This paper investigates a multiple input single output (MISO) downlink communication system in which users are equipped with movable antennas (MAs). First, We adopt a field-response based channel model to characterize the downlink channel with respect to MAs' positions. Then, we aim to minimize the total transmit power by jointly optimizing the MAs' positions and beamforming matrix. To solve the resulting non-convex problem, we employ an alternating optimization (AO) algorithm based on penalty method and successive convex approximation (SCA) to obtain a sub-optimal solution. Numerical results demonstrate that the MA-enabled communication system perform better than conventional fixed position antennas.