Using CPS microdata, 1976-2024, we estimate trend and cyclical components of unemployment and labor force participation for 44 age-gender-education groups. We fit a parsimonious state-space model in which each series is the sum of latent cohort and time-varying age effects and a latent cyclical factor shared across unemployment and participation, without imposing structural covariates. Aggregating group trends with observed population shares, we find that population aging and educational upgrading explain most long-run movements in aggregate trends, while cohort effects drive large gender differences in participation. Combining our estimates with demographic projections and an estimated cohort model of education shares, we forecast that over the next two decades, trend participation declines by about 1.5 pp and trend unemployment falls by about 0.4 pp, remaining historically low.