Assessing Independent Variables Used in Econometric Modeling Forest Land Use or Land Cover Change: A Meta-Analysis

2014 5(7), 1532-1564


We conducted a meta-analysis on 64 econometric models from 47 studies predicting forestland conversion to agriculture (F2A), forestland to development (F2D), forestland to non-forested (F2NF) and undeveloped (including forestland) to developed (U2D) land. Over 250 independent econometric variables were identified from 21 F2A models, 21 F2D models, 12 F2NF models, and 10 U2D models. These variables were organized into a hierarchy of 119 independent variable groups, 15 categories, and 4 econometric drivers suitable for conducting simple vote count statistics. Vote counts were summarized at the independent variable group level and formed into ratios estimating the predictive success of each variable group. Two ratios estimates were developed based on (1) proportion of times the independent variables had statistical significance and (2) proportion of times independent variables met the original study authors’ expectations. In F2D models, we confirmed the success of popular independent variables such as population, income, and urban proximity estimates but found timber rents and site productivity variables less successful. In F2A models, we confirmed success of popular explanatory variables such as forest and agricultural rents and costs, governmental programs, and site quality, but we found population, income, and urban proximity estimates less successful. In U2D models, successful independent variables found were urban rents and costs, zoning issues concerning forestland loss, site quality, urban proximity, population, and income. In F2NF models, we found poor success using timber rents but high success using agricultural rents, site quality, population, and income. Success ratios and discussion of new or less popular, but promising, variables was also included. This meta-analysis provided insight into the general success of econometric independent variables for future forest-use or -cover change research.