Plenty of prominent fund managers seem to enjoy talking their book in the financial media. They reason that it’s free marketing for them and potentially free alpha too, insofar as their celebrity can persuade other investors to buy into their trades and bid up their holdings’ prices.
However, they may be surprised to discover that the sheep are eating the wolves. A new paper by David Lesmond of Tulane University and Roberto Stein of the University of Nebraska at Lincoln outlines an innovative methodology for charting how managers influence each other – and whether it is the influencers or the influenced who outperform.
The sincerest form of flattery
Lesmond and Stein tracked the quarterly portfolio disclosures of actively managed US equity funds between 1996 and 2016, recording all the instances when two funds had a position in the same stock. This inevitably yielded a great many observations. Using a process developed by Nobel laureate Clive Granger to determine whether one time series of data forecasts another, they were able to assess not merely correlation, but also causation in the funds’ trades in these stocks.
This information was then aggregated into what Lesmond and Stein called a ‘copycat score’ based on the percentage of a portfolio’s assets in which the fund’s trades influenced other funds, minus the portion of assets in which it was influenced by others. They deemed funds with the lowest copycat scores ‘leaders,’ those with the highest scores ‘followers,’ and those with a score of zero ‘zeros.’
The Granger test is a well-established measure of causation, but to further verify their findings Lesmond and Stein also looked at passive funds that were excluded from the main investigation. Of course, index products should not be leaders or followers, but their periodic rebalancing could correlate with other funds’ activity. However – reassuringly – this approach did not flag any passive funds as leaders or followers.
Encouragingly, the leaders did beat the zeros: the former group returned a gross monthly average of 68 basis points, ahead of the 52 basis points delivered by the latter. However, the followers did best of all, generating 78 basis points. Even when this was passed through a four-factor alpha model, the results were the same: followers achieved a statistically significant average monthly alpha of 13.6 basis points, compared with just 3.6 basis points for the leaders and a marginally negative reading for the zeros.
Diving into the portfolios, Lesmond and Stein found that the typical holdings of leaders and followers did indeed display similar characteristics: both favored smaller, less liquid and cheaper stocks. They were not, though, momentum stocks; neither type of manager was simply a trend follower, but they were both more akin to traditional small-cap and value stock-pickers.
It also emerged that both the leader and follower classifications were persistent over time, as were the benefits derived from those trading behaviors. Lesmond and Stein drew a number of conclusions. ‘First, imitation or copycatting exists and is part of the toolkit of some mutual fund managers,’ they noted. ‘Second, and most important, it is actually a profitable strategy for the average fund that applies it.’
However, they also noted that the average follower fund appeared to dedicate less than 10% of its assets to aping the leaders. ‘It is clear that these followers limit the imitation of others’ trades to a few stocks within their portfolios,’ Lesmond and Stein said.
‘Nevertheless, they clearly derive a benefit from this copycatting activity. They seem to obtain the added performance of the stock picks made by superior fund managers, the leaders, while at the same time enjoying lower costs as they forego the research required to identify stocks with higher expected returns. Thus, followers may not be as adept as other managers in finding the best investments themselves but, as a group, they seem to excel in identifying fund managers which can.’