This paper studies the reform effect and the role of training vouchers(계좌제) in Korea’s public training programs. Using a rich administrative data set,so called employment insurance data set, we apply non-parametric matchingand dynamic survival analysis regression methods to measure the effect of thetraining reform for the unemployed in Korea by comparing with unemploymenttraining system in the past.Analyzing the most important type of training in Korea, we find a positiveimpact of the voucher which increases the employment probability. We alsoinvestigate the effect and importance of the locking-in-effect(LIE) of previoustraining system by exploiting the variation in the duration of between twotraining.