作者: Andrew J. Plantinga , Ralph J. Alig , SoEun Ahn
DOI: 10.1093/FORESTSCIENCE/46.3.363
关键词: Ordinary least squares 、 Panel data 、 Variables 、 Econometrics 、 Set (abstract data type) 、 Land use 、 Random effects model 、 Component (UML) 、 Mathematics 、 Dummy variable
摘要: Predictions of future forestland area are an important component forest policy analyses. In this article, we test the ability econometric land use models to accurately forecast area. We construct a panel data set for Alabama consisting county and time-series observation period 1964 1992. estimate using restricted sets-namely, from early periods-and out-of-sample values dependent independent variables precise tests model's forecasting accuracy. Three model specifications examined: ordinary least squares, dummy (fixed effects), error components (random effects). find that produces more accurate forecasts at state level than other specifications. This result is related completely control cross-sectional variation in variables. suggests estimated parameters better capture temporal relationship between economic