The determinants of international investment and attention allocation: Using internet search query data

a r t i c l e i n f o This paper explores the joint determination of home bias and attention allocation. We overcome the typical challenge associated with evaluating attention allocation theories by using a new internet search query dataset to measure how much information investors decide to process. Employing an instrumental variables approach, we find empirical evidence of a two-way causality between home bias and attention. Our estimates suggest that if all countries were to receive the same level of attention as the U.S., then the average home bias by U.S. investors would fall from 85.2% to 57.3%. The existence of a domestic bias in international equity holdings was first documented by French and Poterba (1991) and Tesar and Werner (1995). One popular justification focuses on the role of informational asymmetries between domestic and foreign investors. Although over the past decade innovations in telecommunication and information technologies have greatly increased investors' access to information around the globe, significant levels of home bias still remain. To explain this persistent bias, information-based models commonly have to assume implausibly large exogenous information asymmetries. Recent work by Van Nieuwerburgh and Veldkamp (2009), however, generated these large information asymmetries by endogenizing the investors' information structure. In their model, the allocation of investors' limited attention leads to information heterogeneity endogenously and thereby a sizeable bias in portfolios. Our paper provides empirical support for attention-based explanations of the home equity bias. The idea that " most people everyday encounter, or could very easily encounter, much more information that is in principle relevant to their economic behavior than they actually respond to " (Sims, 2006)–known as inattentiveness–has recently been incorporated into finance and economic models. 1 Although attention allocation models are very appealing from a theoretical perspective, empirical evaluation remains a challenge. The main difficulty relies on measuring the pieces of information each agent processes. Our paper overcomes such a challenge by using a new dataset from America Online (henceforth AOL) that includes over 21 million web searches by 657,426 customers. We then combine data on U.S. holdings of foreign securities to analyze the joint determination of home bias and attention allocation by U.S. investors. Our contributions are threefold. First, we derive a simple model based on Van Nieuwerburgh and Veldkamp (2009) to study the channels through which attention allocation affects portfolio choice and vice versa. Second, we provide a new measure of the attention …