AI Driven Productivity surge: Transforming the Indian service sector workforce, Journal of Indian School of Political Economy, ISSN: 0971-0396, Volume: 36, NO: 02, January-June:2024.

This paper primarily analyses the relationship between labor productivity and Artificial intelligence in Indian service sector. For this purpose, variable such as the value added, amount of capital stock and expenditure on research and development chosen. The value added is treated as a dependent variable, on the contrary other variables are treated as explanatory variables. Here the expenditure on R&D is considered as a proxy for investment in AI. Although AI has been playing major role in output maximization, here the attempt is made to capture AI’s impact on Labor productivity, The data of capital stock and labor productivity were taken from RBI KLEMS data base and the expenditure on research and development taken from: NSTMIS, Department of Science & Technology. The data has been filtered for the research period from 1991 to 2023. The data was transformed into log form for easier statistical interpretation. The measure of central tendency and correlation matrix were executed. For each variable stationary of the data has been worked out using Augmented dickey fuller unit root test. It was observed that there were mixed differences. Hence the ARDL (Autoregressive Distributed log) model was chosen to for accurate model building. The results depicted that Labor productivity and capital stock are correlated around 0.993, that of labour productivity and research & development by 0.996, and whereas capital and research and development 0.992 show that there is a high positive link between labour productivity, capital stock, and research and development. This implies that a rise in one variable tends to boost the others as well. With an R-squared value of 0.7457, the regression model explains a considerable amount of the variance in labour productivity. Specifically, variations in capital stock and research & development account for around 74.57% of the variability in Labor Productivity. Key Words: Artificial Intelligence, Labor Productivity, Research and Development, ARDL model. JEL: C30, C32, C67

本论文主要分析印度服务业中劳动生产率与人工智能（Artificial Intelligence）的关联。为此，研究选取了增加值、资本存量额以及研发支出作为变量。其中，增加值被设定为因变量，其余变量则作为解释变量。此处将研发支出作为人工智能投资的代理变量。尽管人工智能在产出最大化中一直发挥着重要作用，但本文旨在探究人工智能对劳动生产率的影响。资本存量与劳动生产率的数据取自印度储备银行KLEMS数据库（RBI KLEMS database），研发支出数据则取自科学技术部下属的NSTMIS（国家科学技术管理信息系统）。研究时段筛选为1991年至2023年，为便于统计解读，所有数据均经过对数变换处理。本研究开展了集中趋势测度与相关矩阵分析，并通过增广迪基-富勒单位根检验（Augmented Dickey-Fuller unit root test）对各变量的数据平稳性进行检验，结果显示变量间存在混合阶数单整。因此，本文选用自回归分布滞后模型（Autoregressive Distributed log, ARDL）开展精准建模。结果表明，劳动生产率与资本存量的相关系数约为0.993，劳动生产率与研发支出的相关系数为0.996，而资本存量与研发支出的相关系数为0.992，这说明劳动生产率、资本存量与研发支出之间存在高度正相关关系。这意味着任一变量的提升均会带动其他变量的增长。该回归模型的决定系数（R-squared）为0.7457，能够解释劳动生产率约74.57%的变异量，即资本存量与研发支出的变动可解释劳动生产率约74.57%的波动。关键词：人工智能（Artificial Intelligence）、劳动生产率、研发支出、自回归分布滞后模型（ARDL model）。JEL分类号：C30、C32、C67