Because Russell reconstitutions can be inferred well in advance, there should be little role for information effects around the announcement or reconstitution dates. There are three prominent price effects that might be relevant. First, the imperfect substitutes hypothesis assumes that stocks have a downward-sloping demand curve. A stock is added to an index faces permanently increased demand, and therefore there is a permanent price increase around additions. Second, the price pressure hypothesis is that predictable demand leads to arbitrage activity pre-addition, causing a temporary price surge which is then fully reversed subsequently. Third, the liquidity hypothesis suggests that firms added to an index will have their liquidity increase and therefore have permanently higher prices. These effects should be symmetric for deletions.

We first look at returns and short interest changes separately. For additions, we find low returns during the reconstitution period and high returns later, suggesting price pressure. For deletions, we find low returns throughout, suggesting the imperfect substitutes effect dominates. For both additions and deletions, we find significant permanent increases in short interest. For additions, this can be explained by increased liquidity; for deletions, this can be explained by imperfect substitution.

We then look at how returns and changes in short interest are related. We estimate a system of equations. For additions, over the reconstitution period, an increase in short interest is associated with lower returns, consistent with a price-pressure effect. In the post-reconstitution period, an increase in short interest is associated with higher returns, suggesting that increased liquidity leads to both, more shorting and higher prices.

For deletions, we find a negative relationship between changes in short interest and returns over both windows. This is consistent with the imperfect substitutes hypothesis.

Finally, we examine the changes in institutional ownership around reconstitutions, and confirm prior work that finds index addition leads to higher institutional ownership.

FoFs primarily invest in other funds in their own families. Over our sample period, between half and three-quarters of our FoF sample only invest in funds within their own family, while between 14% and 27% invest only outside their family. The remainder (9%- 29%) invest both inside and outside. More than 85% of all FoFs invest outside their family only when they do not have an internal option and in recent years when this occurs they ordinarily invest in passive funds.

FoFs selecting passive funds select passive funds that underperform the simple selection rule of picking the lowest expense passive fund tracking the same index. This is true whether the alternative is in the same family or outside the family.

FoFs selecting active funds within their own family have lower alphas than they would have achieved by randomly selecting funds in the same Morningstar category. When comparing the active funds held by FoFs with active funds which could have been selected in the same Morningstar category, the underperformance is about 20 bp per year. When we compare the active funds selected within a fund family with all other funds which could have been selected in the same Morningstar category and in the same fund family they do still worse, now by 34 bp per year.

Clearly FoFs hurt investors when they select funds in general and when they select funds from their own family. While managers of Fund of Funds may have access to special information any benefit from this appears to be outweighed by fund family or manager objectives. We examine whether the manager of a Fund of Funds appears to be following any one of several fund family or manager objectives rather than selecting funds to meet investor objectives.

The fund family or manager objectives we examine are:
(1) Selecting new funds
(2) Selecting high expense funds
(3) Selecting small funds
(4) Selecting funds managed by the FoF manager

FoFs select funds that meet these objectives more often than could arise by chance. This leads to alphas which are more negative than alternatives. When we examine funds that are most likely to be using fund family and manager objectives, we find that this accounts for 82% of the holdings weighted underperformance found when a FoF manager selects funds within his fund family. Finally, we examine which of our four objectives account for the cross sectional difference in FoF alpha. We find that two of the criteria (being in the top 15% of expenses and being managed by the same manager as the FoF) account in part for cross-sectional differences in FoF performance.

For FoFs selecting funds in their own family the pursuit of fund family and manager goals appears to far exceed any value of insider information, and this pursuit accounts for the vast majority of the underperformance.

When we examine FoFs which only invest outside the fund family we find that these "all outside" FoFs outperform FoFs that invest principally within the fund family. These FoFs do not benefit from special access to information nor are they meeting family goals. The fact that they outperform other funds which should benefit from special information but are hurt by pursuing family goals is further evidence that the negative impact of family goals outweighs any benefit from access to special information.

Back to index page