标题： 人工智能时代的基于技能的劳动力市场极化：印度与美国的比较分析 显示英文标题

标题： Skill-Based Labor Market Polarization in the Age of AI: A Comparative Analysis of India and the United States

摘要： 本文通过比较分析2018年至2023年印度和美国基于技能的就业分布和工资分布，研究了劳动力市场极化现象，并特别关注自动化风险差异和人工智能准备情况。 利用详细的职业层面数据、自动化风险指标以及一系列统计测试，包括工资溢价分析、就业份额检验和工资-就业回归，我们记录了发展中经济体与发达经济体之间劳动力市场的显著结构性差异。 我们的分析得出四个关键发现。 首先，我们发现就业分布模式存在统计学上显著的差异，印度相比美国在低技能就业方面显示出不成比例的集中，特别是在高自动化风险的职业中。 其次，回归分析显示两国之间的工资溢价存在系统性差异，印度的基于技能的工资差距显著更大。 第三，我们发现就业规模与工资之间存在稳健的负相关关系，表明发展中经济体的劳动力供给效应更强。 第四，对特定职业自动化风险的分析揭示，发展中经济体面临“双重脆弱性”——不仅集中在低技能职业，还集中在具有更高自动化潜力的工作岗位，且受到较低的人工智能准备得分的复杂影响。 这些发现为发展阶段如何影响劳动力市场极化模式提供了新的实证证据，并对发展中经济体的技能培训和技术采用政策具有重要意义。 我们的结果显示，传统劳动力市场发展的方法可能需要重大调整，以考虑人工智能在不同发展阶段的不同影响。 显示英文摘要

摘要： This paper examines labor market polarization through a comparative analysis of skill-based employment and wage distributions in India and the United States during 2018-2023, with particular attention to differential automation risks and AI preparedness. Using detailed occupation-level data, automation risk metrics, and a series of statistical tests including wage premium analysis, employment share tests, and wage-employment regressions, we document significant structural differences in labor markets between developing and developed economies. Our analysis yields four key findings. First, we find statistically significant differences in employment distribution patterns, with India showing disproportionate concentration in low-skill employment compared to the US, particularly in occupations with high automation risk. Second, regression analysis reveals that wage premiums differ systematically between the two countries, with significantly larger skill-based wage gaps in India. Third, we find robust evidence of a negative relationship between employment size and wages, suggesting stronger labor supply effects in developing economies. Fourth, analysis of occupation-specific automation risk reveals that developing economies face a "double vulnerability" - concentration of employment in both low-skill occupations and jobs with higher automation potential, complicated by lower AI preparedness scores. These findings provide novel empirical evidence on how development stages influence labor market polarization patterns and carry important implications for skill development and technology adoption policies in developing economies. Our results suggest that traditional approaches to labor market development may need significant modification to account for the differential impacts of AI across development stages.