The paper addresses a problem of the manipulator stiffness modeling, which is extremely important for the precise manufacturing of contemporary aeronautic materials where the machining force causes significant compliance errors in the robot end-effector position. The main contributions are in the area of the elastostatic parameters identification. Particular attention is paid to the practical identifiability of the model parameters, which completely differs from the theoretical one that relies on the rank of the observation matrix only, without taking into account essential differences in the model parameter magnitudes and the measurement noise impact. This problem is relatively new in robotics and essentially differs from that arising in geometrical calibration. To solve the problem, several physical and statistical model reduction methods are proposed. They are based on the stiffness matrix sparseness taking into account the physical properties of the manipulator elements and also on the heuristic selection of the practically non-identifiable parameters that employs numerical analyses of the parameter estimates. The advantages of the developed approach are illustrated by an application example that deals with the stiffness modeling of an industrial robot used in aerospace industry.