We investigate the relationship between public sentiment on specific issues including Sewol ferry disaster and the outbreak of MERS and economic performance in Korea. For this work, we implement two tasks: (1) after training sentiment classifiers with several sources of social media datasets, we consider three kinds of feature sets such as feature vector, sequence vector and combination of dictionary-based feature and sequence vectors. Then, the performance of six classifiers including MaxEnt-L1, C4.5 Decision Tree, SVM-kernel, Ada-boost, Naïve Bayes and Max Ent is assessed. Hence, MaxEnt-L1 shows the highest performance amongst other classifiers. (2) we collect datasets pertinent to a number of critical events that public explicitly express their opinions on such as Sewol ferry disaster, the outbreak of MERS, etc. Then, we calculate sentiment value of the collected data with trained classifiers and compare it with economic indices.