Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/57711 
Year of Publication: 
2007
Series/Report no.: 
CFR Working Paper No. 07-08
Publisher: 
University of Cologne, Centre for Financial Research (CFR), Cologne
Abstract: 
This paper documents the tendency of mutual fund managers to follow analyst recommendation revisions when they trade stocks, and the impact of analyst revisioninduced mutual fund herds on stock prices. We find that mutual fund herds follow consensus revisions in analyst recommendations, controlling for common investment signals that affect both analyst revisions and mutual fund trading. Consensus upgrades result in herds of funds buying a stock, while consensus downgrades result in even bigger herds of funds selling. Our most important finding is that mutual fund herding impacts stock prices to a much larger degree during our sample period (1994 to 2003) than during prior-studied periods. Further, we find the first evidence that mutual funds appear to overreact when they herd in their trades - stocks heavily bought by herds tend to underperform their size, book-tomarket, and momentum cohorts during the following year, while stocks heavily sold outperform. These reversal patterns are even stronger when herds of mutual funds (especially funds with poor performance records) follow analyst recommendation revisions. An investment strategy that accounts for the direction of both analyst revisions and mutual fund herding generates a return (adjusted for size, book-to-market, and momentum) exceeding six percent during the following year. Our results remain robust when we condition fund herding on analyst earnings forecast revisions instead of recommendation revisions. Overall, our study finds that the interaction between sell-side analysts and mutual fund managers plays an important role in setting prices in equity markets.
Document Type: 
Working Paper

Files in This Item:
File
Size
600.98 kB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.