Calendar synchronization of gasoline price increases

• Published date: 01 April 2019
• DOI: http://doi.org/10.1111/jems.12267
• Author: Michael D. Noel
• Date: 01 April 2019

355

J Econ Manage Strat. 2019;28:355–370. wileyonlinelibrary.com/journal/jems © 2018 Wiley Periodicals, Inc.

Received: 12 October 2016 Revised: 11 May 2018Accepted: 18May2018

DOI: 10.1111/jems.12267

ORIGINAL ARTICLE

Calendar synchronization of gasoline price increases

Michael D. Noel

Department of Economics, TexasTech Uni-

versity,Lubbock, Texas

Correspondence

MichaelD. Noel, Associate Professor, Depart-

mentof Economics, Texas Tech University,

Lubbock,TX.

Email:michael.noel@ttu.edu

Abstract

In many retail gasoline markets with Edgeworth price cycles, large and regular price

increases occur on the same day of the week every week, that is, they are calendar

synchronized. In this article, I test whether calendar synchronization leads to higher

or lower consumer expenditures on gasoline compared to a world with cycles but

without calendar synchronization. On one hand, firms may attempt to trigger price

increases just prior to periods of normally high demand. On the other, consumers

may be better able to predict and shift purchases to low price days of the cycle. Using

high-frequency gasoline volume data and matching it to high-frequency price data, I

find that the latter effect dominates. All else equal, consumer expenditures on gasoline

fall with calendar synchronization in the study markets. I also calculate intertemporal

price elasticities and find them to be high.

KEYWORDS

calendar effects, dynamic oligopoly, Edgeworth price cycles, prediction, quantity-weighting, retail gaso-

line

JEL CLASSIFICATION:

L1, L2, L4, M3, Q4

1INTRODUCTION

Economists have long been interested in the striking pattern found in retail gasoline prices in many cities around the world. In

these cities, retail gasoline prices follow a high-frequency, asymmetric price cycle in which prices rise quickly, often 10% or

more over the course of a day or two, and then fall back more slowly, often over a week or two. The cycle repeats itself again

and again even in the absence of wholesale price changes. In some cities, the asymmetric cycle can be especially rapid, with

prices rising and falling asymmetrically in a matter of only hours instead of days or weeks.

Figure 1 shows an exampleof the retail gasoline price cycles in Pert h, Australia,in 2012 and 2013. Cycles similar to these have

been documented in other Australian cities, Canada, the United States, and various European countries. The leading explanation

for the cycles is that they are Edgeworth price cycles, the outcome of a dynamic oligopoly game formalized theoretically by

Maskin and Tirole (1988). Empirical studies finding support for the Edgeworth price cycle theory include Eckert (2003), Noel

(2007a, 2007b), Atkinson (2009), Lewis (2009, 2012), Wang (2009a, 2009b), Doyle, Muehlegger, and Samphantharak (2010),

and others.

Although the potential causes of the cycles have been well-studied, little attention has been given to the precise timing of the

ups and downs of the cycles. In particular, the timing of the relenting phase (i.e., when prices increase) has been a controversial

topic among consumers, regulators, and economists. A close inspection of Figure 1 reveals the issue. The cycles in Perth, as

in many other cities, are not just about a week long—they are exactly a week long. Peaks and troughs are synchronized to the

calendar week. Troughs always occurred on Wednesdays and peaks always occurred on Thursdays during this period. Weekly

synchronized cycles such as these have been observed in other major Australian cities as well (Noel & Chu, 2015; Valadkhani,

2JOURNAL OF ECONOMICS & MANAGEMENTSTRATEGY

356

120 130 140 150 160

Price (cpl)

7/1/2012 10/1/2012 1/1/20 13 4/1/2013 7/1/2013

Retail Wholesale

FIGURE 1 Example of retail gasoline price cycles, perth in 2012–2013

2013), along with Norway (Foros & Steen, 2013), the United States (Lewis, 2012) and Canada (Byrne, Leslie, & Ware, 2015;

Noel, 2007a). More recently, daily-frequency price cycles synchronized to the hour of the day have been observed in Canada

(Atkinson, Eckert, & West, 2014; Noel, 2015) and Germany (Haucap, Heimeshoff, & Siekmann, 2016; Siekmann, 2017).

This invites the obvious question—conditional on the presence of retail gasoline price cycles, does the calendar synchroniza-

tion feature of those cycles benefit or harm consumers, all else equal? Are firms timing the relenting phases so that the “high

price days” now coincide with the days of the week when gasoline demand would normally be high anyway? Or are consumers

using the better predictability of the cycle troughs to shift their purchases to “low price days”? The first potential effect is the

“relenting-phase-shifting” effect by firms, which contributes to higher consumer expenditures, and the second is the “intertem-

poral switching effect” by consumers, which contributes to lower consumer expenditures. In this article, I use high-frequency

data on gasoline prices and gasoline volumes to estimate the net effect of calendar synchronization on the prices consumers pay,

conditional on the presence of price cycles, and holding all else about the cycles and the marketplace equal.

The question I am asking is different than the question of whether price cycles are beneficial or harmful to consumers as a

whole, compared to a world with no cycles. Recent research suggeststhat cycles are in fact beneficial overall (Doyle et al., 2010;

Lewis & Noel, 2011; Noel, 2002, 2015; Siekmann, 2017; Zimmerman, Yun, & Taylor, 2013). These articles do not consider the

calendar synchronization feature itself, however, and that is my purpose here. To my knowledge, this is the first study to do so. I

envision two otherwise identical worlds—one in which retail gasolinepr ice cycles are present and synchronizedto the calendar,

and a counterfactual world in which the same basic cycles are present but not synchronized to the calendar.

Past studies have made conjectures about how calendar synchronization might affect consumer expenditures. Noel (2007a)

examines Canadian cycles and suggests that firms may havean incentive to trigger relenting phases at certain times of the week,

prior to days with normally high demand. Foros and Steen (2013) examine cycles in Norway and conjecture that firms may be

strategically coordinating relenting phases on specific days of the week to increase their profits. In contrast, Noel (2012) and

Noel and Chu (2015) conjecture that calendar synchronization might be beneficial to consumers by making it easier for them

to predict cycle troughs and buy at relatively low prices. In policy discussions, a sort of folk theory has developed in recent

years that begins with the presumption that calendar synchronization is an undesirable feature of price cycles that is necessarily

harmful to consumers.

In spite of all the conjectures and presumptions, no study has been able to actually test them. One obstacle has been the

difficulty in obtaining high-frequency gasoline volume data (i.e., gasoline quantity data). Publicly available volume data are

rarely available at better than a monthly frequency, but to address the question at hand in a world with weekly price cycles,

volume data at the daily frequency level is needed to match daily changes in volumes with daily changes in prices.1

I overcome this difficulty with access to a limited amount of high-frequency daily gasoline volume data which I can then

match to daily gasoline price data—for all five major Australian capital cities of Adelaide, Brisbane, Melbourne, Sydney, and

Perth, over several years. High-frequency volume data are largely missing in the gasoline literature and this article represents a

rare exception in that regard.

I offer several contributions. The first is the application methodology itself. In place of presumptions and conjectures, this

article lays out a simple and transparent method for evaluatingthe effects of calendar synchronization and volume shifting for the

first time. It compares consumer expenditures in the presence of calendar synchronized price cycles to consumer expenditures