We use data from the Sloan Digital Sky Survey (SDSS) and the DEEP2 survey to characterize the distribution of stellar mass and light of galaxies in the low-redshift and z = 1 Universe. We investigate the clustering bias of stellar mass and light by comparing these to projected autocorrelations of dark matter estimated from the Millennium Simulations (MS). All of the autocorrelation and bias functions show systematic trends with spatial scale and waveband, which are impressively similar at the two redshifts. This shows that the well-established environmental dependence of stellar populations in the local Universe is already in place at z = 1. The recent MS-based galaxy formation simulation of Guo et al. (2011) reproduces the scale-dependent clustering of luminosity to an accuracy better than 30% in all bands and at both redshifts, but substantially overpredicts mass autocorrelations at separations below ~ 2 Mpc. Further comparison of the shapes of our stellar mass bias functions with those predicted by the model suggests that both the SDSS and DEEP2 data prefer a fluctuation amplitude of σ8 ~ 0.8 rather than the σ8 = 0.9 assumed by the MS.