Daily Price Cycles and Constant Margins: Recent Events in Canadian Gasoline Retailing.

AuthorAtkinson, Benjamin
  1. INTRODUCTION

    Retail gasoline pricing in Canada has typically followed two or three distinct patterns (see Eckert, 2003 and Noel, 2007 for discussion). Many markets have shown considerable price rigidity relative to wholesale prices, with prices remaining constant often for weeks or months at a time, despite wholesale prices typically adjusting daily. Examples of markets that have exhibited price rigidity historically, possibly with occasional price wars, have been cities in western Canada (such as Regina and Winnipeg), northern Ontario (e.g., Timmins), and in Atlantic Canada prior to price regulation.

    In contrast, other cities exhibit the Edgeworth cycle pattern now widely documented in Canada, the U.S., Europe, and Australia. In such a pattern, retail prices increase quickly within one or two days, and fall slowly back to initial levels, typically within days or weeks. With some exceptions, Edgeworth cycles in Canadian retail gasoline prices have been concentrated in Ontario and Quebec (see Eckert, 2003 and Noel, 2007). Finally, Noel (2007) classifies markets that appear to exhibit neither cycles nor price rigidity as following cost-based pricing; Ottawa during the late 1980s and early 1990s is provided as an example.

    While price cycles in Canadian cities have been described and analyzed, certain unexpected changes to cycle pricing patterns in several Canadian cities have not. To illustrate, Figures 1(a) and 1(b) plot, for the city of Toronto, weekly retail and rack (wholesale) prices from January 2000 to December 2010 as well as the retail-rack markup (the "margin"). (1) While no changes in pricing are immediately obvious from Figure 1(a), Figure 1(b) illustrates two decreases in the volatility of the retail margin (the difference between the ex-tax retail price and the wholesale price), which by the end of the sample appears to be fixed. In particular, in early 2007, margins became essentially constant, first at exactly 5.0 cents per liter (cpl), increasing eventually to a roughly constant 7.0 cpl by the end of 2010. (2)

    One purpose of this paper is to provide detailed evidence regarding the changes to pricing patterns in a number of Canadian cities. Establishing these stylized facts is an important first step that should lead to extensions of price cycle theory. Using high frequency retail price data obtained from GasBuddy.com, we demonstrate that the volatility changes exhibited in Toronto appear to correspond to an increased frequency of the price cycle, and replacement of the cycle with fixed retail margins. These changes are identified in three other Ontario cities. While multiple factors may have contributed to the first pricing change, we show that the second change corresponds closely to a refinery fire in southern Ontario; this temporary event seems to have triggered a permanent change in equilibrium behavior.

    As a second contribution, this paper illustrates the effect that price data frequency can have on policy analysis. Much of the analysis of Canadian gasoline retail pricing has been conducted using price data from Kent Marketing Services, in which a sample of stations in a city is surveyed each Tuesday morning. Examples of articles making use of these data include Eckert (2003), Sen (2003, 2005), and Erutku and Hildebrand (2010). Such data have also played an important role in policy analysis, being used repeatedly in reports issued by the Competition Bureau and Natural Resources Canada. (3) In this paper, we combine the Kent price data with high-frequency price data obtained from GasBuddy.com to examine how the discrete changes in pricing patterns are reflected in the different data sources. As well, the changes in pricing patterns in certain Canadian markets provide a useful example of how observing prices at a lower frequency than the price cycle can lead to misleading conclusions regarding the effect of a structural change.

  2. ECONOMIC LITERATURE ON GASOLINE PRICE CYCLES

    The existence of asymmetric price cycles in gasoline retailing has given rise to a large literature aimed at understanding why these cycles exist, and what factors determine where they occur and their main features, including speed and height. (4) The starting point of most such research is the alternating-moves model developed by Maskin and Tirole (1988). (5) In this model, two identical firms maximize their present discounted stream of profits by setting prices for a homogeneous product over an infinite horizon. Marginal cost is constant, there are no fixed costs or capacity constraints, and the lowest-priced firm serves the entire market; when prices are equal, they split market demand evenly. Prices are chosen in discrete time over a finite grid and a firm's strategies depend only on the most recent price set by its rival. Using this framework, Maskin and Tirole (1988) demonstrate the existence of Edgeworth cycles. In an Edgeworth cycle equilibrium, firms undercut each other repeatedly until marginal cost is reached; a war of attrition eventually ends in a price increase and the beginning of a new cycle.

    Where Edgeworth cycles are expected to occur, and what determines their shape and frequency, has been the subject of both theoretical and empirical research. Lower market concentration and an increased presence of independent (non-refiner) retailers has been found by Eckert (2003) and Noel (2007) to be associated with the existence of cycles in Canadian cities. (6) A larger number of firms is also found by Noel (2007) to be associated with shorter undercutting phases. Byrne and Ware (2011), in a recent cross-section analysis of towns in Ontario, find a U-shaped association between concentration and the presence of market size.

    In addition, the existence and nature of price cycles have been associated with the presence of retailers with convenience stores and other secondary revenue sources that exhibit demand complementarity with gasoline; see Doyle, Muehlegger and Samphantharak (2010) for evidence for U.S. cities. Price cycles have also been associated with the presence of particular "maverick" brands with a reputation for undercutting; see for example Eckert and West (2004), and Atkinson (2009). Theoretical and computational analyses by Noel (2008) and Doyle, Muehlegger and Samphantharak (2010) suggest that increased product differentiation and brand loyalty should decrease the incentive to undercut and reduce the likelihood of price cycles.

    The impact of capacity constraints is studied computationally by Noel (2008). He finds that if capacity constraints are symmetric, then as they tighten, cycles initially become longer and more asymmetric (i.e., smaller price decreases during the undercutting phase). However, if constraints become too tight then focal price equilibria replace cycles. The idea that price cycles may not exist where there are capacity constraints finds some support in Byrne and Ware (2011), who find that price cycles are less likely in remote towns in Ontario with reduced access to gasoline inventory shipments.

    When capacity constraints are asymmetric (e.g., major brands are less constrained than independents), Noel (2008) finds that high frequency "Hyper-Edgeworth Cycles" can arise that might be missed entirely depending on data frequency. Specifically, as constraints on the high-constrained firm (e.g., independent brand) tighten, that firm becomes more aggressive in the undercutting phase because its rival (the major brand) is more likely to lead the next restoration.

    Finally, the presence of retail gasoline price cycles has been associated with large chains able to coordinate the price increases at the start of the cycle. Atkinson (2009) describes the role of major brands in leading price increases in Guelph, Ontario. Evidence supporting the coordination role of large chains is also provided by Lewis (2012) for the U.S. Midwest.

    To our knowledge, the only other study to discuss the constant margin price setting behavior that we observe in certain Ontario markets is by Byrne and Ware (2011). (7) The authors use a cross-section of price data for 2007-2008 to document the presence and absence of price cycles. The authors find that over this period, Toronto and other large southern Ontario urban areas exhibit neither rigid prices nor Edgeworth cycles, but rather prices that are set as a fixed markup of the wholesale price. The authors suggest that increased competition means that major brands are unable to co-ordinate price increases; constant margins are therefore indicative of a greater degree of competition. Because of the time horizon covered in the paper, no attention is given to pricing patterns in these markets in earlier years, or the possible causes of the switch to constant margins.

  3. PRICE VOLATILITY PATTERNS AND CHANGES IN MAJOR CANADIAN CITIES

    This section identifies those regions that exhibited shocks to the pricing patterns in retail gasoline, and characterizes the pricing patterns that existed in those regions before and after the shocks. Two types of retail price data are used: weekly average prices and station level prices reported to consumer websites.

    3.1 Timing and Location of Price Volatility Changes

    The 2000s saw dramatic changes to retail gasoline price volatility in certain Canadian markets. As an illustration, Figure 1(b) provides the Toronto weekly retail margin (the difference between the before tax retail price and the wholesale price, in cpl) for the period from 2000 to 2010. Retail prices are obtained from the Kent Marketing website, and consist of the average price of a consistent sample of stations taken each Tuesday morning at 10:00 AM. These average prices are reported also on the Natural Resources Canada website, and are frequently used in Canadian academic and industry studies of gasoline retailing. Weekly wholesale (rack) prices were obtained from the website of Kent Marketing. Margins are used in Figure 1(b) in order to control for price...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT