011 
0.050 
-0.151 
7.852 
0.475 
20.800*** 
22.310*** 
-0.860 
8.488*** 
-2.815*** 
8.551 *** 
1.281 
Denotes significance at: * * * :::;: 1% · 
Measuring Response Differences to 
Changes in Wholesale Price 
71 
The previous results indicate that retail margins are somewhat affected by the 
SBC laws. But a binding price floor could also cause retailers to react more quickly to. 
increases in wholesale prices. This argument assumes that retail prices do not adjust 
instantaneously to changes in wholesale prices and the results reported in Table 5 seem to 
affirm this. The effect of SBC laws on adjustment processes can be measured by 
interacting the LAW variable with current and lagged wholesale prices. Conducting this 
test, however, requires prior knowledge about the lag structure of retail prices. 
The analysis of the lag structure began by experimenting with various free form 
lags on the wholesale price. Equation 4.4 then becomes, 
n 
Retprice1, =Po+ L a.1Whlprice1.,_1 +P1Taxu +P2Retwageu +P3Prvalu + j=O 
T-1 
P4Popdenu +P5NSSu +P6LAWu + L y,W1, +eu . 
t=l 
(4.5) 
A consistent pattern emerges as more lagged terms of the wholesale price are added to 
this equation. The coefficient on the contemporary wholesale price is consistently 
between .55 and .60, similar to its value in equation 4.4. The second through fourth 
weeks are characterized by negative adjustments in retail price. Following this, retail 
prices begin a slow positive adjustment towards a new equilibrium level. 
Long adjustment processes are often modeled by including a lagged dependent 
72 
variable as an explanatory variable in equation 4.4. A simple Koyck lag, however, seems 
inappropriate given the negative adjustment indicated by equation 4.5.11 A model with a 
lagged dependent variable that does capture the negative adjustment after the first period 
lS 
2 
Retprice 11 = a+ .1: PiWhlprice l,t-j +P3Retprice 1,1_1 +P 4 Tax 11 +P 5Retwage 11 + ( 4 .6) 
j=O 
T-1 
P6Prval11 +P1Popden 11 +PgNSS11 +P9LAW11 + L y 1W 11 +E11 • 
t=l 
The results of equation 4.6 are given in Table 6. To test the significance of including the 
one and two period lagged wholesale prices, an F -test was run on the null hypothesis that 
P1 = P2 = 0. This hypothesis is rejected, supporting the time path indicated by equations 
4.5 and 4.6 (F=20.00). Further evidence that is consistent with this result is given when 
considering the following issue. 
It has been argued that retail prices respond more quickly to positive changes in 
wholesale prices than negative. Starting from an initial equilibrium position, a wholesale 
price decrease supposedly affords. a retailer the opportunity for short run profits if the 
current retail price is maintained. However, under competition the higher prices will not 
persist. Asymmetry has been attributed to the· presence of market power amongst 
retailers. Shin, however, points to inventory adjustments as one reason for price 
11L.M. Koyck, Distributed Lags and Investment Analysis, (Amsterdam: North 
Holland Publishing Company, 1954). 
73 
Table 6. Estimation of the Lag Structure of Prices 
· with a Variable 
Whlpricet 
Whlpricet-I 
Whlpricet.2 
Retpricet-I 
Taxt 
Retwaget 
Prvalt (,000) 
Popden. (,000/sq mile) 
NSSt 
LAWt 
0.567 
-0.603 
0.132 
0.887 
0.122 
0.0006 
0.004 
-0.014 
0.676 
0.142 
19.890*** 
-14.490*** 
4.582*** 
113.600*** 
11.260*** 
0.356 
4.383*** 
-1.586 
4.359*** 
2.120** 
Adjusted R2 = .88 Denotes significance at: *** = 1%, ** = 5% 
Number of Observations = 3640 
asymmetry in competitive markets.12 During periods of tight supply, increases in 
wholesale price may be passed along especially fast. If there is excess supply, however, 
firms may purchase for future periods as well, thereby slowing the downward movement 
of retail prices. Moreover, the discussion of gasoline retail markets presented in Chapter 
2 suggests that it is highly competitive, thus bringing into question the arguments that 
price changes are asymmetric for uncompetitive reasons. 
Various studies have been done on price asymmetry in the context of the 
12David Shi.Il, "Do Product Prices Respond Symmetrically to Changes in Crude 
Prices," (Washington, D.C.: American Petroleum Institute, Research Study #068, 1992), 
4. 
74 
petroleum products industry. The results vary, depending on the model and data set 
used. 13 Bacon, for example, uses a non-linear partial adjustment model and fmds only a 
slight amount of asymmetry .14 In this model, a quadratic term is included to measure the 
adjustment process. Changes between wholesale and retail prices were modeled as 
llNRetpr11 =P0(<I>0 + <1>1 Whlprice1,1_1 - NRetpr1,t-1) + (4.9) 
P1( <Po+ <1>1 Whlprice1,t-1 - NRetpr1,1_/ + e11 
where aNRetprit is the change in retail prices net of taxes. If P1 = 0, the model reduces to 
a linear form and no asymmetry is present. Using Bacon's model we find that asymmetry 
is present, with positive changes occurring more quickly than negative. The size of the 
coefficient pb however, is extremely small (.007), indicating that the amount of 
asymmetry is inconsequential. 
An alternative estimating procedure is presented by Borenstein, Cameron, and 
Gilbert (BCG).15 They study price asymmetry at various levels of the petroleum products 
industry with a lag adjustment model that incorporates an error-correction term to account 
for a long run stable relationship between the two price series. If retail and wholesale 
prices tend to have a fairly constant long run relationship, the error correction term will 
13Ibid., 6-11. 
14Robert W. Bacon, "Rockets and Feathers: The Asymmetric Speed of Adjustment 
of U.K. Retail Gasoline Prices to Cost Changes," Energy Economics 13, no. 3 (June 
1991): 211-218. 
15Borenstein, Cameron, and Gilbert, (1994). 
75 
account for this relationship. BCG model this by including in the regression equation an 
error correction term of the form : 
Retprice1 t-1 =<1>0 +<1>1 Whlprice1 ,_1 +£1 t-1 
' ' ' 
(4.10) 
They then partition the wholesale prices into increasing and decreasing first differences. 
For our purposes, this can be represented as: 
AWhlprice1; = Whlprice1, -Whlprice1 t-1' if Whlpricett -Whlprice1 t-1>0, 
' ' 
= zero otherwise 
AWhlprice1; = Whlpricett-Whlprice1,,_1, if Whlprice 11 -Whlprice1,t-1<0, 
= zero otherwise 
(4.11) 
BCG fmd that the positive and nega~ive adjustment processes are statistically 
different from each other up to 8 weeks after a change in wholesale price, with positive 
responses occurring more quickly than negative.16 Following their lead, we include the 
same error correction term in a slightly different model.17 The model estimated is shown 
in equation 4.12 and the results are listed in Table 7. 
n 
ANRetpr1, =a0 + .E (p;Awhlprice1~-k +P;Awhlprice1~-k) 
k=O 
+6/YRetpr1.t-t +61 Whlprice1.t-t +62TIME1, +£11 
16lbid., 26. 
(4.12) 
17Their model also includes lagged positive and negative differences of the 
dependent variable as explanatory variables. Here, the disturbance term is corrected for 
autocorrelation. 
76 
The error-correction term insures that retail prices fully adjust to wholesale prices 
according to their long run relationship.18 Note that dividing 6 1 by 60 yields an estimate 
of the coefficient <l>b shown in equation4.10. The results, reported in Table 7, yield a 4>1 
equal to .8. Although less than unity, this constraint results in plausible time paths of 
adjustment. The time paths using the results presented in Table 7 are given in Figure 10. 
The results indicate a large first period response, followed by a negative adjustment and 
then a gradual ascent to a long run equilibrium. This is consistent with the previous 
results from equations 4.5 and 4.6. A slight amount of asymmetry is indicated using this 
model, although in the opposite direction than that which supports the premise of market 
power.19 An F-test also rejects the hypothesis that the contemporary positive and 
negative changes are equal (F = 5.94). 
Another approach to testing for price asymmetries was proposed by Wolfram.20 
As with the procedure by Borenstein, Cameron, and Gilbert, this procedure partitions the 
data by positive and negative first differences such as: 
18Excluding the error correction term yields results that do not indicate full 
adjustment of retail prices to changes in wholesale prices. 
19 An F -test of the difference between equation 4.12 and a model that does not split 
wholesale prices into positive and negative differences is statistically significant. This 
indicates a statistically significant difference between positive and negative changes. 
Therefore, some asymmetry is present. 
20Rudolf Wolfram, "Positivistic Measures of Aggregate Supply Elasticities: Some 
· New Approaches - Some Critical Notes," American Journal of Agricultural Economics 
53 (1971): 356-59. James P. Houk, "An Approach to Specifying and Estimating 
Nonreversible Functions," American Journal of Agricultural Economics 59 (Aug., 1977): 
570-72, proposes a method of partitioning the independent variable with a slightly 
different specification than that employed by Wolfram. 
77 
Table 7. Test of Asymmetrical Pnce Responses 
Dependent Variable = aNRetpr 
+a Whlpricel 
+a Whlpricet-1 
-a Whlpricel 
-a Whlpricet-1 
Retprice1_1 
Whlprice1_1 
TIME 
0.406 
-0.092 
0.569 
-0.159 
-0.099 
0.080 
-0.003 
9.286*** 
-2.138** 
14.940*** 
-4.038** 
-14.840*** 
10.040***' 
-1.860* 
Adjusted R_2 = .18 Denotes significance at: *** = 1%, ** = 5%, * = 10% 
Number of Observations = 3640 · 
Figure 10. Time Path for the Cumulative Adjustment of Retail Prices 
to Wholesale Price Changes 
0.9 
0.8 
- 0.7 c:: 
Q) 
~ 0.6 
::l 
~ 0.5 
~ 0.4 
"5 
E 8 0.3 
0.2 
0.1 
0 
I 
II 
II 
II 
!j' 
0 
Ill 
I 
I 
I 
I I 
I I 
I 
~ 
-~ -II It-- __, ,__ 
II" 
-
------
--
f.---" 
2 3 4 5 6 
Weeks after Wholesale Price Change 
-'ltl'- Positive Adjustment -a- Negative Adjustment 
I--
7 
78 
AWhlpriceitc.O 
AWhlprice it:S 0 
(4.13) 
Then, a new vector of prices, WP 'it , is formed for the increasing phase. The initial value 
of this vector is formed by adding the first A Whlpriceit ~ 0 to the initial value of the 
vector of observed wholesale prices. Succeeding values of A Whlpriceit ~ 0 are then 
summed to the previous values of WP'it· A new vector of prices, WP"it, is formed for 
the decreasing phase in the same manner using values of A Whlpriceit :s;; 0. The vector of 
lagged whol~sale prices is partitioned following the same procedure. In this case 
however, the first observation is lost. The initial value of the lagged wholesale prices 
thus becomes the base value upon which all succeeding positive or negative changes are 
summed. Applying this procedure for each city yielded mixed results with many of the 
estimated coefficients being unrealistic. This could be the consequence of using a lagged 
term that requires the exclusion of the initial observation. 
The preceding tests indicate only a slight amount of asymmetry and the direction 
of the asymmetry is sensitive to the model being used. Thus, the results are not 
consistent with the hypothesis that the retail gasoline bus~ess is non-competitive. Recall 
that it was argued that retail prices would respond more quickly to positive changes in 
wholesale price than negative changes and the majority of the adjustment would not 
occur in the contemporary period. With these issues in mind, we may now proceed with 
measuring differences in price responsiveness between law and non-law states. 
To accomplish this, consider equation 4.12. This model is particularly 
appropriate since the hypothesized effect will occur for price increases. Interacting the 
79. 
LAW dummy variable with the current and lagged positive changes of wholesale price 
allows a measure of any difference between SBC law states and non-SBC law states. The 
equation estimated is 
n 
I!:JVRetpr11 = a 0 + L [p;.6.whlprice1;_k + p;.6.whlprice1~-k] + 
k=O 
n L [(y;.6.Whlprice1;_k*LAW1)+60NRetpr1,t-t + 
k=O 
61 Whlprice1.t-t +62TIME1, +ett , 
(4.14) 
for n = 2. The hypothesis is supported if the coefficients on the interaction terms (yk+) are 
\ 
positive. The results of this regression are shown in Table 8. None of the yk+ coefficients 
are significantly different from zero.21 
Conclusion 
The above tests, utilizing the retail margin as the dependent variable, indicate that 
SBC laws do affect the pricing behavior of firms in the retail gasoline business, at least to 
some extent. It was initially shown that margins were moderately higher in states with 
SBC laws. The majority of this effect, however, was attributed to active enforcement of 
the laws. These results are also consistent with the premise that predatory pricing is not a 
common occurrence in the retail gasoline business. 
Regression runs using the retail price as the dependent variable and allowing for a 
21These results are not affected if the LAW variable is also interacted with the 
negative changes. 
Table 8. The Differences in Retail Price 
+A Whlpricet 
+A Whlpricet-1 
-A Whlpricet 
·A Whlpricet.1 
+A Whlpricet * LAW 
+A Whlpricet.1 * LAW 
Retpricet_1 
Whlpricet_1 
TIME 
80 
0.400 
-0.093 
0.568 
-0.160 
-0.047 
0.007 
-0.099 
0.081 
-0.003 
to Increases in Wholesale Prices 
8.646*** 
-2.027** 
14.920*** 
-4.051 *** 
0.453 
0.067 
-14.830*** 
10.040*** 
-1.829* 
Adjusted R2 = .18 
Observations= 3640 
Significant at:***= 10%, **= 5%, * = 1% 
distributive lag process also revealed a statistically significant positive effect of the SBC 
laws on price. However, it is also clear that the error structure in those models is 
especially complex. Introducing a more detailed adjustment model allowed for the 
testing of additional hypotheses. These results, however, do not indicate that retail prices 
in states with SBC laws respond more quickly to increases in whol~sale prices, therefore 
suggesting that SBC laws are not always binding. 
81 
CHAPTERS 
CONCLUSION 
The objective of this thesis was to quantify the effects of Sales Below Cost laws 
in the retail gasoline business. To accomplish this, implications of the laws were iriferred 
from economic theory and a workable definition of a predatory price. Based on the 
conditions of ~ost advantage, expected profits, and entry deterrence, the potential for 
predatory pricing in the retail gasoline business appears unlikely. In comparison with the 
Areeda and Turner rule, the SBC laws appear to define cost so as to restrict competition 
and set a binding price floor. 
But if the laws are unnecessary and their effect is to restrict, rather than protect, 
competition, why were they enacted? A plausible answer is that political pressures, 
initiated by special interest groups, namely independent dealers, have resulted in 
regulations that protect less efficient firms. This hypothesis would be supported if retail 
margins in states with SBC laws were higher than states without such laws. Additionally, 
if the laws are binding, retail gasoline stations in states with SBC laws will increase retail 
. 
prices more quickly in response to wholesale price increases. 
Several preliminary analyses were conducted to obtain a sense of how actively the 
laws are enforced. Active enforcement would be prima facie evidence that SBC laws 
are binding. The analysis revealed scant evidence of enforcement, except in the state of 
82 
Alabama. However, this is merely prima facie evidence and the mere threat of penalties 
may be sufficient to make a SBC law binding. Thus we began to ·look for direct evidence 
of a binding price floor. By comparing descriptive statistics, it was revealed that the 
minimum value of margins were higher in states with SBC laws. And in Montana, retail 
margins were typically above the minimum markup. These results suggested that the 
pricing behavior of firms may be affected by these regulations. 
While suggestive, further defining and quantifying the effect required empirical 
tests using time series, cross sectional data. The results indicated that margins are 
affected by the SBC laws, but the majority ofthis effect is the result of active 
enforcement of the law in Alabama. Also, the speed of retail price adjustment in response 
to increases in wholesale prices was not affected. Although the effect of these laws may 
be weak, the results support the premise that the industry is not characterized by 
predatory pricing. Rather than protecting independent dealers from predatory behavior, 
SBC laws appear to be directed at protecting these same firms from competition. 
These results should be understood in the context of several characteristics of the 
data set. First, the cost data used in the margin equation is aggregate city-wide data and 
although the data set includes 3720 observations, the variability in these cost measures 
only occurs acr9ss the 40 cities. Second, the nearly impossible task of obtaining the 
invoice cost of fuel to a typical retailer could result in overestimated margins. Some 
correction for this problem was offered by the inclusion of the population density variable 
to account for expected differences in types of ownership based on demographics. Third, 
measurement of the varying seasonal effect between cities was made difficult by 
83 
inadequate data on the quantity of gasoline consumed. A more extensive time series 
could prove beneficial because seasonal patterns may be identifi~d with more certainty 
and time series variability could be added to the· explanatory variables. 
An interesting and additional aspect ofthis study was the analysis of the lag 
structure of prices and the hypothesis or' asymmetry between positive and negative 
changes in price. The assumption that the retail gasoline business is relatively 
competitive imposes several prior expectations about such price behavior. We expect, for 
example, that retail prices will respond symmetrically to changes in wholesale prices. 
Since the existing literature arg'!les otherwise, further study in this area is warranted. 1 
This would necessitate, however, reliable price data that is preferably measured in time 
periods that allow reasonable estimates. An additional area of interest relating to the 
gasoline business is the types of contractual ownership that exist. Although hampered by 
the accessibility of data, this issue is relevant to recent legislation that, as with Sales 
Below Cost laws, assumes strategic behavior on the part of large firms. 
1For example, Borenstein (1991) argues that price discrimination occurs within 
the retail ga5oline business based on customers willingness switch between retail stations. 
84 
REFERENCES CITED 
85 
References Cited 
Alabama. Motor Fuel Marketing Act. Statutes, Annotated. 1984. 
American Petroleum Institute. Policy Analysis Department. "An Overview of Gasoline 
Prices and Their Determination." Washington D. C.: American Petroleum 
Institute, 1990. 
American Petroleum Institute. State Governmental Relations Department. "State 
Gasoline Excise Tax Rankings- May 4, 1993." Washington D.C.: American 
Petroleum Institute, 1993. 
Areeda, Phillip, and Donald F. Turner. "Predatory Pricing and Related Practices Under 
Section 2 ofthe Sherman Act." Harvard Law Review 88 (1975): 697-733. 
Bain, Joe S. Barriers to New Competition. Cambridge: Harvard University Press, 1956. 
Barron, John M., Mark A Lowenstein, and John R. Umbeck, "Predatory Pricing: The 
Case of the Retail Gasoline Market." Contemporary Policy Issues 3 (Spring 
1985): 131-39. 
Borenstein, Severin. "Selling Costs and Switching Costs: Explaining Retail Gasoline 
Margins~" The Rand Journal ofEconomics 22 (Autumn 1991): 354-69. 
Borenstein, Severin, A Colin Cameron, and Richard Gilbert. "Do Gasoline Prices 
Respond Asymmetrically to Crude Oil Price Changes." Program on 
Workable Energy Regulation, University of California Energy Institute, PWP-
001R, 1994. 
Borenstein, Severin, and Andrea Shepard. "Dynamic Pricing in Retail Gasoline 
Markets." National Bureau ofEconornic Research, Research Paper No~ 1269, 
Sept., 1993. 
Bork, Robert H. The Antitrust Paradox. New York: Basic Books, Inc., 1978. 
Breusch, T.S., and AR. Pagan. "A Simple Test for Heteroscedasticity and Random 
Coefficient Variation." Econometrica 47 (1979): 1287- 1294. 
86 
Burns, Malcolm R. "Predatory Pricing and the Acquisition Cost of Competitors." Journal 
ofPolitical Economy 94, no. 2 (April1986): 266-296. 
Carlton, Dennis W., and Jeffrey M. Perloff. Modern Industrial Organization. New York, 
N.Y.: Harper Collins College Publishers, 1994. 
Colorado. Unfair Practices Act. Revised Statutes, Annotated. 1993. 
Dahl, Carol and Thomas Sterner. "Analyzing Gasoline Demand Elasticities: A Survey." 
. Energy Economics (July 1991): 203:-210. 
Dennison, Mike. "Restriction on Gasoline Prices Widely Criticized." Great Falls 
Tribune, 25 Sept. 1995,.2(A). 
Dougher, Rayola, and Thomas F.Hogarty. "The Impact of State Legislation on the 
Number ofRetail Gasoline Outlets." Washington D.C.: American Petroleum 
Institute, Research Study 062, 1991. 
Easterbrook, Frank H. "Predatory Strategies and Counterstrategies." The University of 
Chicago Law Review 48, no. 2 (1981): 263- 337. 
Florida. Motor Fuel Marketing Practices Act. Statutes, Annotated. 1985. 
Gellerson - Sawides, Evi. "The Effects of State Below Cost Selling Laws on Retail 
Prices ofMotor Gasoline." Washington D.C.: American Petroleum Instititute, 
Research Study 043, 1987. · · 
Hogarty, Thomas F., and Perry M. Lindstrom. "Empirical Examination of Allegations of 
Below Cost Retail Selling of Gasoline by Refiners." Washington D.C.: American 
Petroleum Institute, Research Study 038, 1986. 
Houck, James P. "An Approach to Specifying and Estimating Nonreversible Functions." 
American Journal of Agricultural Economics (Aug. 1977): 570-72. 
Kmenta, Jan. Elements of Econometrics: 2nd ed. New York: Macmillan Publishing 
Company, 1986. 
Koyck, L.M. Distributed Lags and Investment Analysis. Amsterdam: North Holland 
Publishing Company, 1954. 
Massachusetts. Unfair Sales Act for the Retail Sale ofMotor Fuels. General Statutes, 
Annotated. 1950. 
87 
McGee, JohnS. "Predatory Pricing Revisited." The Journal ofLaw and Economics 23, 
no. 2 (Oct. 1980): 289-330. 
Missouri. Missouri Motor Fuel Marketing Act. Code, Annotated. 1993. 
Montana. Montana Retail Motor Fuel Marketing Act. Code Annotated. 1991. 
Motor Fuel Pricing Problems. prepared by Paul E. Verdon, Staff Researcher. Helena, 
MT: Montana Legislative Council, 1990. 
New Jersey. Unfair Motor Fuels Practice Act. Statutes, Annotated. 1953. 
Oil and Gas Journal. (Tulsa, OK) 23 Mar. 1992- 27 Dec. 1993. 
Scherer, F.M. "Predatory Pricing and the Sherman Act: A Comment." Harvard Law 
Review 89 (1976): 869-890. 
Shepard, Andrea. "Contractual Form, Retail Price, and Asset Characteristics in Gasoline 
Retailing." The Rand Journal ofEconomics 24 (Spring 1994): 58-77. 
Shin, David. "Do Product Prices Respond Symmetrically to Changes in Crude Prices?" 
Washington D.C.: American Petroleum Institute, Research Study #068, 1992. 
Solomon, Caleb. "Independent Gas Stations Cry Foul Over Price Wars." The Wall 
Street Journal, 1 April 1991, 7(B). 
Sorenson, Philip E., Rayola S. Dougher, Lisa A. Ho:finan, and Thomas. F. Hogarty. "An 
Economic Analysis of the Distributor-Dealer Wholesale Gasoline Price Inversion 
of 1990: The Effects ofDifferent Contractual Relations." Washington D.C.: 
American Petroleum Institute,-1991. 
The Road Information Program. 1993 State Highway Funding Methods. Washington, 
D.C.: The Road Information Program, 1993. 
Tennessee. Petroleum Trade Practices Act. Code, Annotated. 1977. 
Tennessee. Unfair Sales Law. Code, Annotated. 1937. 
U.S. Department of Commerce. Economics and Statistics Administration. County and 
City Data Book: 1994. Washington D.C.: U.S. Government Printing Office, 
1994. 
88 
____ . 1992 Census ofRetail Trade. Geographic Area Series. Washington D.C.: 
U.S. Government Printing Office, 1994. 
U.S. Department ofEnergy. Energy Information Administration. Petroleum Marketing 
Monthly. Washington D.C.: U.S. Government Printing Office, Mar. 1992- Dec. 
1993. 
----·· Petroleum Stipply Annual1993. Washington D.C.: U.S. Government 
Printing Office, 1994. 
U.S. Department ofEnergy. Minimum Markup Laws in Gasoline Marketing: An 
Economic Analysis and a Legal, Economic and Legislative Review. Washington 
D.C.: U.S. Government Printing Office, 1985. 
U.S. Department of Transportation. Federal Highway Administration. Monthly Motor 
Fuel Reported by States. Washington D.C.: U.S. Government Printing Office, 
Mar. 1992 - Dec. 1993. 
____ . Highway Statistics. Washington D.C.: U.S. Government Printing Office, 
1992, 1993. 
Utah. Motor Fuel Marketing Act. Code, Annotated. 1981. 
Williamson, Oliver E. "Predatory Pricing: A Strategic and Welfare Analysis." The Yale 
Law Journal 87, no. 2 (1977): 284- 340. 
Wolffram, Rudolf. "Positivistic Measures of Aggregate Supply Elasticities: Some New 
Approaches- Some Critical Notes." American Journal of Agricultural Economics 
53 (1971): 356-59. 
89 
APPENDICES 
90 
APPENDIX A 
INTERPOLATION OF WEEKLY QUANTITY DATA 
1) Let: 
91 
Interpolation Method of Weekly Quantity/Seasonality Variable 
a) ~vol = the monthly total volume for the current month. This value will 
become the quantity value for the last day of the current month. 
b) lagmvol = the monthly total volume for the previous month. This value will 
become the quantity value for the last day of the previous month. 
2) Calculate the slope of a line between mvol and lagmvol. 
slope = (mvol - lagmvol)/no. of days in current month 
3) Multiply slope times the number of days corresponding to the date of the price 
observation. 
4) Add result from (3) to lagmvol to get the weekly quantity for that observation. 
92 
APPENDIXB 
DESCRIPTIVE STATISTICS 
93 
Table 9. Descriptive Statistics ofKey Variables (part 1) (3 720 observations) 
Variable Mean St. Dev. Minimum Maximum 
Whlprice (¢) 61.023 7.260 38.000 81.000 
Retprice (¢) 111.870 8.708 85.100 142.500 
Margin(¢) 15.829 5.748 -3.000 37.522 
Retwg ($) 226.210 24.118 187.790 288.340 
Prval ($,000) 84.401 56.950 25.600 298.900 
Popden 5.854 4.580 0.746 23.671 
(,000/sq mile) 
NSS 0.050 0.218 0.000 1.000 
LAW 0.167 0.373 0.000 1.000 
wt 0.011 0.103 0.000 1.000 
AL 0.025 0.156 0.000 1.000 
Tax(¢) 35.015 4.990 22.100 49.999 
Time 47.000 26.849 1.000 93.000 
94 
Table 9. Descriptive Statistics of Key Variables (part 2) (3640 Observations) 
Name Mean St. Dev. Minimum Maximum 
(all values are in cents) 
Nretprit 76.984 7.212 51.533 101.150 
ANRetprit -0.034 2.414 -13.643 ·18.100 
Whlpriceit 61.047 7.328 38.000 81.000 
' 
Whlpricei,t-1 61.234 7.071 38.000 81.000 
Whlpricei,t-2 61.415 6.805 38.000 81.000 
A Whlprice+it 0.495 0.873 0.000 11.700 
A Whlprice -it -0.682 1.048 -13.800 0.000 
A Whlprice+i 1_1 0:498 0.873 0.000 11.700 
A Whlprice -i,t-1 -0.679 1.049 -13.800 0.000 
TIME 48.000 26.271 3.000 93.000 
95 
APPENDIXC 
MOTOR FUELS TAXES BY CITY 
96 
Table 10 .. Motor Fuels Taxes and Computational Formulas as of Dec. 27, 1993 (part 1) 
City sT· FT Misc.Per Sales Tax State Local Sales Tax Formula 
(¢) (¢) .. Gallon Base Sales Sales 
Tax(¢) (STB) Tax Tax 
(¢) (%) (%) 
Albuqurque 23 18.4 N/A N/A 
Atlanta 7.5 18.4 N/A RP-ST 4 1 STB - (STB/1.05) 
Baltimore 23.5 18.4 N/A N/A 
Birmingham 18 18.4 1 N/A 
Boston 21 18.4 N/A N/A 
Buffalo 22.24 18.4 N/A RP-8 4.5 4 STB - (STB/1.08) 
Cheyenne 9 18.4 N/A N/A 
Chicago 19 18.4 NIA RP-ST 6.25 1.75 STB - (STB/1.08) 
Cleveland 22 18.4 N/A N/A 
Dallas 20 18.4 N/A N/A 
Denver 22 18.4 N/A N/A 
Des Moines 20 18.4 NIA N/A 
Detroit 15 18.4 N/A RP-16 4 STB - (STB/1.04) 
Houston 20 18.4. N/A N/A 
Indianapolis 15 18.4 N/A RP-ST-FT 5 STB - (STB/1.05) 
Kansas City 13.03 18.4 N/A N/A 
Los Angeles 17 18.4 N/A RP 7.25 1 STB - (STB/1.0825) 
Memphis 20 18.4 N/A N/A 
Miami 11.8 18.4 10.3 N/A 
Milwaukee 23.2 18.4 N/A N/A 
Minneapolis 20 18.4 N/A N/A 
New York 22.24 18.4 N/A RP-8 4.5 4.25 STB - (STB/1.0825) 
New Orleans 20 18.4 N/A N/A 
Newark 10.5 18.4 4 N/A 
97 
Table 10. Motor Fuels Taxes and Computational Formulas as of Dec. 27, 1993 (part2) 
Norfolk 17.5 18.4 N/A N/A 
Okla. City 17 18.4 N/A N/A 
Omaha 24.4 18.4 N/A N/A 
Pheonix 18 18.4 N/A N/A 
Philadelphia 22.35 18.4 N/A N/A 
Pittsburgh 22.35 18.4 N/A N/A 
Portland 24 18.4 N/A N/A 
Salt Lk. City 19 18.4 N/A N/A 
San Antonio 20 18.4 N/A N/A 
San Diego 17 18.4 N/A RP 7.25 0.5 STB - (STB/1.0775) 
San Francisco 17 18.4 N/A RP 7.25 1.25 STB - (STB/1.085) 
Seattle 23 18.4 N/A N/A 
St. Louis 13.03 18.4 N/A N/A 
Tulsa 17 18.4 N/A N/A 
Wash., D.C. 20 18.4 N/A N/A 
Wichita 18 18.4 N/A N/A 
ST = State per Gallon Tax 
FT = Federal per Gallon Tax 
RP = Retail Price 
N/A =Not Applicable 
• Some state taxes vary during during the sample period . 
.. Prior to Oct. 1, 1993, the Federal Gasoline Tax was 14.1 cents. 
Sources: Federal Highway Administration, Monthly Motor Fuel Reported by States (various da~es); The 
Road Information Program, 1993 State Highway Funding Methods; American Petroleum Institute, 
"State Gasoline Excise Tax Rankings - May 4, 1993 ". 
98 
APPENDIXD 
SALES BELOW COST LAWS BY STATE 
99 
Tablell. Sales Below Cost Laws by State 
State Sales Below Cost Law Minimum Markup 
Provision 
Alabama 
" Arizona 
California 
Colorado 
" District of Columbia 
Florida 
" Georgia 
Illinois 
Indiana 
Iowa 
Kansas 
Louisiana 
Maryland 
Massachusetts 
" " Michigan 
Minnesota 
Missouri 
" Montana 
" " Nebraska 
New York 
New Mexico 
New Jersey 
" Ohio 
Oklahoma 
Oregon 
Pennslyvania 
Tennessee 
" " Texas 
Utah 
" " Virginia 
Washington 
Wisconsin 
Wyoming 
1 t/ denotes the eXIstence o:t a :Sales tle1ow Cost Law or MlDllllum Markup .Provision 
100 
APPENDIXE 
REGRESSION RESULTS OF EQUATION 4.1 
101 
Table 12. Full Regression Results of Equation 4.1 (part 1) 
Dependent Variable = Margin 
Independent Variable 
Retwg 
Prval 
Popden 
NSS 
LAW 
WI 
w2 
w3 
w4 
Ws 
w6 
w7 
Ws 
w9 
WIO 
Wn 
wl2 
Wn 
Wt4 
Wts 
wl6 
wl7 
Wts 
wl9 
W2o 
W21 
W22 · 
w23 
W24 
W2s 
w26 
W21 
W2s 
w29 
W3o 
W31 
. w32 
Coefficient Value 
-0.007 
0.040 
-0.113 
5.810 
0.841 
-12.254 
-13.762 
-16.262 
-15.231 
-13.930 
-14.337 
-15.000 
-14.924 
-13.175 
-13.286 
-12.923 
.-13.016 
-11.736 
-10.067 
-8.356 
-7.439 
-5.815 
-7.333 
-8.138 
-7.545 
-7.271 
-8.768 
-9.979 
-11.259 
-11.190 
-9.907 
. -7.759 
-7.641 
-8.132 
-8.820 
-8.642 
-8.500 
t-statistic 
-0.545 
6.053 
-2.148 
5.732 
1.976 
-18.320 
-20.680 
-24.510 
-23.010 
-21.080 
-21.730 
-22.750 
-22.650 
-20.000 
-20.170 
-19.630 
-19.770 
-17.830 
-15.290 
-12.700 
-11.300 
-8.836 
11.140 
-12.370 
-11.460 
-11.050 
-13.320 
-15.160 
-17.110 
-17.000 
-15.050 
-11.790 
-11.610 
-12.360 
-13.400 
-13.130 
-12.920 
102 
Table 12. Full Regression Results of Equation 4.1 (part 2) 
Independent Variable Coefficient Value 
w33 
w34 
W3s 
w36 
W37 
W3s 
w39 
W4o 
W41 
W42 
w43 
w44 
W4s 
w46 
w47 
W4s 
W49 
Wso 
Wsl 
Ws2 
Ws3 
Ws4 
Wss 
Ws6 
Ws7 
Wss 
Ws9 
W6o 
w61 
w62 
w63 
w64 
W6s 
w66 
w67 
W6s 
w69. 
W7o 
-7.802 
-6.677 
-4.119 
-4.074 
-3.488 
-2.498 
-2.824 
-5.375 
-5.645 
-5.645 
-5.598 
-3.062 
-3.004 
-2.984 
-2.986 
-3.003 
-6.995 
-7.049 
-7.086 
-7.121 
-7.158 
-11.479 
-11.509 
-11.560 
:-12.707 
-12.751 
-12.832 
-12.889 
-11.145. 
-11.193 
-11.218 
-11.234 
-11.223 
-6.510 
-6.489 
-6.624 
-6.487 
-4.485 
t-statistic 
-11.860 
-10.150 
-6.259 
-6.191 
-5.300 
-3.796 
-4.291 
-8.168 
-8.577 
-8.578 
-8.507 
-4.653 
4.566 
-4.535 
-4.537 
4.564 
-10.630 
-10.710 
10.770 
-10.820 
-10.880 
-17.450 
-17.500 
-17.580 
-19.320 
-19.400 
-19.520 
. -19.610 
-16.970 
-17.040 
-17.090 
-17.120 
-17.120 
-9.935 
-9.911 
-10.130 
-9.931 
-6.877 
103 
Table 12. Regression Results of Equation 4.1 (part 3) 
Independent Variable Coefficient Value 
w71 
Wn 
w73 
W74 
W75 
w76 
w77 
w1s 
W79 
Wso 
Wsi 
W~2 
Ws3 
Ws4 
Wss 
Ws6 
Ws7 
Wss 
Ws9 
W9o· 
W91 
w92 
Constant 
Adjusted R2 = .49 
3720 Observations 
-4.484 
-4.482 
-4.492 
-4.518 
-8.393 
-8.372 
-8.521 
-8.529 
-6.590 
-6.618 
-10.961 
-10.978 
-9.450 
-9.453 
-9.475 
-9.497 
,.4.220 
-4.225 
-4.225 
-4.222 
-0.023 
-0.021 
22.549 
t-statistic 
-6.887 
-6.898 
-6.930 
-6.992 
-13.030 
-13.060 
-13.360 
-13.460 
10.480 
-10.620 
-17.780 
-18.050 
-15.800 
-16.140 
-16.600 
-17.190 
-7.966 
-8.425 
-9.075 
10.080 
-0.064 
-0.081 
8.914