·活动安排 Schedule·

主题：投资互动：幼儿学习的微观动力学和测量

主讲嘉宾：詹姆斯·赫克曼教授

嘉宾主持：钱辉环教授

日期：2023年5月21日，星期日

时间：10:00-11:15

地点：港中大（深圳）行政楼西翼W201

语言：英文

Topic: Interactions as Investments: The Microdynamics and Measurement of Early Childhood Learning

Speaker: Prof. James J. Heckman

Guest Host: Prof. Huihuan Qian

Date: Sunday, May 21st, 2023

Time: 10:00 a.m.-11:15 a.m.

Venue: W201, Administration Building, CUHK-Shenzhen

Language: English

詹姆斯·赫克曼是芝加哥大学亨利·舒尔茨经济学和公共政策杰出服务教授。他致力于了解不平等的起源和技能形成，并制定和应用解决这些问题的战略。赫克曼已经发表了350多篇文章和9本书。赫克曼教授曾获2000年诺贝尔经济学奖、丹·大卫奖和中国政府友谊奖等多项荣誉。他还是芝加哥大学人类发展经济学中心主任。该中心调查贫困和社会不稳定的根源以及改善人类繁荣的政策。

James J. Heckman is the Henry Schultz Distinguished Service Professor of Economics and Public Policy at the University of Chicago. He works to understand the origins of inequality, and skill formation, and develops and applies strategies for addressing these issues. Heckman has published over 350 articles and 9 books. Professor Heckman received the 2000 Nobel Prize in Economics, the Dan David Prize, and the Chinese Government Friendship Award, among other recognitions. He is Director of the Center for the Economics of Human Development at the University of Chicago. The center investigates the sources of poverty and social immobility and policies to improve human flourishing.

·摘要 Abstract·

本次演讲使用了一个在中国实施的广泛模仿的幼儿家访项目的新实验数据，并进行了高频测量，以研究技能形成的动力学。我们发现，促进儿童发展的家访干预措施促进了家访者和照顾者之间的高质量互动。我们报告了与动态互补性一致的非参数证据，即早期投资使后期投资更具生产力，而不依赖于对技能的任意衡量。基于这一证据，我们制定并估计了一个针对多种技能的动态随机学习模型，并量化了早期生活学习的来源。我们的模型结合并扩展了两个基本且广泛使用的学习和测量的心理测量模型：IRT模型和BKT模型。使用我们的模型，我们检验了人们普遍认为的技能（“人力资本”）存在恒定单位计量的假设，这些单位计量在不同技能水平和不同年龄段的潜在语言和认知技能方面具有可比性。我们发现有证据支持某些技能水平，但并非能证明所有技能水平。我们的随机增长模型可以解释一个常观察到的现象——一个人的技能将随随着时间的推移产生“衰退”。

This talk uses novel experimental data from a widely-emulated early childhood home-visiting program implemented in China with high-frequency measurements to investigate the dynamics of skill formation. We show that home visiting interventions promoting child development promote quality interactions between home visitors and caregivers. We report non-parametric evidence consistent with dynamic complementarity - that early investment makes later investment more productive - that does not rely on arbitrary measures of skills. Based on this evidence, we formulate and estimate a dynamic stochastic learning model for multiple skills and quantify the sources of early life learning. Our model unites and extends two basic and widely-used psychometric models of learning and measurement: the IRT model and the BKT model. Using our model, we test the widely held assumption of the existence of constant unit measures of skill ("human capital") that are comparable over levels of skills and across ages for latent language and cognitive skills. We find evidence supporting it for certain skill levels but not for all. Our stochastic growth model can explain the frequently observed phenomena of “fadeout” of skills over time for an individual.