Conflicting Family Values in Mutual Fund Families

A major reason for the existence of conglomerates or business groups is to create internal capital markets to promote the efficiency of the group. One of many efficiency measures that internal capital markets can offer is an insurance pool, which provides temporary liquidity to the members of the group in the event of adverse shocks.  If mutual fund families, which are a collection of legally independent entities tied together by the sponsoring management company, are regarded as groups, it seems reasonable to assume that there would be a group interest. If so, it seems natural to ask whether insurance pools could exist in these families where cash-rich mutual funds direct capital to family funds that are facing large redemption requests, as these redemptions could lead to large fire sale losses. However, by law, they cannot. This is because, while the provision of such an insurance pool against temporary liquidity shocks benefits the family, the cost is borne by the shareholders of the fund providing this “free” insurance. A mutual fund owes a fiduciary responsibility only to its own shareholders, and not to its family.

In our paper, Conflicting Family Values in Mutual Fund Families, forthcoming in the Journal of Finance, we address whether such insurance pools exist in mutual fund families. We examine this by analyzing the investments of affiliated funds of mutual funds (AFoMFs). AFoMFs are mutual funds that only invest in other mutual funds within the family. Instead of the investors or their financial advisors choosing which mutual funds of the family to invest in, AFoMFs do that for the investors. Virtually non-existent in the 1990s, these funds have become very popular. In 2007, which is the last year of our sample.  Of the 30 large families that made up around 75% of the industry’s assets, 27 had AFoMFs.

To determine whether AFoMFs provide liquidity to member funds in need, we divide total fund flow to each ordinary mutual fund into AFoMF flow and non-AFoMF (or outside investor) flow. The flows are normalized by the underlying fund’s value. We find that when we sort each family fund into deciles based on the flow from its outside investors, the lowest decile (i.e., the group of distressed funds/funds experiencing the largest withdrawals from their outside investors) has a statistically significantly higher average inflow from its family AFoMFs than any of the other nine deciles. This is our primary evidence showing that AFoMFs offset severe liquidity shortfalls of funds in the family. Interestingly, though we scan through the AFoMF prospectuses, we find that none of them mention liquidity provision as an objective.

We perform several additional tests to confirm that what we find is not a spurious result but evidence that AFoMFs are purposefully providing liquidity to distressed funds. First, if AFoMFs provide liquidity to distressed funds, this liquidity support should not exist for funds that rarely need this support. We find that AFoMFs do not favor distressed funds that are money market funds, Treasury funds, or ETFs. Second, the liquidity position of the AFoMF should not matter. We find that AFoMFs provide liquidity to distressed funds even when the AFoMFs are cash poor. Third, if insurance, the AFoMF should be providing liquidity for transient liquidity shortfalls rather than persistent shortfalls. We find that AFoMFs do not help underlying funds that have persistent liquidity shortfalls. Fourth, if insurance, it should not be provided by unaffiliated funds of mutual funds (UFoMFs), which are funds of funds that can only invest outside the family. We find that UFoMFs do not provide liquidity to distressed funds. Fifth, as fire sale costs are higher for less liquid funds than for more liquid funds, the temporary liquidity provision should be more for less liquid funds. We find that AFoMFs provide greater insurance to less liquid distressed funds than to more liquid U.S. equity funds. Sixth, and finally, if most funds in the same style are trying to sell at the same time, costly fire sales are more likely; therefore, temporary liquidity provision by the AFoMF should be more likely. We find that AFoMFs favor distressed funds more if other funds in the distressed fund’s style are also selling. Multivariate tests confirm the above main univariate tests. In these tests, we control for measures of the underlying fund’s liquidity, the AFoMF’s liquidity, and various characteristics of the underlying fund, such as size, fund fees, and past performance.

Why do AFoMFs favor distressed mutual funds in their families? Thus far, our discussion is biased towards suggesting that they do so solely to help member funds avoid costly liquidity driven trades. This explanation is motivated by prior research. Existing studies show that liquidity induced mutual fund trading is indeed costly. Edelen (1999) argues that these trades are uninformed and, as a result, lead to losses against informed traders in the order of approximately 140 basis points annually. Moreover, Coval and Stafford (2007) find that large redemptions induce fire sales that generate a significant price impact in the markets.

Fund managers dispute our claim that AFoMF investment in distressed funds is used to help these funds.  Their alternative explanation is that many AFoMFs are asset allocation or target date funds that maintain target weights in various asset classes. This implies a mechanical injection of inflows into any distressed fund whose asset class value has fallen below the target. To check for this, we construct a variable that measures the current deviation from the target weight. We find that our liquidity provision, though diminished by the addition of this control variable, remains. A second explanation is that the temporary liquidity provision is only given to the top-performing or high-fee mutual funds, and so it is just a strategy to protect these funds. We find that liquidity provision exists for all types of funds, except the extreme losers. A third explanation is that liquidity provision to another fund only occurs if the manager of an AFoMF manages the other fund as well. We find that this is not true.

A final alternative explanation is that AFoMFs may have inside information that others do not, and so they act as smart contrarian investors. This is conceivable since the AFoMFs are geographically close to their own family (e.g., Lee (2011), Gervais, Lynch, and Musto (2005), Massa and Rehman (2008), Coval and Moskowitz (2001)). If AFoMFs invest in distressed mutual funds because they have superior information and believe that these distressed funds are undervalued, AFoMFs should profit by going against the crowd. We follow the smart money literature (e.g., Gruber (1996), Zheng (1999), and Sapp and Tiwari (2004)) to examine this alternative hypothesis. We find that AFoMFs lose by providing liquidity to the distressed funds.

Finally, to address whether liquidity provision is a rational family strategy, we test whether the sacrifice, which is the cost incurred by AFoMF shareholders from these investments, benefits the family. We first measure the benefit. We find that though liquidity shortfalls hurt fund performance, this hurt is ameliorated by the AFoMFs’ inflow. This amelioration is the fund’s benefit. We next find that if the AFoMF invested in the distressed portfolio the same way it invested in the other portfolios, its performance would have improved. This improvement sacrifice is the AFoMF’s cost. We find that the benefit to distressed funds exceeds the AFoMF cost. Though we cannot draw definitive conclusions from our low frequency data and a back-of-the-envelope calculation, the results hint that the cross-subsidy may be rational for the family.  The cross-subsidy, however, if it exists, is illegal.

The full paper is available for download here.