Using different econometric approaches and based on a panel of 21 OECD countries this thesis investigate whether differences in structural or policy factors significantly affects the price responsiveness of shocks to demand in the short run and in the cases of abrupt movements in real prices. Over such steeper areas of the housing cycle the analysis focus specifically on finding evidence of asymmetric responses of demand and structural factors on price dynamics. The study of asymmetries in the overall business cycle is a well-developed field in econometric analysis but, to my knowledge, no such efforts are put into the study of housing cycles.I find evidence of different influential driving factors involved for a simple predictor of large house price falls and large house price increases. The existence of asymmetries in the driving factors behind housing cycles could make demand relations dependent on a sort of switching regime, in which case any prediction problem become a nonlinear one. This would evidently affect empirical results based on symmetric models. In a symmetric short term price dynamics model these results indicate the income elasticity of housing prices decrease if the degree of banking supervision is high. However, evaluating periods of large movements in prices, this effect dissipates in periods of booms and increases the risk of a bust. Moreover, according to these estimates, rigidities in supply is an important driving factor in a booming period but not influential during periods of a bust. Considering contraction periods only, an in-sample forecasting model of price falls in a similar framework as the more contemporaneous models discussed in the previous section is estimated. Evaluating the models predictive abilities at historical episodes of large price falls, the risk probabilities are in all but a couple of cases much higher than the country specific average at the eve of a bust. For Belgium, Canada, France, the Netherlands ( 79 case) and the US ( 79 case) the calculated probabilities are between 2.5-3.2 standard deviations away from their respective means.