Petrol – Further econmetric analysis undertaken by ACCC
Appendix S contained in the ACCC’s Inquiry Report into the price of unleaded petrol, detailed the results of the econometric analysis undertaken by the ACCC to examine the price effects of the introduction of Fuelwatch into Western Australia from 2 January 2001. In addition to the results included in the ACCC’s Report the ACCC undertook further econometric analysis to examine the influence of other factors on the price effects that were analysed in the Report. Further analysis assessed the price effects of purchasing petrol at different times of the week.
Econometric approach - Methodology and data
The initial tests examined the difference in price that occurred in Perth relative to the eastern capitals before and after the introduction of Fuelwatch.
The data series was constructed using pricing information supplied by Informed Sources and Platts. The series tested was a measure of price margin that removes factors from the retail price that are beyond the scope of Fuelwatch to affect, such as net taxes, fuel quality premiums and ex-refinery petrol prices:
Price margin = (Retail price – lagged Mogas95 price – net taxes – fuel quality premium) Perth - (Retail price – lagged Mogas95 price – net taxes – fuel quality premium) Average of eastern capitals.
The price margin is calculated for Perth relative to the average of the eastern capitals. This allows many items that could be assumed to be relatively common to Australia as a whole to be implicitly ignored.
The price margin explicitly allows for lagged Mogas prices, net taxes and changes in indicative fuel standard premiums. The Mogas price is lagged by one week to reflect the typical lag seen between the effect of changes in Mogas on domestic retail petrol prices.
The data series used extends from 1 August 1998 to 8 June 2007. We do not go back before 1 August 1998 due to the major deregulation of petrol prices at that time. Data after 8 June 2007 was excluded as it was the time of the announcement of this inquiry. There are three data series tested using this price margin:
Weekly average: The primary data series uses weekly averages of prices to remove some of the effects of the price cycle.
Monthly average: This series uses monthly averages of prices rather than weekly averages. This further reduces the effect of the price cycle. It also helps to ensure that any apparent move from typically weekly to typically fortnightly cycles in Perth does not unduly affect the results.
Weekly minimum: This series uses a measure of the low point of the week’s prices. The prices used are the average of the two lowest daily retail prices for each week. This measure represents the options of the most price conscious consumers.
Unit root tests
Testing for a unit root is used to establish which series are stationary and suitable for dummy variable tests for a structural break. No deterministic trend is indicated or assumed. Dummy variable tests for a structural break are performed subsequently. Results using an augmented Dickey Fuller test for a unit root were included in the ACCC’s original report.
Testing for a structural break
A dummy variable approach was used to test for a structural break in the time series data. The tests assume the date of structural break is 2 January 2001, when Fuelwatch began.
The simplest possible test was included in the ACCC report, with an intercept, a break dummy equal to 0 before the break and equal to 1 after the break and no time trend. The intercept coefficient represents the average before the break date. The break dummy coefficient represents the change in average after the break date. The results as included in the ACCC’s original report are presented in the table below.
Structural break test for relative price margin, cpl, August 1998 to June 2007
Series
Average (August 1998 to December 2000)
Change in average (January 2001 to June 2007)
Weekly average
0.83 (0.002)
-1.92 (0.000)
Monthly average
0.88 (0.001)
-1.86 (0.000)
Weekly minimum
0.30 (0.277)
-0.90 (0.003)
Further econometric tests undertaken
The ACCC also undertook further econometric analysis to take account of price effects on different days of the week. In particular the ACCC wanted to check that consumer who benefit from the price cycle by buying on the lowest price day each week would not be harmed by the introduction of the Fuelwatch system. The ACCC looked at price changes for the lowest price day of the week, the highest price day of the week and for the remaining five days of the week. The ACCC’s analysis showed that:
prices decreased an average of 3.5cpl for the highest price day of the week and
prices decreased an average of 0.7 cpl for the lowest price day of the week and
prices decreased an average of 1.8 cpl for the remaining middle five days of the week.
This analysis confirmed that the overall price reductions indicated by the inquiry analysis were not isolated to certain times of the week.
Testing for significant events that may have affected relative prices
The previous tests were aimed at examining the prices before and after the introduction of Fuelwatch. The ACCC decided to undertake further tests to explain whether further factors could explain the difference. This was done by using an econometric technique that aims to show when the most significant events occurred and their impact. This analysis is known as endogenous selection of structural break points. These tests were done on the same series as in the Inquiry report. These tests were designed to identify the timing of a single significant event or alternatively the timing of two significant events in the average of the price margin measure.
The main event identified by all of these tests was the decrease in price margin around from around the time of the establishment of the WA Select Committee on Pricing of Petroleum Products and the establishment of Fuelwatch.
The analysis of the structural breaks indicated that the entry of Coles into Perth was an event that may have had a price impact. However, its impact was small compared to the break around the time of the introduction of Fuel Watch. Further the entry of Coles into the eastern capitals could have induced a similar break favouring the Eastern capitals. Hurricane Katrina was also indicated as a significant event. The results of this analysis are in the table below.
Structural breaks in the pricing measure, cpl, August 1998 to June 2007
Price margin series
Single structural break Timing & price margin change
Two structural breaks Timing & price margin change
Weekly average
May 2000 -1.1cpl
May 2000 -1.1cpl February 2004 -0.4cpl
Monthly average
March 2000 -1.5cpl
March 2000 -1.4cpl February 2004 -0.6cpl
Weekly minimum
March 2000 -0.8cpl
March 2000 -1.0cpl September 2005 +0.4cpl
Source: ACCC estimates.
Conclusion
The purpose of this econometric analysis has been to satisfy the ACCC that the introduction of a Fuelwatch scheme nationally would not, based on the experience in Western Australia, lead to consumers paying higher prices for petrol.
From the econometric analysis, on a conservative basis, the ACCC can say that there is no evidence that the introduction of Fuelwatch in Western Australia led to any increase in prices and it appears to have resulted in a small price decrease overall.
