Nonlinear and asymmetric pricing behaviour in the Spanish gasoline market

Abstract

Over the last decades a transition from a state-own monopoly to a private business took place in the Spanish fuel sector. To figure out whether downstream prices react differently to upstream price increases than to price decreases, alternative dynamic nonlinear and asymmetric error correction models are applied to weekly price data. This paper analyse the existence of price asymmetries in the fuel market in Spain during the 2011-2016 period. In comparison with traditional asymmetric price theory literature, this paper introduces a new double threshold error correction (ECM) model (DT-ECM) and new double logistic ECM models and compares them with more common linear ECM, time varying parameter models (TV-ECM), threshold autoregressive models (T-ECM), smooth transition autoregressive (STAR) models and nonlinear error correction (Logistic-ECM) and double threshold Logistic (DT-Logistic ECM). The nonlinear and asymmetric results found show that sophisticated bivariate long-run asymmetries are present in the prices of the fuel sector and that those price reactions depend on whether the oil price increases or decreases, on the stage of the production, the distribution chain as well as on the period considered.JEL classification: B23, C24, C52, D43, L13, L71.

Nonlinear Interactions and Volatility Spillovers between Stock and Foreign Exchange Markets: The STVEC-STGARCH-DCC Approach

“…unanticipated price increase). In addition to volatility asymmetry, there might exist smooth transition attributes of assets volatility under different economic statuses in the real world (Gonzalez-Rivera, 1996;Lundbergh and Teräsvirta, 1998;Nam, 2002;Sollis, 2009;Yaya and Shittu, 2014;Liu and Chen, 2016;Escribano and Torrado, 2018). Therefore, it is greatly appropriate to connect the nonlinear conditional mean equation to the conditional variance equation with smooth transition mechanism to explain the nonlinear price movements and volatility spillovers of financial assets (e.g., an extension of GJR-GARCH models).…”

European gasoline markets: price transmission asymmetries in mean and variance

“…More general price nonlinearities in mean are considered by the use of the nonlinear error correction models of Escribano (1985Escribano ( , 1986Escribano ( , 1987Escribano ( and 2004, Escribano and Granger(1998) and Escribano and Pfann (2998), by jointly relating the nonlinearity in the ECM to the increases or decrease in the international oil price (rockets and feathers hypothesis). In particular Escribano and Torrado (2018) consider a nonlinear three-dimensional logistic-ECM models (Double threshold-ECM). The main goal of this paper is to study the international reactions in retail European prices to changes in the prices of oil with a joint estimation of the asymmetry parameters of the conditional mean, using Double Threshold-ECM models, and allowing also for asymmetries in their conditional variance.…”

“…Notice that once we allow also for asymmetries in the variance, the nonlinearities in the mean changes. See for example the different nonlinearities found in oil prices for Spain in Escribano and Torrado (2018). The F-statistics for the potential asymmetries derived from the A-ECM.…”

“…Instead of allowing for single type of asymmetries in the mean with the traditional A-ECM, we consider the Double-Threshold Error Correction Model (DT-ECM) for the mean, following Escribano and Torrado (2018), and we extend it here to allow for asymmetries also in the conditional variance. Under this method, asymmetries in the mean are arising from positive or negative input (oil) price changes are considered, together with whether prices are above or below their long run-equilibrium levels.…”

“…Secondly, the econometric methodology used to assess general price asymmetries consists on studying asymmetries from two main sources. On one hand, asymmetries in the mean through the double threshold error correction models (DT-ECM), introduced by Escribano and Torrado (2018) and on the other, with the asymmetric price patterns in the conditional variance by employing GARCH, EGARCH, GJR-GARCH and Log-GARCH models with dummy variables.…”