Dynamical scenarios of terrestrial planets formation involve strong perturbations of the inner part of the solar system by the giant-planets, leading to enhanced impact velocities and subsequent collisional erosion. We quantitatively estimate the effect of collisional erosion on the resulting composition of Earth, and estimate how it may provide information on the dynamical context of its formation. The composition of the Bulk Silicate Earth (BSE, Earth’s primitive mantle) for refractory and lithophile elements (RLE) should be strictly chondritic as these elements are not affected by volatile loss nor by core formation. However, an excess in 142Nd compared to the 144Nd has been emphasized in terrestrial samples compared to most measurements in chondrites. In that case, the Samarium/Neodymium (Sm/Nd) ratio could be roughly 6% higher in the BSE than in chondrites, as suggested from the 146Sm/142Nd isotope system (Boyet and Carlson, 2005). This proposed chemical offset could be the consequence of preferential collisional erosion of the crust during the late stages of Earth’s accretion, leaving a BSE enriched in Sm due to its lower incompatibility compared to Nd (O’Neill and Palme, 2008; Boujibar et al., 2015; Bonsor et al., 2015; Carter et al., 2015, 2018). However, if the present 142Nd of the BSE arises from nucleosynthetic heterogeneities within the protoplanetary disk (Burkhardt et al., 2016; Bouvier and Boyet, 2016; Boyet et al., 2018), then the BSE has no excess in Sm compared to Nd and this hypothesis precludes any significant loss of relatively Nd-enriched component early in the Solar System. Here, we simulate and quantify the erosion of Earth’s crust in the context of Solar System formation scenarios, including the classical model and Grand Tack scenario that invokes orbital migration of Jupiter during the gaseous disk phase (Walsh et al., 2011; Raymond et al., 2018). We find that collisional erosion of the early crust is unlikely to explain the proposed superchondritic Sm/Nd ratio of the Earth for most simulations. Only Grand Tack simulations in which the last giant impact on Earth occurred later than 50 million years after the start of Solar System formation can account for this Sm/Nd ratio. This time frame is consistent with current cosmochemical and dynamical estimates of the Moon forming impact (Chyba, 1991; Walker, 2009; Touboul et al., 2007, 2009, 2015; Pepin and Porcelli, 2006; Norman et al., 2003; Nyquist et al., 2006; Boyet et al., 2015). However, such a late fractionation in the Sm/Nd ratio is unlikely to be responsible for a 20-ppm 142Nd excess in terrestrial rocks due to the half life of the radiogenic system. Additionally, such a large and late fractionation in the Sm/Nd ratio would accordingly induce non-observed anomalies in the 143Nd/144Nd ratio. Considering our results, the Grand Tack model with a late Moon-forming impact cannot be easily reconciled with the Nd isotopic Earth contents.