We investigate the role of public research investment (R&D) and learning by doing (LBD) in improving productivity through an empirical examination of the US agricultural production sector. We construct a dual model and track R&D and LBD impacts on returns to scale, production cost, and input demand utilizing data for more than a century. A Bayesian approach is used to maintain regularity conditions implied by economic theory. We find that US agriculture shows significant evidence of increasing returns to scale when both R&D and LBD are included in the production process. R&D and LBD are complementary in reducing cost as an increase in one stock significantly strengthens the cost-reducing effect of the other. The direct impacts of R&D and LBD on scale economies, cost, and input demands are sensitive to choices of R&D lag structure, LBD proxy, LBD knowledge depreciation rate, and data period. But input demand price elasticities are highly robust across model specification.

中文翻译：

---

美国农业中的知识积累：边做边学

我们通过对美国农业生产部门的实证研究，研究了公共研究投资（R＆D）和边做边学（LBD）在提高生产力中的作用。我们构建了一个双重模型，并利用一个多世纪的数据来跟踪研发和LBD对规模回报，生产成本和投入需求的影响。贝叶斯方法用于维持经济理论所隐含的规律性条件。我们发现，将R＆D和LBD都包括在生产过程中，美国农业显示出规模收益增加的重要证据。研发和LBD在降低成本方面是互补的，因为一种库存的增加会显着增强另一种库存的降低成本的效果。研发和LBD对规模经济，成本和投入需求的直接影响对研发的选择敏感 滞后结构，LBD代理，LBD知识折旧率和数据周期。但是输入需求价格弹性在整个模型规范中都非常强大。