<blockquote id="5nul7"></blockquote>

  1. 學術活動
    當前位置:首頁 ? 學術活動 ?

    華南經濟論壇第264期:Douglas Hanley助理教授講座

    2019-06-13 14:44:00 來源:院科研辦 點擊: 收藏本文

     題目: Geographic Clustering of Firms: A Machine Learning Approach

    主講人:Douglas Hanley,the University of Pittsburgh. 

    時間:2019年6月13日(周四)下午3:00

    地點:學院301會議室


    摘要: Understanding how the geographic arrangement of firms affects outcomes such as productivity and growth is a widely studied question in economics. Much of the literature has focused on constructing statistics that characterize firm clustering and using these as inputs into traditional regression analysis. In this paper, we use techniques developed in the field of machine learning (ML) to tackle this question. In particular, we employ convolutional neural networks (CNN) that are widely used in image classification tasks and apply them to the firm density maps considered in our setting. Using state-of-the-art techniques and tools from ML on comprehensive Chinese firm data, we find that such an approach has predictive power over and above traditional techniques. We investigate methods of interpreting such complex algorithms analogous to the impulse response approach used with DSGE models. Additionally, we explore ways to account for confounding factors such as industry in our analysis, as well as to deal with issues of endogeneity of firm location choice through the use of instrumental variables.


      
    pk10牛牛稳赚技巧