||There is now considerable research devoted to testing the dividend pricing model as a predictor of the actual prices of shares traded in financial markets. A general finding is that present value models are not good predictors of asset prices. This lack of fit is interpreted as excess volatility, or alternatively as a failure of the maintained hypothesis that the discount rate for shares is constant. Considerable discretion in the payout of dividends is vested in the managers of firms who may follow rules of thumb in awarding dividends. Managers may also be reluctant to increase dividends unless they expect that the payout can be maintained subsequently. The failure of the present value model can thus be attributed to the dividend process followed by firm managers with considerable discretion over timing and payout forms, which may be substantially different from the process assumed in econometric models. In this paper, we present rather powerful tests of the dividend pricing relation using a unique data set in which dividends are set by market forces independent of managers' preferences. We rely, not upon observations on shares traded on organized financial markets, but on observations taken from the market for condominium dwellings in Korea – perhaps the only market in which dividends are publicly available to consumers and investors for assets bought and sold over short-term intervals. We test the present value model using large panels of observations on asset price movements and dividends. First, we analyze the cross sectional characteristics of returns to investment in housing of differing types and locations, noting the importance of lags and analyzing simple investment strategies. Second, we analyze the "dividend-price ratio model" proposed by Campbell and Shiller (1988) using panels of housing returns differentiated by type and location. In contrast with much of the existing literature, we find broad support for the dividend pricing model in this more general framework. Lastly, the third body of evidence we present involves tests for the stationarity of these ratios in each of our panels. We conduct a series of unit root tests based upon panels of investment returns and dividends, differentiated by type of housing, investigating the stationarity of dividend price ratios. We find that most of the time series are consistent with stationary processes. This provides further support for our finding that the dividend pricing model is consistent with the pattern of housing returns observed in the Korean housing market.