콘텐츠 바로가기
로그인
컨텐츠
  • HOME
  • SEARCH
PLUS Text Size MINUS RESET
FACEBOOK TWITTER Linked In

Category Open

Resources

tutorial

Collection of research papers and materials on development issues

home

Resources
Economy Financial Policy
Industry and Technology Science/Technology

Print

중소기업 연구개발 지원정책 수혜자 선정모형 연구(Predictive models that select the recipients of R&D grants to maximize the growths of SMEs)

Related Document
Frame of Image


Full Text
Title 중소기업 연구개발 지원정책 수혜자 선정모형 연구(Predictive models that select the recipients of R&D grants to maximize the growths of SMEs)
Similar Titles
Material Type Reports
Author(Korean)

이성호

Publisher

한국개발연구원

Date 2017-12
Series Title; No 정책연구시리즈 / 2017-12
ISBN 979-11-5932-290-7
Pages 118
Subject Country South Korea(Asia and Pacific)
Language Korean
File Type Documents
Original Format pdf
Subject Economy < Financial Policy
Industry and Technology < Science/Technology
Holding 한국개발연구원; KDI국제정책대학원
License

Abstract

The Korean government benchmarked the US Small Business Innovation Research (SBIR) system and has steadily increased its budget for SME R&D support for two decades since 1998. The R&D expenditure of all the ministries granted to SMEs amounted to KRW 2.897 trillion in 2016, accounting for 15.2% of the total government R&D expenditure. Thanks to the enthusiastic support of the government, the total amount of R&D investment by SMEs is second only to the US, surpassing Japan, Germany and France. In the past, there have been many empirical studies on the effect of government R&D support on the firms’ own R&D investment and technological capacity, both domestically and internationally, but the evaluation of economic effects has only recently begun to emerge. This study analyzed the information of 21,265 government R&D projects supporting firms (FY 2010~2014), extracted from the NTIS DB and linked to the KED financial information DB. Among the financial information, ten performance indicators were selected from three categories: operational performance (value-added, sales, and operating profits), capability assets (tangible assets, human capital investment, R&D investment, intellectual property registration, and marketing investment), and fund raising (debt and equity financing). The comparison between the 670,760 observations of non-beneficiaries and the 18,980 observations of beneficiaries finds that the average values of selected beneficiaries are significantly higher for all the indicators. However, by comparing the differences (growths) of the performance index after one to three years, the beneficiary companies grew significantly slower than non-beneficiaries across all the index except IPR. In addition, a causal effect was estimated using a two-step unified method combining a non-parametric genetic matching method and a parametric DID regression model. The estimation found that government R&D support led companies to increase their own R&D investment and IPR registration, to expand debt and equity financing aided by government funds and technology guarantees, and to contribute to increased tangible assets, human assets and marketing investment. However, the two-step unified estimation found that the government supports failed to increase value-added, sales, and operating profits, or even worsened some of them. The reason why the more favorable treatment effect was derived from the two-stage integrated estimation than the simple difference of differences analysis is that the government grants were distributed more to the low growth firms than the high growth firms. Surprisingly, low-growth or negative growth occurred more frequently in companies with larger R&D investments and more IP registrations. This study estimates the heterogeneous treatment effects of each company using the causalTree package developed by Athey et al.(2016), and finds that positive treatment effects are estimated for less than a half of the granted firms. In other words, the negative effect outweighs the positive effect, so that the total effects on the value-added growth was not significant. If the R&D support to be distributed to the companies that are predicted to be negatively affected is reallocated to other companies with positive effects predicted, the positive value increase effect can be more than doubled. Based on the results of the above analysis, this report recommends four policy changes: (1) expanding the use of forecasting models while relying less on the qualitative evaluation of technical experts, (2) increasing the weight of economic values and decreasing the weight of technology capacity in program evaluation, (3) designing the ways of supports aligned with involved risk, and (4) diversifying R&D subjects.