Invest in what you know?

Barcelona GSE Focus

When purchasing mutual funds, do you go for the lesser-known funds or what may be familiar to you? This familiarity bias among investors may help shape the competition among funds, the persistence of fund performance, and the fees charged, according to Ariadna Dumitrescu’s and Javier Gil-Bazo’s Barcelona GSE Working Paper (No. 815)“Familiarity and Competition: The Case of Mutual Funds.”

Familiarity Bias?

This is basically the tendency for non-savvy investors to invest in financial products that are well known, culturally familiar, or even the stock of the company they work for. Investors in the US who may not be familiar with, say, the Japanese economy or business culture may be less interested in investing in Japanese mutual funds or other investment products. That is, their distance from these markets may induce either an underestimation in the returns or overestimation in the risk of such a product vis-à-vis a comparable fund from the US. This bias towards choosing “familiar” financial products decreases diversification, and choosing funds based on characteristics that tend toward this bias (like funds with headquarters close to home) have been shown to be correlated with poor fund choices.

The authors build a model that leads to some salient predictions on how savvy (think, institutional investors or those who invest time researching funds) and non-savvy investors invest in the market for mutual funds, and what type of fund fees one would expect to pay. They then test the prediction of their model using US equity mutual fund data to show how several measures of fund visibility (as a stand-in for familiarity) relate to persistence of fund performance and the fees of the funds.

What are their predictions?

The authors consider high quality and low quality funds (where the expected return of the high quality is always greater than the low quality fund), where each fund may be of two types of visibility, high and low. They assume that non-savvy investor will shun the low visibility funds of whatever quality because their overall lack of information about these funds will lead to a high disutility. Also, when evaluating high visibility funds, non-savvy investors are influenced by their familiarity with them. Familiarity is just the distance between investors and funds (where this distance can be interpreted as either geographic distance – as the Japanese example above – or simply a measure of how well they research and thus exposed to information about a fund). Savvy investors consider all funds and are not prone to familiarity bias.

The predictions from their model are then threefold:

  • In equilibrium the bad, low visibility fund will be driven out of the market by the good, low visibility funds as the good quality fund can undercut the fee that bad funds charge.
  • Low visibility funds cater only to savvy investors while high visibility funds cater only to non-savvy investors, that is, there is market segmentation.
  • If investing in non-familiar funds is personally costlier to non-savvy investors such that this cost is higher than the relative difference in performance between a low and high quality, highly visible fund, then bad, high visibility funds will coexist with good, high visibility funds in this non-savvy segment.

So the stronger competition among the low visibility funds for savvy investors leads to homogeneity in quality and fees, and therefore since only the high quality, low visibility fund survives. In contrast, there is heterogeneity in quality and fees among high-visibility funds. Thus, their model further predicts that differences in performance across funds will persist through time for these high visibility funds; but if there is any difference in performance among the low visibility funds, then this difference is short-lived and only due to luck. Finally, when the familiarity bias is somewhat intermediate the highly visible funds charge higher fees than the low visibility funds. These two outcomes predicted on the performance of funds will be what the authors test for in the data.

How do funds measure up?

The authors use the CRSP Survivor-Bias-Free US Mutual Fund Database for the period 1993-2010 and consider the set of funds that are actively managed, diversified, and domestic. Persistence is measured over a year period accounting for the fiscal year definition of funds for turnover and fee setting. To proxy for familiarity the authors regress separately indicators for characteristics based on the family the fund belongs to (a family is simply the firm that manages the fund): number of investment categories in which the family offers funds; family size; family age based on oldest fund; and advertising expenditures of the family. For each proxy, low visibility funds are then those in families at the bottom of the distribution, while high visibility funds are those at the upper end of the distribution.

Overall, the authors find persistence in performance for high visibility US mutual funds and no persistence in the low visibility funds, consistent with the model predictions. These results are robust in that results hold when also considering whether fund size and investment category may influence persistence, as well as whether recent performance may drive the persistence results. The authors also consider whether other types of segmentation (besides familiarity) might drive the performance persistence results. These other possibilities include institutional vs. retail investors (where the intuition is that institutional investors may react more quickly to after-fee performance, making persistency short-lived in the low visibility funds); and the distribution channel, where fee-sensitive investors may engage more direct channels for investing vs. the fee-insensitive investors who may not care about the distribution channel. Here they find no evidence that alternative segmentation hypotheses generate different results (that is, the persistence in fund performance is statistically no different between retail and institutional investors, nor is it different between a brokered or a direct channel fund).

Regarding fees the authors find that low visibility funds charge significantly less, whether the fund is held for 1 year or 5, as their model predicts when familiarity is at medium levels. High visibility funds, on the other hand, charge fees that are no different than other medium visibility funds.

Why is this important?

To the extent that familiarity bias is segmenting the market and preserving poorly performing (low quality), highly visible funds means that investors lose out when they pay more for inferior products. In contrast, the presence of savvy investors in the low-visibility segment of the market, by driving out poorly performing funds, ensures the homogeneity in the operation of funds thus decreasing the price they pay and the expectation of outperformance. The authors note that, unfortunately, even new entrants into the mutual fund market may not remove the performance persistence among high visibility funds (and thus decrease their fees) precisely because the familiarity bias will prevent non-savvy investors from investing with the new entrant.