Measuring market integration in the presence of transaction costs—a threshold vector error correction approach

Abstract

Applied econometric analyses of market integration based on price data alone have been criticised because they neglect the role of transaction costs. To meet this objection, threshold vector error correction models are used. Threshold models can account for the effects of transaction costs in price transmission without directly relying upon information about these costs, which are often unavailable. It is argued that threshold models that are based on two thresholds provide results that are economically more intuitive than those obtained from one-threshold models. However, to this point no adequate econometric test is available for threshold significance in a two-threshold model; such tests are only available for the one-threshold model. In this paper a restricted two-threshold model is developed in which the significance of the thresholds can be tested. This model is therefore amenable to economic interpretation and statistical inference. The model is used to estimate market integration in the European pig market.

Introduction

Econometric analyses of market integration are becoming increasingly popular. Farmers as well as agricultural economists are interested in the extent to which price shocks are transmitted between markets. For example, some German pig producers use the commodity exchange AEX in Amsterdam to hedge price risk, assuming that German and Dutch markets are highly integrated. Economists often use market integration, defined as the degree of price transmission between either vertically or spatially related markets, as a proxy for market efficiency. Adequate empirical methods are necessary to answer questions about market integration. In recent years these methods have improved considerably. Nevertheless, as pointed out by Meyer and von Cramon-Taubadel (2002), there is still no unified approach to evaluate market integration. Recent studies that rely on price data alone have been criticised by Barrett for their neglect of transaction cost (Barrett, 2001, Barrett and Li, 2002). Due to the lack of information on transaction costs, an approach of measuring market integration based on price data alone, but also considering the effects of transaction costs is proposed. Unlike most studies that analyse price adjustment between vertical related stages of the marketing chain, this analysis focuses on spatial market integration. Hence, the analysis is related to the literature on the ‘Law of one price’ (Lo and Zivot, 2001, Obstfeld and Taylor, 1997). To quantify price adjustment, the vector error correction model (VECM) framework is used. As an extension and innovation testable threshold effects are also taken into account. To illustrate this approach market integration between pig markets in Germany and The Netherlands is analysed.

In the following section of the paper, the empirical method is described. In Section 3, the results of the empirical application to German/Dutch pork markets are presented. In Section 4, conclusions are drawn and suggestions for further research are given.