The structure of the IT industry has always evolved in line with technological progresses and changes in consumer preferences, as well as with regulatory trends. This is why, when assessing the effect that a new technology or industry policy may have on the national economy, companies and policy-makers need to consider dynamic structural changes affecting the IT industry. One of the most popular existing methods for economic impact analysis is based on a traditional input-output table, and is conducted over a period between the current time and a given time in the future. In this study, we compare the accuracy of RAS and Cross Entropy (CE), the two most widely employed methods for updating input-output (IO) tables, by applying them to Korean IT industries. The main results of this study are the following. In terms of the accuracy of input coefficient estimates, we have found that both the RAS and CE methods have a tendency to overestimate or underestimate them. When the Korean industry was first divided into fourteen sectors, and the RAS and CE methods were applied to each of the fourteen industries, it was difficult to discern a consistent trend for the two methods concerning their accuracy in estimation of input coefficients. Secondly, when used to update an IO table in which the IT industry is subdivided into IT equipment and services, neither the CE nor RAS method proved distinctly superior to the other. Third, in light of the above two findings, we concluded that updating IO tables is best done through a hybrid method combining the CE and RAS methods. This paper proposes a procedure consisting of two steps: IO tables are first updated using the two methods, which are once again updated by employing the OLS average approach through the use of optimal weights.
A Field Study on Global R&D Centers in Korea: Focusing on Global IT Industry Policy with Gap Approach