In stark contrast with liquid asset returns, commercial real estate idiosyncratic return means and variances do not scale with the holding period, even after accounting for all cash flow-relevant events. This puzzling phenomenon survives controlling for vintage effects, systematic risk heterogeneity, and a host of other explanations. To explain the findings, I derive an equilibrium search-based asset-pricing model that, when calibrated, provides an excellent fit to transactions data. A structural model of transaction risk seems crucial to understanding real estate price dynamics. These insights extend to other highly illiquid asset classes, such as private equity and residential real estate.