We introduce a hedonic price model that enables us to disentangle the value of a property into the value of land and the value of structure. For given reconstruction costs, we are able to estimate the impact of physical deterioration, functional obsolescence and vintage effects on the structure and the impact of time on sale (and external obsolescence) on the land value simultaneously. Our findings show that maintenance has a substantial impact on the rate of physical deterioration. After 50 years of not or barely maintaining a home, a typical structure has lost around 43% of its value. In contrast, maintaining a home very well results in virtually no physical deterioration in the long run.

A common definition of liquidity in real estate investment is the ability to sell property assets quickly at full value, as reflected by transaction volume. The present paper makes methodological and conceptual contributions in the study and understanding of liquidity. First, we extend the Fisher et al. (2003, 2007) methodology for the separate tracking of changes in reservation prices on the demand (potential buyers) and supply (potential sellers) sides of the asset market. We show how to apply the methodology to a repeat sales indexing framework, allowing application to typical commercial property transaction price datasets, which lack appraisal valuations or complete data regarding property characteristics. We also use a Bayesian, structural time series approach to estimate the indexes. These methodological enhancements enable much more granular supply and demand index estimation, including at the metropolitan level. Second, we propose a Liquidity Metric based on the indexes, and show that the normal liquidity dynamic in commercial property asset markets is "pro-cyclical", that is, price and trading volume tend to move together, with demand tending to lead supply. But we also observe an "anomalous" dynamic that occurs about 25 percent of the time, in which the Liquidity Metric declines while consummated prices are still rising. This anomalous dynamic is often associated with the end of a period of rapid growth in market values.

In this article, we define a new construct for urban economic and investment analysis, which revisits the conventional wisdom that investment in real estate development is riskier than investment in stabilized property assets. This new construct, referred as a "development asset value index" (DAVI), is a value index for newly developed properties (only) in a given geographical property market. It tracks longitudinal changes in the highest and best use (HBU) value of locations, and it reveals developer and landowner behavior taking advantage of the optionality inherent in land ownership. In particular, the DAVI reflects developers' use of flexibility in the exercise of the call option to (re)develop the property to any legal use and density. We empirically estimate a DAVI for commercial property (i.e., central locations) and compare it with a corresponding traditional transaction-price-based property asset price index (PAPI) corrected for depreciation. We believe that the difference primarily reflects the realized value of flexibility in land development. We find that the DAVIs display greater value growth and are smoother over time and less cyclical than their corresponding PAPIs for the same locations. This suggests that developers successfully use flexibility, and that development may be riskier than stabilized property investment due primarily only to leverage effects (construction costs). Practical implications are also discussed.

This article presents a model agnostic methodology for producing property price indices. The motivation to develop this methodology is to include non-linear and non-parametric models, such as Machine Learning (ML), in the pool of algorithms to produce price indices. The key innovation is the use of individual out-of-time prediction errors to measure price changes. The data used in this study consist of 29,998 commercial real estate transactions in New York, in the period 2000–2019. The results indicate that the prediction accuracy is higher for the ML models compared to linear models. On the other hand, ML algorithms depend more on the data used for calibration; they produce less stable results when applied to small samples and may exhibit estimation bias. Hence, measures to reduce or eliminate bias need to be implemented, taking into consideration the bias and variance trade-off.

No abstract