The latest revision of KERI’s macroeconometric model has two noteworthy aspects. First, the revision reflects the need to update the macroeconometric model periodically for efficient provision of economic projections and analysis of policy effects. Second, it reflects quantitative techniques designed to address the limitations of the traditional econometric model used since the 1980s. Problem areas with the KERI 2005 Model, including spurious regression, have been improved by applying an error-correction model, etc. to the estimate of individual equations.

 

The differences between the latest model and KERI 2005 model are summarized as follows. First, the new model reflects the advantages of an error-correction model in the estimate of individual equations in order to resolve spurious regression problems identified in the previous model. Second, it was composed in a way that the GDP gap can be related to real economy sectors, investment function, etc., as the well as the price sector, and also improved the relationship among respective sectors by including the won-U.S. dollar exchange rate, inventories, export unit prices, etc. that were treated as exogenous variables in the previous model. Third, it segmented endogenous variables that reflect changes in economic environments. Considering increases in the ratio of intangible fixed investments, it separated intangible fixed asset investments, which were included with facility investment in the previous model, and estimated the consumption function after distinguishing between durable goods and non-durable goods. In terms of the export function, the model established separate regions-China and others-to reflect the nation’s deepening export dependance on China. Fourth, the model now includes seasonal dummy variables while continuing to used the original series, eliminating possibility of occurrence of errors in the conversion process into seasonal adjustment or original series, and also changed the base year for GDP and various price variables to 2005.

The revised and supplemented KERI 2010 Model has improved predictability over the previous model and its policy simulation results better harmonize theory and empirical data. Taking into account the limitation of macroeconometric models that can not reflect changes in the behavior of economic players, however, it will be necessary to make re-estimates periodically whenever new data, information, etc. are obtained and to improve continuously its ability to predict and explain outcomes.