||This paper incorporates loss aversion, a psychological phenomenon widely observed in consumers’ behavior, in the traditional stock-flow models of housing price determination. According to the behavioral literature, there is a fundamental asymmetry in individuals’ consumption patterns. When there is sufficient income uncertainty, people resist lowering consumption in response to bad news about future income much more than they avoid increasing consumption in response to good news. Several authors have confirmed this behavior in what respects the consumption of everyday goods ¾ utilities, food, supermarket shopping, etc. These studies conclude that individuals display an excessive aversion to lowering consumption below a reference-level, usually an average of the recent consumption patterns. How does this behavior affect house prices? Considering that buying a house is not an everyday decision, consumption loss aversion suggests that a decrease in real income will cause a more than proportional drop in individuals’ demand for houses, due to their unwillingness to sacrifice the usual consumption patterns of those everyday goods. We use a simple reference dependent model to formalize this idea. We argue that loss aversion implies that the elasticity of house prices to changes in income should be much higher when income falls below a reference-level than when it surpasses that level. We test this hypothesis using quarterly data for the Lisbon housing market from 1987 to 1996. This data has the significant advantage that it allows us to test the existence of a loss aversion effect while clearly isolating the contributions of the traditionally relevant variables, given the variety of values that they assume. In fact, during this period, besides three well-defined income cycles, Lisbon was an expanding market without any significant space constraints, interest rates ranged from a very high 19% to a much lower 11%, and no significant external shocks occurred. Also, the period is small enough so not to be affected by relevant demographic changes, with the exception of a slight decrease in family size that the model captures. Our empirical estimates produce highly robust results that are consistent with the model’s theoretical predictions. Three mains facts are derived: There is a slow adjustment of prices. In line with Wheaton and DiPasquale’s (1990) findings, we report a quarterly speed of adjustment to market conditions close to 30 to 40% of their long-run equilibrium value. / Not surprisingly, given this slow price adjustment and the non-efficient characteristics of the housing market, house price movements are strongly determined by changes in the cost of ownership, what we can call the speculation motive. / Using the traditional formulation, we find a significant relationship between income levels and house prices. However, this result does not hold when we split the sample based on income cycles. In fact, increases in real income show a small and non-significant impact on house prices once other factors are taken into account. The same is not true when there is a decrease in income. We find that a 1% drop in real income causes house prices to lower by about 0.75%. // This asymmetry has some interesting implications in terms of house price movements. In particular, it is a possible additional explanation for the pronounced cyclical behavior of prices and for the bubbles and busts in the housing market. For example, after a recession, once income levels start to recover, the negative income effect abruptly disappears allowing house prices to significantly go up. This sudden increase in prices will jump-start the speculation motive, which then takes over in a disproportionate manner (thus the bubble) until gradually losing strength (the price elasticity of the cost-of-ownership variable is much lower than one). Inversely, when real income starts to fall, the negative income effect will kick in. This will lower house prices disproportionately, providing additional strength to the downward speculation movement. We confirm these conclusions simulating the consequences of exogenous and endogenous shocks to the housing market, based on our empirical results. These simulations illustrate the extreme volatility to which house prices are subject due to this asymmetry, even when we ignore the usual lag with which stock supplies respond to changes in prices.