Abstract
National Statistical Institutes (NSIs) must balance between timeliness and accuracy of the indicators they publish. Because some of the house sales transactions are reported several months after they occur, many countries that include Israel, publish provisional house price indices (HPIs) that are subject to large revisions as further transactions are reported. This happens because the late-reported transactions behave differently from the transactions reported on time. In this paper, we propose a novel methodology to minimize the size of the revisions, with illustrations from Israel, but the method can be applied to other countries with appropriate modifications. The proposed methodology consists of nowcasting three types of variables at a subdistrict level and adding them as input data to an extended hedonic model used for the computation of the HPI: (1) the average characteristics of the late-reported transactions such as the average number of rooms and the area size of the sold apartments; (2) the average price of the late-reported transactions; and (3) the number of late-reported transactions. The three variables are nowcasted based on models fitted to data from previous months. Evaluation of our methodology shows more than 50% reduction in the magnitude of the revisions.
Original language | English |
---|---|
Pages (from-to) | 253-266 |
Number of pages | 14 |
Journal | International Journal of Forecasting |
Volume | 38 |
Issue number | 1 |
DOIs | |
State | Published - 1 Jan 2022 |
Bibliographical note
Publisher Copyright:© 2021 International Institute of Forecasters
Keywords
- Delayed reports
- Hedonic model
- Index revision
- Provisional indicator
- Real estate market