The purpose of this study is to test several prevalent non-survey methods and find the best one to construct regional input-output model in Korea. For this purpose, the Leontief inverse matrices are deduced from national input-output tables and regional employment data. The results are comparedwith benchmark models which are issued on survey based model. Empirically, MAD(mean absolute deviation) and MAPE(mean absolute percentage error) were computed to measure how far are the estimated multipliers are from true ones. The FLQ(Flegg et al. Location Quotient) and AFLQ(Augmented Flegg et al. Location Quotient) methods are demonstrated as a better version for constructing regional input-output models in Korea. However all the conventional methods(SLQ, CILQ, RLQ) tend to overestimate the true output multipliers by about 50% which is consistent with recent studies.