We propose newly developed unsmoothing techniques for appraisal-based real estate returns based on a regime-switching threshold autoregressive (TAR) model. We show that when true returns follow a TAR process, conventional linear autoregressive techniques are misspecified and underestimate true variance. Two exogenous variables, equity returns and gross domestic product growth, outperform other variables as regime indicators and appear to capture risks of downturns in real estate. We extend the model to the smoothing equation, allowing for switching behavior by appraisers, using two new techniques: the TAR-AR and TAR-TAR approaches. The “co-switching” specification opens up a new frontier of empirical research. We estimate the TAR-TAR using FT returns as the regime indicator, and we find results that outperform conventional smoothing models and have plausible economic explanations.