Estimating and Comparing Market Efficiency Frontiers
by Yali Mu
Date of Examination:2021-09-10
Date of issue:2021-09-21
Advisor:Prof. Dr. Stephan von Cramon-Taubadel
Referee:Prof. Dr. Stephan von Cramon-Taubadel
Referee:Prof. Dr. Brümmer Bernard
Referee:Prof. Dr. Zhong Funing
Files in this item
Name:Dissertation-Yali Mu-SUB.pdf
Size:2.84Mb
Format:PDF
Description:Main article
Abstract
English
Previous studies employing error correction models (ECMs) that include threshold effects in long-run equilibrium relationships (threshold cointegration) as well as in error correction (threshold autoregression; threshold ECMs; non-parametric ECMs; Markov-switching ECMs; and other forms of ECM with time-varying error correction such as state space models) investigating market integration and efficiency are based on a static concept – given a set of demand and supply conditions on two (or more) markets in space, and trade costs between them, they are either in equilibrium and therefore efficient, or not. Market efficiency, however, has two additional dynamic dimensions. First, trade costs vary over time and space. Second, since moving agricultural products between markets takes time, trade and prices cannot adjust instantaneously to shocks that disturb equilibrium. The main innovation of this thesis is the proposed procedure for benchmarking dynamic market efficiency by estimating stochastic frontier accounting for sampling error in estimated measures of price transmission, which also identifies the strongest and most rapid price transmission that can be attained in a given setting. Another novelty of this study is the comparation of empirical benchmark of market efficiency in China to the empirical benchmark of market efficiency in the EU. Rather than using standard regression techniques to explain variation in estimated measures of the strength and speed of price transmission, stochastic frontiers are estimated. Estimating a stochastic frontier could establish a within-sample benchmark for market efficiency against which performance can be measured. Stochastic frontier methods provide a straightforward and intuitively appealing means of accounting for measurement errors. The proposed method is first illustrated by using monthly data on pork prices on 30 provincial markets in China from 2000 to 2017. First, standard VECMs are used to estimate elasticities of price transmission as well as adjustment parameters that measure the speed of transmission between individual markets. And then benchmarks are estimated using stochastic frontier techniques and covariates such as the geographic distance between the markets. The estimated frontier lies just slightly below the theoretical frontier. And the estimated frontier differs considerably from the OLS estimate of magnitude of market efficiency on distance that has been used in past studies. Furthermore, there are significant province effects in inefficiency term. The inefficiency term is higher for market pairs that include border provinces. Factors such as ethnicity and dietary customs and relative remoteness might contribute to these results. Then, the proposed procedure is applied to hog prices on the 30 provincial markets in China and 23 member-state markets in the EU from 2004 to 2017. The distribution of the inefficiency scores for the magnitude of market efficiency for China shows that for many market pairs the long-run elasticity of price transmission is within shorter distance of the frontier than the EU. The average inefficiency is lower in China than in the EU. Market pairs involving provinces located in the Central China are characterized by the highest magnitude of market efficiency, and that the magnitude of market efficiency tends to be lower for provinces located farther to the South and West. In comparison, market pairs involving members located in the central EU are characterized by the highest magnitude of market efficiency. To sum up, a new method is proposed for benchmarking dynamic market efficiency using cointegration analysis coupled with frontier estimation methods. The use of frontier methods makes it possible to estimate within-sample benchmarks for the magnitude and the speed of restoring market efficiency at interregional level and at the international level as well. It also provides a convenient way of accounting for sampling error in estimated measures of price transmission. The influence of distance is considerably higher at the international level (the EU) than at the interprovincial level (China). International hog market integration in the EU is relatively low and heterogeneous and trade costs are relatively high compared to China, this might reflect more stringent animal welfare-based restrictions on transporting live hogs. To foster market functioning, investments in transportation infrastructure, but also the elimination of informal payments (e.g. administrative costs under animal welfare regulations during transport), are fundamental for reducing pork trade costs in both China and the EU.
Keywords: Spatial price transmission, market efficiency, VECM, stochastic frontier benchmark