A Structural Analysis of Price Discovery Measures

Abstract

We analyze the structural determinants of two widely used measures of price discovery between multiple markets that trade closely-related securities. Using a structural cointegration model, we show that both the information share (IS) and component share (CS) measures account for the relative avoidance of noise trading and liquidity shocks, but that only the IS can provide information on the relative informativeness of individual markets. In particular, the IS of one market is higher if it incorporates more new information and/or impounds less liquidity shocks. Use of the CS in conjunction with the IS can help sort out the confounding effects of the two types of shocks. Furthermore, we find that the IS only accounts for the immediate (one-period) responses of market prices to the news innovation which implies that the IS estimates based on high sampling frequencies may be distorted by transitory frictions and may miss important price discovery dynamics. * We thank

Price discovery of subordinated credit spreads for Japanese mega-banks: Evidence from bond and credit default swap markets

“…It should be noted here, however, that recently, Yan and Zivot (2007) rigorously analyze the determinants of these two price discovery measures under some structural models in which both permanent (fundamental) and transitory shocks are identified. 22 Their main findings can be summarized as follows.…”

“…Second, the GG measure is less appropriate in that it actually contains only contemporaneous responses to transitory shocks. According to the arguments in Yan and Zivot (2007), the GG measure for market 1 is higher when the contemporaneous response of market 2 to transitory shocks is higher than market 1. Third, even without a correlation between residuals from the VECM, the Hasbrouck measure cannot avoid ambiguity in interpreting the result.…”

“…A high value of the Hasbrouck measure for market 1 can be caused by a high value for the contemporaneous response of market 1 to fundamental shocks, a high value for the contemporaneous response of market 2 to transitory shocks, or both. In this regard, Yan and Zivot (2007) argue that combining information from the two price discovery measures enables us to remove some ambiguity associated with the Hasbrouck measure. More specifically, a high Hasbrouck measure together with a low GG measure for market 1 is likely to indicate that market 1 has a more significant response to fundamental shocks; that is, it has a more dominant role in price discovery.…”

“…For example, the Journal of Financial Markets released a special issue about this issue in 2002 (Issue 3). Among others, Lehmann (2002) expresses his concern about the interpretation of these measures because both are based on the residuals from a reduced-form VECM, and it is Yan and Zivot (2007) who rigorously address Lehmann's concern from a structural point of view. 23 We set the starting point at the first business day of FY 2004.…”

Measuring price discovery: The variance ratio, the R2, and the weighted price contribution

“…Notice that our concept of price discovery is different from the information share measure developed in Hasbrouck (1995Hasbrouck ( , 2002 employed by Theissen (2002), Hasbrouck (2003) and Yan and Zivot (2010), among others. The information share recognizes that transaction prices chase an unobservable value (often denoted the efficient price), but deviate from it due to microstructure noise and/or asymmetric information.…”