Longitudinal scaling of pseudorapidity distribution of charged particles (dN_ch/dη) is observed when presented as a function of pseudorapidity (η) shifted by the beam rapidity (η − y_beam) for a wide range of collision systems (e⁺ + e⁻, p + p, d + A, and A + A) and beam energies. Such a scaling is also observed for the elliptic flow (v2) of charged hadrons in A + A collisions. This is a striking observation, as v2 is expected to be sensitive to the initial conditions, the expansion dynamics, and the degrees of freedom of the system, all of which potentially varies with collision system and colliding energies. We present a study of the longitudinal scalings of dN_ch/dη, average transverse momentum (‹pT›), and v2 using transport models UrQMD and AMPT for Au + Au collisions at center of mass energies (√s_NN) of 19.6, 62.4, and 200 GeV and Pb + Pb collisions at 2760 GeV. Only the AMPT models which includes partonic effects and quark coalescence as a mechanism of hadronization shows longitudinal scaling for dN_ch/dη, pT, and v2. Whereas the UrQMD and AMPT default versions show longitudinal scaling only for dNch/dη and ‹pT›. We also discuss the possibility of longitudinal scaling of v2 within two extreme scenarios of models with hydrodynamic and collisionless limits. We find the longitudinal scaling of bulk observables to be an important test for the underlying physics mechanism in models of particle production.