ILLUSTRATION BY BRYAN LEISTER
By Stijn Van Nieuwerburgh and Laura Veldkamp
ersonal business magazines in the US frequently hector their readers to shift their investments to the Next Big Thing. One week it’s emerging markets, the next, Japan’s making a comeback, and a week later, don’t forget about Europe. No doubt, the magazines’ counterparts overseas preach a similar story in reverse. Thing is, despite all the information available in this Internet Age, most investors tend to weight their portfolios toward their local equities. Does this make any sense? We set out to understand this phenomenon, and what we found revealed that gaining an information edge was the critical factor in deciding where to invest, and where better to gain that edge than your own backyard?
The pattern persists even if you’re a relatively sophisticated, worldly investor or portfolio manager with global access to information about international markets. Your portfolio is probably weighted toward domestic equities. Economists call this phenomenon the “home bias.” Home bias is defined as a long position in the home asset that exceeds what is prescribed by the standard world market portfolio. We asked why global information access doesn’t eliminate this asymmetry.
Returns on national equity portfolios suggest there are substantial benefits from international diversification, yet individuals and institutions in most countries hold modest amounts of foreign equity. While restrictions on international capital flows may have been a viable explanation for the home bias 30 years ago, the free flow of investments globally makes that explanation obsolete. An alternative hypothesis contends that home investors have superior access to information about domestic firms or economic conditions. This information-based theory of the home bias assumes that home investors cannot learn about foreign firms. It replaces the old assumption of capital immobility by the similar assumption of information immobility. Our critique of this information-based theory of home bias was that domestic investors are free to learn about foreign firms. It seemed logical that such cross-border information flows could potentially undermine the home bias – that when investors could choose which information to collect, initial information advantages could disappear.
Most existing models of asymmetric information in financial markets are silent on information choice. A small but growing literature studies how much information investors acquire about one risky asset or models a representative agent who, by definition, cannot have asymmetric information. However, instead of asking how much investors learn, we asked which assets they learn about. To answer this question required a model with three features: information choice, multiple risky assets to learn about, and heterogeneous agents so that information asymmetry is possible.
Take your Pick
We developed a two-country, rational-expectations, general-equilibrium model where investors first chose what home or foreign information to acquire, and then chose what assets to hold. The prior information home investors had about each home asset’s payoff was slightly more precise than the prior information foreigners had. The reverse was true for foreign assets. This prior information advantage may have reflected what was incidentally observed from the local environment. Home investors chose whether to acquire additional information about either home or foreign asset payoffs. The interaction of the information decision and the portfolio decision caused home investors to acquire information that magnifies their comparative advantage in home assets.
“The standard asset pricing and portfolio choice models typically assume symmetric information sets across agents. Our study shows that these models are subject to an important criticism: Investors have an incentive to deviate by learning information that others do not know.”
In our model, if home investors chose to undo their information asymmetry by learning about foreign assets, they sacrificed excess returns. When information indicated that the foreign assets’ payoffs would be high, both home and foreign investors knew about it, demanded more of the foreign assets, and bid up their price. If home investors instead learned more about home assets than the average investor did, then when they observed information indicating high home asset payoffs, home asset prices would not fully reflect this information. Prices reflected only as much as the average investor knew. The difference between prices and expected payoffs generated home investors’ expected excess return.
When choosing what to learn, investors made their information set as different as possible from the average investor’s. To achieve the maximum difference, home investors began with home assets, which they started out knowing relatively more about, and specialized in learning even more about them. One of the main results was that information immobility persisted not because investors could not learn what locals know, nor because it was expensive, but because they chose not to. Specializing in what they already knew was a more profitable strategy.
The model’s key mechanism was the interaction between the information choice and the investment choice. To illustrate its importance, we shut down this interaction by forcing investors to take their portfolios as given, when they chose what to learn. These investors minimized investment risk by learning about assets that they were most uncertain about. Our inquiry related to the theory that if there is sufficient capacity, learning can undo the initial information advantage, and therefore also home bias. Thus, this model embodied the logic that the asymmetric information criticisms are founded on.
We next showed that when investors have rational expectations about their future optimal portfolio choices, this logic was reversed. While acquiring information that others did not know increased expected portfolio returns, that alone did not imply that home investors took a long position in home assets, only that they took a large position. But home bias is a comparatively long position in the home asset, not necessarily a large one, and it arises because home assets offer risk-adjusted expected excess returns to informed home investors.
Information about the home asset reduces the risk or uncertainty that the asset poses without reducing its return, hence the high risk-adjusted returns. How does information reduce uncertainty? An asset’s payoff may be very volatile, high one period and low the next. But if an investor has information that tells him when the payoff is high and when it is low, the asset payoff is not very uncertain to that investor. Information drives a wedge between the conditional standard deviation (uncertainty or risk) and the unconditional standard deviation (volatility) of asset payoffs. While foreign assets offer lower risk-adjusted returns to home investors, they are still held for diversification purposes. The optimal portfolio tilts the world market portfolio toward home assets.
Considering how learning affects portfolio risk offers an alternative way of understanding why investors with an initial information advantage in home assets choose to learn more about home assets. Because of the excess risk-adjusted returns, a home investor with a small information advantage initially expects to hold slightly more home assets than a foreign investor would. This small initial difference is amplified because information has increasing returns in the value of the asset it pertains to: As the investor decides to hold more of the asset, it becomes more valuable to learn about. So, the investor chooses to learn more and hold more of the asset, until all his capacity to learn is exhausted on his home asset.
Knowledge is Power
A numerical example showed that learning can magnify the home bias considerably. When all home investors received a small initial advantage in all home assets, the home bias ranged between 5 percent and 46 percent, depending on the magnitude of investors’ learning capacity. When each home investor received an initial information advantage that was concentrated in one local asset, the home bias was amplified. It rose as high as the 76 percent home bias in US portfolio data, for a level of capacity that is consistent with observed excess returns on local asset portfolios.
A variety of evidence supported the model’s predictions, connecting the theory to facts about analyst forecasts, portfolio patterns, excess portfolio returns, cross-sectional asset prices, as well as evidence thought to be incompatible with an information-based home bias explanation. In particular, the theory offered a unified explanation of home bias and local bias. While we cannot claim for any one of these facts that no other theory could possibly explain the same relationship, taken together, they constituted a large body of evidence that is consistent with one fully adequate theory.
Our model studied a common criticism of information-based models of the home bias: If home investors have less information about foreign stocks, why don’t they choose to acquire foreign information, reduce their uncertainty about foreign payoffs, and undo their portfolio bias? The answer to this question required a model where investors chose which risky asset payoffs to learn about. We showed that investors who did not account for the effect of learning on portfolio choice chose to undo their initial advantages. But, investors with rational expectations reinforced informational asymmetries.
e can conclude from this model that investors learn more about risks they have an advantage in because they want their information to be very different from what others know. Thus our main message is that information asymmetry assumptions are defensible, but not for the reason originally thought. We do not need cross-border information frictions. With sufficient capacity to learn, small initial information advantages can lead to a home bias of the magnitude observed in the data.
A problem that many asymmetric information theories face is that unobservable information makes them difficult to evaluate empirically. While information cannot be observed, it can be predicted. A separate contribution of our study is to connect the observed features of assets to predictions about investors’ information sets. This connection provides a new way to bring information-based theories to the data.
An important assumption in our model is that every investor must process his own information. But paying one portfolio manager to learn for many investors is efficient. How might such a setting regenerate a home bias? Because monitoring information collection is difficult, portfolio managers have an incentive to lie about how much research they do. Investors may want to occasionally assess the portfolio manager’s performance. Having a manager from the same region, with similar initial information, is advantageous because evaluating his investment choices is easier. Portfolio managers who want to maximize their risk-adjusted return and have the same initial information advantage as their clients would form a home-biased portfolio, for the same reason that an individual investor would. Thus, even with delegated portfolio management, home bias should emerge. Future work could use the framework in this model to build an equilibrium model of delegated portfolio management that connects a manager’s ability to acquire or process information to the size, style, and fees of his fund.
The broader message is that investors choose to have different information sets. The standard asset pricing and portfolio choice models typically assume symmetric information sets across agents. Our study shows that these models are subject to an important criticism: Investors have an incentive to deviate by learning information that others do not know.
STIJN VAN NIEUWERBURGH is assistant professor of finance and Charles Schaefer Family Fellow and LAURA VELDKAMP is associate professor of economics at NYU Stern.