President, National Institute for Public Policy Analysis (INAPP) Technological change, employment and welfare 技术变革、就业和福利 Prof. Stefano Sacchi President, National Institute for Public Policy Analysis (INAPP) 意大利公共政策分析研究所主任 Component One - 2018 Training Programme “Effects and Tendency of Income Redistribution Policy” “收入在分配政策的调节效果和调整趋势”培训 Italy, October 14th - 28th, 2018 2018年10月14-28日，意大利

Outline 大纲 Context and literature 背景和文献资料 Evidence for Italy 意大利的情况 Key research questions 关键研究和问题 Data and descriptive evidence 数据和描述性证据 Econometric results 计量结果 Conclusions and policy challenges 结论和政策挑战 2

Context 背景 All across the world economy, workers are increasingly exposed to sources of uncertainty affecting (and potentially reshaping) their behaviour and preferences 对于全世界所有的经济体，职工都越来越多的面临着影响（及潜在重新塑造）其行为和倾向的不确定因素来源 Among these sources a privileged place is held by labor-disruptive technological change that is recognized as a key driver of polarization and inequality in terms of employment and income 在这些因素来源中，占最主要地位的是劳动颠覆性技术变革，这被认为是就业和收入的两极化和不公平的主要驱动因素 Tech-related uncertainty may 与技术有关的不确定因素可能 induce precautionary behavior (consumption and savings) 引发有害的行为（消费和储蓄） affect workers’ preferences for redistributive policies 影响职工对再分配政策的倾向性 3

Impact of tech change on economy and society 技术变革对经济和社会的影响 Impact of technological change on: 技术变革带来的影响： Employment and wages 就业和薪资 Welfare state 福利制度 State capacity to fund the welfare state 国家为福利体系提供资金的能力 The politics of the welfare state 福利体系的政策 Political participation 政治参与 4

背景：下一轮工业革命（第四届世界经济论坛）

Labor markets 劳动力市场 Routine Biased Technological Change 对技术变革的一般偏见 Employment and Wages: occupations and tasks 就业和薪资：职业和工作 Polarization of employment opportunities and wages 就业机会和薪资水平两极化 Previous “revolutions” have brought about more employment 之前的“革命”带来了更多的就业 AI and ML: Is this time different? 人工智能和机器语言：这次真的不同？ Labor-saving tech change and role of redistribution of productivity growth 节省劳动力的技术变革和生产率增长的再分配的功能 Platform work: quality of work, industrial relations 平台工作：工作质量、产业关系 Taxation in the digital economy and funding the welfare state 对数字经济的征税和为福利制度提供的资金 6

1980-2005年就业的平滑变化曲线，按技能百分位数计算 Source: Autor and Dorn, 2013

1980-2005年真实小时工资的平滑变化曲线，按技能百分位数计算 Source: Autor and Dorn, 2013

1970年代到2005年总就业占比变化 1980年就业中，包含最低技能五分位数的职业 Source: Autor and Dorn, 2013

Impact of technological change 技术变革的影响 Source: Arntz, Gregory and Zierahn, 2016.

技术变革的政治（法律和伦理）影响了创新的步伐 Mediating factors 调节因素 Politics (law and ethics) of technological change impacts upon pace of adoption of innovations 技术变革的政治（法律和伦理）影响了创新的步伐 Relevance of skills (soft skills) and skills-formation institutions 技能的相关性（软技能）和技能形成机制 In any case: transitional period – management of (new?) social risks and demand for compensation 在任何情况下：过渡时期 —（新的？）社会风险管理和补偿的需求 11

Welfare state 福利机制 Three strands of research (Busemeyer et al): 研究的三个分支 Impact of tech change on preferences for redistribution 技术变革对再分配倾向性的影响 Collective representation, industrial action and workers’ power resources in the digital era: mediated by institutions or market power will have its way? 数字时代中的集体代表、劳工行动和职工权利来源：相关体制或市场势力有自己的方式吗？ Social policy responses 社会政策响应 Iversen and Rehm: External modernization effect (tech change inducing new demands, both compensation and social investment – skills!) 外部现代化效应（技术变革带来了新的需求，包括补偿和社会投资—技能！） Internal modernization effect (use of new tech in welfare state provision) 内部现代化效应（在提供福利体系时使用新的技术） 12

Acceptability of robots 机器人的可接受度 太空探索 制造 搜寻和营救 军队和保安 家用 农业 交通/物流 医疗卫生 休闲 教育 护理 Source: Eurobarometer 2012.

Xiao-Yi passed China’s National Medical Licencing Examination 小易通过了中国国家医学资格考试

Support for redistribution 对再分配的支持 Expectation: tech change related to support for redistribution and insurance (Meltzer and Richard 1981, Rehm 2016); similarly to globalization (Walter 2010). Link btw being a globalization loser and voting for populist right (Colantone and Stanig) 预期：技术变革对再分配和保险起到支持作用，这与全球化相同。无法应对全球化和支持民粹主义权利之间的联系。 Very little research 相关研究很少 Kitschelt and Rehm (2014): occupations play an important role in shaping attitudes towards the welfare state 在形成对福利体系的态度方面，职业起了很大作用。 Dekker et al (2017): fear or robots at work related to labor market vulnerabilities (low education, female, unemployed, manual) 在工作中恐惧使用机器人与在劳动力市场中处于弱势有关（低学历、女性、失业、体力劳动）。 Thewissen and Rueda (2017): exposure to tech-change related employment risks (RTI) correlated to support for redistribution 面对技术变革相关的就业风险与对再分配的支持之间相互联系。 Politics of social status (Gidron and Hall 2017 on Brexit, Hochschild 2016 on US right): importance of subjective perceptions of tech change alongside objective measures 社会现状的政治：对技术变革的主观认识的重要性，及其客观措施。 (Gidron和Hall关于英国脱欧，2017年;Hoschschild关于美国的权利，2016年) 16

To recap 扼要重述 Fear of massive tech-unemployment and economic studies aimed at quantifying such risks 对大规模技术型失业的恐惧，经济研究旨在量化这类风险。 These fears may be among the causes of the rapidly changing political sentiment and attitudes in many advanced countries 这些恐惧的原因可能是很多发达国家快速变化的政治情绪和态度。 Nevertheless, no specific attention (with the exception of Thewissen and Rueda, 2017) in the literature so far devoted to identify the role of exposure to tech-unemployment risks in accounting for political preferences and behavior 然而，目前的文献资料中，没有针对确定技术型失业风险对政治倾向性和政治行为作用的特别关注。 17

Contribution by Sacchi, Guarascio and Vannutelli (2018) Sacchi,Guarascio和Vannutelli的研究成果（2018年） This paper: 这份报告： a) provides evidence of routine-biased technological change in Italy, relying on INAPP ongoing research program and findings; 1）意大利国家公共政策分析研究所持续开展的研究项目及其成果，为意大利技术变革提供了实证资料。 b) links together an economics approach to the technology-employment relationship with a comparative political economy approaches investigating the relationship between labor market insecurity and socio-political preferences 2）用经济方法分析技术与就业的关系，以及用政治经济学比较方法分析劳动力市场不稳定性与社会-政治倾向性的关系 RQ – Is there a relationship between the exposition to tech-unemployment risks and policy preferences with respect to redistributive policies? 未完待续-陷入技术型失业风险和再分配政策的倾向性真的有联系吗？ 18

Data and descriptive evidence 数据和描述性证据 Employment dynamics vs degree of task routinarity (RTI (2012) quintiles, % 2011-2016) 就业的动态变化与工作程序性的程度[技术变革的相关就业风险五分位数（2012年），2011-2016年%] 19

Data and descriptive evidence 数据和描述性证据 对基本收入和最低收入的支持与工作程序性的程度（技术变革相关就业风险五分位数） Support for basic income and minimum income vs degree of task routinarity (RTI quintiles) 20

Support for basic income vs RTI and controls Econometric strategy and results 计量经济学策略和成果 对最低收入的支持与技术变革相关的就业风险及控制 Support for basic income vs RTI and controls   Supports UBI High-RTI 0.012 0.061 0.121** [0.036] [0.041] [0.057] Perceived tech unempl. up to 15-34% -0.004 -0.011 [0.067] Perceived tech unempl. over 35% 0.087 0.072 [0.064] ESS controls Yes Constant 1.033*** 1.060*** 1.511*** 1.318*** [0.231] [0.271] [0.357] [0.362] Observations 1069 790 364 359 R2 0.031 0.047 0.108 0.120 Controls: Positive correlation with vars. capturing estimated (non-tech) unemployment as well as with variable re. income support to migrants working since at least one year Negative correlation with ‘female’ dummy as well as with vars. capturing regional unemployment (puzzling) 21

Support for minimum income vs RTI and controls Econometric strategy and results 计量经济学策略和成果 Support for minimum income vs RTI and controls   Supports GMI High-RTI 0.066** 0.090*** 0.142*** [0.030] [0.034] [0.050] Perceived Tech Unempl 15-34% -0.046 -0.050 [0.057] [0.056] over 35% 0.011 -0.001 [0.055] ESS controls Yes Constant 0.461** 0.675*** 0.656** 0.466 [0.188] [0.220] [0.294] [0.288] Observations 1046 775 359 355 R2 0.058 0.093 0.099 0.118 Controls: Positive correlation with ESS vars. capturing support for generic redistributive policies, estimated (non-tech) unemployment, as well as (strong) with variable re. income support to migrants working since at least one year Negative correlation with interaction btw regional trade openness and regional unemp. rate 22

Econometric strategy and results 计量经济学策略和成果 Opposition to both UBI & GMI vs RTI and controls None highRTI -0.062** -0.069** -0.109** -0.024 -0.106** [0.026] [0.031] [0.044] [0.047] [0.043] Female 0.003 0.025 0.036 0.052 0.034 [0.023] [0.027] [0.040] age 0.001 0.003*** 0.002 [0.001] [0.002] Years of full-time education completed -0.005 -0.006 -0.012** -0.004 -0.011** [0.003] [0.005] [0.008] Supports Govmt Redistribution (dummy) -0.081 -0.090 -0.029 -0.126 -0.028 [0.053] [0.060] [0.077] [0.092] [0.076] Other ESS controls Yes Constant 0.324** 0.176 0.292 0.507 0.323   [0.160] [0.202] [0.263] [0.315] [0.261] Observations 1143 834 382 359 R-sq 0.04 0.08 0.09 0.17 0.10 Controls: Negative correlation with ESS vars. capturing support for estimated (non-tech) unemployment, as well as (strong) with variable re.income support to migrants working from at least one year 23

Summary of findings 结论小结 Key results: 关键结果 Perceived tech-unemployment (based on individuals' assessment of their occupational risk) is somewhat – but only lightly – correlated with an objective measure based on task substitutability (RTI) 对技术型失业的主观判定（根据个人对其职业风险的评估）与根据工作可持续性采取的客观措施有一些相互联系。 The Italian case displays high levels of support for both UBI and GMI, but with the highest support for GMI, thus for a means-tested, conditional measure - support confirmed by both descriptive evidence and econometrics 意大利的情况是对UBI和GMI的支持水平较高，尤其是GMI的支持水平最高，因此需要开展家计调查，采取有限制条件的措施— 这需要描述性证据和计量经济学方法给予确定。 Despite mixed evidence as regards UBI, the RTI is strongly correlated with support for GMI and significantly reduces opposition to the introduction of some sort of income protection measure 虽然在UBI方面存在混合证据，但是技术变革的相关就业风险与对GMI的支持非常有关系，并能够明显降低对使用某些收入保护措施的反对。 24

Conclusions 总结 The risk of technological unemployment is an urgent issue not only per se but also due to its impact on socio-political preferences 技术型失业风险不仅本身是一个紧急的问题，并且还影响了社会-政治倾向。 Additional research (both quantitative and qualitative) is needed on preferences for redistribution and effects of tech-change-related risks on social status and perceptions 需要开展额外研究（定量和定性），方向为再分配倾向性和技术变革带来的风险对社会状况和认知方面的影响。 Research needed to devise how to craft social and labor policies that address such issues. Well-crafted redistributive policies may contribute to reduce fears and reconcile objective and perceived social conditions 需要开展研究，设计能够解决上述问题的社会和劳动政策。设计良好的再分配政策也许对降低恐惧、调和社会环境的客观状况和主观认知有促进作用。 25