The first two papers in this dissertation discuss a fairly recently developed research field, Space Syntax, and how the findings in this field may be used to understand spatial economic patterns such as geographic distribution of market rents. Both papers use standard econometric methods to investigate the relationship between rents and the so called integration value developed within Space Syntax. The integration value may be understood as a measure of the accessibility of a certain location in a street network. The measure is constructed using tools from graph theory and uses the shape of the street network as its only input. The papers estimate hedonic models of office and retail leases from central Stockholm to test whether the integration value can help explain rents. A statistically significant effect of integration value on both office and retail rent is found. It appears as if Space Syntax adds important information to the understanding of intraurban rent patterns. Illiquidity is a main feature of most property markets and market participants are therefore directed to property appraisals to obtain information about market values. The reliability of property appraisals is therefore an important research topic. The third paper studies the “rationality” of valuations by testing if capitalisation (cap) rates from individual discounted cash-flow (DCF) valuations are consistent with economic theory. Standard econometrics is used to study the variation in cap rates. For the most part the results support the hypothesis that appraisers are “rational” in the above mentioned sense. Illiquidity of direct property also poses a problem when constructing property price indices.  Lack of price observations and heterogeneity among the few observations available is likely to introduce noise in price indices based on transactions. Valuations are therefore often used instead to construct indices. These indices however suffer from a bias due to so called “appraisal smoothing”. In the fourth paper it is shown that, given certain assumptions, one may filter out noise in a transaction-based price index by regressing it on a valuation-based index (contemporaneous and lagged one period). The procedure may in some circumstances improve pure valuation- or transaction-based indices. / QC 20101201