Scholarlink

<p>This study examines how efficiently countries convert R&amp;D resources into observable innovation outputs and investigates whether large-scale R&amp;D investment is systematically associated with higher input–output efficiency. Using panel data for 51 countries over 2016–2020, we model gross domestic expenditure on R&amp;D and total researchers as inputs, and patent applications and scientific publications as outputs. In the first stage, we estimate technical, pure technical, and scale efficiency using an output-oriented Data Envelopment Analysis (DEA) framework under both CCR and BCC specifications. To capture temporal variation, we implement a DEA-Window analysis (BCC model) with a three-year window width. In the second stage, we explore environmental correlates of efficiency using a Random Effects Tobit model, incorporating GDP per capita, educational attainment, government-financed BERD share, manufacturing value added, trade openness, and the Corruption Perceptions Index. The DEA results reveal substantial cross-country heterogeneity. A small set of emerging economies consistently lies on the efficiency frontier, while several advanced economies display full pure technical efficiency but lower overall efficiency, indicating scale-related shortfalls. The dynamic analysis suggests that in many mature R&amp;D systems, efficiency scores remain relatively stable over time, implying persistent structural patterns rather than transitory fluctuations. The Random Effects Tobit estimates indicate negative associations of efficiency with GDP per capita and educational attainment, while governance-related indicators exhibit more conditional relationships; the remaining covariates show limited statistical association. Overall, the findings suggest that cross-national differences in R&amp;D performance are not explained by input scale alone and may reflect broader structural and institutional conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Cross-country dynamics in R&D efficiency and environmental correlates (2016–2020): DEA-Window and Random-Effects Tobit

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月22日

查看全文
  • Heung Hee Kim,
  • Dae Geun Kim

摘要

This study examines how efficiently countries convert R&D resources into observable innovation outputs and investigates whether large-scale R&D investment is systematically associated with higher input–output efficiency. Using panel data for 51 countries over 2016–2020, we model gross domestic expenditure on R&D and total researchers as inputs, and patent applications and scientific publications as outputs. In the first stage, we estimate technical, pure technical, and scale efficiency using an output-oriented Data Envelopment Analysis (DEA) framework under both CCR and BCC specifications. To capture temporal variation, we implement a DEA-Window analysis (BCC model) with a three-year window width. In the second stage, we explore environmental correlates of efficiency using a Random Effects Tobit model, incorporating GDP per capita, educational attainment, government-financed BERD share, manufacturing value added, trade openness, and the Corruption Perceptions Index. The DEA results reveal substantial cross-country heterogeneity. A small set of emerging economies consistently lies on the efficiency frontier, while several advanced economies display full pure technical efficiency but lower overall efficiency, indicating scale-related shortfalls. The dynamic analysis suggests that in many mature R&D systems, efficiency scores remain relatively stable over time, implying persistent structural patterns rather than transitory fluctuations. The Random Effects Tobit estimates indicate negative associations of efficiency with GDP per capita and educational attainment, while governance-related indicators exhibit more conditional relationships; the remaining covariates show limited statistical association. Overall, the findings suggest that cross-national differences in R&D performance are not explained by input scale alone and may reflect broader structural and institutional conditions.