|
Article Excerpt
Asymmetric price cycles similar to Edgeworth Cycles are appearing in increasingly many retail gasoline markets in the United States and worldwide. The asymmetry in the cycles can give rise to a finding of asymmetric price responses to cost shocks (asymmetric passthrough). This article estimates asymmetric passthrough for the market of Toronto, which exhibits cycles, and decomposes it into two components--that part explainable by the cycles and that part driven by other unknown sources. Significant asymmetric passthrough is found, with increases passed through more quickly than decreases. A significant cause of the finding is the presence of the cycles themselves.
1. Introduction
* The question of whether firms pass cost increases through to prices more quickly than they do decreases has long generated interest by economists, especially in the context of gasoline markets. A large literature has developed to examine this phenomenon--known as asymmetric passthrough--for gasoline markets in the United States and in many other countries. A widely cited article is Borenstein, Cameron, and Gilbert (1997), who find evidence of asymmetric passthrough from crude prices into retail prices in the United States and suggest retailer market power and inventory asymmetries as some possible causes. Other authors suggest other possibilities (Lewis, 2004; Radchenko, 2004) or find no asymmetry at all (Bachmeier and Griffin, 2003). (1)
A new literature has emerged that has the potential to explain some or much of the asymmetric passthrough found in certain gasoline markets by the earlier studies. The literature examines the rapid and asymmetric retail price cycles recently found in many retail gasoline markets around the world including in Canada (Noel, 2007a, 2007b; Eckert, 2002, 2003), the United States (Castanias and Johnson, 1993; Lewis, 2007), Australia (Wang, 2005), and in several European countries. Although newly discovered with the availability of high-frequency price data, asymmetric cycles have existed in some markets for decades. As an example of the cycles, in Figure 1, I plot the retail prices for two gasoline stations (one major firm, one independent) and the wholesale ("rack") price over four months in 2001 in the market of Toronto, Canada, which experiences cycles. The 12 hourly data show the cycle clearly.
[FIGURE 1 OMITTED]
These authors argue the cycles are the theoretical Edgeworth Cycles of Maskin and Tirole (1988), and I defer to them for further evidence. In an Edgeworth Cycle, firms selling homogeneous goods repeatedly undercut one another by small amounts to steal market share. When margins get too low, one firm "relents" by raising its price significantly higher. Other firms follow quickly by relenting themselves, and then, from the new high price, another round of undercutting begins. The asymmetric price process--large increases and small decreases--repeats over and over. This is true even in the absence of any cost shocks at all.
There is an interaction between the asymmetry inherent in an Edgeworth Cycle and cost shocks, however, that can result in asymmetric passthrough. For example, a positive rack shock can trigger a new relenting phase and, if it does, a large price increase. In contrast, a negative shock will not cause a large price decrease, as it only allows more room for undercutting and undercuts tend to be small. (2) The result is that cost increases can be passed through to retail prices more quickly than decreases, relative to the pre-shock prices. Eckert (2002) observes this in his study of retail price cycles in Windsor, Canada and argues that the cycles are a source for at least some portion of the asymmetric passthrough he finds there.
It is an excellent insight, but such a finding still cannot tell us whether Edgeworth Cycles are the sole cause for the asymmetric passthrough found or whether they are just one of many possible contributing factors. This is important to know and begs the question that is the focus of this article: when asymmetric passthrough is found in a market with Edgeworth Cycles, how much of it is attributable directly to the Edgeworth Cycles themselves, and how much of it remains to be explained with other sources? Can the cycles explain 100%? With the exception of Eckert (2002), prior studies have not considered Edgeworth Cycles as a potential cause and their influence is not well understood. (3) Clearly, it is important to separate the known effects of Edgeworth Cycles from potential other causes of asymmetric passthrough that would warrant further study. In this article, I do this. I show asymmetric passthrough exists for the city of Toronto, where strong retail cycles exist, and then decompose it into two components: that caused solely by cycles--the "Edgeworth-explained" component--and that which remains--the "residual" component. The source of the latter is unknown and may include interactions between those unknown causes and the Edgeworth Cycles themselves. For Toronto, I find that the Edgeworth-explained component is large and plays a significant role in generating asymmetric passthrough.
2. Components of overall asymmetric passthrough
* "Overall" asymmetric passthrough (from all sources) is intuitively found by comparing post-shock prices to pre-shock prices--first after a rack increase, then after a decrease, and then differencing the two series. This is the comparison implicit in the usual techniques used in the literature (e.g., vector autoaggression [VAR] models). The use of pre-shock prices as the reference point is standard and based on the idea that prices would not change but for the shock.
What is unique about markets with cycles, however, is that prices do keep changing and in an asymmetric way, even absent a shock. And we can predict how. It turns out that this predicted "no-shock" price path holds the key to decomposing overall asymmetric passthrough into its Edgeworth-explained and residual components.
The reason is that each individual part of a theoretical Edgeworth Cycle taken separately responds symmetrically to...
|
