Basic Report

Public libraries nowadays are playing a role of cultural space for local communities while still performing their traditional function of providing local people with knowledge information through library materials. This expansion in library functions makes a public library an axis of life. As a result, 160 of the 289 projects selected for the SOC project in 2020 included public library and small library projects. In particular, 73 projects among them included public libraries, and the government aid amounted to 233.1 billion won.
   Moreover, the need to remodel old and aging libraries and the demand for new libraries in new town have increased the number of public library projects with local governments. Therefore, the number of requests for feasibility studies under Article 37 of the Local Finance Act, which must be implemented before the investment examination, has increased.
   Therefore, it is necessary to develop an appropriate methodology to enhance the reliability and accuracy of library demand estimation which is the core of feasibility study. Although one can find a list of methods for demand estimation in 『Guidance for feasibility study for cultural / sports / tour investment projects』(Local Investment Management Center, 2016), it would be better if the list included methods specific to public library feasibility study than otherwise. Ham Yun-ju et al.(2019) conducted an empirical analysis of demand estimation for public libraries located in Anyang city through a gravity model, but there is a lack of research on the estimation of demand related to public libraries.
   This study reviewed the features imposing impacts on the library demand and the existing method of estimating demand, and developed models that predict the demand for public libraries through the microeconometrics (program evaluation) and the data science. Since the National Public Library Statistics provides a vast amount of library-related data, we have constructed a library-visitor-prediction model that can be used for feasibility studies in the future by using machine learning methods including XGboosting. Also, the importance of variables in data was considered through measuring their contribution on making the prediction value.
   Furthermore, this study analyzed the impact of a library provision using the number of visitors as a measure of the impact. To this end, we used a fixed effect model that is a method for controlling for omitted variables in panel data when the omitted variables vary across entities but do not change over time. As a result, we concluded that when investing in the library system of interest we should first consider increasing the number of seats and collections of the existing libraries. Further, the construction of a new library should be legitimate only if the capacity of the library system is short.