House price forecasts are important for several parties. For example, (potential) owner-occupiers want to know whether or not it is a good time to buy or sell a home, and decisions of policy makers and commercial parties may depend on house price expectations. In this paper we are not primarily interested in point estimates, but the focus lies on the distribution of the forecasts. House prices depend on other variables, like interest rates, income, wealth, construction costs, number of households and houses, so forecasts of house prices will depend on forecasts of these variables. For that reason we model the relation between these variables, using a time series of monthly price indices in the Netherlands from January 1974 to December 2010. Some explanatory variables are available on a yearly basis, others on a quarterly or monthly basis. We use two different modeling approaches. The first is an error-correction model, in which a distinction is made between the long and the short-term price developments. The long-term price development depends on the levels of the variables, such as the height of the income and the interest rate. This long-term equation defines for each period an equilibrium price. If the actual price is above the equilibrium price, there is talk of overvaluation. In the short-term equation price changes are explained from previous price changes, changes in the explanatory variables and the deviation between the actual price and the equilibrium price in the previous period. In the second approach we use frequency domain techniques to bring all horizons, observation frequencies and variables together. All time series are decomposed into several components, such as long term and short term, using filtering techniques. The filtered components have zero correlation and fully add up to the original series. Spectral analysis techniques can be used to determine the optimal frequency decomposition. All aspects of the time series behavior can be modeled correctly simultaneously due to separate modeling of the various components. Because the correlations between the components are zero, separate models for the various components can be devised. In the end the results from the component models can then be brought together again. Both models are used to estimate the historical time series, and based on these estimates scenarios of house prices and the corresponding explanatory variables are made. The scenario results from both models are compared.