Abstract: The performances of funds have been paid close attention with the rapidly growth of the funds market. A special phenomenon has been found in funds that performances of funds can be unexpectedly improved at the end of each quarter and at the end of each year. An explanation for this phenomenon is the ‘windows dressing’ of funds, for which the managers will distort the performance of funds by buying the winner and shorting the loser. In this paper, another explanation for the phenomenon will be shown, that is the ‘portfolio pumping’, for which managers will continuously buy lots of the stocks owned by funds at the end of each quarter and at the end of each year. The main difference of windows dressing from portfolio pumping is that the stocks traded by the managers in the windows dressing can not be held by funds at the normal time.
Methods:In order to test the portfolio pumping from windows dressing, the performance of funds is checked with respect to several factors. The measures of performances are multi-dimensions, such as, the daily return, the recess return, abnormal buying or selling volume. And the following regression is used.
LR_it=β₀+β₁ QEND_t+β₂ YEND_t+β₃ QBEG_t+β₄ YBEG_t+ε_it,
where LR_it is the proxy of the daily return, the recess return, abnormal buying or selling volume.
Data:The data from Jan. 2005 to Dec. 2015 for all open-ended funds of China is chosen. There are 2902 open-ended funds including 1122829 daily traded data of funds and 4891112 daily traded data of the heavy warehouse stocks of funds. The benchmark consists of HS300 index and Shanghai Security Exchange Treasury bill.
Results:Our empirical evidences show that, (1) the daily average return is 0.38% at the end of quarter, and 0.62% at the end year higher than the average daily returns. (2) The effects are is indifferent for different investment style. However, the impact of different size funds on pumping is significant. Specially, meddle market value funds have the stronger pumping than small and large market sizes funds, for which are significant at level 1%. (3) The funds with only a single manager is easier to lead to pumping than the funds with team manager, moreover, the daily average return of funds with the single manager at the end of year is 0.14% higher than that of funds with team manager. (4) The findings show from the heavy warehouse stocks of funds that at the end of quarter or year, the declining sale trades, instead of increasing buying trades, may be another reason that the performance of funds is raised.

Key words: portfolio pumping, open-ended fund, funds size, investment style