Abstract

Model selection tests and criteria are employed to identify empirically the most appropriate price variable for the purpose
of studying the price-volume relation. Five different price variables are considered as explanatory variables in model
selection tests, which are carried out on a bilateral basis using data on stock prices and trading volume in eleven markets/
indices. The results show that the most appropriate price variable varies from one market to another although two price
variables appear to be dominant: the extreme value variance and the absolute price change. The results do not provide
much support for the notion of asymmetry in the price-volume relation. It is suggested that it may be useful to use nonnested
model selection tests and criteria to identify the most appropriate price variable before testing for causality, which
is the principal tool used for examining the price- volume relation.

This work is licensed under a Creative Commons Attribution 4.0 License.