Institutional Repository of Vilnius University
How do aggregate macroeconomic variables and valuation ratios perform in forecasting house price growth in lithuania? /
Abstract
dc:description43 pages, 5 figures (without 16 in Annexes), 3 tables, 31 references. This academic paper studies the forecasting performance of 16 potential predictors of house price growth in Lithuania, comparing it between two differently constructed housing price indexes and three forecasting horizons (six model specifications in total), as well as against four techniques of combination forecasts. The paper starts from the analysis of past studies, followed by a description of applied methodology, gathered data, and obtained results. Literature about real estate price determinants and forecasting is discussed in the overview of preceding studies, based on which potential predictors are selected, used forecasting techniques analysed, and expected results formed. A popular forecasting procedure was chosen to provide empirical results, in which indicators under consideration were added into the simple ARDL model as regressors. Forecasts of housing price index growth were computed using this model and compared against benchmark AR forecasts based on produced forecast errors measured by mean squared forecast error. Afterwards, combinations of individual forecasts were constructed utilizing simple mean, median, weighted average, and cluster approaches. Six variables were found to outperform benchmark forecasts in at least half of model specifications, including price-to-national income ratio, debt-to-GDP ratio, and industrial production. The relative usefulness of indicators in forecasting housing price index growth was discovered to improve with increasing forecasting horizon, explained by reducing impact of historical values on future prices with more extended time intervals. Results are consistent across the two housing price indexes and groups of indicators. None of the predictors reduced forecast error in all model specifications, which was achieved by all forecast combination methods (except for weighted average), suggesting that combination forecasts should be used in practice when forecasting housing prices due to difficulties in defining best-performing variables in all scenarios a priori.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Pikūnas, Domantas,
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- eng
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD193224626&prefLang=en_US