Price Transmission in the Austrian
Gasoline Market
Daniela Rroshi
Introduction
The rockets and feather phenomenon or asymmetric pricing is a stylized fact in
many markets (banking, agriculture, gasoline).
„In two out of three markets output prices rise faster than they fall „ (Peltzman,
RJE 2000)
Theory:
Market power :(Borenstein et al. 1997)
Consumer search: (Tappata, 2009; Yang and Ye, 2008 ; Cabral and Fishman 2012
Lewis(2011)
Investigate this question for the Austrian gasoline market
Introduction
 Dataset
Daily data on 282 gasoline stations : January 1st 2003 –
December 5th 2004
Input and output price measures: Brent crude oil index and the
retail diesel price.
Preparatory Analysis
1. Testing for Unit Roots
Standard unit root tests for price and cost:
1. Augmented Dickey-Fuller Test
∆𝒑𝒕 = 𝜸𝒑𝒕−𝟏 +𝜺𝒕
∆𝒄𝒕 = 𝜸𝒄𝒕−𝟏 +𝜺𝒕
If 𝛾 = 0 the series contains a unit root or is I(1)
Cointegrated variables: variables that move together in the long-run
(Granger 1986; Engle and Granger 1987)
𝑝𝑡𝑖 =𝛼0 + 𝛼1 𝑐𝑡𝑖 + 𝐸𝐶𝑇𝑡,𝑖
Cointegrated variables can be characterized by an Error Correction
Model
Preparatory Analysis
 2. Testing for Cointegration
Engle and Granger Approach/Phillips-Ouliaris residual based test
𝑝𝑡𝑖 =𝛼0 + 𝛼1 𝑐𝑡𝑖 + 𝜀𝑖,𝑡
Run a Phillips – Ouliaris Test on the residuals:
∆𝜀𝑡 = 𝛾 ∆𝜀𝑡−1 +∈𝑡
𝐻0 : 𝛾 = 0
Econometric Models
 Symmetric Error Correction model:
𝑅
∆𝑝𝑖,𝑡
=𝜏𝑖 + 𝛼𝑖 𝐸𝐶𝑇𝑖,𝑡−1 +
𝑤ℎ𝑒𝑟𝑒 𝐸𝐶𝑇𝑖,𝑡−1 =∆𝜀𝑡−1
𝑛
𝑅
𝑗=1 𝛿1,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗 +
𝑛
𝐶𝑂
𝑗=1 𝛿2,𝑖,𝑗 ∆𝑐𝑜𝑠𝑡𝑖,𝑡−𝑗
+ 𝜀𝑖,𝑡
Asymmetric Error correction model (Granger and Lee, 1989):
𝑅
∆𝑝𝑖,𝑡
=𝜏𝑖 + λ𝑖+ 𝐸𝐶𝑇𝑖,𝑡−1 +
𝑅
∆𝑝𝑖,𝑡
=𝜏𝑖 +
𝑛
𝑛
𝐶𝑂
𝑅
if 𝐸𝐶𝑇𝑖,𝑡−1 > 0
𝑗=1 𝛿1,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗 + 𝑗=1 𝛿2,𝑖,𝑗 ∆𝐶𝑖,𝑡−𝑗 + 𝜀𝑖,𝑡
𝑛
𝑛
𝐶𝑂
𝑅
λ−
𝑖 𝐸𝐶𝑇𝑖,𝑡−1 +
𝑗=1 𝛿1,𝑖,𝑗 ∆𝑝𝑖,𝑡−𝑗 + 𝑗=1 𝛿2,𝑖,𝑗 ∆𝐶𝑖,𝑡−𝑗 + 𝜀𝑖,𝑡 if 𝐸𝐶𝑇𝑖,𝑡−1 <0
 Where ECT is the lagged residual from regressing price on cost