This study computes the durability of Return on Assets (ROA) in small and medium enterprises from different sample datasets. Utilizing information from the Financial Statements Statistics of Corporations by Industry...This study computes the durability of Return on Assets (ROA) in small and medium enterprises from different sample datasets. Utilizing information from the Financial Statements Statistics of Corporations by Industry, it verifies the precision of correlation coefficients using the Non-iterative Bayesian-based Imputation (NIBAS) and multiple imputation method for all combinations of common variables with auxiliary files. The following are the three important findings of this paper. First, statistical matching estimates of higher precision can be obtained using key variable sets with higher canonical correlation coefficients. Second, even if the key variable sets have high canonical correlation coefficients, key variables that are correlated extremely strongly with target variables and have high kurtosis should not be used. Finally, using auxiliary flies can improve the precision of statistical matching estimates. Accordingly, the durability of ROA in small and medium enterprises is computed. The author finds that the series of ROA correlation fluctuates for smaller enterprises compared to larger ones, and thus, the vulnerability of ROA in small and medium enterprises can be clarified via statistical matching.展开更多
文摘This study computes the durability of Return on Assets (ROA) in small and medium enterprises from different sample datasets. Utilizing information from the Financial Statements Statistics of Corporations by Industry, it verifies the precision of correlation coefficients using the Non-iterative Bayesian-based Imputation (NIBAS) and multiple imputation method for all combinations of common variables with auxiliary files. The following are the three important findings of this paper. First, statistical matching estimates of higher precision can be obtained using key variable sets with higher canonical correlation coefficients. Second, even if the key variable sets have high canonical correlation coefficients, key variables that are correlated extremely strongly with target variables and have high kurtosis should not be used. Finally, using auxiliary flies can improve the precision of statistical matching estimates. Accordingly, the durability of ROA in small and medium enterprises is computed. The author finds that the series of ROA correlation fluctuates for smaller enterprises compared to larger ones, and thus, the vulnerability of ROA in small and medium enterprises can be clarified via statistical matching.