Shock waves drastically alter the nature of Reynolds stresses in a turbulent flow and conventional RANS models cannot reproduce this effect. In the present study, we evaluate Wallin and Johansson's explicit algebraic Reynolds stress model (EARSM) against available DNS data for canonical shock turbulence interaction. The model is found to overpredict the post-shock Reynolds stresses for a range of Mach numbers. Budget of the transport equation of Reynolds stresses, computed using linear interaction analysis, shows that the unsteady shock distortion mechanisms and the pressure velocity correlations are important. We propose an improvement to the EARSM to include these effects and redistribute the kinetic energy between the different normal Reynolds stresses. The new model is found to match DNS data for the amplification of Reynolds stresses across the shock and their post-shock evolution, for a range of Mach numbers.