Information asymmetry, market segmentation, and cross-listing: Implications for event study methodology

Abstract

This paper connects three subjects related to international financial markets-(i) information asymmetry, (ii) market segmentation, and (iii) cross-listings-and highlights their implication for event study methodology. When firms list equities on more than one exchange, and the exchanges are characterized by different information sets, a problem arises as to which exchange(s) to include in the event study sample. If market segmentation impedes the arbitrage of these multiple responses, then the use of a single listing (for a firm that is cross-listed) can yield abnormal return estimates that are biased. In such circumstances, using returns from all the markets in which a firm's securities are listed not only increases the sample size (often an important consideration when undertaking event studies in emerging markets), but also enables full-information abnormal return estimates to be obtained. What is required is a method that extracts the independent information from each listing while counting the common information only once. In this paper, we develop an estimation procedure that achieves these twin objectives. We then apply our approach to an event study of Chinese OMAs and compare results from alternative samples and estimators. We demonstrate that including return data from cross-listings of the same firm can result in substantially different conclusions.

Chinese overseas M&A performance and the Go Global policy1

“…Gu and Reed (2012) develop a procedure for estimating the ρ ij,ik ‘s when they are unknown. The extension to multiple time periods and the calculation of CARs is straightforward.…”

“…We employ event-study methodology to evaluate the effect of Chinese OMAs on shareholder wealth. A noteworthy feature of our analysis is that we apply a new, GLS procedure for event studies with multi-listed observations (Gu and Reed, 2012).…”

“…In the context of our problem, the appropriate procedure estimates β using the standardized model [], and then applies the corresponding sandwich estimator (incorporating cross‐sectional correlation) for the standard error and associated hypothesis testing (Gu and Reed, 2012). These are respectively given by:…”

“…This raises econometric issues. To address these, we employ a new, feasible, generalized least squares (GLS) estimator developed by Gu and Reed (2012).…”

“…The aggregation of share price data across multiple markets, and the listing of firms in multiple exchanges, raise econometric issues for the standard event‐study methodology. To address these, we use a new, feasible generalized least squares (GLS) procedure developed by Gu and Reed (2012). On the basis of an analysis using both aggregated and disaggregated samples, and of daily and cumulative abnormal returns, we find consistent evidence that (i) Chinese OMAs have not lowered the wealth of shareholders of Chinese acquiring firms, and (ii) shareholders of Chinese acquiring firms have not fared worse under under China's ‘Go Global’ policy of encouraging outward investment by Chinese firms.…”